Chronotype changes after sex hormone use 209 a unique and novel way of studying effects of sex hormones on chronotype in hormone users. Secondly, the prospective setup enables within-person comparisons of chronotype and sleep duration, meaning the resulting estimates are more reliable for assessing possible causal effects of sex hormones. A methodological limitation of this study is that a subgroup of participants in this study used alarm clocks on free days. It is preferable to determine someone's natural rhythm on free days, without the interference of an alarm clock. Therefore, participants who use an alarm clock on free days should be excluded from the calculation of MSFsc. The number of participants using an alarm clock on free days was quite large in our cohort, and excluding these subjects would significantly reduce the size of our study sample. Therefore, we decided not to exclude participants who used an alarm clock on free days. We have instead opted to conduct a sensitivity analysis for participants who did not use an alarm clock on free days. We found a 30-minute later MSFsc in the TM group and a significant 17-minute earlier MSFsc in the TF group, indicating that including participants who use alarm clocks on free days did not change the direction of the found effects. Another limitation in our study setup is that starting GAHT is associated with changes in depression and anxiety, since starting the use of genderaffirming hormones was found to most likely reduce depressive symptoms in both TM and TF groups (Costa & Colizzi, 2016). Depression is related to disturbed sleep and circadian rhythm alterations, meaning that the interaction between depression and chronotype after starting GAHT could be affecting our results (Dai & Hao, 2019; Soria & Urretavizcaya, 2009). Therefore, our results should be interpreted taking into account that participants’ depressive symptoms might change after the start of GAHT. Future research that examines sleep timing in interaction with depressive and anxiety symptoms in transgender persons during GAHT use could contribute to a better understanding of our findings. The final limitation is that chronotype and related sleep variables were measured through a self-reported questionnaire. Participants estimate their own sleep patterns, which might be an over- or underestimation since only two questions about sleep onset and wake-up time are included in the questionnaire. There is a well-known paradox in sleep and sex differences,
RkJQdWJsaXNoZXIy MTk4NDMw