Higher Frequency of A

trial Fibrillation Linkedto Colder Seasons and Air Temperature on the Day

of Ischemic Stroke Onset

Osvaldo Fustinoni, MD,* Gustavo Saposnik, MD, MSc, FAHA,†‡xjj{Mar�ıa Martha Esnaola y Rojas, MD,# Susan G. Lakkis, PhD,** and

Luciano A. Sposato, MD, MBA, FAHA,††‡‡ on behalf of ReNACer Investigators

From the *Cerebrovas

Buenos Aires, Buenos Ai

Center, St Michael’s Hosp

edge Institute, St Michae

of Medicine, University

of Health Policy Manage

Ontario, Canada; {InstituCanada; #Department of

nos Aires, Argentina; **E

Cambio Global, Pontific

Madero, Argentina; ††Str

University Hospital, Fav

476

Background:Whether a seasonal variation of atrial fibrillation among acute ischemic

stroke (AIS) patients occurs is unknown. We studied the distribution of atrial fibril-

lation across seasons and air temperatures in a cohort of AIS patients. Methods: We

selected 899 AIS patients from the Argentinean Stroke Registry (ReNACer), who

were admitted to 43 centers in the Province of Buenos Aires. We recorded the min-

imum and maximum temperatures at local weather centers on the day and the city

where each stroke occurred. We used the goodness-of-fit c2 test to assess the distri-

bution of atrial fibrillation across seasons and air temperatures and the Pearson cor-

relation coefficient to assess the relationship between these variables. We developed

a regression model for testing the association between seasons and atrial fibrillation.

Results: We found a seasonal variation in the occurrence of atrial fibrillation, with

a peak inwinter and a valley in summer (23.1% versus 14.0%, P,.001). The semester

comprised by autumn and winter was associated with atrial fibrillation (Pearson

P , .001). Atrial fibrillation showed a nonhomogeneous distribution across ranges

of temperature (P , .001, goodness-of-fit test), with a peak between 5�C and 9�C,and was associated with minimum (Pearson P 5 .042) and maximum (Pearson

P 5 .002) air temperature. After adjusting for significant covariates, there was a 2-

fold risk of atrial fibrillation during autumn and winter. Conclusions: In this cohort

of AIS patients, atrial fibrillation showed a seasonal variation and a nonhomoge-

neous distribution across air temperatures, with peaks in cold seasons and low tem-

peratures on the day of stroke onset. Key Words: Ischemic stroke—atrial

fibrillation—risk—season—temperature—weather.

� 2013 by National Stroke Association

cular Diseases, Instituto de Neurociencias

res, Argentina; †Stroke Outcomes Research

ital, Ontario, Canada; ‡Li Ka Shing Knowl-

l’s Hospital, Ontario, Canada; xDepartment

of Toronto, Ontario, Canada; jjDepartment

ment and Evaluation, University of Toronto,

te for Clinical Evaluative Sciences, Ontario,

Neurology, Hospital Dr Cesar Milstein, Bue-

quipo Interdisciplinario para el Estudio del

ia Universidad Cat�olica Argentina, Puerto

oke Center at the Institute of Neurosciences,

aloro Foundation, Buenos Aires, Argentina;

and ‡‡Vascular Research Unit, INECO Foundation, Buenos Aires,

Argentina.

Received February 7, 2013; revision received March 6, 2013;

accepted March 7, 2013.

Address correspondence to Luciano A. Sposato, MD, MBA, FAHA,

Stroke Center at the Institute of Neurosciences, University Hospital,

Favaloro Foundation and Vascular Research Unit, INECO Founda-

tion, Rosales 2616, Olivos, Buenos Aires, Argentina. E-mail:

lucianosposato@gmail.com.

1052-3057/$ - see front matter

� 2013 by National Stroke Association

http://dx.doi.org/10.1016/j.jstrokecerebrovasdis.2013.03.009

Journal of Stroke and Cerebrovascular Diseases, Vol. 22, No. 4 (May), 2013: pp 476-481

ISCHEMIC STROKE, SEASONS, AND TEMPERATURE 477

Background

Winter peaks in cardiac events and stroke have been

observed in various regions of the world.1-7 Atrial

fibrillation (AF), one of the commonest cardiac

arrhythmias, is a well-established risk factor for embolic

stroke8 and peaks in winter as well.9-14 A few studies

have reported a stroke risk increase in winter in patients

with AF,13,14 but, to the best of our knowledge, no study

has specifically investigated the seasonal variation of AF

in patients admitted with an acute ischemic stroke

(AIS). Furthermore, the reasons for the higher incidence

of AF and stroke during winter are uncertain. Some

studies have linked it to lower mean monthly air

temperatures.15,16 However, studies carried out in

warmer climates seeking to demonstrate an association

between low air temperature and a higher incidence of

AF have yielded conflicting results.16,17 One study

reported that approaching cold temperature fronts were

followed by an increase in hospital AF-related admis-

sions,18 so perhaps steep temperature drops, rather than

stable low air temperatures, might be associated to AF on-

set. The aim of this study was to assess seasonal variation

of AF in patients admitted with AIS in a region with rel-

atively warm air temperatures and to investigate whether

this seasonal variation could be explained by colder air

temperatures on the day of AIS onset.

Methods

ReNACer

ReNACer is an Argentinean prospective, countrywide,

hospital-based stroke registry comprising 74 medical in-

stitutions. We included every consecutive patient admit-

ted for AIS to each participating center between

November 2004 and October 2006. Further details are

reported elsewhere.19,20

Buenos Aires Province

Argentina has a total surface area of 2,780,400 km2

(1,073,500 mi2). Because of its North–South extension

(3940 km2/448 mi2), the weather is highly variable

throughout its different regions. Assessing the seasonal

variation of AF in patients admitted with AIS across,

the whole country would have been therefore limited by

weather variability. Consequently, we selected patients

admitted to 43 centers from 17 cities of Buenos Aires, Ar-

gentina’s largest and most crowded province (307,571

km2/118,754 mi2, population of 15.6 million, over 39%

of the country’s total). As part of the ‘‘Pampa,’’ a wide

plain in the center and east of the country, highly influ-

enced by oceanic climate, it has a humid subtropical cli-

mate21 with a very homogeneous weather, with uniform

hot summers (mean maximum temperatures of

22.1�C 6 3.2�C or 71.7�F 6 10.7�F for the period

1990-2009) and moderate cold winters (mean minimum

temperatures of 10.5�C 6 1.6�C or 50.9�F 6 6.9�F for the

period 1990-2009) Fig 1.

Patients

The original cohort of ReNACer comprised 1991 pa-

tients with a diagnosis of AIS. For the present analysis,

we selected patients aged 18 years and older, admitted

within 72 hours of AIS onset to every institution of the

Province of Buenos Aires. The final study cohort com-

prised 899 patients with a confirmed diagnosis of AIS

admitted to 43 centers. Patients with AIS were admitted

to the general ward and those with more severe strokes

were admitted to the intensive care unit or the coronary

care unit (CCU). We recorded and completed the data

prospectively on a real-time basis during each patient’s

hospital stay up to discharge.

Every patient underwent cerebral computed tomogra-

phy, magnetic resonance imaging, or both.

We assessed demographic and vascular risk factors.

Seasonal and Weather Data

A specialist in atmospheric sciences (S.G.L.) retrieved

the minimum and maximum temperatures collected

from the local weather centers on the day and the city

where each of the 899 AIS occurred. The data were pro-

vided by the National Weather Service and the Intergov-

ernmental Panel on Climate Change Fourth Assessment

Report. We thereby obtained the actual temperatures of

the day and place of stroke onset.

Definitions

We defined AIS in accordance with the World Health

Organization criteria22 and the absence of blood in neuro-

imaging studies. AF was defined by the absence of

P waves or evidence of atrial flutter with irregular ventric-

ular response.

Diagnosis of AF

On admission, each patient with AIS underwent

a 12-lead electrocardiographic (ECG) evaluation. Further

ECGs were ordered only when a patient became symp-

tomatic while hospitalized in the general ward (eg, palpi-

tations, chest pain). Patients admitted to intensive care

unit and coronary care unit underwent ECG monitoring.

Statistical Analysis

Patients were grouped according to the season of stroke

occurrence during the 2-year study period. We used the

goodness-of-fit c2 test to assess the distribution of AF

across seasons and air temperatures, and the Pearson cor-

relation coefficient to assess the relationship between

these variables. The null hypothesis was a homogeneous

Table 1. Seasonal pattern of AF among AIS patients

Season Proportion of AF % (95% CI)

Autumn 22.9 (17.7-29.2)

Winter 23.1 (17.3-30.2)

Spring 17.4 (13.1-22.8)

Summer 14.0 (10.6-18.4)

P (goodness of fit) ,.001

Abbreviations: AF, atrial fibrillation; CI, confidence interval.

O. FUSTINONI ET AL.478

distribution of the rate of AF along seasons and tempera-

tures. A 2-tailed P value ,.05 was deemed statistically

significant for all the analyses. We plotted the relationship

of air temperatures with the total number of AIS patients

with AF across seasons (Fig 3).

We used the c2 and Fisher exact tests to compare cate-

gorical variables, and the Mann–Whitney U test and Stu-

dent t test to compare continuous variables in a univariate

analysis (Table 2). Finally, we developed a forward step-

by-step logistic regression model for adjusting the rela-

tionship between seasons and temperatures with the

rate of AIS patients with AF. We included variables with

a P value ,.2 in the univariate analysis and those biolog-

ically relevant (age, gender, hypertension, history of coro-

nary artery disease, and left atrial enlargement). Results

were expressed as odds ratios (ORs) with 95% confidence

intervals (CIs).

Ethical Considerations

The protocol was approved by the Argentinean Neu-

rological Society and the Argentinean Ministry of Health.

We conducted the study in accordance with Good Clini-

cal Practice, all applicable subject privacy and confiden-

tiality requirements, and the guiding principles of the

declaration of Helsinki. We deleted any personal infor-

mation, so that the collected data remained anonymous.

Patient-informed consent was thus not necessary accord-

ing to regulatory laws. Only authorized individuals had

access to study-related documents. The database was

registered at the Argentinean National Office for the

Protection of Personal Data (Ministry of Justice and

Human Rights).

Results

A total of 899 AIS patients were included in the study,

of whom 518 (57.6%) were males. The mean age of the

study cohort was 70 6 13 years. The distribution of pa-

tients across seasons was as follows: autumn 205 (22.8%,

95% CI 20.2-25.7), winter 160 (17.8%, 95% CI 15.4-20.4),

spring 235 (26.1%, 95% CI 23.4-29.1), and summer 299

(33.3%, 30.3-36.4).

A total of 167 patients had AF (18.6%, 95% CI 16.2-21.3).

Among patients with AF, 39 where diagnosed after ad-

mission (4.3%, 95% CI 3.2-5.9).

As shown in Table 1, the rate of AF among AIS patients

showed a marked seasonal variation (P , .001) with

a peak in winter (23.1%) and a valley in summer

(14.0%). There was also a correlation between the semes-

ter comprised by autumn and winter and the occurrence

of AF (Pearson P , .001).

The frequency of AIS with AF showed a nonhomoge-

neous distribution across ranges of temperature

(P , .001, goodness-of-fit test), with peaks for minimum

and maximum temperatures ranging from 5�C to 9�C(Fig 2). There was an association between the percentage

of patients with AF and the minimum (Pearson P 5 .042)

and maximum (Pearson P5 .002) air temperatures on the

day of stroke onset. Figure 3 shows the inverse association

of temperature with AF across seasons. The higher AF

rates were found during the colder seasons (autumn

and winter) and vice versa.

The univariate analysis comparing patients with and

without AF is shown in Table 2. There were no differences

in the rate of fever in patients with or without AF. In the

forward step-by-step regression model, AIS occurring

during autumn and winter was associated with a higher

AF rate (OR 2.0, 95% CI 1.2-3.4) after adjusting for age,

gender, diabetes mellitus, coronary heart disease, smok-

ing, and left atrial enlargement. Age (OR 1.0, 95% CI

1.0-1.1), left atrial enlargement (OR 5.9, 95% CI 3.5-10.4),

and diabetes mellitus (OR 2.3, 95% CI 1.2-4.7) were also

significantly associated with AF in the model.

Discussion

In a fairly large sample of AIS patients hospitalized in

a wide area with homogeneous weather conditions, we

Figure 1. Average weather characteristics for

Buenos Aires province stratified by month

(decade 2000-2010). Source: Argentinean

National Weather Service.

Figure 2. Frequency of AF across ranges

of maximum and minimum temperatures.

Abbreviations: AF, atrial fibrillation; MaxT,

maximum temperature; MinT, minimum

temperature.

ISCHEMIC STROKE, SEASONS, AND TEMPERATURE 479

found a marked seasonal pattern in the frequency of cases

with AF, with a peak in winter and a valley in summer.

After adjusting for significant covariates, patients experi-

encing a stroke during autumn and winter had a 2-fold

higher risk of having an AF than those admitted during

summer and spring. Also, there was an inverse correla-

tion between minimum and maximum air temperatures

on the day of stroke onset and the percentage of patients

with AF, with a nonhomogeneous distribution.

AFhasbeen shown toprevail inwinter.9-14 Furthermore,

we have shown a higher frequency of AF in winter among

our relatively large cohort of AIS. To date, there is no

definite explanation for the higher frequency of AF

Figure 3. Minimum and maximum temperatures plotted against the oc-

currence of AF across seasons. Abbreviations: AF, atrial fibrillation; MaxT,

maximum temperature; MinT, minimum temperature.

during cold seasons among the general population.

Lower air temperatures enhancing sympathetic function

by activation of central angiotensin or hypothalamic

receptors for mineralocorticoids could trigger episodes

of AF.23-26 Also, blood pressure rises because of cold

air temperatures (cold-induced hypertension) could

increase intra-atrial pressure, leading to atrial enlargement

and subsequent AF.27 Additionally, cold could induce ele-

vations of endothelin 1, renin, and angiotensin II plasmatic

levels, resulting in an increased risk of AF.28-31 Finally, the

lower sunlight exposure during winter could bring on

changes in cardiac ion channels, which in turn could

result in shorter action potentials, further facilitating

re-entry mechanisms.32

Fewer explanations have been proposed to account for

the higher frequency of AF during wintertime among

AIS patients. However, our finding of a positive associ-

ation between cold temperatures on the day of stroke

onset and a higher frequency of AF could be partially ex-

plained by the aforementioned relationship between

those temperatures and a higher risk of AF. Additionally,

a variety of thrombotic conditions, such as ischemic

stroke,33,34 acute myocardial infarction,35 deep vein

thrombosis,36 and pulmonary embolism,37 have been

shown to prevail in winter. The underlying mechanism

could be a higher occurrence of inflammatory and infec-

tious events in cold seasons subsequently triggering pro-

thrombotic states and thrombotic incidents.38,39 The

concurrent occurrence of a higher number of cases of

AF9-14 and a variety of prothrombotic conditions38,39

during winter could partially explain the higher

incidence of AF among AIS under cold temperatures.

As an increase in hospital AF-related admissions has

been seen to follow approaching cold temperature

fronts,18 abrupt air temperature falls might activate

mechanisms such as these, leading to AF, to stroke,

and maybe to other thrombotic incidents as well. In

line with this hypothesis, some studies have demon-

strated a higher risk of stroke among AF patients during

winter.13,14

Table 2. Univariate analysis comparing AIS patients with and without AF

With AF (n 5 167) Without AF (n 5 732) P

Demographic data

Age, y, mean (SD) 76 (11) 69 (13) ,.001Gender, men, % 52.1 59.0 .10

Risk factors

Hypertension, % 80.4 78.9 .66

Diabetes mellitus, % 16.8 24.2 .039

Coronary heart disease, % 17.2 28.2 .001Smoking, % 32.1 14.9 ,.001Left atrial enlargement, % 71.1 30.9 ,.001

Seasons and temperatures

Autumn/winter, % 50.3 38.4 .005Median MaxT (IQR), �C 22.3 (18.0-27.5) 24.0 (19.8-28.0) .004Median MinT (IQR), �C 13.4 (8.0-17.8) 14.3 (9.7-18.7) .027

Fever 11.4 10.2 .67

Abbreviations: AIS, acute ischemic stroke; AF, atrial fibrillation; IQR, interquartile range; MaxT, maximum temperature; MinT, minimum

temperature; MeanT, mean temperature.

Bold values are those with P , 0.05 (significant values)

O. FUSTINONI ET AL.480

Our study has several limitations. First, infections and

fever could have triggered some episodes of AF. We

were not able to document which patients had fever or

were diagnosed with an infection on admission. How-

ever, we could determine which patients developed fever

during hospitalization and our data did not show differ-

ences in the rate of fever in patients with or without AF,

suggesting that the higher rates of AF were not necessar-

ily linked to fever. Second, hospitalized patients might

not be representative of the whole spectrum of cerebro-

vascular disease in the study area, given that a consider-

able number of stroke patients are often not referred for

admission. Only a population-based approach could

have avoided this bias. Third, the observational design

of our study limits its inference of causation. Fourth, we

acknowledge the ecological nature of our study. However,

ecological studies can answer important questions that

cannot be addressed by other study designs (eg, random-

ized clinical trials).40 Fifth, because of the low number of

patients with newly diagnosed AF, we were not able to

analyze the influence of air temperature on the develop-

ment of AF after AIS.

Despite the aforementioned limitations, our study is

novel in the sense that we did not only analyze the sea-

sonal variation of the rate of AF among AIS patients but

also recorded the actual air temperature on the day each

stroke occurred, thereby avoiding the inconsistencies of

calculations or estimations. In addition, we were able to

demonstrate a higher risk of AF among ischemic patients

living under a temperate climate. Finally, though our

study was not population based, it comprised a relatively

high number of centers covering every region from the

wide territory of the Province of Buenos Aires.

In conclusion, among our cohort of AIS patients from

Buenos Aires, Argentina, we found a positive and inde-

pendent association between the occurrence of AF and

wintertime, as well as with low air temperatures on the

day of stroke onset. These results point to cold weather

as a hitherto unrecognized risk factor for stroke and could

have a bearing on prevention strategies in the future.

References

1. Sheth T, Nair C, Muller J, et al. Increased winter mortalityfrom acute myocardial infarction and stroke: the effect ofage. J Am Coll Cardiol 1999;33:1916-1919.

2. Kelly-Hayes M, Wolf PA, Kase CS, et al. Temporal pat-terns of stroke onset: the Framingham Study. Stroke1995;26:1343-1347.

3. Douglas AS, Russell D, Allan TM. Seasonal, regional andsecular variations of cardiovascular and cerebrovascularmortality in New Zealand. Aust N Z J Med 1990;20:669-676.

4. Khan FA, Engstrom G, Jerntorp I, et al. Seasonal patternsof incidence and case fatality of stroke in Malmo, Swe-den: The STROMA study. Neuroepidemiology 2005;24:26-31.

5. Sobel E, Zhang ZX, Alter M, et al. Stroke in the LehighValley: seasonal variation in incidence rates. Stroke1987;18:38-42.

6. Turin TC, Kita Y, Murakami Y, et al. Higher strokeincidence in the spring season regardless of conventionalrisk factors: Takashima Stroke Registry, Japan, 1988-2001.Stroke 2008;39:745-752.

7. D�ıaz A, Gerschcovich ER, D�ıaz AA, et al. Seasonal varia-tion and trends in stroke hospitalizations andmortality ina South American Community Hospital. J Stroke Cere-brovasc Dis 2012. [Epub ahead of print, PubMed PMID:22609318].

8. Goldstein LB, Bushnell CD, Adams RJ, et al, AmericanHeart Association Stroke Council; Council on Cardiovas-cular Nursing; Council on Epidemiology and Prevention;Council for High Blood Pressure Research; Council onPeripheral Vascular Disease, and Interdisciplinary Coun-cil on Quality of Care and Outcomes Research.

ISCHEMIC STROKE, SEASONS, AND TEMPERATURE 481

Guidelines for the primary prevention of stroke: a guide-line for healthcare professionals from the American HeartAssociation/American Stroke Association. Stroke 2011;42:517-584.

9. Murphy NF, Stewart S, MacIntyre K, et al. Seasonal vari-ation in morbidity and mortality related to atrial fibrilla-tion. Int J Cardiol 2004;97:283-288.

10. Watanabe E, Kuno Y, TakasugaH, et al. Seasonal variationin paroxysmal atrial fibrillation documented by 24-hourHolter electrocardiogram. Heart Rhythm 2007;4:27-31.

11. Frost L, Johnsen SP, Pedersen L, et al. Seasonal variationin hospital discharge diagnosis of atrial fibrillation: a pop-ulation-based study. Epidemiology 2002;13:211-215.

12. Moutzouris DA, Hassid VJ. Seasonal variation of atrial fi-brillation: further supportive evidence. Epidemiology2003;14:127.

13. Frost L, Vukelic Andersen L, Mortensen LS, et al. Sea-sonal variation in stroke and stroke-associated mortalityin patients with a hospital diagnosis of nonvalvular atrialfibrillation or flutter. A population-based study in Den-mark. Neuroepidemiology 2006;26:220-225.

14. Christensen AL, Rasmussen LH, Baker MG, et al. Season-ality, incidence and prognosis in atrial fibrillation andstroke in Denmark and New Zealand. BMJ Open 2012;2:e001210.

15. Kupari M, Koskinen P. Seasonal variation in occurrenceof acute atrial fibrillation and relation to air temperatureand sale of alcohol. Am J Cardiol 1990;66:1519-1520.

16. Kiu A, Horowitz JD, Stewart S. Seasonal variation inAF-related admissions to a coronary care unit in a ‘‘hot’’climate: fact or fiction? J CardiovascNurs 2004;19:138-141.

17. Kountouris E, Korantzopoulos P, Dimitroula V, et al. Isthere a seasonal variation in hospital admissions foracute-onset atrial fibrillation? Cardiology 2005;103:79-80.

18. G1uszak A, Koco�n S, Zuk K, et al. Episodes of atrial fibril-lation and meteorological conditions. Kardiol Pol 2008;66:958-963.

19. Sposato LA, Esnaola MM, Zamora R, et al, ReNACer In-vestigators; Argentinian Neurological Society. Quality ofischemic stroke care in emerging countries: the Argenti-nian National Stroke Registry (ReNACer). Stroke 2008;39:3036-3041.

20. Sposato LA, Ioli P, Povedano G, et al, ArgentineanNeuro-logical Society and ReNACer Investigators. Unemploy-ment: a social risk factor associated with early ischemicstroke mortality? Results from the Argentinean NationalStroke Registry (ReNACer). J Stroke Cerebrovasc Dis2012;21:679-683.

21. Peel MC, Finlayson BL, McMahon TA. Updated worldmap of the K€oppen-Geiger climate classification. HydrolEarth Syst Sci 2007;11:1633-1644.

22. Hatano S. Experience from a multicentre stroke register:a preliminary report. Bull World Health Organ 1976;54:541-553.

23. Papanek PE, Wood CE, Fregly MJ. Role of the sympa-thetic nervous system in cold-induced hypertension inrats. J Appl Physiol 1991;71:300-306.

24. Sun Z, Cade R, Morales C. Role of central angiotensin IIreceptors in cold-induced hypertension. Am J Hypertens2002;15:85-92.

25. Sun Z, Bello Roufai M, Wang X. RNAi inhibition of min-eralocorticoid receptor prevents the development of cold-induced hypertension. Am J Physiol Heart Circ Physiol2008;294:H1880-H1887.

26. Kapa S, Venkatachalam KL, Asirvatham SJ. The auto-nomic nervous system in cardiac electrophysiology: anelegant interaction and emerging concepts. Cardiol Rev2010;18:275-284.

27. Kalifa J, Jalife J, Zaitsev AV, et al. Intra-atrial pressure in-creases rate and organization of waves emanating fromthe superior pulmonary veins during atrial fibrillation.Circulation 2003;108:668-671.

28. Chen GF, Sun Z. Effects of chronic cold exposure on theendothelin system. J Appl Physiol 2006;100:1719-1726.

29. Sun Z, Fregly MJ, Cade JR. Effect of renal denervation onelevation of blood pressure in cold-exposed rats. Can JPhysiol Pharmacol 1995;73:72-78.

30. Mayyas F, Niebauer M, Zurick A, et al. Association of leftatrial endothelin-1 with atrial rhythm, size, and fibrosis inpatients with structural heart disease. Circ ArrhythmElectrophysiol 2010;3:369-379.

31. Schneider MP, Hua TA, B€ohm M, et al. Prevention ofatrial fibrillation by renin-angiotensin system inhibitiona meta-analysis. J Am Coll Cardiol 2010;55:2299-2307.

32. Yamashita T, Sekiguchi A, Iwasaki YK, et al. Circadianvariation of cardiac K channel gene expression. Circula-tion 2003;107:1917-1922.

33. Emsley HCA, Tyrrell PJ. Inflammation and infection inclinical stroke. J Cereb Blood Flow Metab 2002;22:1399-1419.

34. Myint PK, Vowler SL, Woodhouse PR, et al. Winter excessin hospital admissions, in-patient mortality and length ofacute hospital stay in stroke: a hospital database studyover six seasonal years in Norfolk, UK. Neuroepidemiol-ogy 2007;28:79-85.

35. Spencer FA, Goldberg RJ, Becker RC, et al. Seasonaldistribution of acute myocardial infarction in the secondNational Registry of Myocardial Infarction. J Am CollCardiol 1998;31:1226-1233.

36. Gallerani M, Boari B, de Toma D, et al. Seasonal variationin the occurrence of deep vein thrombosis. Med Sci Monit2004;10:CR191-CR196.

37. Boulay F, Berthier F, Schoukroun G, et al. Seasonal varia-tions in hospital admission for deep vein thrombosis andpulmonary embolism: analysis of discharge data. BMJ2001;323:601-602.

38. Woodhouse PR, Khaw KT, Plummer M, et al. Seasonalvariations of plasma fibrinogen and factor VII activityin the elderly: winter infections and death from cardio-vascular disease. Lancet 2004;343:435-439.

39. Mattila KJ, Valtonen VV, Nieminen MS, et al. Role of in-fection as a risk factor for atherosclerosis, myocardial in-farction, and stroke. Clin Infect Dis 1998;26:719-734.

40. Tu JV, Ko DT. Ecological studies and cardiovascularoutcomes research. Circulation 2008;118:2588-2593.