Efficiency and cost-effectiveness of three surgical strategies for primary osteoarthritis of the elbow: a multicenter propensity score-weighted cohort study.

Abstract

Background: Primary osteoarthritis of the elbow causes progressive stiffness and pain, yet no study has directly compared the efficiency and cost-effectiveness of the three principal surgical strategies: total elbow arthroplasty (TEA), open osteocapsular arthroplasty (OCA), and arthroscopic OCA. We hypothesized that joint-preserving OCA would achieve greater motion recovery at lower cost than TEA.

Methods: This multicenter retrospective cohort study identified 542 patients with primary elbow OA from three Chinese Tertiary A hospitals (surgery January 2017-December 2023; minimum two year follow-up through December 2025): 142 TEA, 221 open OCA, and 179 arthroscopic OCA. Because propensity-score matching left substantial residual imbalance in age and baseline motion, primary inference was based on generalized overlap weighting, which achieved covariate balance and estimates the average treatment effect in the overlap population, that is, the sub-population in clinical equipoise [25, 40]. Matching with covariate-adjusted mixed models is reported as a supporting analysis. Co-primary outcomes were the percentage of lost motion recovered and the flexion-extension (FE) arc at two years. Because pain relief rather than motion is the principal indication for TEA, cost-effectiveness was additionally assessed on a pain axis. Willingness-to-pay thresholds were not applied, as gross-domestic-product-based thresholds are defined per quality-adjusted life year and are not commensurate with effects expressed in natural units [4, 27].

Results: Overlap weighting achieved acceptable balance (age standardized mean difference 0.11; baseline arc 0.08) with an effective sample size of 246 of 542 patients (45.4%; TEA 41, open OCA 144, arthroscopic OCA 62). In the overlap population, the percentage of lost motion recovered at 2 years was 15.8% (TEA), 35.9% (open OCA), and 32.2% (arthroscopic OCA), and the 2-year FE arc was 94.0°, 108.0°, and 104.5°. Both OCA procedures exceeded TEA by 16-20 percentage points of lost motion recovered, whereas the difference between the two OCA approaches was below the minimal clinically important difference. On the pain axis, visual analog scale reduction was 1.93, 2.44, and 2.77 points; all between-group differences (0.33-0.84) fell below the 1.4-point minimal clinically important difference, indicating clinically equivalent pain relief. Overlap-weighted total costs were ¥86,728 (TEA), ¥32,503 (open OCA), and ¥24,677 (arthroscopic OCA). TEA therefore cost approximately ¥62,000 more per patient than arthroscopic OCA without a clinically important advantage on either axis. Corrected complication rates in the TEA group were radiographic aseptic loosening 13.6%, loosening requiring revision 8.2%, and revision surgery 9.4%, consistent with published mid-term series.

Conclusions: Within the sub-population in clinical equipoise, joint-preserving OCA recovered more motion than TEA at substantially lower cost, while pain relief was clinically equivalent across all three strategies. These findings describe real-world treatment strategies applied to their usual indications and should not be read as evidence that TEA and OCA are interchangeable, nor as a general first-line recommendation: patients with end-stage ulnohumeral destruction, for whom TEA is indicated, are under-represented in the overlap population. For patients in whom both joint preservation and arthroplasty are genuinely being considered, OCA appears the more efficient choice.