Results

During the 2 years of the study, 149 incident patients with spontaneous subarachnoid haemorrhage were enrolled and interviewed. Detailed information, including evaluation, treatment, and outcome of these patients, is described elsewhere.17 Of the 149 patients enrolled in the study, 74 could be interviewed. The remaining 75 case patients could not be interviewed because of death or disability, and their proxies were interviewed instead. A detailed description of the completeness and accuracy of interview data from proxy respondents is found elsewhere.24 Through random digit dialing, 298 matched control subjects were enrolled and interviewed. Mean (SD) age was 56.3 (16.7) years for case patients and 56.2 (16.5) years for control subjects. Sixty nine per cent of subjects were women. White people comprised 86.6% of case patients and 91.9% of control subjects. Education of less than 12 years applied to 19.7% of case patients and 12.7% of control subjects.

Physical activities immediately before the bleed were reported by 38 of 149 (26%) case patients and are listed in table 1, along with their MET values when known. In the case-control analysis, the risk of subarachnoid haemorrhage was increased for those engaged in vigorous physical activity at reference time compared with those engaged in non-vigorous or no physical activity (RR 11.6; 95% CI 1.2–113.2, table 2). Similar results were found in the case cross over analysis, in which each case patient served as his or her own control (see appendix for details on calculation of RR). The RR was 15.0 (95% CI, 4.3–52.2) of having a subarachnoid haemorrhage during a 15 minute period of vigorous physical activity compared with a 15 minute period of non-vigorous or no physical activity.

Higher long term regular physical activity tended to be protective against spontaneous subarachnoid haemorrhage, but none of the RRs or the test for trend were significant (table 3). Exploration for confounding showed smoking status as the only confounder. When examining the risk for subarachnoid haemorrhage for “somewhat active” and “very active” persons compared with “inactive” persons according to age, sex, hypertension status, body mass index, and smoking status, there was no evidence for effect modification (tests for heterogeneity all p>0.05).

Discussion

Although only a few subjects in this study engaged in vigorous physical activity at the reference time, we found a strong association between spontaneous subarachnoid haemorrhage and acute vigorous physical activity which could not be explained by potential confounding factors, including hypertension or baseline physical activity. This association was similar whether considering the traditional case-control analysis (RR 11.6) or the case cross over analysis (RR 15.0). We are not aware of other studies that have examined this question with these types of analyses, but these findings are consistent with previous descriptive reports. In a study from Rochester, MN, USA12 activity at the time of haemorrhage was recorded for 84 patients with subarachnoid haemorrhage. Forty per cent of the bleeds occurred during periods of physical or emotional stress. Tsementzis et al 11found that among 557 patients with stroke, subarachnoid haemorrhage (n=194) occurred significantly more often during sexual intercourse, sporting activity, and toileting when compared with intracerebral haemorrhage and cerebral infarction. In the Cooperative Study, two thirds of 2288 case patients with aneurysmal subarachnoid haemorrhage were related to physical activity, emotional strain, trauma, or surgical operation.13 Another study found that strenuous physical activity was associated with 25% of patients with aneurysmal subarachnoid haemorrhage compared with 14% of patients with non-aneurysmal subarachnoid haemorrhage .14

Although long term regular physical activity over the 12 months before the reference time was associated with a lower risk for subarachnoid haemorrhage, this association did not reach significance. This finding is consistent with previous findings examining the protective nature of regular physical activity against stroke.1 2 4-7 15 25 Only two studies have considered this question specifically for subarachnoid haemorrhage. The Honolulu Heart Program used an estimate of current 24 hour habitual physical activity to prospectively study the impact of baseline physical activity on subarachnoid haemorrhage in men.15 In that study, men were divided into equal tertiles of “inactive,” “partially active,” and “active.” The study found an excess of subarachnoid haemorrhage in inactive older men (ages 55 to 68). For haemorrhagic stroke, which included intracerebral as well as subarachnoid haemorrhage, a greater than threefold excess of haemorrhage was present among inactive compared with active older men, after controlling for hypertension, diabetes, and left ventricular hypertrophy. Another study found no relation between subarachnoid haemorrhage and physical activity, but included only 14 patients with subarachnoid haemorrhage and assessed physical activity with a crude four point scale.26 Studies with large sample sizes and detailed measures of physical activity are needed to clarify the protective potential of habitual physical activity level against subarachnoid haemorrhage.

The association between acute physical activity and subarachnoid haemorrhage is likely mediated through hypertension induced by exertion. Hypertension has been documented as a risk factor for subarachnoid haemorrhage in longitudinal,27-29case-control,30-33 and ecological studies.11 34 35 One study showed that the probability of intracranial aneurysm rupture increases with increased intra-aneurysmal pressure, which is the same as systemic arterial pressure.36 When the stress on the wall of an aneurysm exceeds wall strength aneurysmal rupture occurs. A study of the formation and rupture of intracranial aneurysms showed the importance of properties of the aneurysm wall and flow dynamics through the aneurysm, which are influenced by heart rate and blood pressure.37 Exercise is associated with a rapid rise in systolic blood pressure due to an increase in cardiac output.38 During defecation, micturition, and other postural activities the predominant physiological response is a Valsalva manoeuvre.39 During the course of this manoeuvre, arterial blood pressure rises and intracranial cerebrospinal fluid pressure drops, creating a large transmural pressure gradient that has been implicated in the pathogenesis of aneurysmal rupture.40 Physical activity has also been found to activate the fibrinolytic system.41-44 This haemostatic change may also play a part in the initiation or progression of subarachnoid haemorrhages. The possible protective nature of long term regular physical activity is likely mediated through improved cardiovascular conditioning, leading to lower baseline blood pressure and attenuated acute rise in blood pressure and heart rate from acute physical activity.45

The study has several strengths and limitations in its ability to determine the relation between acute and baseline physical activity and subarachnoid haemorrhage. Case patients and control subjects came from a defined geographical population and were asked detailed questions about their physical activity. Such questionnaires have been found to be accurate estimates of physical fitness.20 The current study was able to collect physical activity data that were more sensitive and quantifiable than that obtained in many other studies. Information about exposures was obtained from proxies for about half the case patients and their matched control subjects due to death or disability from their bleed. The comparison of responses from index and proxy control subjects showed good reliability for broad indicators of physical activity such as whether or not they were engaged in recreational physical activity in the past 12 months and whether they were inactive, active in only non-vigorous activities, or active in vigorous activities.24 However, only moderate reliability was found for detailed physical activity measures such as minutes of exercise/week and average kcal expended/week in recreational activity. The odds ratios obtained from proxy data were slightly biased toward the null compared with those obtained from index subject data, which may in part explain our inability to find a significant protective effect of long term regular physical activity.

Another potential limitation relates to recall bias. We asked subjects to recall their activities at a specific time. For the case patients the time was the onset of their subarachnoid haemorrhage. Control subjects lacked such an anchoring event. Consequently, we only required them to recall events from the same time of day and day of week as for their matched case patient, but in the week before their interview. Such physical activity recall surveys have been shown to be reliable and valid, particularly for recalling vigorous physical activity.46 Despite these efforts, the case patients and their proxies may have recalled with more details and with greater accuracy the events around the reference time than the control subjects.

Given our concern that differential recall might bias the results of our traditional case-control analysis, we also performed a case cross over analysis. This technique involves only information from case patients, whereby each patient serves as his or her own control. It was developed to identify transient effects of exposures, such as physical activity, on the risk of diseases with acute onset, such as subarachnoid haemorrhage. As with the more traditional case-control analysis, the case cross over analysis indicated that vigorous physical activity increased the risk of subarachnoid haemorrhage. These results add to the evidence indicating that the case cross over method is useful when the exposure in question is highly intermittent. The freedom of the case cross over method from confounding between subjects makes it uniquely suited for the study of physical activity, which may be subject to many personal factors, such as the frequency of exercise, emotional states at the time of exercise, and preferences for exercise at a particular time of day.

In this study, vigorous physical activity was associated with an acutely increased risk of spontaneous subarachnoid haemorrhage but was present in only a few patients. Although a regular exercise programme may reduce the overall risk of stroke, perhaps including subarachnoid haemorrhage, the risk of bleed seems to be increased during vigorous physical activity.