Letters Section Editor: Robert M. Golub, MD, Senior Editor.
To the Editor: Dr Thompson and colleagues1 studied erectile dysfunction as a predictor of cardiovascular disease (CVD). We have concerns about the validity and generalizability of their findings.
Because erectile dysfunction and CVD share risk factors, bias in study design may have important impact on results, particularly when the measured associations are weak, as in this study. The Berkson fallacy is one such bias.2 This selection bias would operate if the study population was nonrepresentative, and this biased selection was related both to the risk factor and outcome, eg, was enriched with persons with both erectile dysfunction and an independent propensity to develop CVD. The high incidence rate of erectile dysfunction (65% in 7 years vs 23% in 8.8 years in the Massachusetts Male Aging Study3) and low incidence of CVD (compared with data from the American Heart Association4) suggest that selection bias could have taken place if it resulted in a relative underrepresentation of CVD in the men without erectile dysfunction.
Schouten B, Bohnen AM, Thomas S. Erectile Dysfunction and Incidence of Cardiovascular Disease. JAMA. 2006;295(17):1998-2000. doi:10.1001/jama.295.17.1999-a