Testosterone Therapy and Risk of Cardiovascular Disease in Men

Abstract

The clinical question about which men should receive testosterone therapy is controversial, with data from short-term clinical trials suggesting benefits for improving sexual function, strength, and well-being. What is missing from the literature are data from randomized trials that include a sufficient number of men for an adequate amount of time to assess the longterm benefits and risks of testosterone therapy. There is no study involving men that is equivalent to the Women’s Health Initiative, nor is it likely that there will be a trial of equal scale. Because testosterone therapy is available and prescribed for an estimated 2.9% of US men aged 40 years or older, 1 observational data from existing cohorts of men can contribute meaningfully to assessment of therapeutic risk. In this issue of JAMA, 2 Vigen and colleagues present retrospective analyses from the Veterans Affairs system of men who had undergone coronary angiography, had subsequent total testosterone assessment, and were found to have a testosterone level of less than 300 ng/dL. Through linkage with pharmacy data, 1223 men, mean age 60.6 years, who initiated testosterone therapy were compared with 7486 men, mean age 63.8 years, who did not. The testosterone users were found to have an increased risk of the composite end points: 67 died, 23 had myocardial infarctions, and 33 had strokes, whereas among those who did not receive testosterone therapy 681 died, 420 had myocardial infarctions, and 482 had strokes, for an absolute 3-year event rate of 25.7% vs 19.9% (hazard ratio, 1.29; 95% CI, 1.04-1.58). Importantly, this estimate did not differ between the men with and without coronary artery disease, which was ascertained in all men by coronary angiography, and was similar when revascularization was included in the outcome. Any pharmacoepidemiology study is susceptible to confounding by indication. The authors have incorporated a wealth of data on potential confounders and a sophisticated weighted analysis with testosterone use as a timedependent covariate in an attempt to mitigate the effects of confounding. The men who were prescribed testosterone were slightly healthier than those who were not, with the exception of testosterone level and obesity. Ordinarily, this would raise concerns about prescribing bias, as was seen with prescription of estrogen therapy to more health conscious women. However, in the study of Vigen et al, the