Demographic, Clinical, and Psychosocial Predictors of Exercise Adherence: The STRRIDE Trials.

Abstract

Purpose: To identify baseline demographic, clinical, and psychosocial predictors of exercise intervention adherence in the Studies of a Targeted Risk Reduction Intervention through Defined Exercise (STRRIDE) trials.

Methods: A total of 947 adults with dyslipidemia or prediabetes were enrolled into an inactive control group or one of ten exercise interventions with doses of 10-23 kcal/kg/week, intensities of 40-80% of peak oxygen consumption, and training for 6-8-months. Two groups included resistance training. Mean percent aerobic and resistance adherence were calculated as the amount completed divided by the prescribed weekly minutes or total sets of exercise times 100, respectively. Thirty-eight clinical, demographic, and psychosocial measures were considered for three separate models: 1) clinical + demographic factors, 2) psychosocial factors, and 3) all measures. A backward bootstrapped variable selection algorithm and multiple regressions were performed for each model.

Results: In the clinical and demographic measures model (n=947), variables explained 16.7% of the variance in adherence (p<0.001); lesser fasting glucose explained the greatest amount of variance (partial R² = 3.2%). In the psychosocial factors model (n=561), variables explained 19.3% of the variance in adherence (p<0.001); greater 36-Item Short Form Health Survey (SF-36) physical component score explained the greatest amount of variance (partial R² = 8.7%). In the model with all clinical, demographic, and psychosocial measures (n=561), variables explained 22.1% of the variance (p<0.001); greater SF-36 physical component score explained the greatest amount of variance (partial R² = 8.9%). SF-36 physical component score was the only variable to account for >5% of the variance in adherence in any of the models.

Conclusions: Baseline demographic, clinical, and psychosocial variables explain approximately 22% of the variance in exercise adherence. The limited variance explained suggests future research should investigate additional measures to better identify participants who are at risk for poor exercise intervention adherence.