Beyond binary sex/gender variables in clinical algorithms: a critical integrative review and the Algorithmic Gender Misclassification Framework for transgender health

Background Clinical medicine embeds binary sex/gender classifications within laboratory reference intervals, kidney-function equations, cardiovascular risk calculators, medication-dosing rules, and electronic decision support. These instruments were derived and validated in populations presumed to be largely cisgender, usually without reporting how sex or gender was ascertained, creating uncertainty when they are applied to transgender people, including but not limited to those receiving gender-affirming hormone therapy (GAHT).Aim This critical integrative review examines how binary sex/gender coding operates across clinical decision systems used in transgender healthcare and develops a theory-building account of how apparently neutral computational rules may contribute to inequity.Method A purposive, iterative corpus of 82 empirical, theoretical, guideline, standards, and review sources was integrated across laboratory medicine, nephrology, cardiovascular medicine, pharmacology, oncologic screening, and health informatics. Sources were retained when they clarified how the classification of sex or gender alters risk estimation, interpretation, eligibility, dosing, or the architecture of clinical classification.Results Across the synthesized literature, a recurrent pattern of classification discordance was identified: the variable an algorithm requests may not correspond to the physiological, social, or clinical construct it is intended to represent. Evidence directly demonstrates classification-dependent differences in the estimation of kidney function and prognosis, laboratory interpretation, cardiovascular risk estimation, and treatment thresholds. I propose the Algorithmic Gender Misclassification Framework (AGMF), comprising five mechanisms: binary input compression, construct proxy failure, classification-dependent divergence, decision amplification, and cumulative clinical disadvantage.Conclusions Algorithmic gender misclassification should be investigated as an underrecognized structural mechanism of transgender health inequity, distinct from interpersonal discrimination. Clinical systems should replace indiscriminate binary sex/gender inputs with construct-specific variables such as hormonal milieu, organ inventory, body composition, and directly measured biomarkers when clinically appropriate. Transgender-inclusive validation of clinical algorithms is required and would improve care for people of all genders.