American Journal of Epidemiology
ª The Author 2008. Published by the Johns Hopkins Bloomberg School of Public Health.
All rights reserved. For permissions, please e-mail: [email protected].
Vol. 168, No. 6
DOI: 10.1093/aje/kwn174
Advance Access publication August 6, 2008
Original Contribution
Inflammatory Markers and Longitudinal Lung Function Decline in the Elderly
Rui Jiang1, Gregory L. Burke2, Paul L. Enright3, Anne B. Newman4, Helene G. Margolis5, Mary
Cushman6,7, Russell P. Tracy7, Yuanjia Wang8, Richard A. Kronmal9, and R. Graham Barr1,10
1
Department of Medicine, College of Physicians and Surgeons, Columbia University, New York, NY.
Department of Public Health Sciences, Wake Forest University School of Medicine, Winston-Salem, NC.
3
Department of Medicine, University of Arizona, Tucson, AZ.
4
Department of Epidemiology, University of Pittsburgh, Pittsburgh, PA.
5
Department of Internal Medicine, School of Medicine, University of California, Davis, CA.
6
Department of Medicine, University of Vermont, Burlington, VT.
7
Department of Pathology, University of Vermont, Burlington, VT.
8
Department of Biostatistics, Mailman School of Public Health, Columbia University, New York, NY.
9
Collaborative Health Studies Coordinating Center and Department of Biostatistics, University of Washington, Seattle, WA.
10
Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, NY.
2
Received for publication January 9, 2008; accepted for publication May 21, 2008.
Longitudinal studies examining associations of the inflammatory markers fibrinogen and C-reactive protein
(CRP) with lung function decline are sparse. The authors examined whether elevated fibrinogen and CRP levels
were associated with greater longitudinal lung function decline in the elderly. The Cardiovascular Health Study
measured fibrinogen and CRP in 5,790 Whites and African Americans from four US communities aged 65 years or
older in 1989–1990 or 1992–1993. Spirometry was performed in 1989–1990 and 4, 7, and 16 years later. Fibrinogen
and CRP were inversely associated with lung function at baseline after adjustment for multiple potential confounders. In mixed models, the rate of decline in forced expiratory volume in 1 second (FEV1)/forced vital capacity (FVC)
ratio with increasing age was faster among those with higher baseline fibrinogen (0.032%/year per standard
deviation higher fibrinogen (95% confidence interval: 0.057, 0.0074)) but not among those with higher CRP
(0.0037%/year per standard deviation higher CRP (95% confidence interval: 0.013, 0.0056)). Longitudinal
analyses for FEV1 and FVC yielded results in the direction opposite of that hypothesized, possibly because of
the high mortality rate and strong inverse association of FEV1 and FVC but not FEV1/FVC with mortality. An
alternative approach to missing data yielded similar results. In conclusion, higher levels of fibrinogen, but not
CRP, independently predicted greater FEV1/FVC decline in the elderly.
aged; biological markers; C-reactive protein; fibrinogen; forced expiratory volume; inflammation; spirometry; vital
capacity
Abbreviations: CI, confidence interval; COPD, chronic obstructive pulmonary disease; CRP, C-reactive protein; FEV1, forced
expiratory volume in 1 second; FVC, forced vital capacity.
Lung function deteriorates gradually throughout adult
life. Accelerated decline in forced expiratory volume in 1
second (FEV1) and the ratio of FEV1/forced vital capacity
(FVC) characterizes chronic obstructive pulmonary disease
(COPD) (1). COPD is currently the fourth leading cause of
death in the United States and much of the developed world
Correspondence to Dr. R. Graham Barr, Columbia University Medical Center, 622 West 168th Street, PH 9 East - Room 105, New York,
NY 10032 (e-mail: [email protected]).
602
Am J Epidemiol 2008;168:602–610
Inflammatory Markers and Lung Function Decline
(2–4). COPD morbidity and mortality continue to rise. By
the year 2020, COPD is projected to be the third leading
cause of death worldwide (5).
Recent research on the pathogenesis of COPD emphasizes
the role of inflammation. It is hypothesized that systemic
inflammation may lead to permanent loss of lung function
over time by inducing endothelial dysfunction and subsequently causing lung alveolar destruction. Several papers reported that inflammation markers fibrinogen and C-reactive
protein (CRP) were inversely associated with lung function in
cross-sectional analyses (6–11), and other studies have reported that COPD patients had significantly higher levels of
fibrinogen and CRP compared with controls (12, 13).
It is uncertain whether systemic inflammation contributes
to accelerated decline in lung function or whether aspects of
COPD cause systemic inflammation. Longitudinal studies
reporting the association of fibrinogen and CRP levels with
lung function decline are sparse. In the Cardiovascular
Health Study, a population-based, observational study of
cardiovascular disease in the elderly, spirometry was conducted four times over 16 years of follow-up. We examined
the association between the inflammatory markers, fibrinogen and CRP, and longitudinal decline in lung function in
this large elderly cohort, hypothesizing that higher levels of
baseline fibrinogen and CRP predicted greater declines in
FEV1 and the FEV1/FVC ratio.
MATERIALS AND METHODS
Study population
The Cardiovascular Health Study is a prospective study of
5,888 men and women aged 65 years or older at baseline
designed to investigate risk factors for cardiovascular disease
in older adults (14, 15). A total of 5,201 original cohort subjects (primarily Whites) were initially recruited in 1989–1990
from four US communities: Forsyth County, North Carolina;
Sacramento County, California; Washington County, Maryland; and Pittsburgh (Allegheny County), Pennsylvania. An
additional 687 African Americans were recruited in 1992–
1993 from three of the four counties (Forsyth, Sacramento,
and Allegheny) to increase minority representation. The elderly participants were randomly selected from the US Health
Care Financing Administration Medicare eligibility lists. Individuals who were institutionalized; were likely to move from
the area within the next 3 years; were wheelchair-bound at
home; or were receiving hospice treatment, radiation therapy,
or chemotherapy for cancer were not eligible to participate.
The institutional review board at each participating center
approved the study, and all participants gave informed consent.
Extensive data collection was performed at baseline.
A subset of the standardized American Thoracic Society
DLD-78 Respiratory Questionnaire was administered by
a centrally trained interviewer (16). Other questionnaires
were also administered, which collected information on
health behaviors and medical conditions. The baseline examination included spirometry, seated blood pressure by
random-zero sphygmomanometer, fasting blood chemistry
and lipid analysis, and a physical examination. Clinic examinations and telephone contacts occurred annually
Am J Epidemiol 2008;168:602–610
603
through 1998–1999, followed by annual phone calls. Spirometry testing was conducted in 1989–1990, 1993–1994,
1996–1997, and 2005–2006 (17), the latter measure via an
ancillary study on lung function in the Cardiovascular
Health Study/All Stars examination, an in-person followup examination in 2005–2006 (18).
In the present study, 39 participants who reported their
race as American Indian/Alaskan Native, Asian/Pacific
Islander, or other were excluded. Of 5,849 Whites and African
Americans, we excluded 98 without fibrinogen and 81 without CRP measures. We further excluded those with unusable
baseline spirometry data (missing or spirometry values read
from unacceptable curves or unreasonably high or low spirometry values: FEV1 >5.5, FEV1 <0.5, FVC >6.0, or FVC
<0.5 l) (19). After the exclusions, 5,011 participants remained for the analysis of fibrinogen and lung function and
5,021 for the analysis of CRP and lung function.
Assessment of inflammatory markers
Blood samples were analyzed at the University of Vermont
(Burlington, Vermont) (20). For the initial cohort, baseline
fibrinogen levels in 1989–1990 were measured with a BBL
fibrometer (Becton Dickinson, Cockeysville, Maryland) by
the Clauss method with Dade fibrinogen calibration reference (Baxter-Dade, Bedford, Massachusetts) and bovine
thrombin (Parke-Davis, Lititz, Pennsylvania) (20). The mean
monthly coefficient of variation was 3.1 percent. Baseline
CRP levels were measured by using an enzyme-linked immunosorbent assay (ELISA) developed at the Cardiovascular
Health Study central blood laboratory (21). This colorimetric
competitive immunoassay uses purified protein and polyclonal anti-CRP antibodies. The interassay coefficient of variation was 5.5 percent (21). Fibrinogen and CRP levels were
also available in 1992–1993 (3 years of follow-up) for the
initial cohort. Fibrinogen levels at 3 years of follow-up were
measured by the same method as that used for fibrinogen
levels at baseline. The latter measurement of CRP was measured by using the BNII nephelometer from Dade Behring
(N High Sensitivity CRP; Dade Behring Inc., Deerfield, Illinois), utilizing particle-enhanced immunonepholometry.
The coefficient of variation for this assay was 5.0 percent.
Because different assays were used for CRP at baseline and
3 years later, an adjustment was made for the baseline CRP
values so that the values obtained from the two assays were
directly comparable (CRP_BNII ¼ exp(ln(CRP_ELISA) þ
0.2781)) (22).
For the African-American cohort, fibrinogen and CRP
levels were measured only once at their baseline (1992–
1993). CRP levels that had been measured by the enzymelinked immunosorbent assay at baseline were remeasured by
using the automated BNII assay. In both cohorts, fibrinogen
was measured in the same year that blood samples were
collected, whereas CRP was measured from stored serum
collected at the year.
Spirometry testing
Lung function was measured by using a water-sealed spirometer in 1989–1990, 1993–1994, and 1996–1997 (17)
604 Jiang et al.
according to American Thoracic Society recommendations
(23, 24). Each spirometer was connected to a personal computer programmed to provide expiratory curves, calculate
lung function test parameters, and determine the acceptability
of the tests. Checks for leaks and volume accuracy of the
spirometer using a 3-l syringe were repeated prior to each
session. Internal spirometer temperature was measured for
each maneuver and was used for corrections to body temperature, ambient pressure, and saturated water vapor. Participants were asked to perform five to eight forced expiratory
maneuvers or until three acceptable and two reproducible
maneuvers were obtained. The highest values of FEV1 and
FVC from acceptable maneuvers were used in this analysis.
The quality of the spirometry testing conducted by the
technicians was monitored centrally throughout testing. All
test sessions were reviewed at the Pulmonary Function
Reading Center by a single quality control supervisor
(P. L. E.). Five quality control grades were computed separately for FEV1 and FVC. A grade of A, B, and C met the
American Thoracic Society 1994 recommendations (grades
from test sessions with two acceptable maneuvers and reproducibility of FEV1 and FVC values <200 ml) (24). Identical spirometers, software, procedures, and reading center
personnel were used at all three examinations, and the majority of the technicians at each of the four clinical sites did
not change over time.
The first spirometry for the African-American cohort was
performed 1 year later (1993–1994), after the recruitment.
Therefore, we treated the 1993–1994 spirometry measures as
the baseline spirometry measures for the African-American
cohort in the analyses.
The Cardiovascular Health Study All-Stars examination
was performed in 2005–2006 in participants’ homes, nursing homes, or any other convenient location because of the
advanced age of the cohort 16 years after its inception. The
use of a water-seal spirometer was therefore impractical.
Instead, spirometry measures were performed by using the
EasyOne Diagnostic portable spirometer (ndd Medical
Technologies, Chelmsford, Massachusetts). The reproducibility and validity of this flow-sensing unit has been previously established (25, 26). The protocol was otherwise the
same as for earlier examinations, and all test sessions were
reviewed at the Pulmonary Function Reading Center by two
quality control supervisors (P. L. E. and R. G. B.). Quality
control grades were computed by the EasyOne software and
were confirmed by the quality control supervisors.
Test sessions with no acceptable curves at baseline and
follow-up examinations were eliminated from the analysis.
Statistical analysis
We analyzed the cross-sectional associations of fibrinogen and CRP levels with lung function at baseline by using
generalized linear regression. Longitudinal analyses were
conducted by using mixed linear regression models. We
modeled the interactions between the inflammatory markers
and time as fixed effects, with random subject-specific intercepts. In multivariate models, we controlled for a variety
of potential confounders. Smoking status and quality control
factors were treated as time varying in longitudinal analyses.
A relatively large proportion of spirometry data was missing in the later years of follow-up because of the high rates
of mortality and morbidity in this elderly cohort. Since
FEV1, FVC, and inflammatory markers predict mortality,
these missing data had the potential to bias analyses of
longitudinal change in lung function. We used two different
approaches to deal with missing spirometry values.
First, we used mixed-effects linear regression models for the
main analyses, which are relatively well suited to such problems since they treat missing data as missing at random conditional on the covariates in the model and observed
spirometry values. Second, we imputed missing longitudinal
spirometry values by using the Markov chain Monte Carlo
multiple imputation method based on baseline characteristics
including age, gender, height, height squared, FVC, physician
diagnosis of emphysema, and updated smoking status (27). We
imputed data sets five times. Each imputed full data set was
then analyzed by using a mixed-effects model as described
above, providing five sets of parameter estimates. Parameter
estimates from each replication of analysis were averaged to
provide a single estimate. The standard error of the estimate
was computed based on the weighted average of the withinand between-imputation variability of the parameter estimates.
All p values were two tailed, and p values of <0.05 were
considered statistically significant. We used SAS software
(version 9; SAS Institute, Inc., Cary, North Carolina) for the
analyses.
RESULTS
Study population
A total of 5,011 participants had adequate information on
fibrinogen and lung function, and 5,021 participants had
adequate information on CRP and lung function at baseline.
Among the participants, mean age was 72.7 years at baseline; 42.9 percent were male and 87.2 percent were White.
Mean baseline FEV1, FVC, and FEV1/FVC were 2,059 ml,
2,929 ml, and 70.1 percent, respectively. Those participants
excluded because of missing measures of fibrinogen and
CRP or usable lung function data were more likely to be
African American (34.1 percent vs. 12.8 percent in the study
population). The distributions of age, gender, height, and
smoking status did not differ substantially between the
two groups. Those with unusable lung function data had
slightly higher median levels of fibrinogen (321 mg/dl vs.
311 mg/dl in the study population) and CRP (2.68 mg/dl vs.
2.51 mg/liter in the study population).
Table 1 shows baseline characteristics of study participants by the first, third, and fifth quintiles of baseline levels
of fibrinogen and CRP. Compared with participants in the
lowest quintile of fibrinogen levels, those in the highest
quintile were more likely to be females, African Americans,
and current smokers; were heavier; and had a larger waist
circumference. Those in the highest fibrinogen quintile were
also more likely to have low educational attainment and
higher prevalences of self-reported physician-diagnosed
asthma, pneumonia, emphysema, and bronchitis. Distributions of participant characteristics by quintiles of CRP levels
were similar to those by quintiles of fibrinogen.
Am J Epidemiol 2008;168:602–610
Inflammatory Markers and Lung Function Decline
605
TABLE 1. Characteristics by first, third, and fifth quintiles of fibrinogen and CRP* levels, Cardiovascular Health Study baseline
examination, 1989–1990 and 1992–1993y
Fibrinogen quintile
1
CRP quintile
3
5
1
3
5
Age (years)
72.2 (5.4)
72.9 (5.6)
73.2 (5.7)
73.2 (5.7)
72.5 (5.2)
72.5 (5.5)
Male gender
47.7
39.1
40.7
45.5
45.5
39.3
Race
White
88.9
86.4
73.4
90.2
89.1
81.3
African American
11.1
13.6
26.6
9.8
10.9
18.7
Never
45.1
49.6
41.7
50.5
46.2
40.9
Past
46.5
40.5
39.6
41.5
43.4
43.0
Current
8.4
9.9
18.7
8.0
10.5
16.1
30.3 (29.5)
35.8 (29.1)
37.8 (28.8)
29.0 (28.3)
34.9 (28.0)
41.4 (31.3)
Less than grade 8
6.2
7.4
9.6
5.9
6.5
9.0
Grade 8–grade 11
18.0
22.4
25.4
17.4
18.8
26.1
Smoking status
Pack-years among ever smokers
Education
Grade 12–GED*
27.3
27.6
26.7
27.9
28.4
26.6
1 year vocational school4 years of college
35.3
33.3
31.4
33.7
35.7
31.5
Graduate or professional
13.2
9.3
6.9
15.1
10.7
6.8
166 (9.2)
165 (9.5)
164 (9.6)
165 (9.7)
165 (9.4)
164 (9.5)
Height (cm)
Weight (poundsz)
158 (30.5)
160 (32.7)
163 (32.2)
149 (29.6)
163 (31.3)
169 (34.5)
Waist circumference (cm)
92.4 (12.0)
94.7 (13.0)
96.8 (13.4)
88.6 (12.0)
95.6 (12.2)
99.1 (13.9)
Use of beta blocker
12.4
12.5
13.8
11.1
12.8
15.1
Current asthma
2.5
3.2
4.5
3.1
3.2
3.3
Pneumonia
28.1
28.0
31.2
25.6
27.9
30.7
Emphysema
4.3
3.8
5.9
3.7
4.4
7.8
Bronchitis
20.2
23.5
25.6
18.9
21.9
28.6
* CRP, C-reactive protein; GED, general equivalency diploma.
y Values are given as mean (standard deviation) or percentage.
z One pound ¼ 0.454 kg.
Cross-sectional analysis of inflammatory markers and
lung function
Inflammatory markers and longitudinal change in lung
function
Baseline fibrinogen and CRP levels were moderately correlated (Spearman correlation coefficient ¼ 0.46; p <
0.0001). Higher levels of fibrinogen were associated with
lower FEV1 and FEV1/FVC ratio at baseline (table 2). For
each standard deviation higher fibrinogen (66.5 mg/dl),
there was a 43.9-ml (95 percent confidence interval (CI):
57.9, 29.9 ml) lower FEV1 and a 0.33 percent (95 percent CI: 0.60, 0.067 percent) lower FEV1/FVC ratio.
Associations similar to those for FEV1 were observed for
FVC (table 2).
Higher levels of CRP were associated with lower FEV1
and FVC at baseline (table 2). CRP levels were not associated with FEV1/FVC ratio (table 2). Every standard deviation higher CRP (2.80 mg/liter) was associated with a
14.3-ml (95 percent CI: 19.1, 9.59 ml) decline in FEV1.
No linear trend was observed between baseline CRP levels
and FEV1/FVC ratio in the multivariate model.
Retention in the Cardiovascular Health Study, defined as
number of participants completing each examination/(baseline number – deaths), was 94.5 percent at the end of the
main study period and 85 percent for the Cardiovascular
Health Study All-Stars examination. However, the original
cohort mortality rates were 9.7 percent after 4 years, 21.1
percent after 7 years, and 62.6 percent after 16 years (16
years of follow-up from 1989–1990 to 2005–2006); the
African-American cohort mortality rates were 11.6 percent
after 4 years and 50.5 percent after 13 years (13 years of
follow-up from 1992–1993 to 2005–2006). Consequently,
74.8 percent, 53.5 percent, and 17.2 percent of the initial
cohort had usable spirometry data after 4, 7, and 16 years,
respectively; 72.1 percent and 24.6 percent of the AfricanAmerican cohort had usable spirometry data after 4 and 13
years, respectively. A total of 1.90, 2.69, 0.64, and 7.39
percent of the initial cohort at baseline and 4, 7, and 16
Am J Epidemiol 2008;168:602–610
606 Jiang et al.
TABLE 2. Cross-sectional associations of fibrinogen and CRP* levels (in quintilesy) with lung function, Cardiovascular Health Study
baseline examination, 1989–1990/1992–1993
Quintile 2
Mean
differencez
95%
confidence
interval
Quintile 3
Mean
differencez
Quintile 4
95%
confidence
interval
Mean
differencez
Quintile 5
95%
confidence
interval
Mean
differencez
95%
confidence
interval
p for
trend
Fibrinogen
FEV1* (ml)
Model 1§
65.2
110, 20.8
105
150, 60.6
121
166, 76.6
223
270, 176
<0.0001
Model 2{
53.6
96.5, 10.8
83.8
127, 40.8
91.1
134, 47.8
159
205, 114
<0.0001
Model 3#
47.2
89.4, 5.11
70.8
113, 28.4
82.7
126, 39.9
130
175, 85.2 <0.0001
Model 1§
47.4
96.7, 1.82
133
183, 83.5
132
182, 82.1
217
269, 165
<0.0001
Model 2{
38.0
87.1, 11.2
116
165, 66.7
109
159, 59.6
168
220, 116
<0.0001
Model 3#
33.5
81.8, 14.8
94.4
143, 45.9
86.6
136, 37.6
122
173, 70.5 <0.0001
Model 1§
0.81
1.69, 0.06
0.58
1.46, 0.31
0.88
1.77, 0.01
2.35
3.27, 1.43 <0.0001
Model 2{
0.61
1.46, 0.24
0.20
1.05, 0.65
0.34
1.20, 0.52
1.21
2.11, 0.31
0.03
Model 3#
0.54
1.37, 0.29
0.33
1.17, 0.51
0.68
1.52, 0.16
1.30
2.19, 0.41
0.006
FVC* (ml)
FEV1/FVC (%)
CRP
FEV1 (ml)
Model 1§
65.3
109, 21.2
116
160, 71.6
210
254, 166
270
314, 225
<0.0001
Model 2{
46.2
88.8, 3.58
91.8
134, 49.2
162
205, 119
198
241, 154
<0.0001
Model 3#
37.4
79.5, 4.77
72.8
115, 30.1
135
179, 92.0
162
207, 117
<0.0001
Model 1§
106
155, 57.3
172
221, 124
279
328, 230
343
392, 294
<0.0001
Model 2{
92.3
141, 43.7
154
203, 106
245
293, 196
291
341, 242
<0.0001
Model 3#
73.9
122, 25.7
113
162, 64.6
193
242, 143
218
269, 166
<0.0001
Model 1§
0.67
0.20, 1.55
0.11
0.77, 0.99
0.49
1.37, 0.39
1.01
1.89, 0.12
0.0008
Model 2{
1.10
0.21, 1.90
0.57
0.28, 1.42
0.38
0.47, 1.23
0.36
0.51, 1.22
0.76
Model 3#
0.89
0.06, 1.72
0.20
0.64, 1.05
0.02
0.84, 0.87
0.26
1.15, 0.62
0.10
FVC (ml)
FEV1/FVC (%)
* CRP, C-reactive protein; FEV1, forced expiratory volume in 1 second; FVC, forced vital capacity.
y Quintile 1 was considered the reference.
z Mean differences in lung function between the second, third, fourth, or fifth quintile and the lowest quintile (reference).
§ Model 1 was adjusted for gender, height, height squared, and race/ethnicity.
{ Model 2: model 1 þ smoking status (never, past, and current) and pack-years.
# Model 3: model 2 þ educational attainment, weight, waist circumference, use of beta blocker, quality control factors, diagnosed asthma
(current vs. other), and diagnosed pneumonia.
years later, and 6.31, 3.50, and 13.2 percent of the AfricanAmerican cohort at their baseline and 4 and 13 years later,
respectively, were excluded because of nonacceptable curves.
In longitudinal analyses using mixed models, higher baseline fibrinogen levels were associated with a faster rate of
decline in FEV1/FVC with increasing age (0.032 percent/
year per standard deviation higher fibrinogen (95 percent CI:
0.057, 0.0074; p ¼ 0.01)) during 16 years of follow-up.
A trend in the same direction was observed for higher baseline
CRP levels, but it was not statistically significant (0.0037
percent/year per standard deviation higher CRP (95 percent
CI: 0.013, 0.0056; p ¼ 0.44)) (table 3).
Given the large quantity of missing data in the later years
of follow-up, we performed additional analyses by using an
alternative approach for missing data (table 3). Qualitatively
similar results were obtained with Markov chain Monte Carlo
multiple imputation, although the rate of decline in FEV1/
FVC was marginally nonsignificant for fibrinogen levels.
The rate of decline in the FEV1/FVC ratio did not change
significantly after we excluded current smokers (0.033
percent/year per standard deviation higher fibrinogen (95
percent CI: 0.059, 0.007; p ¼ 0.01) and 0.0068 percent/year per standard deviation higher CRP (95 percent CI:
0.016, 0.0028; p ¼ 0.17)). We also examined the rate of
decline in FEV1/FVC stratified by gender. The differences
in the rate of FEV1/FVC decline between men and women
were not statistically significant (p for interaction was 0.26
for fibrinogen and 0.19 for CRP).
Am J Epidemiol 2008;168:602–610
Inflammatory Markers and Lung Function Decline
607
TABLE 3. Longitudinal associations of baseline levels of fibrinogen and CRP* with repeated measures of
lung function, Cardiovascular Health Study, 16 years of follow-up
Mixed model
Change in lung
functiony per unit
increase in
fibrinogen/CRP, age,
or fibrinogen 3 age/
CRP 3 age
Imputation
95%
confidence
interval
Change in lung
functiony per unit
increase in
fibrinogen/CRP, age,
or fibrinogen 3 age/
CRP 3 age
95%
confidence
interval
Fibrinogen
FEV1* (ml)
Fibrinogen (SD*)
194
264, 124
190
256, 123
Age (years)
44.0
48.5, 39.5
43.8
47.9, 39.6
Fibrinogen (SD) 3 age (years)
2.06
1.15, 2.98
2.00
1.13, 2.93
FVC* (ml)
Fibrinogen (SD)
389
493, 286
377
472, 281
Age (years)
77.8
84.4, 71.1
76.1
82.2, 70.0
Fibrinogen (SD) 3 age (years)
4.74
3.40, 6.09
4.59
3.33, 5.99
Fibrinogen (SD)
1.90
0.0088, 3.79
0.86
1.00, 2.79
Age (years)
0.40
0.27, 0.52
0.29
Fibrinogen (SD) 3 age (years)
0.032
FEV1/FVC (%)
0.057, 0.0074
0.018
0.17, 0.41
0.043, 0.0067
CRP
FEV1 (ml)
CRP (SD)z
79.1
107, 51.5
70.2
99.6, 40.8
Age (year)
35.6
36.8, 34.4
35.4
36.5, 34.3
CRP (SD)z 3 age (years)
0.90
0.54, 1.26
0.78
0.39, 1.17
FVC (ml)
CRP (SD)z
125
165, 85.9
112
151, 73.0
Age (years)
57.6
59.4, 55.8
56.5
58.1, 54.8
CRP (SD)z 3 age (years)
1.52
1.00, 2.03
1.34
0.84, 1.85
FEV1/FVC (%)
CRP (SD)z
0.17
0.54, 0.88
Age (years)
0.24
0.21, 0.28
CRP (SD)z 3 age (years)
0.0037
0.013, 0.0056
0.10
0.20
0.000053
0.87, 0.67
0.17, 0.23
0.010, 0.01
* CRP, C-reactive protein; FEV1, forced expiratory volume in 1 second; SD, standard deviation; FVC, forced vital
capacity.
y Results were adjusted for gender, height, height squared, race/ethnicity, smoking status (never, past, and
current), pack-years, educational attainment, weight, waist circumference, use of beta blocker, quality control
factors, diagnosed asthma (current vs. other), and diagnosed pneumonia.
z Because the distribution of CRP was highly skewed, the back-transformed SD of lnCRP was used.
Compared with those for the 16 years of follow-up, analyses with a shorter-term follow-up (7-year follow-up) generated similar, somewhat stronger results for baseline
fibrinogen (0.046 percent change in FEV1/FVC/year per
standard deviation higher fibrinogen (95 percent CI:
0.080, 0.013; p ¼ 0.006)) and CRP (0.005 percent
change in FEV1/FVC/year per standard deviation higher
CRP (95 percent CI: 0.016, 0.006; p ¼ 0.38)) levels.
In contrast, the rate of decline in FEV1 with increasing
age was not faster among those with higher baseline fibrinogen or CRP levels. The interaction terms for fibrinogen 3
age and CRP 3 age were positive, implying a small, but
Am J Epidemiol 2008;168:602–610
statistically significant slower decline in FEV1 among those
with higher fibrinogen and higher CRP levels at baseline
(table 3). Findings for FVC were similar to those for FEV1.
Updated analyses using time-varying fibrinogen and CRP
levels yielded similar results. We further examined the associations of baseline fibrinogen and CRP levels with rate of
decline in lung function stratified by baseline lung function,
and we found that the association was stronger (greater decline) for FEV1/FVC ratio but weaker (slower decline) for
FEV1 and FVC among those with lower baseline lung function
(FVC <80 percent predicted or FEV1/FVC 0.7) (table 4).
Among those with likely COPD at baseline (FEV1/FVC <0.65
608 Jiang et al.
TABLE 4. Longitudinal associations of baseline levels of fibrinogen and CRP* with repeated measures of
lung function, stratified by baseline lung function, Cardiovascular Health Study, 16 years of follow-up
95%
confidence
interval
Change in
lung
functiony
per increase
in CRP
(SD)z 3 age
(years)
95%
confidence
interval
1.73
0.62, 2.79
0.56
0.13, 1.01
2.39
0.93, 3.92
1.20
0.64, 1.73
FEV1/FVC >0.7 and FVC >80% predicted
3.59
2.00, 5.26
0.90
0.23, 1.57
[FVC 80% predicted] or FEV1/FVC 0.7
6.12
3.92, 7.98
2.18
1.37, 2.99
Change in
lung
functiony
per increase
in fibrinogen
(SD*)z 3 age
(years)
FEV1/FVC* >0.7 and FVC >80% predicted
[FVC 80% predicted] or FEV1/FVC 0.7
FEV1* (ml)
FVC (ml)
FEV1/FVC (%)
FEV1/FVC >0.7 and FVC >80% predicted
0.018
0.044, 0.0086
0.0015
0.012, 0.009
[FVC 80% predicted] or FEV1/FVC 0.7
0.056
0.093, 0.015
0.0062
0.021, 0.0084
* CRP, C-reactive protein; SD, standard deviation; FEV1, forced expiratory volume in 1 second; FVC, forced vital
capacity.
y Results from mixed-effects models were adjusted for gender, height, height squared, race/ethnicity, smoking
status (never, past, and current), pack-years, educational attainment, weight, waist circumference, use of beta
blocker, quality control factors, diagnosed asthma (current vs. other), and diagnosed pneumonia.
z Because the distribution of CRP was highly skewed, the back-transformed SD of lnCRP was used.
and FEV1 <80 percent of predicted values), the interaction
terms for fibrinogen 3 age and CRP 3 age were negative but
did not achieve statistical significance.
DISCUSSION
In this large elderly cohort, baseline fibrinogen levels
were cross-sectionally, inversely associated with FEV1 and
FEV1/FVC ratio, and baseline CRP levels were crosssectionally, inversely associated with FEV1 and FVC. Elevated levels of fibrinogen predicted greater longitudinal
decline in FEV1/FVC with increasing age, and a similar,
but nonsignificant association was found for CRP. However,
contradictory to our hypothesis, higher fibrinogen and CRP
levels were associated with small, but statistically significant
slower declines in FEV1 and FVC with increasing age.
Our cross-sectional analyses of fibrinogen and CRP associations with lung function confirmed the findings of previous epidemiologic studies. In the CARDIA Study, FEV1
and FVC were lower by 166 ml in the highest versus lowest
quartile of plasma fibrinogen, although FEV1/FVC ratio was
unrelated to fibrinogen (6). Similar inverse associations for
fibrinogen and FEV1 and/or FVC were found in three other
studies (none of them examined the associations with FEV1/
FVC ratio) (7–9). Several studies have explored the crosssectional associations between CRP and lung function.
A recent study in the United Kingdom analyzed CRP and
lung function in 1991 and 2000, respectively, and found that
each milligram-per-liter increase in serum CRP was associated with a 9-ml reduction in FEV1 and an 11-ml reduction
in FVC in 1990, and with a 7-ml reduction in FEV1 and an
8-ml reduction in FVC in 2000 (10). In the European Community Respiratory Heath Survey, participants in the top
tertile of CRP levels had lower FEV1 and FVC but not
a lower FEV1/FVC ratio compared with those in the lowest
tertile (3.29 liters/second vs. 3.50 liters/second for FEV1 and
3.97 liters/second vs. 4.16 liters/second for FVC) (11). In the
Third National Health and Nutrition Examination Survey,
higher levels of CRP were associated with lower FEV1 (8).
Increased levels of fibrinogen and CRP have also been
reported in individuals with COPD (12, 13).
Although fibrinogen and CRP have been cross-sectionally
associated with lung function and COPD, longitudinal studies on inflammatory markers and change in lung function
over time are sparse. We are aware of two published studies
that have examined the relation between fibrinogen levels
and longitudinal change in lung function. Thyagarajan
et al. (6) found that higher fibrinogen levels were associated
with greater declines in FEV1 and FVC but not with FEV1/
FVC ratio in the much younger CARDIA sample. Donaldson
et al. (28) found that higher fibrinogen level was associated
with a faster FEV1 decline in elderly COPD patients. To our
knowledge, our study is the first to show an accelerated decline in lung function (FEV1/FVC) among those with higher
levels of fibrinogen in a population-based elderly cohort. The
longitudinal relation between CRP levels and change in lung
function has been examined in three published studies, to our
knowledge. Fogarty et al. (10) found no association between
CRP and decline in FEV1 and FVC among those aged 18–70
years. Shaaban et al. (29) found that, among those aged 20–
44 years, FEV1 decline tended to increase from the lower to
the upper tertile for baseline CRP levels, but the associations
were not statistically significant (p ¼ 0.09). Man et al. (30)
found that CRP levels were associated with accelerated decline in FEV1 in cigarette smokers aged 35–60 years with
mild to moderate COPD. Of the four longitudinal studies
Am J Epidemiol 2008;168:602–610
Inflammatory Markers and Lung Function Decline
(three previous studies plus ours) on CRP and lung function,
three population-based studies did not find a significant relation between CRP and accelerated lung function decline
(although a relation for CRP was suggested in two of
the studies). The inconsistent results between the three
population-based studies and Man et al.’s study of COPD
patients could be due to the different study populations.
Although contradictory to the hypothesis, it was not surprising to observe that high levels of fibrinogen and CRP
were associated with slower decline in FEV1 and FVC in the
Cardiovascular Health Study. First, a previous longitudinal
study in the Cardiovascular Health Study found that a diagnosis of emphysema, which was associated with higher fibrinogen and CRP levels, predicted slower decline in FEV1
and FVC compared with no emphysema (17). Because emphysema is a subphenotype of COPD, which is characterized by accelerated decline in FEV1, this finding suggests
strong informative censoring by emphysema/COPD deaths
in the Cardiovascular Health Study. Second, in the Cardiovascular Health Study, FEV1 and FVC strongly (relative risk ¼
6.0) and independently predict death but ratio does not (relative risk ¼ 1.3) (31). Hence, the deaths are likely to strongly
bias results for FEV1 and FVC, but the ratio will be relatively
less biased. Third, in our study, the rate of decline in FEV1
and FVC with increasing fibrinogen or CRP levels was
weaker among those with lower baseline lung function (vs.
those with normal baseline lung function), suggesting survival bias by FEV1 and FVC.
In addition to the survival bias, another potential limitation
of this study is that the results may not be generalizable to
other age groups. Individuals aged 65 years or older may have
developed strong resistance to the deterioration in lung function. The strengths of this study include the large cohort with
long duration of follow-up, repeated spirometry measures (up
to four times), and multiple potential confounders collected.
In conclusion, higher levels of baseline CRP and fibrinogen levels were cross-sectionally associated with lower lung
function. Higher levels of fibrinogen, but not CRP, were
associated with greater longitudinal declines in FEV1/FVC
ratio in the elderly.
ACKNOWLEDGMENTS
This study was supported by contracts N01-HC-35129,
N01-HC-45133, N01-HC-75150, N01-HC-85079 through
N01-HC-85086, N01-HC-15103, N01-HC-55222, R01-HL77612, R01-HL-75476, R01-AG-23629, and U01-HL80295 from the National Heart, Lung, and Blood Institute,
Bethesda, Maryland.
Conflict of interest: none declared.
REFERENCES
1. Pauwels RA, Buist AS, Calverley PM, et al. Global strategy for
the diagnosis, management, and prevention of chronic obstructive pulmonary disease. NHLBI/WHO Global Initiative
for Chronic Obstructive Lung Disease (GOLD) Workshop
summary. Am J Respir Crit Care Med 2001;163:1256–76.
Am J Epidemiol 2008;168:602–610
609
2. Hurd S. The impact of COPD on lung health worldwide: epidemiology and incidence. Chest 2000;117(suppl):1S4S.
3. Hoyert DL, Kung HC, Smith BL. Deaths: preliminary data for
2003. Natl Vital Stat Rep 2005;53:1–48.
4. Devereux G. ABC of chronic obstructive pulmonary disease.
Definition, epidemiology, and risk factors. BMJ 2006;332:
1142–4.
5. Murray CJ, Lopez AD. Alternative projections of mortality
and disability by cause 1990–2020: Global Burden of Disease
Study. Lancet 1997;349:1498–504.
6. Thyagarajan B, Jacobs DR, Apostol GG, et al. Plasma fibrinogen and lung function: the CARDIA Study. Int J Epidemiol
2006;35:1001–8.
7. Engstrom G, Lind P, Hedblad B, et al. Lung function and
cardiovascular risk: relationship with inflammation-sensitive
plasma proteins. Circulation 2002;106:2555–60.
8. Gan WQ, Man SF, Sin DD. The interactions between cigarette
smoking and reduced lung function on systemic inflammation.
Chest 2005;127:558–64.
9. Dahl M, Tybjaerg-Hansen A, Vestbo J, et al. Elevated plasma
fibrinogen associated with reduced pulmonary function and
increased risk of chronic obstructive pulmonary disease. Am J
Respir Crit Care Med 2001;164:1008–11.
10. Fogarty AW, Jones S, Britton JR, et al. Systemic inflammation
and decline in lung function in a general population: a prospective study. Thorax 2007;62:515–20.
11. Kony S, Zureik M, Driss F, et al. Association of bronchial
hyperresponsiveness and lung function with C-reactive protein
(CRP): a population based study. Thorax 2004;59:892–6.
12. Gan WQ, Man SF, Senthilselvan A, et al. Association between
chronic obstructive pulmonary disease and systemic inflammation: a systematic review and a meta-analysis. Thorax
2004;59:574–80.
13. Pinto-Plata VM, Mullerova H, Toso JF, et al. C-reactive protein in patients with COPD, control smokers and non-smokers.
Thorax 2006;61:23–8.
14. Fried LP, Borhani NO, Enright P, et al. The Cardiovascular Health
Study: design and rationale. Ann Epidemiol 1991;1:263–76.
15. Tell GS, Fried LP, Hermanson B, et al. Recruitment of adults
65 years and older as participants in the Cardiovascular Health
Study. Ann Epidemiol 1993;3:358–66.
16. Ferris BG. Epidemiology standardization project II: recommended respiratory disease questionnaires for use with adults
and children in epidemiological research. Am Rev Respir Dis
1978;118:7–52.
17. Griffith KA, Sherrill DL, Siegel EM, et al. Predictors of loss of
lung function in the elderly: the Cardiovascular Health Study.
Am J Respir Crit Care Med 2001;163:61–8.
18. The Cardiovascular Health Study. CHS All Stars. Study
overview. (http://www.chs-nhlbi.org/AllStars.htm).
19. Hankinson JL, Odencrantz JR, Fedan KB. Spirometric reference values from a sample of the general U.S. population. Am
J Respir Crit Care Med 1999;159:179–87.
20. Cushman M, Cornell ES, Howard PR, et al. Laboratory
methods and quality assurance in the Cardiovascular Health
Study. Clin Chem 1995;41:264–70.
21. Macy EM, Hayes TE, Tracy RP. Variability in the measurement of C-reactive protein in healthy subjects: implications for
reference intervals and epidemiological applications. Clin
Chem 1997;43:52–8.
22. Analysis tips. (http://www.nhlbi.nih.gov/resources/deca/chs/
labels/analysistips.pdf).
23. Standardization of spirometry—1987 update. Statement of the
American Thoracic Society. Am Rev Respir Dis 1987;136:
1285–98.
610 Jiang et al.
24. Standardization of spirometry, 1994 update. American
Thoracic Society. Am J Respir Crit Care Med 1995;152:
1107–36.
25. Barr RG, Taveras B, Basner RC. Reproducibility and validity
of a portable spirometer. (Abstract). Am J Respir Crit Care
Med 2003;167:A84.
26. Perez-Padilla R, Vazquez-Garcia JC, Marquez MN, et al. The
long-term stability of portable spirometers used in a multinational study of the prevalence of chronic obstructive pulmonary disease. Respir Care 2006;51:1167–71.
27. Schafer JL. Analysis of incomplete multivariate data. New
York, NY: Chapman and Hall, 1997.
28. Donaldson GC, Seemungal TA, Patel IS, et al. Airway and
systemic inflammation and decline in lung function in patients
with COPD. Chest 2005;128:1995–2004.
29. Shaaban R, Kony S, Driss F, et al. Change in C-reactive protein
levels and FEV1 decline: a longitudinal population-based
study. Respir Med 2006;100:2112–20.
30. Man SF, Connett JE, Anthonisen NR, et al. C-reactive protein
and mortality in mild to moderate chronic obstructive pulmonary disease. Thorax 2006;61:849–53.
31. Barr RG, Lovasi GS, Enright PL, et al. Spirometry for the risk
stratification of cardiovascular disease among the elderly.
(Abstract). Presented at the International Conference of the
European Respiratory Society 2007:311s2s.
APPENDIX TABLE 1. Selected ATS*-DLD-78 items from the
Cardiovascular Health Study Spirometry Questionnaire
ATS-DLD # 17
Have you ever had pneumonia? Yes/No
If you answered Yes, please answer the
following:
Was it confirmed by a doctor? Yes/No
ATS-DLD # 18
Have you ever had chronic bronchitis? Yes/No
If you answered Yes, please answer the
following:
Was it confirmed by a doctor? Yes/No
ATS-DLD # 19
Have you ever had emphysema? Yes/No
If you answered Yes, please answer the
following:
Was it confirmed by a doctor? Yes/No
ATS-DLD # 20
Have you ever had asthma? Yes/No
If you answered Yes, please answer the
following:
Do you still have it? Yes/No
Was it confirmed by a doctor? Yes/No
* ATS, American Thoracic Society.
Am J Epidemiol 2008;168:602–610