Considerations for collecting and analyzing longitudinal data in observational cohort studies of transgender, non-binary, and gender diverse people.

The health and well-being of transgender, non-binary, and gender-diverse people is receiving increasing attention from epidemiologists and public health researchers, including those utilizing longitudinal observational cohort studies. These longitudinal studies are advantageous over cross-sectional observational study designs given their scope over several timepoints rather than one, and when exposures and outcomes are prospectively measured this improves validity of causal claims. However, within these longitudinal studies, gender is often collected inconsistently (e.g. only asked at a single timepoint), or inadequately (e.g. questions that use limiting notions of gender). Due to the temporal nature of gender, this introduces potential including misclassification error and may provide an incomplete picture of gender diversity in a sample. This article considers these methodological issues and offers evidence-based recommendations to ensure longitudinal data on trans, non-binary, and gender-diverse people is treated with epidemiological rigor, while maintaining inclusivity.