Determination of Pain Phenotypes in Knee Osteoarthritis Using Latent Profile Analysis

Abstract

Abstract Objective To identify clinical phenotypes of knee osteoarthritis (OA) using measures from the following domains: 1) multimorbidity; 2) psychological distress; 3) pain sensitivity; and 4) knee impairment or pathology. Design Data were collected from 152 people with knee OA and from 31 pain-free individuals. In participants with knee OA, latent profile analysis (LPA) was applied to the following measures: normalized knee extensor strength, Functional Comorbidity Index (FCI), Pain Catastrophizing Scale (PCS), and local (knee) pressure pain threshold. Comparisons were performed between empirically derived phenotypes from the LPA and healthy older adults on these measures. Comparisons were also made between pheonotypes on pain intensity, functional measures, use of health care, and history of knee injury. Results LPA resulted in a four-group solution. Compared with all other groups, group 1 (9% of the study population) had higher FCI scores. Group 2 (63%) had elevated pain sensitivity and quadriceps weakness relative to group 4 and healthy older adults. Group 3 (11%) had higher PCS scores than all other groups. Group 4 (17%) had greater leg strength, except relative to healthy older adults, and reduced pain sensitivity relative to all groups. Groups 1 and 3 demonstrated higher pain and worse function than other groups, and group 4 had higher rates of knee injury. Conclusion Four phenotypes of knee OA were identified using psychological factors, comorbidity status, pain sensitivity, and leg strength. Follow-up analyses supported the replicability of this phenotype structure, but future research is needed to determine its usefulness in knee OA care.