Testosterone Intention-to-Treat Trial: Duration of Treatment Matters

Recent studies of testosterone replacement therapy with short treatment duration and follow-up have suggested that it may increase risk of heart attack and stroke, and prompted the US Food and Drug Administration to issue a caution.

The last large-scale controlled observational study of testosterone treatment in older men published in Lancet Diabetes & Endocrinology reports different — and somewhat surprising – results.

This study enrolled men whose physicians gave them new prescriptions for testosterone (n=10,311), using the date of first prescription and the corresponding date for controls (n=28,029) as the study entry date. The study was not randomized, which prompted the investigators to use statistical tools to adjust for as many confounders as possible. This was essentially an intention-to-treat (ITT) observational study. Statisticians often described ITT analysis as “once randomized, always analyzed.”

The investigators identified a subtle association between testosterone exposure and outcomes.

Men who took testosterone for shorter periods of time had an increased risk of mortality, with a hazard ratio of 1.11, and cardiovascular events, with a hazard ratio of 1.26. These men had taken testosterone for 120 days or fewer, with a median duration of 2 months.

Men who took testosterone for the longest periods of time — 511 days or more – had reduced risks of mortality, cardiovascular events, and prostate cancer. The hazard ratios were 0.84, 0.6 and 0.67, respectively.

Risk of prostate cancer did not increase among men in the lowest tertile of testosterone exposure, but it did fall with testosterone longer exposure.

These findings indicated that use of testosterone and its possible downstream effects is not a simple association. The decrease in prostate cancer risk may reflect bias; however, men prescribed testosterone are generally screened for prostate cancer before stating therapy.

In an accompanying essay, an expert in the field compared this study to the Women’s Health Initiative (WHI) and its findings concerning long-term hormone-replacement therapy. The commentator indicates that although early findings from the WHI observational analyses seemed to have shown benefit, ultimately, it appeared that confounders, such as socioeconomic status, might have been overlooked.

Had the investigators used intention-to-treat design earlier, they may have identified what they eventually came to understand: that in the long run, HRT has health implications.