Aims Predictive modelling studies are increasingly popular, but the reporting quality in developing and validating these models remains suboptimal. This review aimed to evaluate the methodological quality of predictive models for patient-reported outcomes following total hip arthroplasty (THA) and total knee arthroplasty (TKA), identifying gaps in reporting and biases. Methods The review followed PRISMA guidelines, appraising studies that developed and/or validated multivariate predictive models. Methodological quality was assessed using the Prediction model Risk Of Bias ASsessment Tool (PROBAST) tool, and reporting quality was evaluated using Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) guidelines. An electronic search was conducted across MEDLINE, EMBASE, and CINAHL up to 29 May 2025 and several studies from expert recommendation. Studies involving adults (aged ≥ 18 years) undergoing elective primary or revision THA or TKA were included, while univariate analyses and literature reviews were excluded. Results The search identified 6,194 results, with 3,793 unique articles. A total of 58 studies were screened, and 41 were included. TRIPOD compliance ranged from 58% to 68%. Overall, 98% of studies had a low risk of bias in participant selection, but 83% showed a high risk of bias in analysis. Applicability concerns were low in 93% of studies. Conclusion The review reveals significant methodological limitations in predictive models for THA and TKA outcomes, especially in analysis. Improving adherence to reporting guidelines is essential for enhancing transparency and reliability, ultimately supporting better clinical decision-making and patient outcomes. Cite this article: Bone Jt Open 2026;7(1):115–129.
Lin et al. (Fri,) studied this question.