Exploring Activity Patterns In Chronic Low Back Pain Patients Using Accelerometry-based Step Count Data

PURPOSE: Physical activity (PA) benefits people with chronic low back pain (cLBP), yet objective data on PA and pain experience are limited. This study aims to (1) characterize 24-hour PA patterns via accelerometry-derived step counts in individuals with cLBP, (2) examine associations between PA patterns and patient-reported outcomes (PROs), and (3) cluster individuals by PA patterns to identify potential cLBP phenotypes. METHODS: We analyzed data from the ongoing comeBACK study (NIH Back Pain Consortium-BACPAC, 1 U19AR076737-01), including 283 participants (mean age 56.8 years, 57.0% female). Participants wore an Actigraph GT3X+ sensor on the right hip for 7+ days. Step count data and 24-hour PA patterns were analyzed. Further individual clusters were identified using K-means clustering based on activity pattern distribution. Nonparametric comparisons of PA and PROs (PROMIS-29 Physical Function and Pain Interference) across clusters were performed, and Spearman correlation analysis examined associations between PA pattern distribution and PROs. RESULTS: Eight distinct 24-hour activity patterns were identified: Inactive (17.3%), Low PA (25.3%), Morning (9.8%), Mid-Day (20.5%), Evening (5.6%), All Day (6.4%), Bi-Phase (14.5%), and High PA (0.6%). Based on the distribution of these patterns, participants were grouped into five clusters: Mainly Inactive (N = 53), Mainly Low PA (N = 65), Mainly Bi-Phase (N = 38), Mainly Mid-Day (N = 68), and Various PA Patterns (N = 59). The Inactive cluster exhibited the lowest step count, poorest physical function, and highest pain interference compared to other clusters. In contrast, the Mid-Day and Various PA Pattern clusters showed above-average step counts, higher physical function, and lower pain interference. Correlation analysis indicated that the percentages of Inactive and Mid-Day patterns were consistently associated with various pain-related PROs. CONCLUSIONS: We identified distinct 24-hour physical activity patterns among individuals with cLBP and explored their relationship with patient-reported physical function and pain interference. These findings suggest that clustering based on PA patterns could help identify potential cLBP phenotypes, supporting future tailored physical activity interventions for cLBP management. Supported by: National Institute Of Arthritis And Musculoskeletal And Skin Diseases of the National Institutes of Health under Award Number U19AR076737, The Back Pain Consortium (BACPAC) Research Program