分担研究報告小児心停止救命率向上のための AED を

平成 22 年度厚生労働科学研究費補助金
循環器疾患・糖尿病等生活習慣疾病対策総合研究事業
循環器疾患等の救命率向上に資する効果的な救急蘇生法の普及啓発に関する研究
（H21-心筋-一般-001）
（研究代表者
丸川征四郎）
平成 22 年度研究報告
分担研究報告
小児心停止救命率向上のための AED を含めた包括的研究
研究分担者
清水
直樹
東京都立小児総合医療センター救命・集中治療部
平成 23(2011)年 3 月
医長
目
１．研究者名簿
次
················································
２
················································
３
２．研究報告書
研究要旨
課題１．小児「院内」「院外」心肺停止症例登録基盤の確立に関する研究
A.研究目的、B.研究方法、C.研究結果、D.考察
················
３
課題２．小児に対する胸骨圧迫の強さと心肺蘇生の品質モニタリングに関する研究
A.研究目的、B.研究方法、C.研究結果、D.考察
··················
５
課題３．Heart rate variability(HRV)を用いた小児心肺停止予測に関する研究
A.研究目的、B.研究方法、C.研究結果、D.考察
··················
７
Ｅ．結論
··········································
９
Ｆ．健康危険情報
··········································
９
Ｇ．研究発表
··········································
９
Ｈ．知的財産権の出願、登録情報
··········································
９
３．資料
清水
1
研究者名簿
研究分担者清水
直樹
研究協力者黒澤
茶茶
本間
順
太田邦雄
新田雅彦
斉藤
修
新津健裕
井上信明
池山貴也
帯包エリカ
丸川征四郎
清水
東京都立小児総合医療センター救命・集中治療部
国立成育医療センター研究所成育政策科学研究部
東京都立小児総合医療センター救命・集中治療部
国立成育医療センター研究所成育政策科学研究部
静岡県立こども病院救急総合診療科
東京都立小児総合医療センター救命・集中治療部
千葉大学医学部小児病態学
金沢大学医薬保健研究域医学系血管発生発達病態学
大阪医科大学救急医学
東京都立小児総合医療センター救命・集中治療部
東京都立小児総合医療センター救命・集中治療部
東京都立小児総合医療センター救命・集中治療部
東京都立小児総合医療センター救命・集中治療部
国立成育医療センター研究所成育政策科学研究部
フィラデルフィア小児病院集中治療部
亀田総合病院小児科
医誠会病院
2
小児心停止救命率向上のための AED を含めた包括的研究
清水直樹 1)2)、黒澤茶茶 1)2)3)、本間順 4)1)、太田邦雄 5)、新田雅彦 6)、斉藤修 1)、
新津健裕 1)、井上信明 1)、池山貴也 1)2)7)、帯包エリカ 8)、丸川征四郎 9)
1)
東京都立小児総合医療センター救命・集中治療部、2)国立成育医療センター研究所成育政策
科学研究部、3)静岡県立こども病院救急総合診療科、4)千葉大学医学部小児病態学、5)金沢大
学医薬保健研究域医学系血管発生発達病態学、6)大阪医科大学救急医学、7)フィラデルフィア
小児病院集中治療部、8)亀田総合病院小児科、9)医誠会病院
研究要旨：
平成 18−20 年度の厚生労働科学研究費補助金「循環器疾患等生活習慣疾病対策総合
研究事業」「自動体外式除細動器 AED を用いた心疾患の救命率向上のための体制構築に関する研
究」
（代表研究者：丸川征四郎）の「小児心肺停止例への AED 普及にかかわる研究」の継続として、
「小児心停止救命率向上のための AED を含めた包括的研究」を行った。今年度研究としては、主
に以下３課題を更に発展させ、最終年度研究へ継続することとした。
課題１：
小児「院内」「院外」心肺停止症例登録基盤の確立に関する研究
課題２：
小児に対する胸骨圧迫の強さと心肺蘇生の品質モニタリングに関する研究
課題３： Heart rate variability; HRV を用いた小児心肺停止予測に関する研究
小児「院内」心停止症例登録基盤については、小児 ECPR 症例登録をも包括するかたちで、全国展
開準備が完了した。前年度課題として見出されたデータクリーニングと解析のプロセスを解決し、
最終年度には全国登録システムとしての運用を開始、海外との共同研究へとつなげたい。小児「院
外」心停止症例登録基盤については、総務省全国データなど別の研究スキームで実施した。小児
心肺蘇生における胸骨圧迫の新しい指標として、新たな絶対値指標を提唱し、ガイドライン 2010
作成過程において国際的貢献を果たした。これらの目標値に対して実際の圧迫の深さをモニタリ
ングすることが必須であると考えられ、それを可能とする小児用の各種器機開発に結びつける研
究も行った。最終年度にはこれらを基盤として、PBLS 指導対象となる市民への啓蒙方略、救急隊
による小児への AED 使用状況や AED ホームユースに関する研究を進め、さらに小児病院前救護に
関する国際比較研究にもとづき、コンセンサス 2010 を前提とした教育体制と併せた提言をはかっ
てゆく。小児院内心停止においては、その直接的原因に関する分析結果より、MET 導入による早
期の介入で、小児院内心肺停止の救命率向上を期待することができると考えられた。これを目的
として、パルスオキシメーターを用いた heart rate variability; HRV 解析による小児心肺停止
予測と合わせたデータ収集をさらに進め、最終年度研究で総括報告を行うこととした。
課題１：
小児「院内」「院外」心肺停止症
担者：清水直樹）において、わが国における
例登録基盤の確立に関する研究
小児 AED の効果的な普及法を検討するにあた
っては、小児心原性心停止の国内疫学調査が
１−Ａ．
研究目的
不可欠であると考えた。
平成 18−20 年度の厚生労働科学研究費補助
日本全国の小児心停止症例の疫学調査を目
金「循環器疾患等生活習慣疾病対策総合研究
的として、小児心肺蘇生レジストリの web 登
事業」
「自動体外式除細動器 AED を用いた心疾
録基盤を完成させた。昨年度は試験的に数施
患の救命率向上のための体制構築に関する研
設からのデータ入力と一部のデータ解析を行
究」
（研究代表者：丸川征四郎）の「小児心肺
い、様々な問題点が認識された。今年度はそ
停止例への AED 普及にかかわる研究」
（研究分
れらの解決を図り、システムの齟齬を解決し
清水
3
たうえで、全国展開の準備を整え、海外との
例であった。年齢は 0−31 歳（平均値 3.9 歳、
共同研究への展開に備えた。
中央値 1 歳）で、性別は、
男性 64 例（41.0％）、
女性 88 例（56.5％）、不詳/記載なし４例
小児においては MET（Medical Emergency
（2.6％）であった。
Team）の重要性が再認識されたことから、MET
事例発生時の波形とその予後に関しての検
対応となった症例の登録フォームの作成にも
討を行ったところ、156 例中 39 例はデータ欠
着手した。
損のため解析対象となった症例は、117 例で、
また、成人領域（平成 19-21 年度厚生労働
科学研究費補助金
循環器疾患等生活習慣病
全体の生存退院率は 35％であった。脈拍が触
対策総合研究事業
「急性心筋梗塞症と脳卒
れない症例は 56 例（生存退院率 25％）、脈拍
中に対する超急性期診療体制の構築に関する
は触れるが循環不全を伴う症例は 61 例（生存
研究」主任研究者野々木宏：国立循環器病セ
退院率 38％）で、脈拍は触れるが循環不全が
ンター心臓血管内科）との連携も継続した。
ある症例の方が予後が良い結果となった。CPA
なお、小児院外心肺停止の疫学調査につい
前の既往（基礎疾患）では、呼吸障害が最も
ては、総務省ウツタイン全国調査データを利
多く（約 40％）、その他、心血管系の問題や、
用することで、別の研究スキームで進めるこ
中枢神経障害、先天奇形が続いた。心停止の
ととなった（日本循環器学会蘇生科学委員
直接の原因としは、循環不全（47％）が最も
会調査研究）
。
多く、呼吸不全、不整脈、代謝電解質異常の
順となっていた。
１−Ｂ．
発生時の状況は、目撃のある心停止が多く、
研究方法
昨年度は、対象施設 4 施設（国立成育医療
110 例（70.5％）、反対にコード発令なしが 32
センター、長野県立こども病院、静岡県立こ
例（20.5％）であったが、今回の登録施設は
ども病院、兵庫県立こども病院）から小児心
すべて PICU を持つ小児病院であり、PICU で
肺蘇生レジストリに登録した。今年度はさら
の発生が多いことに起因していると考えられ
にデータ集積を重ね、登録されたデータ（2002
る。経過中に VF/VT になった症例は 28 例
年 3 月〜20010 年 12 月）を解析した。
（18％）、そのうち 18 例に除細動が行われて
いた。
さらにデータの一部を成人主体のデータベ
ースに登録し、その枠組みにおいて小児と成
自己心拍再開は 88 例（56.4％）で、発見か
人における違いを検討することは昨年度と同
ら循環再開までの時間は 1-144 分（平均 20.8
様である。
分、中央値 10 分）であった。
今年度は、データ収集のプロセスにおける
Web 登録画面の主な改良点は、①必須項目
データ欠損の問題を解決するため、 we ｂ
（項目を入力していなければ、データの保存
登録画面に工夫を加え良質なデータが収集さ
が行えない）を設定することによりデータの
れやすい環境を整えた。また、入力に際して
欠損を少なくした。②時間軸の間違いを少な
の解釈の相違が発生しないように、データ入
くするため、日付や時刻を自動計算で入力出
力ガイドのマニュアルも邦語で作成した。
来るようにした。③事例発症と時間経過が合
わないものに関しては、アラートを出す。④
１−Ｃ．
予想される入力値から極端に外れた値に関し
結果
ては、確認のアラートを出す。といった点で
2002−2009 年のデータの集計、解析結果を
ある。
添付に示す（第３８回日本集中治療医学会学
術集会発表スライド）。登録された症例は 179
新しい登録画面を用いて、2010 年の症例の
例であったが、解析の対象となった症例胸骨
入力を依頼した（都立小児総合医療センター、
圧迫または除細動が実施された症例）は 156
静岡県立こども病院、長野県立こども病院、
清水
4
兵庫県立こども病院）。登録された症例は、44
は、各地域独自のデータを包括的に収集する
例。事例発生時の波形とその予後に関しての
作業に加えて総務省全国データの併用が望ま
検討を行ったところ、データ欠損はなく、44
しく、病院外心停止記録活用研究会との協力
例全てが解析対象となった。
体制を確立しつつある。次年度以降はこれを
解析を行った一部のデータの結果は、添付
基盤として、救急隊による小児への AED 使用
に示す。全体（44 例）の生存退院率は 51％と
状況に関する研究を進め、さらには小児病院
高く、脈拍が触れない症例は 13 例（生存退院
前救護に関する国際比較研究にもとづき、コ
率 62%）、脈拍はあるが循環不全を伴う症例が
ンセンサス 2010 を前提とした提言と、遠隔シ
31 例（生存退院率 45％）であった。
ミュレーションシステムを含めた教育体制の
提言とをはかってゆきたい。
また、データの入力が一時保存のままで放
置されることのないよう、さらに未入力項目
を極力減らせるよう定期的にデータ管理を行
課題２：小児に対する胸骨圧迫の強さと心
う方向とした。
肺蘇生の品質モニタリングに関する研究
また、Web 登録画面の改良が終わったと
２−Ａ．
ころで、対象施設を広げるにあたって、デー
研究目的
タ入力に関するマニュアル（添付）の作成を
平成 18−20 年度の厚生労働科学研究費補助
行い、不明点等を減らし、データの質を上げ
金「循環器疾患等生活習慣疾病対策総合研究
るよう努めた。
事業」
「自動体外式除細動器 AED を用いた心疾
患の救命率向上のための体制構築に関する研
１−Ｄ．
究」
（研究代表者：丸川征四郎）において、小
考察
児心肺蘇生における至適な胸骨圧迫の深さに
今回の研究から、登録 Web 画面の改良によ
関する検討を行った。
り、データの欠損が明らかに少なくなった。
1−8 歳の胸部 CT 画像から検討した結果より、
2010 年の結果は、2002−2009 年のデータと解
離があるが、今回は症例数も少なく、重複例
小児における至適な胸骨圧迫の深さは、
「胸の
もみられたことが要因と考えられた。今後は
厚みの１/3」と考えられたが、その後行った
対象施設を拡大し、症例集積を行う予定であ
人形を用いた「胸骨圧迫の深さの検証」に関
る。
する研究では、実際の圧迫の深さは目標値よ
さらにはデータ解析のためのソフト開発、
りも浅くなることが示された。以上２つの論
あるいはデータ解析の方法を明確にすること
文は、日本集中治療医学会雑誌に掲載された
が必要であると考えられ、今後の課題として
（日本集中治療医学会雑誌 2009;16:27-31、
残っている。
2010;17:173-177）。
昨年度は、1-8 歳の年齢範囲をひろげて、0
心停止の直接原因に関する分析の結果より、
小児では、循環不全、呼吸不全を経て心停止
歳から 15 歳までの胸部 CT 画像からの検討結
に至る経路が容易に推察される。それゆえ、
果に関して、引き続き解析を行った。
MET 導入による早期の介入が予後を改善させ
歳から 15 歳までの CT 画像から胸郭前後径を
る可能性が期待される。
計測し、各年齢における胸郭前後径の平均値
0
MET 導入に際しては、研究課題３の heart
を算出し、各年齢層（１歳未満、1−8 歳、8
rate variability; HRV を用いた小児心肺停
歳以上）における胸骨圧迫の指標を検討する
止予測と合わせて検討を進めることにより、
こととした。
今年度は、この結果をもってして国際蘇生
小児院内心肺停止の救命率向上を期待するこ
連絡委員会（ILCOR）の 2010 Consensus on
とができると考えている。
Science
なお、小児院外心停止の疫学研究について
清水
5
with
Treatment
Recommendation
(CoSTR)に対して、国際的影響を与えた。ま
ける胸骨圧迫の至適な深さを胸郭前後径と
た、わが国の JRC（日本版）ガイドライン 2010
1/3 であると考え、各年齢の胸郭前後径の 1/3
へ反映させることができた。
に相当する深さを算出した。算出されたデー
タより、各年齢層（１歳未満、1−8 歳、8 歳以
現行のコンセンサス 2010 や各地域のガイ
ドラインにおける小児の胸骨圧迫の深さの指
上）における適切な数値での指標を検討した。
標は「胸の厚みの 1/x」で示されている。そ
今年度は、これらの研究結果を ILCOR の国
の至適な深さは「胸の厚みの 1/3」と考えら
際会議、ならびに JRC（日本版）ガイドライ
れるが、今後は、実際の現場における胸骨圧
ン作成合同委員会の場で発表し、CoSTR やガ
迫の質に関するモニタリングが必須であり、
イドライン作成への影響を与えた。
また、胸骨圧迫深度測定機器開発について
その際には、成人同様の数値での指標が必要
も、各種企業との共同開発を始めた。
であると考えられ、各種医療機器の開発基盤
を整えた。
２−Ｃ．
２−Ｂ．
結果
0−15 歳の小児 447 例の内訳と、胸郭前後径
研究方法
（①）および胸骨後面－椎体前面間距離（②）
対象は、2002 年 3 月から 2008 年 8 月まで
に国立成育医療センター（現国立成育医療研
の計測結果を、各年齢の身長・体重の平均値
究センター）で胸部 CT を施行した０歳から
ならびに SD 値と共に Table 1 に示した。身
15 歳までの小児患者 3068 名。このうち胸郭
長・体重は各年齢標準値より小さい傾向がみ
内に病変を有するもの、胸郭の形態に影響を
られたが、対象症例には血液腫瘍疾患等の全
及ぼす可能性のある基礎疾患を有するもの、
身消耗性疾患が含まれており、その影響も一
年齢および身長が各年齢の標準偏差から外れ
因と考えられた。
胸郭前後径の 1/2 ならびに 1/3 の深さで胸
るものを除外し、最終的に、447 例について
骨圧迫したと仮定して算出された、残存する
検討を行った。
計測対象 447 例の計測にあたっては、胸部
胸骨後面－椎体前面間距離を Fig. 2 に記載し
CT 検査結果で乳頭線上に一致する断面の画
た。また、胸郭前後径の 1/2 ならびに 1/3 の
像を選択し、この画像上で胸郭前後径と胸骨
深さで圧迫したと仮定した際に、残存する胸
後面—椎体前面間距離を計測した。胸郭前後径
骨後面－椎体前面間距離が 0mm 未満および
とは胸骨正中線上での皮膚表面から背面の皮
10mm 未満となる頻度を Table 2 に記載した。
膚表面までの距離とし（Fig. 1 ①）、胸骨後
次に、胸郭前後径とその 1/2 ならびに 1/3
面—椎体前面間距離とは、この正中線上での胸
に相当する深さの平均を計算した（Fig.3,
骨後面から椎体前面までの距離とした（Fig.
Table 3）。さらに、1/3 に相当する深さに関
1 ②）。
しては、各年齢層（１歳未満、1−8 歳、8 歳以
更に、胸郭前後径の 1/2 ならびに 1/3 の深
上）でその分布を表に示した（Fig.4）。これ
さで胸骨圧迫したと仮定し、その際に残存す
らの結果より、胸骨圧迫の深さの目標値は、1
る胸骨後面－椎体前面間距離を①・②から
歳未満の乳児では 30±5mm（25-35mm）、1−8
各々推計して、年齢毎に解析した。また、胸
歳では 40±5mm（35-45mm）、8 歳以上では 50
郭前後径の 1/2 の深さで圧迫したと仮定した
±5mm（45-55mm）が適当であると考えられた。
際に、残存する胸骨後面－椎体前面間距離が
２−Ｄ．
0 mm 未満および 10 mm 未満となる頻度を算出
CoSTR 作成にあたり、胸骨圧迫の深さに
した。これらの算出では胸郭の変形や組織の
関する話題は小児のみならず成人においても
圧縮性などは考慮していない。
大きくとりあげられた。その中では、指標を
次に、これまでの研究結果より、小児にお
清水
考察
6
価することを目的とする。
決めることも大切であるが、今後はその決め
られた目標値に対して適切な圧迫が行われて
昨年度研究をふまえ、今年度はさらに症例
いるかどうかのモニタリングが重要視されて
数を蓄積し、最終年度の結論につなげること
いる。
とする。
CoSTR2005 においては、胸骨圧迫の深さは、
３−Ｂ．
成人では「1.5-2 インチ（38−51mm）」
、「小児
研究方法
パルスオキシメーター（OxiMax®、N-600x）
では胸の厚みの 1/3」と推奨されていたが、
議論の結果、CoSTR2010 では、
「乳児では胸の
によるプレチスモグラフより算出された心拍
厚さの少なくとも 1/3、あるいは 4cm、小児で
信号を、心電図信号とを合わせて
は胸の厚さの少なくとも 1/3、あるいは 5cm」
Memcalc/Tonam®（GMS Co., Ltd.)で両者同時
と改定された。
に心拍変動解析を行う。その両者の相関関係
今回の研究では、心肺蘇生時に胸骨圧迫の
評価するべく、健常者を対象とした研究設計
深さのモニタリングを行うことを前提として、
を行う。
Memcal®とは、時系列解析システムの一つで
小児においても数値での指標を検討した。そ
の結果、その目標値は、1 歳未満の乳児では
以下のような特徴を有する。
30± 5mm（ 25-35mm ）、 1− 8 歳では 40± 5mm
時系列として生体情報を処理解析する場合、
（35-45mm）、8 歳以上では 50±5mm（45-55mm）
時間ドメインのみではその解析は不十分で、
が適当であると考えられた。この結果より、
その解析には周波数ドメインを用いて、出力
日本人小児に対して CoSTR2010 をそのまま適
機序の解明を行うことが重要である。そのた
応することは圧迫が深くなりすぎる可能性が
めに最大エントロピー法（Maximum Entropy
示唆され、JRC ガイドラインでは「胸の厚さ
Method ; mem）を用いた Memcal®は、生体情
の約１/3」が推奨されることとなった。
報を有限長離散データおよび特性ゆらぎを有
今後は、この目標値を使って、現在成人に
する多重周期構造としたみた場合に、高い分
使用されている心肺蘇生モニタリング機器を
解能をもつ時系列解析方法として確立した方
小児に対して応用可能であるかの検討や、小
法である。従って、従来の FFT や RA といった
児に使用出来るモニタリングシステムの開発
解析法よりも有用といわれている。
が必要であると考えている。
３−Ｃ．
研究課題３：
Heart rate variability; HRV
小児 22 症例で心電図およびパルスオキシ
を用いた小児心肺停止予測に関する研究
メーターの 2 チャンネルで HRV 解析を行った
（後述
３−Ａ．
結果
研究目的
研究結果
頁 1-5）。
パルスオキシメーターは HRV 解析を行うこと
小児においてパルスオキシメーターを用い
ができる可能性が高いと考えられ、今後心肺
た経皮酸素飽和度測定は、その非侵襲性から
停止予測モニターとして有効である可能性が
広く適応を有した生体情報モニターとして認
示唆された。
められている。
一方で、心拍変動（ Heart rate
３−Ｄ．
variability ; HRV）が、成人の突然死を予測
考察
はじめに、パルスオキシメーターで得られ
する因子として注目を集めている。
た PP データおよび心電図で得られた RR デー
この二つの方法を用いて、パルスオキシメ
タそれぞれの外れ値に対して外れ値処理を
ーターから HRV 解析可能か、またそれが小児
Memcalc/Win で施行した。その上で双方の除
心肺停止の予測パラメーターとなりうるか評
外データがない部分で 300sec をセグメント
清水
7
長として解析を行った。このような処理の上
る小児用自動胸骨圧迫装置と、その品質モニ
では、上記の結果を得ることができ、今後パ
タリング器機開発を開始した。
ルスオキシメトリーの波形取得改善が得られ
れば、より一層心肺停止予測モニターとして
最終年度にはこれらを基盤として、PBLS 指
の有用性が示していけることと考えられた。
導対象となる市民への啓蒙方略、救急隊によ
る小児への AED 使用状況や AED ホームユース
に関する研究を進め、さらに小児病院前救護
Ｅ．結論
小児「院内」心停止症例登録基盤について
に関する国際比較研究にもとづき、コンセン
は、小児 ECPR 症例登録をも包括するかたちで
サス 2010 を前提とした遠隔シミュレーショ
国内データ収集基盤がほぼ整ったが、昨年度
ンシステムを含めた教育体制教育体制と併せ
研究において、データクリーニングと解析の
た提言をはかってゆく。
プロセスに課題が残った。
小児心停止の直接原因に関する分析の結果
より、小児では、循環不全、呼吸不全を経て
今年度はこの点を様々な方略で解決し、東
収集
心停止に至る経路が容易に推察される。それ
に参画することでデータ規模を飛躍的に拡張
ゆえ、MET 導入による早期の介入が予後を改
し、最終的には小児集中治療ネットワークに
善させる可能性が期待された。その際には、
重畳することで、全国症例登録基盤の完成に
heart rate variability; HRV を用いた小児
向かう準備も整った。今後は、海外との共同
心肺停止予測と合わせて検討を進めることに
研究へとつなげたい。
より、小児院内心肺停止の救命率向上を期待
京都立小児総合医療センターもデータ
することができると考えた。
小児「院外」心停止症例登録基盤について
は、各地域独自のデータを包括的に収集する
これを目的として、パルスオキシメーター
作業を継続的に進める努力に加え、総務省全
を用いた heart rate variability; HRV 解析
国データの併用が望ましいため、別の研究ス
による小児心肺停止予測と合わせたデータ収
キームで実施することとなった。
集をさらに進め、最終年度研究で総括報告を
行うこととした。
小児心肺蘇生における胸骨圧迫の新しい指
Ｆ．研究危険情報
標として、新たな絶対値指標を提唱し、ガイ
なし
ドライン 2010 作成過程において国際的貢献
を果たした。これらの目標値に対して実際の
圧迫の深さをモニタリングすることが必須で
Ｇ．研究発表
あると考えられ、それを可能とする小児用の
Comparison of North American and Japanese
各種器機開発に結びつける研究が必要とされ
pediatric chest depths during simulated
た。
chest
compressions
using
computer
Care
Medicine
tomography
Society
このための具体的数値として、1 歳未満の
of
Critical
乳児では 30±5mm（25-35mm）、1−8 歳では 40
Scientific Meeting, Miami, USA, January
± 5mm （ 35-45mm ）、 8 歳以上では 50 ± 5mm
2010
（45-55mm）を提唱した。これらの目標値に対
(Braga M)
して実際の圧迫の深さをモニタリングするこ
小児の心肺蘇生〜予防から PCAS まで〜
とが必須であると考えられ、それを可能とす
清水
8
日本集中治療医学会学術総会（第３８回）、横
栃木、2010 年 9 月【シンポジウム】
浜、2011 年２月【シンポジウム】
（演者：新津健裕）
（演者：清水直樹）
公表されている Work Sheet について
新生児・小児集中治療委員会
日本蘇生科学シンポジウム（第 3 回）、
委員会報告（3）
小児の院内心停止症例レジストリ
埼玉、2010 年 9 月
日本集中
（演者：黒澤茶茶）
治療医学会新生児小児集中治療委員会・
JSICM-PICU network・小児院内心停止レジス
トリグループ
小児心肺蘇生レジストリ始動
〜日本からの
日本集中治療医学会学術総会（第３８回）、横
エビデンス創出に向けて〜
浜、2011 年２月【シンポジウム】
日本小児科学会総会（第 113 回）、
（演者：黒澤茶茶）
岩手、2010 年 4 月
（演者：黒澤茶茶）
東京都立小児総合医療センターにおける小児
ECMO 管理の展開
ガイドライン 2010 から 2015 へ
〜日本から
日本集中治療医学会学術総会（第 38 回）、
の情報発信をめざして〜
横浜、2011 年 2 月【シンポジウム】
日本小児科学会総会（第 113 回）、
（演者：本間順）
岩手、2010 年 4 月【シンポジウム】
（演者：新田雅彦）
開院 10 ヶ月間の院内蘇生の検討
千葉集中治療研究会、
ガイドライン 2005 から 2010 へ
〜科学が示
千葉、2011 年 1 月
す新しいガイドライン〜
（演者：本間順）
日本小児科学会総会（第 113 回）、
岩手、2010 年 4 月【シンポジウム・座長】
（演者：太田邦雄）
「小児蘇生 2010 からの展望」
小児心肺蘇生法の品質改善に関わる研究
日本蘇生学会（第 29 回大会）、
心肺蘇生に関わる Consensus 2010 最新の話
栃木、2010 年 9 月【シンポジウム】
題
（演者：黒澤茶茶）
日本集中治療医学会総会（第 37 回）、広島、
2010 年 3 月【シンポジウム】
（演者：黒澤茶茶）
小児蘇生の疫学：院内心停止
日本蘇生学会（第 29 回大会）、
栃木、2010 年 9 月【シンポジウム】
Domestic and international comparison of
（演者：本間順）
pediatric vs. adult in-hospital cardiac
わが国の小児蘇生疫学：小児院外心停止の現
arrest
状と課題
日本循環器学会総会（第 74 回）、
日本蘇生学会（第 29 回大会）、
京都、2010 年 3 月
栃木、2010 年 9 月【シンポジウム】
（演者：黒澤茶茶）
（演者：新田雅彦）
黒澤茶茶:
小児蘇生の新しい潮流
新しい小児救急医学に向けた変革小児救命・
日本蘇生学会（第 29 回大会）、
集中治療医学としての再定義：蘇生
清水
9
救急医学、2010; 34(9): 1051-54
Ｈ．知的財産権の出願、登録情報
なし
清水
10
代表記入者名
3.事例
4.直接原因
2009/08/11
14:25
身長
白人
女性
日
その他
療養型リハビリ施設
分
尿路感染症
外傷
マグネシウムの使用
前子癇
子癇
糖尿病
内診
介助分
骨盤位
外診
自然経腟分
頭位
胎位
様式
胎児モニタリング
分
経過
不詳/記載なし
帝王切開
不詳/記載なし
計算
時刻記載なし
カレンダー
その他
不詳
未施行
麻薬依存症またはその治療中
覚醒剤の使用
コカインの使用
アルコール摂取
実施されたが方法が不詳
出後4時間以内の麻薬投与
絨毛膜羊膜炎
母体感染症
母体GBS陽性
なし
妊娠高血圧
あり
帝王切開
特記事項なし
母体の状態
母体の妊婦検診履歴
なし/記載なし
不詳/記載なし
不詳/記載なし/適応外（新生児）
その他（以前の入院時）
記載なし
年齢不詳/記載なし
00:00
いいえ
出生日不詳/記載なし
1.2 新生児（入院中の出生、または他院出生後の搬送）
詳細
入院時のCPC/PCPCスコア
院外（今回の入院の原因）
以前の心肺蘇生履歴(該当項目全てをチェック)
ホスピス（在宅を含む）
精神医療/福祉施設
身長不詳/記載なし
体重不詳/記載なし
新生児（入院中の出生）
ｃｍ
その他
黒人
推定年齢
急性期医療施設/病院
ｋｇ
53
日本人以外の東洋人
日本人
男性
2.172
0
出生時刻（はいの場合のみ記載）
はい
自宅
主要な居住先
体重 *
検索画面
人種
性別 *
入院時年齢 *
ログアウト
一時保存
管理ユーザー
ユーザ名
入院中の出生（または他院出生後の搬送）
2009/08/11
出生日 *
カレンダー
いいえ/不詳（そうであれば登録適応外）
時刻記載なし
はい
※外来患者は外来登録あるいは救急外来受け入れ時刻を記載
静岡県立こども病院
入院日 *
1.1 入院時データ
当事例において、胸骨圧迫または除細動を実施された *
病院名
8.CPRの質の評価と問題点
7.事例の予後
6.治療
1.0 事例
関連データ2
5.事例発生時の状態
※フリーコメント欄(最大２０文字) 施設毎通し番号など。
sch2010-2
関連データ1
2.事例発生前の状態
※入力データを修正する場合に検索項目として利用できます。
※システムが自動的に発番する管理番号です。(入力、修正は行えません)
00000003-2011-10
管理コード
データ管理
1.入退院時データ
CPA
不詳/記載なし
多胎妊娠
先天奇形（非心疾患）
退院時転帰 *
詳細
退院時のCPC/PCPCスコア
その他
不詳/記載なし
不詳/記載なし
精神医療/福祉施設
療養型リハビリ施設
時刻記載なし
なし
時刻記載なし
ホスピス（在宅を含む）
なし
その他
肩甲難産
前置胎盤
転帰未決（90日以上経過）
急性期医療施設/病院
あり
2010/02/27
あり
07:18
自宅
退院先
生存退院の場合
生命維持療法の終了の有無
死亡症例の場合
DNARの日時
入院中のDNARの有無
退院（死亡）日時 *
転院
胎盤早期剥離
一過性徐脈
2010/03/01
臍帯巻絡
臍帯脱出
1.3 退院時データ
備考
生存
胎便吸引
チアノーゼ性心疾患
死亡
胎児水腫
非チアノーゼ性心疾患
特記事項なし
出生時に存在した特殊な状況（該当項目を全てチェック）
臍帯血ｐＨ
不詳/記載なし
不詳/記載なし
2
1
APGAR（5分値）
不詳/記載なし
37
推定在胎週数
APGAR（1分値）
カレンダー
カレンダー
検索画面
ログアウト
一時保存
管理ユーザー
ユーザ名
静岡県立こども病院
病院名
8.CPRの質の評価と問題点
7.事例の予後
6.治療
5.事例発生時の状態
4.直接原因
3.事例
2.事例発生前の状態
1.入退院時データ
CPA
その他
ペーシング
その他
不詳
不詳
不詳
17:47
2010/02/12
/
呼吸機能障害
急性脳伷塞、急性脳出血
脳伷塞以外の急性中枢神経障害
慢性的な中枢神経機能低下
肝機能障害
腎機能障害
悪性腫瘍（固形転移または血液腫瘍）
心筋伷塞/狭心症（入院時初発）
心筋伷塞/狭心症（入院前発症）
鬱血性心不全（入院時初発）
鬱血性心不全（入院前発症）
非チアノーゼ性心奇形
チアノーゼ性心奇形
弁膜症
心筋症
外傷
AIDS（HIV陽性なら）
HIV陽性
敗血症
代謝疾患/電解質異常
糖尿病
94
94
SpO2（％）
TAA/AAA
大動脈解離
肺塞栓
36.5
36.8
なし
PartA*
胸骨圧迫や除細動の必要性が最初に認識された時、すでに施工されていた処置(該当するもの全て選択)
2.3 すでに施行されていた処置
先天奇形（心奇形を除く）
肺炎
不整脈（洞性頻拍を除く）
21
26
呼吸（回/分）
低血圧、循環不全
特記事項なし
CPA前の既往 *
/ 35
/ 38
カレンダー
65
53
血圧（mmHg）
/
112
103
心拍（回/分）
カレンダー
カレンダー
カレンダー
2.2 CPA前の既往（該当項目を全てチェック）
20:51
日時
2010/02/12
体温(℃)
時刻記載なしカレンダー
バイタルサイン（CPA前4時間以内に記録された全てのもの、もしくは最後に記載されたもの）
徐脈
頻脈
洞調律
確認時心電図波形
自発呼吸なし（挿管中を含
む）
意識混濁（呼吸・循環あり）
意識清明
意識レベル
鎮静中
その他
病棟内（廊下）
外来（緊急・専門・総合）
検査室
ラウンド（病室）
確認方法
急変前最後の確認時間
分前
いいえ
はい
鎮静や全身麻酔後24時間以内のCPA
0
いいえ
はい
救急外来受診（滞在）後24時間以内のCPA
急変前の患者：確認方法
いいえ
いいえ
はい
はい
麻酔観察室（PACU）退室後24時間以内のCPA
はいの場合、ICU退室日時
ICU退室後のCPA
2.1 心肺停止（CPA）前の状況
(詳細)
他の抗不整脈薬
アドレナリン（エピネフリン）
プロカインアミド
(詳細)
その他
フェニレフリン
ノルアドレナリン
ニトログリセリン（ノルエピネフ
リン）
ドパミン（＞ 3γ）
ドブタミン
血管作動薬の持続投与
パルスオキシメーター
リドカイン
アミオダロン
抗不整脈薬の持続投与
なし
PartB
ペースメーカー
気管切開チューブ
無呼吸/徐拍
無呼吸
侵襲的気道確保
気管チューブ
心電図
モニタリング
動脈ライン
人工呼吸/補助呼吸（CPAP/BiPAP含む）
PGI2製剤
(詳細)
他に施行されていた処置
胃管挿入
酸素投与（鼻カヌラ，マスク，フード，
テント）
鎮静薬/麻薬持続投与（PCA含む）
肺動脈カテーテル
プロスタグランジン持続静注（新生児）
大動脈バルーンパンピング
埋込型除細動器（ICD）
一酸化窒素吸入療法（新生児）
透析/血液浄化（使用中）
鎮静
胸腔ドレーン
処置
臍帯動脈路
臍帯静脈路
骨髄路
中心静脈路
末梢静脈路
薬剤投与経路
21:47
新生児
精神医療/福祉施設入院中
時刻記載なし
記載なし
時刻記載なし
時刻記載なし
時刻記載なし
記載なし
そのコードは、胸骨圧迫または除細動の必要性が認識される前に発令されたものか？
そうであれば発令日時
いいえ
14:25
2009/08/11
心電図モニターが装着された日時
はい
21:47
2010/02/12
医師が到着した日時
院内蘇生コードが発令されたか？
21:47
2010/02/12
いいえ
小児ICU (PICU)
手術室
新生児室
目撃の有無 *
はい
はい
いいえ
カレンダー
カレンダー
カレンダー
カレンダー
不詳/記載なし
一般病棟
応援要請時刻
集中治療
その他
新生児ICU (NICU)
病棟内廊下
トイレ
救急外来
室
検査室（心カテ室を除く）
分
記載なし
重症観察室 (Step-down Unit)
日帰り手術室
成人ICU
心臓カテーテル検査室
リハビリ、介護、精神病棟
麻酔後回復室
計算
成人CCU (Coronary Care Unit)
外来
検索画面
事例発生場所（名称）
年齢不詳/記載なし
事例発生場所（領域） *
推定年齢
カレンダー
その他（見舞い客，従業員を含む）
外傷
産科
外科系（心疾患以外）
療養型リハビリ施設入院中
見舞い客，従業員
外科系（心疾患）
入院中
内科系（心疾患以外）
救急外来
ヵ月
内科系（心疾患）
疾患区分 *
6
時刻記載なし
一般外来
対象区分 *
年齢（事例発生時） *
2010/02/12
胸骨圧迫（または発見時の心電図波形がVF/pulseless VTの際は除細動）が必要と認識された最初の日時 *
3.1 事例
ログアウト
一時保存
管理ユーザー
ユーザ名
静岡県立こども病院
病院名
8.CPRの質の評価と問題点
7.事例の予後
6.治療
5.事例発生時の状態
4.直接原因
3.事例
2.事例発生前の状態
1.入退院時データ
CPA
検索画面
ログアウト
一時保存
管理ユーザー
ユーザ名
静岡県立こども病院
病院名
8.CPRの質の評価と問題点
7.事例の予後
6.治療
5.事例発生時の状態
4.直接原因
3.事例
2.事例発生前の状態
1.入退院時データ
CPA
イオン化Ca < 1mmol/L または < 4mg/dL
ナトリウム < 125 mEq/L
マグネシウム > 4 mEq/L
呼吸不全
気胸
鎮静
低体温
急性脳伷塞、急性脳出血
痙攣重積
人工呼吸器の作動異常
侵襲的エアウエイの位置異常
侵襲的エアウエイの閉塞（LMA含む）
エアウエイの閉塞
中毒
薬物過量投与
薬物副作用
中毒の問題
カリウム > 6.5 mEq/L
血糖 < 40mg/dl
肺塞栓
肺水腫
アシドーシス（pH < 7.2，動脈，静脈，毛細血管）
心筋虚血（急性冠症候群）/心筋伷塞
代謝異常/電解質異常
不整脈（洞性頻拍を除く）
低血圧、循環不全
不詳/記載なし
4.1 直接の原因（該当項目全てをチェック）
検索画面
ログアウト
一時保存
管理ユーザー
ユーザ名
静岡県立こども病院
病院名
8.CPRの質の評価と問題点
7.事例の予後
6.治療
5.事例発生時の状態
4.直接原因
3.事例
2.事例発生前の状態
1.入退院時データ
CPA
いいえ
洞調律（洞性頻拍を含む）
脈のある心室頻拍（VT）
徐脈
ペースメーカー調律
いいえ
不詳/記載なし
いいえ
死戦期呼吸
補助換気
いいえ
5.3 換気
日時
除細動の回数
カレンダー
カレンダー
カレンダー
カレンダー
カレンダー
AEDが使用された日時
除細動器の種
類/モード
自動体外式除細動器（AED）は使用されたか *
除細動は行われたか。 *
波形
心室細動あるいは無脈性心室頻拍が最初にみられた日時
はい
出力(J)
いいえ
はい
時刻記載なし
記載なし
記載なし
脈拍の再開は
除細動と胸骨圧迫の後の心
電図波形（１項目に○を付
け、詳細はop-defs参照）
カレンダー
カレンダー
カレンダー
開胸心臓マッサージ
記載なし
時刻記載なし
不詳/記載なし
いいえ
不詳/記載なし
不詳/記載なし
カレンダー
時刻記載なし
はい
記載なし
今回の事例の経過中に心室細動あるいは無脈性心室頻拍はみられたか *
5.2 自動体外式除細動器と心室細動/無脈性心室頻拍
はい
脈拍のある患者が、胸骨圧迫が必要となった際に、呼吸があったか
はい
脈拍のある患者が、胸骨圧迫が必要となった際に、意識があったか
上室性頻脈（SVT）
頻脈性心室固有調律（AIVR）
当事例で脈拍のある患者に胸骨圧迫が開始された際の心電図波形
脈拍はあるが胸骨圧迫が行われた場合
無脈性心室頻拍(無脈性VT）
不詳/記載なし
時刻記載なし
記載なし
無脈性電気活動（PEA）
いいえ
21:47
機械的なピストン式CPR
いいえ，事前指示により
心室細動（VF）
はい
2010/02/12
標準の胸骨圧迫
はい
心静止
脈拍がなかった際の最初の心電図波形
最初に脈拍がなくなった日時
蘇生中のいずれかに脈拍がなかった場合
Impedance Threshold Deviceが使用されたか
胸骨圧迫が開始された日時 *
胸骨圧迫の手段
胸骨圧迫（開胸心マを含む）を受けたか *
脈はあるが循環不良があり、胸骨圧迫が必要であったが、一貫して脈拍がなくなることはなかった
脈拍はあるが循環不良があり、胸骨圧迫が必要であったが、その後、脈拍がなくなった
脈拍がなく、胸骨圧迫または除細動が必要であった
当事例に該当するもの *
5.1 発見時の状態
なし
不詳/記載なし
呼気終末CO2検知器
（詳細）
なし/記載なし
施行されているが、方法が不詳/記載なし
その他
レントゲン写真（蘇生事象終了前に結果が出ているもの）
ファイバー喉頭鏡
食道挿管検知器
気管切開チューブ
その他の侵襲的な気道確保/換気
聴診
ラリンゲアルマスクエアウェイ
侵襲的な気道確保後の位置確認の方法(該当項目全てをチェック)
時刻記載なしカレンダー
侵襲的な気道確保（再確保）が行われた日時
いいえ、行われなかった（不成功例も含む）
いいえ、非侵襲的な治療に反応した
いいえ、事前指示により
いいえ、すでに確保済み
はい、侵襲的な気道確保が再確保で成功した
はい、侵襲的な気道確保が1回目で成功した
蘇生中に侵襲的な気道確保（あるいは再確保）が行われたか
気管チューブ
侵襲的な気道確保用具
（詳細）
その他の非侵襲的な換気
口対口
口対バリアデバイス
マスクおよび/あるいはNasal CPAP/BiPAP
バッグバルブマスク
非侵襲的な用具/補助換気(該当項目全てをチェック)
あり
換気/気道確保の種類
検索画面
ログアウト
一時保存
管理ユーザー
ユーザ名
静岡県立こども病院
病院名
8.CPRの質の評価と問題点
7.事例の予後
6.治療
5.事例発生時の状態
4.直接原因
3.事例
2.事例発生前の状態
1.入退院時データ
CPA
記載なし
気管/気管切開チューブ
記載なし
気管/気管切開チューブ
回
単位
バゾプレッシンの投与量 *
いいえ
記載なし
不詳/記載なし
時刻記載なしカレンダー
cc of 1：10000
OR
不詳/記載なし
時刻記載なしカレンダー
記載なし
胸腔刺（針で）
輸血
人工心肺/体外循環を使用したCPR(ECPR)
中心静脈ライン確保
胸腔ドレーン挿入
心エコー
氷で頭部を冷却
なし（下記の選択肢をよく確認すること）
6.3 非薬物的介入
その他の非薬物的介入
ペースメーカー、経皮的
ペースメーカー、経皮的経静脈的または心外膜
心嚢刺
胸腔刺
気管切開/輪状甲状間膜切開（今回の蘇生中に）
その他の薬剤的介入
その他のアルカリ化剤
アトロピン
塩化カルシウム/グルコン酸カルシウム
デキストロースボーラス投与
インスリン併用デキストロースボーラス投与
循環血液量増加のための輸液ボーラス投与
グリコピロレート
その他の抗不整脈薬
硫酸マグネシウム
アドレナリン(ネピネフリン)とバゾプレッシンのボーラス投与以外
神経筋遮断剤/筋弛緩剤
の強心薬
プロスタグランジンE1
ドブタミン
拮抗薬（ナロキソン、フルマゼニル、ネオスチグミ
ドパミン（＞ 3γ ）
ン等）
ノルアドレナリン（ノルネルエピネフリン）
鎮静/導入薬
フェニレフリン
炭酸水素ナトリウム
その他の強心薬
THAM（アルカリ化剤）
抗不整脈薬
アデノシン/Adenocard
アミオダロン/Cordarone
リドカイン
プロカインアミド
なし（下記の選択肢をよく確認すること）
6.2 その他の薬剤的介入
（蘇生時に開始、あるいは事前に開始されていて蘇生中も継続されたものは全て選択）
回
はい
OR
cc of 1：1000
mg.
いいえ
不詳/記載なし
はい
投与回数 *
初回静脈内/骨髄内ボーラス投与の日時 *
その他
静脈内/骨髄内
投与経路（該当項目全てをチェック） *
バゾプレッシンのボーラス投与は行われたか。 *
静脈内/骨髄内投与されたアドレナリンの最大量 *
投与回数 *
初回静脈内/骨髄内ボーラス投与の日時 *
その他
静脈内/骨髄内
投与経路（該当項目全てをチェック） *
アドレナリンのボーラス投与は行われたか *
6.1 アドレナリン/バゾプレッシン
検索画面
ログアウト
一時保存
管理ユーザー
ユーザ名
静岡県立こども病院
病院名
8.CPRの質の評価と問題点
7.事例の予後
6.治療
5.事例発生時の状態
4.直接原因
3.事例
2.事例発生前の状態
1.入退院時データ
CPA
いいえ
℃
37.2
最高
はい
皮膚
皮膚
カレンダー
カレンダー
時刻記載なしカレンダー
時刻記載なしカレンダー
不詳/記載なし
不詳/記載なし
mg/dl
最高
時刻記載なしカレンダー
時刻記載なしカレンダー
mg/dl
記録された日時
06:39
最低
血糖値
2010/02/13
時刻記載なし
時刻記載なし
記載なし
記録された日時
01:21
記載なし
21:47
21:47
2010/02/13
いいえ
2010/02/12
2010/02/12
測定部位（運用上の定義参照）
循環再開後24時間以内の最低/最高血糖値
℃
36
最低
体温
循環再開後24時間以内の最低/最高体温
循環再開後に低体温療法は開始されたか
7.2 蘇生後管理
循環再開した（２０分以上持続）
あるいは蘇生努力が中止された日時 *
死亡-家族からの中止
死亡-医学的理由
死亡-蘇生努力を中止、循環の再開なし
死亡-事前指示
生存-循環再開
蘇生終了の理由 *
最初の循環再開日時 *
はい
胸骨圧迫なしに循環の再開の記録があるか（触診、聴診、ドップラー、動脈圧波形、血圧測定による脈拍/心拍再開の確認） *
7.1 事例の予後
コメント
利用上の問題
機器
ALS/PALS
プロトコールからの逸脱
機能
その他(コメント欄に詳細記載)
その他(コメント欄に詳細記載)
その他(コメント欄に詳細記載)
薬剤/プロトコールについての知識
NRP
メンバーが多すぎる
機器に関する知識
役割に関する知識
チームの監修
リーダー決定の遅延
リーダーシップ
その他(コメント欄に詳細記載)
選択
投与量
投与経路
遅延
薬剤投与経路（血管確保）
検索画面
ログアウト
確認画面表示
一時保存
管理ユーザー
ユーザ名
静岡県立こども病院
病院名
8.CPRの質の評価と問題点
7.事例の予後
6.治療
5.事例発生時の状態
4.直接原因
3.事例
2.事例発生前の状態
1.入退院時データ
CPA
はい
はい
感染防御
いいえ
記載なし
いいえ
記載なし
記載なし
記載なし
記載なし
記載なし
記載なし
遅延
病院全体へ蘇生コードを呼びかける
その他(コメント欄に詳細記載)
薬剤投与経路の記載なし
その他のサインの不備
記載不備
初期の心電図所見の記載なし
刺
機器（除細動器）の故障
指示なしに行われている
指示されているが、行われていない
その他(コメント欄に詳細記載)
開始の遅延、患者への除細動器のアクセスの問題
開始の遅延、パッドあるいはパドルの位置の問題
その他(コメント欄に詳細記載)
開始の遅延、蘇生者が除細動を行えない
背板なし
不詳/記載なし
出力レベルが推奨されているよりも低い/高い
除細動
遅延
胸骨圧迫
その他(コメント欄に詳細記載)
漏出/接続不良
予期しない動脈
遅延
薬剤投与経路（血管確保）
その他(コメント欄に詳細記載)
複数回の気管挿管の試み → 気管挿管の回数
気管挿管を試みたが、不成功
挿管チューブの位置異常の認識遅延
遅延
気道確保に関連した誤嚥
ポケベル・PHS等の問題
その他(コメント欄に詳細記載)
チームメンバー全員には行われなかった(コメント欄に詳細記載)
記載なし
コードチームリーダーのコードシートへのサインなし
気道
いいえ
いいえ
いいえ
いいえ
いいえ
いいえ
換気回数は挿管チューブの位置確認の時間を除いて10/分（小児は20/分）を超えていたか
8.2 蘇生関連事例と問題点
記録
はい
はい
はい
はい
はい
はい
CPR中、侵襲的な気道確保のような処置のために胸骨圧迫が15秒以上（新生児では20秒以上）中断されたか
CPR中、圧迫が10秒以上中断されたか
CPR中、圧迫の速度は100/分を維持していたか
もし使用されていれば
胸骨圧迫の質の評価のために道具あるいは技術が使用されたか
動脈ラインの拡張期血圧は胸骨圧迫の質の評価のために測定されたか
もし測定されていれば、呼気終末CO2は＞10mmHgに保たれていたか
呼気終末CO2はCPRの質の評価のために持続的に測定されていたか
8.1 CPRの質
コメント
利用上の問題
機器
ALS/PALS
プロトコールからの逸脱
機能
その他(コメント欄に詳細記載)
その他(コメント欄に詳細記載)
その他(コメント欄に詳細記載)
薬剤/プロトコールについての知識
NRP
メンバーが多すぎる
機器に関する知識
役割に関する知識
チームの監修
リーダー決定の遅延
リーダーシップ
その他(コメント欄に詳細記載)
選択
投与量
投与経路
遅延
薬剤投与経路（血管確保）
11.5.31
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
第38回日本集中治療医学会シンポジウム4
背景
「新生児・小児集中治療委員会委員会報告」
小児の院内心肺停止症例
レジストリ
•  小児重症患者の集約化の遅れと治療戦略コンセンサ
スの不足が指摘され、各種症例登録基盤の必要性を
認識
日本集中治療医学会新生児小児集中治療委員会
JSICM-PICU network・小児院内心肺停止レジストリグループ
黒澤茶茶1), 2), 3) 清水直樹2), 3) 本間順2) 植田育也4) 志馬伸朗5) 中川聡6) 丸川征四郎7)
1)
静岡県立こども病院救急総合診療科 2) 東京都立小児総合医療センター
救命・集中治療部 3) 国立成育医療研究センター研究所成育政策科学研究部
4) 静岡県立こども病院小児集中治療科 5) 京都府立大学集中治療部 6) 国立成育医療研究センター手術・集中治療部 7) 医誠会病院
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
•  厚生労働科学研究「AEDを用いた心疾患の救命率
向上のための体制の構築に関する研究（丸川班）」
「小児心肺停止例へのAED普及にかかわる研究（清
水分担班）」の枠組みの中で、小児心肺蘇生レジスト
リの構築を検討
•  一昨年、本学会において、「データベース構築・小児
心肺蘇生レジストリ」を報告
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
小児心肺停止症例レジストリ
本システムとJ-RCPR・NRCPRの関係
日本語Web入力
•  米国を中心に展開する NRCPR (National registry of
Cardiopulmonary Resuscitation) に基づくレジストリの
登録項目を選択
•  登録作業はWeb上で展開
•  全国からの症例集積が必要であり、日本集中治療医学会
新生児小児集中治療委員会JSICM-PICU networkと連携
•  国内の院内心停止登録システムである J-RCPR (Japanese
registry of Cardiopulmonary Resuscitation) （厚生労働
科学研究野々木班）＝成人領域との連携
わが国固有の
薬剤・医療制度
を加味し、
ニーズに応じた
データマイニング
必要項目を
自動出力
国内レポート 作成と配布
自施設データ 抽出可能
•  NRCPR＝国際的な連携
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
各施設からの症例登録開始(2008年
NRCPR quarterly report
J-RCPR
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
結果：症例情報
)
＊期間：2002年3月∼2009年12月
度数
日本小児総合医療施設（29施設）
179
総登録データ
156
胸骨圧迫と/または除細動を実施された症例
入院時データ
長野県立こども病院
都立小児総合医療センター
（2010年3月）
国立成育医療研究センター
静岡県立こども病院
兵庫県立こども病院
英訳レポート自動出力
(NRCPR data collection form)
平均
中央値
年齢（歳）
3.9
体重（kg）
13.4
85.1
身長（cm）
入院時データ
性別不詳/記載なし
度数パーセント
4
2.6
最小値
最大値
標準偏差
1
0
31
6.44
7.5
0.742
64.5
14.94
70
36.7
169
39.78
入院時：主な居住先
度数パーセント
73
46.8
男性
64
41.0 急性期医療施設/病院
28
18.0
女性
88
56.5 新生児（院内出生）
26
16.7
2
1.3
12
7.7
15
9.6
人種不詳/記載なし
東洋人
6
150
自宅
3.8 療養型リハビリ施設
96.2 不詳/記載なし
不明
1
11.5.31
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
Modified Utstein Template (1)
結果：症例情報
退院時データ
度数パーセント
死亡
102
生存
41
転帰未決（90日以上）
6
不明
7
退院時：退院先
65.4
パーセント
4/41
9.8
4
3/41
5
4/41
0/41
0
22/41
53.7
70.7 6
12.2 不明
5/41
療養型リハビリ施設
2/41
4.9
不明
5/41
12.2
12.2
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
*Insufficient record
(n=39, 25%) was
excluded
Pulseless
56
% Disch. Alive
25 %
7.3
9.8
% Disch. Alive
25 %
PEA
16
% Disch. Alive
22 %
Asystole
22
% Disch. Alive
17%
Unkown
17
% Disch. Alive
40 %
Poor Perfusion,
Pulse Present
61
% Disch. Alive
38 %
Bradycardia
48
% Disch. Alive
35 %
Other Rhythms
11
% Disch. Alive
45 %
Unkown
2
% Disch. Alive
0%
院内心停止症例：2002年3月∼2009年12月
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
結果：CPA前の状況
CPA前の既往（基礎疾患）
パーセント
12
7.7
4
2.56
救急外来受診（滞在）24時間以内のCPA
8
5.13
鎮静や全身麻酔後24時間以内のCPA
23
14.7
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
結果：事例
発生場所
35 %
8
（％）
度数
麻酔観察室退室後24時間以内のCPA
% Disch. Alive
VF/Pulseless VT
結果：CPA前の状況
ICU退室後のCPA
117
(156*)
7.3
5/41
急性期医療施設/病院
Total Cardiopulmonary Index
Events
41
3/41
4.5 2
3
度数パーセント
29/41
自宅
度数
26.3 生存退院
3.9 1
41
生存退院
退院時：
CPC/PCPC
結果：原因
（％）
直接の原因
（％）
2
11.5.31
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
結果：発生時の状況
平均
目撃の有無
度数
パーセント
110
あり
なし/記載なし
32
不明
14
最小値
0.37
発見から胸骨圧迫までの時間（分）
結果：発生時の状況
最大値
0
標準偏差
13
コード発令の有無
1.67
度数
70.5 あり
20.5 なし/記載なし
9.0 不明
パーセント
非侵襲的気道確保
度数パーセント
BVM
46
29.5
侵襲的気道確保
2
1.3 気管切開チューブ
Mouth to mouth
2
1
26.3
その他
100
64.1
なし
15
9.6
105
度数
パーセント
132
標準の胸骨圧迫
度数
84.6
経過中のVF/VT
開胸心マッサージ
12
7.7 除細動の実施
不明
12
7.7 AEDの使用
パーセント
6.4
1.3 LMA
1
1.3
0.6 その他
1
0.6
24
15.4
67.3 なし
28
18.0
18
11.5
3
1.9
66
既に確保済
実施なし（不成功含む）
5 非侵襲的治療に反応
5
1回で成功
44
再確保で成功
14
侵襲的気道確保
の確認方法
42.3
度数パーセント
3.2 聴診
3.2 食道挿管検知器
61
0
0
28.2 呼気CO2検知器
9.0 胸部X線写真
41
26.3
55
33.3
25
16.0
不詳
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
度数
パーセント
125
あり
静脈内/骨髄内
80.1
111/125
気管内
6/125
不明
8/125
バゾプレッシン
ボーラス投与
度数
その他の薬剤
度数パーセント
5.8
静脈内/骨髄内
9/9
100
気管内
0/9
0
69
44.2
心電図
41
26.3
リドカイン
16
10.3
カルシウム
51
32.3
CVライン確保
26
16.7
3
1.9
輸液ボーラス
42
26.9
輸血
27
17.3
2
1.3
鎮静/導入薬
28
18.0
PM（経静脈/心外膜）
7
4.5
1
0.6
アトロピン
22
14.1
PM（経皮）
1
0.6
その他
7
4.5
筋弛緩薬
15
9.6
胸腔ドレーン
3
1.9
強心剤
55
35.3
ドブタミン
13
8.3
ドパミン>3γ
30
ノルアドレナリン
フェニレフリン
マグネシウム
9
5.8
胸腔穿刺
1
0.6
グルコース
9
5.8
心嚢穿刺
3
1.9
19.2
インスリン＋Glu
2
1.3
その他
2
1.3
16
10.3
PGE1
2
1.3
2
1.3
12
7.8
16
4.5
結果：心肺蘇生の質の評価
蘇生終了の理由
度数パーセント
88
56.4
生存ー循環再開
85
54.5
なし/記載なし
54
34.6
死亡ー循環再開なし
46
29.5
不明
14
2
1.3
23
14.7
9.0 死亡ー家族からの中止
度数パーセント
7
蘇生後２４時間以内
最低体温（℃）
不明
4.5
平均
発見ー循環再開時間（分）
20.8
平均
中央値
最小値
最大値
標準偏差
10
1
144
32.85
中央値
最小値
最大値
標準偏差
35.2
35.6
30
37
1.43
37.3
37.3
35
40.6
1.24
最低血糖値（mg/dl）
101.1
96
27
296
51.02
最高血糖値（mg/dl）
209.3
181
91
513
95.92
最高体温（℃）
その他
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
あり
低体温療法
度数パーセント
炭酸水素ナトリウム
結果：予後
実施あり
非薬剤的介入
13.5
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
度数パーセント
度数パーセント
21
その他
自己心拍再開
その他の薬剤
抗不整脈薬
6.4 アデノシン
9
あり
結果：薬剤的/非薬剤的介入
88.9 アミオダロン
4.8 プロカインアミド
パーセント
39.1
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
結果：薬剤的介入
アドレナリン
ボーラス投与
76.9
10
侵襲的気道確保の実施度数パーセント
胸骨圧迫の手段
120
NPPV
41
度数パーセント
挿管チューブ
評価項目
ETCO2持続モニター
CO2>10mmHg
度数
パーセント
38
24.4
28/38
（73.7）
動脈ライン拡張期血圧の測定
31
胸骨圧迫の質の評価のための機器/技術使用
26
19.9
16.7
圧迫速度：100/分を維持
21/26
（80.8）
圧迫が10秒以上中止されなかった
13/26
（50.0）
処置のために胸骨圧迫が15秒以上中止されなかった
14/26
（53.8）
換気回数：10/分（小児は20/分）を超えていた
17/26
（65.4）
3
11.5.31
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
データ登録に関する問題点
① 誰が、どの時点で、登録するか
② 未入力を減らすためにのシステム上の工夫
–  必須項目の設定
③ データ管理のための人材確保
–  入力されたデータを確認し、不明な点等に関して
は、問い合わせの必要あり
問題点に対する対策
① データの入力時期については、事例発生直後に可能な
限り入力し、3ヶ月を経過した時点で最終登録とする
② データの欠損・誤入力を少なくするために、昨年末に
システムを改定
–  必須項目の設定
–  時間軸の不一致に対するアラート
–  異常値に対するアラート
③ データマネージャーの採用
④ 個人を同定出来る情報がないので、確認
作業が難しい
–  入力不備をチェックし、直接問い合わせ
④ それぞれの施設で登録台帳の作成
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
Modified Utstein Template (2)
Total Cardiopulmonary
Index Events
44
% Disch. Alive
51 %
Pulseless
13
% Disch. Alive
62 %
*No insufficient record
Poor Perfusion,
Pulse Present
31
% Disch. Alive
45 %
今後の方向性
•  登録対象施設を拡大し、全国からの症例
集積を開始予定
•  データ解析のためのシステムを構築
VF/Pulseless VT
2
Bradycardia
29
% Disch. Alive
100 %
% Disch. Alive
45 %
PEA
4
Other Rhythms
1
% Disch. Alive
75 %
% Disch. Alive
0%
Asystole
7
Unkown
1
% Disch. Alive
43 %
% Disch. Alive
100 %
•  成人データや海外のデータと比較検討
•  小児心肺停止症例の予後の改善へ
院内心停止症例：2010年1月∼12月
The 38th Annual Meeting of Japan Society of Intensive Care Medicine 2011.2.24 in Yokohama
謝辞
上谷良行先生
兵庫県立こども病院副院長
福原信一先生
兵庫県立こども病院救急集中治療科
植松悟子先生
国立成育医療研究センター救急診療科
黒澤寛史先生
静岡県立こども病院小児集中治療科
大崎真樹先生
静岡県立こども病院心臓血管外科
阿部世紀先生
長野県立こども病院麻酔集中治療部
森川和彦先生
東京都立小児総合医療センター臨床試験科
金子徹治先生
東京都立小児総合医療センター臨床試験科
4
11/05/29
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
Domestic and international
comparison of pediatric vs. adult
in-hospital cardiac arrest ...
children are not small adults ...
黒澤茶茶1),3) 清水直樹2),3) 横山広行4) 米本直裕5)
6) 野々木宏4)
丸川征四郎
1)
静岡県立こども病院救急総合診療科 2)
東京都立小児総合医療センター救命・集中治療部
3)
4)
5)
国立成育医療センター研究所成育政策科学研究部
国立循環器病センター心臓血管内科緊急治療科京都大学医学部研究科社会健康医学系専攻 6) 医誠会病院
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
背景1
•  国内外での院外心停止に関する疫学調査より、
成人では心原性心停止が、小児では呼吸原性
心停止がその主な原因と考えられている
•  一方、国内での院内心停止の疫学調査に関して
は、明確なデータがない
•  心肺蘇生症例の予後改善のためには、成人・
小児を含めた院内心停止における疫学調査が
必須である
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
背景2
対象と方法
•  厚生労働科学研究（野々木班）において、
２００８年より院内心停止の症例登録（J-RCPR :
The Japanese Registry of CPR for In-hospital
Cardiac Arrest）を開始
•  対象
•  同（丸川班）において、2006年より小児心肺蘇生
レジストリを構築し、2008年より試験的に始動
•  方法
•  これらに登録されたデータより小児と成人に関して
比較検討を行った
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
–  J-RCPR に登録された
成人（18歳以上）症例：251例（2008年）
–  小児心肺蘇生レジストリに登録された
小児（18歳未満）症例：116例（2002-2008年）
–  登録データから、発症時心電図所見、直接原因、
発生場所、予後（自己心拍再開率、生存退院率）
について成人例と小児例で比較検討を行った
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
結果1
結果2
発症時心電図所見
直接原因
（％）
（％）
50
成人 (251)
小児 (116)
40
60
成人 (251)
小児 (116)
50
40
30
30
20
20
10
10
0
VF/VT
心静止
PEA
徐脈
その他
不明
0
不整脈
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
低血圧
呼吸不全代謝・電解質
ACS
その他
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
1
11/05/29
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
結果3
結果4
発生場所
（％）
自己心拍再開率
（％）
60
成人 (251)
小児 (116)
50
90
成人 (251)
小児 (116)
80
70
40
60
30
50
40
20
30
10
20
10
の
他
0
全体
そ
テ
室
救
急
外
来
カ
手
術
室
一
般
病
棟
集
中
治
療
室
0
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
結果5
心静止
PEA
不明
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
結果のまとめ
生存退院率
（％）
VF/VT
•  蘇生対象となる小児症例の約40%は徐脈の症例であり、
心停止前の介入の必要性が再認識された
60
成人 (251)
小児 (116)
50
40
30
•  直接原因としては、成人が不整脈が最多であるのに対し、
小児では低血圧、呼吸不全、不整脈の順であり、多くの
症例では呼吸不全／循環不全を経て心停止へ至ると
推察される
•  発生場所は、小児において集中治療室での発生率が
高かったが、対象施設の特性による影響も大きいと
考えられる
20
10
•  予後に関しては、自己心拍再開率は大きな差はないもの
の、生存退院率は小児の方が高い傾向がみられた
0
全体
VF/VT
心静止
PEA
不明
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
背景
小児心肺蘇生レジストリ
•  小児救急・集中治療領域においては、小児重症患者の集約化
の遅れと治療戦略コンセンサスの不足が指摘され、各種症例
登録基盤の必要性を認識
•  厚生労働科学研究（平成18-20年）「AEDを用いた心疾患の
救命率向上のための体制の構築に関する研究（丸川班）」
「小児心肺停止例へのAED普及にかかわる研究（清水分担班）」
の枠組みの中で、小児心肺蘇生レジストリを構築
•  同（平成20-22年）「成育疾患のデータベース構築・分析と
その情報提供に関する研究（原田班）」とも連携
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
2
11/05/29
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
NRCPRとは？
小児心肺蘇生レジストリ
•  米国を中心に展開する NRCPR (National registry of
Cardiopulmonary Resuscitation) に基づくレジストリの
登録項目を選択
•  登録作業はWeb上で展開
•  全国からの症例集積が必要であり、日本集中治療医学会
新生児・小児集中治療委員会PICU-EBM作業部会と連携
•  国内の院内心停止登録システムであるJ-RCPR
(Japanese registry of Cardiopulmonary Resuscitation)
（厚生労働科学研究野々木班）＝成人領域との連携
•  NRCPR＝国際的な連携
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
本システムとJ-RCPR・NRCPRとの関係性
日本語Web入力
•  2000年から米国を中心にスタート
した院内心肺蘇生事例の国際的
データベース
•  米国、カナダ、ドイツ、ブラジル、日本の430以上の施設が参加、
100,000件以上の蘇生事例集積
CPA； cardiopulmonary arrest
ARC； acute respiratory compromise
MET； medical emergency team （2006年）
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
各施設からの症例登録開始（2008年
）
英訳レポート自動出力
(NRCPR data collection form)
日本小児総合医療施設（29施設）
わが国固有の
薬剤・医療制度
を加味し、
ニーズに応じた
データマイニング
国内レポート 作成と配布
自施設データ 抽出可能
長野県立こども病院
都立小児総合医療センター
（2010年3月）
国立成育医療センター
必要項目を自動出力
NRCPR quarterly report
静岡県立こども病院
J-RCPR
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
小児院内心停止の症例ボリウム
兵庫県立こども病院
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
小児心停止症例の国際比較1
•  Nadkarni, et al, JAMA 2006
–  First documented rhythm and clinical outcome
from in-hospital cardiac arrest among children
and adults.
•  January 2000 – March 2004
•  253 US and Canadian hospitals
•  37,782 cases registered
•  880 cases (2.3 %) were children (<18 y)
•  Only about 200 cases / year
発症時心電図所見
（％）
50
JAPAN (50)
NRCPR (880)
40
30
20
10
0
•  小児心肺蘇生レジストリ：心停止 50例と比較
VF/VT
心静止
PEA
不明
Nadkarni, et al, JAMA. 2006;295(1):50-70
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
3
11/05/29
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
小児心停止症例の国際比較2
心停止の直接原因
（％）
小児心停止症例の国際比較3
JAPAN (50)
NRCPR (880)
心停止発生場所
（％）
70
70
60
30
10
20
0
10
の
他
40
20
JAPAN (50)
NRCPR (880)
0
集中治療室一般病棟
そ
中
毒
30
AC
S
50
肺
水
腫
40
不
整
脈
代
謝
・電
解
質
気
道
閉
塞
60
低
血
圧
呼
吸
不
全
50
手術室
カテ室
Nadkarni, et al, JAMA. 2006;295(1):50-70
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
Nadkarni, et al, JAMA. 2006;295(1):50-70
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
（％）
80
60
70
結語1
•  院内心停止登録システム（J-RCPR）に登録された成人と
小児に関しての比較検討を行った
生存退院率
（％）
JAPAN (50)
NRCPR (880)
50
40
30
40
30
•  小児では、蘇生対象となる症例の約40%は “循環不全を
伴う徐脈”であった
•  直接原因は、成人では、 “不整脈”が最も頻度が高いのに対
し、小児では、 “低血圧”・“呼吸不全”がその主な原因であっ
た
60
50
その他
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
小児心停止症例の国際比較4
自己心拍再開率
救急外来
20
20
10
10
0
0
全体
VF/VT
心静止
PEA
不明
全体
VF/VT
心静止
PEA
不明
•  成人と小児では、心停止の原因およびその経過の違いが
示唆され、小児においては心停止に至る前の徐脈の段階
での介入がその予後を改善する可能性がある
Nadkarni, et al, JAMA. 2006;295(1):50-70
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
結語2
•  小児蘇生においてはその症例ボリウムの少なさから
国際協働が不可欠であり、NRCPRがその基盤となる
•  小児症例の国際比較では、類似した傾向が見られたが、
発生場所の差に関しては、北米におけるPICU設置の充実が
要因として挙げられる
•  今後はMET対応症例や呼吸不全症例の登録も視野に入れて
おり、患者安全向上や病院危機管理において重要な情報源と
なりえ、科学的のみならず社会的にも重要なシステムである
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 74th Annual Scientific Meeting of the Japanese Circulation Society 2010.3.7�in Kyoto
発表内容の一部は、
•  平成19-21年度厚生労働科学研究費補助金
「・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・・（主任研究者野々木宏
：国立循環器病センター
心臓血管内科）の「・・・・・・・・・・・・・・・」
•  平成18-20年度同循環器疾患等生活習慣病対策総合研究事業
「自動体外式除細動器（AED）を用いた心疾患の救命率向上のた
めの体制の構築に関する研究」（主任研究者丸川征四郎：兵庫医
科大学救急・災害医学教授）の「小児AEDの効果的な普及法にか
かわる研究」
•  平成21年度同循環器疾患等生活習慣病対策総合研究事業「
循環器の救命率向上に資する効果的な救急蘇生法の普及啓発に
関する研究」（主任研究者丸川征四郎）の「小児心停止救命率向上
のためのAEDを含めた包括的研究」
の一環として行われた．
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
4
11/05/29
【はじめに】
開院10ヵ月間の
院内蘇生の検討
東京都立小児総合医療センター
救命・集中治療部本間順清水直樹井上信明新津健裕齊藤修池田次郎鶴和美穂中林洋介水城直人新津麻子
藤浪綾子
小児院内蘇生の発生数は、入院患者総数の
0.11∼1.4 % と報告されており、決して稀な事象
ではない（成人 0.1∼0.5 %）。
東京都立小児総合医療センターは 2010年3月
に開院しており、今回、開院後10か月間の院内
蘇生に関して、海外文献との比較も含めて報告
する。
参考文献
Tibballs et al
2006 Resuscitation
Reis et al
2002 Pediatrics
Donoghue et al 2007 Intensive Care Med
【病院概要】
  東京都立小児総合医療センター
【概要】 2010.3.1-12.31
(府中市)
都立清瀬小児病院、八王子小児病院、梅ヶ丘病院、府中病院小児科が
合併して2010.3.1に開院
延べ蘇生延べ蘇生回数
症例数
回数 /延べ入院患者数
性別
(例)
(回)
(%)
17
病床数: 561床
0.35
PICU/HCU 22床
NICU 24床、GCU 48床
14 男7:女7
年齢
(中央値)
イベント
目撃
(例)
日齢2-9歳
(2歳1ヵ月)
17
【基礎疾患】
体外循環蘇生時間
心肺蘇生
(分)
(例)
(中央値)
3
1-77
(3)
重複あり
診療科: 35診療科
医師数：約190人
延べ入院患者数
4735人 (2010.3.1~12.31)
延べPICU/HCU入室患者数
376人 (2010.3.1~12.31)
【蘇生成績】
【発生場所】
海外 vs 国内 vs 当院
80
院内蘇生 60
海外国内当院
52-73 %
56 %
94 %
生存退院率
16-35 %
(or 3か月後生存率) 25 %
64 %
神経学的転帰良好
20 %
57 %
ROSC
(%)
40
(return of spontaneous
circulation)
20
0
15-18 %
国内報告
海外多施設 880例 Nadkarni et al 2006
国内 4施設 156 例 Kurosawa et al 2010
当院
4施設 156 例：厚生労働省科学研究班報告 2010 Kurosawa et al
海外報告
Wu et al Nadkarni et al Tibballs et al Reis et al
Suominen et al 2009 Resuscitation
2006 JAMA
2006 Resuscitation
2002 Pediatrics
2000 Resuscitation
1
11/05/29
【考察】
【 ECPR】
  国内では海外に比し一般病棟で蘇生事象が多い
補助循環を使用したCPR (extracorporeal CPR; ECPR) の
有用性が1992年に報告されて以来、小児領域では de Mosらの
報告 (Crit Care Med 2006) をはじめ、一般的な蘇生に不応な
院内蘇生症例に対するECPRの有用性が報告されている。
当院でも、各科・コメディカルとの連携のもと、ECPRを積極的
に導入している。
• 
一般病棟で小児の重症の患者を見ざるを得ない日本
望まれる対策
•  院内心肺停止を未然に防ぐMET (medical emergency team) の普及
•  後方病棟としてのPICUの整備
  当院でROSC、神経学的予後が良好
• 
• 
全例目撃例であった→PALSﾌﾟﾛﾊﾞｲﾀﾞｰの迅速な対応
通常蘇生に不応時の積極的な補助循環導入
緊急ECMO物品
【ECPR 症例】
症例１
心臓外科との連携
臨床工学技士との連携
【結語】
症例２
症例３
年齢、体重
2y10mo 7.8kg
10mo 2.9kg
2day 4.2kg
疾患分類
原因
先天性心疾患
心不全の悪化
先天性心疾患
開心術後
ペースメーカー不全
先天性心疾患
ショック
場所
一般病棟
PICU
PICU
心電図
徐拍 → PEA
徐拍 → PEA
徐拍 → PEA
導入時間
77分
48分
44分
ECMO期間
11日間
12日間
6日間
離脱
不可(心ポンプ不全)
可
可
転帰
死亡
死亡 (sepsis)
生存退院
神経学転帰
ECMO中
瞳孔散大なし
蘇生前と同様の反応
（追視、手を動かすetc）
良好
  日本では重症患者を一般病棟で管理せざるをえない状況が続いており、院内心肺停止を未
然に防ぐMET の普及と、PICUの整備が必要
と思われる。
  小児院内蘇生例の中には先天奇形、先天代謝
疾患など、小児特有の疾患を持つ場合が多く、
蘇生後の神経予後を改善するには、集中治療、
小児疾患共に精通した医師が必要と思われる。【小児院内心停止報告】
著者 (年)
国
Wu ET et al
(2009)
台湾
Nadkarni et al (2006)
Tibballs and Kinny
(2006)
施設数症例数
アメリカ、
カナダ
オーストラリア
生存
性差年齢 ROSC
神経学的転帰
退院率
（%） median (%)
良好 (%)
(%)
1
316
M
56
― 61
21
16
253
880
M
54
1.8
52
27
18
1
111
―
0.5
73
35
―
―
64
16
15
―
63
18
―
Reis et al (2002)
ブラジル
1
129
M
58
Suominen et al (2000)
フィンランド
1
129
M
59
2000年以降、n>100例以上の報告
2
11/05/29
【はじめに】
小児蘇生の疫学 院内心停止
小児では、入院患者数に対する院内心停止数
は入院患者数の0.11∼1.4 %と報告されており、
決して稀な事例ではない。（成人0.1∼0.5 %）
本発表では、小児の院内心停止の疫学につい
て、成人との比較、院外心停止との比較、国内
外の比較の3つの視点から検討する。
東京都立小児総合医療センター
救命・集中治療部本間順
参考文献
J tibballs et al (Resuscitation 2006)
A reis et al (Pediatrics 2002)
Donoghue et al (Intensive Care Med 2007)
【院内心停止小児 vs 成人
①】
【小児院内心停止報告】
著者 (年)
国
Wu ET et al
(2009)
台湾
Nadkarni et al (2006)
アメリカ、
カナダ
生存
性差年齢 ROSC
神経学的転帰
退院率
（%） median (%)
良好 (%)
(%)
施設数症例数
1
316
M
56
― 61
21
16
253
880
M
54
1.8
52
27
18
111
―
0.5
73
35
―
Tibballs and Kinny
オーストラリア 1
(2006)
Reis et al (2002)
ブラジル
1
129
M
58
―
64
16
15
Suominen et al (2000)
フィンランド
1
129
M
59
―
63
18
―
353
M
57
0.9
全例
49
44
Moler et al
(2009) ※
アメリカ
15
※ CPR20min 以内にROSCに至った症例
小児
成人
性差
M 54-58 %
M 57 %
ROSC
52-73 %
47 %
生存退院率
16-35 %
18 %
神経学的転帰良好
15-18 %
11 %
2006 V.Nadkarni et al （NRCPR）より
成人院内心停止36902例
2000年以降、n>100例以上の報告
【院内心停止小児 vs 成人
②】
直接原因
％
100
小児 (%)
【小児心停止院内 vs 院外 ①】
院内
院外
性差
M 54-59 %
M 58-62 %
年齢(median)
0.8-1.5
1.5-2.9
ROSC
52-73 %
10-27 %
生存退院率
16-35 %
2-12 %
神経学的転帰良好
15-18 %
1-4 %
成人 (%)
80
60
院内蘇生 65
61
57
44
40
49
41
20
12 11
11
2
0
低血圧
呼吸不全
不整脈
代謝
電解質異常
心筋梗塞
2006 M.Nadkarni et al (NRCPR)より
院内心停止成人36902例ｖｓ小児880例
院外心停止参考文献
Sirbough PE et al (Ann Emrg med 1999)
Kelly D et al (Pediatrics 2004)
Donoghue et al (Ann Emrg med 2005)
Dianne L et al (Circulation 2009 )
1
11/05/29
【小児心停止院内 vs 院外 ②】
直接原因
％
100
％
院内 (%)
80
100
院内 (%)
院外 (%)
72
80
院外 (%)
69
60
40
【小児心停止院内 vs 院外 ③】
死亡原因 (ROSC後)
60 20
42
37
36
40 30
4
2 4
0
心原性
呼吸原性
神経原性
6
22
20
20 15
心原性
50
8
11
0
脳神経
外傷性
心血管
その他
（不整脈、ショック、（CHD)
心筋炎など）
2009 F.Moler et al
院内353例 vs 院外138例より
F.Moler et al 2009
院内353例 vs 院外138例より
【院内心停止国内】
・国外の小児院内心停止の疫学に関しての文献
的報告は多いが、国内はない。
・厚生労働科学研究、小児心肺停止例へのAED
普及に関わる研究（清水分担班）の枠組みで小児
心肺蘇生レジストリを構築中
・国際的に展開する NRCPR(National registry of Cardiopulmonary
Resuscitation)に基づく登録項目を選択し各領域と連携
小児心肺蘇生レジストリ 【】
国際、国内連携
日本語Web入力
英訳レポート自動出力
(NRCPR data collection form)
わが国固有の
薬剤・医療制度
を加味し、
ニーズに応じた
データマイニング
国際連携
必要項目を自動出力
国内レポート 作成と配布
NRCPR quarterly report
J-RCPR
成人領域との連携
【院内心停止国内 vs 国外 ①】
国内：研究班報告4施設（PICUあり） 156 例
(2010 S Kurosawa et al)
国内
【院内心停止国内 vs 国外 ②】
直接原因
％
100
国内 (%)
国外
国外 (%)
80
性差
M 42 %
M 54-58 %
60
61
57
49
46
年齢(median)
1.0
0.8-1.5
40
ROSC(>20min)
56 %
52-73 %
20
生存退院率
25%
16-35 %
神経学的転帰良好
19.5 %
15-18 %
0
36
30
20
12
65
8
4
22
低血圧呼吸不全不整脈代謝気道閉塞肺水腫 ACS
電解質異常
02
41
中毒
その他
国内 50例 Kurosawa et al 2010
国外 880例 V.Nadkarni et al 2006
2
11/05/29
【院内心停止国内 vs 国外 ③】
心停止発生場所
％
100
80
国外 (%)
65
60
20
・院内と院外では、院内の方が、生存退院率、神経
転帰が共に良好であった。
42
・国際比較では、類似した傾向が見られていたが、
国内データは4施設と現状を反映するには少ない。
38
14
6 3
0
・小児院内心停止では、想定よりも不整脈による心
停止が多かった。
国内 (%)
40
【結語】
10 13
0 2
4 2
集中治療室一般病棟手術室検査室救急外来その他
国内 50例 Kurosawa et al 2010
国外 880例 V.Nadkarni et al 2006
・小児心肺蘇生レジストリに多施設の参加を呼びか
け日本の現状を把握する必要があると思われた。
3
10.9.11
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
Introduction
シンポジウム5 「小児蘇生2010からの展望」
小児心肺蘇生法の
品質改善に関わる研究
静岡県立こども病院救急総合診療科東京都立小児総合医療センター救命・集中治療部国立成育医療研究センター研究所成育政策科学研究部 •  小児心肺停止症例の生存率は決して高くない
•  生存率改善のために出来ることは？
•  CPRは蘇生治療の根幹
•  効果的な胸骨圧迫を絶え間なく行うことが重要
黒澤茶茶 Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
CPR : Chest Compression
CPR : adequate depth of CC
Schema
Chest CT image
•  Adequate Depth?
Sternum –  1/2 AP chest diameter or 1/3 or・・・
Heart ① •  Method
② –  for Children : single or doubble hands
–  for Infant : “circumferential squeeze”
in the use of two-thumb chest compression
② ① thorax Vertebral body ① AP chest diameter (external AP) ② Between sternum and vertebral body (internal AP) Kurosawa S, Jpn. Soc. Intensive care med. 2009
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
CPR : adequate depth of CC
Residual AP with compression of 1/3 or 1/2 external AP
1/2 external AP 1/3 external AP standard deviation
40 Residual AP (mm) Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
CPR : adequate depth of CC
Percentage of Patients with Greater than 10mm
during 1/2 AP CC, 1/3 AP CC, 51mm CC, and 38mm CC
US 1-‐3 yo Japan 1-‐3 yo US 4-‐7 yo Japan 4-‐7 yo ① 30 20 10 Average
1/2 : 1.5±3.4
1/3 : 22.6±4.4
0 -‐10 1 2 3 4 Age (years) 5 6 7 Kurosawa S, Jpn. Soc. Intensive care med. 2009
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Braga M, SCCM's 39th Critical Care Congress. 2010
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
1
10.9.11
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
CPR : CC methods for infants
CPR : CC methods for infants
Shimizu N, Circulation (abstract). 2008
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
CPR : CC quality monitoring
乳児の胸骨圧迫の深さの検証
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
New target for CC depth
P=0.016* < 1year 1-‐7 year 8-‐15 year Chest compression data PC skill reporOng system 0 32 34 0 5 10 Time（second）
15 Depth of compression (mm) 35
1/3 平均値＝40.54 mm 標準偏差＝4.112 度数＝241
平均値＝53.97 mm 標準偏差＝6.709 度数＝124
30
25
20 20
30.0±4.2 before 32.4±1.5 aSer Kurosawa S, Jpn. Soc. Intensive care med. 2010
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
New target for CC depth
（mm）
平均値＝32.08 mm 標準偏差＝3.177 度数＝82
AP chest diameter 1/2 1/3 1/3 AP depth 1/3 AP depth 1/3 AP depth Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
150 New target for CC depth
100 200 50 150 30mm
0 100 30mm
0 0-‐2 3-‐5 6-‐8 9-‐11 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
months 50 50mm
40mm
50mm
40mm
years •  実数規定で年齢差がない方が、自動胸骨圧迫装置
やQ-CPR類似機器の小児応用に際して利便性が高い
0-‐2 3-‐5 6-‐8 9-‐11 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
months years Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
2
10.9.11
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
CPR Quality in Pediatrics
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
CC quality monitoring devices
•  8歳から18歳までの小児入院患者の保護者を対象とした研究：VAM
（voice advisory manikin）を用いて、的確なフィードバックを行いながら
トレーニングを行ったほうが、インストラクターと1対1で行うよりもBLS手技
の習得はより正確になる
–  Sutton RM, et al. The voice advisory manikin (VAM): an innovative approach to pediatric lay
provider basic life support skill education. Resuscitation. 2007;75: 161-8.
•  8歳以上の小児の院内心肺蘇生事例に対し、Q-CPRを使ってCPRの質を
解析した研究：CPRのトレーニングを受けている者が蘇生を行った場合で
も、CPRの質は決して高くない
–  Sutton RM, et al. Quantitative analysis of CPR quality during in-hospital resuscitation of older
children and adolescents. Pediatrics. 2009;124(2):494-9.
POCKET CPR：
Bio-Detek, Incorporated製
ハートスタートMRx（Q-CPR）：PHILIPS製
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
New method of CC depth measure
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
Automatic CC systems •  Applying flow sensor rather than accelerometer LUCAS：Jolife社製
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
Application to children AutoPulse：ZOLL社製
KOMSTAT2300：コーケン社製
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
New technology of auto. CC device
•  安全性、正確性の検証 •  固定の問題 Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
3
10.9.11
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
The Japanese society of Reanimatology 29th Annual Meeting, 2010.9.11 in Utsunomiya
Future Developments
発表内容の一部は、
•  小児では、新しく挑戦的な蘇生手段を拒否されがち –  少ない小児症例、年齢における体格の違い •  小児心肺蘇生における品質改善のためには年齢の壁を越えた挑戦が必要 •  品質改善のためのモニタリング機器、自動胸骨圧迫
装置の小児への応用ならびに新たな機器の開発 •  今日は不可能でも、明日は可能に！ Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
•  平成18-20年度厚生労働科学研究費補助金循環器疾患等生活
習慣病対策総合研究事業「自動体外式除細動器（AED）を用いた
心疾患の救命率向上のための体制の構築に関する研究」
（主任研究者丸川征四郎：兵庫医科大学救急・災害医学教授）の
「小児AEDの効果的な普及法にかかわる研究」
•  平成21-22年度同循環器疾患等生活習慣病対策総合研究事業「循環器の救命率向上に資する効果的な救急蘇生法の普及啓発
に関する研究」（主任研究者丸川征四郎：医誠会病院）の
「小児心停止救命率向上のためのAEDを含めた包括的研究」
の一環として行われた．
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
4
11.6.1
Table 1. Demographics and anthropometric characteristics
Figure 1. Measurement of the external and internal anteriorposterior diameter of the chest
1. Schema
Age category Age* Male(n) Female(n) total(n)
Infants
2. Chest CT image
Sternum
Heart
Younger children
B
A
B
A
thorax
Vertebral body
Older children
A.  External AP: External anterior-posterior chest diameter
B.  Internal AP: Internal chest diameter between sternum and vertebral body
Weight(kg) external AP(mm) internal AP(mm)
Height(cm)
0-2
13
15
28
51.0 ± 3.5 (+0.12SD)
3.6 ± 1.0 (+0.22SD)
87.3 ±6.9
46.0 ±3.7
3-5
8
10
18
63.0 ± 2.5 (-0.43SD)
7.0 ± 0.9 (+0.16SD)
99.6 ±6.7
48.4 ±6.0
6-8
11
7
18
66.5 ± 2.7 (-0.73SD)
7.7 ± 0.9 (-0.41SD)
100.8 ±7.4
51.3 ±6.6
9-11
8
10
18
71.3 ± 3.3 (-0.16SD)
8.5 ± 0.9 (-0.18SD)
102.2 ±8.1
51.5 ±6.5
1
31
25
56
78.9 ± 3.6 (-0.37SD)
10.2 ± 1.2 (-0.05SD)
109.4 ±5.3
56.1 ±4.5
2
29
17
46
87.4 ± 3.9 (-0.29SD)
12.2 ± 1.4 (+0.2SD)
115.3 ±7.1
59.8 ±5.9
3
15
15
30
93.9 ± 3.4 (-0.53SD)
13.5 ± 1.3 (-0.44SD)
118.1 ±7.1
60.6 ±5.0
4
18
12
30
102.8 ± 4.8 (-0.1SD)
16.2 ± 2.2 (-0.13SD)
125.1 ±8.9
63.9 ±6.9
5
20
9
29
107.9 ± 4.3 (-0.17SD)
18.1± 2.2 (-0.1SD)
131.4 ±6.9
67.4 ±5.1
6
11
11
22
114.1 ± 5.4 (-0.17SD)
19.8 ± 2.8 (-0.32SD)
131.6 ±7.5
67.5 ±7.8
7
15
13
28
121.2 ± 6.6 (-0.14SD)
23.1 ± 3.8 (-0.25SD)
138.3 ±9.9
73.0 ±8.4
8
8
3
11
123.9 ± 5.6 (-0.68SD)
22.8 ± 3.6 (-0.84SD)
142.4 ±12.2
73.0 ±13.1
9
5
14
19
131.0± 7.1 (-0.35SD)
27.1 ± 6.5 (-0.54SD)
144.1 ±12.0
73.2 ±9.4
10
8
10
18
137.1 ± 6.4 (-0.29SD)
30.4 ± 5.5 (-0.55SD)
150.0 ±12.5
78.3 ±10.2
11
14
14
28
141.4 ± 6.5 (-0.64SD)
35.8 ± 6.0 (-0.43SD)
163.5 ±16.4
82.2 ±9.3
12
7
5
12
152.1 ± 9.0 (+0.19SD)
43.3 ± 8.5 (-0.07SD)
170.3 ±17.4
85.1 ±11.0
13
5
5
10
156.7 ± 5.7 (-0.11SD)
47.7 ± 6.9 (-0.2SD)
174.2 ±11.8
85.7 ±12.0
14
6
9
15
161.6 ± 5.7 (+0.31SD)
53.2 ± 7.4 (+0.11SD)
184.9 ±15.6
91.5 ±14.6
15
6
5
11
164.0 ± 8.7 (+0.14SD)
51.0 ± 6.0 (-0.44SD)
175.9 ±15.9
82.9 ±13.6
n=238
(53%)
n=209
(47%)
n=447
* infants ; months, children ; years
Table 2. Residual AP with compression of 1/2 or 1/3 external AP
Fig.2 Residual AP at 1/2 and 1/3 chest compression depth
Overcompression
Age
(Residual AP<0mm)
(years)
1/3 CC depth
1/2 CC depth
1/3 CC depth
25/82
0/82
82/82
5/82
1
17/56
0/56
56/56
1/56
2
13/46
0/46
45/46
0/46
3
10/30
0/30
30/30
0/30
4
9/30
0/30
30/30
0/30
5
7/29
0/29
28/29
0/29
6
7/22
0/22
22/22
1/22
7
6/28
0/28
24/28
0/28
8
6/11
0/11
8/11
0/11
8/19
0/19
19/19
0/19
2/18
0/18
17/18
1/18
11
13/28
0/28
27/28
0/28
12
6/12
0/12
12/12
0/12
13
6/10
0/10
9/10
0/10
14
5/15
0/15
14/15
1/15
15
9/11
0/11
11/11
0/11
total
149/447 (33.3%)
0/447 (0%)
434/447 (97.1%)
9/447 (2.0%)
10
0
-10
0-2 3-5 6-8 9-11 1
(months)
2
3
4
5
6
7
8 9
(years)
10 11
12
13 14
15
Age
Fig.3 Averages of depth equivalent to 1/3 and 1/2 external AP
AP chest diameter
1/3 CC depth
20
10
(mm)
1/2 CC depth
30
Residual AP
9
(mm)
(Residual AP<10mm)
1/2 CC depth
0
1/2
Table 3. Average depths equivalent to 1/3 and1/2 AP diameter
1/3
200
150
100
Infants
Children
Older children
1/2 AP diameter depth
(average, mm)
48.1±4.8
60.8±6.2
81.0±10.1
1/3 AP diameter depth
(average, mm)
32.1±3.2
40.5±4.1
54.0±6.7
50
0
0-2 3-5 6-8 9-11 1
(months)
2
3
4
5
6
7 8 9
(years)
10
11 12 13 14 15
Age
1
11.6.1
Fig.4 Frequency distribution of 1/3 external AP depth
Infants
Average = 32.08 mm
Standard deviation = 3.177
Frequency = 82
Children
Older children
Average = 40.54 mm
Standard deviation = 4.112
Frequency = 241
Average = 53.97 mm
Standard deviation = 6.709
Frequency = 124
20.0 25.0 30.0 35.0 40.0 30.0 35.0 40.0 45.0 50.0 55.0 1/3 AP depth
1/3 AP depth
Table 4. Residual AP and Proportions of patients with CC
targewithout overcompression (residual AP <10mm) for each 30,
40, and 50mm target for each infants, children, and older children
Infants
(30mm)
Children
(40mm)
Older children
(50mm)
Residual AP (mm)
18.9 ± 6.0
22.7 ± 8.1
31.2 ± 12.4
Proportion of patients with
residual AP ≥10mm
77/82
(94.0%)
233/241
(96.7%)
119/124
(96.0%)
Proposed target depth (mm)
40.0 50.0 60.0 70.0 1/3 AP depth
2
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
C2010 Peds Worksheets
第3回日本蘇生科学シンポジウムトピック「公表されているWork Sheetについて」
小児領域における 日本からの情報発信
•  55 items
•  75 Worksheets
•  4 WS authors nominated from Japan
–  Naoki Shimizu, Research Inst., NCCHD
–  Masahiko Nitta, Osaka Medical College
–  Kunio Ohta, Kanazawa University
–  Sasa Kurosawa, Research Inst., NCCHD
黒澤茶茶
静岡県立こども病院救急総合診療科
東京都立小児総合医療センター救命・集中治療部
国立成育医療研究センター研究所成育政策科学研究部
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
C2010 Peds topics
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
Access : IV vs IO, ET vs IV
Airway: BMV, Tube, ETCO2, Oxygen, cricoid pressure
Bypass : ECPR
CPR : CV ratio, Compression only CPR, Compression depth, methods
Defibrillation : AED, Doses, Adequate energy, Pad/Padlle
Monitoring : methods, invasive monitoring
Pharmacology : Epinephrine, Vasopressin, Atropine・・・
Prognosis : predict ROSC
Recognition : pulse check accuracy
Shock : fluid, intubation, inotropes, etomidate, corticosteroids・・・
Special Circumstances : family, resuscitation for Fontan circulation
Systems of Care : MET/RRT
Temperature : induced hypothermia
Trauma : fluid resuscitation, traumatic arrest
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
CPR : Chest Compression
•  Adequate Depth?
C2010 Peds topics
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
Access : IV vs IO, ET vs IV
Airway: BMV, Tube, ETCO2, Oxygen, cricoid pressure
Bypass : ECPR
CPR : CV ratio, Compression only CPR, Compression depth, methods
Defibrillation : AED, Doses, Adequate energy, Pad/Padlle
Monitoring : methods, invasive monitoring
Pharmacology : Epinephrine, Vasopressin, Atropine・・・
Prognosis : predict ROSC
Recognition : pulse check accuracy
Shock : fluid, intubation, inotropes, etomidate, corticosteroids・・・
Special Circumstances : family, resuscitation for Fontan circulation
Systems of Care : MET/RRT
Temperature : induced hypothermia
Trauma : fluid resuscitation, traumatic arrest
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
CPR : adequate depth of CC
Chest CT image
Schema
Sternum
–  1/2 AP chest diameter or 1/3 or・・・
•  Method
②
Heart
①
②
①
thorax
–  for Children : single or double hands
–  for Infant : “circumferential squeeze”
in the use of two-thumb chest compression
Vertebral body
① AP chest diameter (external AP)
② Between sternum and vertebral body (internal AP)
Kurosawa S, et al, Jpn. Soc. Intensive care med. 2009
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
1
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
CPR : adequate depth of CC
CPR : adequate depth of CC
Residual AP with compression of 1/3 or 1/2 external AP
Percentage of Patients with Greater than 10mm
during 1/2 AP CC, 1/3 AP CC, 51mm CC, and 38mm CC
1/2 external AP 1/3 external AP standard deviation
Residual AP (mm)
40
US 1-3 yo
Japan 1-3 yo
US 4-7 yo
Japan 4-7 yo
①
30
20
10
Average
1/2 : 1.5±3.4
1/3 : 22.6±4.4
0
-10
1
2
3
4
Age (years)
5
6
7
Kurosawa S, et al, Jpn. Soc. Intensive care med. 2009
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
CPR : CC quality monitoring
乳児の胸骨圧迫の深さの検証
CPR Quality in Pediatrics
P=0.016*
PC skill reporting system
0
32
34
0
5
10
15
Depth of compression (mm)
35
Chest compression data
Braga M, et al, SCCM's 39th Critical Care Congress. 2010
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
1/3
•  8歳から18歳までの小児入院患者の保護者を対象とした研究：VAM
（voice advisory manikin）を用いて、的確なフィードバックを行い
ながら
トレーニングを行ったほうが、インストラクターと1対1で行うより
もBLS手技の習得はより正確になる
–  Sutton RM, et al. The voice advisory manikin (VAM): an innovative approach to pediatric lay
provider basic life support skill education. Resuscitation. 2007;75: 161-8.
30
•  8歳以上の小児の院内心肺蘇生事例に対し、Q-CPRを使ってCPRの質
を解析した研究：CPRのトレーニングを受けている者が蘇生を行っ
た場合でも、CPRの質は決して高くない
25
–  Sutton RM, et al. Quantitative analysis of CPR quality during in-hospital resuscitation of older
children and adolescents. Pediatrics. 2009;124(2):494-9.
20
Time（second）
20
30.0±4.2
before
32.4±1.5
after
Kurosawa S, et al, Jpn. Soc. Intensive care med. 2010
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
New target for CC depth
< 1year
平均値＝32.08 mm
標準偏差＝3.177
度数＝82
1-7 year
平均値＝40.54 mm
標準偏差＝4.112
度数＝241
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
8-15 year
平均値＝53.97 mm
標準偏差＝6.709
度数＝124
New target for CC depth
AP chest diameter
（mm）
200
1/2
1/3
150
100
50
1/3 AP depth
1/3 AP depth
1/3 AP depth
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
0
0-2 3-5 6-8 9-11 1
months
50mm
40mm
30mm
2
3
4
5
6
7
8
years
9
10 11 12 13 14 15
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
2
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
150
New target for CC depth
100
50
0
0-2 3-5 6-8 9-11 1
50mm
40mm
30mm
2
3
4
5
6
months
7
8
9
10 11 12 13 14 15
years
•  成人の胸骨圧迫（4-5 cm）が浅すぎる？
•  実数規定で年齢差がない方が、QCPR類似機器の
小児応用に際して利便性が高い
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
C2010 Peds topics
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
Access : IV vs IO, ET vs IV
Airway: BMV, Tube, ETCO2, Oxygen, cricoid pressure
Bypass : ECPR
CPR : CV ratio, Compression only CPR, Compression depth, methods
Defibrillation : AED, Doses, Adequate energy, Pad/Padlle
Monitoring : methods, invasive monitoring
Pharmacology : Epinephrine, Vasopressin, Atropine・・・
Prognosis : predict ROSC
Recognition : pulse check accuracy
Shock : fluid, intubation, inotropes, etomidate, corticosteroids・・・
Special Circumstances : family, resuscitation for Fontan circulation
Systems of Care : MET/RRT
Temperature : induced hypothermia
Trauma : fluid resuscitation, traumatic arrest
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
MET/RRT for Pediatrics
Authors
Hospital
Pre-MET/RRT
period
Post-MET/RRT
period
Rapid Response
system
MET/RRT for Pediatrics
Main findings (per 1000 Admissions/Discharges)
Authors
Code
Cardiac / Resp.Arrest
Brilli et al
(2007)
1.54 to 0.62/1000 A
(p=0.02)
0.56 to 0.24/1000 A (NS)
Hunt et al
(2008)
2.1 to 1.1/1000 D
(NS)
Resp.Arrest
1.46 to 0.40/1000 D
(p=0.04)
Brilli et al
(2007)
Children’s Hospital 15 months
8 months
Two-tiered
Doctors & Nurses
(MET attend <15 min.)
Hunt et al
(2008)
Tertiary academic
hospital
12 months
One-tiered
Doctors & nurses
Mistry et al
(2006)
Children’s Hospital 6 months
5 months
One-tiered
Mistry et al
(2006)
Sharek et al
(2007)
264 bed
pediatric hospital
56 months
19 months
One-tiered
Doctors & nurses
Sharek et al
(2007)
Tibballs et al
(2009)
215 bed
pediatric hospital
41 months
48 months
One-tiered
Doctors & nurses
Tibballs et al
(2009)
Zenker et al
(2007)
Children’s Hospital 23 months
12 months
Two-tiered
Nurses
12 months
Zenker et al
(2007)
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
NRCPR and MET/RRT
decreased 58% (p=0.001)
2.45 to 0.69/1000 A
(P=0.008)
Mortality
0.43 to 0.24 (NS)
decreased 48% (p=0.005)
10.1 to 8.3/1000 D
(p=0.007)
0.19 to 0.17/1000 A (NS)
preventable CA :
0.16 to 0.07/1000 A
(p=0.04)
8.0 to 5.1/1000 D
(36%) (NS)
4.38 to 2.87/1000 A
(p<0.0001)
from 4.3 to 4.5/1000 D
(NS)
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
NRCPRとは？
•  2000年から米国を中心にスタート
した院内心肺蘇生事例の国際的
データベース
•  米国、カナダ、ドイツ、ブラジル、日本の430以上の施設が参加、
100,000件以上の蘇生事例集積
CPA； cardiopulmonary arrest
ARC； acute respiratory compromise
MET； medical emergency team （2006年）
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
3
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
小児死亡の症例ボリウム
小児心肺蘇生レジストリ
平成19年度厚生労働科学研究（池田班）
幼児死亡の分析と提言に関する研究
（件・人）
300
死亡施設数
死亡症例数
250
200
•  米国を中心に展開する NRCPR (National registry of
Cardiopulmonary Resuscitation) に基づくレジストリの
登録項目を選択
年間死亡症例 5人/年以上の施設数は
施設数総計の10％未満
•  登録作業はWeb上で展開
年間死亡症例10人/年以上の施設数は
施設数総計の 2 ％未満
•  全国からの症例集積が必要であり、日本集中治療医学会
新生児・小児集中治療委員会PICU-EBM作業部会と連携
年間死亡症例10人/年以上施設で死亡
する小児は、小児死亡総数の10％程度
•  国内の院内心停止登録システムであるJ-RCPR
(Japanese registry of Cardiopulmonary Resuscitation)
（厚生労働科学研究野々木班）＝成人領域との連携
150
100
50
平成17年死亡小票
0
1-4歳のデータから
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
•  NRCPR＝国際的な連携
単位施設あたりの年間死亡症例数 (人/年）
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
各施設からの症例登録開始（2008年
）
Modified Utstein Template
日本小児総合医療施設（29施設）
長野県立こども病院
都立小児総合医療センター
（2010年3月）
国立成育医療研究センター
静岡県立こども病院
兵庫県立こども病院
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
Total Cardiopulmonary Index
Events
132
(168*)
% Disch. Alive
35 %
Pulseless
63
% Disch. Alive
29 %
VF/Pulseless VT
8
% Disch. Alive
25 %
PEA
17
% Disch. Alive
29 %
Asystole
21
% Disch. Alive
19 %
Unkown
17
% Disch. Alive
41 %
*Insufficient record
(n=36, 21%) was
excluded
Poor Perfusion,
Pulse Present
69
% Disch. Alive
41 %
Bradycardia
48
% Disch. Alive
44 %
Other Rhythms
11
% Disch. Alive
55 %
Unkown
10
% Disch. Alive
10 %
院内心肺停止症例：2002年3月∼2009年12月
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
Immediate causes （％）
今後の展望
•  CoSTR 2010 Treatment Recommendation :
60
「In infants and children in ・・・, there is
insufficient evidence to support or refute・・・」
50
40
30
•  小児蘇生領域におけるエビデンス創成の必要性
51
20
30.2
22.4
10
0
•  小児重症患者の集約化と多施設共同臨床研究
20.3
12.6
10.6
循
呼
環
不
吸
不
全
不
全
整
脈
代
謝
電
エア
ウ
解
質
異
常
ェイ
4.2
2.8
2.1
鎮
AC
S
痙
静
の
不
攣
重
明
積
•  小児蘇生領域での国際恊働
異
常
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
4
The 3rd Japanese Resuscitation Science Symposium, 2010.9.12 in Omiya
発表内容の一部は、
•  平成18-20年度厚生労働科学研究費補助金循環器疾患等
生活習慣病対策総合研究事業「自動体外式除細動器(AED)
を用いた心疾患の救命率向上のための体制の構築に関する
研究」（主任研究者丸川征四郎：兵庫医科大学救急・災害医
学教授）の「小児AEDの効果的な普及法にかかわる研究」
•  平成21-22年度同循環器疾患等生活習慣病対策総合研究
事業「循環器の救命率向上に資する効果的な救急蘇生法の
普及啓発に関する研究」（主任研究者丸川征四郎：医誠会
病院）の「小児心停止救命率向上のためのAEDを含めた包括
的研究」
の一環として行われた．
Department of Emergency and general Pediatrics, Shizuoka Children’s Hospital, Shizuoka, Japan
5
→
1
ち外れ値は 457 点、解析対象として残ったのは 17104 点である。
また、下図は PP データの原系列(上)と外れ値処理済系列(下)である。PP データは 17561 点、う
値は 178 点、解析対象として残ったのは 16825 点である。
下図は RR データの原系列(上)と外れ値処理済系列(下)である。RR データは 17003 点、うち外れ
れ値処理を行った。
それぞれの時系列を MemCalc/Win に読込み、
「実行」
「修正系列の計算」にて、つぎの条件で外
【外れ値処理】
記録された値を用いる。
RR データは 20103300646RR.csv ファイル、PP データは 20103300646SBPPP.csv ファイルに
【比較検討する時系列】
当する情報が得られるか、検討する。
パルスオキシオメータ脈波から取り出した PP データにより、心電波形 RR データの HRV に相
【目的】
心電波形の RR とパルスオキシオメータ脈波の PP データの対応について
開始 2011 年 5 月 10 日
最終更新 2011 年 5 月 10 日
他方、次図は別の 5 分長区間である。
いるように見える。
2
赤い折れ線(＝PP データ)は緑の点列(＝RR データ)と一致しないが、緑の点列の推移をなぞって
次図は上図の一部を拡大したもので、5 分長のデータである。
図で赤い折れ線は PP データ、緑の点列は RR データである(いずれも外れ値を取り除いたデータ)。
て読み込んで比較したのが次図である。
MemCalc/Win にてそれぞれの外れ値処理済系列をファイルに記録し、ビューア(MvView.exe)に
【外れ値処理済系列の比較】
上図では、区間中央付近で PP データは欠落しており、従って RR データの推移を PP データは
セグメントにおいて RR データと PP データがともに欠落値のない場合のみ、解析対象と
2.
3
列(欠落値のない、等間隔データ)を得る。つぎに上記条件にて MEM-PSD を求める。次図が RR の
MemCalc/Win に RR の最初のセグメントデータを読込み、外れ値処理等、一切行わずに修正系
秒)相当、すなわち 10%とする(25÷300×100∼8.3%)。
1. なめらかなスペクトルを得るため、周波数分解能との兼ね合いでラグ値は 1/0.04 秒(＝25
一例として最初のセグメントを解析する。このとき、解析条件(ラグ値)をつぎのように設定する。
【解析】
して切り出した 21 個のセグメントデータは添付のとおりである。
する、(g)「実行」キーを押す、以上である。二つ目以降のセグメントでは(c)以降を繰り返す。こう
ダ・時系列 A と B のファイル名を設定する、(f)記録する時間帯を「描画範囲」ボタンを押して指定
セグメントを探す、(d)フォーム下端のフロッピーディスクの絵の記録キーを押す、(e)記録先フォル
のキリのよい値を入力し Enter キーを押す、(c)フォーム下端の左右向きの三角キーで欠落値のない
作は、(a)「描画巾」に「300」と入力し Enter キーを押す、(b)「左端時刻」にデータ左端時刻以降
時刻を 10 進数表示＝時分秒表示でない、として立ちあげた MvView.exe から切り出す)。具体的操
上記 1.と 2.を満足するセグメントを(MvView.exe にて)目視確認しながら複数抽出する(このとき、
する。
セグメント長を 300 秒とする。
1.
るため、つぎの条件を設定する。
LF 周波数帯(0.04∼0.15Hz)と HF 周波数帯(0.15∼0.40Hz)で「安定した」パワースペクトルを得
【解析対象データの抽出】
が妥当と思われる。
注目区間で一方のデータが欠落する場合、その区間は解析対象としないこと
冒頭の目的に照らし、注
増大し、他方 PP データについては LF 成分が過少に評価されるなどの結果が予想される。従って
この区間のデータを解析した場合、RR データについては区間中央付近の挙動により LF 成分が
「追随」していない。この間の PP データの挙動と外れ値処理のアルゴリズムの結果と思われる。
それとよく対応する。
4
LFとHFはそれぞれ 342msec2と 215msec2、LF/HFは 1.59 である。この 3 つの値はRRデータの
また、つぎは同じ時間帯の PP データの結果である。
LFのパワーは 371msec2、HFのそれは 244msec2、LF/HFは 1.52 である。
最初のセグメントの結果である。
5
する LF、HF、LF/HF を得られる可能性がある。
以上
すなわち、パルスオキシオメータ脈波からの PP データにより RR データの解析結果によく対応
測定波形を改善することにより、対応するセグメントとなる可能性もある。
2. 対応するセグメントを構成できなかった区間の一部については、パルスオキシオメータの
データの解析結果によく対応する PP データのそれを得る可能性がある。
1. 上述の手続きにて RR データと PP データのセグメントを作成・解析することにより、RR
以上より、
࿑ߢ޽ࠆ‫ޕ‬
ࠇ୯ߪ 525 ୘ߢᄖࠇ୯₸ߪ 6.69㧑ߢ޽ࠆ‫ޕ‬ᄖࠇ୯ಣℂߒߡୃᱜᤨ♽೉ࠍ૞ᚑߒߚ߽ߩ߇ਅ
ౝኈߢࡈࠔࠗ࡞ฬ‫࠲࡯࠺ޔ‬ᢙߥߤ߇⴫␜ߐࠇᄖࠇ୯ಣℂࠍⴕߞߚ⚿ᨐ‫ޔ‬ᱜᏱ୯ 7326 ୘ߢᄖ
ߥ޿ㇱಽ߇ᄖࠇ୯ಣℂࠍⴕ޿ਅߦ✛ߩㇱಽ߇⹺⼂ߢ߈ߥ޿ㇱಽߣಽ߆ࠆ‫ޕ‬ฝ਄ߪ࠺࡯࠲ߩ
਄࿑ߪේ♽೉࠺࡯࠲ᮮゲ߇ᤨ㑆‫❑ޔ‬ゲߪ P- P 㑆㓒ߢࡁࠗ࠭ߥߤߦࠃࠅᵄᒻ⹺⼂߇ߢ߈ߡ޿
⸃ᨆ଀ ࡈࠔࠗ࡞ͳ201010280001SBPPP.csv ࠍ⸃ᨆ‫ޕ‬Excel ࠺࡯࠲ࡈࠔࠗ࡞ 0001PP
㑆㓒ߣߒߚ‫ޕ‬
࡯ࠢߣࡇ࡯ࠢߩᤨ㑆㑆㓒㧕ߪ଻ሽߐࠇߥ޿ߩߢ෼❗ᤨᵄᒻߩᤨ㑆࠺࡯࠲߆ࠄ⸘▚ߒߡ P-P
MemCalc/Tonam2C ߢߪᔃ㔚࿑ R-R 㑆㓒ߪ⥄േ⊛ߦ଻ሽߢ߈ࠆ߇⣂ᵄ߆ࠄߩ P-P 㑆㓒㧔ࡇ
LF/HF ࠍ੤ᗵ␹⚻ᯏ⢻ߩᜰᮡߣߒߚ‫ޕ‬
⢻ࠍ⴫ߔ HF ᚑಽࠍ 0.15㨪0.5Hz ߣߒߚ‫ޕ‬
ⴕ޿੤឵␹⚻ᯏ⢻ߣ೽੤ᗵ␹⚻ᯏ⢻ࠍวࠊߖߚ LF ᚑಽࠍ 0.04㨪0.15Hz ߣߒ೽੤ᗵ␹⚻ᯏ
ࠣ࡜ࡓ MemCalc/Win ࠍ૶↪ߒ࠮ࠣࡔࡦ࠻㐳 5 ಽ㧔㊀ߨวࠊߖߪ 2.5 ಽ㧕ߣߒ๟ᵄᢙ⸃ᨆࠍ
ࠗࡓ⸃ᨆࡊࡠࠣ࡜ࡓ MemCalc/Tonam2C ࠍ↪޿ᵄᒻ଻ሽࠍⴕ޿‫ߩߘޔ‬ᓟ‫♽ᤨޔ‬೉⸃ᨆࡊࡠ
ߘࠇߙࠇߩࠕ࠽ࡠࠣ಴ജࠍ A/D ᄌ឵ߒߡࡁ࡯࠻ࡄ࠰ࠦࡦߩขࠅㄟߺ‫ޔ‬ᔃᜉ࡮ⴊ࿶࡝ࠕ࡞࠲
ᣇᴺ
ࡔ࡯ࠞ࡯ߪࡈࠖ࡝࠶ࡊࠬ␠⵾ࡕ࠾࠲࡯ߣࠦࡧࠖ࠺ࠖࠛࡦ␠⵾ࡄ࡞ࠬࠝࠠࠪࡔ࡯࠲࡯‫ޕ‬
ᔃ㔚࿑ᵄᒻߣࡄ࡞ࠬࠝࠠࠪࡔ࡯࠲࡯⣂ᵄߣߩ HRV ߩᲧセᬌ⸛
⸃ᨆ⚿ᨐ ࠮ࠣࡔࡦ࠻㐳 5 ಽ‫ޔ‬LF0.04㨪0.15Hz‫ޔ‬HF0.15㨪0.5Hz ߣߒߚ‫ޔ‬
LF/HF ߪ 1.0 ߣߥࠆ‫ޕ‬㧔ᷝઃߩ Excel ࠺࡯࠲ࡈࠔࠗ࡞ 0001PP ߦ޽ࠆ㧕
ฝਅߩ⸃ᨆ⚿ᨐߢ LF ߇ 1002.4㧔ms㨊2㧕HF ߇ 991.4㧔ms㨊2㧕‫ޕ‬
Ꮐ࿑ߪ⸃ᨆߔࠆ 5 ಽߩ࠺࡯࠲ߢฝ࿑ߪࠬࡍࠢ࠻࡞ኒᐲ㧔PSD㧕ߩ⚿ᨐ‫ޕ‬
࠮ࠣࡔࡦ࠻ᢙߪ 42 ୘ߢ޽ࠆ‫ޕ‬
⸃ᨆ᧦ઙߪห᭽ߢᱜᏱ୯߇ 9969 ୘‫ޔ‬ᄖࠇ୯ߪ 130 ୘ߢᄖࠇ୯₸ߪ 1.29㧑ߢ޽ࠆ‫ޕ‬
หᤨ᷹ቯߒߚ R-R 㑆㓒ߩ⸃ᨆ
⸃ᨆ⚿ᨐ ࠮ࠣࡔࡦ࠻㐳 5 ಽ‫ޔ‬LF0.04㨪0.15Hz‫ޔ‬HF0.15㨪0.5Hz ߣߒߚ‫ޔ‬
Ყセ
LF/HF ߪ 1.2 ߣߥࠆ‫ޕ‬㧔ᷝઃߩ Excel ࠺࡯࠲ࡈࠔࠗ࡞ 0001RR ߦ޽ࠆ㧕
ฝਅߩ⸃ᨆ⚿ᨐߢ LF ߇ 1127㧔ms㨊2㧕HF ߇ 921.1㧔ms㨊2㧕‫ޕ‬
࠮ࠣࡔࡦ࠻ᢙߪ 41 ୘ߢ޽ࠆ‫ޕ‬
ో૕ PP㧔21 ଀ 1 ଀ਇᢛ⣂㒰ߊ㧕4.062㧑 ⚂ 8 ୚ᄖࠇ୯߇ᄙ޿‫ޕ‬
22 ଀ߩᄖࠇ୯₸ ో૕ RR㧔21 ଀ 1 ଀ਇᢛ⣂㒰ߊ㧕0.522㧑
22 ଀ߩᲧセ (ᷝઃࡈࠔࠗ࡞ෳᾖ)
ో૕ߣߒߡߪ⦟ᅢߦ⷗߃ࠆ߇ߘࠇߙࠇߩ⸃ᨆ୯ߩ࠻࡟ࡦ࠼ࠣ࡜ࡈࠍᲧセߔࠆ‫ޕ‬
P-P ߪ 5 ಽ㑆ߥߩߢ޿ߊߟ߆⹺⼂ߢ߈ࠇ߫ᰳ⪭ㇱಽ߇ߥߊߥࠆ‫ޕ‬
HF‫ޔ‬LF/HF ߩᰳ⪭ㇱಽߪᄖࠇ୯߇ᄙ޿ߚ߼ߦ⸃ᨆਇน⢻‫ޕ‬
0001PP ߩ࠻࡟ࡦ࠼ࠣ࡜ࡈ
ᄖࠇ୯߇ዋߥ޿ߚ߼ߦᰳ⪭ㇱಽ߇ߥ޿‫ޕ‬
0001RR ߩ࠻࡟ࡦ࠼ࠣ࡜ࡈ
ᢙ⴫ߢߪ 0.5 ⑽ߩ㆑޿޽ࠆ‫ޕ‬
ᤨ㑆ߩߕࠇߪࡄ࡞ࠬࠝࠠࠪࡔ࡯࠲࡯ౝߢߩಣℂᤨ㑆ߥߩ߆㧫
0001PP ߣ RR ߩᲧセ
0001 LF㧔P-P㧕ߣ LF㧔RR㧕ߩᲧセ
HF ߩᲧセ
LF/HF ߩᲧセ
& $""..
%!
&
#&
' &
!&$
#"'
#!%
$$#%
&&#
%( &("
" (!($
--
$ % &..
--
% %"..
$
%"
%#"
%!%#
(
--
' "..
!(&(#$
!$&
'!$
' (!($
& %!$..
$
'#
--
%$
$'
( ( %
"
# %..
!%
%
--
!#
!"
$#&
"$$
# !(!#("#
" ..
%#
$
$&
%!("#(#
"$
! %#!..
" (!(!
"'%
$$..
""'
--
$
'%
--
'"$$
!&'
& !(#(
--
%!"..
&
&&!
"!%
''"
"
' &$#
%
'&
&"#
'"$
"#
!"!
%"
' "'
"'
"
" %
$
&$&
'
%%"
!(!#(
(( &
& !
$!&&
'
$'
--
%
&"&"
$"!
$
"&'%
!'
%..
' ..
--
--
(!(#
(!(#
!("%(
% $%..
--
'
# ( ( &
!!((!"
' #(&($
$ % "..
--
# "'..
--
" %..
--
! $!..
%!
'
# !("(!"
$" ..
--
#! !
"" #
8
%!$
*)
+)
--
!!(!( #
20
6457
$"$..
--
..
/
&
"
#
'
!
&
&
#
$!
#
$
#
!&
&
$
!
"
"
#
#
#
%
'
"
!!
!
&
!
'
$
"
"
%
+)*)
$
"%
#$&#&
$#&%$
'"#&
!
"&%
!!' '
$"
$&$
$%!!
% #$
!%!#
!&&
"& !
#$ "!
" &!
"
" !&!
"%$%&
%&#
''
#
!#&
!"#'&
" !&"
"%!!
""!$'
"'%"&
# !%$
%$!%
%'
$''
$
!"$%!
"'"
%#$
%!
..
--
--
!!..
--
$#..
--
&..
' !((!(!
&("($
% (#(
"
'
'$
#!
%'
!'$
!$
!!
%&
'!
"%#
$
"$
!&'
$
'
!
!#&!!
$ "''
$'! '
!"& !'!"
#$!&
#$&%
#'%
%"
""
$!
%
!#$&
"
&& $
$#&$'
#! '
"!
"%$
!#&
%&&
!'#
!!"
""''
"'&
$$&!
'!
% ##
$%'$
%
!% "#
"#&
!'& #
'
" %
!&&
"#
"'%&
$
&%
"%!'
&"$
# #
%! $
"'%%
!
#$ !%
' $
$
''$'
%&'
!$'
'
&'
#$'
"
'#
"%'!
!! '
##
""!"
!%
"%#
#'
"#&
$"#
"'#%
!
!
# !$#
$$&
!&%#
!%'&
"#%
"&"%
%&
%"
3
$& #
,,
1
"$
$ &
#
"
$
!
%"
%"
'
"
#
!!
$
$$'
'
! " "#!
!
'
!!&
$'
%
&
&
!$
$
#
1
#
3
$!"!
$'!
!" "
#$
%
$'$
#
'"
##"!'
!'!!#
#"
&
#$%$
!#
$
&#
$%
%
PLoS ONE | www.plosone.org
Cardiac diseases are a major cause of mortality in the world.
Studies carried out in 2006 in Colombia establish that heart
diseases produced circa 30875 deceases with an overall increase of
19.6% since 1999. Therefore, there has been great interest in the
development of computational tools for prognosis and diagnosis.
The main aim of these tools is to improve performance of
cardiologists on prognostic and diagnostic tasks, i.e., reducing both
the number of missed diagnoses or prognoses and the time taken
to reach such decisions. Under these conditions, it is expected that
detecting cardiac signs helps to decrease the mentioned decease
rates. At the same time, the introduction of computational systems
offers additional benefits, since the early identification of patients
would help the specialists to deal efficiently with certain cardiac
diseases. Moreover, as traditional risk stratifiers are commonly
used for prognosis, their positive predictive value is not as high as
the clinical practice demands.
Heart rate variability (HRV) has been often related with the
diagnosis and prognosis of certain cardiac diseases and, in fact, is a
standard method for studying the autonomic nervous system
(ANS) in heart control [1]. Several findings on HRV analysis have
demonstrated that geometric, statistical, spectral, multi-resolution
and non-linear approaches are powerful tools for the assessment of
cardiovascular health [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,
18]. Several patterns observed from HRV dynamics are often
related with myocardial infarction (MI) [18,19], sick sinus
syndrome (SSS) [1], multiple cardiac arrhythmias [1], atrial
fibrillation (AF) [20], congestive heart failure (CHF) [21], complete
heart block (CHB) [2], and ischemic cardiopaty [2], amongst
Introduction
* E-mail: [email protected]
1
February 2011 | Volume 6 | Issue 2 | e17060
others. Additionally there are some risk factors that affect HRV
directly or indirectly, such as blood pressure (BP), alcohol, smoking
and drug consumption [2].
Cardiovascular risk in general terms is often related to low or
excessive fluctuations of the NN intervals given by intrinsic and
extrinsic factors [10]. The relation between reduced HRV and
mortality risk was first shown by Wolf et al. in 1977 [22].
Furthermore, during the last 25 years, the significance of HRV in
assessing cardiac health has been recognized and various techniques
have been developed in order to analyze the fluctuations of NN
intervals. Time and frequency domain analyses are often referred to
as classical analysis as they were the very first methods being used
for the HRV processing [23,24,25,26,27,28,29]; these methods are
often inconvenient as they are linear and stationary methods
intending to model a highly non-linear phenomenon. These
methods as well as visual assessment of the raw HRV data are the
most common approaches used in clinical practice.
In 1996, the Task Force of the European Society of Cardiology
and the North American Society of Pacing and Electrophysiology
(ESC/NASPE) published standards on HRV analysis. They
proposed several time and frequency domain parameters and
their clinical uses, based on short-term (5 min) and long-term
(24 h) HRV data [12,21].
Relatively recent findings have shown that frequency domain
methods in HRV are related to hypertension [29]. Subjects with
risk factors such as hypertension, obesity, insulin resistance, among
others, generally show a high sympathetic activity which is often
presented before the clinical manifestation of hypertension. As
spectral methods are useful to assess the changes in sympathovagal
balance, hypertension has been accurately predicted.
Competing Interests: The authors have declared that no competing interests exist.
Funding: All the financial resources to carry out this research were provided by Universidad Autonoma de Occidente, Cali, Colombia. They provided the
necessary medical equipment for the data recording and analysis. The funders had no role in the preparation of the manuscript.
Copyright: ß 2011 Ramirez-Villegas et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Received November 2, 2010; Accepted January 17, 2011; Published February 28, 2011
Editor: Kelvin Wong, Royal Melbourne Institute of Technology, Australia
Citation: Ramirez-Villegas JF, Lam-Espinosa E, Ramirez-Moreno DF, Calvo-Echeverry PC, Agredo-Rodriguez W (2011) Heart Rate Variability Dynamics for the
Prognosis of Cardiovascular Risk. PLoS ONE 6(2): e17060. doi:10.1371/journal.pone.0017060
Statistical, spectral, multi-resolution and non-linear methods were applied to heart rate variability (HRV) series linked with
classification schemes for the prognosis of cardiovascular risk. A total of 90 HRV records were analyzed: 45 from healthy
subjects and 45 from cardiovascular risk patients. A total of 52 features from all the analysis methods were evaluated using
standard two-sample Kolmogorov-Smirnov test (KS-test). The results of the statistical procedure provided input to multilayer perceptron (MLP) neural networks, radial basis function (RBF) neural networks and support vector machines (SVM) for
data classification. These schemes showed high performances with both training and test sets and many combinations of
features (with a maximum accuracy of 96.67%). Additionally, there was a strong consideration for breathing frequency as a
relevant feature in the HRV analysis.
Abstract
1 Computational Neuroscience, Department of Physics, Universidad Autonoma de Occidente, Cali, Colombia, 2 Engineering Faculty, Department of Automatics and
Electronics, Universidad Autonoma de Occidente, Cali, Colombia
Juan F. Ramirez-Villegas1*, Eric Lam-Espinosa2, David F. Ramirez-Moreno1, Paulo C. Calvo-Echeverry2,
Wilfredo Agredo-Rodriguez2
Heart Rate Variability Dynamics for the Prognosis of
Cardiovascular Risk
PLoS ONE | www.plosone.org
Two distinct source materials were employed in this study: (a) A
database of risk and non-risk individuals given by Coomeva IPS
(Health Provider Company) experts and (b) the corresponding
2. ECG Database
This study was approved by the Institutional Review Board of
Universidad Autonoma de Occidente (UAO), Cali, Colombia.
Each patient in this study was informed in detail about the
procedure and signed an informed consent which guaranteed the
transparence of the test and the records’ future usage.
1. Ethics Statement
Materials and Methods
Classical methods of analysis are not absolutely suitable for
analysis of HRV [21,30,31,32]. Consequently, some approaches
have applied multi-resolution methods to HRV analysis. Multiresolution methods are related to the wavelet transform multi-level
decomposition; given the non-stationarity of the HRV signals [31],
the discrete wavelet transform (DWT) and wavelet packet
transform calculate the required high and low frequency subbands, enabling more accurate HRV analysis. Moreover, wavelet
entropy measures have been introduced for the implementation of
pattern recognition schemes and seem to provide high performances in diagnosing cardiac diseases [21]. However, to our
knowledge, previous work has not considered utilizing wavelet
energy measures analysis for HRV assessment.
Recently, new dynamic methods of HRV quantification have
been used to uncover nonlinear fluctuations in heart rate that
otherwise are not apparent. Several methods have been proposed:
Return map (Poincaré plot) calculation [13,14,15,33,34,35,36,37];
Lyapunov exponents/spectrum [2,38]; 1/f slope [39]; approximate and sample entropy (ApEn and SmEn, respectively) [20,40];
and detrended fluctuation analysis (DFA) [41]. Moreover, for the
last years these analysis techniques have been useful to understand
the HRV dynamics as the response of a highly non-linear system,
and therefore to produce discriminative enough features to reach
high success rates when several pattern recognition techniques are
implemented [2,20]. Pattern recognition in HRV has been used
for a variety of applications from prognosis to diagnosis of heart
diseases. The most commonly used schemes include: Artificial
neural networks (ANN) frames [20]; support vector machines
(SVMs) [1,20]; and linear statistical classifiers [21]. In general
terms, the performance of these classifiers in prognostic or
diagnostic tasks is relatively high (80% to 95% sensitivity in the
best cases); however, they have been used for the recognition of
several patterns in specific cardiac diseases (e.g., CHF, paroxysmal
AF, MI, cardiac arrhythmias, amongst others) rather than for the
prognosis of cardiovascular risk.
In this work, HRV analysis methods and pattern recognition
schemes (namely, artificial neural networks and support vector
machines) were used to discriminate between healthy control
subjects and cardiovascular risk patients. Extensive experiments
were carried out regarding the overall usefulness of the features
with emphasis on the prognostic values associated to classical and
non-linear analysis methods. We determined the potential
application of such methods to clinical practice in order to
increase the success rates of cardiovascular risk assessment. There
is a strong consideration for breathing frequency as a relevant
feature of the HRV analysis, given the respiratory sinus
arrhythmia (RSA) phenomenon [25], [42,43,44]. Additionally,
we provide a brief explanation on the implementation of advanced
HRV analysis software using the analyses performed in this work
and for automatic cardiovascular risk prognosis.
2
February 2011 | Volume 6 | Issue 2 | e17060
This phenomenon is known as respiratory sinus arrhythmia (RSA),
which is a rhythmic fluctuation of the heart beat intervals in a
phase relation with the inspiration and the expiration. The
autonomous nervous system (ANS) is the part of the nervous
system which extrinsically controls the vital functions and organs
such as the heart, lungs and glands. ANS is divided into two major
subsystems: The Sympathetic Nervous System and the
Parasympathetic Nervous System. These systems are antagonists
and responsible for the tuning of some physiologic mechanisms.
Intrinsic as well as extrinsic factors may affect such balance and
sometimes provoke different nervous activity patterns (sympathetic
or parasympathetic), which are often the cause of functional
irregularities; those patterns are given by hyper/hypoactivity of
such subsystems.
When the respiration is being monitored in a controlled
environment, the R-R intervals tend to be shorter during the
2.1. HRV and Respiratory Sinus Arrhythmia Considerations. Respiration has an important influence in HRV.
electrocardiographic data extracted by us using a medical expert
protocol.
The requirement for patient’s clinical history and positive or
negative cardiovascular risk verification was assessed by Coomeva
experts as this verification concedes extreme importance on the
validity and quality of the subsequent results of this work. The
main purpose of this step was to measure the relationship between
HRV indices and the subjects with risk factors. Diabetes Mellitus,
high blood cholesterol and other lipids, high blood pressure,
metabolic syndrome, overweight, obesity, physical inactivity,
tobacco and drug consumption are common risk factors of
cardiovascular heart diseases (CHD) and heart failure; nearly all of
these risk factors are associated with HRV reduction or excessive
fluctuations [45,46,47,48,49]. All the risk subjects (patients)
showed at least 3 significant risk factors according to the experts’
risk assessment.
In order to get the ECG and respiratory signals, a PowerLab
device -ref. ML865- (ADInstruments) and a piezoelectric band
-ref. MLT1132/D- were used. PowerLab is a data acquisition
system used in a variety of experiments and applications with
humans’ biopotentials. The unit can record more than 200000
samples per second and has individually selectable input
sensibilities. Additionally, it has a bioamplifier (used to record
any biological signal from the human body or other source) and an
internal processor with low and high pass filters. The hardware of
the PowerLab uses the software package Chart and Scope in order
to record and analyze each acquired dataset.
On the other hand, a non-invasive blood pressure (NIBP)
measure was taken into account for the HRV evaluation as it is
considered one of the main risk factors in the cardiovascular
assessment [49]. Such measure was extracted by using a
multiparameter monitor Spacelabs ref. 90309 which allows us to
monitor the following parameters: Electrocardiography, respiration,
temperature, non-invasive blood pressure and pulse oximetry.
Each register (a total of 90 electrocardiographic records) was
recorded following a medical protocol designed by the teamwork
using the frontal-bipolar derivation (D2) of the electrocardiogram.
The duration of each record was 5 minutes, following the
international standards established by the Task Force of the ESC/
NASPE [12]. In addition, the following information was taken
from each patient: Age, gender, weight, height, diagnosed
cardiovascular diseases, risk valuation given previously (without
risk, medium risk or high risk), diagnosed risk factors. All the
subjects of this study were in sinus rhythm during the ECG
recordings, furthermore, the mean breathing frequency for all was
12:1+2:1breaths/min.
HRV for the Prognosis of Cardiovascular Risk
220
50
0
2
3
4
FN
0
0
0
270
PLoS ONE | www.plosone.org
doi:10.1371/journal.pone.0017060.t001
FP
240
Decomposition
Level
1
100.00
99.19
96.53
96.06
Sensitivity
(Se - %)
100.00
99.19
96.53
91.98
Accuracy
(Ac - %)
Table 1. Performance of the R-peak detection algorithm at
various wavelet decomposition levels (sampling frequency of
200 kHz) using the Haar wavelet.
3.1. R-Peak Extraction.
The R peaks were extracted using
the Pan-Tompkins algorithm and the wavelet transform by
keeping the detail coefficients from 21 to 24 using the Haar
wavelet [51].
In order to know the effectiveness of the wavelet decomposition
levels, the performance of the algorithm was measured using the
detail coefficients from the first decomposition level to the fourth
decomposition level using on 6120 heart beats recorded with the
implemented medical protocol. The results of this evaluation
procedure are summarized in Table 1.
According to Table 1, the most appropriate wavelet decomposition level to extract the R-peak is 24 , with a sampling frequency
of 200 kHz for the ECG signal and using the Haar wavelet.
However, the diverse resolution levels on the R-peak extraction
procedure may affect the temporal resolution of the signal; thereby
a comparison of the estimation of the R-R series was made using
the resolutions taken into consideration for this work. We found
that, despite the detected false positives, there are no significant
spatial and frequency changes on the signal given by the
decimation process; thereby, there are no significant variations
on the spectral indexes estimation.
3.2. Outlier Removal Procedure. Past publications have
shown that eliminating the ectopic HRV data is often better than
interpolating them or doing any other cumbersome procedure
3. ECG Data Pre-processing
inspiration and larger during the expiration. Various theories have
been exposed about this phenomenon, according to many
experiments on animals [43]. The RSA comes from the control
given by the oscillations in the firing rates of the medullar neural
networks (central pattern generators, CPGs); the medullar neural
network shows periodic oscillations even when the afferent inputs
are interrupted. When those oscillations are carried out by afferent
stimuli from the receptors in the lungs and the thoracic wall, there
is a cardiac rhythm oscillation known as RSA.
Obtaining the respiratory frequency has been used to evaluate
the magnitude of the RSA, defined as the sum of the power
spectral density estimations on the respiratory band. The
variations induced by the respiratory rate introduce a significant
change on the measurement of HRV; however, if the respiratory
frequency is a constant (i.e., 12 breaths/min) and the tidal volume
is relatively constant for the maximum capacity, the error in the
measurement caused by respiratory irregularities is eliminated and
allows a better stability and an effective comparison of the RSA
magnitude between patients [50]. In this study, the above criteria
were used to measure HRV in the patients. Consequently, both
the recorded ECG signals in resting conditions and specific
conditions of the environment were taken into account for the vital
signs stabilization.
3
maxxj {x
, for i~0, 1,2 ::: k:
s
ð1Þ
February 2011 | Volume 6 | Issue 2 | e17060
4.1. Statistical or Time-domain Measures. In this
approach, a set of 7 well-known statistical indexes were
calculated: First, the mean and the standard deviation (SDNN)
of the NN intervals of each 5 min record; second, the square root
of the mean of the sum of the square of differences between
adjacent NN intervals (RMSSD); third, the so-called pNN50 was
computed as the NN50 count value divided by the total of all NN
intervals, where NN50 is the count of adjacent intervals differing
by more than 50 ms in the entire HRV record; fourth, the
interquartile margin of the NN intervals (MIRR) was also
calculated, i.e., the first quartile subtracted from the third
quartile of the NN series; in addition, the median of the
absolute differences of the NN intervals (MDARR) and the
standard deviation of the differences between adjacent NN
intervals (SDSD) were calculated.
4.2. Spectral or Frequency-domain Measures. Spectral
or frequency-domain measures are based on the power spectral
density (PSD) analysis of the R-R series. In this kind of analysis
some processing techniques, such as interpolation and detrending,
are necessary. The spectral measures have the advantage of
relating the power of variation in different frequency bands to
different physiological modulating effects [2,24]. Extensive
experiments have shown that parametric methods (AR spectrum)
tend to produce better results than classical nonparametric
methods (Welch’s periodogram) when the data length of the
signal is relatively short, as is the case with HRV data [20]. For this
reason, we applied parametric PSD estimation. Three main
spectral measures are distinguished from the spectrum of the R-R
series: The power of the very low frequency band (VLF band, 0–
0.04 Hz), the power of the low frequency band (LF band, 0.04–
0.15 Hz) and the power of the high frequency band (HF band,
0.15–0.4 Hz). These frequency components and their normalized
values (NLF and NHF) were calculated using a standard
integration procedure (area under the curve) of the spectrum
regions. In addition, the ratio of LF to HF was calculated as it
indicates the balance of ANS.
For parametric spectral methods, the data can be modeled as
the output of a discrete and causal filter whose input is white
noise.
4. HRV Classical Measures
where x is the mean and s is the standard deviation of the
sample ðn{iÞ and j is the position of one given value of the
sample. In this way, the critical values of the test are determined by
specifying a and by calculating b and lðbÞ in order to calculate the
t-student statistical test [53].
Ri ~
[52]. Grubbs Test extended by Rosner method was used in this
work [53,54]. Assuming a normal distribution, Grubbs’ outlier test
can be used to remove one outlier. Nevertheless, if we decide to
remove this outlier, we might be tempted to run Grubbs’ test again
to see if there is a second outlier in the data; however, the rejection
criteria changes. Rosner has extended Grubbs’ method to detect
several outliers in one dataset. Rosner’s several outliers detection
method seems to be compatible with HRV signals in general ways
[54].
For a specified limit k of the number of outliers, the procedure is
calculated by using reduced samples of length n, n{1, :::,
n{kz1, respectively. For each sample (n{i):
HRV for the Prognosis of Cardiovascular Risk
s
ð2Þ
ð3Þ
Ej
:
Etotal
ð4Þ
j
X
pj log2 (pj ):
ð5Þ
PLoS ONE | www.plosone.org
In [21], the wavelet entropy is calculated as a unique multiresolution measure. However, in this work, both the energy and
entropy measures are taken into account due to their high
cardiovascular risk prognostic value.
A problem with wavelet packets implementation is coping with
redundant information coming from the wavelet transform
computation for the approximation and detail coefficients [54].
In order to avoid the redundant information due to wavelet packet
decomposition, a standard Principal Component Analysis (PCA)
was computed. PCA chooses a dimensionality reduction by linear
projection that maximizes the scattering of all projected samples.
The resultant feature space is the projection of the original data set
over the covariance matrix eigenvectors, which constitutes the
WS~{
In this way, using the definition of entropy given by Shannon,
the entropy can be calculated as follows:
pj ~
Then, the total energy can be calculated as the mean value of
the energy for each coefficient.
On the other hand, the wavelet entropy can be calculated as the
probability distribution of the coefficients or the wavelet energy
into normalized values:
2
Ej ~ Cj :
According to the literature consulted [21], there is a three step
procedure to calculate the wavelet measures: First, calculate the
wavelet packet coefficients; second, calculate the wavelet energy;
and third, calculate the wavelet entropy.
The wavelet packet analysis in HRV is used to separate the
signal into multiple scales. This method allows us to analyze both
frequency and spatial domains and removes polynomial nonstationarities of the signal [11]. Due to this property, wavelet
analysis is much more suitable for analyzing HRV signals than
statistical and spectral methods. In this work, the wavelet packet
analysis was implemented by using the DB4 function as mother
wavelet [21]. The decomposition was performed at a level of 5
[21,56,57].
Once the wavelet coefficients are known, it is possible to
calculate the energy for each coefficient:
5. Wavelet Packet Measures
where a(k) are the recursive coefficients calculated by covariance
method [2]. An important factor in the implementation of the AR
method is the selection of the order [28]. The order p = 16 for the
AR model was taken into account for this study [26,54,55].
k~1
ep
^ AR ð f Þ~ 1 ,
P
p
P
{2pjkf fs ^p ðkÞexp
a
1z
f
The spectral power of an AR process is given by:
4
February 2011 | Volume 6 | Issue 2 | e17060
HRV has been often evaluated using non-linear methods
[6,20,58]. These methods seem to be very relevant in feature
extraction of the HRV series; on the other hand, HRV dynamics is
highly non-linear and, actually, HRV series is the response of a
chaotic system, i.e., a system with high sensitivity on the initial
conditions [59,60,61].
6.1. Poincaré Map-Based Features. The most popular
non-linear technique to assess the HRV is the so-called Poincaré
map (also called return map or Lorenz map). The Poincaré map
corresponds to the reconstruction of the attractor of the system
based on the HRV experimental series [60]. This map can be
constructed by plotting each RR interval against the next interval.
This plot is very useful in summarizing beat-to-beat information
on heart behavior. The Poncaré plot is a simple visual
interpretation technique and it has proved to be a very powerful
predictor of disease and cardiac dysfunction [21].
In order to extract features of the Poincaré map, two methods
were applied in this work: The ellipse fitting technique and the
histogram technique [17].
A group of axes with orientation onto the identity line or
principal diagonal is the main feature of the ellipse fitting
technique. The axes of the plot are related to a new group of
axes with a rotation of p=4 [17].
On the new reference system axes, the dispersion of the points
through the x1 axis is measured by its standard deviation, denoted
SD1. On the other hand, the magnitude of the points through the
identity line shows the level of long term variability, denoted SD2,
i.e., the standard deviation over the x2 axis.
On the other hand, the ellipse approximation is satisfactory in
many cases; however, the shortening that occurs on short R-R
intervals is not taken into account by this technique [35]. The
histogram approximation has been used to evaluate the distribution of the data into several time ranges. There are three types of
histograms: The width histogram, the NN interval histogram and
the length histogram [17].
As the visual interpretation of the histogram can be useful to
extract information about the heart, it is necessary to parameterize
it. The computation of the width of the three histograms is a very
strong feature for the prognosis of cardiovascular risk, as it gives
the absolute statistical ranges of the NN intervals and its
projections. These features are useful for assessing short term
HRV, long term HRV and the distribution of NN intervals itself.
These histogram widths were taken into account in this study.
6.2. Complexity Analysis. There are several approximations for estimating regularities of different kinds of signals.
The most widely used complexity measures for short and noisy
data series are approximate entropy (ApEn) and sample entropy
(SmEn). These features assign a non-negative number to temporal
series in order to quantify the regularity of its fluctuations. Given
this fact, complexity measures have been highly useful for the
analysis of HRV signals [20].
To calculate ApEn and SmEn from temporal series it is
necessary to choose two parameters: A length m and a window size
r. ApEn measures the logarithmic similarity amongst neighboring
input patterns (those with a separation radius less than r) for m
contiguous observations. On the other hand, SmEn is an unbiased
estimator introduced to avoid the self-couplings and to quantify
the regularity of highly irregular temporal series. SmEn is equal to
6. Non-Linear Measures
weights of each feature on the input space according to how much
of the model’s variability is explained by them. The most
important entropy and energy features (in terms of variability)
were selected.
HRV for the Prognosis of Cardiovascular Risk
ð6Þ
fj ðxÞ~
expðxÞ{expð{xÞ
:
expðxÞzexpð{xÞ
ð7Þ
where wmji is the synaptic weight i of the neuron, hj is the bias and fj
is the activation function. In the current model we have two nonlinear transfer functions corresponding to the hidden and the
output layers, given by the following equations, respectively.
i~1
!
N
X
Yjm ~fj smj ~fj
wmji xmi zhj ,
Multilayer Perceptrons (MLP) are frequently implemented for
classification tasks, given their generalization capabilities. In this
work, a standard three-layer network has been proposed.
Let m (xmi ) be an input pattern, the output of a single artificial
neuron of the hidden layer is given by the following equation:
8.1. Multi-layer Perceptron (MLP) Neural Network.
Multilayer Perceptrons (MLP), Radial Basis Function (RBF)
networks and several Support Vector Machines (SVM) were
evaluated for the classification stage of this work. All classification
schemes were trained to capture the difference between cardiovascular risk subjects and healthy ones.
8. Classification
PLoS ONE | www.plosone.org
5
February 2011 | Volume 6 | Issue 2 | e17060
ð8Þ
Ni
1 X
î ,
x
Ni i~1
ð9Þ
Normalization,
Validation
and
Performance
PLoS ONE | www.plosone.org
Measures. In order to train the classifiers and perform the
KS-tests, all samples were normalized using the MinMax
normalization [21].
There are several ways to compute the performance of a
recognition system. The pattern recognition schemes were
evaluated using two different procedures: The calculation of
several performance measures such as sensitivity (Se), specificity
8.4.
where Ni is the number of input vectors associated to each
Gaussian node of the hidden layer. This procedure is computed
until the stabilization of the synapses is reached (the weights do not
change from a training cycle to another).
The Gaussian scale parameters for each hidden neuron can be
determined by the approximated magnitude of the influence
radius of each neuron on the input space in relation to other
neurons near to the j neuron.
This network consisted of three layers (the sensory input layer,
the hidden layer and the output layer), with 5, 3 and 1 neurons,
respectively.
8.3. Support Vector Machines (SVMs). In the most
cumbersome case, the patters are not linearly separable. The
main objective of the SVM schemes is to map the input data from
the N-dimensional space to the M-dimensional space (M.N),
where the classes are supposed to be linearly separable and can be
classified by the calculation of a standard separating hyperplane
[64,65] (see Appendix S2 for details on the implementation of
SVM).
In our work, polynomial and radial basis function (RBF) as well
as linear SVM were used to classify the input data.
^ (tz1)~
w
This architecture guarantees that the network’s output will run
over 0 and 1 (given the sigmoid output function).
We used the Levenberg-Marquardt backpropagation algorithm
to train the MLP neural network, as it is the most popular and
successful learning method for training MLPs. The algorithm
employs iterative mean squared error minimization using least
squares curve fitting [64].
The network consisted of three layers (the sensory input layer,
the hidden layer and the output layer), with 5, 200 and 1 neurons,
respectively.
8.2. Radial Basis Function (RBF) Neural Network. RBF
network is a well-known classifier which combines supervised and
unsupervised learning. A standard three-layer RBF network was
implemented in this work. The hidden layer of the network is
responsible for producing a non-linear expansion of the input
space to a hidden space where the classes are linearly separable by
unsupervised learning [64]. The most popular unsupervised
learning rule for the hidden layer is the so-called k-means.
In this procedure, we establish a number of neurons (k), whose
synaptic weights wji are randomly distributed on the input space
and then a similarity measure is calculated, i.e., the Euclidean
distance.
When this procedure is applied to the whole layer, the weight
update rule is then calculated from:
1
:
1zexpð{xÞ
6
February 2011 | Volume 6 | Issue 2 | e17060
All the feature analysis results obtained in this work were
reported using standard box diagrams, given their suitability for
this statistical analysis and since their interpretation can be
performed in a remarkably easy way, regardless of the fact that
variables could present a great deviation from the normal
distribution. The tops and bottoms of each box are the 25th and
75th percentiles of the samples, respectively. The line in the middle
of each box is the sample median; this illustrates the skewness of
the samples. The dashed lines extending below and above each
box are drawn from the ends of the interquartile ranges to the
furthest observation within the dashed line length. Crosses are the
outliers of the samples; they represent atypical data sufficiently
distant from the limits of the box. Note that their elimination is not
justified, provided that the objective of box diagrams is to give a
complete knowledge of the shape of the data distribution.
Statistical, spectral, multi-resolution and non-linear features
were calculated from the recorded HRV database. The statistical
analysis for the classical measures is shown in Figure 3.
1. Statistical, Spectral, Multi-resolution and Non-linear
Analysis of Extracted HRV Data
Results
For each subject statistical, spectral, multi-resolution and nonlinear features were calculated using MatlabTM 7.6.0. The flow
diagram of the whole system is depicted in Figure 2.
The application works with 5-min electrocardiographic (ECG)
signals; thereby, a preprocessing step is involved in the procedure,
i.e., the R-peak extraction and NN intervals calculation.
Therefore, there is an ectopic beat removing algorithm, i.e.,
Grubbs Test extended by Rosner outlier detector. The next stage
contains the feature extraction, i.e., statistical, spectral, multiresolution and non-linear features calculation. For spectral features
calculation, the 4 Hz cubic interpolation and the smoothness
priors l~1000 as detrending method are performed in the whole
HRV dataset. The wavelet packet-based features are extracted
using DB4 mother wavelet and the decomposition is done to a
level m~5 [21]. Ellipse fitting and histogram features are
extracted from the first-order Poincaré plot. SmEn and ApEn
complexity measures are extracted using r~0:1s and m~1 : 4.
All the features mentioned above can be displayed by the user.
Feature selection is performed via KS-tests and the top-5 features
are used to distinguish normal subjects (N) from cardiovascular risk
ones (R). The classification scheme is responsible for giving the
final prognosis result.
The respiratory rate mean and standard deviation feature can
be used to corroborate that respiratory rate is relatively constant
over the whole ECG record and that is approximately equal to 12
breaths/min. If this condition is not fulfilled, the results would be
invalid.
9. Computational Implementation
(Sp), positive predictive value (Pp), negative predictive value (Np)
and accuracy (Ac), all in the interval [0.00,100.00]. On the other
hand, Receiver Operating Characteristic (ROC) curve was used in
order to measure the accuracy of ANNs. The ROC curve is the
plot of the true positive rate (Se) versus the false positive rate (1 –
Sp) for different testing points in a diagnostic test. An ROC curve
illustrates various aspects: First, it shows the tradeoff between the
sensitivity and the specificity in the evaluation of a model; and
second, it is a measure of the accuracy of the algorithm given by
the area under the curve, i.e., the algorithm’s probability of giving
correct classifications when a new input pattern is presented [29].
fk (x)~
Statistical significance of these results was tested using a
standard two-sample KS-test. A level pv0:05 was considered a
statistically significant difference [63]. On the other hand, in order
to use the best features on the classification stage, a level
pv0:0001 was considered statistically relevant enough as KS
test-based selection criteria.
Figure 1. Cumulative distribution plots comparison between a given feature considered in this work and the standard normal
distribution.
doi:10.1371/journal.pone.0017060.g001
Up to this moment, several statistical methods have been related
to feature selection for training and testing classifiers and for
improving their overall performance. As most of computational
methods in pattern recognition require a feature selection step, the
nature of such procedure depends largely on the structure of the
data. As a result, many computational approaches use parametric
statistics, e.g., the so-called multivariate analysis of variance
(MANOVA), which often reports adequate results for such
purposes. However, we cannot assume that all the data has
resemblance to a standard normal distribution at high statistical
significance. Therefore, we implemented two methods in order to
test the normality of the data: First, we conducted the Pearson’s
chi-square test; and second, the one-sample KS-test. According to
our results, the data (for all the features taken into account), has a
remarkably different distribution when compared to the normal
distribution (with pv10{9 ). As both methods reported quite
similar results, we concluded that non-parametric (distributionfree) statistical methods needed to be implemented at this stage.
These methods, unlike parametric statistics, make no assumptions
about the probability distributions of the variables being assessed.
Figure 1 depicts the empirical cumulative distribution plot for a
given feature (note the significant difference to the standard
normal distribution).
7. Statistical Significance Tests and Feature Selection
the negative of the natural logarithm of a conditional probability.
It is the probability that sequences close to each other for m
consecutive data points will also be close to each other when one
more point is added to each sequence [20].
For both, ApEn and SmEn calculation, it is recommended to
take r~ks, where s is the standard deviation of the data series
and k runs over 0.1 to 0.2 [62] (for details regarding the calculation
of complexity analysis measures see Appendix S1).
HRV for the Prognosis of Cardiovascular Risk
HRV for the Prognosis of Cardiovascular Risk
PLoS ONE | www.plosone.org
The KS-test showed that all statistical features (mean, standard
deviation, RMSSD, pNN50, MIRR, MDARR and SDSD) are
statistically significant (pv0:05) for the comparison of normal (N)
and cardiovascular risk (R) subjects. The most statistically
significant features were the standard deviation, RMSSD, MIRR
and SDSD (with pv0:001); the remaining features reported
significances close to the alpha value.
In addition, LF power, LF/HF ratio, NLF power and NHF
power showed statistically significant differences (pv0:05) be-
7
February 2011 | Volume 6 | Issue 2 | e17060
tween normal and cardiovascular risk subjects; however, VLF and
HF powers do not discriminate between these two groups
(pw0:05).
Standard PCA was applied to the multi-resolution measures in
order to obtain the most relevant features in terms of variance.
The total of selected groups of wavelet coefficients was 26 out of 62
(the total of wavelet packet coefficients from a decomposition level
of 5). These 26 groups of coefficients, according to the PCA, retain
approximately 98% of the variance of the model; however, the
Figure 2. Flow diagram of the computational implementation of the computational tool reported in this work.
doi:10.1371/journal.pone.0017060.g002
HRV for the Prognosis of Cardiovascular Risk
were statistically significant with0:003vpv0:05 (1 from the first
decomposition level, 1 from the second one, 2 from the third one,
3 form the fourth one and 5 from the last one). The remaining
entropy and energy components were not taken into account
because they did not discriminate between the two groups (normal
and cardiovascular risk subjects) with statistical significance.
Non-linear analysis KS-test results are illustrated in Figure 5. The
results of the non-linear analysis showed that SD1 and SD2 ellipse
fitting features are statistically significant (pv0:01, in the best case);
SD1 being the less significant one. Additionally, there is statistical
difference in histogram technique parameters, i.e., the widths of the
NN intervals, the width and the length histograms of the HRV
records, among normal and cardiovascular risk subjects. According
to the statistical analysis, statistical significance increases with the
NN intervals histogram width (pv0:00001) and the length
histogram width (pv0:00001). For the case of the width of the
width histogram the statistical difference is relatively high (pv0:01).
PLoS ONE | www.plosone.org
8
February 2011 | Volume 6 | Issue 2 | e17060
Figure 4. PCA transform of multi-resolution features of the 5-min HRV records from Normal (black diamond) and cardiovascular
risk (red square) subjects. (a) entropy features and, (b) energy features.
doi:10.1371/journal.pone.0017060.g004
features projections given by this transformation were not used to
train the classifiers due to the decreasing statistical significance of
the projected features. The main results of PCA for the two
principal components are illustrated in Figure 4. The second
principal component projection showed statistical significance in
both energy and entropy wavelet features; these two features retain
approximately 50% of the variance of the model. In this analysis,
even the third component showed statistical significance; however,
the rest of the projected data were not significantly different or
discriminative between normal and cardiovascular risk subjects.
A total of 27 wavelet packet-based features were selected.
Statistical significance levels of the entropy measures showed that
15 wavelet entropy components were statistically significant
with0:00001vpv0:05 (1 from the first decomposition level, 2
from the second one, 1 from the third one, 3 form the fourth one
and 8 from the last one). On the other hand, the significance levels
for energy features showed that 12 wavelet energy components
Figure 3. Box diagrams. (a) Statistical measures and, (b) Spectral measures from 5-min HRV records from normal (N) and risk (R) subjects in
normalized values (y axis).
doi:10.1371/journal.pone.0017060.g003
HRV for the Prognosis of Cardiovascular Risk
2.1. KS test-based Feature Selection Results.
In order to
evaluate each artificial intelligence (AI) scheme, the cross
validation method was used. This method allows generating the
indexes for the validation of the N observations by choosing
randomly the training and test observations. Each AI scheme was
trained using approximately 66% of the observations (60 HRV
records, 30 from normal subjects and 30 from cardiovascular risk
subjects) and tested using the remaining 33% of them (30 HRV
records, 15 from normal subjects and 15 from cardiovascular risk
2. Artificial Intelligence Schemes Classification for the
Prognosis of Cardiovascular Risk
non-linear and 2 multi-resolution features. These features were
used to train and test the ANN and SVM classifiers. Additional
experiments were conducted in order to compare the performance
of the four principal PCA feature projections to those chosen by
KS-tests.
PLoS ONE | www.plosone.org
9
February 2011 | Volume 6 | Issue 2 | e17060
Figure 6. Box diagrams of complexity measures. (a) ApEn and, (b) SmEn (r~0:1s and m~1{4) of the 5-min HRV records from normal (N) and
risk (R) subjects in normalized values (y axis).
doi:10.1371/journal.pone.0017060.g006
ApEn and SmEn significance test results are contained in
Figure 6. ApEn shows increased statistical significance (pv10{7 )
for m~1 : 4. On the other hand, SmEn shows statistical
significance (pv0:001) only for m~1 : 3, for m~4 there is no
statistical significance for the comparison of normal (N) and
cardiovascular risk (R) subjects. Table 2 contains the p values of
ApEn and SmEn at various m values; the m value varies from 1 to
4. The subsequent values did not allow obtaining statistically
significant ApEn and SmEn values due to the relatively short
duration of the HRV records (5-min per HRV record) [20].
The total number of features extracted from spatial and
frequency domains, multi-resolution and non-linear algorithms is
equal to 52 (7 statistical, 6 spectral, 27 multi-resolution and 12
non-linear features). The total number of optimal features
extracted by KS-tests is equal to 5 due to the statistically
significant discrimination presented by them (the next section will
illustrate this fact clearly). The resulting feature set combined 3
Figure 5. Box diagrams of Poincaré map-based features. (a) SD1 and SD2 ellipse fitting features and, (b) NN histogram (NN), width histogram
(WNN) and length histogram (LNN) features of the 5-min HRV records from normal (N) and risk (R) subjects in normalized values (y axis).
doi:10.1371/journal.pone.0017060.g005
HRV for the Prognosis of Cardiovascular Risk
1.21E–09
4.87E–09
6.83E–08
1
2
3
4
p-value (SmEn)
0.6101
2.52E–06
1.60E–04
2.38E–07
domain) and 15 features (5 from non-linear domain, 6 from multiresolution domain and 4 from statistical domain) selected by KStest. The results of MLP and RBF neural networks as well as of
SVM classifications of HRV records from normal (N) and
cardiovascular risk (R) subjects are illustrated in Table 3. The
ROC curves for both neural network schemes are depicted in
Figure 7; the areas under the curve are 0.9822, 0.8889 and 0.9024
for MLP; and 0.8800, 0.8667 and 0.8400 for RBFNN, for 5, 10
and 15 features, respectively. The linear SVM C value was fixed to
1 by default. The SVM with polynomial and RBF kernels’ C and c
parameters produced the best classification performances. All C=
and c were determined into a heuristic way.
According to Table 3, the higher performance was reached by
MLP using the top-5 features selected by KS-test. It is important to
note that the performance of the classifiers was similar in all cases;
therefore, many combinations of features are suitable for the
prognosis of cardiovascular risk as it is proposed in our work. For
the 5 optimal features, linear SVM selected 36 support vectors;
SVM (RBF kernel)*
SVM (Polynomial kernel)
SVM (Linear)
RBFNN
MLP*
SVM (RBF kernel)*
SVM (Polynomial kernel)
SVM (Linear)
RBFNN
MLP*
SVM (RBF kernel)
SVM (Polynomial kernel)
SVM (Linear)
PLoS ONE | www.plosone.org
10
93.94
Test set
90.91
100.00
Training set
Test set
77.27
100.00
Test set
Training set
100.00
Training set
60.00
Test set
73.33
83.33
Training set
Test set
96.97
100.00
Test set
Training set
100.00
Training set
86.36
Test set
86.36
100.00
Training set
Test set
60.00
100.00
Test set
Training set
96.67
Training set
86.67
Test set
74.24
100.00
Training set
Test set
84.09
100.00
Test set
Training set
100.00
Training set
72.73
Test set
85.71
71.88
Training set
Test set
93.33
96.67
Training set
Test set
Se (%)
100.00
Training set
*Classifiers that presented the higher performances on each experiment.
doi:10.1371/journal.pone.0017060.t003
15
10
MLP*
5
RBFNN*
Classifier
#Features
77.27
100.00
70.45
100.00
77.27
100.00
100.00
83.33
100.00
100.00
81.82
100.00
75.00
100.00
77.27
100.00
100.00
90.00
80.00
100.00
78.79
100.00
70.45
100.00
72.73
75.00
93.33
100.00
100.00
100.00
Np (%)
92.73
100.00
88.57
100.00
77.27
100.00
72.43
83.33
78.95
100.00
96.43
100.00
84.62
100.00
85.00
100.00
71.43
96.43
85.71
100.00
75.36
100.00
81.58
100.00
72.73
70.00
87.50
96.67
93.75
100.00
Pp (%)
80.52
100.00
75.47
100.00
77.27
100.00
100.00
83.33
100.00
100.00
84.21
100.00
77.55
100.00
79.17
100.00
100.00
90.63
81.25
100.00
77.78
100.00
74.00
100.00
72.73
76.67
92.31
100.00
100.00
100.00
Ac (%)
85.61
100.00
80.68
100.00
77.27
100.00
80.00
83.33
86.67
100.00
89.39
100.00
80.68
100.00
81.82
100.00
80.00
93.33
83.33
100.00
76.52
100.00
80.00
100.00
72.73
73.33
89.66
98.33
96.67
100.00
February 2011 | Volume 6 | Issue 2 | e17060
Sp (%)
Table 3. MLP, RBF neural networks and linear SVM (C~1), SVM with polynomial kernel (C~1; c~4) and SVM with RBF kernel
(C~1; c~3) classifications using the top 5, top 10 and top 15 features selected via KS-test of the HRV records from normal (N) and
cardiovascular risk (R) subjects.
patients). This division gave an optimal performance (high
generalization levels) for the ANNs as well as for SVMs.
Experiments were performed using 5 (3 from non-linear domain
and 2 from multi-resolution domain), 10 (4 from non-linear
domain, 3 from multi-resolution domain and 3 from statistical
doi:10.1371/journal.pone.0017060.t002
p-value (ApEn)
1.87E–08
m
Table 2. Statistical significance for complexity measures at
different m values.
HRV for the Prognosis of Cardiovascular Risk
100.00
77.27
Test set
79.55
Training set
Sp (%)
87.88
100.00
86.36
100.00
90.91
100.00
PLoS ONE | www.plosone.org
doi:10.1371/journal.pone.0017060.t004
RBF
Test set
86.36
100.00
Test set
Training set
Linear
Polynomial
Se (%)
100.00
Training set
Kernel
86.44
100.00
85.37
100.00
86.96
100.00
Np (%)
79.45
100.00
80.85
100.00
90.48
100.00
Pp (%)
82.58
100.00
82.95
100.00
88.64
100.00
Ac (%)
Table 4. Results of linear SVM (C~1), SVM with polynomial
kernel (C~1; c~4) and SVM with RBF kernel (C~1; c~3)
classifications using the total of features of the HRV records
from normal (N) and cardiovascular risk (R) subjects.
non-linear and multi-resolution features on HRV analysis perform
better than the conventional and clinical-applied statistical and
spectral analysis methods [2,27,28,29,30,31]. On the basis of
determining the real usefulness of the non-linear and multiresolution features in terms of the prognostic value (Se, Sp, Pp, Np
and Ac), extensive experiments were carried out in this work.
2.2. Do Multi-resolution and Non-linear Features
Perform Better than Conventional Statistical and Spectral
Features? A topic of remarkable discussion has been whether
polynomial kernel SVM selected 28 support vectors; and RBF
kernel SVM selected 43 support vectors from the same training
dataset.
In addition, as part of the experiments, we evaluated the effect
of using all the features from all the analysis methods shown in this
paper. The results showed that for SVM classifiers, the three
schemes evaluated (linear, polynomial kernel and RBF kernel
SVM) reached higher performances than ANN ones. The main
results of these experiments are registered in Table 4. The results
of ANN classification were not included due to the poorness of the
classification performances, this is a common effect given by the
overfitting produced by the high dimensionality of the input space.
11
February 2011 | Volume 6 | Issue 2 | e17060
The proposed classifiers were trained using statistical and
spectral features (classical analysis features) and using the multiresolution and non-linear features (novel features) separately,
especially in order to know the overall suitability of each group of
features with respect to the HRV analysis. The main results of our
experiments are depicted in Table 5. The ANN and SVM schemes
were implemented under the same setup as described in the
previous section. When statistical and spectral features were used
to classify the data, the number of support vectors chosen by linear
SVM, polynomial kernel SVM and RBF kernel SVM was 36, 24
and 46, respectively. On the other hand, when multi-resolution
and non-linear features were used, linear SVM, polynomial kernel
SVM and RBF kernel SVM selected 30, 24 and 46 support
vectors, respectively.
Table 5 illustrates at least two important findings from the
experiments carried out for this part of the research: First, nonlinear and multi-resolution features have remarkably higher
prognostic value than the features referred to as classical analysis
(this fact was also confirmed by the statistical significance of the
features already reported in this work); and second, the results
indicate indirectly the overall usefulness of certain combinations of
statistical and spectral measures and their expected effectiveness in
clinical applications (it is important to note that even when these
classical features are not very suitable to HRV analysis, the nonlinear nature of the classifiers is the main catalyst to reach
moderately good performances).
Due to the non-linearity and non-stationarity of the HRV
signal, many authors prefer using non-linear and multi-resolution
features rather than statistical or spectral ones in order to train
classifiers [20,21]. Nonetheless, it is important to note that as the
complexity of the features increases, also its medical interpretation
becomes obscure and cumbersome. Furthermore, especially multiresolution decomposition produces an important loss on the
potential interpretation of features and signals, regardless of the
remarkable increase of the statistical significance of the features
reported using such analyses.
2.3. PCA-based Feature Selection Results. In addition to
the results reported above, several classification experiments
regarding PCA features’ projections were performed in our
work. They were carried out using the cross validation method
Figure 7. ROC curves. (a) For MLP neural networks, (b) RBF neural networks obtained from the classification of the HRV test records using 5, 10 and
15 top features selected by KS-test; the areas under the curves are equal to 0.9822, 08889 and 0.9024 for MLP and 0.8800, 0.8667 and 0.8400 for
RBFNN, respectively.
doi:10.1371/journal.pone.0017060.g007
HRV for the Prognosis of Cardiovascular Risk
SVM (RBF kernel)
73.33
95.45
88.64
90.91
RBFNN
SVM (Linear)
SVM (Polynomial kernel)
SVM (RBF kernel)
80.00
68.18
SVM (Polynomial kernel)
MLP
SVM (Linear)
PLoS ONE | www.plosone.org
Heart rate variability has been related to numerous cardiac
diseases. According to the literature, the short variations on
Discussion
under the same setup reported above. These experiments were
made on the basis of two main goals: First, in order to investigate
whether the significance levels of the projected features remained
similar to the original ones; and second, to classify the data using
the 3, 4, 5 and 10 principal features’ projections from such PCA
analysis, as the variance retained by the 3, 4, 5 and 10 first
components for both groups –normal healthy subjects and
cardiovascular risk patients– was equal to 77.65%, 83.33%,
86.96% and 94.45%, respectively. Furthermore, the 99% of the
variance of the model was retained by the first 21 components.
According to the KS-test results, from the five first PCA
projections, three of them remained to be statistically significant
(pv0:03), and in one case, the seventh PCA projection also
showed high statistical significance (pv0:04); the rest of the
components were not statistically significant for the comparison of
normal (N) and cardiovascular risk (R) subjects.
The results of the ANN and SVM schemes classification of HRV
records from normal (N) and cardiovascular risk (R) subjects are
depicted in Table 6. With respect to the classification results using
the projected features, linear SVM selected 46, 32, 27 and 23
support vectors from the whole training dataset; polynomial kernel
SVM chose 28, 22, 21 and 21 support vectors; and RBF kernel
SVM picked up 46 support vectors in all cases. Table 6 clearly
illustrates important improvements for the performance of SVMbased classifiers. On the other hand, MLP and RBFNN schemes
presented highly limited performances over the entire track of
experiments regarding the implementation of PCA. However, the
greatest performance reported by our work was achieved by MLP
when the KS test-based feature selection was performed.
The main aim of using these four different numbers of PCA
projections was to compare the performance reached by the
classifiers and to identify whether the data structure was suitable
for the proposed classifiers architecture. Indeed, there exists more
than one group of features that reported high success rates at
classifying the HRV records. Amongst all the classifiers, the
maximum overall performance was reached by the polynomial
kernel SVM with overall sensitivity, specificity and accuracy of
90.91%, 93.18% and 92.05%, respectively.
doi:10.1371/journal.pone.0017060.t005
Non-linear + Multi-resolution
26.67
72.73
68.18
RBFNN
Statistical + Spectral
12
Se (%)
66.67
Classifier
MLP
Features
Np (%)
90.16
87.80
94.44
78.95
83.33
70.00
68.89
76.00
56.00
64.29
Pp (%)
84.51
82.98
80.77
100.00
100.00
72.58
69.77
84.21
80.00
62.50
Ac (%)
87.12
85.23
86.36
86.67
90.00
71.21
69.32
79.55
60.00
63.33
February 2011 | Volume 6 | Issue 2 | e17060
statistical features of the NN intervals are related to: Complete
heart block (CHB), left bundle branch block (LBBB) and ischemic
cardiopaty. On the other hand, the long variations on statistical
features of the NN intervals are related to: Premature ventricular
contractions (PVCs), sinus syndrome (SSS) and atrial fibrillation
(AF). There are several modifications on AR spectrums that can be
noticed by the calculation of the power in frequency bands. Short
variations of HRV segments usually lead to high VLF and LF
bands power. Conversely, long variations of HRV segments,
usually leads to higher HF band power. As cardiovascular risk is
highly related to variations on statistical and spectral components,
one of the major disadvantages of these methods is the linearity
and consequently, their poor suitability for highly non-stationary
HRV dynamics and, certainly, the NN intervals fluctuation by
nervous mechanisms. The Fourier transform techniques (frequency domain methods such as Welch periodogram or AR spectrum
as well) resolve the time domain signal into complex exponential
functions, along with information about their phase shift measured
with respect to a specific reference instant. Here the frequency
components extend from {? to ? in the time scale. That is,
even finite length signals are expressed as the sum of frequency
components of infinite duration. Besides, the phase angle, being a
modular measure, fails to provide the exact location of an ‘event’
along the time scale. This is a major limitation of the Fourier
transform approach [2]. Thereby, considering the cardiovascular
system as nonlinear in nature, can lead to a better understanding
of its dynamics.
The patterns in HRV are directly related to the Poincaré map
patterns in visual assessment. In the case of short fluctuations,
HRV segments are not much dispersed and torpedo shapes [62]
are predominant in many cases. In the case of long fluctuations,
HRV segments appear to be very dispersed forming complex-like
and fan-like return map shapes. All these Poincaré plot patterns
are directly related with cardiovascular risk [66].
One of the main contributions of this work is the prognosis of
cardiovascular risk in a general way, not only for specific cardiac
diseases or the prediction of specific cardiac episodes like other
publications have shown [1,8,13,14,16,18,19,20,21,29,55,61].
Moreover, it has been confirmed that multi-resolution and nonlinear analysis are much more suitable for the assessment and
prognosis of cardiovascular risk than statistical and spectral
classical analysis. KS significance tests also confirmed that those
features lead to higher statistical significance levels (pv0:001 for
83.33
81.82
77.27
100.00
100.00
74.24
70.45
86.36
93.33
60.00
Sp (%)
Table 5. MLP, RBF neural networks and linear SVM (C~1), SVM with polynomial kernel (C~1; c~4) and SVM with RBF kernel
(C~1; c~3) classifications using classical and non-linear/multi-resolution features of the HRV records from normal (N) and
cardiovascular risk (R) subjects.
HRV for the Prognosis of Cardiovascular Risk
SVM (RBF kernel)
95.45
81.82
81.82
SVM (Polynomial kernel)
SVM (RBF kernel)
13
95.45
81.82
SVM (Linear)
PLoS ONE | www.plosone.org
the case of ApEn and SmEn features). We developed a method
that combines statistical, spectral, multi-resolution and non-linear
features as well as ANN and SVM schemes for the prognosis of
cardiovascular risk. Exactly 90 HRV records were analyzed, 60 of
them were used to train and 30 to test each classification scheme.
From the classification schemes, MLP provided the best
classification rates for the prognosis of cardiovascular risk, with
an area under the ROC curve equal to 0.9800. On the other
hand, schemes such as the RBF network and SVMs showed
relatively high classification performances too. Another remarkable finding is the improvement of the classification rates of SVM
using all the extracted features (not only the features selected by
KS significance tests) and PCA, which can be attributable to the
computation of the decision surface and the apparent SVM bias
towards the positive and negative cases. Furthermore, as expected,
ANN schemes often presented overfitting using all those features.
The comparison between the performances of all the implemented
classifiers was limited to sensitivity, specificity, positive predictive
value, negative predictive value and accuracy as a consequence of
ROC curves limitations, i.e., usually its contributions become
cumbersome when a comparison between different classifiers is
needed. Besides, their transformation to objective values is usually
limited to the calculation of the area under the curve [67].
According to findings reported in the literature [5], breath rate
modifies the fluctuation of the NN intervals in a HRV sample
record, i.e., there is an evident modification of the HRV when the
analyzed subject is breathing at different frequencies (e.g., 6
breaths/min or 12 breaths/min) provided that the rhythmic
fluctuations can be larger or shorter given the regulation
84.00
84.00
81.82
N/A
78.57
84.00
84.00
87.50
N/A
75.00
93.65
91.11
90.91
61.90
80.00
86.00
81.58
73.68
60.87
64.71
Np (%)
94.74
94.74
81.82
N/A
75.00
94.74
94.74
95.00
N/A
100.00
89.86
93.02
90.91
77.78
80.00
71.95
74.00
68.00
85.71
69.23
Pp (%)
88.64
88.64
81.82
N/A
76.67
88.64
88.64
90.91
N/A
83.33
91.67
92.05
90.91
66.67
80.00
77.27
77.27
70.45
66.67
66.67
Ac (%)
February 2011 | Volume 6 | Issue 2 | e17060
mechanisms and the RSA dynamics as an effect of the activity
of neural oscillators. As a direct consequence, the experimental
results from the comparison between records at different breathing
rates would be invalid. Thus, another contribution of this work is
the consideration of the breathing rate as an additional variable for
the assessment of HRV; this feature was included in every record
taken for the HRV database reported in this work.
Besides the strong considerations of the breathing rate and the
breathing signal, our computational implementation allows
working with the ECG and HRV signals for medical analysis.
There are new possibilities of analysis that commercial and
conventional HRV analysis software [55] have not yet considered;
additionally, there is the possibility of performing an automatic
prognosis using a trained ANN scheme embedded in the program.
The main objective of this module is to give support to the
specialist criteria on the HRV assessment. This computational
application needs strong validation and medical feedback, which
will be a topic for future research.
According to the results of this study, we strongly suggest
working with multi-resolution and non-linear analysis in order to
achieve more reliable cardiovascular risk prognosis, especially for
classification schemes such as ANNs and SVMs. It is evident that a
nonlinear deterministic approach is more appropriate to describe
more complex phenomena, indicating that apparently erratic
behavior can be generated even by a simple deterministic system
with nonlinear structure. In general terms, the fluctuations of
heartbeats during normal sinus rhythm could be partially
attributed to deterministic chaos, and a decrease in this type of
nonlinear variability could be observed in different cardiovascular
81.82
N/A
73.33
95.45
N/A
SVM (RBF kernel)
95.45
80.00
81.82
SVM (Polynomial kernel)
N/A
95.45
MLP
81.82
SVM (Linear)
100.00
89.39
93.18
90.91
86.67
80.00
65.15
70.45
63.64
93.33
73.33
Sp (%)
RBFNN*
N/A
86.36
RBFNN*
66.67
93.94
SVM (Polynomial kernel)
MLP
90.91
90.91
SVM (Linear)
46.67
RBFNN
89.39
SVM (RBF kernel)
80.00
84.09
SVM (Polynomial kernel)
MLP
SVM (Linear)
*The results for this classifier were not reported due to their poorness.
doi:10.1371/journal.pone.0017060.t006
10
5
4
40.00
77.27
RBFNN
60.00
MLP
3
Se (%)
Classifier
Number of PCA
Projections
Table 6. MLP, RBF neural networks and linear SVM (C~1), SVM with polynomial kernel (C~1; c~4) and SVM with RBF kernel
(C~1; c~3) classifications using the projections of the features from PCA analysis of the HRV records from normal (N) and
cardiovascular risk (R) subjects.
HRV for the Prognosis of Cardiovascular Risk
PLoS ONE | www.plosone.org
1. Mohammadzadeh Asl B, Kamaledin Setarehdan S, Mohebbi M (2008) Support
vector machine-based arrhythmia classification using reduced features of heart
rate variability signal. Artificial Intelligence in Medicine 44: 51–64.
2. Rajendra Acharya U, Paul Joseph K, Kannathal N, Min Lim C, Suri JS (2006)
Heart rate variability: a review. Med Bio Eng Comput 44: 1031–1051.
3. Bezerianos A, Papadimitriou S, Alexopoulos D (1999) Radial basis function
neural networks for the characterization of heart rate variability dynamics.
Artificial Intelligence in Medicine 15: 215–234.
4. Belova NY, Mihaylov SV, Piryova BG (2007) Wavelet transform: A better
approach for the evalation of instantaneous changes in heart rate variability.
Autonomic Neuroscience: Basic and Clinical 131: 107–122.
5. Lopes P, White J (2006) Heart rate variability: Measurement methods and
practical implications. In: Maud PJ, Foster C, eds. Physiological Assessment of
Human Fitness. pp 39–61.
6. Garcı́a-González MA (1998) Estudio de la Variabilidad del ritmo cardı́aco
mediante técnicas estadı́sticas, espectrales y no-lineales. PhD Thesis, Universitat
Politècnica De Catalunya.
7. Montano N, Porta A, Cogliati C, Costantino G, Tobaldini E, et al. (2008) Heart
rate variability explored in the frequency domain: A tool to investigate the link
between heart and behavior. Neuroscience and Biobehavioral Reviews. In Press.
References
This work has been focused on 4 major aspects: (1) The HRV
database conformation integrating the breathing signal; (2)
integrating the breathing frequency to the analysis of HRV
records; (3) the analysis of statistical, spectral, multi-resolution and
non-linear features linked with classification schemes for the
prognosis of cardiovascular risk as well as the confirmation of the
properties of these features as they are reported in the literature;
(4) a brief illustration of the software implementation of advanced
HRV analysis integrating an automatic prognosis tool using a
standard 5 min ECG record.
This work has developed a method for the cardiovascular risk
prognosis using statistical, spectral, multi-resolution and non-linear
features extracted from HRV data. The database used in this work
was recorded using a proper medical protocol and contains a total
of 90 HRV and breathing signal records from normal and
cardiovascular risk subjects. The suitability of each HRV analysis
feature has been shown by using KS-tests and neural and support
vector classifiers. Short-term HRV features are highly useful for
the prognosis of cardiovascular risk; nonetheless, long-term
features can be useful for increasing the positive predictive value
of the proposed classifiers.
In addition, it has become obvious that the performance
reported by the different feature selection strategies depends
largely on the pattern recognition scheme implemented to classify
the data. For the short-term HRV features analyzed in this work,
there were two main observed effects: First, the ANN schemes
were more suitable for the KS test-based feature selection as well
as for feature spaces of low dimension; and second, SVM schemes
were more suitable for the PCA-based feature selection as well as
to high dimensional feature spaces. Nonetheless, as the PCA
projected features improved by far the performance of SVM
classifiers, their statistical significance showed influential decrease
that affected negatively the performance of the ANN classifiers.
Breathing signal and breathing frequency were employed in this
work for the analysis of HRV, given the findings about RSA
phenomena on the modification of the fluctuations of NN intervals
at different breathing frequencies [5]. All the HRV measurements
in this work were done approximately at 12 breaths/min for every
single subject in the database; this was guaranteed by the
procedures established in the medical protocol that we designed
Conclusions
diseases [58]. Further investigation is needed for the incorporation
of chaotic dynamics and fractals to the analysis of HRV for the
prognosis of cardiovascular risk proposed in this study.
14
February 2011 | Volume 6 | Issue 2 | e17060
8. Acharyaa UR, Bhatb PS, Iyengarc SS, Raod A, Dua S (2003) Classification of
heart rate data using artificial neural network and fuzzy equivalence relation.
Pattern Recognition 36: 61–68.
9. Kuss O, Schumann B, Kluttig A, Greiser KH, Haerting J (2008) Time domain
parameters can be estimated with less statistical error than frequency domain
parameters in the analysis of heart rate variability. Journal of Electrocardiology
41: 287–291.
10. Urbanowicza K, Zebrowski J, Baranowskic JR, Holysta JA (2007) How random
is your heart beat?. Physica A 384: 439–447.
11. Bilgin S, Çolak OH, Koklukaya E, Ari N (2008) Efficient solution for frequency
band decomposition problem using wavelet packet in HRV. Digital Signal
Processing 18: 892–899.
12. Task force of the European society of cardiology and the North American
society of pacing and electrophysiology (1996) Heart Rate Variability Standards
of Measurement, Physiological Interpretation, and Clinical Use. Circulation 93:
1043–1065.
13. D’Addio G, Acanfora D, Pinna GD, Maestri R, Furgi G, et al. (1998)
Reproducibility of Short -and Long-Term Poincaré Plot Parameters Compared
with Frequency-Domain HRV Indexes in Congestive Heart Failure. Computers
in Cardiology 25: 381–384.
Conceived and designed the experiments: JFR ELE. Performed the
experiments: JFR ELE PCE WAR. Analyzed the data: JFR ELE.
Contributed reagents/materials/analysis tools: DFR. Wrote the paper:
JFR DFR. Performed the feature extraction methods: JRV. Gave
important feedback on statistical procedures: ELE. Made the PCA-based
feature selection and classification experiments: JRV. Gave important
feedback and medical validation of the reported results: WAR. Gave
important scientific background: DFR. Suggested suitable classification
schemes and architectures to overcome the classification problems: DFR.
Gave important feedback for the enhancement of the paper: DFR PCE.
Author Contributions
We acknowledge the support of the Occupational Health Office
(Universidad Autonoma de Occidente), the Coomeva IPS S. A. experts
and the patients which formed part of this study. We also acknowledge the
special reviewing collaboration of Dr. Odelia Schwartz (Albert Einstein
College of Medicine, NY, USA). Finally, we are grateful for the important
suggestions given by anonymous reviewers.
Acknowledgments
algorithm.
(DOC)
Appendix S2 Detailed description of SVM learning
sures calculation.
(DOC)
Appendix S1 Detailed description of complexity mea-
Supporting Information
to carry out this research; moreover, the acquired breathing
signals confirmed it.
Finally, HRV signal –as both a traditional and non-linear signal
in nature– has been an important predictor of cardiovascular risk.
Furthermore, this work represents an important step in comprehensively understanding such dynamics, and more importantly, on
the prognosis of cardiovascular risk as a stated goal established by
Chattipakorn et al. (2007) and a plethora of researchers and
clinicians: ‘‘It is necessary to understand the mechanisms of HRV
components and improve their sensitivity and specificity as a
prognostic marker’’ [68]. Furthermore, the combination of the
methods reported in this work and other ECG and physiological
parameters is expected to lead to a solution for the prognosis of
cardiovascular mortality at reasonable cost-effectiveness.
HRV for the Prognosis of Cardiovascular Risk
PLoS ONE | www.plosone.org
14. D’Addio G, Pinna GD, Maestri R, Acanfora D, Picone C, et al. (1999)
Correlation Between Power-law Behavior and Poincaré Plots of Heart Rate
Variability in Congestive Heart Failure Patients. Computers in Caridiology 26:
611–614.
15. Thong T (2007) Geometric Measures of Poincaré Plots for the Detection of
Small Sympathovagal Shift. Proceedings of the 29th Annual International
Conference of the IEEE EMBS Cité Internationale. pp 4641–4644.
16. Woo MA, Stevenson WG, Moser DK, Trelease RB, Harper RM (1992) Patterns
of beat-to-beat heart rate variability in advanced heart failure. Am Heart J 123:
704–710.
17. Brennan M, Planiswami M, Kamen P (2001) Do existing measures of Poincare
plot geometry reflect nonlinear features of heart rate variability?. IEEE Trans
Biomed Eng 48: 1342–1347.
18. Rueda OL, López EJ, Vargas CA, Delgado MB, Murillo CA (2007) La
variabilidad de la frecuencia cardı́aca como factor pronóstico de mortalidad del
infarto del miocardio: Revisión sistemática de estudios observacionales. Médica
Sanitas 22.
19. Malik M, Farrell T, Cripps TR, Camm AJ (1989) Heart rate variability in
relation to prognosis after myocardial infarction: selection of optimal processing
techniques. Eur Heart J 10: 1060–1074.
20. Chesnokov YV (2008) Complexity and spectral análisis of the heart rate
variability dynamics for distant prediction of paroxysmal atrial fibrillation with
artificial intelligence methods. Artificial Intelligence in Medicine 43: 151–165.
21. Ísler Y, Kuntalp M (2007) Combining classical HRV indexes with wavelet
entropy measures improves to performance in diagnosing congestive heart
failure. Computers in Biology and Medicine 37: 1502–1510.
22. Wolf MM, Varigos GA, Hunt D, Sloman JG (1978) Sinus arrhythmia in acute
myocardial infarction. Med J Aust. pp 52–53.
23. Penaz J, Roukenz J, Van Der Waal HJ (1968) Spectral Analysis of Some
Spontaneous Rhythms in the Circulation.. Leipzig Germany: Biokybernetik,
Karl Marx University. pp 223–241.
24. Pomeranz M, Macaulay R, Caudill M, Ma K, Kutz I, et al. (1985) Assessment of
autonomic function in humans by heart rate spectral analysis. Am J Physiol 248:
H151–H153.
25. Akselrod S, Gordon D, Ubel FA, Shannon DC, Barger AC, et al. (1981) Power
spectrum analysis of heart rate fluctuation: a quantitative probe of beat to beat
cardiovascular control. Science. pp 220–222.
26. Akselrod S, Lishner M, Oz O, Bernheim J, Ravid M (1987) Spectral analysis of
fluctuations in heart rate: an objective evaluation of autonomic nervous control
in chronic renal failure. Nephron 45: 202–206.
27. Berntson GG, Bigger JT, Jr., Eckberg DL, Grossman P, Kaufmann PG, et al.
(1997) Heart rate variability: Origins, methods, and interpretive caveats.
Psychophysiol 34: 623–648.
28. Broadman A, Schlindwein FS, Rocha AP, Leite A (2002) A study on the
optimum order of autoregressive models for heart rate variability. Physiol Meas
23: 324–336.
29. Pagani M, Lombardi F, Guzzetti S, Rimoldi O, Furlan R, et al. (1986) Power
spectral analysis of heart rate and arterial pressure variabilities as a marker of
sympathovagal interaction in man and conscious dog. Circ Res 58: 178–193.
30. Pal GK, Pal P, Nanda N, Amudharaj D, Karthik S (2009) Spectral analysis of
heart rate variability (HRV) may predict the future development of essential
hypertension. Medical Hypotheses 72: 183–185.
31. Ravichandran IT, Ramasuhha Reddy I, Avudainayagam A (2003) Estimation
and Power Spectral Analysis of Heart Instantaneous Frequency (HIF) - A
Wavelet Approach. Bio-Signal Processing TENCON. pp 223–226.
32. Fukuda O, Nagata Y, Homma K, Tsuji T (2001) Evaluation of heart rate
variability by using wavelet transform and a recurrent neural network.
Proceedings of the 23rd Annual EMBS International Conference. pp 1769–1772.
33. Lerma C, Infante O, Perez-Grovas H, Jose MV (2003) Poincaré plot indexes of
heart rate variability capture dynamic adaptations after haemodialysis in chronic
renal failure patients. Clin Physiol & Func Im 23: 72–80.
34. Marciano F, Migaux ML, Acanfora D, Furgi G, Rengdf F (1994) Quantification
of Poincaré Maps for the Evaluation of Heart Rate Variability. Computers in
Cardiology. pp 577–580.
35. Singh D, Vinod K (2005) Effect of RR Segment Duration on Short-Term HRV
Assessment Using Poincaré Plot. Proceedings of ICISIP. pp 430–434.
36. Brennan M, Palaniswami M, Kamen P (2002) Poincaré plot interpretation using
a physiological model of HRV based on a network of oscillators. Am J Physiol
Heart Circ Physiol 283: 1873–1886.
37. Hnatkova K, Copie X, Staunton A, Malik M (1995) Numeric processing of
Lorenz plots of R–R intervals from long-term ECGs. Comparison with timedomain measures of heart rate variability for risk stratification after myocardial
infarction. J Electrocardiol 28: 74–80.
38. Wolf A, Swift JB, Swinney HL, Vastano JA (1985) Determining Lyapunov
exponents from a time series. Physica 16D: 285–317.
39. Kobayashi M, Musha T (1982) 1/f fluctuation of heartbeat period. IEEE Trans
Biomed Eng 29: 456–257.
15
February 2011 | Volume 6 | Issue 2 | e17060
40. Goldberger AL (1996) Non-linear dynamics for clinicians: chaos theory, fractals,
and complexity at the bedside, Lancet 347: 1312–1314.
41. Peng CK, Havlin S, Hausdorf JM, Mietus JE, Stanley HE, et al. (1995) Fractal
mechanisms and heart rate dynamics. Long-range correlations and their
breakdown with disease. J Electrocardiol 28: 59–65.
42. Hirsh JA, Bishop B (1981) Respiratory sinus arrhythmia in humans: how
breathing pattern modulates heart rate. Am J Physiol Heart Circ Physiol 241:
H620–H629.
43. Piepoli M, Sleight P, Leuzzi S, Valle F, Spadacini G, et al. (1997) Origin of
Respiratory Sinus Arrhythmia in Conscious Humans. Circulation 95:
1813–1821.
44. Katona PG, Jih F (1975) Respiratory sinus arrhythmia: noninvasive measure of
parasympathetic cardiac control. J Appl Physiol 39: 801–805.
45. Lloyd-Jones D, Adams R, Carnethon M, de Simone G, Ferguson TB, et al.
(2009) Heart disease and stroke statistics – 2009 update: A report from the
American Heart Association Statistics Committee and Stroke Statistics
Subcommittee. Circulation 119: e21–e181.
46. Beltrán-Bohórquez JR (2010) Guı́as colombianas de cardiologı́a. Sı́ndrome
coronario agudo con elevación del ST. Revista Colombiana de Cardiologı́a 17:
111–275.
47. Gallagher D, Terenzi T, de Meersman R (1992) Heart rate variability in
smokers, sedentary and aerobically fit individuals. Clinical Autonomic Research
2: 383–387.
48. Liao D, Carnethon M, Evans GW, Cascio WE, Heiss G (2002) Lower heart rate
variability is associated with the development of coronary heart disease in
individuals with diabetes: the atherosclerosis risk in communities (ARIC) study.
Diabetes 51: 3524–3531.
49. Williams JS, Brown SM, Conlin PR (2009) Blood-pressure Measurement.
N Engl J Med 360: e6.
50. Motivala SJ, Hurwitz BE, Lagreca AM, Llabre MM, Marks JB, et al. (1999)
Aberrant parasympathetic and Hemodynamic function distinguishes a subgroup
of psyologically distressed individuals with asymptomatic type-I diabetes mellitus,
International Journal of Behavioral Medicine 6: 78–94.
51. Gutierrez A, Lara M, Hernandez PR (2005) Evaluación de un detector de
complejo QRS basado en la wavelet de Haar usando las bases de datos MITBIH de arritmias y Europea del segmento ST y de la onda T. Computación y
Sistemas 8: 293–302.
52. Lippman N, Stein KM, Lerman BB (1994) Comparison of methods for removal
of ectopy in measurement of heart rate variability. Am J Physiol Heart Circ
Physiol 267: H411–H418.
53. Rosner B (1983) Percentage Points for a Generalized ESD Many-Outlier
Procedure. Technometrics 25: 165–172.
54. Ramirez-Villegas JF, Lam-Espinosa E (2009) Análisis de la variabilidad de la
frecuencia cardı́aca integrando la señal de la frecuencia respiratoria. B. Eng.
Thesis, Universidad Autonoma de Occidente, Cali, Colombia.
55. Niskanen JP, Tarvainen MP, Ranta-Aho PO, Karjalainen PA (2002) Software
for advanced HRV analysis. Computer Methods and Programs in Biomedicine.
In Press.
56. Rosso OA, Blanco S, Yordanova J, Kolev V, Figliola A, et al. (2001) Wavelet
entropy: a new tool for analysis of short duration brain electrical signals. J
Neurosci Methods 105: 65–75.
57. Blanco S, Figliola A, Quian QR, Rosso OA, Serrano E (1998) Time-Frequency
analysis of electroencephalogram series (iii): Information transfer function and
wavelet packets. Phys Rev E 57: 932–940.
58. Lombardi F (2000) Chaos Theory, Heart Rate Variability, and Arrhythmic
Mortality. Circulation 101: 8–10.
59. Meiss JD (2007) Differential dynamical systems. Society for Industrial and
Applied Mathematics (SIAM). 434 p.
60. Sano M, Sawada Y (1985) Measurement of the Lyapunov spectrum from a
chaotic time series. Physical Review Letters 55: 1082–1085.
61. Korpelainen JT, Sotaniemi KA, Mäkikallio A, Huikuri HV, Myllylä VV (1999)
Dynamic Behavior of Heart Rate in Ischemic Stroke. Stroke 30: 1008–1013.
62. Pincus SM (2001) Assessing Serial Irregularity and its Implications for Health.
Annals of the New York Academy of Sciences 954: 245–267.
63. Rencher AC (2002) Methods of Multivariate Analysis. Wiley Series in
Probability and Statistics. 738 p.
64. Haykin S (1999) Neural Networks: A Comprehensive Foundation. Prentice Hall,
inc. 823 p.
65. Wang L, Liu B, Wan C (2005) Classification using support vector machines with
graded resolution. Proceedings of the international conference on granular
computing. pp 666–670.
66. Woo MA, Stevenson WG, Moser DK, Trelease RB, Harper RM (1992) Patterns
of beat-to-beat heart rate variability in advanced heart failure. Am Heart J 123:
704–710.
67. Fawcett T (2006) An introduction to ROC analysis. Pat Rec Lett 27: 861–874.
68. Chattipakorn N, Incharoen T, Kanlop N, Chattipakorn S (2007) Heart rate
variability in myocardial infarction and heart failure. International Journal of
Cardiology 120: 289–296.
HRV for the Prognosis of Cardiovascular Risk
|
doi: 10.1111/j.1365-2796.2010.02321.x
J. M. Huston1 & K. J. Tracey2
Inflammation is induced by factors that are exogenous (e.g. pathogens and microbial products) and
endogenous [e.g. High Mobility Group Box-1
(HMGB1) released from injured cells] to the host [1,
2]. These inducing agents interact with genome encoded pattern recognition receptors expressed on
Amongst the leading causes of morbidity and mortality in Western societies are heart disease, cancer,
stroke, diabetes and sepsis. Recent advances in
immunology reveal a significant pathogenic role for
inflammation in the development and progression of
these disorders. Inflammation accelerates deposition
of atherosclerotic plaques leading to myocardial and
cerebral infarction; mediates insulin resistance;
stimulates tumour growth; and causes organ damage
in lethal sepsis. Knowledge of these mechanisms has
elevated the importance of understanding both the
molecular basis of inflammation and the regulatory
systems that keep it in check during health.
Introduction
Biological therapeutics targeting TNF, IL-1 and IL-6
are widely used for treatment of rheumatoid arthritis,
inflammatory bowel disease and a growing list of
other syndromes, often with remarkable success.
Now advances in neuroscience have collided with this
therapeutic approach, perhaps rendering possible
Abstract. Huston JM, Tracey KJ (Department of
Surgery, Division of General Surgery, Trauma,
Surgical Critical Care, and Burns, Stony Brook
University Medical Center, Health Sciences Center,
Stony Brook; and Laboratory of Biomedical Science, The Feinstein Institute for Medical Research,
Manhasset; NY, USA). The pulse of inflammation:
heart rate variability, the cholinergic anti-inflammatory pathway, and implications for therapy (Key
Symposium). J Intern Med 2011; 269: 45–53.
ª 2010 The Association for the Publication of the Journal of Internal Medicine
45
monocytes, macrophages and other cells of the innate immune system [3]. These receptor families,
including the toll-like receptors and NOD-like receptors, transduce intracellular signals leading to the
production and release of cytokines, eicosanoids and
other inflammatory molecules that directly mediate
cellular responses causing inflammation [3]. The
cardinal signs of inflammation, including pain,
erythema, oedema, warmth and loss of function, can
be produced by exposing tissues to inflammatory
cytokines. Nonresolving or persisting exposure to
cytokine damages tissue, impairs organ function and
can be lethal. Biological therapeutics that specifically
inhibit cytokine mediators of inflammation are widely
used to treat arthritis, colitis, psoriasis and a growing
list of other disabling illnesses. Therefore, the
dangers of uncontrolled inflammation are inherent
to the molecular activity of cytokines themselves, and
maintenance of health requires tight control over the
steps leading to the production and release of cytokines.
Keywords: heart rate variability, inflammation, neuroimmunology, therapeutics, vagus nerve stimulation.
the development of nerve stimulators to inhibit cytokines. Action potentials transmitted in the vagus
nerve culminate in the release of acetylcholine that
blocks cytokine production by cells expressing acetylcholine receptors. The molecular mechanism of
this cholinergic anti-inflammatory pathway is attributable to signal transduction by the nicotinic alpha 7
acetylcholine receptor subunit, a regulator of the
intracellular signals that control cytokine transcription and translation. Favourable preclinical data
support the possibility that nerve stimulators may be
added to the future therapeutic armamentarium,
possibly replacing some drugs to inhibit cytokines.
Department of Surgery, Division of General Surgery, Trauma, Surgical Critical Care, and Burns, Stony Brook University Medical Center, Stony
Brook; and 2Laboratory of Biomedical Science, The Feinstein Institute for Medical Research, Manhasset; NY, USA
1
The pulse of inflammation: heart rate variability, the
cholinergic anti-inflammatory pathway and implications for
therapy
Key Symposium
|
46
Under basal conditions, the cholinergic anti-inflammatory pathway exerts a tonic, inhibitory influence
on innate immune responses to infection and tissue
injury. Interrupting this pathway, by either severing
the vagus nerves, or by knocking out the a7 gene
(CHRNA7), produces an inflammatory phenotype
characterized by exaggerated responses to bacterial
products and injury [5, 8]. For example, electrical
stimulation of the cervical vagus nerve in wild-type
The ‘cholinergic anti-inflammatory pathway’ is the
descending, or motor arc of the inflammatory reflex
(Fig. 1) [6]. It is comprised of vagus nerve signals leading to acetylcholine- dependent interaction with the
alpha 7 nicotinic acetylcholine receptor subunit
(a7nAChR) on monocytes and macrophages, resulting in reduced cytokine production [8, 22]. The cholinergic anti-inflammatory pathway can be activated
experimentally by electrical or mechanical vagus
nerve stimulation, or through administration of a7
agonists, to inhibit inflammatory cytokine production, prevent tissue injury and improve survival in
multiple experimental models of systemic inflammation and sepsis [4, 5, 9–14, 16–18]. Mononuclear cells
express muscarinic and nicotinic acetylcholine
receptors on their cell surface and experimental evidence suggests that a7 is required for the regulation
of cytokine release by acetylcholine [8]. This pathway
is unique in the context of vagus nerve signalling because in contrast to classical parasympathetic nervous system signalling through muscarinic receptors
on target organs, this circuit requires signalling
through a specific nicotinic receptor subunit.
ª 2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 45–53
To date, the inflammatory reflex is the best- studied
anti-inflammatory neural circuit, but undoubtedly,
as expertise in this field continues to improve, it is
likely that other pathways will be elucidated [21].
These findings now raise some fundamental ques-
In the late 1990’s, whilst studying CNI-1493, an
inhibitor of the p38 MAP kinase developed by one of
us (KJT) as an anti-inflammatory molecule, we discovered an anti-inflammatory neural circuit [4, 5].
Termed ‘the inflammatory reflex’, this neurological
mechanism involves the vagus nerve, which can
sense peripheral inflammation and transmit action
potentials from the periphery to the brain stem [6].
This in turn leads to the generation of action potentials in the descending vagus nerve that are relayed to
the spleen, where pro-inflammatory cytokine production is inhibited [7]. The molecular basis of this
anti-inflammatory circuit, termed the cholinergic
anti-inflammatory pathway, includes the neurotransmitter acetylcholine interacting with the alpha7 nicotinic acetylcholine receptor subunit expressed
on monocytes, macrophages and other cytokine
producing cells [8]. Signal transduction through
this receptor inhibits cytokine release, suppresses
inflammation and confers protection against tissue
damage in polymicrobial sepsis, arthritis, colitis, diabetes, atherosclerosis, ischaemia-reperfusion injury,
pancreatitis, myocardial ischaemia and haemorrhagic shock [5, 9–20].
The inflammatory reflex suppresses inflammation
The cholinergic anti-inflammatory pathway
tions about understanding clinical disease pathogenesis. Pre-eminent amongst these is whether it is possible to measure activity within the inflammatory
reflex in humans in order to understand or predict
the risk of uncontrolled inflammation. Here, we review available data addressing this question and focus on the potentially informative data regarding
measurements of vagus nerve signalling. We discuss
the concept that measuring changes in vagus nerve
activity may provide useful information about realtime activity of the inflammatory reflex in patients
with inflammatory diseases. We also examine the evidence that measuring underlying vagus nerve activity may have clinical utility for predicting damage
from ongoing inflammation and review the potential
implications of utilizing this diagnostic information
to then guide therapeutic modulation of immune responses.
Key Symposium: The pulse of inflammation: implications for therapy
Factors that trigger inflammation also enhance the
activity of anti-inflammatory pathways, which function to counter-balance inflammation. This concurrent activation of pro- and anti-inflammatory mechanisms is analogous to other homoeostatic systems,
such as coagulation and fibrinolysis, which act in
concert to coordinate haemostasis during haemorrhage. Anti-inflammatory pathways that exert critical
roles in suppressing cytokine production include the
hypothalamic-pituitary-adrenal axis, which culminates in glucocorticoid release; release of neutralizing
soluble cytokine receptors; and production of antiinflammatory cytokines (e.g. IL-10 and TGFb). Interruption of these pathways (e.g. adrenalectomy), or
failure of their function (genetic deficiency of IL-10)
leads to excessive inflammation. This knowledge has
enabled the development of novel therapies that
suppress inflammation in humans by either directly
targeting the activities of cytokines (e.g. anti-TNF
antibodies), or by preventing cytokine release (glucocorticoids).
J. M. Huston & K. J. Tracey
|
47
In addition to regulating cytokine release, the cholinergic anti-inflammatory pathway modulates the
expression of activation markers on circulating leukocytes that transit the spleen. Vagus nerve stimulation down-regulates neutrophil activation by attenuating expression of CD11b, a surface molecule
required for cell adhesion and chemotaxis [28]. The
mechanism of CD11b suppression requires F-actin
polymerization, the rate-limiting step for CD11b surface expression. Using a carrageenan air pouch
ª 2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 45–53
Fig. 1 Action potentials transiting the vagus nerve synapse
in the coeliac ganglion, the origin of the splenic nerve. The
splenic nerve controls lymphocytes in the spleen, which can
produce acetylcholine that interacts with a7 nicotinic acetylcholine receptors expressed on cytokine producing macrophages. Intracellular signal transduction through this receptor inhibits the activity of nuclear factor-jB to suppress
cytokine production. Nerve stimulators can provide an identical signal to initiate the anti-inflammatory pathway, an approach that reverse signs and symptoms in preclinical disease models of arthritis, inflammatory bowel disease,
ischaemia-reperfusion injury, heart failure, pancreatitis, sepsis and other syndromes .
Organs of the reticuloendothelial system, including
the lungs, liver and spleen, contain innate immune
cells that mediate the immediate, early response to
pathogens and injury. During endotoxemia, an experimental model of Gram-negative bacterial shock
produced by administration of lipopolysaccharide,
the spleen is the major organ source of systemic TNF
that accumulates in blood [7, 25]. TNF production in
spleen accounts for >90% of the total TNF burden that
reaches the circulation, so it is perhaps not surprising that splenectomy significantly reduces circulating TNF levels in endotoxin-challenged mice. Vagus
nerve stimulation fails to inhibit systemic TNF production in splenectomized mice or in mice following
interruption of either the common coeliac branch of
the abdominal vagus nerve, or the splenic nerve, indicating that cholinergic anti-inflammatory pathway
control of TNF culminates in spleen [7, 26, 27]. As expected by the observation that vagus nerve stimulation suppresses serum TNF, vagus nerve stimulation
significantly reduces TNF synthesis in spleen, an effect that requires a7nAChR. Moreover, administration of selective a7nAChR agonists to splenectomized
mice fails to reduce cytokine levels; rather, this exacerbates pro-inflammatory cytokine production and
increases lethality [7, 13, 25]. These findings indicate
that the spleen is a critical physiological interface between cholinergic anti-inflammatory signalling and
regulation of systemic immune responses.
cholinergic anti-inflammatory pathway. Signal transduction mechanisms involving the a7nAChR are an
area of active study by a number of groups and more
data are needed to fully understand this mechanism.
There is general agreement that the cytokine suppressing signals from a7nAChR are not dependent
upon activation of ion channels, the principal mechanism, by which a7nAChR mediates signalling in neurons. Rather, the current signal transduction model
indicates that receptor ligand interaction activates
JAK-STAT dependent inhibition of the nuclear translocation of nuclear factor (NF)-jB, resulting in decreased transcription of cytokine genes [22–24].
Key Symposium: The pulse of inflammation: implications for therapy
mice reduces pro-inflammatory cytokine production,
but a7nAChR-knockout animals are resistant to
these effects and produce higher levels of cytokines
despite vagus nerve stimulation. Even in the absence
of vagus nerve stimulation, a7nAChR-knockout mice
generate a significantly elevated pro-inflammatory
cytokine response following challenge with endotoxin
giving direct evidence that the a7 receptor is necessary to maintain the tonic inhibitory influence of the
J. M. Huston & K. J. Tracey
|
48
A widely used approach to measuring vagus nerve
activity in humans is based on cardiac physiology.
Measuring vagus nerve activity: Heart rate variability
Neural circuits function reflexively to maintain physiological stability in visceral organs. Each reflex is
comprised of a sensory or afferent arc that detects
environmental and chemical changes. This information is relayed to neural centres in the central nervous
system, which integrate the input and relay neural
signals to an output or motor nucleus. Efferent signals are then relayed by motor neurons travelling to
the innervated organ to produce a response that
maintains an ‘appropriate’ or healthy level of function. The net physiological effect of this reflexive
behaviour is a system output that varies according to
a set point function curve. For example, consider the
control of heart rate: increases in heart rate activate a
reflex circuit that leads to increased activity in the vagus nerve, which slows heart rate and restores homoeostasis. A major unanswered question in clinical
immunology is whether it is possible to record neural
activity in the vagus nerve as a surrogate marker of
activity in the inflammatory reflex to determine the
sensitivity of the innate immune system to inflammation.
Clinical implications of the inflammatory reflex
ª 2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 45–53
The original concept of the inflammatory reflex followed experimental observations that administration
of extremely low quantities of CNI-1493 into the brain
inhibited systemic TNF production during endotoxemia [31]. This effect was not attributable to either
leakage of the drug from the brain into the periphery,
or to stimulation of the hypothalamic-pituitary-adrenal axis. Surprisingly at the time, the systemic cytokine inhibitory actions of intracerebral CNI-1493
were dependent upon the vagus nerve because vagotomy abolished the TNF-suppressing effects of intracerebral administration [9]. This was ultimately explained by the fact that CNI-1493 is a weak agonist of
the M1 class of muscarinic receptors, which stimulate activity within descending forebrain cholinergic
neural pathways that mediate increased efferent va-
Central activation of the cholinergic anti-inflammatory pathway
It is interesting to consider the clinical experience
with splenectomized patients. Even with the availability of effective vaccines, splenectomized patients
are at increased risk for potentially lethal bacterial
infections from the syndrome of overwhelming postsplenectomy sepsis. The pathogenesis of this syndrome is debated, but it has been attributed to inadequate immune responsiveness, particularly to
encapsulated bacteria [30]. Based on these new insights provided by the cholinergic anti-inflammatory
pathway, it is intriguing to consider whether patients
dying of postsplenectomy complications develop
uncontrolled, and ultimately lethal, cytokine responses due to the absence of a functional inflammatory reflex.
gus nerve activity [32, 33]. Intracerebral M1 muscarinic receptors had been implicated in the control of
visceral functions by the vagus nerve, including glycogen synthesis in the liver, pancreatic exocrine
secretion and cardiovascular reflexes [34, 35].
Administration of selective muscarinic receptor agonists into the brain stimulates vagus nerve mediated
suppression of cytokine production during endotoxemia [32, 33]. These effects are directly attributable
to intracerebral signals because peripheral administration of muscarinic receptor agonists that cannot
cross the blood-brain barrier fails to inhibit cytokine
release. There is now widespread interest in studying
the anti-inflammatory effects of clinically approved,
centrally acting acetylcholinesterase inhibitors,
which increase brain acetylcholine levels and enhance M1 signalling. Preclinical studies indicate that
these agents increase the activity of the cholinergic
anti-inflammatory pathway and suppress inflammation in the periphery. It may be possible to exploit this
approach in clinical trials of treating inflammatory
diseases, including rheumatoid arthritis, inflammatory bowel disease and psoriasis because these diseases can be controlled by inhibiting cytokine activity.
Key Symposium: The pulse of inflammation: implications for therapy
model to study recruitment of neutrophils to sites of
local soft tissue inflammation, stimulating the vagus
nerve significantly inhibits neutrophil recruitment
[28, 29]. An intact spleen is required for this response
because vagus nerve stimulation of splenectomized
animals fails to inhibit neutrophil recruitment to the
air pouch [25, 28]. In the absence of exogenous vagus
nerve stimulation, removing the spleen also interferes with endogenous neutrophil trafficking. These
results indicate that vagus nerve signals to spleen
control the activity of circulating immune cells, regulating the ability of these cells to respond to inflammatory stimuli and migrate to local zones of ongoing
inflammation, even when these regions are not innervated directly by the vagus nerve. Together these findings point to a specific, centralized neural pathway
innervating the spleen that is positioned to both suppress inflammatory cytokine production and downregulate the activity of circulating inflammatory cells.
J. M. Huston & K. J. Tracey
|
49
Planned and ongoing clinical studies in patients with
cytokine-mediated diseases, including rheumatoid
arthritis, inflammatory bowel disease, sepsis, psoriasis and depression are providing new insights into
measures of vagus nerve activity as a direct correlate
to cholinergic anti-inflammatory pathway activity.
We recently assessed RR interval variability in rheumatoid arthritis and observed that vagus nerve activity was significantly decreased in patients as com-
ª 2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 45–53
Extensive physiological and pharmacological studies
have examined the neural contributions to the fre-
Frequency domain analysis, which is more widely
used to analyse HRV, utilizes spectral methods to
interpret the RR tachogram. Power spectral density
analysis provides information of how power (variability) distributes as a function of frequency. A mathematical algorithm, Fast fourier transform, generates
spectral (frequency) components that are labelled ultra-low frequency (ULF), very low frequency (VLF),
low frequency (LF) and high frequency (HF) power
components. Power components can be expressed in
absolute values (ms2) or in normalized units, which
are used to represent the relative contribution of each
power component to the total variance (power) in the
recording.
Analysis of the time differences between successive
heartbeats to assess HRV can be accomplished with
reference to time (time domain analysis) or frequency
(frequency domain analysis). The former is based on
the normal-to-normal (NN) interval, or the time difference between successive QRS complexes (RR interval) resulting from sinus node depolarization on a
standard, continuous EKG. Statistical analysis of
measurements of the NN intervals, or those derived
from the differences between NN intervals yields various measures of inter-beat variability. Examples include the standard deviation of the NN interval, i.e.
the square root of variance, the standard deviation of
the average NN interval calculated over short periods
and the square root of the mean squared differences
of successive NN intervals >50 ms.
Measures of HRV have been strongly correlated to
morbidity and mortality from diverse diseases. Early
clinical findings, first observed more than 50 years
ago, revealed that variability in RR intervals predict
the onset of foetal distress before any measurable
changes in absolute heart rate [38–40]. There is now
extensive experience using HRV measures in diverse
disease syndromes and these data indicate that decreased vagus nerve activity is associated with increased morbidity and mortality. These correlations
include increased morbidity and mortality following
cardiac surgery or myocardial infarction, increased
mortality from sepsis and progression or disease
severity in autoimmune diseases, including rheumatoid arthritis, inflammatory bowel disease, systemic
lupus erythematosus and sarcoidosis [41–49]. Prior
to knowledge of the inflammatory reflex, it was
thought that decreased vagus nerve activity in these
cases resulted from neural damage associated with
the underlying diseases. It is now possible to consider
an alternative explanation that decreased vagus
nerve activity and the associated loss of the tonic
inhibitory influence of the cholinergic anti-inflammatory pathway on innate immune responses and cytokine release, may enable significantly enhanced cytokine responses to stimuli that would have been
otherwise harmless in the presence of a functioning
neural circuit.
quency components of HRV. For example, administration of acetylcholine antagonists or vagotomy
down modulates the HF power component and electrical vagus nerve stimulation increases HF power
[37]. These results indicate that the HF power component reflects efferent vagus nerve activity to the sinoatrial node. The interpretation of LF power is less clear,
but most agree that the LF component is a measure of
sympathetic activity or a combination of sympathetic
and parasympathetic activity [37]. The ratio of low-tohigh- frequency spectral power (LF ⁄ HF) has been proposed as an index of sympathetic to parasympathetic
balance of heart rate fluctuation. A consensus has
not been achieved concerning the physiological correlates of VLF and ULF power.
Key Symposium: The pulse of inflammation: implications for therapy
Heart rate is controlled by action potentials transmitted via the vagus nerve to the sinoatrial node of the
heart, where vagus nerve-dependent acetylcholine
release essentially ‘prolongs’ the time to the next
heartbeat, thus slowing the pulse. Measuring the
time between individual hearts beats, as can be
accomplished with software that captures the distance between R waves on the electrocardiogram
(EKG) tracing, provides information about the instantaneous heart rate. These data are then plotted as a
function of time to provide analysis of heart rate variability (HRV), or the dynamic variation of heart rate
under control of the sympathetic and parasympathetic nervous input [36]. Heart rate variability represents the time differences between successive heartbeats (also known as the beat-to-beat interval), and is
synonymous with RR variability, referring to the R
waves on the electrocardiogram corresponding to
ventricular depolarization.
J. M. Huston & K. J. Tracey
50
Sloan et al. [68] recently assessed whether aerobic
exercise training can modulate vagus nerve activity
and whole blood production of TNF. In a study of 61
healthy sedentary subjects (age 20–45 year), those
ª 2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 45–53
Numerous studies have investigated the relationship
between depression, systemic cytokine production
and HRV. Depression is associated with abnormalities in innate and adaptive immune function, including increased production of pro-inflammatory cytokines, decreased production of anti-inflammatory
cytokines and increased expression of surface markers associated with immune cell activation [54–56]. It
is plausible that over-expression of cytokines in the
brain may influence depressive behaviour because
cognitive impairment, behavioural dysfunction and
sickness syndrome effects are mediated by cytokines,
including TNF. Current data are unable to determine
whether the onset of a major depressive episode precedes the development of a dysfunctional immune response, or vice versa. Patients with major depressive
disorder also exhibit decreased HRV and the severity
The role of inflammation in the development and progression of atherosclerosis is another area of significant interest and a number of recent studies have explored the potential relationship between the
inflammatory reflex and atherogenic risk [62–65]. For
example, C-reactive protein (CRP), implicated as an
independent risk factor for cardiovascular mortality
and morbidity, was measured in 678 healthy subjects. CRP levels were significantly inversely related
with vagus nerve activity, as assessed by HRV, thus
providing clinical evidence that vagus nerve activity
may modulate systemic inflammatory responses in
cardiovascular disease [36]. Another large study of
CRP and IL-6 levels was conducted in 682 outpatients with coronary heart disease and vagus nerve
activity was inversely correlated with CRP and IL-6
levels [66]. The relationship between circulating
inflammatory markers has also been studied in
healthy university students, (20.56 ± 0.82 years)
and those subjects in the highest tertile of hs-CRP
had significantly decreased vagus nerve activity [67].
Moreover, vagus nerve activity was inversely correlated with hs-CRP, with the lowest hs-CRP levels observed in the most physically active subjects, who
also had the highest levels of vagus nerve activity.
Thus, loss of the inflammation suppressing activity of
vagus nerve signals may contribute to overproduction of CRP, which in turn is controlled by cytokines,
including IL-1 and IL-6.
Depressed vagus nerve activity has been implicated
in exaggerated inflammation in peripheral organs following brain death, and as expected, HRV decreases
significantly following brain death in rats [53]. Interestingly, before harvesting donor organs, vagus nerve
stimulation significantly decreases cytokine concentrations in serum and reduces the expression of proinflammatory cytokines, E-selectin, IL-1b and ITGA6.
Moreover, assessment of renal function reveals significant improvements in recipients of grafts from donors, who had been subjected to vagus nerve stimulation as compared with unstimulated donor grafts
[53]. These results agree with a direct, contributory
role of impaired vagus nerve signalling in excessive
cytokine release during ischaemia before organ harvesting.
Key Symposium: The pulse of inflammation: implications for therapy
of the impairment correlates with the clinical severity
of depression [57]. Moreover, implantable vagus
nerve stimulators are used in patients with treatment-resistant depression, with improvements observed in a significant percentage of patients [58–60].
It is now interesting to consider whether these observations regarding the relationship between depression and maladaptive immune responses result from
impaired vagus nerve regulation of cytokine release.
It should be possible to design clinical studies to address the question of whether increasing vagus nerve
activity using a nerve stimulator corrects the dysregulated cytokine response and lowers the exposure of
the brain to cytokines that disrupt behaviour. End
point selection of these studies will be is critically
important and should include assays of stimulated
cytokine release (e.g. whole blood endotoxin stimulated cytokine release) because basal cytokine levels
are not well correlated to disease severity [61].
|
pared with healthy controls [50, 51] Moreover, serum
levels of HMGB1, a cytokine that has been implicated
in the pathogenesis of rheumatoid arthritis and other
inflammatory syndromes, are significantly related to
RR interval variability. There was no significant relationship between disease severity and vagus nerve
activity, which is consistent with the hypothesis that
impaired vagus nerve activity, is not the result of
advancing disease. Vagus nerve activity was predictive of the innate immune response to endotoxin and
administration of endotoxin to healthy human subjects revealed a significant correlation of basal highfrequency variability to the magnitude of TNF release
[52]. Together these results support the hypothesis
that diminished vagus nerve signals, which normally
provide an inhibitory influence on cytokine production, which contribute to enhanced or unregulated
production of TNF and other inflammatory mediators.
J. M. Huston & K. J. Tracey
|
51
It is also theoretically possible that monitoring HRV
and vagus nerve activity may prove to be useful as a
long-term measure of inflammation in chronic diseases. Similar to tracking haemoglobin A1c levels in
patients with diabetes, or daily blood pressure monitoring in patients with hypertension, HRV monitoring
could theoretically be developed to monitor the activity of inflammatory risk in these and other cytokinemediated diseases. Noninvasive methods to determine HRV are available and it will be of interest to assess the usefulness of portable monitoring devices at
home that interface with central monitoring stations
to provide online analysis of changes in the inflammatory reflex. Correction of chronic, maladaptive levels
of inflammation using nerve stimulators might prevent the progression of debilitating and deadly diseases, potentially replacing the need for some biological therapeutics.
ª 2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 45–53
As clinical studies of vagus nerve- mediated inflammatory responses expand, it may be possible to dissociate the neural pathways that regulate immunity
from those that regulate other vagus nerve functions.
In animal studies of direct stimulation of the vagus
nerve, it is possible to activate the cholinergic antiinflammatory pathway by delivering an electrical
charge that is below the threshold required to significantly change heart rate [13]. Thus, the neural tracts
descending in the vagus nerve to modulate immune
responses function at a lower firing threshold than
the cardio-inhibitory fibres. It is likely that there are
anatomical and physiological differences that underlie these responses. For example, cardio-inhibitory
vagus nerve fibres in mammals are B and C fibre
types, which require significantly higher stimulation
intensities to fire as compared with myelinated A-type
fibres, which do not participate in heart rate regulation. These ‘lower threshold’ A signalling fibres may
Are the cardiac and immune regulatory vagus nerve pathways linked?
There is a need to determine whether augmenting vagus nerve activity in patients who are deficient in this
activity will reduce cytokine production and the levels
of other inflammatory factors, including CRP and IL6. This has not been reported in patients with autoimmune or other active inflammatory diseases, but a
study of 183 healthy adults (mean age = 45) revealed
that higher vagus nerve activity is significantly associated with lower production of TNF and IL-6 in endotoxin- stimulated whole blood assays [69]. This association was independent of demographical and
health characteristics, including age, gender, race,
years of education, smoking, hypertension and white
blood cell count. These authors concluded that vagus
nerve activity is inversely related to the activity of
inflammatory mediators, which has potential implications for studying mechanisms linking psychosocial factors to risk for inflammatory diseases.
The relationship between activation thresholds of the
cholinergic anti-inflammatory pathway and regulation of HRV is an area of intensive study and the available clinical evidence indicates that when vagus
nerve activity is deficient, inflammation is excessive.
There are theoretical and practical advantages to
developing devices that can selectively activate the
cholinergic anti-inflammatory pathway without
stimulating cardiac fibres. It may be possible to draw
correlative analysis from measurements of HRV to
identify individuals with reduced vagus nerve
signalling, who are susceptible to tissue damage from
inflammation. Heart rate variability could serve as a
biomarker to identify patients, who may benefit from
pharmacological or electrical stimulation of the
cholinergic anti-inflammatory pathway. As Holter
monitors are used to track changes in heart rhythm,
HRV monitors may one day provide indices of diminished or enhanced vagus anti-inflammatory activity.
During therapy for inflammation, it may be possible
to measure the physiological level of exogenous cholinergic stimulation delivered to each patient and to
modulate the delivery of therapy by altering voltage,
pulse and time in order to tailor the treatment to the
individual, based on changes in HRV. Autoimmune
diseases are characterized by waxing and waning
clinical episodes and HRV measurements may one day be used to predict impending relapse by revealing
declining activity in the cholinergic anti-inflammatory pathway, thus signalling the need for additional
treatment to enhance the neural network.
convey the anti-inflammatory signal of the cholinergic anti-inflammatory pathway to peripheral immune
cells [13].
Key Symposium: The pulse of inflammation: implications for therapy
receiving the highest intensity aerobic training had
significant reductions in TNF production. These data
suggest that in healthy young adults, a 12-week
high-intensity aerobic training programme, sufficient to increase VO2 max, can inhibit cytokine
release from blood monocytes. A larger study is
presently underway to assess the role of vagus nerve
activity in conferring protection against inflammatory cytokine release in humans [68]. It has been
proposed that a major cardio-protective benefit of
exercise is derived from enhanced vagus nerve
activity, which inhibits inflammatory risk and
atherogenesis.
J. M. Huston & K. J. Tracey
|
52
ª 2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 45–53
1 Lotze MT, Tracey KJ. High-mobility group box 1 protein
(HMGB1): nuclear weapon in the immune arsenal. Nat Rev Immunol 2005; 5: 331–42.
2 Yang H, Hreggvidsdottir HS, Palmblad K et al. A Critical Cysteine
is Required for HMGB1 Binding to TLR4 and Activation of Macrophage Cytokine Release. Proc Natl Acad Sci USA 2010; 26:
11942–7.
3 Medzhitov R. Inflammation 2010: new adventures of an old
flame. Cell 2010; 140: 771–6.
4 Borovikova LV, Ivanova S, Nardi D et al. Role of vagus nerve signaling in CNI-1493-mediated suppression of acute inflammation. Auton Neurosci 2000; 85: 141–7.
5 Borovikova LV, Ivanova S, Zhang M et al. Vagus nerve stimulation
attenuates the systemic inflammatory response to endotoxin.
Nature 2000; 405: 458–62.
6 Tracey KJ. The inflammatory reflex. Nature 2002; 420: 853–9.
7 Huston JM, Ochani M, Rosas-Ballina M et al. Splenectomy inactivates the cholinergic antiinflammatory pathway during lethal
endotoxemia and polymicrobial sepsis. J Exp Med 2006; 203:
1623–8.
8 Wang H, Yu M, Ochani M et al. Nicotinic acetylcholine receptor alpha7 subunit is an essential regulator of inflammation. Nature
2003; 421: 384–8.
9 Bernik TR, Friedman SG, Ochani M et al. Pharmacological stimulation of the cholinergic antiinflammatory pathway. J Exp Med
2002; 195: 781–8.
10 Bernik TR, Friedman SG, Ochani M et al. Cholinergic antiinflammatory pathway inhibition of tumor necrosis factor during ischemia reperfusion. J Vasc Surg 2002; 36: 1231–6.
11 van Westerloo DJ, Giebelen IA, Florquin S et al. The vagus nerve
and nicotinic receptors modulate experimental pancreatitis
severity in mice. Gastroenterology 2006; 130: 1822–30.
12 van Westerloo DJ, Giebelen IA, Florquin S et al. The cholinergic
anti-inflammatory pathway regulates the host response during
septic peritonitis. J Infect Dis 2005; 191: 2138–48.
13 Huston JM, Gallowitsch-Puerta M, Ochani M et al. Transcutaneous vagus nerve stimulation reduces serum high mobility group
box 1 levels and improves survival in murine sepsis. Crit Care
Med 2007; 35: 2762–8.
14 Yeboah MM, Xue X, Duan B et al. Cholinergic agonists attenuate
renal ischemia-reperfusion injury in rats. Kidney Int 2008; 74:
62–9.
15 van Maanen MA, Lebre MC, van der PT et al. Stimulation of nicotinic acetylcholine receptors attenuates collagen-induced arthritis in mice. Arthritis Rheum 2009; 60: 114–22.
16 Mioni C, Bazzani C, Giuliani D et al. Activation of an efferent cholinergic pathway produces strong protection against myocardial
ischemia ⁄ reperfusion injury in rats. Crit Care Med 2005; 33:
2621–8.
References
No conflict of interest was declared.
Conflict of interest statement
17 Guarini S, Cainazzo MM, Giuliani D et al. Adrenocorticotropin
reverses hemorrhagic shock in anesthetized rats through the
rapid activation of a vagal anti-inflammatory pathway. Cardiovasc Res 2004; 63: 357–65.
18 Guarini S, Altavilla D, Cainazzo MM et al. Efferent vagal fibre
stimulation blunts nuclear factor-kappaB activation and protects against hypovolemic hemorrhagic shock. Circulation 2003;
107: 1189–94.
19 Ottani A, Giuliani D, Galantucci M et al. Melanocortins counteract inflammatory and apoptotic responses to prolonged myocardial ischemia ⁄ reperfusion through a vagus nerve-mediated
mechanism. Eur J Pharmacol 2010; 637: 124–30.
20 Li J, Xie H, Wen T, Liu H, Zhu W, Chen X. Expression of high
mobility group box chromosomal protein 1 and its modulating
effects on downstream cytokines in systemic lupus erythematosus. J Rheumatol 2010; 37: 766–75.
21 Tracey KJ. Understanding immunity requires more than immunology. Nat Immunol 2010; 11: 561–4.
22 Parrish WR, Rosas-Ballina M, Gallowitsch-Puerta M et al. Modulation of TNF Release by Choline Requires Alpha7 Subunit Nicotinic Acetylcholine Receptor-Mediated Signaling. Mol Med 2008;
14: 567–74.
23 de Jonge WJ, van der Zanden EP, The FO et al. Stimulation of the
vagus nerve attenuates macrophage activation by activating the
Jak2-STAT3 signaling pathway. Nat Immunol 2005; 6: 844–51.
24 Wang H, Liao H, Ochani M et al. Cholinergic agonists inhibit
HMGB1 release and improve survival in experimental sepsis. Nat
Med 2004; 10: 1216–21.
25 Huston JM, Wang H, Ochani M et al. Splenectomy protects
against sepsis lethality and reduces serum HMGB1 levels.
J Immunol 2008; 181: 3535–9.
26 Rosas-Ballina M, Tracey KJ. The neurology of the immune system: neural reflexes regulate immunity. Neuron 2009; 64: 28–32.
27 Rosas-Ballina M, Ochani M, Parrish WR et al. Splenic nerve is
required for cholinergic antiinflammatory pathway control of TNF
in endotoxemia. Proc Natl Acad Sci USA 2008; 105: 11008–13.
28 Huston JM, Rosas-Ballina M, Xue X et al. Cholinergic neural signals to the spleen down-regulate leukocyte trafficking via
CD11b. J Immunol 2009; 183: 552–9.
29 Saeed RW, Varma S, Peng-Nemeroff T et al. Cholinergic stimulation blocks endothelial cell activation and leukocyte recruitment
during inflammation. J Exp Med 2005; 201: 1113–23.
30 Brigden ML, Pattullo AL. Prevention and management of overwhelming postsplenectomy infection–an update. Crit Care Med
1999; 27: 836–42.
31 Tracey KJ. Suppression of TNF and other proinflammatory cytokines by the tetravalent guanylhydrazone CNI-1493. Prog Clin
Biol Res 1998; 397: 335–43.
32 Pavlov VA, Parrish WR, Rosas-Ballina M et al. Brain acetylcholinesterase activity controls systemic cytokine levels through the
cholinergic anti-inflammatory pathway. Brain Behav Immun
2009; 23: 41–5.
33 Pavlov VA, Ochani M, Gallowitsch-Puerta M et al. Central muscarinic cholinergic regulation of the systemic inflammatory
response during endotoxemia. Proc Natl Acad Sci USA 2006; 103:
5219–23.
34 Li Y, Wu X, Zhu J, Yan J, Owyang C. Hypothalamic regulation of
pancreatic secretion is mediated by central cholinergic pathways
in the rat. J Physiol 2003; 552: 571–87.
35 Shimazu T, Matsushita H, Ishikawa K. Cholinergic stimulation of
the rat hypothalamus: effects of liver glycogen synthesis. Science
1976; 194: 535–6.
Key Symposium: The pulse of inflammation: implications for therapy
Supported in part by grants from the NIH (NIGMS) to
KJT.
Acknowledgements
J. M. Huston & K. J. Tracey
|
ª 2010 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine 269; 45–53
53
Correspondence: Jared M. Huston, MD, Department of Surgery,
Stony Brook University Medical Center, T18-040, Health Sciences
Center, Stony Brook, NY 11794, USA.
(fax: +631-444-6176; e-mail: [email protected]).
54 Dantzer R, O’Connor JC, Freund GG, Johnson RW, Kelley KW.
From inflammation to sickness and depression: when the
immune system subjugates the brain. Nat Rev Neurosci 2008; 9:
46–56.
55 Raison CL, Capuron L, Miller AH. Cytokines sing the blues:
inflammation and the pathogenesis of depression. Trends Immunol 2006; 27: 24–31.
56 Miller AH, Maletic V, Raison CL. Inflammation and its discontents: the role of cytokines in the pathophysiology of major
depression. Biol Psychiatry 2009; 65: 732–41.
57 Kemp AH, Quintana DS, Gray MA, Felmingham KL, Brown K,
Gatt JM. Impact of depression and antidepressant treatment on
heart rate variability: a review and meta-analysis. Biol Psychiatry
2010; 67: 1067–74.
58 Bajbouj M, Merkl A, Schlaepfer TE et al. Two-year outcome of vagus nerve stimulation in treatment-resistant depression. J Clin
Psychopharmacol 2010; 30: 273–81.
59 Rush AJ, Sackeim HA, Marangell LB et al. Effects of 12 months of
vagus nerve stimulation in treatment-resistant depression: a
naturalistic study. Biol Psychiatry 2005; 58: 355–63.
60 Marangell LB, Rush AJ, George MS et al. Vagus nerve stimulation
(VNS) for major depressive episodes: one year outcomes. Biol Psychiatry 2002; 51: 280–7.
61 Dowlati Y, Herrmann N, Swardfager W et al. A meta-analysis of
cytokines in major depression. Biol Psychiatry 2010; 67: 446–57.
62 Lampert R, Bremner JD, Su S et al. Decreased heart rate variability is associated with higher levels of inflammation in middleaged men. Am Heart J 2008; 156: 759.e1–7.
63 Haensel A, Mills PJ, Nelesen RA, Ziegler MG, Dimsdale JE. The
relationship between heart rate variability and inflammatory
markers in cardiovascular diseases. Psychoneuroendocrinology
2008; 33: 1305–12.
64 Carney RM, Freedland KE, Stein PK et al. Heart rate variability
and markers of inflammation and coagulation in depressed
patients with coronary heart disease. J Psychosom Res 2007; 62:
463–7.
65 Janszky I, Ericson M, Lekander M et al. Inflammatory markers
and heart rate variability in women with coronary heart disease.
J Intern Med 2004; 256: 421–8.
66 Frasure-Smith N, Lesperance F, Irwin MR, Talajic M, Pollock BG.
The relationships among heart rate variability, inflammatory
markers and depression in coronary heart disease patients.
Brain Behav Immun 2009; 23: 1140–7.
67 Soares-Miranda L, Negrao CE, Antunes-Correa LM et al. High levels of C-reactive protein are associated with reduced vagal modulation and low physical activity in young adults. Scand J Med Sci
Sports 2010; July 6. [Epub ahead of Print].
68 Sloan RP, Shapiro PA, Demeersman RE et al. Aerobic exercise
attenuates inducible TNF production in humans. J Appl Physiol
2007; 103: 1007–11.
69 Marsland AL, Gianaros PJ, Prather AA, Jennings JR, Neumann
SA, Manuck SB. Stimulated production of proinflammatory cytokines covaries inversely with heart rate variability. Psychosom
Med 2007; 69: 709–16.
Key Symposium: The pulse of inflammation: implications for therapy
36 Sloan RP, McCreath H, Tracey KJ, Sidney S, Liu K, Seeman T. RR
interval variability is inversely related to inflammatory markers:
the CARDIA study. Mol Med 2007; 13: 178–84.
37 Heart rate variability: standards of measurement, physiological
interpretation and clinical use. Task Force of the European Society of Cardiology and the North American Society of Pacing and
Electrophysiology. Circulation 1996; 93: 1043–65.
38 Gaziano EP, Freeman DW. Analysis of heart rate patterns preceding fetal death. Obstet Gynecol 1977; 50: 578–82.
39 Sampson MB, Mudaliar NA, Lele AS. Fetal heart rate variability
as an indicator of fetal status. Postgrad Med 1980; 67: 207–10,
213–15.
40 Lauersen NH, Wilson KH, Bilek A, Rao VS, Kurkulos M. A new
modality in nonstress testing: evaluation of beat-to-beat fetal
heart rate variability. Am J Obstet Gynecol 1981; 141: 521–6.
41 Kleiger RE, Miller JP, Bigger JT Jr, Moss AJ. Decreased heart rate
variability and its association with increased mortality after
acute myocardial infarction. Am J Cardiol 1987; 59: 256–62.
42 La Rovere MT, Bigger JT Jr, Marcus FI, Mortara A, Schwartz PJ.
Baroreflex sensitivity and heart-rate variability in prediction of
total cardiac mortality after myocardial infarction. ATRAMI
(Autonomic Tone and Reflexes After Myocardial Infarction) Investigators. Lancet 1998; 351: 478–84.
43 Schwartz PJ, La Rovere MT, Vanoli E. Autonomic nervous system
and sudden cardiac death. Experimental basis and clinical
observations for post-myocardial infarction risk stratification.
Circulation 1992; 85: I77–91.
44 Garrard CS, Kontoyannis DA, Piepoli M. Spectral analysis of
heart rate variability in the sepsis syndrome. Clin Auton Res
1993; 3: 5–13.
45 Schmidt H, Muller-Werdan U, Hoffmann T et al. Autonomic dysfunction predicts mortality in patients with multiple organ dysfunction syndrome of different age groups. Crit Care Med 2005;
33: 1994–2002.
46 Aydemir M, Yazisiz V, Basarici I et al. Cardiac autonomic profile
in rheumatoid arthritis and systemic lupus erythematosus.
Lupus 2010; 19: 255–61.
47 Stojanovich L, Milovanovich B, de Luka SR et al. Cardiovascular
autonomic dysfunction in systemic lupus, rheumatoid arthritis,
primary Sjogren syndrome and other autoimmune diseases.
Lupus 2007; 16: 181–5.
48 Sharma P, Makharia GK, Ahuja V, Dwivedi SN, Deepak KK. Autonomic dysfunctions in patients with inflammatory bowel disease
in clinical remission. Dig Dis Sci 2009; 54: 853–61.
49 Lindgren S, Stewenius J, Sjolund K, Lilja B, Sundkvist G. Autonomic vagal nerve dysfunction in patients with ulcerative colitis.
Scand J Gastroenterol 1993; 28: 638–42.
50 Bruchfeld A, Goldstein RS, Chavan S et al. Whole blood cytokine
attenuation by cholinergic agonists ex vivo and relationship to vagus nerve activity in rheumatoid arthritis. J Intern Med 2010;
268: 94–101.
51 Goldstein RS, Bruchfeld A, Yang L et al. Cholinergic anti-inflammatory pathway activity and High Mobility Group Box-1
(HMGB1) serum levels in patients with rheumatoid arthritis. Mol
Med 2007; 13: 210–5.
52 Jan BU, Coyle SM, Macor MA, Reddell M, Calvano SE, Lowry SF.
Relationship of basal heart rate variability to in vivo cytokine
responses after endotoxin exposure. Shock 2010; 33: 363–8.
53 Hoeger S, Bergstraesser C, Selhorst J et al. Modulation of brain
dead induced inflammation by vagus nerve stimulation. Am
J Transplant 2010; 10: 477–89.
J. M. Huston & K. J. Tracey
PLoS ONE | www.plosone.org
Congestive heart failure (CHF) remains to be one of the
major cardiovascular disorders in the world [1]. Despite its
high expenditure in healthcare budgets [2], the mortality rate
of CHF patients can be up to 8 times higher than the agematched control population [3]. The present treatment
protocols of CHF patients, such as administrating angiotensin
converting enzyme inhibitors (ACE-I) and b blockers, have
been proven to lower the mortality and hospital admission rate
[4]. Nevertheless, the residual risk for mortality and morbidity
of CHF remains high even under such treatment protocols
[5,6]. Therefore, further novel prognostic predictor is needed to
strengthen the treatment strategy in addition to neurohormonal
inhibition therapy.
Conventional linear heart rate variability (HRV) analyses,
including frequency and time domain analyses, have been
reported as prognostic factors for CHF [7,8]. However, heart
rate fluctuations have been recognized as complex behaviors
Introduction
* E-mail: [email protected]
1
April 2011 | Volume 6 | Issue 4 | e18699
originated from nonlinear processes and often with nonstationary
property [9–11]. Applying linear algorithms to those seemingly
irregular and ‘‘patchy’’ patterns of heart rate fluctuations [12]
may cause the intrinsic computational errors of the linear
algorithms [11,13,14]. Properly use of the analyses based on
fractals and chaos theory [15–17] to qualify or quantify the
characteristics of heart rate time series are suggested to serve as a
more reliable index of physiological systems in many clinical
studies [10,11,18]. As one of such mathematic methods,
multiscale entropy (MSE) analysis has focused specifically on
characterizing heterogeneous complexity [19]. Such complex
structure is ‘‘breakdown’’ (loss of information richness) and
points to poor prognosis in CHF patients [19,20]. We
hypothesized that MSE could yield a prognostic marker which
was not relevant to neurohormonal inhibition therapy in CHF
patients. The aims of this study were 1) to evaluate the influences
of b-blockers on parameters derived from MSE; 2) to assess the
prognostic significance of parameters derived from MSE for
CHF patients.
Competing Interests: The authors have declared that no competing interests exist.
Funding: M-T Lo was supported by NSC (Taiwan, ROC), Grant No 99-2627-B-008-003, joint foundation of CGH and NCU, Grant No CNJRF-99CGH-NCU-A3,
VGHUST100-G1-4-3 and NSC support for the Center for Dynamical Biomarkers and Translational Medicine, National Central University, Taiwan (NSC 99-2911-I-008100). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Copyright: ß 2011 Ho et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted
use, distribution, and reproduction in any medium, provided the original author and source are credited.
Received December 31, 2010; Accepted March 8, 2011; Published April 13, 2011
Editor: Matjaz Perc, University of Maribor, Slovenia
Citation: Ho Y-L, Lin C, Lin Y-H, Lo M-T (2011) The Prognostic Value of Non-Linear Analysis of Heart Rate Variability in Patients with Congestive Heart Failure—A
Pilot Study of Multiscale Entropy. PLoS ONE 6(4): e18699. doi:10.1371/journal.pone.0018699
Conclusion: The area under the MSE curve for scale 6 to 20 is not relevant to b-blockers and could further warrant
independent risk stratification for the prognosis of CHF patients.
Methods and Results: Patients with systolic heart failure were enrolled in this study. One month after clinical condition
being stable, 24-hour Holter electrocardiogram was recording. MSE as well as other standard parameters of heart rate
variability (HRV) and detrended fluctuation analysis (DFA) were assessed. A total of 40 heart failure patients with a mea age
of 56616 years were enrolled and followed-up for 6846441 days. There were 25 patients receiving b-blockers treatment.
During follow-up period, 6 patients died or received urgent heart transplantation. The short-term exponent of DFA and the
slope of MSE between scale 1 to 5 were significantly different between patients with or without b-blockers (p = 0.014 and
p = 0.028). Only the area under the MSE curve for scale 6 to 20 (Area6–20) showed the strongest predictive power between
survival (n = 34) and mortality (n = 6) groups among all the parameters. The value of Area6–20!21.2 served as a significant
predictor of mortality or heart transplant (p = 0.0014).
Aims: The influences of nonstationarity and nonlinearity on heart rate time series can be mathematically qualified or
quantified by multiscale entropy (MSE). The aim of this study is to investigate the prognostic value of parameters derived
from MSE in the patients with systolic heart failure.
Abstract
1 Graduate Institute of Clinical Medicine, National Taiwan University Hospital and National Taiwan University College of Medicine, Taipei, Taiwan, 2 Division of Cardiology,
Department of Internal Medicine, National Taiwan University Hospital and National Taiwan University College of Medicine, Taipei, Taiwan, 3 Center for Dynamical
Biomarkers and Translational Medicine, National Central University, Chungli, Taiwan, 4 Research Center for Adaptive Data Analysis, National Central University, Taoyuan,
Taiwan, 5 Institute of Systems Biology and Bioinformatics, National Central University, Taoyuan, Taiwan
Yi-Lwun Ho1,2, Chen Lin3,4,5, Yen-Hung Lin2, Men-Tzung Lo3,4*
The Prognostic Value of Non-Linear Analysis of Heart
Rate Variability in Patients with Congestive Heart
Failure—A Pilot Study of Multiscale Entropy
PLoS ONE | www.plosone.org
DFA is a modified root-mean-square analysis used to evaluate
the fractal correlation beneath the heart rate fluctuation originated
from the interacted regulatory mechanisms. The algorithm has
Detrended fluctuation analysis (DFA)
Nonlinear analysis enables the researchers to probe the
fundamental characteristics of the signals. However, unwanted
inferences such as noise and nonstationarity may introduce
spurious features to the signals [24,25]. The underlying mechanisms of the irregular and unpredicted behavior of the signals can
be misinterpreted and the reliability of the results of analysis can
be compromised. Two methods had been chosen for their ability
to evaluate the main properties of the signals [14,26].
Nonlinear methods
Standard deviation of normal RR intervals (SDNN) and
percentage of absolute differences in normal RR intervals greater
than 50 ms (pNN50) were calculated to represent the total variance
and vagal modulation of the HR. In addition, the spectrum
analysis was carried out in accordance with the recommendations
of the European Society of Cardiology and the North American
Society of Pacing Electrophysiology [23]. The spectral density of
each frequency band-high frequency (HF) (0.15–0.4 Hz), low
frequency (LF) (0.04–0.15 Hz), and very low frequency (VLF)
(0.003–0.04) were computed by average power spectrum.
Time and frequency domain analysis
Each digitalized 24-hour ECG data was annotated by an
automated algorithm and the annotated file was then carefully
inspected and corrected by technicians for extracting the RR
intervals. The ectopic beats (including atrial or ventricular
premature beats) were interpolated by its adjacent RR intervals.
A four-hour period of RR intervals in daytime (between 9AM–
5PM) was selected from each recording to avoid confounding
effects on nonlinear or linear analysis caused by different sleep
stage or diurnal rhythm [21,22]. Only subjects consisted of more
than 80% of qualified normal sinus beats were included for further
analysis (the typical RR-interval tracings and the corresponding
recurrent plots for survival and mortality groups (Figure S1) as well
as all of the RR-interval data are provided as supplementary
materials (Materials S1)).
Data pre-processing
Patients with manifestation of exertional dyspnea, leg edema
and systolic heart failure (LVEF,45% by echocardiography) at
the National Taiwan University Hospital were enrolled after
giving their inform consents. Baseline information, including age,
sex, etiologies for heart failure,diabetes mellitus, hypertension,
dyslipidemia (total cholesterol .220 mg/dl), and cardiovascular
medication use (b-blockers, ACE-I, angiotensin-II receptor
blockers, and spironolactone) was reviewed in medical records
and charts. Patients with renal dysfunction (defined by creatine§2.0 mg/dl) were excluded. One month after clinical condition
being stable, standard ambulatory 24-hour electrocardiogram
(ECG) recorders were placed on all participants. The ECG signals
were sampled at 250 Hz and stored in SD memory card for offline
analysis on a microcomputer. Subsequently, these patients were
followed up and mortality or heart transplantation will be noted as
end-point for follow-up. The Ethics Committee of National
Taiwan University Hospital approved the study and all patients
provided written informed consent.
Study Population
Methods
2
April 2011 | Volume 6 | Issue 4 | e18699
Figure 1. Demonstrative graph of MSE derived parameters. The
profile of MSE can be assessed by a) its linear-fitted slope between
certain scales which represent the complexity behaviors of the signals.
The negative slope may indicate a random-like structure over certain
timescale. B) the area under curve between certain scales that may
represent its quantitative feature of the underlying physiological
mechanisms in certain time scales (ex. area under scale1,5 may
respond to the ability of respiratory sinus arrhythmia).
doi:10.1371/journal.pone.0018699.g001
Instead of simply using single time scale to estimate the complex
pattern (irregularity) of a time series, MSE extended this concept to
evaluate the complexity of physiological signals on multiple time
scales. It comprises of two steps: 1) coarse-graining the signals into
different time scales; 2) quantify the degree of irregularity in each
coarse-grained time series using sample entropy (SpEn) [27].
Finally, the entropy is calculated as a function of scale, providing
a measure of information richness embedded in different time
scales. In addition, it has shown that different features of small and
large scales in different groups of subjects may assist the clinical
categorization [19] and thus three different parameters were
derived from the MSE profile: the summations of quantitative
values of scale 1–5 (Area5) or scale 6–20 (Area6–20) which represent
the complexity exhibit in short and long time scales, respectively;
and the linear-fitted slope of the first 5 scales (Slope5) (Figure 1).
Although MSE was successfully applied in physiological signals
[18,19,24], nonstationary artifacts especially trends can compromise the estimation of entropy-based analysis by increasing the
standard deviation of the data. Hence, detrending process was used
to attenuate the spurious influence caused by nonstationarity [24].
MSE analysis
been described in detail elsewhere [14]. To briefly introduce this
method, at first, it eliminates the environmental inferences by
removing the linear-fitted ‘‘local’’ trend over different time scales
(‘‘box sizes’’) in an integrated time series. Next, the root-meansquare fluctuation of this integrated and detrended time series is
calculated. This procedure is repeated over different time scales
and then the slope of the curve (a exponent) can be estimated on
the log-log plot of fluctuations versus box sizes.
In addition, a crossover phenomenon of a exponent in heart
rate dynamics between short (4–11 beats) and long (11–64 beats)
time scales has been proposed. The short-term (a1) as well as longterm (a2) fractal correlation exponents were calculated to provide
better understanding of the fractal correlation property in
physiological system [14].
Heart Failure and Multi-Scale Entropy
129685
185635
13.661.9
8.963.2
Triglyceride (mg/dl)
Cholesterol (mg/dl)
Hemoglobin (g/dl)
Uric acid (mg/dl)
3
113647
Fasting sugar (mg/dl)
PLoS ONE | www.plosone.org
Among all parameters, Area5, Area6–20, and LF were
significantly lower (p = 0.027, p = 0.021, and p = 0.004 respectively)
in the mortality group. (Table 3) Moreover, the ROC of the
Application of MSE in prognostic prediction
There was no significant difference when any single sample
entropy value of scales 1 to 20 was compared between patient
groups with or without b-blockers. However, the value of Slope5
was not only significantly lower (p = 0.028) in patients without bblocker therapy but also exhibit a negative value (Table 2).
Effect of b-blockers on the dynamical complexity
assessed by the MSE analysis
While the conventional HRV measurements showed no
significant different between these two groups, DFAa1 showed
significantly higher values (p = 0.014) in patients with b-blocker
treatment (Table 2).
Effect of b-blockers on autonomic activities and its fractal
properties
A total of 40 heart failure patients (30 males and 10 females)
with a mean age of 56616 years were enrolled and followed-up for
6846441 days. Twenty-five patients received b-blockers (either
carvedilol or metoprolol). Carvedilol was titrated from 3.25 mg
per day and metoprolol was titrated from 12.5 mg per day to
maximal tolerable doses. Six patients died or received heart
transplantation during follow-up period of this study. The
demographic and clinical data were showed in Table 1. No
clinical variable was significantly different between these two
groups (with or without b-blocker therapy).
Characteristics of Patients
Results
1.0760.25
Creatinine
25.665.3
5
0
25.465.2
IV
9
4
6
Non-coronary artery Diseases
Hypertension
Diabetes mellitus
10
9
6
Loop diuretics
Digoxin
Spironolactone
7
10
18
19
10
11
14
11
9
VLF
p = 0.189
p = 0.845
p = 0.924
p = 0.929
April 2011 | Volume 6 | Issue 4 | e18699
Although the effects of b-blockers on HRV indices have been
extensively studied [30–32], the effects of b-blockers on MSE are
still unclear. The present study was the first study to assess the
relationship between b-blockers and MSE in CHF patients. The
main findings of this preliminary study were that the b-blocker
therapy may change the short-term complexity (the slope of MSE
between scale 1 to 5). But the most significant predictor of
Discussion
previously proposed predictors (LF, VLF, and DFAa1) and the
new derived parameters (Area5 and Area6–20) were depicted in
Figure 2. Area6–20 (AUC:0.85860.075) showed the best overall
discriminative power than LF (AUC: 0.78460.087), VLF
(AUC:0.73560.117), DFAa1 (AUC:0.7160.145) and Area5
(AUC:0.79460.108) in mortality or heart transplant prediction.
Therefore, Area6–20 was adopted to perform the analysis of
Kaplan-Meier survival curves. The value of Area6–20!21.2 was a
significant predictor of mortality or heart transplant (p = 0.0014).
(Figure 3)
13.163.0
5.361.2
20.0260.07
1.2760.09
0.9160.22
484.86321.7
56.3643.0
31.43620.91
0.9160.85
53.2617.8
14.362.7
5.561.1
0.0360.08
1.2660.18
1.1060.34
563.06462.9
90.96102.9
27.9635.3
0.7661.13
49.1626.9
b-blockers(+)
(n = 25)
p = .211
p = .719
p = .028
p = .679
p = .014
p = .956
p = .659
p = .088
p = .100
p = .332
p value
pNN50
87.6690.1
575.46420.4
LF
VLF
22.363.5
16.364.5
4.561.1
20.0360.11
1.1560.23
0.8460.45
297.46290.5
23.3622.6
12.669.9
0.4660.37
38.4622.9
Mortality
group
(n = 6)
p = .004
P = .021
p = .197
p = .239
p = .127
p = .071
p = .027
p = .092
p = .939
p = .225
p value
PLoS ONE | www.plosone.org
Among all parameters, Area5, Area6–20, and LF were significantly lower
(p = 0.027, p = 0.021, and p = 0.004 respectively) in the mortality group and
those indices may potentially serve as outcome predictors.
Slope5 = the linear-fitted slope of the first 5 scales, Area5 = the summations of
quantitative values of scale 1–5, Area6–20 = the summations of quantitative
values of scale 6–20.
doi:10.1371/journal.pone.0018699.t003
5.661.0
Area5
Area6–20
Slope5
0.0260.07
1.2860.13
a2
Multiscale entropy
1.0660.27
a1
Detrended fluctuation analysis
32.1632.0
HF
Frequency domain analysis
52.8623.5
0.8861.09
SDNN
Time domain analysis
Survival
group
(n = 34)
Table 3. Prognostic value of parameters of HRV.
mortality or heart transplantation was the long-term complexity
(the area under the MSE curve for scale 6 to 20). Therefore, we
found an alternative prognostic predictor of CHF in addition to
While the conventional HRV measurements showed no significant different
between these two groups, nonlinear indices, DFAa1 and the value of Slope5,
were significantly lower in patients without b-blocker therapy.
Slope5 = the linear-fitted slope of the first 5 scales, Area5 = the summations of
quantitative values of scale 1–5, Area6–20 = the summations of quantitative
values of scale 6–20.
doi:10.1371/journal.pone.0018699.t002
Area6–20
Area5
Slope5
Multiscale entropy
a2
p = 0.138
p = 0.660
a1
p = 0.712
Detrended fluctuation analysis
LF
p = 0.988
HF
p = 0.280
p = 0.573
Frequency domain analysis
pNN50
p = 0.684
SDNN
p = 0.158
Time domain analysis
b-blockers(2
2)
(n = 15)
Table 2. Effect of the b-blockers on the autonomic activities,
fractal properties and MSE.
p = 0.793
p = 0.073
A total of 40 heart failure patients (30 males and 10 females) were enrolled in
this study. No clinical variable was significantly different between the patients
with or without b-blocker therapy.
NYHA = New York Heart Association; ACE-I = angiotensin converting enzyme
inhibitor; ARB = angiotensin receptor blocker.
doi:10.1371/journal.pone.0018699.t001
12
ACE-I/ARB
Medication
6
Coronary artery disease
Etiology of heart failure
Body mass index
2
8
III
10
2
II
4
7.162.6
13.462.3
185665
1956177
122647
1.1860.35
35614
I
NYHA functional class
33612
LVEF(%)
For the independence of different nominal variables between
groups, the chi-square test or Fisher exact test were performed.
The continuous variables were represented as mean value 6 SD
and the normality of those variables was evaluated by using the
Shapiro-Wilk test. Then, the Mann-Whitney U test or Student’s t
test was applied to the between-group comparison accordingly
while the Wilcoxon sign test or Student’s paired t test was
calculated for the intra-group comparison. The receiver operating
characteristic curve (ROC) was constructed by the sensitivity and
specificity of the continuous variable in predicting the end-point.
Area under the ROC curve (AUCs) gave an estimate of the overall
discriminate ability. Furthermore, the most predictive indexes will
be selected to seek the optimal cut point within the 30th to 70th
percentile for all patients in 5th-percentile step. The maximal
hazards ratio and independent correlation of variables with event
status (mortality) was determined by Cox regression analysis.
Then, Kaplan-Meier event probability curves and log rank
analysis of the dichotomized groups were obtained. The statistical
significance was set at p,0.05.
87615
21/4
9/6
86613
Male/Female
51.8613.5
61.7614.7
2) b-blockers (+)
b-blockers(2
(n = 15)
(n = 25)
p value
Heart rate (bpm)
Age (years)
Patient characteristics
Table 1. The clinical characteristics between patients with
and without using b-blockers.
Statistical analysis
In this study, the empirical mode decomposition (EMD) method
was adopted as an adaptive filter to eliminate the oscillations
slower than VLF range in the original R-R interval signals [28].
The data subsequently evaluated by the MSE analysis after
detrending. This algorithm, instead of removing trend with a priori
mathematical formulas such as linear or polynomial functions
[14,29], could evaluate the hidden dynamics of heart beat
fluctuations better [9,10,29].
Heart Failure and Multi-Scale Entropy
4
April 2011 | Volume 6 | Issue 4 | e18699
Figure 3. Using MSE Area6–20!21.2 as a clinical predictor,
significant difference in survival was noted from the KaplanMeier survival curve (P = 0.0014).
doi:10.1371/journal.pone.0018699.g003
All linear HRV measurement showed no difference between
patients with or without b-blocker treatment in the present study.
Effect of b-blocker therapy on linear and nonlinear
properties of HRV
neurohormonal inhibition therapy by assessing the nonlinear
characteristics of the heart rate fluctuations.
Figure 2. Very-low-frequency component (VLF; black, solid
line), low-frequency component (LF; black, dashed line), shortterm fractal exponent (DFAa1; black, dotted line), the summations of quantitative values of scale 1–5 (Area5; grey, dashed
line), and the summations of quantitative values of scale 6–20
(Area6–20; grey, dash-dotted line) receiver operating characteristic (ROC) curves. The area under each ROC curves (AUC) was
calculated for each parameters. The AUCs were 0.735 for VLF, 0.784 for
LF, 0.701 for DFAa1, 0.794 for Area5, and 0.858 for Area6–20.
doi:10.1371/journal.pone.0018699.g002
Heart Failure and Multi-Scale Entropy
PLoS ONE | www.plosone.org
1. Wang TJ, Evans JC, Benjamin EJ, Levy D, LeRoy EC, et al. (2003) Natural
history of asymptomatic left ventricular systolic dysfunction in the community.
Circulation 108: 977–982.
2. Stewart S, Jenkins A, Buchan S, McGuire A, Capewell S, et al. (2002) The
current cost of heart failure to the National Health Service in the UK. Eur J Heart
Fail 4: 361–371.
3. van Jaarsveld CH, Ranchor AV, Kempen GI, Coyne JC, van Veldhuisen DJ,
et al. (2006) Epidemiology of heart failure in a community-based study of
subjects aged . or = 57 years: incidence and long-term survival. Eur J Heart
Fail 8: 23–30.
4. Flather MD, Yusuf S, Kober L, Pfeffer M, Hall A, et al. (2000) Long-term ACEinhibitor therapy in patients with heart failure or left-ventricular dysfunction: a
systematic overview of data from individual patients. ACE-Inhibitor Myocardial
Infarction Collaborative Group. Lancet 355: 1575–1581.
5. Dahlof B, Devereux RB, Kjeldsen SE, Julius S, Beevers G, et al. (2002)
Cardiovascular morbidity and mortality in the Losartan Intervention For
References
Applying Fourier-based method to nonlinear or nonstationary
signals may result in inaccurate estimations [9,29] and compromise the sensitivity of linear HRV measurements. Recently,
DFAa1 has been proposed to be a better predictor for CHF
patients [34]. In the present study, SDNN and DFAa1 failed to
predict the prognosis of CHF patients. Makikallio et al. noticed that
HRV indexes such as SDNN is less sensitive in CHF patient with
NYHA.III [35]. Sixteen patients (40%) were classified as NYHA
III–IV in our study and five out of them were in mortality group.
Moreover, administration of beta-blockade has shown a reversal
effect in either linear and nonlinear parameters such as SDNN,
HF, LF and DFAa1 [30,32,33]. Half of the patients with poor
outcome were treated with beta blocker which may potentially
influence the parameters. SDNN and DFAa1 may be, therefore,
Complexity analysis as a prognosis predictor
Other researchers have proposed that the restoration of autonomic
function can be assessed by HRV indices [30,31,33]. The
discrepancy between our finding and previous studies could be
due to several factors. The first, the b-blockers were titrated
according to patients’ tolerance in our study. Therefore, the
duration and dosage of b-blockers were variable. The second, the
ECG was recorded 1 month after clinical condition being stable in
our study. However, significant changes of most linear HRV
parameters are found after 12 weeks of treatments. The short-term
fractal scaling correlation index, DFAa1, was markedly higher in
b-blocker group. The reversal of DFAa1 is also reported after
administrating b-blocker in patients with CHF [33]. Our data
showed similar results.
The new nonlinear method, MSE, allows us to evaluate the
information richness in heart beat time series over different time
scales. Although the underlying mechanisms responsible for the
quantitative feature of the MSE in different time scales are still
unclear, previous studies have shown that the complexity
decreased significantly during aging and further deteriorated in
patients with CHF [19]. We applied three different parameters,
Area5, Area6–20, and Slope5, to compare patients with or without
b-blocker treatment. The summation of entropy values at different
time scales may give the quantitative estimation of information
richness over certain time scales. That is, Area5 and Area6–20 can
probe the complexity structure of the heart rate dynamics in short
(e.g., 1 to 5 heart beats) and longer (e.g., 6 to 20 heart beats) time
scales, respectively. The Slope5 also outlined the structure of heart
rate dynamics in short time scale [19]. The negative value of
Slope5 indicates random-like patterns in short time scales.
Therefore, the significant difference between b-blocker and non
b-blocker groups might due to dysfunction of the short-term
regulatory mechanisms coincided with the results assessed by
DFAa1.
5
April 2011 | Volume 6 | Issue 4 | e18699
Endpoint reduction in hypertension study (LIFE): a randomised trial against
atenolol. Lancet 359: 995–1003.
6. Cohn JN, Tognoni G (2001) A randomized trial of the angiotensin-receptor
blocker valsartan in chronic heart failure. N Engl J Med 345: 1667–
1675.
7. Fauchier L, Babuty D, Cosnay P, Fauchier JP (1999) Prognostic value of heart
rate variability for sudden death and major arrhythmic events in patients with
idiopathic dilated cardiomyopathy. J Am Coll Cardiol 33: 1203–1207.
8. La Rovere MT, Pinna GD, Maestri R, Mortara A, Capomolla S, et al. (2003)
Short-term heart rate variability strongly predicts sudden cardiac death in
chronic heart failure patients. Circulation 107: 565–570.
9. Lo MT, Tsai PH, Lin PF, Lin C, Hsin YL (2009) The nonlinear and
nonstationary properties in EEG signals: probing the complex fluctuations by
Hilbert Huang Transform. Adv Adapt Data Anal 1: 461–482.
10. Peng CK, Costa M, Goldberger AL (2009) Adaptive data analysis of complex
fluctuations in physiologic time series. Adv Adapt Data Anal 1: 61–70.
Conceived and designed the experiments: YLH MTL. Performed the
experiments: YHL CL. Analyzed the data: MTL CL. Contributed
reagents/materials/analysis tools: YLH MTL. Wrote the paper: YLH CL.
Author Contributions
output in a one-column Ascii file and categorized into
two groups according to their outcomes after 2.5 years
follow up. All of the files were packed into a compressed RAR
file.
(RAR)
Materials S1 The RR intervals of each patient was
Figure S1 The 4-hour time tracings of RR intervals of
survived patient (A) and patient who died after 1 year (B)
and the return map traits of three-dimensional RR time
series reconstruction (x-axis for RRn, y-axis for RRn++1
and z-axis for RRn+2) of survived patient (C) and expired
patients (D). Note that the trait of the map in survival patient
was similar to that in expired patient.
(TIFF)
Supporting Information
First, our study had small sample size and no placebo-controlled
group. Second, all ECG data were recorded in normal ‘‘freerunning’’ conditions with possible confounding factors (e.g.,
physical activities, different breathing patterns, and so on). The
additive or dynamical noises may affect the properties of the
signals. Although we examined and interpreted the results of the
analysis cautiously and the differences characteristics of the
patients were unlikely due to the noise, we did not assess the
features or level of noise for more detailed information that may
benefit the exploration of the underlying deterministic rules.
Finally, some parameters related to possible physiological
mechanisms of MSE were not collected, such as baroreflex
sensitivity, catecholamine levels, and chemoreflex activities.
In conclusions, the area under the MSE curve for scale 6 to 20 is
not relevant to b-blockers and could further warrant independent
risk stratification for the prognosis of CHF patients.
Limitations of study
insensitive to predict their prognosis. Area5 and Area6–20 derived
from MSE were markedly lower in the mortality group. This
phenomenon was in agreement with those found by Costa et al. in
MUSIC study [19]. Although the underlying mechanisms were
still unclear, this preliminary study provided a new insight for the
prognosis of CHF by probing the dynamical complexity on the
system level. It could potentially offer an alternative marker for the
outcome of CHF in addition to neurohormonal inhibition therapy.
Heart Failure and Multi-Scale Entropy
PLoS ONE | www.plosone.org
11. Goldberger AL, Amaral LA, Hausdorff JM, Ivanov PC, Peng CK, et al. (2002)
Fractal dynamics in physiology: alterations with disease and aging. Proc Natl
Acad Sci U S A 99 Suppl 1: 2466–2472.
12. Buchman TG (2002) The community of the self. Nature 420: 246–251.
13. Lombardi F (2000) Chaos theory, heart rate variability, and arrhythmic
mortality. Circulation 101: 8–10.
14. Peng CK, Havlin S, Stanley HE, Goldberger AL (1995) Quantification of scaling
exponents and crossover phenomena in nonstationary heartbeat time series.
Chaos 5: 82–87.
15. Jagric T, Marhl M, Stajer D, Kocjancic ST, Jagric T, et al. (2007) Irregularity
test for very short electrocardiogram (ECG) signals as a method for predicting a
successful defibrillation in patients with ventricular fibrillation. Transl Res 149:
145–151.
16. Perc M (2005) Nonlinear time series analysis of the human electrocardiogram.
European Journal of Physics 26: 757–768.
17. Perc M (2005) The dynamics of human gait. European Journal of Physics 26:
525–534.
18. Yuan HK, Lin C, Tsai PH, Chang FC, Lin KP, et al. (2011) Acute increase of
complexity in the neurocardiovascular dynamics following carotid stenting. Acta
Neurol Scand 123: 187–192.
19. Costa M, Goldberger AL, Peng CK (2005) Multiscale entropy analysis of
biological signals. Phys Rev E Stat Nonlin Soft Matter Phys 71: 021906.
20. Costa M, Cygankiewicz I, Zareba W, Bayes de Luna A, Goldberger AL, et al.
(2006) Multiscale Complexity Analysis of Heart Rate Dynamics in Heart Failure:
Preliminary Findings from the MUSIC Study. Comput Cardiol 33: 101–103.
21. Pikkujamsa SM, Makikallio TH, Sourander LB, Raiha IJ, Puukka P, et al. (1999)
Cardiac interbeat interval dynamics from childhood to senescence: comparison
of conventional and new measures based on fractals and chaos theory.
Circulation 100: 393–399.
22. Vanoli E, Adamson PB, Ba L, Pinna GD, Lazzara R, et al. (1995) Heart rate
variability during specific sleep stages. A comparison of healthy subjects with
patients after myocardial infarction. Circulation 91: 1918–1922.
23. (1996) Heart rate variability: standards of measurement, physiological
interpretation and clinical use. Task Force of the European Society of
Cardiology and the North American Society of Pacing and Electrophysiology.
Circulation 93: 1043–1065.
6
April 2011 | Volume 6 | Issue 4 | e18699
24. Costa M, Priplata AA, Lipsitz LA, Wu Z, Huang NE, et al. (2007) Noise and
poise: Enhancement of postural complexity in the elderly with a stochasticresonance-based therapy. Europhys Lett 77: 68008.
25. Kantz H, Schreiber T (2004) Nonlinear Time Series Analysis. Cambridge:
Cambridge University Press.
26. Costa M, Goldberger AL, Peng CK (2002) Multiscale entropy analysis of
complex physiologic time series. Phys Rev Lett 89: 068102.
27. Richman JS, Moorman JR (2000) Physiological time-series analysis using
approximate entropy and sample entropy. Am J Physiol Heart Circ Physiol 278:
H2039–H2049.
28. Wu Z, Huang NE, Long SR, Peng CK (2007) On the trend, detrending, and
variability of nonlinear and nonstationary time series. Proc Natl Acad Sci U S A
104: 14889–14894.
29. Lo MT, Novak V, Peng CK, Liu Y, Hu K (2009) Nonlinear phase interaction
between nonstationary signals: a comparison study of methods based on HilbertHuang and Fourier transforms. Phys Rev E Stat Nonlin Soft Matter Phys 79:
061924.
30. Sanderson JE, Chan SK, Yip G, Yeung LY, Chan KW, et al. (1999) Betablockade in heart failure: a comparison of carvedilol with metoprolol. J Am Coll
Cardiol 34: 1522–1528.
31. Mortara A, La Rovere MT, Pinna GD, Maestri R, Capomolla S, et al. (2000)
Nonselective beta-adrenergic blocking agent, carvedilol, improves arterial
baroflex gain and heart rate variability in patients with stable chronic heart
failure. J Am Coll Cardiol 36: 1612–1618.
32. Sanderson JE, Yeung LY, Chan S, Tomlinson B, Kay R, et al. (1999) Effect of
beta-blockade on baroreceptor and autonomic function in heart failure. Clin Sci
(Lond) 96: 137–146.
33. Lin LY, Lin JL, Du CC, Lai LP, Tseng YZ, et al. (2001) Reversal of deteriorated
fractal behavior of heart rate variability by beta-blocker therapy in patients with
advanced congestive heart failure. J Cardiovasc Electrophysiol 12: 26–32.
34. Ho KK, Moody GB, Peng CK, Mietus JE, Larson MG, et al. (1997) Predicting
survival in heart failure case and control subjects by use of fully automated
methods for deriving nonlinear and conventional indices of heart rate dynamics.
Circulation 96: 842–848.
35. Makikallio TH, Hoiber S, Kober L, Torp-Pedersen C, Peng CK, et al. (1999)
Fractal analysis of heart rate dynamics as a predictor of mortality in patients with
depressed left ventricular function after acute myocardial infarction. TRACE
Investigators. TRAndolapril Cardiac Evaluation. Am J Cardiol 83: 836–839.
Heart Failure and Multi-Scale Entropy
(Second presenter) C2010
In infant and children with bradycardia that
unresponsive to oxygenation and/or ventilation
(P), does the use of atropine (I), as compared
with epinephrine (C), improve patient outcome
return to age-appropriate heart rate,
subsequent pulseless arrest, survival) (O)? Worksheet identifier: Peds-052(B)
Author:
Sasa Kurosawa, MD
Affiliation:
1) Research Inst., National Center for Child Health & Dev.
2) Shizuoka Children’s Hospital
Taskforce:
PEDS
Other Worksheet Authors: Susan Fuchs, Masahiko Nitta(042)
(Second Presenter) C 2010
Worksheet Specific Conflict
Of Interest Disclosure
!  Commercial/industry
!   none
!  Potential intellectual conflicts
!   none
(Second presenter) C2010
(Second presenter) C2010
Differences in supporting evidence
Differences in COS Statement and
Treatment Recommendation compared
to first presenter
Summary of evidence
Evidence Supporting Clinical Question
Good
Fair
!   I almost agree with first presenter (Susan Fuchs).
Fullerton,1991(E )
Poor
1
2
3
Level of evidence
4
5
A = Return of spontaneous circulation C = Survival to hospital discharge E = Other endpoint
B = Survival of event
D = Intact neurological survival Italics = Animal studies
denotes key article(s)
(Second presenter) C2010
Differences in Neutral evidence
(Second presenter) C2010
Differences in Opposing evidence
Summary of evidence
Summary of evidence
Evidence Neutral to Clinical Question
Evidence Opposing Clinical Question
Good
Tibballs, 2006 (ABCD)
Guay, 2004 (ABCD)
Reis, 2002 (ABCD)
Good
Brady, 1999, (E)
Smith, 1994, (E)
Chadda, 1977, (E)
Fair
Zimmerman,
1986(E)
Poor
1
2
3
Level of evidence
4
Fair
Coon, 1981
Poor
(AC)
5
1
A = Return of spontaneous circulation C = Survival to hospital discharge E = Other endpoint
B = Survival of event
D = Intact neurological survival Italics = Animal studies
denotes key article(s)
2
3
Level of evidence
4
5
A = Return of spontaneous circulation C = Survival to hospital discharge E = Other endpoint
B = Survival of event
D = Intact neurological survival Italics = Animal studies
denotes key article(s)
1
(Second presenter) C2010
(Second presenter) C2010
Differences in interpretation of Key studies
Differences in interpretation of Key studies
Key study #1 Reis, Pediatrics 2002
Key study #2 Tibballs, Resuscitation 2006
(Second presenter) C2010
Summary Of 3 In-hospital Arrest
Studies (Tibballs, Guay, Reis)
Study
# pts
Tibballs
111
Rhythm
73 hypotension
& bradycardia
Epi used
97/111
(Second presenter) C2010
Differences in interpretation of Key studies
Key study #3 Coon, Ann Emerg Med 1981
Survivors
38% at 1 yr
Guay
203
102 bradycardia
101/203
30% at 1 yr
Reis
129
43 bradycardia
86 arrest
(55% asystole)
107/129
(atropine in 26)
33% brady at 24 hr
21 total to hosp d/c
Table 3
PARAMETERS OF
THERAPEUTIC EFFECACY
Brady-Asystolic Rhythm
Control
Ca2+,
Steroid,
Isoproterenol
HCO3, Epinephrine
Intubation
ABGs
Pacemaker
Pressors
Atropine
No
Coltrol Atropine Rhythm
10/11 8/10
Pulse
6/11
4/10
Admit
2/11
2/10
Discharge 1/11 0/10
Atropine
(1-2mg)
New
Rhythm
! 
Appropriate Drug
Atropine may not affect the
outcome of prehospital bradyasystolic cardiac arrest of
primary cardiac etiology.
Fig.2. Protocol for treating brady-asystolic rhythm.
(Second presenter) C2010
Differences in interpretation of Key studies
Key study (Fuchs)
Horigome, Acta Paediatrica Japonica 1993
!   Atropine increased HR and mean blood pressure,
but had no effect on left ventricular end diastolic dimension
(LVEDD), or LV shortening fraction (LVSF).
!   Myocardial depression caused by halothane was not
improved by atropine
But I think if atropine increase HR and mBP,
it is effective to improve systemic circulation.
(Second presenter) C2010
Different Consensus on Science statements
!   Several adult studies and one pediatric study showed the efficacy of
atropine for bradycardia caused by vagal stimulation, AV block,
intoxication…(LOE 4, 5) [Brady, 1999, 47; Smith, 1994, 245;
Zimmerman, 1986, 320; Chadda, 1977, 503].
!   However, in pediatrics, bradycardia is mainly
caused by hypoxia poor perfusion.
Bradycardia is a pre-arrest state.
!   Large pediatric Utstein studies demonstrated the
significant frequency of hypotension-bradycardia,
in pediatrics (LOE 4) [Guey, 2004, 373; Tibballs,
2006, 310; Reis, 2002, 200].
In this context, epinephrine is the first line drug
rather than atropine.
2
(Second presenter) C2010
Different Treatment Recommendations
!   There is no supportive evidence to demonstrate that
atropine is superior to epinephrine in the case of
bradycardia unresponsive to oxygenation and/or
ventilation (excluding the case with vagal stimulation).
!   Epinephrine is the first line drug for poorly perfused infants and
children with bradycardia (heart rate <60 beats/min) that is
unresponsive to oxygenation and/or ventilation.
The first thing start chest compressions!
3
Page 1 of 21
Date Submitted for review: December 25, 2009
Nitta (042B):
Databases searched: MEDLINE, Cochrane, ECC EndNote Master Library
Previous 2005 WS related to this question was reviewed and added several hand search.
#1 "Heart Arrest"[Mesh] AND "Atropine"[Mesh]
#2 "bradycardia"[Mesh] AND "Atropine"[Mesh]
#3 #1 or #2 advanced search child: 0-18 years
#4 #2 and (symptomatic bradycardia or resuscitation or shock or cardiopulmonary failure)
#10 What is the optimal drug therapy for significant bradycardia?(WS91 Lim Swee Han, 20 Oct 2004) → 2 articles(#6,
#7)
#11 Additional hand search → 2 articles(#8, #9)
#7 "Atropine"[Mesh] AND "Bradycardia"[Mesh] AND "Child"[Mesh] → hits 49
#8 "Epinephrine"[Mesh] AND "Bradycardia"[Mesh] AND "Child"[Mesh] → hits 5
#9 #7 OR #8 → hits 53, but there is no appropriate article.
#1 Bradycardia, Atropine, Child → hits 122
#2 Bradycardia, Epinephrine, Child → hits 25
#3 #1 OR #2 → hits 136 → 1 article(#1)
#4 Bradycardia, Resuscitation, Atropine → hits 81
#5 Bradycardia, Resuscitation, Epinephrine → hits 72
#6 #4 OR #5 → hits 119 → 4 article(#2 - #5)
Kurosawa (052B): Databases searched: Medline, ECC EndNote Master Library
Previous 2005 WS related to this question was reviewed and added several hand search.
Hand searching references of articles
Review of 2005 ILCOR worksheets on similar topic
Recommendation from Dr Morley
Search strategy (including electronic databases searched).
Fuchs (052A):
Embase: bradycardia, atropine, epinephrine: 47 articles- none useful
Embase June 4, 2009: bradycardia, epinephrine, atropine : 11 articles from combined search-none useful, 21 from
bradycardia and epinephrine- 1 useful-had already, 108 from bradycardia and atropine-1 useful-had already
Pubmed: bradycardia (mesh), atropine(mesh), atropine derivative(mesh), epinephrine(mesh) 31 articles results- none
were useful
Cochrane library: bradycardia (mesh), epinephrine (mesh), atropine (mesh): 97 articles, 1 useful (epinephrine for the
resuscitation of apparently stillborn or extremely bradycardic newborn infants AJA Zino, MW Davies, PG Davis Cochrane
Database of Systemic Reviews 2002, Issue 3, Art no: CD003849.
Cochrane library June 4, 2009- no additional information that previously
Pubmed: bradycardia (mesh), atropine(mesh), atropine derivative(mesh), epinephrine(mesh) : 69 articles, 5 possibly
useful
Is this question addressing an intervention/therapy, prognosis or diagnosis? Intervention
State if this is a proposed new topic or revision of existing worksheet:
Conflict of interest specific to this question
Do any of the authors listed above have conflict of interest disclosures relevant to this worksheet? No
Nitta: In infants and children with cardiac arrest (out-of-hospital and in-hospital) or symptomatic bradycardia (P), does the
use of atropine (I) compared with no atropine use (C), improve outcome (O) (eg. ROSC, survival)?
Clinical question.
Fuchs and Kurosawa: In infant and children with bradycardia that unresponsive to oxygenation and/or ventilation(P),
does the use of atropine(I), as compared with epinephrine(C), improve patient outcome(return to age-appropriate heart
rate, subsequent pulseless arrest, survival)(O)?
Worksheet author(s)
Susan Fuchs, Sasa Kurosawa, Masahiko Nitta
WORKSHEET for Evidence-Based Review of Science for Emergency Cardiac Care
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
Page 2 of 21
#5 "Heart Arrest"AND "Atropine"
#6 "symptomatic bradycardia" AND "Atropine"
#7 #5 or #6 advanced search child: 0-18 years
#8 Additional hand search
• State inclusion and exclusion criteria
Fuchs: None
Kurosawa: Included all human clinical trials, meta-analysis, case reports, guidelines.
English only.
Review articles and animal experiments were excluded.
Nitta: The following studies were excluded: Not true cardiac arrest cases.
Included all human clinical trials, meta-analysis, and animal experiments. English Language limitations.
Review articles were excluded.
• Number of articles/sources meeting criteria for further review:
Fuchs :12
Kurosawa: 9 articles relevant to the topic (from 308 hits and previous WS) were reviewed in detail
Nitta: • Number of articles/sources meeting criteria for further review:
#3 9 articles relevant to the topic (from 123 hits)
#1 17 articles relevant to the topic (from 183 hits)
#4 4 articles relevant to the topic (from 57 hits)
#7 9 articles relevant to the topic (from 85 hits)
#8 2 articles relevant to the topic
#3 or #1 or #4 or #6 or #8 22 articles relevant to the topic and 12 articles were reviewed in detail
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
Summary of evidence
Page 3 of 21
1
1
A = Return of spontaneous circulation
B = Survival of event
* Pediatric Study
Poor
Fair
Good
2
C = Survival to hospital discharge
D = Intact neurological survival
3
Level of evidence
Meaney (C)
Reis (B)
2
C = Survival to hospital discharge
D = Intact neurological survival
Lienhart A *
Zimmerman (E)
McNamara (E)
Polin (E)
Sacchetti (e)
Rothrock (E)
3
4
Level of evidence
5
Brown –A,B,C,E
(sinus rhythm)
Stuevan A
Sorensen (A)
Stiell (B)
Blecic (AE)
E = Other endpoint
Italics = Animal studies
5
Iseri AC
Steuvan C
Yilmaz (E)
Redding (A)
Chow E (histochemical
studies)
Brady (B,E)
Smith (E)
Chadda (E)
Oshige (BE)
Niemann (A)
Angelos A
Kaplan B
McCaul B
E = Other endpoint
Italics = Animal studies
4
Fullerton E
(prevent
recurrence of
bradycardia) *
Thrush (E)
Evidence Neutral to Clinical question
A = Return of spontaneous circulation
B = Survival of event
* Pediatric Study
Poor
Fair
Good
Evidence Supporting Clinical Question
Studies in Bold-Fuchs
Studies in Normal font-Kurosawa
Studies in Italics- Nitta
Studies in Bold and Underlined-Fuchs and Kurosawa
Studies in Bold & Italics- Fuchs and Nitta
Studies in Italics and underlined-Kurosawa and Nitta
Only study cited in all 3 (bold, italics and underlined)-Coon
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
1
Horigome E *
A = Return of spontaneous circulation
B = Survival of event
* Pediatric Study
Poor
Fair
Good
Page 4 of 21
2
C = Survival to hospital discharge
D = Intact neurological survival
3
Level of evidence
Herlitz 1994
(BCE)
Herlitz 2003 (BCE)
Engdahl (BCE)
5
DeBehnke (A)
Coon (ABCE)
E = Other endpoint
Italics = Animal studies
4
Tibballs (ABCD)
Guay (ABCD)
Reis (ABCD)
Evidence Opposing Clinical Question
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
Page 5 of 21
There were only 2 studies that opposed the use of atropine. The Coon study (LOE 5) evaluated atropine after
epinephrine (and other drugs).Twenty one prehospital adults who suffered cardiac arrest and were then in asystole or
pulseless idioventricular rhythm (PIVR) were studied. There was no difference between the 2 groups (control n=11) and
atropine (N=10) with respect to rhythm changes, pulse, admission or hospital discharge, so atropine did not improve
outcome. The other study to oppose the use of atropine was a randomized study by Horigome (LOE1), which involved
34 children undergoing halothane anesthesia for minor surgery. Echocardiographic recordings and indices were before,
after induction of general anesthesia, and after IV atropine administration (either 0.01 mg/kg or 0.02 mg/kg). Following
halothane anesthesia, heart rate and mean blood pressure fell from preinduction levels in both groups. The left systolic
time interval ration (LSTI), and the left ventricular end-diastolic dimension (LVEDD/preload) were reduced, and the left
ventricular shortening fraction (LVSF) was increased in both groups. After injection of atropine, the heart rate increased
above baseline (more for the 0.02 mg/kg group), the mean blood pressure returned to baseline in both groups. The
other parameters LSTI, LVEDD and LVSF remained unchanged in both groups, but the LSTI and LVEDD were larger
and the LVSF smaller than pre-induction. What this research demonstrated was that atropine can counteract the
bradycardia caused by halothane anesthesia, but myocardial contractility remains depressed. There was no advantage
Most of the studies were neutral with respect to the question. Angelos (LOE 5) evaluated ROSC in a rat model utilizing
epinephrine vs placebo, and found that as the duration of cardiac arrest increased, epinephrine is more important to
attain ROSC, but there was greater post-ROSC myocardial,depression. Brady (LOE 5) was a study involving adult
patients with hemodynamically compromising bradycardia or AVB with evidence of spontaneous circulation who
received atropine delivered by EMS personnel, and in the hospital. 86 patients had bradycardia and approximately onethird of patients who received atropine in the prehospital setting for bradycardia had either a partial or complete
response to therapy. The Chow study (LOE 5) was a histiochemical and immunochemical analysis of the autonomic
innervation of the human cardiac conduction system that demonstrated initial sympathetic dominance in infancy with
gradual transition to sympathetic and parasympathetic innervation in adulthood. Kaplan (LOE 5) compared BG-9719, a
selective A1AdoR antagonist to atropine or saline in a vagotomized, hypoxic rat model. Atropine and saline did not
prolong survival or attenuate posthypoxic decreases in heart rate compared to BG-9719. McCaul (LOE 5) evaluated a
brief asphyxial arrest rat model utilizing 10 or 30 mcg/kg of epinephrine vs saline, and demonstrated that epinephrine
increased mortality, especially with the higher dose. The retrospective study by Iseri (LOE 5) evaluated adult patients in
the prehospital setting in ventricular fibrillation or extreme bradycardia or asystole. Epinephrine or isoproterenol was
given to 14 bradycardic patients, and atropine in 12 patients (either as treatment 1, 2 or 3), but results were inconsistent
and only 1 patient survived to hospital day 12. Leinhart (LOE 4) was a case report on a 20 month old infant who
suffered a near drowning. The first dose of epinephrine and atropine has no effect, but another dose of epinephrine and
vasopressin resulted in ROSC.
The only pediatric study to support the use of atropine involved 4 infants who had undergone cardiac surgery (Fullerton
LOE 4). Three experienced hypotension, followed by bradycardia and asystole. Two did not respond to epinephrine, but
all responded to atropine. One other infant had episodes of hypotension without bradycardia, and responded to
atropine. It was thought that these episodes were secondary to the Bezold-Jarisch reflex. This reflex is initiated by
activation of mechanical stretch receptors located mainly in the wall of the left ventricle, with afferents carried by the
right vagus nerve. Activation of this reflex results in hypotension and bradycardia mediated by cholinergic vasodilatation
and withdrawal of sympathetic tone. The continuous administration of atropine in all cases prevented the recurrence of
this reflex in these patients. This concept may support the use of atropine in vagally mediated bradycardia, as the
current PALS bradycardia algorithm suggests.
The studies that were supportive of atropine in some aspect include the case series by Brown (LOE 5) which involved 8
adults in cardiac arrest, all whom were in asystole as the initial rhythm or as a result of defibrillation. Six failed to
respond to epinephrine, but responded to atropine. And 50% of these patients were discharged from the hospital. In the
prehospital study by Stueven (LOE 5), which was a retrospective review, there were 84 adult patients in asystole who
remained in asystole after epinephrine (and sodium bicarbonate). Of these patients, 43 also received atropine, and 6
had successful resuscitation, which was defined as conveyance of a patient with a rhythm and pulse to an emergency
department vs 0/41 of the epinephrine, sodium bicarbonate group, (P<.04). However, none of the patients who received
atropine survived to hospital discharge. Therefore, this study is neutral with respect to survival to hospital discharge.
Fuchs: There were no randomized controlled, quasi-randomized controlled trials or any study in children which looked at
the specific question: In infants and children with bradycardia that is unresponsive to oxygenation and/or ventilation,
dose the use of atropine as compared to epinephrine improve patient outcome.
REVIEWER’S FINAL COMMENTS AND ASSESSMENT OF BENEFIT / RISK:
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
Page 6 of 21
There are one adult (Sorensen, LOE 5) and two pediatric case series (Fullerton, Thrush, both LOE 4), evaluating
treatment of bradycardia presumed to be caused by vagal stimulation. Four cardiac surgical patients exhibited
From the National Registry of Cardiopulmonary Resuscitation (NRCPR), Meany. (LOE 3) reported 464 pediatric ICU
arrests The odds ratio (95% CIs) for survival to discharge with atropine was 2.38 (1.20-4.74) using multivariate logistic
regression analysis. In a second NRCPR study, (Reis 2002) 129 pediatric patients who received cardiopulmonary
resuscitation for cardiac arrest or symptomatic bradycardia were studied. The relative risk of death at 24 hours with
atropine was 0.98 (0.58-1.90) using multivariate logistic regression analysis. These results suggest that atropine may be
effective in cardiac arrest or resuscitated symptomatic bradycardia. But it was not clear whether atropine was given
while CPR in progress.
Stiell (LOE 5) reported a large observational cohort study of cardiac arrest in adult cases. 529 adult patients who
suffered in-hospital and out-of-hospital cardiac arrest were studied. The odds ratio (95% CIs) for successful
resuscitation with atropine, after multivariate adjustment for potential confounders, was 1.2 (1.0-1.3).
There are few articles related to this topic in pediatrics which are supportive of atropine. The Stueven and Brown studies
are mentioned above. Using a prospective, controlled, blinded, adult canine model of PEA induced by ventricular
fibrillation followed by external defibrillation, Blecic (LOE 5) demonstrated that atropine increased survival and
shortened CPR time compared to the control group treated with D5W. In treatment group, 91% (10/11) achieved of
ROSC with atropine, but only 67% (8/12) with 5% dextrose, control. Additionally, the atropine treated group had higher
arterial pressure, heart rate, cardiac output, stroke volume and decreased time to recovery. The administration of
atropine with epinephrine enhanced the recovery in this experimental canine model.
Nitta: Symptomatic bradycardia is the most common initial rhythm in pediatric cardiac arrest (asphyxial arrest).
Epinephrine is the first line drug in this setting. In CoSTR 2005, there was no evidence to eliminate atropine for asystole
in pediatrics. Atropine for asystole in adults was reviewed in two work sheets (W97A, W97B).
There are some Utstein reports in pediatrics. In these reports (Tibballs, Guay, Reis, all LOE 4), the percentage of
hypotensive bradycardia in all patients who received cardiopulmonary resuscitation was relatively high. And most of
these patients received epinephrine rather than atropine during resuscitation. Tibballs (LOE 4) Resuscitation was
attempted in total of 111 cases. Of these patients who had cardiac arrests, hypotensive-bradycardia was identified in 73
cases, which is two third of all cases, and epinephrine was used in around 90% (97/111) of all cases as the first line
drug. Guay (LOE 4) Of all 203 cases, initial rhythm was bradycardia in 102 cases (50%), and 101 cases(50%) were
received epinephrine. There was no mention of administration of atropine. Reis (LOE 4) Chest compressions and
assisted ventilation were provided for total of 129 patients. Of all cases, bradycardia with hypoperfusion, unresponsive
to oxygenation and ventilation, was noted in 43 cases which is one third of all cases. Most of the patients received
epinephrine rather than atropine during resuscitation which is almost over 80% of all cases. Even in adults, one study
(see Coon LOE 5 above) shows that atropine may not superior to epinephrine in pre-arrest state.
Kurosawa: Some adult studies (Smith, Yilmaz ,Brady, Chadda, all LOE 5) and a pediatrics study ( Zimmerman LOE 4)
indicated that iv atropine improved heart rate in symptomatic bradycardia. Atropine is effective for vagal stimulation, AV
block, intoxication, etc.
Zimmerman (LOE 4). As common causes of intraoperative bradycardia are hypoxia or vagal activity, such therapy
should consist of ventilation with oxygen and iv atropine. Yilmaz (LOE 5) 66 patients were hospitalized with a variety of
symptoms (nausea, vomiting, salivation, dizziness, weakness, hypotension, bradycardia and syncope) several hours
after the ingestion of small amount of honey. All patients had hypotension, and majority had bradycardia. These
features resolved completely in 24h with iv fluids and atropine. Brady (See above). Smith (LOE 5) 64 adult patients who
received general anesthesia (sufentanyl + N2O + vecuronium) were allocated randomly to received either atropine,
5μg/kg(Group 1), glycopyrrolate, 2.5μg/kg(Group 2), or transesophageal atrial pacing (Group 3) after the onset of
bradycardia, defined as a heart rate of < 50bpm ( or < 60bpm with concurrent of hypotension). Bradycardia occurred in
15 patients of each group. The therapeutic response was significantly more rapid in Group 3. But the therapy was
effective in all groups and there is no great difference in incidence of side effects between the three groups. Chadda
(LOE 5) The effect of 0.4 to 1.5 mg intravenously administered atropine were evaluated in 100 adult patients with a
heart rate < 60/min following acute myocardial infarction. There was a statistically significant correlation between the
dose of atropine and the increment in heart rate.
of a dose of 0.02 mg/kg over 0.01 mg/kg. While this does not answer the question about if epinephrine is better than
atropine in bradycardia, it is known that epinephrine increases the heart rate and myocardial contractility, so it could be
of more benefit in bradycardia unresponsive to oxygenation and ventilation.
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
042: There is extremely limited data on the use of atropine in pediatric cardiac arrest, and poor quality data supporting
its use in adult cardiac arrest. Evidence from two LOE 3 studies of in-hospital pediatric cardiac arrest [Meany 2006,
2424, Reis 2002, 200] showed some improvement in survival to discharge, but another demonstrated no decrease in
risk of death. Evidence from one LOE 4 study [Fullerton, 1991, 534] and one LOE 5 study in adults [Brown, 1979, 448]
demonstrated the some efficacy of atropine in symptomatic bradycardia requiring resuscitation. Evidence from adults
studies, one level 5 study [Stueven, 1984, 815], one level 5 study [Brown, 1979, 448] and one level 5 animal study
[Blecic, 1992, 515] demonstrated the efficacy of atropine in cardiac arrest (ROSC). Evidence from one level 5 animal
study [DeBehnke, 1995, 1034] demonstrated no improvement with atropine in canine asphyxial cardiac arrest.
Large pediatric Utstein studies demonstrated the significant volumes of hypotensive-bradycardia, which is indicating this
specific feature in pediatrics (LOE 4) [Tibballs, 2006, 310; Guay, 2004, 378; Reis, 2002, 200]. In this context,
epinephrine is the first line drug rather than atropine. Even in adults, one study shows that atropine may not superior
than epinephrine in pre-arrest state (LOE 5)(Coons, 1981, 462).
Evidence from several adult studies (LOE 5: Brady 1999,47, Chow 2001,169, Coon 1981,462, Iseri 1978,741, Steuven
1984,815), animal studies (LOE 5: Angelos 2008,101, Kaplan 2003,923, McCaul 2006,102), and 1case series on a child
(LOE 4: Lienhart 2005,486) demonstrate no benefit from the use of atropine after epinephrine. One randomized
pediatric study (LOE 2: Horigome 1993,513) demonstrated that atropine only increased heart rate and mean blood
pressure, but did not improve myocardial depression induced by halothane anesthesia. One LOE 5 adult study (Coon
1981,462) also demonstrated no benefit of atropine over epinephrine with respect to ROSC, survival or event or hospital
discharge.
Several adult studies and one pediatric study showed the efficacy of atropine for bradycardia caused by vagal
stimulation, AV block, intoxication (LOE 5) Yalmaz, 2006, 405; Brady, 1999, 47; Smith, 1994, 245; Zimmerman, 1986,
320; Chadda, 1977, 503). One pediatric case series (LOE 4: Fullerton 1991,534) demonstrated a beneficial effect of
atropine after epinephrine for children who developed hypotension after cardiac surgery (Bezold-Jarisch refllex
mediated).
CONSENSUS ON SCIENCE:
052: Evidence from 2 adults studies (LOE 5, Brown 1979,448-adult case series, Stueven 1984,815-adult retrospective
control), demonstrated a change of cardiac rhythm from asystole after atropine, when there had been no change after
epinephrine, however survival to hospital discharge was no different.
Conclusion
In pediatric patients, hypoxemia, hypothermia, acidosis, hypotension, hypoglycemia, central nervous system insults and
excessive vagal stimulation may produce symptomatic bradycardia and asystole. Asystole can be exacerbated by
excessive vagal tone and the administration of atropine is reasonable for its physiological effects.
Two cohort studies of out-of-cardiac arrest showed no evidence that treatment with atropine increase the chance
of survival among asystolic patients (Engdahl, 2000) and cardiac arrest patients (Herlitz, 2003). Although these studies
included pediatric patients in cardiac arrest, they were not separated in the analysis.
Citation List
Atropine can have a place during cardiopulmonary resuscitation (CPR) in the management of asystole, where
parasympathetic influence might be excessive. However, the beneficial effects of atropine in electromechanical
dissociation (EMD) have not been clearly demonstrated. The authors studied the effects of atropine in combination with
epinephrine on an experimental model of EMD in the closed-chested dog. In 15 pentobarbital-anesthetized, mechanically
ventilated dogs (mean weight 20 kg), EMD was induced by ventricular fibrillation followed by an external countershock,
and was observed for 2 minutes before CPR was started. After 5 minutes of chest compression using a CPR thumper,
either atropine 0.5 mg or D5W was administered, and the same injection was repeated every 5 minutes until recovery.
Epinephrine 1 mg was administered in alternans. Each dog was submitted to two successive episodes of CPR, using
either atropine or D5W, in a randomized order. Of a total of 28 CPRs, five were successful with chest compression alone.
In the treatment groups, 10 of 11 were successful with atropine, but only eight of 12 with D5W (P < .01). The duration of
Blecic, S., C. Chaskis, et al. (1992). "Atropine administration in experimental electromechanical dissociation." Am J
Emerg Med 10(6): 515-518.
Comments: Level 5; good. Supportive
Objective: Epinephrine (adrenaline) is widely used as a primary adjuvant for improving perfusion pressure and
resuscitation rates during cardiopulmonary resuscitation (CPR). Epinephrine is also associated with significant
myocardial dysfunction in the post-resuscitation period. We tested the hypothesis that the cardiac effects of epinephrine
vary according to the duration of cardiac arrest. Methods and materials: Cardiac arrest (CA) was induced in SpragueDawley rats with an IV bolus of KCl (40 mug/g). Three series of experiments were performed with CPR begun after 2, 4,
or 6 min of cardiac arrest. Epinephrine (0.01 mg/kg) IV or placebo was given immediately in the 2 and 4 min CA groups.
In the 6 min group, CPR was started after 6 min CA and epinephrine was given at 15 min if no return of spontaneous
circulation (ROSC) occurred. Time to ROSC was recorded in all groups. Cardiac function was determined with transthoracic echocardiography at baseline, 5, 30 and 60 min after ROSC. Results: After 2 min CA, 8/8 (100%) placebo
animals and 8/8 (100%) epinephrine animals attained ROSC. Cardiac index was significantly increased during the first
60 min in the epinephrine group compared with the placebo group (p < 0.01). After 4 min of cardiac arrest, 14/29 (48%)
placebo animals and 14/16 (88%) epinephrine animals attained ROSC (p < 0.01). Cardiac index after ROSC returned to
baseline in both groups, although tended to be lower in the epinephrine group. After 6 min CA, 10/31 (32%) animals
attained ROSC without epinephrine and 17/21 (81%) animals with epinephrine (p < 0.01). Post-ROSC depression of
cardiac index was greatest in the epinephrine group (p < 0.05). Conclusions: As the duration of cardiac arrest increases,
a paradoxical myocardial epinephrine response develops, in which epinephrine becomes increasingly more important to
attain ROSC, but is increasingly associated with post-ROSC myocardial depression.
Summary: Rat model study; epi vs placebo-as duration of cardiac arrest increases, a paradoxical myocardial epinephrine
response develops, in which epi becomes more important to attain ROSC, but is associated with increasing post-ROSC
myocardial depression.
Angelos MG, Butke RL, Panchal AR et al. Cardiovascular response to epinephrine varies with increasing duration of
cardiac arrest. Resuscitation 2008; 77:101-110.
LOE 5, neutral. Animal (rat) study Quality of evidence good.
Sponsored by Roessler Scholarship Fund, Ohio State Univ, SAEM Institutional Training Award, Ohio State Initiatives
Grant and The American Heart Association, Ohio Affiliate.
Acknowledgements:
042: Nitta (Atropine vs no atropine): There is insufficient evidence to support or refute the use of atropine for
symptomatic bradycardia or pediatric cardiac arrest.
If bradycardia is caused by increased vagal tone, cholinergic drug toxicity, or primary AV block, administer atropine
rather than epinephrine.
Epinephrine is the first line drug for poorly perfused infants and children with bradycardia (heart rate <60 beats/min) that
is unresponsive to oxygenation and/or ventilation.
Page 8 of 21
There are some studies which do not support the use of atropine The Coon study is mentioned above. A prospective,
controlled, blinded canine asphyxial PEA model; pediatric asphyxial cardiac arrest model was used by
DeBehnke(1995). After 10 minutes untreated PEA, the animals were block randomized to one of five groups and each
group was treated with different dose of atropine. The standard dose of atropine did not improve ROSC rate compared
with control group. Increasing dose of atropine tended to decrease ROSC rates compared with control group and
standard dose group.
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
TREATMENT RECOMMENDATION:
(052A: Combined Fuchs and Kurosawa): There is no supportive evidence to demonstrate that atropine is superior to
epinephrine in the case of bradycardia unresponsive to oxygenation and/or ventilation (excluding those with vagal
mediation).
Page 7 of 21
cardiovascular collapse in early postoperative course and were resuscitated. The vaso-vagal reflex was suspected and
treated with atropine (Fullerton). Severe hypertension with reflex bradycardia progressed to cardiac arrest caused by
drug and treated with atropine and cardiopulmonary resuscitation (Thrush). These studies support atropine use for
pediatric in-hospital asystolic cardiac arrest or symptomatic bradycardia caused by increased vagal activity.
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
Page 9 of 21
Chadda KD, Lichstein E, et al. Effects of atropine in patients with bradyarrhythmia complicating myocardial infarction.
Usefulness of an optimum dose for overdrive. Am J Med 1977; 63(4): 503-10.
Parasympathetic tone may be high during ventricular asystole because of reflex vagal stimulation from a number of
sources. Eight patients in cardiac arrest were treated with cardiopulmonary resuscitation. All eight patients had
ventricular asystole as the initial rhythm or as the result of defibrillation. Six patients failed to respond to 5 cc to 20 cc of
1:10,000 epinephrine intravenously (IV). In all eight cases a regular rhythm (sinus in seven, idioventricular in one)
appeared within 30 seconds of administration of the last dose of atropine (1 mg to 2 mg IV). Five patients (62.5%) lived
12 hours, three (37.5%) were discharged from the hospital. These results suggest that atropine may be of value in the
treatment of ventricular asystole.
Summary: Case series in adults atropine better than epinephrine
Brown DC, Lewis AJ, Criley JM. Asystole and its treatment: The possible role of the parasympathetic nervous system in
cardiac arrest. JACEP1979; 8: 448-452.
LOE 5, atropine better than epinephrine. Adult study. Quality of evidence poor.
No industry funding.
OBJECTIVE: To determine the efficacy of atropine therapy in patients with hemodynamically compromising bradycardia
or atrioventricular block (AVB) in the prehospital and emergency department settings. METHODS: DESIGN:
Retrospective review of prehospital, emergency department, and hospital records. PARTICIPANTS: Prehospital patients
with hemodynamically compromising bradycardia or AVB with evidence of spontaneous circulation who received atropine
as delivered by emergency medical services personnel (advanced life support level). SETTING: Urban/suburban fire
department-based emergency medical service system with on-line medical control serving a population of approximately
1.6 million persons. DEFINITIONS: Hemodynamic instability was defined as the presence of any of the following:
ischemic chest pain, dyspnea, syncope, altered mental status, and systolic blood pressure less than 90 mmHg.
Bradycardia was defined as sinus bradycardia, junctional bradycardia, or idioventricular bradycardia (grouped as
bradycardia) while AVB included first-, second- (types I and II), or third-degree (grouped as AVB). The response that
occurred within one minute following each dose of atropine was defined as none, partial, complete, or adverse. MAIN
RESULTS: Of 172 patients meeting entry criterion complete data was available for 131 (76.1%) and constitutes the
study population. The mean age was 71 years. Fifty-one percent were female. Forty-five patients had AVB and 86
bradycardia. Patients with AVB were more likely to have a presenting systolic blood pressure less than 90 mmHg than
those with bradycardia. In the 131 patients, responses to atropine were as follows: 26 (19.8%) = partial, 36 (27.5%) =
complete, 65 (49.6%) = none, and 4 (2.3%) = adverse. Patients presenting with bradycardia (compared to AVB) more
commonly: (1) received a single dose of atropine; (2) a lower total dose of atropine in the prehospital interval; (3) were
more likely to arrive in the ED with a normal sinus rhythm; and (4) were less likely to receive additional atropine or
isoproterenol in the ED. Those patients who achieved normal sinus rhythm over the total course of care were likely to
have achieved that rhythm during the prehospital interval. There was no difference between groups in the likelihood of
leaving the ED with a normal sinus rhythm achieved during the ED interval. Acute myocardial infarction was more
common in patients presenting with AVB (55.5%) than with bradycardia (23.2%, P = 0.001). CONCLUSIONS:
Approximately one-half of patients who received atropine in the prehospital setting for compromising rhythms had either
a partial or complete response to therapy. Adverse responses were uncommon. Those patients who presented with
hemodynamically unstable bradycardia to EMS personnel responded more commonly to a single dose and a lower total
dose of atropine compared to similar patients with AVB. Those patients who achieve normal sinus rhythm by ED
discharge were likely to have achieved it during the prehospital interval.
Summary: Retrospective review of adult patients with hemodynamically compromising bradycardia or AVB with evidence
of spontaneous circulation; approx 50% had complete or partial response to atropine
Brady WJ, Swart G, DeBehnke DJ et al:. The efficacy of atropine in the treatment of hemodynamically unstable
bradycardia and atrioventricular block: prehospital and emergency department considerations. Resuscitation 1999;41:4755.
LOE 5, neutral. Adult population. Quality of evidence fair.
No industry funding.
CPR was also significantly shorter when atropine was used (9 minutes 56 seconds (plus or minus) 14 seconds versus 12
minutes 08 seconds (plus or minus) 43 seconds, P < .001). During the recovery period, atropine-treated animals had
higher arterial pressure, heart rate, cardiac output and stroke volume. On this experimental model, the administration of
high doses of atropine together with epinephrine enhances the recovery from EMD and results in a better cardiac
function during recovery.
Summary: This study is small prospective, controlled, blinded canine coutershock PEA model with higher dose of
atropine (0.05 mg/kg) and adrenaline.
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
Page 10 of 21
The efficacy of atropine in treating prehospital cardiac arrest patients developing asystole slow pulseless idioventricular
rhythms (PIVR) was evaluated in a controlled, prospective study. Twenty-one prehospital cardiac-arrested patients
developing asystole or PIVR (less than 40) were divided into atropine-treated or non-atropine (control) groups. Control
group patients received treatment including bicarbonate, epinephrine, calcium, isoproterenol, dexamethasone, and
transthoracic pacing. Atropine-treated patients received 1 mg atropine intravenously with a repeat dose at one minute if
no rhythm change occurred. These patients then received the same therapy as the control group. In both groups, rhythm
changes were treated as appropriate for the specific circumstances. No differences in mortality or effected rhythm
changes were observed. Ten of the 11 controls and eight of 10 atropine patients developed rhythms other than asystole
Coon GA, Clinton JE, Ruiz E. Use of atropine for brady-asystolic prehospital cardiac arrest. Ann Emerg Med
1981;10:462-467.
LOE 5: Controlled, prospective study in adults, neutral with respect to change in rhythm or survival.
No industry funding. Quality of evidence good
In order to study the changes in the pattern of autonomic innervation of the human cardiac conduction system m relation
to age, the innervation of the conduction system of 24 human hearts (the age of the individuals ranged from newborn to
80 years), freshly obtained at autopsy, was evaluated by a combination of immunofluorescence and histochemical
techniques. The pattern of distribution and density of nerves exhibiting immunoreactivity against protein gene product 9.5
(PGP), a general neural marker, dopamine beta-hydroxylase (DBH) and tyrosine hydroxylase (TH), indicators for
presumptive sympathetic neural tissue, and those demonstrating positive acetylcholinesterase (AChE) activity, were
studied. All these nerves showed a similar pattern of distribution and developmental changes. The density of innervation,
assessed semiquantitatively, was highest in the sinus node, and exhibited a decreasing gradient through the
atrioventricular node, penetrating and branching bundle, to the bundle branches. Other than a paucity of those showing
AChE activity, nerves were present in substantial quantities in infancy. They then increased in density to a maximum in
childhood, at which time the adult pattern was achieved and then gradually decreased in density in the elders to a level
similar to or slightly less than that in infancy. In contrast, only scattered AChE-positive nerves were found in the sinus
and atrioventricular nodes, but were absent from the bundle branches of the infant heart, whereas these conduction
tissues themselves possessing a substantial amount of pseudocholinesterase. During maturation into adulthood,
however, the conduction tissues gradually lost their content of pseudocholinesterase but acquired a rich supply of AChEpositive nerves, comparable in density to those of DBH and TH nerves. The decline in density of AChE-positive nerves in
the conduction tissues in the elders was also similar to those of DBH and TH nerves. Our findings of initial sympathetic
dominance in the neural supply to the human cardiac conduction system in infancy, and its gradual transition into a
sympathetic and parasympathetic codominance in adulthood, correlate well with the physiologic alterations known to
occur in cardiac rate during postnatal development. The finding of reduction in density of innervation of the conduction
tissue with ageing is also in agreement with clinical and electrophysiological findings such as age-associated reduction in
cardiac response to parasympathetic stimulation. Finally, our findings also support the hypothesis that, in addition to the
para-arterial route, the parafascicular route of extension along the conduction tissue constitutes another pathway for the
innervation of the conduction system of the human heart during development
Summary: An interesting study which demonstrated increased sympathetic tone in infant hearts, as compared to
child/adolescent and adult hearts which have increased parasympathetic tone.
Cheung Chow TS, Ming Chow SS, Anderson RH at al. Autonomic innervation of the human cardiac conduction system:
changes from infancy to senility-An immunohistochemical and histochemical analysis. Anat Rec 2001;264:169-182.
LOE 5: Neutral. Mechanical model/histochemical analysis. Quality of evidence fair.
No industry funding.
The effects of 0.4 to 1.5 mg intravenously administered atropine were evaluated in 100 patients with a heart rate <60/min
following acute myocardial infarction. Prior to the administration of atropine the mean heart rate was 45 + 7/min, the
incidence of ventricular premature beats was 12 + 4/min, and two patients had angina Based on the maximum heart rate
response to atropine, the patients were divided into two groups. Heart rate increased to <100/min in 70 patients (group 1)
and >100/min in 21 patients (group II) Ventricular premature beats decreased significantly (p<0.01) in both groups (2 +
4/min in group I and 2.4 + 0.8 in group II). In only one patient in group II did ventricular premature beats first appear
following the administration of atropine. Blood pressure did not change significantly after the administration of atropine in
either group. In two patients (group I), angina disappeared and in two others (group II) angina appeared following the
administration of atropine. Dose-heart rate analysis revealed that 70per cent of the patients in group II and only 17
percent of those in group I received >0.8 mg atropine (p<0.01). There was a statistically significant correlation between
the dose of atropine and the increment in heart rate (r=0.41 p<0.01). Thus, a relatively lower dose of atropine (<0.8 mg)
had a beneficial antiarrhythmic effect. Higher doses caused an inappropriate tachycardia, angina and ventricular
premature beats and should be avoided.
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
Fullerton DA, St Cyr JA, Clarke DR, et al; Bezold-Jarisch reflex in postoperative pediatric cardiac surgical patients Ann
Thorac Surg 1991;52:534-536.
LOE 4, Case series. Atropine better than epinephrine. Quality of evidence poor.
No industry funding
We describe the epidemiology, prognosis, and circumstances at resuscitation among a consecutive population of
patients with out-of-hospital cardiac arrest (OHCA) with asystole as the arrhythmia first recorded by the Emergency
Medical Service (EMS), and identify factors associated with survival. We included all patients in the municipality of
Goteborg, regardless of age and etiology, who experienced an OHCA between 1981 and 1997. There were a total of
4,662 cardiac arrests attended by the EMS during the study period. Of these, 1,635 (35%) were judged as having
asystole as the first-recorded arrhythmia: 156 of these patients (10%) were admitted alive to hospital, and 32 (2%) were
discharged alive. Survivors were younger (median age 58 vs 68 years) and had a witnessed cardiac arrest more often
than nonsurvivors (78% vs 50%). Survivors also had shorter intervals from collapse to arrival of ambulance (3.5 vs 6
minutes) and the mobile coronary care unit (MCCU) (5 vs 10 min), and they received atropine less often on scene. There
were also a greater proportion of survivors with noncardiac etiologies of cardiac arrest (48% vs 27%). Survivors to
discharge also displayed higher degrees of consciousness on arrival to the emergency department in comparison to
nonsurvivors. Multivariate analysis among all patients with asystole indicated age (p = 0.01) and witnessed arrest (p =
0.03) as independent predictors of an increased chance of survival. Multivariate analysis among witnessed arrests
indicated short time to arrival of the MCCU (p < 0.001) and no treatment with atropine (p = 0.05) as independent
predictors of survival. Fifty-five percent of patients discharged alive had none or small neurologic deficits (cerebral
performance categories 1 or 2). No patients > 70 years old with unwitnessed arrests (n = 211) survived to discharge.
Summary: 1635 patients were judged as having asystole as the first-recorded arrhythmia. Only 32 patients were
survived. 16% of survivors and 34% of nonsurvivors received atropine. Multivariate analysis among witnessed arrests
indicated no treatment with atropine (p=0.05) as independent predictors of survival. In this treatment algorithm,
adrenaline is given before atropine. And 50% of survivors to discharge did not receive any prehospital intravenous
medication at all. The author concluded that the survivors more often responded with rhythm changes after the initial
adrenalin injection, and therefore, no longer needed atropine. This study did not show how many pediatric patients were
included.
Engdahl, J., A. Bang, et al. (2000). "Can we define patients with no and those with some chance of survival when found
in asystole out of hospital?" Am J Cardiol 86(6): 610-4.
Comments: Level 3; Poor. Opposing
DeBehnke, D. J., G. L. Swart, et al. (1995). "Standard and higher doses of atropine in a canine model of pulseless
electrical activity." Acad Emerg Med 2(12): 1034-41.
Comments: Level 5; Good. Opposing
Herlitz, J., A. Bang, et al. (2003). "Factors associated with survival to hospital discharge among patients hospitalised
alive after out of hospital cardiac arrest: change in outcome over 20 years in the community of Goteborg, Sweden." Heart
89(1): 25-30.
Comments: Level 3; Poor. Opposing
BACKGROUND: A large proportion of patients who suffer out-of-hospital cardiac arrest have asystole as the initial
recorded arrhythmia. Since they have a poor prognosis, less attention has been paid to this group of patients. AIM: To
describe a consecutive population of patients with out-of-hospital cardiac arrest with asystole as the first recorded
arrhythmia and to try to define indicators for an increased chance of survival in this population. SETTING: The
community of Gothenburg. PATIENTS: All patients who suffered out-of-hospital cardiac arrest during 1981 to 1992 and
were reached by our emergency medical service (EMS) system and where cardiopulmonary resuscitation (CPR) was
attempted. RESULTS: In all there were 3434 cardiac arrests of which 1222 (35%) showed asystole as the first recorded
arrhythmia. They differed from patients with ventricular fibrillation by being younger, including more women and having a
longer interval between collapse and arrival of the first ambulance. In all 90 patients (7%) were hospitalized alive and 20
(2%) could be discharged from hospital. Independent predictors for an increased chance of survival were: (a) a short
interval between the collapse and arrival of the first ambulance (P < 0.001) and the time the collapse occurred (P < 0.05).
Initial treatment given in some cases with adrenaline, atropine and tribonate were not associated with an increased
survival. CONCLUSIONS: Of all the patients with out-of-hospital cardiac arrest, 35% were found in asystole. Of these,
7% were hospitalized alive and 2% could be discharged from hospital. Efforts should be made to improve still further the
interval between collapse and arrival of the first ambulance.
Summary: 3434 patients who suffered OHCA was attempted CPR during 1981 to 1992. Initial treatment given in some
cases with adrenaline, atropine and tribonate were not associated with increased chance of survival. Only 24% of
asystolic patients received atropine. In multivariate analysis the administration of atropine were worse of discharged alive.
This study did not show how many pediatric patients were included.
Herlitz, J., L. Ekstrom, et al. (1994). "Predictors of early and late survival after out-of-hospital cardiac arrest in which
asystole was the first recorded arrhythmia on scene." Resuscitation 28(1): 27-36.
Comments: Level 3; Poor. Opposing
PURPOSE: Evaluate the efficacy of advanced life support interventions using the pediatric Utstein guidelines.
METHODS: Charts from all patients for whom a cardiorespiratory arrest code was called during a six-year period in a
university affiliated centre were reviewed. Data were recorded according to the pediatric Utstein guidelines and a P <
0.05 was considered significant. RESULTS: Of the 234 calls, 203 were retained for analysis. The overall survival rate at
one year was 26.0% of which 10% had deterioration of their neurologic status compared to the pre-cardiorespiratory
arrest evaluation. Time to achieve sustained return of spontaneous circulation (ROSC; P < 0.0001) and sustained
measurable blood pressure (P = 0.002), to perform endotracheal intubation (P = 0.04) and the dose of sodium
bicarbonate (P < 0.0001) were indicators of long-term survival. Two patients were alive at one year with unchanged
neurologic status despite a time to achieve sustained ROSC longer than 30 min (38 and 44 min). The mean first
epinephrine dose of patients for whom ROSC was achieved but unsustained was higher than those for whom ROSC was
achieved and sustained (0.038 +/- 0.069 mg*kg(-1) vs 0.011 +/- 0.006 mg*kg(-1); P = 0.004). Survival rate and mean first
epinephrine dose of patients who received their first epinephrine dose endotracheally (13.3%; 0.011 +/- 0.004 mg*kg(-1))
were comparable to those of patients who received their first epinephrine dose intravenously (7%; 0.015 +/- 0.027
mg*kg(-1)). CONCLUSIONS: For intravenously administered epinephrine, a dose of 0.01 mg*kg(-1) seems appropriate
as the first dose. The endotracheal route is a valuable alternative for epinephrine administration and, for infants, the dose
does not need to be increased. A minimal resuscitation duration time of 30 min can be misleading if ROSC is used as the
indicator.
Guay J, Lortie L. An evaluation of pediatric in-hospital advanced life support interventions using the pediatric Utstein
guidelines: a review of 203 cardiorespiratory arrests. Can J Anaesth 2004; 51(4): 373-8.
Page 12 of 21
OBJECTIVE: To determine whether standard or increased doses of atropine improve the return of spontaneous
circulation (ROSC) rate in a canine model of pulseless electrical activity (PEA). METHODS: A prospective, controlled,
blinded laboratory investigation was performed using an asphyxial canine cardiac arrest model. After the production of
asphyxial PEA, 75 dogs remained in untreated PEA for 10 minutes and then were randomized to receive placebo (group
1) or one of four doses of atropine (group 2, 0.04 mg/kg; group 3, 0.1 mg/kg; group 4, 0.2 mg/kg; group 5, 0.4 mg/kg). All
the animals received mechanical external CPR and epinephrine (0.02 mg/kg every 3 minutes) throughout resuscitation.
RESULTS: The ROSC rates were not significantly different between the groups (group 1, 73%; group 2, 67%; group 3,
40%; group 4, 47%; group 5, 27%; p = 0.06). The heart rates and hemodynamics during resuscitation were not
significantly different between the groups. CONCLUSION: In this canine model of asphyxial PEA cardiac arrest,
standard-dose atropine did not improve ROSC rates, compared with placebo. Increasing doses of atropine tended to
decrease ROSC rates, compared with placebo and standard-dose atropine.
Summary: This animal model may be more representative of pediatric asphyxial cardiac arrest than adult PEA. In their
previous study of canine asphyxial PEA model, they had found that vagotomized animals had better ROSC rate. But this
study shows conflicting results. They suggest the pharmacologic actions of atropine on vagal afferent activity.
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
The Bezold-Jarisch reflex is an inhibitory reflex that originates from the heart, is mediated by the vagus nerve, and is
manifested by hypotension and bradycardia. We present 4 pediatric cardiac surgical patients, aged 1 day to 9 months,
who exhibited cardiovascular collapse in their early postoperative course. In each patient, cardiovascular deterioration
was marked by an insidious decrease in arterial blood pressure without an associated change in heart rate, central
venous pressure, or airway pressure. Bradycardia followed the decrease in blood pressure. The Bezold-Jarisch reflex
was suspected and atropine was administered, first as a bolus injection at 0.01 mg/kg, and later, as a continuous infusion
at 0.01 mg.kg-1.h-1. Atropine prevented recurrent episodes of hypotension and bradycardia. We believe the BezoldJarisch reflex is more prevalent than previously suspected in postoperative pediatric cardiac surgical patients.
Summary: Case series in children; developed hypotension followed by bradycardia-all responded to atropine (2 had epi
first); all were being ventilated (so not hypoxic)
Page 11 of 21
or PIVR less than 40. However, only two patients in each group were successfully resuscitated in the emergency
department and only one control group patient was discharged alive. Our findings are not in agreement with those of
previous authors who have advocated the use of atropine in cardiac arrest patients with these arrhythmias. We question
the usefulness of atropine in this setting. More study is necessary in order to clearly define its role in the resuscitation of
patients who have sustained brady-asystolic arrests.
Summary: Prospective, controlled study in adults: epinephrine=atropine
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc
Page 13 of 21
Of 133 persons with spontaneous cardiac arrest attended by paramedics within 10 minutes, 100 (75%) had ventricular
fibrillation as the initial rhythm and 33 (25%) had extreme bradycardia or asystole. The latter group of arrhythmias was
characterized by sinus arrest or severe sinus bradycardia (90%) and complete A-V block (10%). Junctional escape
rhythm was also absent or markedly retarded. Despite cardiopulmonary resuscitation and the administration of
epinephrine, atropine, isoproterenol, and sodium bicarbonate, recovery of the sinus and junctional tissues was infrequent.
Ventricular fibrillation developed in 11 cases (33%). One patient lived 12 days, but all others were dead on arrival or died
Iseri LT, Humphrey SB, Siner EJ. Prehospital brady-asystolic cardiac arrest. Ann Int Med 1978;88:741-745.
LOE 5. Neutral. Adult study. Quality of evidence poor.
NHLBI funded
The aims of this study were to define the antagonistic effects of atropine sulfate to halothane-induced cardiovascular
depression in children, and to clarify whether or not a larger dose of atropine is more effective in attenuating the
cardiovascular depression. Thirty-four children aged 1-12 years who had undergone minor surgery, free from cardiac or
pulmonary disease, were assigned at random to two groups. M-mode echocardiographic evaluation of left ventricular
function in each patient was performed at three points (before induction, point A; after induction, point B; and following
administration of atropine, point C). Results were compared between points A and B, B and C and C and A, and between
the two study groups with different doses of atropine (0.01 mg/kg vs 0.02 mg/kg). Heart rate (HR), mean blood pressure
(MBP) and left ventricular shortening fraction (LVSF) decreased, and left ventricular end-diastolic dimension (LVEDD)
were increased significantly by halothane induction. Although HR and MBP recovered following atropine, LVSF and
LVEDD remained unchanged. There were no differences found between the values after vagolysis in both study groups,
except for HR and mean velocity of circumferential fiber shortening (mVcf). Heart rate increased above that of preinduction, even following the smaller dose of atropine. The myocardial depression cannot be necessarily attenuated by
vagolysis regardless of the dosage of atropine. The smaller dose (i.e. 0.01 mg/kg) seems to be sufficient only to
antagonize the bradycardia and hypotension during halothane anesthesia in children.
Summary: Randomized study of children undergoing halothane anesthesia. Monitored HR, Mean BP left ventricular end
diastolic dimension (LVEDD), LV shortening fraction(LVSF) with 2 doses of atropine (0.01 mg/kg and 0.02 mg/kg).
Atropine only increased HR and MBP, but had no effect on LVSF (which was decreased by halothane), or LVEDD (which
was increased by halothane). Conclusion-atropine does not improve the myocardial depression caused by halothane,
and a dose o 0.02 mg/kg is no better than 0.01 mg/kg.
Horigome H, Tsuji M, Yamashita M, et al. Echocardiographic evaluation of vagolytic effects of atropine sulfate during
pediatric halothane anesthesia. Acta Paediatrica Japonica 1993;35:513-517.
LOE 1, randomized, controlled study in children. Against the use of atropine to improve myocardial depression caused by
halothane anesthesia. Quality of evidence good.
No industry funding
OBJECTIVE: To describe the change in survival and factors associated with survival during a 20 year period among
patients suffering from out of hospital cardiac arrest and being hospitalised alive. PATIENTS: All patients hospitalised
alive in the community of Goteborg after out of hospital cardiac arrest between 1 October 1980 and 1 October 2000 were
included. METHODS: Patient data were prospectively computerised with regard to factors at resuscitation. Data on
medical history and hospitalisation were retrospectively recorded. Patients were divided into two groups (the first and
second 10 year periods). SETTING: Community of Goteborg, Sweden. RESULTS: 5505 patients suffered from cardiac
arrest during the time of the survey. Among them 1310 patients (24%) were hospitalised alive. Survival (discharged alive)
was 37.5% during the first part and 35.1% during the second part (NS). The following were independent predictors of an
increased chance of survival: ventricular fibrillation/tachycardia as the first recorded rhythm (odds ratio (OR) 3.46, 95%
confidence interval (CI) 2.36 to 5.07); witnessed arrest (OR 2.50, 95% CI 1.52 to 4.10); bystander initiated
cardiopulmonary resuscitation (OR 2.00, 95% CI 1.42 to 2.80); the patient being conscious on admission to hospital (OR
6.43, 95% CI 3.61 to 11.45); sinus rhythm on admission to hospital (OR 1.53, 95% CI 1.12 to 2.10); and treatment with
lidocaine in the emergency department (OR 1.64, 95% CI 1.16 to 2.31). The following were independent predictors of a
low chance of survival: age > 70 years (median) (OR 0.65, 95% CI 0.47 to 0.88); atropine required in the emergency
department (OR 0.35, 95% CI 0.16 to 0.75); and chronic treatment with diuretics before hospital admission (OR 0.59,
95% CI 0.43 to 0.81). CONCLUSION: There was no improvement in survival over time among initial survivors of out of
hospital cardiac arrest during a 20 year period. Major indicators for an increased chance of survival were initial
ventricular fibrillation/tachycardia, bystander cardiopulmonary resuscitation, arrest being witnessed, and the patient being
conscious on admission. Major indicators for a lower chance were high age, requirement for atropine in the emergency
department, and chronic treatment with diuretics before cardiac arrest.
Summary: They suggested that the patients who require atropine are those in whom the resuscitation attempt was not
immediately successful. This study did not show how many pediatric patients were included.
C2010 Worksheet:Peds-052A 25-Dec-2009 .doc

Download Report