Table 3. Relative Risk for Ischemic Stroke After Surgery (within 30 days)
* Data were analyzed using conditional logistic regression making use of the 1:1 matched pair design (Breslow and Day, 1980). 17An indicator variable for surgery within 30 days before the index date was included in a multivariate model along with all variables from the “final” model identified by Whisnant (Neurology, 1996) 10with the exception of smoking status. The odds ratio (OR) and corresponding P value cannot be computed directly from the information provided.
† There were 18 cases and 0 controls that underwent one or more high-risk (cardiac, neurologic, or vascular) surgeries in the 30 day period before the index date. To determine whether non–high-risk surgery was associated with ischemic stroke, an analysis was performed that excluded these 18 cases and their matched controls.