Characterizing Extreme Phenotypes for Perceived Improvement From Treatment in Persons With Chronic Pain Following Traumatic Brain Injury: A NIDILRR and VA TBI Model Systems Collaborative Project

Abstract

Objective: To define and characterize extreme phenotypes based on perceived improvement in pain for persons with chronic pain following traumatic brain injury (TBI). Setting: Eighteen Traumatic Brain Injury Model System (TBIMS) Centers. Participants: A total of 1762 TBIMS participants 1 to 30 years post-injury reporting chronic pain at their most recent follow-up interview. Primary Measures: The Patient's Global Impression of Change (PGIC) related to pain treatment. Sociodemographic, injury, functional outcome, pain, and pain treatment characteristics. Results: Participants were mostly male (73%), White (75%), middle-aged (mean 46 years), injured in motor vehicle accidents (53%), or falls (20%). Extreme phenotypes were created for an extreme improvement phenotype (n = 512, 29.8%) defined as “moderately better” or above on the PGIC and an extreme no-change group (n = 290, 16.9%) defined as no change or worse. Least absolute shrinkage and selection operator (LASSO) regression combined with logistic regression identified multivariable predictors of improvement versus no-change extreme phenotypes. Higher odds of extreme improvement phenotype were significantly associated with being female (odds ratio [OR] = 1.85), married versus single (OR = 2.02), better motor function (OR = 1.03), lower pain intensity (OR = 0.78), and less frequent pain, especially chest pain (OR = 0.36). Several pain treatments were associated with higher odds of being in the extreme improvement versus no-change phenotypes including pain medication (OR = 1.85), physical therapy (OR = 1.51), yoga (OR = 1.61), home exercise program (OR = 1.07), and massage (OR = 1.69). Conclusion: Investigation of extreme phenotypes based on perceived improvement with pain treatment highlights the ability to identify characteristics of individuals based on pain treatment responsiveness. A better understanding of the biopsychosocial characteristics of those who respond and do not respond to pain treatments received may help inform better surveillance, monitoring, and treatment. With further research, the identification of risk factors (such as pain intensity and frequency) for treatment response/nonresponse may provide indicators to prompt changes in care for individuals with chronic pain after TBI.