Figure. Analysis of the Homeostasis Model Assessment of Insulin Resistance (HOMA2-IR) in First-Episode, Antipsychotic-Naive Patients With Schizophrenia and Healthy Control Individuals.

A, Association of HOMA2-IR with clinical status. B, Association of HOMA2-IR with the 108 loci schizophrenia polygenic risk score. C, Association of antipsychotic drug switching during the initial 12 months of treatment with baseline HOMA2-IR. Plots show adjusted mean with standard error (A) and marginal effects with 95% confidence intervals (B and C). Statistical tests included analysis of covariance (A) and multivariable linear (B) and logistic (C) regression. Covariates were selected using bidirectional elimination and Bayesian information criterion from age, sex, body mass index, race/ethnicity, smoking, alcohol consumption, cannabis use, and previous psychiatric medication (A-C); baseline Brief Psychiatric Rating Scale, Scale for the Assessment of Positive Symptoms, and Scale for the Assessment of Negative Symptoms scores (B; schizophrenia group); and the initial treatment drug (C). Only cases with complete data were analyzed. P values were obtained by permutation testing (1000 permutations). P less than .05 was considered significant. Numbers: A, 54 control individuals and 56 patients with schizophrenia; B, 53 control individuals and 49 patients with schizophrenia; C, 20 patients with schizophrenia with no drug switch and 21 patients with schizophrenia with drug switch (13 owing to low efficacy, 5 owing to adverse effects, and 3 owing to noncompliance).

^aP < .001.

^bP < .05.