Incidence, related risk factors, imaging characteristics analysis and prediction model of radiographic patellofemoral osteoarthritis in a Chinese suburban area: Shunyi osteoarthritis study

Abstract

Purpose: The incidence of radiographic patellofemoral osteoarthritis(RPFOA) remains unclear, few studies have reported the prevalence of RPFOA before, no research has performed a prediction model of RPFOA worldwide and few studies have discussed the imaging characteristic of PFOA. This study aims to describe the incidence of RPFOA in a Chinese suburban area, examine the risk factors for RPFOA incidence and establish a prediction model of RPFOAin a populationbased longitudinal study,a simple analysis of CT imaging features of new onset population is made simultaneously. Methods: Shunyi Osteoarthritis Study was a population-based, longitudinal and prospective study of knee osteoarthritis in a Chinese suburban area. 1295 residents aged over 50 years were recruited by randomized cluster sampling with fully informed in 2014 and were followed 3 years later. At the time of baseline and follow-up visits, subjects completed a home interview (including socio-demographic factors, history of knee injury, joint symptoms, job-related physical activity and work history) and received a clinical examination including height, weight, range of motion(ROM), chair stand test, 50 feet walk test, and weight-bearing posterior-anterior semi-flexed view of radiographs at tibiofemoral(TF) joints and skyline view of radiographs at patellofemoral(PF) joints included. We defined a subject as having patellofemoral OA on skyline view when (1) the osteophyte score was 2 ;(2) the JSN score was 2 and the osteophyte score was 1;(3) the JSN score was 3. For each batch of Shunyi PFOA Study films(n¼100), 20 films from the 2014 Shunyi study were added to test inter-reader reliability. In addition, 20 previously read PF radiographs from the 2017 Shunyi study were fed back to test intra-reader reliability. None-weighted Kappa statistics for inter-reader and intra-reader reliability were 0.82 and 0.85 respectively. Multiple logistic regression was used to examine all the factors at baseline to the risk of incident RPFOA. Multivariate logistic regression was used to estimate the coefficients of each risk factors. The continuous variable was cut into groups, the median value of the group was used as a reference value. The distance(D) from the baseline to each risk factor was calculated by multipling the regression coefficient of the age and the group space(10). Set the unit distance B of 1 point, score the Point¼D/B of each risk factor. The probability(P) can be calculated by the logistic probability calculation formula. Results: A total of 1295 participants (2590 knees) were recruited at baseline in 2013, and 967(74.7%) residents were followed in 2017. The rest of 328(25.3%) residents were excluded for the loss to follow-up due to loss of contact, death, rejection, move out, etc. Of all the knees(n¼1537) without RPFOA at baseline,139 knees (9.04%) developed incident RPFOA. That is to say, the incidence of RPFOA in knee level is annually 3.01% in this study. Different factors were tested for the relationship with incident RPFOA. Advanced age(OR¼1.037, 95% C...