Baseline Body Mass Index and Anthropometric Responses to an Immersive Virtual Reality-Based Exercise Intervention: An Exploratory Secondary Analysis of a Randomized Controlled Trial.

Background/Objectives: Whether baseline body mass index (BMI) modifies the anthropometric response to immersive virtual reality (IVR)-based exercise interventions remains unclear. This exploratory secondary analysis of a randomized controlled trial evaluated whether baseline BMI was associated with heterogeneity in the effect of an IVR-based exercise intervention on post-intervention anthropometric outcomes in university students. The treatment effect modification analysis was not prespecified in the parent trial protocol or registry. Methods: Sixty-four participants were randomized to IVR or control groups, of whom 60 (30 per group) provided post-intervention anthropometric data and were included in the principal complete-case analysis. Post-intervention BMI was designated as the principal outcome of the present secondary analysis, whereas body weight, waist circumference, and hip circumference were secondary outcomes. Treatment effect modification was evaluated using baseline-adjusted ordinary least-squares regression models including treatment allocation, centered baseline BMI, and the treatment × baseline BMI interaction. Sensitivity analyses included adjustment for sex, HC3 heteroscedasticity-consistent standard errors, influence diagnostics, exploratory assessment of nonlinearity, and multiple imputation including all 64 randomized participants. Categorical BMI analyses were conducted post hoc and considered exploratory. Results: No statistically significant linear treatment-by-baseline BMI interaction was detected for post-intervention BMI (B = -0.039, 95% CI: -0.098 to 0.020; p = 0.189) or for body weight, waist circumference, or hip circumference (all p > 0.05). Model-estimated IVR-control differences in post-intervention BMI became progressively more negative with increasing baseline BMI, but uncertainty widened toward the upper end of the observed BMI distribution. Multiple imputation including all 64 randomized participants yielded a nearly identical interaction estimate (B = -0.039, 95% CI: -0.097 to 0.020; p = 0.192). Post hoc categorical analyses yielded different patterns from the principal continuous analysis and were considered hypothesis-generating. Conclusions: No statistically significant linear treatment-by-baseline BMI interaction was detected for post-intervention BMI or the secondary anthropometric outcomes. However, the modest sample size, limited representation at the upper end of the BMI distribution and widening uncertainty in the conditional estimates preclude excluding potentially meaningful treatment effect heterogeneity. These exploratory findings should not be interpreted as evidence of equivalent anthropometric responses across baseline BMI levels and require confirmation in trials specifically designed and adequately powered to evaluate effect modification.