Bias from depletion of susceptibles: the example of hormone replacement therapy and the risk of venous thromboembolism

Abstract

Purpose: The data on the association between hormone replacement therapy and the increased risk of venous thromboembolism (VTE) in postmenopausal women are conflicting. The observed differences between oral estrogen and oral estrogen–progestogen combination formulations may be the result of bias from depletion of susceptibles. Methods: We used United Kingdom's Clinical Practice Research Datalink to identify the cohort of all women aged 50 to 79 during 1987–2008, with all incident cases of VTE occurring during the study period identified. Using a nested case–control approach, the rate ratios (RRs) of VTE with current use of oral estrogen and oral estrogen–progestogen combinations were estimated as a function of duration of use using conditional logistic regression with cubic splines. Results: The cohort of 955 582 postmenopausal women included 23 505 cases of VTE matched to 231 562 controls. The risk of VTE was increased with current use of oral estrogen (RR 1.49; 95% confidence interval: 1.37 to 1.63) and oral estrogen–progestogen (RR 1.54; 95% confidence interval: 1.44 to 1.65), relative to non‐use. When assessed by duration of use, the risks with oral formulations were particularly elevated during the first year of use and were reduced subsequently. Conclusion: The phenomenon of depletion of susceptibles should be considered in cohort studies evaluating acute side effects of medications. This can be achieved by estimating the risk as a function not only of current use but also of duration of use. Copyright © 2017 John Wiley & Sons, Ltd.