(Stroke. 2000;31:1538.)
© 2000 American Heart Association, Inc.
Original Contributions |
From the Istituto di Clinica delle Malattie Nervose e Mentali (S.P.) and Istituto di Scienze Neurologiche (G.C.), Universita di Siena, Italy, and U.O. di Neurochirurgia (F.R.) and U.O. di Anestesia e Rianimazione I (E.Z.), Azienda Ospedaliera Senese, Siena, Italy.
Correspondence to S. Passero, MD, Istituto di Clinica delle Malattie Nervose e Mentali, Universita di Siena, Viale Bracci, I-53100 Siena. Italy.
| Abstract |
|---|
|
|
|---|
MethodsThe daily, weekly, and yearly variations in occurrence of ICH together with the relationship between ICH occurrence and changes in air temperature were evaluated in 1018 patients. Patients were grouped according to the presumed etiology of ICH: hypertensive ICH, secondary ICH, and ICH of undetermined origin. The contribution of demographic and clinical factors to the temporal distributions of ICH was also evaluated.
ResultsMarked differences in seasonal and diurnal patterns of ICH onset were observed in the different groups. The incidence of hypertensive ICH reflected seasonal and circadian changes in blood pressure, whereas the latter did not seem related to the onset of nonhypertensive ICH. The seasonal pattern was more evident in elderly patients with hypertensive ICH than in younger subjects. No significant weekly variations were observed; however, risk was greater on Monday in the working population.
ConclusionsOur results suggest that the higher incidence of ICH in the colder months is due to the effect of low temperatures on blood pressure and that the clustering of ICH events in the morning is due to the increase in sympathetic tone, and consequent increase in blood pressure, on awakening.
Key Words: circadian rhythm hypertension intracerebral hemorrhage seasons
| Introduction |
|---|
|
|
|---|
However, blood pressure is only a precipitating factor of bleeding in certain types of ICH; 25% to 50% of ICH is due to other etiologic factors, and one would expect to find a higher incidence of ICH in hypertensive patients during the colder months or in the morning because of higher absolute values. Blood pressure may also be influenced by factors such as hypertensive status, age, sex, smoking, and diabetes, and this influence may also affect temporal patterns of ICH onset. No previous studies have analyzed all temporal patterns of ICH onset in the same population together with the relation to changes in air temperature. A few studies present partial analyses related to the etiology of ICH6 or to major clinical and demographic variables.5 15 29
The aim of the present study was to further analyze the circadian, weekly, and yearly variations in occurrence of ICH on the basis of the time of onset of symptoms in a large population of patients, to evaluate the relationship between the occurrence of ICH and changes in air temperature, and to determine whether or not subgroups with specific clinical characteristics exhibit different patterns of onset.
| Subjects and Methods |
|---|
|
|
|---|
The time of onset of ICH was obtained from the patients or from persons witnessing the event and was defined as the time of the first reported symptom or sign, such as headache, vomiting, focal deficits, or change in mental status. All patients had a documented month and day of the week of ICH onset. In 901 patients, it was possible to determine the exact (n=646) or approximate (n=252) time of onset. The time of onset could not be obtained in 117 patients because of diffuse onset of symptoms or because they were found unconscious.
Data from the study period were pooled for analysis, and each year was divided into 4 seasons: winter (December through February), spring (March through May), summer (June through August), and autumn (September through November). The time of day was divided into four 6-hour intervals: night (midnight to 6:00 AM), morning (6:00 AM to noon), afternoon (noon to 6:00 PM), and evening (6:00 PM to midnight). Patients who became aware of symptoms on waking or who were found unconscious in bed were considered to have had ICH during the night. To eliminate the confounding influence of changes in mean temperature in the same month of different years, the distribution of ICH was also analyzed in relation to monthly mean minimum temperature and monthly mean temperature divided into six 4°C intervals. The area served by our hospital enjoys a climate with relatively cold winters and hot summers. During the study period, the monthly mean temperature ranged from 3°C to 26.1°C, and the monthly mean minimum temperature ranged from -0.8°C to 20.1°C. The temperature data were obtained from 4 weather stations in areas where the patients lived.
Location of ICH was classified as basal ganglia and thalamus, lobes, brain stem, cerebellum, and ventricular system of the brain. Patients were grouped according to the probable etiology of the ICH: (1) hypertensive ICH, which was diagnosed in patients with a history of hypertension (previous diagnosis of arterial hypertension based on systolic blood pressure >160 mm Hg or diastolic >90 mm Hg or both and/or past or present use of antihypertensive agents) and in patients who needed antihypertensive treatment on discharge but showed no apparent cause of ICH; (2) secondary ICH, namely, ICH due to all other causes (vascular malformations, long-standing anticoagulation, brain tumors, coagulopathy, thrombocytopenia, and presumed cerebral amyloid angiopathy), as demonstrated by imaging studies, cerebral angiography, biopsy, autopsy, blood analysis, and follow-up; and (3) ICH of undetermined origin.
Clinically probable cerebral amyloid angiopathy was diagnosed in
patients aged >60 years with dementia and lobar hemorrhage or
in patients who had
1 lobar (cortical or corticosubcortical)
hemorrhage at follow-up and no other cause of
ICH.30
Distributions of ICH onset in the hourly, daily, monthly, seasonal, and
temperature intervals were tested for uniformity in the overall
population and in the 3 etiologic subgroups by a 1-way
2 test for goodness of fit applied to the
number of observed versus expected episodes of ICH. For seasonal,
monthly, and temperature intervals, the expected numbers of ICH were
weighted by the number of days in each season, month, or temperature
interval. Stratified analyses were used to assess the
contribution of demographic and clinical factors to the temporal
distributions of ICH. Variables used were age, sex, diabetes
(previous diagnosis of diabetes and/or past or present use of
antidiabetic agents or need of antidiabetic treatment on discharge),
current smoking (smokers, nonsmokers), alcohol abuse (assumption of
>400 mL/wk pure ethanol), and working status. The observed number of
events in each time or temperature interval was compared with the
expected number (O/E). The relative risk of ICH occurring in specific
time or temperature intervals was evaluated by comparing the O/E
proportions on the basis of the total number observed.
| Results |
|---|
|
|
|---|
|
Season, Month, and Temperature
Table 2
shows the distribution of
ICH across the 4 seasons and demonstrates that when the entire sample
of patients is analyzed, ICH onset follows a clear seasonal
pattern. The highest number of ICH occurred in winter (O/E 1.17), and
the lowest occurred in summer (O/E 0.78), with intermediate values in
spring (O/E 1.08) and autumn (O/E 0.98) (P<0.0001).
However, the analysis of the etiologic subgroups showed that
this was true only for the subgroup of patients with hypertensive ICH
(P<0.0000), whereas patients with secondary ICH or ICH of
undetermined origin showed no significant seasonal pattern. In
particular, these subgroups lacked the winter peak of incidence and
showed a statistically nonsignificant increase in spring and autumn.
The distribution of ICH for the entire sample and for the 3 etiologic
subgroups across the 12 months (Table 3
)
was in line with the seasonal pattern.
|
|
Table 4
shows the distribution of ICH in
relation to monthly mean minimum temperature. In the entire population,
ICH was significantly related to air temperature in the sense that
onset of ICH was more frequent when temperature was low
(P<0.005). Subgroup analysis showed that this
pattern was only evident for patients with hypertensive ICH
(P<0.002). Because mean minimum temperature was highly
correlated with mean temperature (r=0.98,
P<0.000), similar results were obtained when the
distribution of ICH was analyzed in relation to monthly mean
temperature.
|
In patients with hypertensive ICH, the seasonal and temperature variations were more pronounced in older than in younger patients. Among those aged >60 years, there was 103% more ICH in winter than in summer (P<0.0000). At younger ages, seasonal changes were limited, with 31% more ICH in winter than in summer (P=0.62). Similarly, hypertensive ICH in older patients was significantly more frequent when the minimum temperature was low (<6°C), with an increased relative risk of 25% (P<0.0035), whereas younger patients had an increased relative risk of 10% (P<0.07).
The other factors (sex, diabetes, smoking, and alcohol abuse) had no significant effect on the seasonal or temperature distribution of ICH.
Day of the Week
The distribution of ICH onset across days of the week showed no
significant periodicity in the whole sample of ICH patients or in the 3
etiologic subsamples.
When the working status of patients was considered, a significant trend toward increased occurrence of ICH on Monday emerged in the total working population, with a 26% increased relative risk. This trend was also found in hypertensive ICH (42% increased relative risk) but not in secondary ICH or in ICH of undetermined origin.
Time of Onset
In the whole population, the diurnal distribution of ICH onset
(Table 5
) showed a clear circadian
variation, with a peak of incidence between 6:00 AM and
noon, corresponding to an 80% increased relative risk, and a minimum
from midnight to 6:00 AM, corresponding to a 65% lower
relative risk (P<0.0000). This pattern of onset persisted
in the etiology subgroups. However, at night, the risk of ICH was
significantly higher in patients with secondary ICH (O/E 0.72) or ICH
of undetermined origin (O/E 0.56) than in patients with hypertensive
ICH (O/E 0.16). Similarly, the predominant morning peak was more
evident in patients with hypertensive ICH (O/E 2.09) than in patients
with secondary ICH (O/E 1.31) or ICH of undetermined origin (O/E 1.43).
Compared with the expected number of ICH incidents, the morning peak
was equivalent to an increased relative risk of 109% in hypertensive
ICH, 31% in secondary ICH, and 43% in ICH of undetermined origin.
When analysis was limited to the 3-day periods, only the
distribution of the hypertensive ICH continued to be statistically
significant (P<0.0000) (Table 5
).
|
Sex, diabetes, smoking, and alcohol abuse were not found to influence the distribution of ICH onset in a significant manner. In very young patients (aged <40 years), the diurnal distribution of ICH was more uniform and significantly different from that of older patients. Younger patients lacked the morning peak of incidence (O/E 1.07 versus 1.85 in older patients) and showed a significant elevation at night, with a relative risk 2-fold that of older subjects (O/E 0.71 versus 0.35).
| Discussion |
|---|
|
|
|---|
Among studies on the seasonal distribution of ICH, some found increased occurrence of ICH1 2 3 4 5 6 7 8 9 10 13 in winter, whereas others indicated peaks in other seasons11 14 37 or no significant seasonal variation at all.16 31 32 33 34 35 38 39 However, the most consistent seasonal variation described was a seasonal pattern with a peak of incidence in winter and a minimum in summer.41
Attempts to relate the occurrence of ICH to climatic conditions have usually considered air temperature3 5 6 8 10 29 31 33 35 but have also considered hours of sunshine,6 10 39 humidity,6 10 24 29 atmospheric pressure,6 10 29 39 rainfall,3 10 31 39 and windchill.39 The most constant finding has been a significant inverse relationship between ambient temperature and the occurrence of ICH.3 5 6 8 10 29 35 In the study of Woo et al33 conducted in a subtropical climate, this relationship was found but did not reach statistical significance. The only study that failed to find such a relationship was that of Sobel et al.31 However, the data reported suggest that 8.2 ICH incidents per month occurred in the coldest months (December through February) of the study period compared with 6.9 ICH incidents per month in the other periods. Other climatological parameters were associated, to a different degree, with the occurrence of ICH, and significant associations were found with parameters strongly related to air temperature, such as hours of sunshine6 10 and rainfall.3 10
In almost all the studies, the inverse relationship between ICH occurrence and air temperature was associated with a significant seasonal variation, with peak incidence in the winter months.3 5 6 8 10 29 Only in the study of Rothwell et al35 was the higher incidence of ICH at low temperatures not associated with a significant seasonal variation, presumably because of the small number of patients.
In the present study, a seasonal variation with a characteristic winter peak of incidence was observed for the entire group of ICH patients, and this pattern was associated with a significant inverse relation with air temperature. The separation into 3 etiologic subgroups revealed a different pattern for each group. The winter peak of incidence only held true for hypertensive ICH, whereas secondary ICH and ICH of undetermined origin lacked the winter peak and showed a statistically irrelevant elevation in spring and autumn. These results differed from those of Capon et al,6 who found no significant differences between 109 patients with hypertensive ICH and 127 patients with hemorrhage due to cerebral amyloid angiopathy (n=27) or clotting disorders (n=10) or hemorrhage of undetermined origin (n=90). This discrepancy may be partly due to the small number of patients with secondary ICH and to the different inclusion criteria used. Albeit, in a small number of cases, Shinkawa et al5 observed a significant seasonal variation only in patients with arterial hypertension.
It has been documented that exposure to cold causes peripheral constriction and an increase in blood pressure17 18 19 20 ; these changes may explain the increased incidence of ICH in cold weather. Other parameters influenced by changes in temperature, such as clotting factors,42 may of course play a role. Our results strongly sustain the existence of a seasonal blood pressuresensitive variation in occurrence of ICH due to changes in air temperature. An acute coldinduced rise in blood pressure43 is probably a much more significant precipitating factor of bleeding than are slight seasonal variations in blood pressure.
In the subgroup of hypertensive ICH, older patients showed a greater increase in ICH occurrence in winter than did younger patients. A similar effect of age has been noted for myocardial infarction, ischemic stroke, and ICH mortality,44 45 although in the studies in question, the seasonal variation in mortality may have been due to the contribution of a fluctuation in the occurrence of associated diseases. The elderly may have exaggerated responses to cold with greater increases in blood pressure,17 46 and this may explain why elderly hypertensive patients are subject to greater winter increases in ICH than are younger individuals. Other authors have found stronger seasonality in the younger than the older age groups in patients with ICH5 and in patients with ischemic events.9 Differences in demographic characteristics, work routine, environmental conditions, housing, and heating in different geographic areas may explain these discrepancies.
The other traditional risk factors analyzed do not seem to have significant effects on seasonality or the relationship between air temperature and incidence of ICH.
There are only a few reports on the weekly distribution of ICH onset. Kelly-Hayes et al34 found an increased incidence of ICH on Monday, and Pasqualetti et al4 found a peak of incidence between Saturday and Tuesday, with greater incidence on Monday. In the present study, we failed to find any significant day-of-the-week variations in the entire population or in the 3 etiologic subgroups. When patients were divided on the basis of modifying factors, we found an increased risk on Monday for the working population with hypertensive ICH. Interestingly, increased risk of myocardial infarction on Mondays was documented in the working population by Willich et al,47 and Kelly-Hayes et al34 found that men who had strokes on Monday were twice as likely to work than not. There is no clear explanation for these findings; however, Willich et al47 hypothesized that external factors, such as the sudden changes in physical and mental activity on the transition from the weekend to workdays, may trigger vascular events. Stress associated with the return to work after the weekend may cause changes in physiological parameters such as blood pressure, which could trigger certain subtypes of strokes, particularly ICH and SAH.
Studies on diurnal variations in ICH onset have produced similar results, and many of these studies demonstrate a rhythm with a peak of incidence between 6:00 AM and noon and a minimum at night.7 11 12 13 14 15 16 This pattern also emerged from a recent meta-analysis performed by Elliot.48 Our results confirm this circadian rhythm, but again, this periodicity was more evident in patients with hypertensive ICH.
Other vascular events, including ischemic stroke,13 14 16 36 40 48 49 50 51 52 53 SAH,15 54 55 56 myocardial infarction,57 58 59 60 61 62 and sudden cardiac death,63 follow a similar temporal pattern, and several important physiological changes occurring in the morning waking hours may be responsible for the morning increase in cardiovascular and cerebrovascular events. However, the principal physiological process that may precipitate ICH in the morning is the increase in sympathetic tone on awakening and the associated increase in blood pressure.64 65 That the morning rise in blood pressure may act as a trigger for the onset of bleeding has also been suggested for SAH. In a study of 273 patients with SAH, Kleinpeter et al66 found that only hypertensive patients displayed a morning peak of incidence, whereas normotensive patients showed no significant circadian variation. In a study of 243 patients, Sloan et al15 found a clustering of events in the morning but only in hypertensive patients. Hypertensive subjects may be particularly vulnerable in the morning because although they have circadian blood pressure variations similar to those of normotensive subjects, they have higher average blood pressure than the latter.21 23 27 28
Among the traditional risk factors, only age seemed significantly related to the diurnal variations in ICH onset; however, we found that this was due to the fact that younger patients had almost exclusively secondary hemorrhages.
In conclusion, marked differences in seasonal and diurnal patterns of ICH onset were found between subgroups of patients divided according to the etiology of ICH. The incidence of hypertensive ICH reflected seasonal and circadian variations in blood pressure, whereas blood pressure variations did not seem to influence the onset of nonhypertensive ICH. Our results strongly suggest that the higher incidence of ICH in colder months is due to the effects of low temperature on blood pressure and that clustering of ICH events in the morning is due to the increase in sympathetic tone on awakening and the associated increase in blood pressure.
| Acknowledgments |
|---|
Received December 2, 1999; revision received March 16, 2000; accepted April 4, 2000.
| References |
|---|
|
|
|---|
-sympathetic
vasoconstrictor activity. N Engl J Med. 1991;325:986990.[Abstract]
This article has been cited by other articles:
![]() |
S Omama, Y Yoshida, A Ogawa, T Onoda, and A Okayama Differences in circadian variation of cerebral infarction, intracerebral haemorrhage and subarachnoid haemorrhage by situation at onset J. Neurol. Neurosurg. Psychiatry, December 1, 2006; 77(12): 1345 - 1349. [Abstract] [Full Text] [PDF] |
||||
![]() |
C A Jackson and C L M Sudlow Is hypertension a more frequent risk factor for deep than for lobar supratentorial intracerebral haemorrhage? J. Neurol. Neurosurg. Psychiatry, November 1, 2006; 77(11): 1244 - 1252. [Abstract] [Full Text] [PDF] |
||||
![]() |
V. L. Feigin, C. S. Anderson, N. E. Anderson, J. B. Broad, M. J. Pledger, and R. Bonita Is There a Temporal Pattern in the Occurrence of Subarachnoid Hemorrhage in the Southern Hemisphere? : Pooled Data From 3 Large, Population-Based Incidence Studies in Australasia, 1981 to 1997 Stroke, March 1, 2001; 32(3): 613 - 619. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Stroke Home | Subscriptions | Archives | Feedback | Authors | Help | AHA Journals Home | Search Copyright © 2000 American Heart Association, Inc. All rights reserved. Unauthorized use prohibited. |