Fig. 2

Smoothed curve fitting analysis of OBS and the prevalence of TD. (A) Dose–response association between OBS and total testosterone level. (B) The association of the prevalence of TD with OBS score in logistic regression adjusting for age, race, PIR, education level, marital status, diabetes, hypertension, total energy intake, and seasonal timing of serum sample collection. The red solid and blue dotted lines represent the central risk estimates and 95% CIs.