abstract
- OBJECTIVES: No clinical prediction rules were found for estimating the likelihood of developing incident radiographic tibiofemoral osteoarthritis (OA) with rapid progression. Such a tool would enhance prognostic capability for clinicians and researchers. DESIGN: We used two longitudinal datasets to independently derive (Multicenter Osteoarthritis Study) and validate (Osteoarthritis Initiative) a prognostic clinical prediction rule for estimating the probability of incident rapidly progressing radiographic knee OA in the following 4-5 years. Eligible subjects had at least one knee with a Kellgren and Lawrence (K&L) graded tibiofemoral joint of 0 or 1. Several potential risk factors were examined including obesity, age, knee alignment, frequent knee symptoms, contralateral knee OA and knee injury history. Multiple logistic regression was used to identify significant predictors and area under the receiver operating characteristic curve (AUC) was used to assess discrimination. RESULTS: A total of 1690 subjects participated in the derivation and 2422 subjects participated in the validation of the clinical prediction rule. The multivariable model displayed good discrimination with AUC of 0.79 in the derivation dataset and 0.81 in the validation dataset. CONCLUSIONS: Persons with contralateral knee OA, a baseline index knee OA grade of 1, higher body mass index (BMI) and higher baseline Western Ontario and McMaster Universities arthritis index total scores were more likely to develop K&L grade of 3 or 4 within 5 years. Frequent knee symptoms at baseline were not a significant predictor. The prediction rule and nomogram can assist clinicians in estimating the probability of rapidly progressing radiographic knee OA and the nomogram can assist researchers conducting epidemiologic studies and clinical trials.