We read with great interest the study by Xu et al, which examined the association between premature menopause, both natural and surgical, and the incidence of depression requiring hospitalization in the UK Biobank. The large sample size, extended follow-up, and thoughtful mediation analysis through frailty and menopausal hormone therapy provide valuable insights into women’s mental health. However, we believe that the main conclusion may be influenced by a structural bias related to the handling of time at risk. ,
In this design, exposure status was retrospectively determined at baseline, and only women who had already reached menopause were included. Depression hospitalizations were counted only if they occurred after enrollment, while those who experienced such events before baseline were excluded. The period between menopause and enrollment, during which participants had to remain event-free to be eligible, constitutes an immortal time. This interval tends to be longer among women with earlier menopause, particularly surgical menopause, creating unequal opportunities for the outcome to occur. Women who experienced menopause earlier had to survive longer without depression to enter the cohort, which could lead to a systematic underestimation of the true association. Although several sensitivity analyses addressed comorbidities, data truncation, and missingness, none directly accounted for this time-related selection mechanism.
To clarify whether and to what extent this bias might affect the results, the authors could align the time scale of exposure and outcome more precisely. Analyses that start follow-up at a uniform age would help equalize observation windows. Using attained age as the time scale with left truncation could accurately separate risk time before and after menopause, treating menopause as a time-varying exposure. Alternatively, defining a common starting point such as age 40 or the first study contact and updating exposure status thereafter could emulate a trial-like design. Stratifying by or restricting the interval between menopause and baseline, and reporting the number of excluded women due to prebaseline depression by exposure group, would also help quantify the potential impact. Addressing this issue would strengthen the causal interpretation of the findings and enhance understanding of the link between premature menopause and women’s risk of severe depression.
The authors report no conflict of interest.
References
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
