For patients seeking care for osteoarthritis (OA), psychological and behavioral characteristics are important contributors to physical impairment and pain. Clinical practice guidelines now recommend a “whole person” approach to OA management that addresses the biological, psychological, behavioral, and social needs of the patient. This approach requires the development and implementation of new interdisciplinary models of care.  We developed the Duke Joint Health Program (JHP) in 2017 to address this need within the Duke University Health System (DUHS). In the JHP, physical therapists serve as Primary Osteoarthritis Providers (POPs).  POPs are trained in psychologically-informed practice and are responsible for directing a patient’s care across the JHP provider network that includes psychologists, nutritionists, weight management experts, and orthopedists. A key responsibility of the POP is to identify psychological and behavioral needs that may affect rehabilitation. The POP then directly addresses those needs through treatment, or initiating      referral to the appropriate provider. Because there are no standard practices for comprehensive psychological and behavioral screening in OA, we developed our own screening process with a goal of designing a process that was feasible for implementation at scale.

We used  the 10-item OSPRO Yellow Flag (OSPRO-YF) Assessment Tool      which measures 11 psychological constructs related to negative mood (e.g., depression, anxiety), fear avoidance (e.g., kinesiophobia) and negative pain coping (e.g., low self-efficacy).  Prior to this project, there was limited guidance on how to integrate such a tool into practice to inform decision-making for treatment or referral. Working with a team of physical therapists, pain researchers, and orthopedists, we determined that using the tool to develop psychological phenotypes would allow us to identify common psychological profiles on which to develop recommended treatment pathways. We used a statistical method, latent class analysis, to identify how the 11 psychological factors clustered among patients in the program. We found 4 distinct phenotypes: high distress, low distress, negative pain coping, and low self-efficacy and acceptance. The highest prevalence phenotype was high distress, which characterized 52% of the JHP sample. Using published best practice recommendations for psychologically-informed practice, our team of physical therapists, psychologists, orthopedic surgeons, and researchers then iteratively designed tailored treatment recommendations to target the specific psychological and behavioral needs of each phenotype. We integrated recommendations into smart phrases within Epic to aid with clinical implementation of recommendations.

We are currently piloting the use of this phenotype-specific treatment model in the JHP. At the beginning of the pilot phase, we trained clinicians on how to identify phenotypes with OSPRO scores and on the recommended treatments for each phenotype. We experienced challenges in coming to a consensus on which treatments to prioritize for which phenotype. We were also challenged to find effective methods for treatment documentation that could be exported and analyzed. Ongoing and future work involves scaling up multidisciplinary resources and assessing program outcomes.