Sixteen studies including three cohort studies and thirteen case-control studies were included.
Current OC use was associated with increased risk of ischemic stroke. RR = 2.75 (95% CI: 2.24, 3.38; P < 0.001). Heterogeneity was present (P = 0.01). After eliminating each study in turn RR ranged from 2.63 to 3.00 with the lower 95% CI never crossing 1. Stroke risk was not associated with past use of OC (9 studies). RR (past vs never use) = 0.86 (95% CI: 0.69, 1.08). No evidence of heterogeneity. Stroke risk appeared to decrease with time (P = 0.006).
Smaller estrogen doses were associated with lower risk (P = 0.01 for trend) but risk was significantly elevated for all doses. RR for low dose estrogen (7 studies) 2.08 (95% CI: 1.55, 2.80). Heterogeneity was present among the low dose studies.
Newer generation progesterones tended to be associated with less elevated risk. RR (3 first generation studies) = 3.21 (95% CI: 2.16, 4.77) compared to 3 third generation studies RR = 2.11 (95% CI: 0.96, 4.64). Trend across generations was non significant. Within progesterone generations there was a non significant trend for higher estrogen doses to be associated with higher risk of stroke.
No differences were seen between generations at equivalent estrogen doses.
Cohort studies tended to produce higher summary RR than case-control studies.
Studies that did not control for smoking and those using hospital based controls found higher RR but no other patient characteristics or elements of study design were important.
Stratification by potential confounders: no evidence of heterogeneity in studies controlling for alcohol use, for smoking, hypertension or both smoking and hypertension. The risk of stroke was minimally affected by the presence of other risk factors. Similar RR were found for smokers vs non smokers, in those with and without a history of migraine or hypertension and for those < 35 years of age vs those >= 35 years of age.
Risk of stroke with low dose estrogen after adjustment for smoking and hypertension (6 studies): RR = 2.04 (95% CI: 1.51, 2.76).
Risk of stroke with low dose estrogen in population based studies after adjustment for smoking and hypertension (5 studies): RR = 1.93(95% CI: 1.35, 2.74).
Risk of CVST with OC use (2 studies): increased risk in current users of OC. RR = 15.9 (95% CI: 6.98, 36.2).
Neither the funnel plot not Kendall tau demonstrated evidence of publication bias.