Hazard ratios (95% confidence intervals) for haemorrhagic strokes adjusted for age, multivariables, and propensity score according to migraine status in Women’s Health Study (n=27 860)
| No migraine history (n=22 730) | Any history of migraine (n=5130) | Active migraine | Previous migraine* (n=1518) | ||
|---|---|---|---|---|---|
| With aura (n=1435) | Without aura (n=2177) | ||||
| No with haemorrhagic stroke (n=85) | 70 | 15 | 9 | 3 | 3 |
| Age adjusted | 1.00 | 1.04 (0.59 to 1.82), P=0.90 | 2.31 (1.15 to 4.64), P=0.019 | 0.52 (0.16 to 1.66), P=0.27 | 0.63 (0.20 to 1.99), P=0.43 |
| Multivariable adjusted (parsimonious) model† | 1.00 | 1.03 (0.59 to 1.81), P=0.91 | 2.34 (1.16 to 4.70), P=0.017 | 0.54 (0.17 to 1.72), P=0.29 | 0.59 (0.19 to 1.88), P=0.37 |
| Propensity score adjusted‡ | 1.00 | 0.98 (0.56 to 1.71), P=0.93 | 2.25 (1.11 to 4.54), P=0.024 | 0.49 (0.14 to 1.70), P=0.26 | 0.45 (0.14 to 1.48), P=0.19 |
*History of migraine but no active migraine in year before completion of baseline questionnaire.
†Adjusted for age, history of hypertension, smoking, BMI, alcohol consumption, and total cholesterol concentration.
‡One propensity score calculated for overall migraine and one for having migraine with aura. Both models included information on age, systolic blood pressure, use of antihypertensive medication, smoking, BMI, alcohol consumption, exercise, total cholesterol concentration, postmenopausal status, postmenopausal hormone use, history of oral contraceptive, history of diabetes, family history of myocardial infarction before age 60, annual household income, level of education, multivitamin use, ethnicity, randomised treatment, and interaction terms for age with smoking, BMI, and oral contraceptives. Hazard ratios estimated by weighted Cox proportional hazard models.
