Objective: To determine whether diurnal variation occurs in the onset of stroke.
Design: Community based study over four years.
Setting: Oxfordshire, United Kingdom.
Subjects: 105,000 people, of whom 675 had a first ever stroke. 545 had a cerebral infarction, 66 had primary intracerebral haemorrhage, 33 had subarachnoid haemorrhage, and in 31 the type of stroke was not known.
Main outcome measures: Time of stroke and degree of activity at onset.
Results: In the 578 patients for whom it was known whether onset occurred while asleep or awake, the proportion with onset during sleep was 25% (135/545) for cerebral infarction, 17% (11/66) for primary intracerebral haemorrhage, and 0% (0/33) for subarachnoid haemorrhage. This difference persisted if patients in whom it was not known whether they were asleep or awake at onset were classed as asleep. For all stroke types together there was a significant (chi 2 = 218.7, p less than 0.001) diurnal variation with a morning peak between 0800 and 1000, which persisted even after allowing for strokes first noted on waking by redistributing the hour of onset through the preceding eight hours (chi 2 = 47, p less than 0.001). A significant diurnal variation was also found in the onset of cerebral infarction (peak 0800-1000, chi 2 = 208.4, p less than 0.001). Fewer patients had other forms of stroke and the diurnal variations for primary intracerebral haemorrhage (peak 1000-1200) and subarachnoid haemorrhage (peaks 0800-1000 and 1800-2000) were not significant. There seemed to be a second smaller peak for all types of stroke.
Conclusions: All types of stroke are most likely to occur after waking in the morning. The cause of the circadian variation requires further study.