Development of a Prognostic Nomogram for Predicting the Probability of Nonresponse to Total Knee Arthroplasty 1 Year After Surgery

Abstract

Background: Indications for total knee arthroplasty (TKA) currently depend on clinical judgment. Up to one fifth of those who undergo primary TKA do not report a clinically meaningful improvement in pain and function after surgery. Our aim was to develop and internally validate a prognostic tool for predicting the probability of nonresponse to surgery at 12 months.

Methods: Patients from 1 center who underwent primary TKA (N = 615) between 2012 and 2013. The Western Ontario and McMaster Universities Arthritis Index was collected pre- and 12 months after TKA from which nonresponse to surgery was determined using the Outcome Measures in Rheumatology-Osteoarthritis Research Society International responder criteria. Using independent prognostic correlates of postoperative nonresponse observed in adjusted modeling, we derived a prognostic nomogram to estimate the probability of nonresponse to TKA based on this suite of explanatory variables.

Results: A total of 90/615 (15%) cases were nonresponders to TKA. The degree of contribution (odds ratio, 95% confidence interval) of each explanatory factor to nonresponse nomogram points was body mass index ≥40 kg/m(2) (3.48; 1.97-6.12), Kellgren and Lawrence <4 (2.59; 1.58-4.24), mental disability on Short Form Health Survey (SF-12) mental component score (3.30; 1.44-7.58), and every 10-point increase in preoperative Western Ontario and McMaster Universities Arthritis Index score (0.81; 0.68-0.97). The concordance index for this model was 0.74.

Conclusion: We have created a prognostic nomogram that displays the predictive probabilities of nonresponse to TKA as a source of decision support for clinicians and patients, about their likely functional outcome from TKA. Although our own internal validation suggested good nomogram performance, external validation in a comparable surgical population is required to confirm generalizability of the nomogram.

Keywords: decision support; nomogram; patient-reported outcomes; risk prediction; total knee arthroplasty.