Objective: This study aimed to paradigmatically show the development of a gender score that can be used as either an adjustment or a matching variable to separate the effects of gender versus biological sex in a sample of older adults.
Methods: Our sample comprised 1100 participants from the Berlin Aging Study II (52% women, mean [standard deviation] age = 75.6 [3.8] years). The gender score included a multitude of gender-related variables and was constructed via logistic regression. In models of health outcomes, it was used as an adjustment variable in regression analyses as well as a matching variable to match older men and women according to their gender.
Results: Matching by gender substantially reduced sample size to n = 340. Analyses (either adjusting for gender or matching men and women according to gender) revealed that female sex was independently associated with lower grip strength (B = -14.47, 95% confidence interval [CI] = -15.51 to -13.44), better cognitive performance (B = 3.47, 95% CI = 1.94 to 5.0), higher pulse wave velocity (B = 0.19, 95% CI = 0.06 to 0.31), lower body mass index (B = -0.97, 95% CI = -1.74 to -0.21), and lower rates of metabolic syndrome (odds ratio = 0.53, 95% CI = 0.37 to 0.77). In addition, both sex and gender were independently associated with cognitive performance and depression.
Conclusions: Calculating a gender score allows for the inclusion of a large number of variables, creating parsimonious models that are adaptable to different data sets and alternative gender definitions. Depending on the research question and the sample properties, the gender score can be used as either an adjustment or a matching variable.Trial Registration: DRKS-Deutsches Register Klinischer Studien (Study ID: DRKS00016157).
Copyright © 2022 by the American Psychosomatic Society.