Background/Purpose:  A better understanding of the factors associated with structural progression of knee osteoarthritis (OA) may help identify individuals not only at risk for more rapid OA progression who can benefit from early intervention, but also help classify high-risk OA patients for inclusion in clinical trials of new treatments. Using the complete 4-year follow-up data from the Osteoarthritis Initiative (OAI), we tested baseline measures from six domains [1) knee examination, 2) anthropometric characteristics, 3) sociodemographic characteristics, 4) medical and family history, 5) physical functioning, and 6) self-reported pain and symptoms] for their association with radiographic knee OA (RKOA) progression.

Methods:  Subjects with evidence of RKOA in one or both knees at baseline and with at least 1 follow-up exam over the 4-year period (n = 3,204) were eligible for analysis; knees with end-stage disease were excluded from the analysis. We defined RKOA progression as an increase in Kellgren-Lawrence (KL) grade or osteophyte score or joint space narrowing score. We used Cox regression models to test the association of individual level measures with RKOA progression, adjusting for age, gender, and race. Within each of the six domains, we performed univariate and multivariable stepwise analyses. Significant predictors from all domain-level multivariable models were then included in an overall multivariable analysis with stepwise selection. We also conducted knee-specific analyses with the same approach, but included a robust sandwich estimate of the covariance matrix to account for non-independence of two knees within an individual. 

Results:  51% of subjects with baseline RKOA (KL grade 1-3) experienced RKOA progression in one or both knees. Progressors were 58% female, 19% Black, and had mean (SD) age of 61 (9) years and mean body mass index of 29 (5) kg/m². Subjects with no RKOA progression had a similar profile. In the multivariable model for each domain, the following baseline factors were found to be significantly associated with RKOA progression: knee joint effusion, pain on flexion, patellofemoral crepitus, patellar grind, and flexion contracture; number of hand bony enlargements, history of knee injury, and use of analgesics; repeated chair stand pace, 400-meter walk pace, and pain during 400-meter walk; KOOS sports and recreational activities score and KOOS symptoms score. Similar results were obtained with the knee-specific analyses with the additional significant finding of KOOS pain score. Final, overall multivariable models for the individual and knee-specific analyses are presented in Table 1.

Conclusion:  Out of the six domains investigated, four domains yielded significant measures associated with RKOA progression. These measures may help identify individuals at increased risk of RKOA progression who may benefit from early intervention. 

Table 1. Hazard Ratios for Associations with RKOA Progression