Statistical Analyses

Means, SD, and percentages were provided to describe the sample. Bivariate analyses utilized t-tests and tests of proportions to compare diagnostic groups. Multivariable analyses utilized logistic regression to examine predictors of chair-rise and balance limitations. Adjusted odds ratios (OR) and 95% confidence intervals were reported. Due to the small sample sizes examined, we used a nested modeling approach in the multivariable analysis. Model 1 included participant characteristics; Model 2 added chronic conditions; Model 3 added the use of specific classes of medications. Pseudo-R² was used to examine model fit and to assess relative improvements in model fit when incorporating additional measures. Logistic regression was also used to examine associations between psychotic symptom dimensions and both chair and balance limitations. In these analyses, Model 1 examined components of positive and negative symptoms, whereas Model 2 considered the potential for different symptoms to explain physical limitations. Ordinary least squares regression was used to conduct sensitivity analyses on chair-rise speed as a continuous outcome. Additional analyses were also conducted examining the role of depressive symptoms. Continuous analysis was not attempted using balance times because balance times were severely skewed.

Analyses were conducted using Stata 15/SE (StataCorp).