Patient-level simulation models in cancer care: a systematic review.

Background: Patient-level simulation (PLS) models overcome some major limitations of conventional cohort models and have broad applicability in healthcare, yet limited knowledge exists about their potential in cancer care. Objectives: This systematic review aims to: (1) describe the application areas of PLS models in cancer care, (2) identify commonly used model structures, (3) evaluate the quality of reporting based on established guidelines, and (4) critically discuss the potential and limitations of PLS models in this context. Methods: A systematic literature search was completed in Web of Science, PubMed, EMBASE and EconLit. Reasons underlying the use of PLS models were identified with a conventional inductive content analysis and reporting quality was assessed with an 18-item checklist based on the ISPOR-SMDM guidelines. Results: The number of publications increased over time and most studies used state-transition microsimulation (49.25%) or discrete event simulation (48.51%). Two main application areas could be discerned, namely disease progression modelling (DPM) (78.36%) and health and care systems operation (HCSO) (21.64%). In the DPM domain, the use of PLS models was mainly motivated by the need to represent patient heterogeneity and history. In the HCSO domain, PLS models were used to better understand and improve cancer care delivery. Average reporting quality was 65.2% and did not improve over time. Conclusion: PLS models can be used to simulate the progression of cancer and to model cancer care delivery. In the DPM domain more direct comparisons with cohort models are required to establish the relative advantages of PLS models and in the HCSO domain the impact on clinical practice needs to be systematically assessed. Furthermore, adherence to the ISPOR-SMDM guidelines should be improved.