Effects of Poststroke Pyrexia on Stroke Outcome
A Meta-Analysis of Studies in Patients
Cother Hajat, MRCP; Shakoor Hajat, MSc; Pankaj Sharma, PhD
Downloaded from http://stroke.ahajournals.org/ by guest on July 12, 2017
Background and Purpose—The effect of pyrexia on cerebral ischemia has been extensively studied in animals. In humans,
however, such studies are small and the results conflicting. We undertook a meta-analysis using all such published
studies on the effect of hyperthermia on stroke outcome.
Methods—Three databases were searched for all published studies that examined the relationship of raised temperature
after stroke onset and eventual outcome. Combined probability values and odds ratios were obtained. A heterogeneity
test was performed to ensure that the data were suitable for such an analysis. Morbidity and mortality were used as
outcome measures.
Results—Nine studies were identified totaling 3790 patients, providing our study with 99% power to detect a 9% increase
in morbidity and 84% power to detect a 1% increase in mortality for the pyrexial group. The combined odds ratio for
mortality was 1.19 (95% CI, 0.99 to 1.43). A heterogeneity test was highly nonsignificant (P⬎0.05) for mortality,
suggesting that the data were sufficiently similar to be meta-analyzed. Combined probability values were highly
significant for both morbidity (P⬍0.0001) and mortality (P⬍0.00000001).
Conclusions—The results from this meta-analysis suggest that pyrexia after stroke onset is associated with a marked
increase in morbidity and mortality. Measures should be taken to combat fever in the clinical setting to prevent stroke
progression. The possible benefit of therapeutic hypothermia in the management of acute stroke should be further
investigated. (Stroke. 2000;31:410-414.)
Key Words: fever 䡲 meta-analysis 䡲 outcome 䡲 stroke
S
We sought to undertake a meta-analysis on all published
studies to investigate the effect of body temperature on stroke
outcome in humans. Combining the results of such studies
has the advantage of increasing the power of small, and often
underpowered, studies by pooling data.
troke remains one of the leading causes of morbidity and
mortality in the western world. Several factors have been
implicated in influencing the extent of cerebral damage in
acute stroke, such as high or low systolic blood pressure,1
elevated blood glucose, and a high temperature.2 Rodent
models have consistently shown hyperthermia to be a reliable
predictor of stroke outcome. One study found that elevation
of rat body temperature to 40°C 24 hours after induced
cerebral infarction resulted in a 3-fold increase in cerebral
infarct volume.3 Significant results were also obtained after
experimental traumatic injury, with a 2.6-fold increase in rat
mortality and a 13-fold increase of cerebral contusion volume.3 Other studies have found significant effects on stroke
outcome in rats with lesser degrees of hyperthermia.2,4,5
Furthermore, it has been shown that induced hypothermia has
a protective effect up to 1 hour after focal permanent
ischemia.6 In humans, however, the relationship between
fever and stroke outcome has been far less extensively
investigated, with studies incorporating small numbers of
patients and, as recently emphasized,7 providing conflicting
results. There are no large-scale prospective studies assessing
outcome between normothermic and hyperthermic patients.
Methods
Three computer databases (MEDLINE, BIDS, and the Cochrane
Library) were searched for all published studies that investigated the
effect of body temperature after the onset of stroke on stroke
outcome. In addition, all articles identified with our electronic search
had their bibliographies searched for other potential sources of data,
revealing 1 additional study suitable for inclusion.8 Unpublished data
for 2 of these studies7,8 were acquired through personal communication from the authors. Morbidity and mortality were used as
measures of outcome. Demographic details such as age and sex were
documented together with statistical information such as sample size,
exclusion criteria, outcome measures, probability value, and odds
ratios (ORs). Most of the studies estimated the OR by multiple linear
regression with the exception of 3 studies that used univariate
analysis9,10 and Cox regression analysis.7
A combined OR was obtained for mortality by taking weighted
means by the method of Woolf11 and was tested for heterogeneity
with a ␹2 index. Individual probability values were amalgamated for
Received March 9, 1999; final revision received November 2, 1999; accepted November 2, 1999.
From Guy’s, King’s, and St Thomas’s School of Medicine (C.H.); Department of Primary Care and Population Sciences, Royal Free Hospital School
of Medicine and University College London Medical School (S.H.); and the National Hospital for Neurology and Neurosurgery (P.S.), London, UK.
Correspondence Dr Cother Hajat, Guy’s, King’s, and St Thomas’s School of Medicine, 5th Floor, Capital House, 42 Weston St, London SE1 3QD,
UK. E-mail [email protected]
© 2000 American Heart Association, Inc.
Stroke is available at http://www.strokeaha.org
410
Hajat et al
Poststroke Pyrexia: A Meta-Analysis of Studies in Patients
411
Demographic and Statistical Patient Details
Results
Outcome Measures
Groups
n
Mean Age,
y/% Male
Reith et al
(1996)14
Admission ⬍6 h
(1) temp ⬍37.4°C
(2) temp ⬎37.4°C
293
97
74.2 (⫾11.3)/
50%
Jorgensen
et al
(1996)13
Admission 6–12 h
(1) temp ⱕ37.4°C
(2) temp ⬎37.4°C
Admission 12–24 h
(1) temp ⱕ37.4°C
(2) temp ⬎37.4°C
Study
166
*
Downloaded from http://stroke.ahajournals.org/ by guest on July 12, 2017
Admission at any time
(1) temp ⬍37.9°C
(2) temp ⱖ37.9°C
Admission ⱕ24 h
(1) temp ⱕ37.5°C
(2) temp ⬎37.5°C
123
54
Terent and
Andersson
(1981)9
Admission at any time
(1) temp ⬍38.0°C
(2) temp ⱖ38.0°C
196
42
Castillo et al
(1998)17
Admission ⬍24 h
(1) temp ⱕ37.5°C
(2) temp ⬎37.5°C
158
102
69.8 (⫾10.2)/
59%
Sharma and
Ross
(1998)7
Admission ⬍24 h
(1) temp ⱕ37.0°C
(2) temp ⬎37.0°C
209
87
74/
44.8%
MacWalter
et al
(1998)8
Admission ⬍24 h
(1) temp ⱕ37.5°C
(2) temp ⬎37.5°C
1555
73
73.3 (⫾10.9)/
48.6%
Huo and
Zhao
(1997)10
Admission ⬍24 h
(1) temp ⱕ37.5°C
(2) temp ⬎37.5°C
203
17
Range 40–78/
59.5%
141
41
77.2 (⫾10.1)/
42.6%
66.6/
50.8%
*
P
Mortality
Morbidity
Mortality
Scandinavian
Stroke Scale
at discharge
During
hospital
stay
0.0024
*
Scandinavia
Stroke Scale
at discharge
During
hospital
stay
0.37
232
Azzimondi
et al
(1995)15
Castillo et al
(1994)16
Morbidity
OR
P
OR
Conclusion
0.0126
1.8
(1.1–2.8)
Positive
*
0.04
2.1
(1.04–4.4)
Negative
for morbidity
0.19
*
0.13
*
Negative
0.0063
3.4
(1.2–9.5)
Positive
䡠䡠䡠
30 d
*
*
Neurological,
Barthel, and
Glasgow
scores
at 6 mo
6 mo
⬍0.001
*
⬍0.001
*
Positive
Complete
paresis
3 mo
⬍0.01
*
⬍0.001
*
Positive
Canadian
Stroke Scale
at 3 mo,
Barthel Index
at 3 mo
3 mo
1.68
(0.84–3.4)
1.85
(0.88–3.88)
⬍0.001
*
Negative
for morbidity
Negative
for morbidity
䡠䡠䡠
3 mo
䡠䡠䡠
䡠䡠䡠
0.498
0.91
(0.6–1.4)
Negative
12 mo
䡠䡠䡠
䡠䡠䡠
0.648
1.012
(0.79–1.31)
Negative
During
hospital
stay
䡠䡠䡠
䡠䡠䡠
*
Positive
0.142
0.213
䡠䡠䡠
䡠䡠䡠
⬍0.01
Temp indicates temperature.
*Data not given by that study.
morbidity and mortality with the Fisher method12; P⬍0.05 was
considered statistically significant.
Results
Nine studies were identified as suitable for inclusion,7–10,13–17
1 of which was not in the English language10 and 1 of which
was published in abstract form,8 totaling 3790 patients
(49.8% male; mean age, 73.0 years) (Table). This provided
our study with 99% power to detect a 9% increase in
morbidity and 84% power to detect a 1% increase in mortality
for the pyrexial group.
Three of the 9 studies9,14,16 showed significantly higher
morbidity and mortality in pyrexial patients (Table). In 2
additional studies,10,15 mortality was the only outcome measured and was significantly higher with associated pyrexia.
Another 2 studies13,17 found mortality, but not morbidity, to
be associated with pyrexia. In the first of these studies,10 this
was true only of a subgroup of patients admitted within 6 to
12 hours of onset of stroke; a second subgroup admitted 12 to
24 hours after onset of stroke showed no association between
pyrexia and mortality. Mortality was not found to be associated with pyrexia in the remaining 2 studies,7,8 in which it was
used as the sole outcome measure (Table).
Three additional studies18 –20 were identified but were not
suitable for inclusion. The first of these18 found that recovery
of neurological deficit after stroke was greater in apyrexial
than in pyrexial patients, but the significance of these results
could not be calculated because of the small numbers of
patients used. Furthermore, these authors were unable to
accurately distinguish between cerebral infarction and hemorrhage because of lack of availability of CT scans when this
study was published, thus compromising the potential accu-
412
Stroke
February 2000
ORs with 95% CIs for mortality.
Downloaded from http://stroke.ahajournals.org/ by guest on July 12, 2017
racy of our meta-analysis. The second study19 compared
differences in body temperature between patients with stable
and progressing stroke and concluded that body temperature
was independently related to progressing stroke (P⬍0.0001).
This study was not, however, included in the meta-analysis
for 2 reasons: first, it did not directly compare survival or
outcome between pyrexial and apyrexial patients, and second,
the study analyzed temperature as a continuous variable and
did not stratify patients into pyrexial and apyrexial groups. A
12th study in Japanese was identified with our search protocol.20 However, the authors sought to find a relationship
between changes in blood pressure and body temperature and
the site of the lesion only in stroke patients who were
autopsied. Hence, this study did not meet the inclusion
criteria of our meta-analysis and was therefore also excluded.
The combined OR for mortality was 1.19 (95% CI, 0.99 to
1.43) (Figure ). The result of the heterogeneity test (P⬎0.05)
indicated that the studies were similar enough to be meta-analyzed. Combining the probability values of individual studies suggested that both morbidity (P⬍0.0001) and mortality
(P⬍0.00000001) were significantly higher in the pyrexial
versus apyrexial patient groups.
Discussion
There is a considerable body of evidence from the animal
literature that raised body temperature after the onset of
stroke is associated with poor outcome. However, most
human studies have been undertaken on small numbers of
patients and give conflicting results. In our study there were
inadequate OR values available for a combined OR to be
calculated for morbidity. The combined OR for mortality
suggested a 19% increase in mortality in pyrexial versus
apyrexial patients, but the lower CI just crossed the line for no
effect. This result may be because of the 5 OR values
available, the vast majority of patients were from 2 of the
studies7,8 that showed no correlation between pyrexia and
mortality and hence may have heavily influenced the combined OR. However, the combined probability value, calcu-
lated with data from all 9 studies, is more representative and
importantly was highly significant for both morbidity and
mortality. From the results of our meta-analysis, we can
conclude that hyperthermia after the onset of stroke has a
significantly detrimental effect on stroke outcome.
The etiology of hyperthermia after stroke is not always
evident. Patients with intraventricular or subarachnoid hemorrhages and brain stem infarcts are thought to have greater
“central” or “neurogenic” fever.14 In our analysis patients
with subarachnoid hemorrhage were excluded by 1 of the
studies.15 A second study9 excluded subdural and epidural
hemorrhages, and 2 additional studies16,17 excluded all hemorrhages. The remaining investigators included all causes of
stroke.7,8,10,13,14 Clearly, this could lead to discrepancies in the
results of individual studies and should be borne in mind
during the interpretation of this meta-analysis. Naturally,
superimposed infection may also account for pyrexia. One
study16 excluded patients with any evidence of infection.
Three other studies9,14,17 also investigated evidence of infection. Of these, 1 study14 found infection to be present in
⬍20% of their pyrexial patients. Unlike body temperature,
neither infection nor leukocytosis had a significant effect on
stroke outcome. The second study9 found the presence of
bronchopneumonia in 50% of pyrexial patients. The third
study17 found that 57.6% of hyperthermic patients had an
infectious cause (mainly pulmonary and urinary sources).
However, coexistent infection within the first 3 days of stroke
was not independently associated with poor prognosis.17
Results from previous investigators generally support these
findings but also emphasize that fever may be directly related
to the size of the cerebral lesion.21 Combined, this suggests
that the presence of infection in the studies included in our
meta-analysis would not have accounted for, or even contributed to, the associated poor outcome.
Castillo et al17 studied the effect of hyperthermia at
different times after the onset of stroke. Hyperthermia within
72 hours of stroke significantly increased mortality. Only
Hajat et al
Poststroke Pyrexia: A Meta-Analysis of Studies in Patients
Downloaded from http://stroke.ahajournals.org/ by guest on July 12, 2017
hyperthermia within the first 24 hours after stroke, however,
caused significantly greater morbidity. When hyperthermia
occurred after 24 hours, it was not an independent risk factor
for poor outcome.17 Since none of the other studies stratified
patients according to time of onset of hyperthermia, only the
72-hour data were used in our analyses. Clearly, the extent of
cerebral damage is related to the timing of hyperthermia
onset, and this important relationship has been shown previously in both experimental and clinical studies.2,5,14
The mechanism for the poor outcome seen after hyperthermia remains speculative. The area of reversibly impaired
neuronal function surrounding the infarcted tissue, known as
the ischemic penumbra, is thought to be the site where
temperature-dependent stroke progression occurs. Several
mechanisms have been postulated to explain this effect of
hyperthermia. Neurotransmitters associated with poor cerebral infarct outcome, such as glutamate, ␥-aminobutyric acid,
and glycine, have been shown to increase during hyperthermia and to diminish with hypothermia.22 Increased free
radical production is another possible mechanism.23 The
temperature-sensitive blood-brain barrier is a possible means
of stroke progression due to hyperthermia. Indeed, protein
transfer across the blood-brain barrier that occurs after
periods of normothermic global ischemia is attenuated during
intraischemic hypothermia (30°C to 33°C) and markedly
increased during intraischemic hyperthermia (39°C) in rats.24
Temperature has a significant influence on intracerebral
metabolism. Animal studies have shown temperaturedependent changes in the levels of ATP,25,26 phosphocreatine,26 and calcium/calmodulin-dependent protein kinase II27
after periods of global cerebral ischemia.
Although the results of this study are strongly positive, its
limitations must be recognized. These include the welldocumented problems associated with all meta-analyses,28
particularly, but not confined to, publication bias. Furthermore, there were clear differences between the parameters
used for patient inclusion into the individual studies. The
subtypes of stroke included in each study are heterogeneous.
Since different etiologies and severities of stroke are thought
to cause neurogenic pyrexia to differing extents, this could
have influenced the final result. Other discrepancies between
studies included the statistical methods used, the means of
testing for morbidity, the duration at which morbidity and
mortality were measured, and, in particular, the measurement
of pyrexia. Indeed, pyrexia was measured at varying times
after the onset of stroke in different studies, ranging from a
single reading on admission14 to pyrexia within 7 days of
admission.9,15 The definition of pyrexia was also discrepant
between studies, ranging from ⬎37.0°C7 to ⱖ38.0°C.9 One
study further subdivided pyrexial patients into those with low
and high fever.15 However, for the purpose of our analyses
patients were dichotomized into those with no (or low) fever
and those with high fever (Table ). The discrepancy in the
proportion of patients with pyrexia, which varied from 4.5%8
to 60.8%,17 could be explained neither by the definition of
pyrexia nor by the duration of time for which body temperature was monitored after the stroke.
If these limitations of meta-analyses are accepted, our
results clearly suggest a detrimental effect of hyperthermia on
413
stroke outcome, with some evidence of greater effect with
early onset of pyrexia.17 This obviously leads to the question
of whether hypothermia could, conversely, offer a beneficial
effect. In the setting of acute traumatic brain injury, induced
hypothermia has been shown to significantly improve outcome for up to 6 months in patients with initial Glasgow
Coma Scale scores of 5 to 7.29 Interestingly, induced hypothermia after head injury has been shown to reduce lactate
production (representative of cerebral ischemia) as well as
intracranial hypertension.30 Moreover, induced moderate hypothermia (33°C) for 48 to 72 hours in 25 patients with
middle cerebral artery territory infarction with elevated intracranial pressure (within 14⫾7 hours of stroke onset) significantly reduced intracranial pressure, with the investigators
suggesting improved long-term outcome among survivors.31
The message from this study is that careful emphasis
should be placed on the fastidious control of pyrexia, particularly in the early poststroke period. We await further clinical
trials on the possible beneficial effects of therapeutic hypothermia. Until then, normothermia must remain the goal in
the management of acute stroke.
Acknowledgments
We would like to thank Dr J.C. Sharma and Dr R. MacWalter for
kindly supplying unpublished data from their studies, and Dr J. Hon
for his help with translation. Pankaj Sharma is a BHF Clinician
Scientist and a Fulbright Scholar.
References
1. Carlberg B, Asplund K, Hagg E. Course of blood pressure in different
subsets of patients after acute stroke. Cerebrovasc Dis. 1991;1:281–287.
2. Busto R, Dietrich WD, Globus MY, Valdes I, Scheinberg P, Ginsberg
MD. Small differences in intraischaemic brain temperature critically
determine the extent of ischaemic injury. J Cereb Blood Flow Metab.
1987;7:729 –738.
3. Kim Y, Busto R, Dietrich WD, Kraydieh S, Ginsberg MD. Delayed
postischemic hyperthermia in awake rats worsens the histopathological
outcome of transient focal cerebral ischemia. Stroke. 1996;27:
2274 –2281.
4. Maher J, Hachinski V. Hypothermia as a potential treatment for cerebral
ischaemia. Cerebrovasc Brain Metab Rev. 1993;5:277–300.
5. Ginsberg MD, Sternau LL, Globus MY, Dietrich WD, Busto R. Therapeutic modulation of brain temperature: relevance to ischaemic brain
injury. Cerebrovasc Brain Metab Rev. 1992;4:189 –225.
6. Baker CJ, Onesti ST, Solomon RA. Reduction by delayed hypothermia of
cerebral infarction following middle cerebral artery occlusion in the rat:
a time-course study. J Neurosurg. 1992;77:438 – 444.
7. Sharma JC, Ross IN. Antipyretic therapy in acute stroke. Lancet. 1998;
352:740 –741.
8. MacWalter R, McMahon A, Fraser H, Bruce V, Hendrick S. Does body
temperature on admission predict long-term outcome after an acute
stroke? Cerebrovasc Dis. 1998;8(suppl 4):11. Abstract.
9. Terent A, Andersson B. Prognosis for patients with cerebrovascular
stroke and transient ischaemic attacks. Ups J Med Sci. 1981;86:63–74.
10. Huo CN, Zhao YX. Relationship between fever and mortality in patients
with acute ischemic cerebrovascular stroke [in Chinese]. Chin J Nursing.
1997;32:640 – 642.
11. Woolf B. On estimating the relation between blood group and disease.
Ann Hum Genet. 1955;19:251–253.
12. Fisher RA. Statistical Methods for Research Workers. Edinburgh,
Scotland: Oliver and Boyd; 1932.
13. Jorgensen HS, Reith J, Pedersen PM, Nakayama H, Olsen TS. Body
temperature and outcome in stroke patients. Lancet. 1996;348:193. Letter.
14. Reith J, Jorgensen HS, Pedersen PM, Nakayama H, Raaschou HO, Jeppesen
LL, Olsen TS. Body temperature in acute stroke: relation to stroke severity,
infarct size, mortality and outcome. Lancet. 1996;347:422–425.
414
Stroke
February 2000
Downloaded from http://stroke.ahajournals.org/ by guest on July 12, 2017
15. Azzimondi G, Bassein L, Nonino F, Fiorani L, Vignatelli L. Fever in acute
stroke worsens prognosis: a prospective study. Stroke. 1995;26:2040–2043.
16. Castillo J, Martinez F, Leira R, Prieto JM, Lema M, Noya M. Mortality
and morbidity of acute cerebral infarction related to temperature and basal
analytic parameters. Cerebrovasc Dis. 1994;4:56 –71.
17. Castillo J, Davalos A, Marrugat J, Noya M. Timing for fever-related brain
damage in acute ischemic stroke. Stroke. 1998;29:2455–2460.
18. Hindfelt B. The prognostic significance of subfebrility and fever in
ischaemic cerebral infarction. Acta Neurol Scand. 1976;53:72–79.
19. Davalos A, Castillo J, Pumar JM, Noya M. Body temperature and
fibrinogen are related to early neurological deterioration in acute ischemic
stroke. Cerebrovasc Dis. 1997;7:64 – 69.
20. Tohgi H, Yamanouchi H, Ogawa M, Uchiyama S, Tabouchi M. Changes in
blood pressure and body temperature during the acute stage of stroke and its
implication on prognosis [in Japanese]. Jpn J Geriatr. 1976;13:334–339.
21. Przelomski MM, Roth RM, Gleckman RA, Marcus EM. Fever in the
wake of a stroke. Neurology. 1986;36:427– 429.
22. Ginsberg MD, Busto R. Combating hyperthermia in acute stroke: a
significant clinical concern. Stroke. 1998;29:529 –534.
23. Globus MY, Busto R, Lin B, Schnippering H, Ginsberg MD. Detection of
free radical activity during transient global ischaemia and recirculation:
effects of intra-ischemic brain temperature modulation. J Neurochem.
1995;65:1250 –1256.
24. Dietrich WD, Halley M, Valdes I, Busto R. Interrelationships between
increased vascular permeability and acute neuronal damage following
25.
26.
27.
28.
29.
30.
31.
temperature-controlled brain ischaemia in rats. Acta Neuropathol (Berl).
1991;81:615– 625.
Ginsberg MD, Busto R, Martinez E. The effects of cerebral ischaemia on
energy metabolism. In: Schousboe A, Diemer NH, Kofod H, eds. Drug
Research Related to Neuroactive Amino Acids: Alfred Benzon Symposium
32. Copenhagen, Denmark: Munksgaard; 1992:207–224.
Chopp M, Welch KM, Tidwell CD, Knight R, Helpern JA. Effect of mild
hyperthermia on recovery of metabolic function after global cerebral
ischemia in cats. Stroke. 1988;19:1521–1525.
Busto R, Globus MY-T, Neary JT, Ginsberg MD. Regional alterations of
protein kinase activity following transient cerebral ischaemia: effects of
intraischaemic brain temperature modulation. J Neurochem. 1994;63:
1095–1103.
Prins JM, Buller HR. Meta-analysis: the final answer, or even more
confusion? Lancet. 1996;348:199. Letter.
Marion DW, Penrod LE, Kelsey SF, Obrist WD, Kochanek PM, Palmer
AM. Treatment of traumatic brain injury with moderate hypothermia.
N Engl J Med. 1997;336:540 –546.
Metz C, Holzschuh M, Bein T, Woertgen C, Frey A, Frey I, Taeger K.
Moderate hypothermia in patients with severe head injury: cerebral and
extracerebral effects. J Neurosurg. 1996;85:533–541.
Schwab S, Schwarz S, Spranger M, Keller E, Bertram M, Hacke W.
Moderate hypothermia in the treatment of patients with severe middle
cerebral artery infarction. Stroke. 1998;29:2461–2466.
Effects of Poststroke Pyrexia on Stroke Outcome: A Meta-Analysis of Studies in Patients
Cother Hajat, Shakoor Hajat and Pankaj Sharma
Downloaded from http://stroke.ahajournals.org/ by guest on July 12, 2017
Stroke. 2000;31:410-414
doi: 10.1161/01.STR.31.2.410
Stroke is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 2000 American Heart Association, Inc. All rights reserved.
Print ISSN: 0039-2499. Online ISSN: 1524-4628
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://stroke.ahajournals.org/content/31/2/410
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published
in Stroke can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.
Once the online version of the published article for which permission is being requested is located, click
Request Permissions in the middle column of the Web page under Services. Further information about this
process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Stroke is online at:
http://stroke.ahajournals.org//subscriptions/