Background: Pharmacological treatments have transformed the management of rheumatic and musculoskeletal conditions (RMDs). However, these medications can cause a wide range of adverse events (AEs), which can impact patients’ quality of life and, in severe cases, lead to hospitalizations or even death. The likelyhood of AEs often determines the next selected pharmacological treatment as part of shared decision making between patients and health care professionals. Clinical prediction models (CPMs) are promising statistical tools that can help identify an individual’s risk of experiencing AEs. Despite their potential, the use of CPMs in clinical practice remains limited.

Objectives: This systematic review aims to identify, summarise, and assess the methodological quality of developed CPMs to predict the risk of AEs associated with RMD medication.

Methods: We conducted a search in the PubMed, Embase, and Medline databases for relevant studies published from their inception up to March 2024. In addition, both forward and backward techniques were employed. The selected studies were to have developed at least one CPM that predicts AEs in adult patients using approved medications for their RMDs. Exclusion criteria included studies focused on illicit/recreational drugs, randomised controlled trials, systematic reviews, and meta-analyses. Three reviewers independently screened titles and abstracts, followed by full-text review. Conflicts were resolved by a fourth reviewer. Data extraction and quality assessment were performed using the Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS) and Prediction Model Risk Of Bias Assessment Tool (PROBAST) checklists. This approach ensured thorough data reporting and assessment of the risk of bias (RoB).

Results: Out of a total of 2,406 studies, 1,734 had their titles and abstracts screened, and 38 studies were screened in full. A total of 12 studies with 17 developed models. The majority of the CPMs (76.4%), focused on rheumatoid arthritis, which is primarily treated with disease-modifying anti-rheumatic drugs (DMARDs) such as methotrexate (69.2%) and, biologic drugs (15.3%). The most commonly used statistical models were the Cox proportional hazards and logistic regression. The overall RoB was high in 70.5% of the studies, primarily due to inappropriate variable selection methods and inadequate sample sizes. Additionally, almost all of the included models (94.1%) focused on the prediction of a single AE, without considering the potential influence of one AE on another.

Conclusion: This review is the first comprehensive analysis summarising CPMs that predict AEs related to medications for RMDs. It underscores several methodological shortcomings in existing CPMs, such as inappropriate variable selection and insufficient justification for sample sizes. Future models should aim to encompass a broader spectrum of RMDs and associated medications. Employing techniques such as machine learning, which can handle complex interactions alongside multi-outcome CPMs designed to predict multiple AEs, has the potential to significantly enhance prediction accuracy.

REFERENCES: NIL.

Acknowledgements: NIL.

Disclosure of Interests: None declared.

© The Authors 2025. This abstract is an open access article published in Annals of Rheumatic Diseases under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ ). Neither EULAR nor the publisher make any representation as to the accuracy of the content. The authors are solely responsible for the content in their abstract including accuracy of the facts, statements, results, conclusion, citing resources etc.