Preoperative prognostic factors for residual pain following total joint arthroplasty: a systematic review and meta-analysis of prospective and retrospective studies

Background Total joint arthroplasty (TJA) reduces pain in patients with end-stage osteoarthritis, but many still experience residual pain after one year. We aimed to identify preoperative prognostic factors for residual pain and describe their relative strength. Methods We systematically searched Embase, MEDLINE, Web of Science, and the Cochrane Central Register of Controlled Trials from inception to 12 December 2023 and screened the 200 top-ranked Google Scholar records. Eligible studies involved patients undergoing primary TJA for osteoarthritis and comprised prospective or retrospective cohort studies, case-control studies, and prognostic analyses of randomised trials developing or updating an internally or externally validated multivariable prognostic model. Associations were converted to a common correlation scale and pooled using random-effects meta-analysis. Prediction intervals, certainty of evidence, and risk of bias were assessed. Results Fifty-nine studies (335’921 patients) addressed hip, knee, or shoulder arthroplasty; no eligible studies addressed other prespecified joints. Across joints, poorer mental health (mean correlation coefficient − 0.11; 95%-CI -0.13 to -0.09; 95%-PI -0.15 to -0.07, 9 studies), higher pain levels (0.18; 95%-CI 0.14 to 0.23; 95%-PI -0.04 to 0.42, 35 studies), and more comorbidities (0.08; 95%-CI 0.04 to 0.13; 95%-PI -0.10 to 0.27, 17 studies) were associated with residual pain (moderate- to high-certainty evidence). Age and sex showed near-zero associations with low-certainty evidence. In knee TJA, lower osteoarthritis severity (-0.16; 95%-CI -0.28 to -0.04; 95%-PI -0.42 to 0.09, 4 studies), greater anxiety (0.10; 95%-CI 0.02 to 0.18; 95%-PI -0.05 to 0.25, 6 studies), and pain catastrophising (0.23; 95%-CI 0.11 to 0.35; 95%-PI -0.08 to 0.56, 7 studies) showed additional pooled associations with uncertain consistency. In hip TJA, preoperative pain showed a small pooled association with uncertain consistency. REML–Hartung–Knapp sensitivity analysis produced similar estimates, but 14 of 35 associations lost statistical significance, including osteoarthritis severity and anxiety in knee TJA. Conclusions These findings support comprehensive preoperative assessment, counselling, and future prognostic model development that consider mental health, pain burden and comorbidities across joint groups. They also support evaluating additional factors, including osteoarthritis severity, anxiety and pain catastrophising in knee TJA patients. PROSPERO registration number CRD42023393334. Clinical trial number Not applicable.