Site-specific pain dynamics: associations between accelerometer-measured physical activity patterns and pain in older...

Physical activity (PA) has emerged as a promising non-pharmacological intervention for pain management, the relationship between objectively measured PA patterns and multi-site pain remains poorly understood. This exploratory study investigated associations between accelerometer-derived PA patterns and pain across various anatomical sites in older adults, and evaluated the predictive utility of machine learning model for pain outcomes.

This study utilized data from the National Health and Aging Trends Study in 2021-2022. Wrist-worn accelerometers measured PA, derived total activity, sedentary time, and activity/sedentary fragmentation) and time-frequency domain features. Cross-sectional and longitudinal analyses examined associations between PA patterns and site-specific pain using multivariable logistic regression with false discovery rate correction, while restricted cubic splines explored non-linear dose-response relationships. Random forest models with recursive feature elimination were developed to predict current pain status and pain relief.

Cross-sectional analysis indicated that moderate sedentary time was associated with back pain (OR = 2.24, 95%CI: 1.12-4.47) and neck pain (OR = 2.12, 95%CI: 1.07-4.20), while moderate-to-vigorous activity fragmentation was associated with lower leg pain prevalence (OR = 0.34, 95%CI: 0.15-0.78), and moderate sedentary fragmentation with foot pain (OR = 1.89, 95%CI: 1.07-3.33). Longitudinal analysis revealed that moderate-to-vigorous activity fragmentation was associated with pain persistence in head (OR = 0.21, 95%CI: 0.05-0.90), though associations did not survive FDR correction. Machine learning prediction models achieved performance for pain status (AUC: back = 0.56, wrist = 0.54) and pain relief prediction (AUC: wrist = 0.85, back = 0.72).

Site-specific tailoring of PA intensity and fragmentation is warranted for effective chronic pain management in older adults based on associations between PA and anatomical pain distribution.