Temporal Patterns of Stroke Onset
The Framingham Study
Background and Purpose Several studies have shown stroke onset to vary by season, day of the week, and time of day. These temporal patterns, which may provide insights into pathogenesis, were found mainly in clinical series, which can be subject to selection bias. To obtain a less distorted picture of stroke onset, we examined the month and season, day of the week, time, and place stroke occurred in a community-based cohort.
Methods Over a 40-year period of surveillance of the Framingham Study cohort of 5070 people aged 30 to 62 years and free of stroke and cardiovascular disease at entry, 637 completed initial strokes occurred. Month, season, day of the week, time of day, and place of occurrence of stroke were ascertained systematically and related prospectively to stroke incidence, subtype, and gender.
Results Winter was the peak season for cerebral embolic strokes. Significantly more stroke events occurred on Mondays than any other day, particularly for working men. For intracerebral hemorrhages, a third happened on Mondays in both genders. The time of day when strokes most frequently occurred was between 8 am and noon. This pattern was true for all stroke subtypes. This pattern persisted when individuals whose onset occurred while sleeping or on awakening were excluded. Stroke in general occurred more at home, with hemorrhagic strokes occurring outside the home and cerebral embolisms in the hospital more than other subtypes.
Conclusions Temporal patterns of stroke onset were observed for season, day of the week, time of day, and place in a community-based population. These findings suggest that there are periods of increased risk of stroke that may be amenable to preventive strategies.
The circumstances surrounding the onset of a stroke are not well understood. Recognizing the temporal patterns associated with cardiovascular disease and stroke can provide insight into what triggers the acute onset of stroke and eventually what preventive strategies can be developed. In clinical series the time of onset in myocardial and cerebral infarctions has been related to diurnal variations and daily activity patterns. In this study of a general population, complete case ascertainment, including those who were not hospitalized, should expand our current knowledge of patterns surrounding the acute stroke event.
Data on the temporal patterns of stroke in community-based samples are sparse.1 2 Most studies of diurnal variation have been based on clinical series.3 4 5 6 7 8 9 Several studies have documented seasonal variation in stroke incidence using both community-based1 10 and hospital-based2 11 12 13 14 15 16 populations. Few reports document the pattern of stroke onset by day of the week.17 18 To date, no community-based studies have examined these factors together.
We studied the temporal patterns surrounding all initial strokes occurring in the Framingham Study cohort over four decades from 1950 to 1990. By identifying these patterns of stroke occurrence we hope to generate hypotheses concerning the pathogenesis and precipitants of stroke.
Materials and Methods
The Framingham Study, consisting of a two-thirds sample of adults resident in the town of Framingham, Mass, in 1948, has been in existence for more than 40 years. The population of 5070 men and women aged 28 to 62 years who were free of cardiovascular disease and stroke at entry to the study in 1948-50 has been followed prospectively with biennial examinations. At each examination the participants give an informed consent that includes permission for an acute neurological evaluation if a stroke occurs. The biennial examinations have systematically characterized the subjects according to personal habits, medical history, physical examinations, and laboratory testing. Methods of recruitment, examination, and follow-up and criteria for diagnoses have been reported elsewhere.19 20 Surveillance has been maintained for cardiovascular disease and stroke on a daily basis, and since 1968 the study neurologists have evaluated subjects at the local hospital in Framingham at the time of the acute stroke. Information on stroke type, date, and time of first symptoms was determined for all cases by a panel of neurology investigators using the results of a detailed clinical examination, history of the occurrence, medical records, and imaging confirmation when the technology became available.
With the use of established criteria for stroke type confirmation, cases were classified according to mechanism as ABI, CE, ICH, or SH. Transient ischemic attacks, recurrent stroke events, and strokes determined to have resulted from medical interventions/procedures or trauma were excluded.
Temporal Pattern Determinations
Month, season, and day of the week were calculated from the date of stroke. Seasons were divided into spring (March, April, May), summer (June, July, August), autumn (September, October, November), and winter (December, January, February). For determination of the time of stroke occurrence, each 24-hour day was divided into the following six 4-hour intervals: 12:01 to 4 am, 4:01 to 8 am, 8:01 am to noon, 12:01 to 4 pm, 4:01 to 8 pm, and 8:01 pm to midnight. When an individual’s exact time of stroke onset could not be confirmed, it was estimated from available documentation and interview, and the midpoint of the approximation period was determined. In some cases there was no information describing the onset, and the individual was unable to be queried. Place or location where the stroke occurred was defined as home or permanent residence, outside the home, or acute hospital. Day of the week was abstracted from the date. Because we were interested in behavior affecting day of the week, we examined the association between stroke onset and alcohol consumption (yes or no), cigarette smoking (yes or no), systolic and diastolic blood pressure, and work status (working or retired/at home). These personal behaviors/risk factors were abstracted from the individual’s last biennial exam preceding the stroke event.
Statistical analysis tests for random (uniform) distribution of strokes by month, season, day of the week, time of the day, and place of stroke onset were performed separately by a one-way goodness-of-fit χ2 test21 as a means of identifying significant deviations from expected frequencies. We used χ2 tests and the t test to investigate the relationship between Monday versus non-Monday occurrence of stroke and categorical and continuous personal behaviors/risk factors, respectively. All analyses were performed on the total stroke sample and also for individual stroke types. Furthermore, analyses were performed for sexes combined and separate. Finally, to minimize the possibility of bias in determining the exact time of onset, we performed two analyses: first, with the total population, and second, eliminating those cases in which it was determined that the stroke occurred during sleep or that the symptoms were present on awakening.
During the 40-year period (1950 to 1990) of follow-up in the Framingham Study cohort, 637 initial incident strokes occurred, of which 635 had exact dates; in 2 cases we were able to determine the month but not the exact date of the stroke. In the sample of 635, 45% were men (286 of 635), and mean ages were 69.7 years for men and 73.4 years for women. At the last biennial examination before the development of the eventual stroke, 56% of the cases consumed alcohol, 37% smoked cigarettes, and 35% were employed. For stroke type classification 61% were ABIs (390 of 635), 25% were CEs (156 of 635), 7% were ICHs (43 of 635), and 7% were SHs (46 of 635). Eighty-seven percent of the total population were hospitalized subsequent to the stroke event. All 635 subjects had a documented month, season, and day of the week of the stroke. For 63% of the total stroke cases (401 of 635) we were able to determine either the exact or approximate time of onset; of the 401 cases with a time of onset, 52% (210 of 401) were exact times.
Month, Season, and Day of the Week
We examined month and season separately. For month, August (n=71, P=.01) and January (n=67, P=.04) were the two most frequent months when strokes occurred and July (n=48) the least frequent. For seasons, the overall numbers were fairly evenly distributed, with no discernible peak; this changed when we examined the seasons by subtype and gender. Winter emerged as a significant peak season for the CE cases (53 of 156, P=.01) as seen in Table 1⇓. This continued to be a significant finding in women (35%, P=.02) but not for men (32%).
In our population the most frequent day for stroke events was Monday (109 of 635, or 17.2%). When compared with non-Monday cases, strokes on Mondays were significantly more frequent (P=.038). Categorizing by stroke subtype showed an even more definite pattern. For the ICH cases, we documented that a third of them occurred on Mondays (14 of 43, or 32%; P=.001; Table 2⇓). This was true in both men (33%, P=.008) and women (32%, P=.03). For the SH cases when examined separately by gender, combining Sunday and Monday together this sequence was significant for men only (P=.02); for women Friday and Saturday was the most frequent SH sequence. Although not significant, ABI cases were most frequent on Monday, and CE cases occurred more often midweek.
Because of the emerging pattern of Monday strokes by gender and hemorrhagic subtype, we examined several personal behaviors/risk factors for an association. In general, people who had Monday stroke cases were more likely to smoke, drink alcohol, and work and had slightly higher blood pressure. When risk factors were analyzed for an association with Monday strokes, the following patterns emerged by hemorrhagic subtype and gender (Table 3⇓). First, in the Monday SH cases both men and women were significantly more likely to smoke versus the non-Monday cases. Second, men who worked were significantly more likely to have a Monday stroke versus a non-Monday stroke. Sixty-eight percent of the men who had strokes on Monday were working at the time, compared with 49% of the men working with non-Monday strokes or 52% of the total group.
Time of Stroke Onset
Of the stroke cases in the study the time of stroke onset was ascertained in 63% (401 of 635) of the cases. For this group, the most frequent time period for the onset of a stroke was the later morning (Fig 1⇓). Thirty-five percent (139 of 401) of all the stroke events occurred between 8:01 am and noon (P<.001). This pattern continued when examined by stroke subtype, with ICH cases having the same frequency from 8 am to noon and 12:01 to 4 pm. The least frequent time period for all stroke cases in our population was late evening, from 8 pm to midnight, when only 5% (21 of 401) of the strokes occurred.
Since we were unable to determine the precise time of stroke onset in those individuals whose strokes occurred during sleep or were present on awakening, we performed a second analysis excluding those individuals. In this group of 323 cases, the most frequent time for strokes continued to be between 8:01 and noon (P<.001, Fig 2⇓).
Location or Place of Stroke Onset
Most stroke events in our population occurred at home. By subtype, 79% of the ABI cases, 75% of the CE cases, 71% of the SH cases, and 69% of the ICH cases occurred in the home. There was a significant difference (P<.001) in place of onset by the specific stroke subtype. In our population 21% of CEs developed in the hospital, compared with less than 10% in the other subtypes; for ICH and SH strokes more than 25% occurred outside the home (Table 4⇓).
The results of this study add support to the belief that several patterns are associated with stroke onset. Our data, derived from a general population sample, identified peak occurrences for stroke onset by month, season, day of the week, time of day, and place of occurrence in both men and women and by stroke subtype.
A number of epidemiological and clinical studies over the past decade have examined seasonal patterns of stroke occurrence.1 10 12 13 16 22 23 24 25 Most of these studies have reported that stroke incidence peaks in winter and have postulated that a decrease in temperature causes an increase in blood pressure. In the present study, when we examined the total population we found that the most frequent months for stroke were August and January, with the least frequent being July. Examining all the cases together, we were unable to identify weather-related associations. However, when analyzed by stroke subtype, in the infarction cases we found that winter, New England’s coldest season, was a peak time for the CE subtype in both men and women. In determining seasonal patterns in cerebral infarction, both Ricci et al1 and Shinkawa et al10 found peak incidence in the winter. Sobel et al12 reported that infarctions peaked in February through April. By classifying the infarctions as CE and ABI subtypes we found a more specific pattern than that reported in the previous studies.
Data on hemorrhagic strokes and seasonality show more variation. The Stroke Data Bank project9 found no seasonal periodicity in the occurrence of ICH or SH. Capon et al23 reported the highest incidence of ICH in November-December and lowest in July-August, and Ricci et al1 and Biller et al13 also found an increase in ICH incidence in winter months. For SH, Chayette et al16 found seasonal and gender fluctuations, men peaking in the late fall and women in the spring. In our study, ICH cases were more frequent in summer and SH cases more frequent in spring and fall. However, none of our patterns were significant. This could be because of the small number of cases of hemorrhagic stroke in our population. In addition, we did not examine seasonal patterns by decades because of the smaller numbers of stroke in the first decade of the study.
An important finding of this study was the significant association between the day of the week and the onset of stroke. The greatest number of strokes for the total population occurred on Mondays. For the hemorrhagic cases a third of the men and women with ICH had stroke events on Mondays, whereas for men with SH the events were clustered on Sunday and Monday. When we examined the association between Monday stroke occurrences and work status, alcohol use, cigarette smoking, and hypertension, we found several relationships. In comparing Monday versus non-Monday stroke cases by work status, we found that men who had Monday strokes were twice as likely to work than not. This pattern did not hold true for women; however, only 20% of the women worked at the time of the stroke.
Recent studies18 26 27 have implicated heavy alcohol consumption and long-term cigarette smoking as independent risk factors for aneurysmal SH. Both Hillbom and Kaste18 and Juvela et al26 identified a weekend increase in SH related to alcohol intoxication. These studies quantified the amount of alcohol ingested 24 hours before the SH and hypothesized that there was an association between weekend behavior and hemorrhagic stroke. In our study in the determination of risk factors we did not include quantification of the amount of alcohol or cigarette smoking just before the stroke event and did not collect information on “binge” weekend drinking. Rather, we used information reported on the biennial exam conducted most recently before the event and did not stratify by amount. The results of this analysis found a Monday increase in ICH cases rather than on a weekend; for SH we found a Sunday and Monday increase, but for men only, and that cigarette smoking was significantly associated with Monday onset of SH. No significant difference was found between mean systolic or diastolic blood pressures in those who had a Monday stroke and those who had strokes on other days of the week. Our analysis points to a need for further exploration in deciphering these behavior/risk factor differences, both in daily life habits and the time immediately preceding the acute onset.
There are only a few reports in the literature on day of the week occurrence and stroke. Brackenridge17 found no day of the week variation in a community-based population in Australia; however, for those whose strokes occurred in the hospital, Wednesday had the highest frequency (35%) and weekends the lowest (14%). Other references for day of the week events relate to stroke mortality and cardiovascular disease. Macfarlane and White28 in a study of weekly cycles for death found that there were more stroke deaths on Monday than any other day. For cardiovascular events Willich et al29 documented an increased risk of myocardial infarction on Monday in working people, whereas nonworking individuals had a more even weekly distribution. Willich et al have hypothesized that external triggering factors may play a role in the acute causation of a cardiovascular event. This in turn could apply to certain subtypes of strokes, particularly in men. Subtle physiological changes in activities from the weekend to the weekday, particularly in workers, deserve more exploration.
In our investigation of circadian time patterns and stroke, we found that stroke events did not happen randomly throughout the day but rather were concentrated in the morning hours between 8:01 am and noon. This was true for all stroke subtypes, with ICH cases continuing to peak until 4 pm. This adds to the epidemiological data that support the notion that stroke occurs in the morning,4 6 8 9 peaking in late morning and being least frequent at night. Because the different times of day may reflect the different pathophysiological mechanisms of stroke, subanalyses seem important for establishing patterns. Trends of morning onset have been documented in both infarction6 and hemorrhagic stroke cases.9 Ricci et al1 found the highest peak between 6 and 9 am in their community-based study. Reports from the Stroke Data Bank showed the peak occurrence of subarachnoid and intracerebral hemorrhage in mid to late morning. Stroke incidence did not increase after meals; anticoagulants and antiplatelet agents before an ischemic stroke did not affect late morning occurrence.6
The discovery of a circadian variation in the onset of myocardial infarction has led to the same inquiry for stroke. However, for stroke the circadian rhythm is much more difficult to explain than for myocardial infarction. The circadian patterns observed in this and others studies may help to identify the underlying physiological or pathophysiological processes that could be modified to prevent stroke. The physiological processes that may trigger the acute stroke include changes in blood pressure, heart rate, platelet aggregability, and fibrinogen level. Ideally, it would be an advantage to examine these factors before the stroke using longitudinal data and to document the changes that occur at the time of the acute stroke process.
One interesting area for future investigation that has thus far not been explored is the relationship of variations in physical activity and daily routines to day of the week and time of stroke occurrence. Muller et al30 in cardiovascular research have theorized that the rest/activity cycle as it relates to circadian rhythms is an important consideration, particularly in elders whose schedule is not controlled by external obligations. Even though there is a morning peak incidence of stroke onset, it could be triggered by activity. Our data presented here on the place where the stroke event occurred add to the emerging descriptions of external influences on the acute event. For example, the vast majority of stroke events occurred in the home. When we examined place by stroke subtype, we found even more detail regarding possible triggers. More than 25% of both the ICH and SH strokes occurred outside the home, in public places. For CE cases more than 20% happened in the hospital, possibly relating to other health events. Analysis of work stress, hemodynamic variations, neuroendocrine responses, fibrinolytic activities, and coinciding illnesses in men and women and by subtype should provide further information about the patterns of stroke onset.
The data from this community-based study add evidence to the emerging concept that there are identifiable internal and external triggers that may increase the possibility of an acute stroke and explain some of the temporal patterns now being reported. These findings add to the reports suggesting that there are periods of increased risk of stroke that may be amenable to preventive interventions.
Selected Abbreviations and Acronyms
|ABI||=||atherothrombotic brain infarction|
This work was supported by grant 2-RO1 NS-17950-13 (National Institute of Neurological Disorders and Stroke) and contract NIH-NO1-HC-38038 (National Heart, Lung, and Blood Institute).
Reprint requests to Margaret Kelly-Hayes, EdD, RN, Boston University School of Medicine, B-603C Neurology, 80 E Concord St, Boston, MA 02118.
- Received March 24, 1995.
- Revision received May 9, 1995.
- Accepted May 9, 1995.
- Copyright © 1995 by American Heart Association
Wroe SJ, Sandercock P, Bamford J, Dennis M, Slattery J, Warlow C. Diurnal variation in incidence of stroke: Oxfordshire community stroke project. BMJ. 1992;304:155-157.
Marshall J. Diurnal variation in occurrence of strokes. Stroke. 1977;8:230-231.
van der Windt C, van Gijn J. Cerebral infarction does not occur typically at night. J Neurol Neurosurg Psychiatry. 1988;51:109-111.
Marler JR, Price TR, Clark GL, Muller JE, Robertson T, Mohr JP, Hier DB, Wolf PA, Caplan LR, Foulkes MA. Morning increase in onset of ischemic stroke. Stroke. 1989;20:473-476.
Argentino C, Toni D, Rasura M, Violi F, Sacchetti ML, Allegretta A, Balsano F, Fieschi C. Circadian variation in the frequency of ischemic stroke. Stroke. 1990;21:387-389.
Sloan MA, Price TR, Foulkes MA, Marler JR, Mohr JP, Hier DB, Wolf PA, Caplan LR. Circadian rhythmicity of stroke onset. Stroke. 1992;23:1420-1426.
Shinkawa A, Ueda K, Hasuo Y, Kiyohara Y, Fujishima M. Seasonal variation in stroke incidence in Hisayama, Japan. Stroke. 1990;21:1262-1267.
Sobel E, Zhang Z, Alter M, Lai S, Davanipour Z, Friday G, McCoy R, Isack T, Levitt L. Stroke in the Lehigh Valley: seasonal variation in incidence rates. Stroke. 1987;18:38-42.
Chayette D, Chen T, Bronstein K, Brass L. Seasonal fluctuation in the incidence of intracranial aneurysm rupture and its relationship to changing climactic conditions. Stroke. 1993;24:187. Abstract.
Brackenridge CJ. Daily variation and other factors affecting the occurrence of cerebrovascular accidents. J Gerontol. 1977;36:176-179.
Hillbom M, Kaste M. Alcohol intoxication: a risk factor for primary subarachnoid hemorrhage. Stroke. 1982;32:706-711.
Shurtleff D. Some characteristics related to the incidence of cardiovascular disease and death: Framingham Study. 18 year follow-up. In: Kannel WB, Gordon T, eds. Epidemiological Investigation of Cardiovascular Disease. Washington, DC: Government Printing Office; 1974. US Dept of Health, Education, and Welfare publication (NIH) 74-599.
Cupples A, D’Agostino RR. Some risk factors related to the incidence of cardiovascular disease and death using pooled repeated biennial measurements: the Framingham Study. 30 year follow-up. In: Kannel W, Wolf P, Garrison R, eds. The Framingham Study: An Epidemiological Investigation of Cardiovascular Disease. Bethesda, Md: National Heart, Lung, and Blood Institute; 1987. National Institutes of Health publication 87-2703.
Snedecor G, Cochran W. Statistical Methods. 7th ed. Ames, Iowa: Iowa State University Press, 1980.
Barer D, Ebrahim S, Smith C. Factors affecting day to day incidence of stroke in Nottingham. Br Med J. 1984;289:662.
Capon A, Demeurisse G, Zheng L. Seasonal variation of cerebral hemorrhage in 236 consecutive cases in Brussels. Stroke. 1992;23:24-27.
Beringer VM, Goldsmith JR, Batz U, Vardi H, Shapiro Y. Clustering of strokes in association with meteorologic factors in the Negev Desert of Israel: 1981-1983. Stroke. 1989;20:65-69.
Juvela S, Hillbom M, Numminen H, Koskinen P. Cigarette smoking and alcohol consumption as risk factors for aneurysmal subarachnoid hemorrhage. Stroke. 1993;24:639-646.
Longstreth WT, Nelson LM, Koepsell TD, van Belle G. Cigarette smoking, alcohol use, and subarachnoid hemorrhage. Stroke. 1992;23:1242-1249.
Macfarlane A, White G. Death: the weekly cycle. Population Trends. 1977;7:7-8.
Willich SN, Lowel H, Lewis M, Hormann A, Arntz HR, Keil U. Weekly variation of acute myocardial infarction: increased Monday risk in the working population. Circulation. 1994;90:87-93.
Muller JE, Tofler GH, Stone PH. Circadian variation and triggers of onset of acute cardiovascular disease. Circulation. 1989;79:733-743.