Predictive value of digital radiographic characteristics in combination with clinical evaluation scores in assessing the need for knee replacement

Abstract

To investigate the possibility of a combination of knee digital radiography (DR) characteristics and pain numerical rating scale (NRS) scores in predicting the need for knee replacement (KR). This follow-up study was approved by the local medical ethics review boards, and informed consent was obtained from the participants. Knee osteoarthritis (KOA) patients who underwent KR (n = 106; 90 females) and those who did not (n = 103; 78 females) along with normal controls (n = 104; 81 females) were followed up from December 2016 to March 2018. The minimum knee joint space width (MJSW) on radiographs, Kellgren–Lawrence (KL) grade, NRS scores, age at operation, and body weight were statistically analysed by binary logistic regression, ROC curves, and Kaplan–Meier analysis. Measurement reliability was assessed by using intraclass correlation coefficient (ICC) values. The results showed a high reliability (ICC > 0.98). NRS and MJSW showed a higher value for predicting KR (odds ratios 2.76 and 2.21, respectively). The combination of MJSW ≤ 2.08 mm and NRS score > 3.75 had better predictive value than either index alone. The MJSW, an imaging characteristic derived from DR, and NRS scores have some predictive value for determining the need for KR in patients with KOA.