Original Paper

Neuroepidemiology 2010;34:2533 DOI: 10.1159/000255463

Received: February 19, 2009 Accepted: August 5, 2009 Published online: November 5, 2009

Diurnal Variation in Onset of Hemorrhagic Stroke Is Independent of Risk Factor Status: Takashima Stroke Registry
Tanvir Chowdhury Turin a Yoshikuni Kita a Nahid Rumana a Naoyuki Takashima a Masaharu Ichikawa b Hideki Sugihara b Yutaka Morita c Kunihiko Hirose d Yoshitaka Murakami a Katsuyuki Miura a Akira Okayama e Yasuyuki Nakamura f Robert D. Abbott a, g Hirotsugu Ueshima a
Department of Health Science, Shiga University of Medical Science; b Takashima General Hospital; c Makino Hospital, Takashima, and d Otsu Red Cross Hospital, Otsu, e The First Institute for Health Promotion and Health Care, Tokyo, and f Kyoto Womens University, Kyoto, Japan; g Division of Biostatistics and Epidemiology, University of Virginia, School of Medicine, Charlottesville, Va., USA
a

Key Words Hemorrhagic stroke Epidemiology Time of day Risk factors Registry Japan

Abstract Background: We examined the circadian periodicity of hemorrhagic stroke onset to identify any existing specific pattern and its relationship with conventional stroke risk factors using 14-year stroke registration data. Methods: Data were obtained from the Takashima Stroke Registry, which covers a stable population of approx. 55,000 in Takashima County in central Japan. Out of 499 registered first-ever hemorrhagic stroke events during 19902003, there were 429 (186 men, 243 women) events with classifiable onset time. Hemorrhagic stroke incidence was categorized as occurring at night (midnight to 6 a.m.), morning (6 a.m. to noon), afternoon (noon to 6 p.m.) or evening (6 p.m. to midnight). The OR (with 95% CI) of having a stroke in the morning, afternoon or evening were calculated, with night serving as reference. Results: There was significant diurnal variation in hemorrhagic stroke incidence (p ! 0.001). The proportion of hemorrhagic strokes was highest in the morning (36.1%, 95% CI: 31.740.8)

and lowest in the night (11.9%, 95% CI: 9.115.3). An excess stroke incidence in the morning was observed in both genders, in subjects !65 years and 665 years, and in both intracerebral hemorrhage and subarachnoid hemorrhage. A second surge was also observed during the later part of the day. The higher daytime risk persisted after adjusting for age, gender, and risk factors. Conclusion: In the examination of circadian variation of hemorrhagic stroke onset, a 2-peak temporal distribution was observed, which was independent of conventional risk factors.
Copyright 2009 S. Karger AG, Basel

Introduction

Several studies have reported diurnal variation in the onset of hemorrhagic stroke (HS) events [110]. Although it is well established that HS incidence is affected by risk factors such as hypertension, diabetes mellitus, cigarette smoking and alcohol intake [11], the effects of these risk factors on diurnal or circadian variation in HS incidence have not been clearly addressed. Information about whether the chronobiological patterns in different cliniTanvir Chowdhury Turin Department of Health Science Shiga University of Medical Science Seta Tsukinowa-cho, Otsu City, Shiga, 520-2192 (Japan) Tel. +81 77 548 2191, Fax +81 77 543 9732, E-Mail turin @ belle.shiga-med.ac.jp

2009 S. Karger AG, Basel 02515350/10/03410025$26.00/0 Fax +41 61 306 12 34 E-Mail karger@karger.ch www.karger.com Accessible online at: www.karger.com/ned

cal subgroups of patients with HS differ might contribute to primary prevention. The primary immediate value of recognizing the circadian variation of acute onset of stroke in relation to the risk factors is the emphasis that might be placed on the pharmacologic protection or lifestyle modification against the probable associated factors that promote stroke during high-risk times of the day. Addressing any periodicity in the time of HS onset might be used as a surrogate to predict HS onset or high-risk period. This also has practical implications for management of stroke patients. Hospitals that propose to treat stroke patients with acute interventions will require heightened levels of awareness during the high-risk period. It is obvious, however, that 24-hour-a-day availability and monitoring will be required by acute stroke teams. HS incidence and case fatality have also been shown to have circannual and circaseptan variations [1215], but little is known regarding the effect of seasonality and day of the week on the circadian variation of HS onset. In the present study, we examined the circadian variation in HS occurrence to explore whether there was a period during 24 h with excess risk. We also examined the modification of these variations by history of traditional risk factors, as well as other temporal variability. We considered the role of hypertension, diabetes mellitus, cigarette smoking and alcohol intake in contributing to any excesses in HS incidence during any time of the day where risk might be highest.

Event Identification and Registration Method All suspected stroke cases in the population during the study period were identified and evaluated according to the criteria established by the Monitoring System for Cardiovascular Disease commissioned by the Ministry of Health and Welfare, Japan [19]. These criteria are in accordance with the WHO-MONICA Project [20]. The stroke registration was carried out and evaluated on the basis of the medical records from all the hospitals inside the county and the county ambulance records. Also registered were stroke patients who were residents of Takashima County but had visited or were referred to the 3 tertiary hospitals outside the county. For out-of-hospital stroke deaths, diagnoses were based on the patients location and symptoms at onset and their history of stroke. This information was collected from emergency room records and ambulance records, and supplemented with death certificate data. Original death certificates were seen at the county health center with the permission of the Ministry of Public Management, Home Affairs, Post and Telecommunications, Japan, in order to establish the cause of death. Research physicians and epidemiologists cross-checked the records for absolute verification regarding eligibility for inclusion in the registry as a stroke event. In cases of disagreement between the physician and epidemiologist, based largely on the mutual decision, events are classified as unclassified stroke. In 22 instances when there was disagreement in a diagnosis, a decision was made to categorize the event as an unclassified stroke. This study was approved by the Institutional Review Boards of the Shiga University of Medical Science. Diagnostic Criteria and Items Registered Diagnosis of stroke type was based on clinical symptoms as well as neurological imaging by CT or MRI. An intracerebral hemorrhage (ICH) was defined on the basis of a region of highdensity absorption in the causal region due to a hematoma as shown on a CT scan. A subarachnoid hemorrhage (SAH) was defined on the basis of a region of high-density absorption in the cerebrospinal fluid due to a hemorrhage as shown on a CT scan. Neuroimaging was performed on 94.2% of the HS patients. The severity of neurological deficit on admission was determined according to the Japan Coma Scale to categorize stroke events into minor, moderate and severe categories [13]. Items recorded at registration of an event were the date and time of onset, the situation and symptoms at the event, the extent of neurological symptoms and clinical observations at the event, past history, family history, treatment given, rehabilitation, fatality, cause of death, recurrence in acute stage, neurological imaging observations, etc. Details of the case finding, registration process, diagnostic criteria and items registered have been described elsewhere [1215]. Analysis Parameters and Statistical Method Our analysis included all patients from the Takashima Stroke Registry who suffered their first-ever HS. The present study covered the time period from the 1 January 1990 to 31 December 2003. Assuming that HS onset is not related to the time of day, then the time of onset would be evenly distributed throughout the day. To investigate the daily distribution of HS, the frequency of observed cases was determined in each 1-hour interval for 24 h, and was compared with the expected frequency using a 2 test for a single population (for goodness of fit to the null model of equal

Population and Methods

Takashima Stroke Registry The Takashima Stroke Registry covers an entire community of approximately 55,000 individuals. This stroke registry is an integrated part of the Takashima Cardiocerebrovascular Disease Registration System [16, 17] established in 1988 in Takashima County, Shiga, Japan. The objective of this disease registration system for stroke is to measure trends in the incidence and case fatality of stroke, and to compare these to sources inside and outside of Japan. Registration Catchments Population Takashima County is located in the rural part of Shiga prefecture in the central part of Japan. The population of Takashima County remained stable over the 14-year study period. Takashima County is a community with inhabitants mainly classified culturally into a single subgroup and with similar standards of living. With an aging populace, the population included 55,451 inhabitants (men 49.2% and women 50.7%) in the year 2000 [18]. Among the population, 22.3% were 665 years, which is higher than the all Japan proportion of 17.4%.

26

Neuroepidemiology 2010;34:2533

Turin et al.

Table 1. Population and registered stroke patients of Takashima

Stroke Registry, Japan (19902003) Characteristics Overall Men Women <65 years 65 years Takashima County Population1 55,451 27,323 (49.2) 28,128 (50.7) 43,081 (77.7) 12,354 (22.3) Registered stroke cases
Classifiable onset time Unclassifiable p onset time

All stroke Total incidence Average age at onset, years Gender Men Women Stroke subtype Intracerebral hemorrhage Subarachnoid hemorrhage Severity Minor Moderate Severe Fatal Stroke 7-day fatal cases 28-day fatal cases Risk factor history Hypertension Diabetes mellitus Drinking Smoking

429 69.0813.6 186 (82.7) 243 (88.7) 298 (85.9) 131 (86.2) 134 (85.4) 87 (86.1) 190 (89.6) 104 (24.2) 139 (32.4) 239 (55.7) 78 (18.2) 91 (21.2) 117 (27.3)

70 65.6815.3 39 (17.3) 31 (11.3) 49 (14.1) 21 (13.8) 23 (14.7) 14 (13.9) 22 (10.4) 14 (20.0) 17 (24.3) 31 (44.3) 8 (11.4) 8 (11.4) 11 (15.7)

0.193

0.054

0.928

0.427

0.439 0.174 0.075 0.165 0.057 0.040

onsets within those periods for wake-up stroke and stoke of unknown onset status were distributed evenly among the preceding hours within the nocturnal and nap sleep periods. Fourth, assuming that symptoms present on awakening or unknown status at onset indicated a stroke occurring at times distributed evenly over the preceding 8 h, the onset times of these strokes were distributed evenly across the 8 h preceding the reported onset time. The results of all the 4 approaches described above showed a similar pattern and significant distribution of stroke onset across the hours of the day. Based on this observation, the reported stroke onset time was used as the surrogate of the onset time for the patients with unknown onset status or wake-up stroke in further analyses. HS incidence was categorized as having occurred at night (midnight to 6 a.m.), morning (6 a.m. to noon), afternoon (noon to 6 p.m.) or evening (6 p.m. to midnight) based on its time of onset. The proportions of HS were measured across each 6-hour time block by gender, age group, subtype and other subgroups. To examine the effect of age on the patterns of HS incidence, patients were stratified into !65 and 665 years. Confidence intervals (CI) for proportions were calculated by Wilsons method [21]. Using multinomial logistic regression, HS cases across time blocks were modeled as a dependent variable. From such models, 3 regression coefficients were estimated which provide estimates of the odds ratio of HS occurring in the morning, afternoon and evening versus the night as a reference. The odds ratio was calculated as e with 95% CI = e( 8 1.96 ! SE). Here, is the regression coefficient corresponding to either morning, afternoon, or evening (with night as a reference), and SE is the standard error of . Risk factor status (presence or absence of hypertension, diabetes mellitus, and drinking and smoking status) were modeled as independent variables to calculate the respective odds ratio. In further analyses, age, gender and risk factors were collectively modeled as independent variables. All statistical analyses were performed using SAS version 9.1 (SAS Institute, Cary, N.C., USA).

Results

Figures in parentheses are percentages, with average age at onset reported as means 8 SD. Comparisons between the groups undertaken using 2-tailed Students t test or 2 test. 1 Based on the 2000 population census of Japan (age reported unknown for 16 persons).

distribution of strokes). For wake-up stroke (symptoms first recognized at awakening) and stroke of unknown onset status (no information on status at onset), the time of awakening/symptom recognition was used as the time of onset. We performed a sensitivity analysis assuming different scenarios regarding the onset time of these strokes. First, the distinction was ignored and the data were analyzed without regard to onset status. Second, the data were analyzed omitting wake-up stroke and unknown onset status stroke. Third, assuming a nocturnal sleep period from 11 p.m. to 7 a.m. and a noon nap period from 1 p.m. to 3 p.m., the

There were 499 first-ever HS patients during the study period of 19902003. Among them, there were 347 ICH (women 52.6%) and 152 SAH (women 60.5%). The average age of HS onset was 65.3 years (SD: 13.8) for men and 71.1 years (SD: 13.5) for women. For 70 patients, the time of HS onset could not be determined. Table 1 shows the demographic characteristics for the registered HS cases, and summarizes the characteristics of patients with classifiable and unclassifiable onset time. Apart for the finding that smoking was less common in patients with unclassifiable onset time, there was no significant difference between the patients with or without classifiable onset time. Among the 429 patients (186 men and 243 women) with classifiable onset time, HS occurred when awake in 346 patients and stroke symptoms were first noted on awakening in 83 patients. Figure 1 shows the frequency
Neuroepidemiology 2010;34:2533

Circadian Periodicity of Stroke

27

All HS (p < 0.001)

40 35 30 25 20 15 10 5 0 1 2 3 4 5 6 7 8 Time of stroke onset

Unknown Wake-up stroke Awake

Events (n)

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

HS occurrence while awake (p < 0.001)

35 30 25 20 15 10 5 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 Time of stroke onset

Redistributing cases with symptoms during sleep and unknown onset across preceding hours of the assumptive night sleep and siesta sleep period (p < 0.001)

Events (n)

Fig. 1. Bar graphs showing a sensitivity

analysis for the incidence of HS onset across the time of the day. Four different methods were explored for incorporating wake-up stroke (symptoms first recognized at awakening) and stroke of unknown onset status (no information on status at onset) into this analysis of data from the Takashima Stroke Registry, Shiga, Japan (19902003). a Distribution of all stroke onsets. b Distribution for only patients with stroke onset while awake. c Results after redistributing cases of wake-up stroke and unknown onset status evenly among the preceding hours within the period of nocturnal sleep (11 p.m. to 7 a.m.) and nap sleep (1 p.m. to 3 p.m.). d Results after redistributing cases with wake-up stroke and unknown onset status across the 8 h preceding the reported onset time.

40 35 30 25 20 15 10 5 0 1 2 3 4 5 6 7 8 Time of stroke onset

Sleep and unknown Awake

Events (n)

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Redistributing cases with symptoms on awakening across preceding 8 h of the wake-up time (p < 0.0001)

40 35 30 25 20 15 10 5 0 1 2 3 4 5 6 7 8 Time of stroke onset

Sleep and unknown Awake

Events (n)

9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

28

Neuroepidemiology 2010;34:2533

Turin et al.

Table 2. Proportion of stroke across time blocks of the day by gender, age group, subtype, severity, fatality, season and day of the week

in the Takashima Stroke Registry, Japan (19902003) Total Night (midnight to 6 a.m.) % All cases Gender Men Women Age group <65 years 65 years Stroke subtype Cerebral hemorrhage SAH Stroke severity Minor Moderate Severe Fatal stroke 7-day fatal cases 28-day fatal cases Season of the year Winter stroke (Dec.Feb.) Spring stroke (MarchMay) Summer stroke (JuneAug.) Autumn stroke (Sep.Nov.) Day of the week Weekend stroke Weekday stroke 429 186 243 150 279 298 131 134 87 190 104 139 104 125 92 108 124 305 11.9 16.7 8.2 14.0 10.8 11.7 12.2 11.2 12.6 12.6 13.5 12.2 7.7 14.4 12.0 13.0 11.3 12.1 95% CI 9.115.3 11.922.8 5.312.5 9.220.6 7.615.0 8.515.9 7.519.1 6.717.8 6.921.5 8.518.2 8.021.6 7.618.9 3.714.7 9.121.8 6.520.4 7.720.8 6.718.3 8.916.3 Morning (6 a.m. to noon) % 36.1 39.8 33.3 35.3 36.6 34.9 38.9 39.6 31.0 36.8 42.3 44.6 38.5 32.8 35.9 38.0 33.9 37.0 95% CI 31.740.7 33.047.0 27.739.5 28.043.4 31.142.4 29.740.5 30.947.6 31.648.1 22.141.6 30.244.0 33.152.1 36.553.0 29.548.2 25.141.6 26.646.3 29.247.5 26.042.7 31.842.6 Afternoon (noon to 6 p.m.) % 31.0 30.6 31.3 29.3 31.9 32.2 28.2 32.1 35.6 26.3 25.0 24.5 30.8 29.6 34.8 29.6 29.8 31.5 95% CI 26.835.6 24.437.7 25.737.4 22.537.2 26.737.6 27.137.8 21.136.6 24.740.5 26.246.3 20.533.1 17.534.3 17.932.4 22.540.4 22.238.2 25.645.1 21.739.0 22.438.5 26.536.9 Evening (6 p.m. to midnight) % 21.0 12.9 27.2 21.3 20.8 21.1 20.6 17.2 20.7 24.2 19.2 18.7 23.1 23.2 17.4 19.4 25.0 19.3 95% CI 17.425.1 8.718.6 21.933.1 15.428.7 16.426.0 16.826.2 14.428.5 11.624.6 13.330.6 18.630.9 12.628.0 13.026.1 15.832.2 16.531.5 10.826.6 12.928.1 18.133.4 15.324.2

of HS onset in 24 1-hour intervals. There was a clear and significant diurnal variation in HS incidence (p ! 0.001). When the cases with symptoms on awakening were redistributed across the preceding hours within the assumptive nocturnal and nap sleep periods, the variation was still significant (p ! 0.001). Even when the strokes with onset during sleep were redistributed throughout the preceding 8 h, the significant variation persisted (p ! 0.001). A second surge was observed during the later part of the day. Table 2 shows the incidence of HS and its subgroups across 6-hour time blocks. The proportion of HS was highest in the morning (36.1%, 95% CI: 31.740.8) followed by afternoon (31.0%, 95% CI: 26.835.6), and lowest in the night (13.2, 95% CI: 11.315.4). An excess HS
Circadian Periodicity of Stroke

incidence in the morning followed by afternoon was observed in all subgroups. The HS onset showed a similar trend across seasons (spring, winter, summer and autumn) and days of the week (weekend and weekday). For all HS, morning excess was higher (odds ratio: 3.04, 95% CI: 2.224.17) compared to the night period. Figure 2 presents the odds ratio of HS stratified by the presence or absence of a history of risk factors: hypertension, diabetes mellitus, smoking and drinking for morning, afternoon, and evening with night as a reference. In all instances, the highest risk of HS occurred in the morning and the lowest risk was in the night relative to the other time blocks. Similar trends persisted after adjusting for age, gender and risk factors (fig. 3).

Neuroepidemiology 2010;34:2533

29

Hypertension
7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0

No hypertension

Odds ratio

Night

Morning

Afternoon

Evening

Night

Morning

Afternoon

Evening

Diabetes
14 12 10 8 6 4 2 0 7 6 5 4 3 2 1 0

No diabetes

Odds ratio

Night

Morning

Afternoon

Evening

Night

Morning

Afternoon

Evening

Smoking
7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0 Odds ratio

No smoking

Night

Morning

Afternoon

Evening

Night

Morning

Afternoon

Evening

Fig. 2. The odds ratios and 95% CI are shown for the occurrence of stroke stratified by the presence and absence of a history of hypertension, diabetes mellitus, smoking, and drinking in morning (6 a.m. to noon), afternoon (noon to 6 p.m.), and evening (6 p.m. to midnight) compared with night (midnight to 6 a.m.). Takashima Stroke Registry, Shiga, Japan (1990 2003).

Drinking
8 7 6 5 4 3 2 1 0 7 6 5 4 3 2 1 0
Night Morning Afternoon Evening

No drinking

Odds ratio

Night

Morning

Afternoon

Evening

Discussion

The present study shows a diurnal variation in the incidence of HS events in both men and women, in younger and older age groups, for both ICH and SAH, and for all severity levels. This periodicity is also reflected in HS fatality. We observed that a higher proportion of fatal HS occurred during the day time with a morning surge followed by a second surge during the later part of the day. Similar patterns were observed across seasons of the year and day of the week. When HS were clustered within 6hour time intervals, the crude odds ratios for the morning, afternoon and evening mimic the 2 peaks that were
30
Neuroepidemiology 2010;34:2533

observed for the 24 1-hour distributions. After adjustment for age, gender, and other risk factors, the bimodal distribution becomes less apparent. In all instances, however, the risk was significantly lower during the night. Characterizations of the time of stroke onset have been plagued by the problem of determining time of onset of events detected when the patient awakens. The proportion of time patients spend asleep and the amount of sleep during the daytime are difficult to measure. However, even if the presumed hour of onset was distributed over the preceding 8 h to minimize diurnal variation, the variation was still significant. We also hypothesized a
Turin et al.

OR Crude Night Morning Afternoon Evening 1.00 3.04 2.61 1.76

95% CI OR (log scale) 2.224.17 1.893.60 1.252.49

OR with 95% CI

Adjusted for age and gender Night 1.00 Morning 3.50 Afternoon 3.26 Evening 3.39 Further adjusted for risk factors Night 1.00 Morning 2.76 Afternoon 2.88 Evening 2.71

1.776.91 1.636.51 1.666.95

1.266.05 1.306.37 1.186.21

10

Fig. 3. The odds ratios and 95% CIs are shown for the occurrence of stroke; Unadjusted, adjusted for age-gender and further adjusted for the risk factors hypertension, diabetes mellitus, dyslipidemia, smoking, drinking, transient ischemic attack, and heart disease in morning (6 a.m. to noon), afternoon (noon to 6 p.m.), and evening (6 p.m. to midnight) compared with night (midnight to 6 a.m.). Takashima Stroke Registry, Shiga, Japan (1990 2003).

sleeping period and distributed the presumed hour of onset over the probable hours slept within that period. Significant diurnal variation also persisted for this approach. The present study shows a diurnal variation in the incidence of HS events in both men and women, in younger and older age groups, for both ICH and SAH and all severity levels. This periodicity also reflected in the fatality of HS. We observed that a higher proportion of fatal HS occurred during the morning period. Similar patterns were observed across the season of the year and day of the week. The diurnal pattern of HS onset seems to hold regardless of the presence or absence of conventional risk factor histories. Our findings are similar to other studies around the world where stroke incidence was found to be higher during the early morning period [110]. We have also observed a second peak for the HS during the later part of the day. Similar to our findings, some other studies also have reported a relatively smaller second peak for stroke onset [1, 3, 4, 7]. In our study, the periodic surge in HS incidence appeared to be independent of history of hypertension, diabetes mellitus, drinking and smoking. Similarly, other studies showed no difference in the pattern of onset between patients with or without hypertenCircadian Periodicity of Stroke

sion [2, 8, 9]. The fact that none of the common HS risk factors seem to explain the circadian pattern suggests that the diurnal variability of HS onset may be triggered by some unknown exogenous and endogenous factors intrinsic to those hours. Although the mechanisms underlying the diurnal variation in HS are not yet clearly understood, probable contributors include the observed day/night rhythmicity in biological factors such as blood pressure [3, 22]. The circadian variation in blood pressure seems to mirror the time of onset of HS [8]. The daily increases in the blood pressure occur in the morning and afternoon and lower levels occur at night [23]. Studies have also reported that fasting plasma glucose level and glycemic abnormalities were significant determinants of morning and waking blood pressure surge [24]. Smoking status and alcohol consumption are also reported to be associated with the magnitude of morning blood pressure [24, 25]. In our study, the circadian variation in HS in various patient subsets stratified according to risk factors was similar. Patients with and without hypertension had the same chronobiological pattern of stroke onset. This suggests that blood pressure, with its circadian variability, and not strictly hypertension, plays an important role in the cirNeuroepidemiology 2010;34:2533

31

cadian pattern of stroke onset. Similarly diabetes, smoking and drinking habits had no major effect on the observed chronobiological patterns of stroke onset. One exception occurred in figure 3, where risk factor adjustment modestly increased the odds of HS in the evening versus the night. Regardless, findings suggest that the morning surge in stroke onset relative to the night has little to do with endogenous circadian rhythms, but may be more of a response to physiologic stress, life-style activities, or unknown physical parameters that are unique to the morning. Physical activities have been shown to have a diurnal variation [3]. Morning and evening peaks in blood pressure, pulse rate and physical activity are reported to have occurred in parallel to the bimodal peaks of stroke onset [3]. Researchers have suggested the occurrence of ICH and SAH during working hours may reflect an activity precipitant for ICH or SAH [23, 26]. Moreover, acceleration of physical activity is regarded as the major determinant of the morning surge in blood pressure [27], which may increase the risk of stroke. Abrupt shifting from the absence of physical activity during sleep to sudden physical activity with a change in rest/activity cycle could serve as a trigger factor for HS. The physiological processes related to these physical activity components may trigger the acute event. In a recent study, it was observed that transition to or out from daylight saving time is coupled with an immediate shift of the time pattern of stroke onset [5]. This suggests an association between the exogenous factors associated with awakening and the diurnal pattern of stroke. The quality of our registration system was assured by its completeness. Our registry system was planned to capture all the cases in the study area by covering all hospitals in the county. It has been estimated that more than 98% of all hospital admissions of Takashima County are seen in these institutions [16, 17]. To ensure that eligible patients hospitalized outside the county were not excluded, registration procedures were also conducted at 3 high-level medical facilities outside the county. In Japan, almost 100% of the residents are covered by health insurance under the control of the Ministry of Health and Welfare [18]. In addition to that, 24hour emergency ambulance service is available for residents without any charge. The usual practice in Japan is to take patients with any acute disease condition to an emergency facility. Furthermore, individuals with mild stroke who visit general physicians in the community are almost always and immediately referred to hospitals for their treatment. Thus, we believe that extremely few patients would have been left out of our registration sys32
Neuroepidemiology 2010;34:2533

tem. Japan has the most MRI units (35 units per million population) and CT scanners (93 units per million population) per capita among developed countries [28]. In our registered stroke cases, the diagnoses were verified by neurological imaging in more than 93% of the cases [1214]. Therefore, in our registration system, identification of stroke cases among the study area was almost complete and stroke categorization was highly accurate. Similar to most studies, the main limitation in the current report is the missing information on the time of onset for a number of registered patients. We analyzed the HS events excluding the cases with missing information on the time of onset. Another limitation of this study is that we could not determine the disease duration for hypertension and diabetes mellitus or quantify the amount of alcohol or cigarette smoking, both in cases of habitual quantity or just before the HS onset. Rather, we used information reported about the presence or absence of a history of risk factors. Additionally, our registry covers a rural and semi-urban population in Japan that may be different from the metropolitan population. In conclusion, we found that there was a chronological pattern of increased HS incidence during the morning hours. This was observed in both young and old subjects, in both men and women, and for both ICH and SAH. This diurnal surge of HS incidence was independent of the presence or absence of conventional risk factors.

Funding
Supported in part by grants from The Research on Cardiovascular Disease (3A-1, 6A-5 and 7A-2) and The Comprehensive Research on Cardiovascular and Lifestyle Related Diseases (H18CVD-Ippan-029) of the Ministry of Health and Welfare, and from the Grants-in-Aid Scientific Research (C-213670361, B17390186 and B-20390184) of Ministry of Education, Culture, Sports, Science and Technology. T.C.T. is supported by the fellowship and Research Grants-In-Aid (P-20.08124) from the Japan Society for the Promotion of Science.

Turin et al.

References
1 Omama S, Yoshida Y, Ogawa A, Onoda T, Okayama A: Differences in circadian variation of cerebral infarction, intracerebral haemorrhage and subarachnoid haemorrhage by situation at onset. J Neurol Neurosurg Psychiatry 2006;77:13451349. 2 Tsementzis SA, Gill JS, Hitchcock ER, Gill SK, Beevers DG: Diurnal variation of and activity during the onset of stroke. Neurosurgery 1985;17:901904. 3 Stergiou GS, Vemmos KN, Pliarchopoulou KM, Synetos AG, Roussias LG, Mountokalakis TD: Parallel morning and evening surge in stroke onset, blood pressure, and physical activity. Stroke 2002; 33:14801486. 4 Spengos K, Vemmos KN, Tsivgoulis G, Synetos A, Zakopoulos NA, Zis VP, Vassilopoulos D: Two-peak temporal distribution of stroke onset in Greek patients: a hospitalbased study. Cerebrovasc Dis 2003; 15: 70 77. 5 Foerch C, Korf HW, Steinmetz H, Sitzer M, Arbeitsgruppe Schlaganfall Hessen: Abrupt shift of the pattern of diurnal variation in stroke onset with daylight saving time transitions. Circulation 2008;118:284290. 6 Ricci S, Celani MG, Vitali R, La Rosa F, Righetti E, Duca E: Diurnal and seasonal variations in the occurrence of stroke: a community-based study. Neuroepidemiology 1992;11:5964. 7 Wroe SJ, Sandercock P, Bamford J, Dennis M, Slattery J, Warlow C: Diurnal variation in incidence of stroke: Oxfordshire community stroke project. BMJ 1992; 304:155157. 8 Nyquist PA, Brown RD Jr, Wiebers DO, Crowson CS, OFallon WM: Circadian and seasonal occurrence of subarachnoid and intracerebral hemorrhage. Neurology 2001;56: 190193. 9 Sloan MA, Price TR, Foulkes MA, Marler JR, Mohr JP, Hier DB, Wolf PA, Caplan LR: Circadian rhythmicity of stroke onset: intracerebral and subarachnoid hemorrhage. Stroke 1992;23:14201426. 10 Kelly-Hayes M, Wolf PA, Kase CS, Brand FN, McGuirk JM, DAgostino RB: Temporal patterns of stroke onset: The Framingham Study. Stroke 1995;26:13431347. 11 Grysiewicz RA, Thomas K, Pandey DK: Epidemiology of ischemic and hemorrhagic stroke: incidence, prevalence, mortality, and risk factors. Neurol Clin 2008;26:871895. 12 Turin TC, Kita Y, Murakami Y, Rumana N, Sugihara H, Morita Y, Tomioka N, Okayama A, Nakamura Y, Abbott RD, Ueshima H: Higher stroke incidence in the spring season regardless of conventional risk factors: Takashima Stroke Registry, Japan, 19882001. Stroke 2008; 39:745752. 13 Turin TC, Kita Y, Rumana N, Murakami Y, Ichikawa M, Sugihara H, Morita Y, Tomioka N, Okayama A, Nakamura Y, Abbott RD, Ueshima H: Stroke case fatality shows seasonal variation regardless of risk factor status in a Japanese population: 15 year results from Takashima Stroke Registry. Neuroepidemiology 2009; 32:5360. 14 Turin TC, Kita Y, Murakami Y, Rumana N, Sugihara H, Morita Y, Hirose K, Okayama A, Nakamura Y, Ueshima H: Increase of stroke incidence after weekend regardless of traditional risk factors: Takashima Stroke Registry, Japan, 19882003. Cerebrovasc Dis 2007; 24:328337. 15 Turin TC, Kita Y, Rumana N, Ichikawa M, Sugihara H, Morita Y, Tomioka N, Okayama A, Nakamura Y, Ueshima H: Case fatality of stroke and day of the week: Is the weekend effect an artifact? Takashima Stroke Registry, Japan (19882003). Cerebrovasc Dis 2008;26:606611. 16 Kita Y, Turin TC, Rumana N, Sugihara H, Morita Y, Hirose K, Okayama A, Nakamura Y, Ueshima H: Surveillance and measuring trends of stroke in Japan: The Takashima Stroke Registry (1988 present). Int J Stroke 2007;2:129132. 17 Turin TC, Kita Y, Rumana N, Sugihara H, Morita Y, Tomioka N, Okayama A, Nakamura Y, Ueshima H: Registration and surveillance of acute myocardial infarction in Japan monitoring an entire community by the Takashima AMI Registry. Circ J 2007;71: 16171621. 18 Ministry of Public Management, Home Affairs, Posts and Telecommunications of Japan: 2000 Population Census of Japan (in Japanese). Tokyo, Ministry of Public Management, Home Affairs, Post and Telecommunications, 2001. 19 Ministry of Health and Welfare: Study Project of Monitoring System for Cardiovascular Disease Commissioned by the Ministry of Health and Welfare: Manual for the Registry and Follow-up of Stroke (in Japanese). Osaka, National Cardiovascular Center, 1998. 20 World Health Organization MONICA Project: Event Registration Data Component, MONICA Manual Version 1.1. Document for meeting of MONICA Principal Investigators. 1986;S-4:911. 21 Rothman KJ: Epidemiology: An Introduction. New York, Oxford University Press, 2002, pp 137139. 22 Elliot WJ: Circadian variation in blood pressure: implication for the elderly patient. Am J Hypertens 1999;12:43S49S. 23 Kleinpeter G, Schatzer R, Bock F: Is blood pressure really a trigger for the circadian rhythm of subarachnoid hemorrhage? Stroke 1995;26:18051810. 24 Shimizu M, Ishikawa J, Eguchi K, Hoshide S, Shimada K, Kario K: Association of an abnormal blood glucose level and morning blood pressure surge in elderly subjects with hypertension. Am J Hypertens 2009;22:611 616. 25 Neutel JM, Schumacher H, Gosse P, Lacourcire Y, Williams B: Magnitude of the early morning blood pressure surge in untreated hypertensive patients: a pooled analysis. Int J Clin Pract 2008;62:16541663. 26 Vermeer SE, Rinkel GJ, Algra A: Circadian fluctuations in onset of subarachnoid hemorrhage: new data on aneurismal and perimesencephalic hemorrhage and a systematic review. Stroke 1997; 28:805808. 27 Khoury AF, Sunderajan P, Kaplan NM: The early morning rise in blood pressure is related mainly to ambulation. Am J Hypertens 1992;5:339344. 28 Organization for Economic Co-operation Development: OECD Health Data 2006: Statistics and Indicators for 30 Countries (version of 26 June 2006). www.oecd.org/health/ healthdata (accessed 22 March 2007).

Circadian Periodicity of Stroke

Neuroepidemiology 2010;34:2533

33

Copyright: S. Karger AG, Basel 2010. Reproduced with the permission of S. Karger AG, Basel. Further reproduction or distribution (electronic or otherwise) is prohibited without permission from the copyright holder.