To the Editor.
—In a randomized clinical trial, Dr Frank and colleagues1 assessed the effects of maintenance of normothermia in the operating room on perioperative cardiac events. Using a multivariate logistic regression model, the authors concluded that hypothermia was associated with a 2.2 times greater rate of adverse perioperative cardiac events than was observed in normothermic patients. Although preoperative β-adrenergic blocker use also was significantly associated with fewer perioperative cardiac events, there was minimal discussion of this finding. The authors cite several studies demonstrating that the adverse cardiac effects of hypothermia can be attributed to an increased adrenergic response. This suggests that preoperative β-adrenergic blockade may be an effect modifier on the relationship between hypothermia and cardiac events, although this possibility was not explored.When logistic regression is used for data analysis, the first step is to assess for effect modification by evaluating interaction terms (product terms) consisting of