Beyond clinical predictors: Determining the impact of psychosocial factors in chronic pain after spinal cord injury.

Abstract

▒: Although chronic pain is common after traumatic SCI, prognostic models have traditionally prioritized clinical injury characteristics, and the relative contribution of sociodemographic and psychological factors remains under-characterized. Secondary analysis of a national longitudinal cohort from United States inpatient rehabilitation centers participating in the Spinal Cord Injury Model Systems (SCIMS). Adults with traumatic SCI (N = 1,303) completed SCIMS Form 1 and a 1-year follow-up Form 2 interview. Pain intensity at 1 year was assessed using the Numeric Rating Scale (0-10) for the prior 4 weeks (mean 4.3 [SD 2.8]; 85.4% reported pain). Hierarchical linear and quantile regression models were used to evaluate the incremental predictive value of biopsychosocial domains, alongside elastic net regression and Boruta feature selection to identify robust predictors. Sequential addition of biopsychosocial domains improved model fit across hierarchical regressions, regardless of entry order (full model R_Adjusted² ≈ 12%). Social/demographic variables accounted for the largest share of predictive signal and were consistently prioritized over clinical factors and psychological measures. EN regression selected 43 predictors across domains, while Boruta confirmed 27. Six predictors were consistently identified across all methods: median household income, race, education, tobacco use, primary payer, and depression severity. Sociodemographic factors accounted for most predictive value for chronic pain after SCI, often outperforming traditional clinical and psychological measures. Converging results across methods support the biopsychosocial model and highlight potential targets for screening and intervention. Prospective studies are needed to clarify causal pathways and inform pain prevention efforts.