Correction for full version of this article
Risk of ischaemic stroke in people with migraine: systematic review and meta-analysis of observational studies
The authors of this paper, Mahyar Etminan and colleagues, alerted us shortly before publication to the fact that some of the results were incorrect, although the conclusions are unaffected (BMJ 2005;330:63-5, 8 Jan). We managed to correct the abridged version in the journal but not the full version on the web. Here are the corrections relating to the full web version.
In table 2, the abstract, and the results section, the correct relative risks of stroke are 2.88 (95% confidence interval 1.89 to 4.39; Ri=0.33; P=0.16) in people who had migraine with aura and 1.56 (1.03 to 2.36; Ri=0.50; P=0.07) for migraine without aura. These data came from eight studies (references 13, 14, 17-19) and not six as originally stated.
In table 1, the relative risks of migraine for Carolei should be 8.6 (1.0 to 75) for migraine with aura and 1.0 (0.5 to 2.0) without aura. The relative risks for migraine with aura should be 3.81 (1.26 to 11.5) for Chang and 2.79 (0.94 to 8.31) for Donaghy. For Schwaag the correct relative risks are 2.11 (1.16 to 3.82) for any migraine, 1.0 (0.28 to 3.53) for migraine with aura, and 2.5 (1.3 to 6.3) for migraine without aura. In column 5, sex, age, race, and date of discharge should be listed as covariates and confounders for Henrich, and age and sex for Schwaag. In column 6 (cases:controls or cohort size), the correct entry for the Collaborative Group is 430:151, the entry for Buring should be 213:21 952, and that for Merikangas should be 423:12 220. In the final column (which should be titled "Cases:controls or cohort size with migraine"), corrected entries are 115:212 for the Collaborative Group, 26:26 for Chang, and 19:1479 for Buring.