Latent Class Growth Analysis predicts long term pain and function trajectories in total knee arthroplasty: a study of 689 patients.

ABSTRACT

OBJECTIVE: To characterize groups of subjects according to their trajectory of knee pain and function over 1 to 5 years post total knee arthroplasty (TKA). METHODS: Patients from one centre who underwent primary TKA (N = 689) between 2006 and 2008. The Knee Society Score (KSS) was collected pre-operatively and annually post-operatively. Latent Class Growth Analysis (LCGA) was used to classify groups of subjects according to their trajectory of knee pain and function over 1-5 years post-surgery. RESULTS: LCGA identified a class of patients with persistent moderate knee pain (22.0%). Predictors (OR, 95% CI) of moderate pain trajectory class membership were pre-surgery SF12 mental component summary (MCS) per 10 points (0.65, 0.54-0.79) and physical component summary (PCS) per 10 points (0.50, 0.33-0.76), Charlson Comorbidity Index (CCI) one (1.70, 1.07-2.69) and ≥two (2.82, 1.59-4.81) and the absence of computer-navigation (2.26, 1.09-4.68). LCGA also identified a class of patients with poor function (23.0%). Predictors of low function trajectory class membership were, female sex (3.31, 1.95-5.63), advancing age per 10 years (2.27, 1.69-3.02), pre-surgery PCS per 10 points (0.50, 0.33-0.74), obesity (1.69, 1.05-2.72), morbid obesity (3.12, 1.55-6.27) and CCI ≥two (2.50, 1.41-4.42). CONCLUSIONS: Modifiable predictors of poor response to TKA included baseline co-morbidity, physical and mental well-being and obesity. This provides useful information for clinicians in terms of informing patients of the expected course of longer term outcomes of TKA and for developing prediction algorithms that identify patients in whom there is a high likelihood of poor surgical response.