Hoes and colleagues [1] concluded that "the use of non-potassium-sparing diuretics and ß-blockers is associated with an increased risk for sudden cardiac death." Their data, however, do not fully support (or at least explain) this conclusion. First, none of the "bivariate" associations (their Table 2 between exposure to various categories of diuretic or ß-blocker use and outcome as case–control status were clinically or statistically significant. For example, 21.8% of case-patients [n = 56] were taking non-potassium-sparing diuretics, regardless of ß-blocker use, compared with 17.9% of controls (n = 46) (odds ratio, 1.5 [95% CI, 0.9 to 2.4]). The other associations had similarly small and nonsignificant distinctions between case-patients and controls. Thus, one interpretation of the study results is that the study is essentially "negative" and does not warrant further analysis.

The authors, however, adjusted the crude results with data from their Table 1"for differences in the risk for sudden cardiac death in a multivariate statistical analysis." As expected, case-patients (who had hypertension and subsequent sudden cardiac death) were much more likely than controls (who also had hypertension and were selected from general practitioners' practices) to have other cardiac risk factors, such as previous myocardial infarction, heart failure, angina, and arrhythmias. The presence of these additional risk factors among the case-patients should decrease the odds ratio for the association of non-potassium-sparing diuretics or ß-blockers, given that these drugs are often used to treat the conditions themselves. Surprisingly, the "adjusted" odds ratios (their Table 2 for the associations between exposure and outcome are increased and are statistically significant. Accordingly, the study requires additional information to convince readers that the results are not merely the result of applying multivariate methods of analysis—an approach that can cause problems if mathematical assumptions are not met [2].