The value of total knee replacement revisited: A comparative analysis of its effectiveness and impact on healthcare costs using data from the osteoarthritis initiative

Abstract

Purpose: Total knee replacement (TKR) utilization in the U.S. has more than doubled since 2000, primarily due to expanding eligibility to less symptomatic knee osteoarthritis (OA) patients. Yet the benefit and costs of TKR have not been evaluated for a contemporary U.S. patient population within a comparative analysis. We evaluated the impact of TKR on quality of life, lifetime costs and quality-adjusted life years (QALYs) while varying TKR utilization by preoperative functional status. Methods: Data for these analyses are from the Osteoarthritis Initiative (OAI) public use data set(s) and Multicenter Osteoarthritis Study (MOST). We estimated the effect of TKR on generic (Short Form [SF] 12 score), and OA-specific quality of life measures among 1,327 OAI participants with knee OA, aged 45 to 79 (median age 61), and 9 years of follow-up. Marginal structural models were used. Effects were validated in 965 knee OA patients in the MOST cohort. We subsequently modeled the virtual life courses of OAI patients until age 100 using microsimulation, tracking occurrence of primary TKR, revision TKR, and death. We calculated QALYs and costs from a U.S. health system perspective and compared scenarios ranging from current TKR practice, to TKR practice limited to those with severe functional status, to no TKR. Cost-effectiveness thresholds of $100,000 and $200,000 per QALY were considered for decision-making. We used 500 bootstrap equations 95% to calculate confidence intervals (CIs). Cost of TKR was varied in deterministic sensitivity analyses. Results: In both OAI and MOST, quality of life outcomes generally improved after TKR, with small effect sizes becoming larger with decreasing preoperative functional status. After adjustment for baseline and time-varying confounders, TKR had a marginal effect of 1.70 (95% CI 0.26 to 3.57) on the SF-12 PCS, and changed SF-12 MCS by −0.22 (95% CI -1.49 to 1.31). For OA-specific measures of quality-of-life, TKR improved WOMAC score by 10.69 (95% CI 8.01 to 13.39) and KOOS by 9.16 (95% CI 6.35 to 12.49) points. TKR reduced the odds of OA pain medication use, but this effect was uncertain: 0.81 (95% CI 0.55 to 1.12). In the scenario of current TKR practice, the lifetime risk of undergoing TKR, as predicted for OAI participants with knee OA, was 40.2% (95% CI 35.1 to 44.1). The average QALYs were 11.93 and the average costs were $24,982 per patient. None of the scenarios’ incremental cost-effectiveness ratios (ICERs) were below $100,000/QALY. Allocating TKR to patients with SF-12 PCS scores below 35 was the optimal scenario given a cost-effectiveness threshold of $200,000/QALY, gaining 0.0065 QALYs against $1,126 additional costs as compared to the previous best scenario resulting in an ICER of $172,338/QALY. The current TKR practice scenario was dominated (see Figure). TKR would become cost-effective when performed in patients with SF-12 PCS scores up to 40 if the hospitalization costs fell below $12,000. Conclusions: TKR practice in a contemporary U.S. knee OA patient population had minimal effects on quality-of-life and was economically unattractive. However, if TKR practice were restricted to patients with more severe functional status, effect size would rise and the procedure may be considered to be cost-effective.