A systematic review of microsimulation models for skin cancer.

BACKGROUND

Simulation modelling can assist with health care decision making. To inform the development and improvement of skin cancer focussed microsimulation models, we conducted a systematic review and narrative synthesis of published skin cancer models to assess the structure, parameterisation, and assumptions.

METHODS

The electronic databases OVIDMedline including Embase and the Cost-Effectiveness Analysis (CEA) Registry were searched up to 7 May 2025. Studies that included microsimulation of individuals who developed or had skin cancer were eligible for inclusion. No restrictions on publication date or language were applied. The outcomes of interest were the purpose of the models, characteristics of the models and applicability for modelling skin cancer screening.

RESULTS

Twenty-two models were identified from the systematic review. Nineteen papers modelled melanoma, and two papers modelled keratinocyte or non-melanoma skin cancer, and one paper modelled both melanoma and keratinocyte cancer. The models were developed to assess treatment strategies (n = 10), skin cancer screening programs (n = 7), diagnostic techniques (n = 3), post-diagnosis surveillance (n = 3), preventative interventions (n = 1) and time to treatment (n = 1), with three models reporting dual aims. There was substantial variation in the simulation of the natural history of melanoma between models, with more recent models having separate natural history and clinical modules. The quality was assessed using the Quality Assessment Reporting for Microsimulation Models (QARMM) checklist and the majority of models were assessed to be of moderate quality. Limitations from these models included assuming an average tumour behaviour and constant melanoma development and progression over time. Data availability was also noted as a limitation for some models.

CONCLUSIONS

Most microsimulation models related to skin cancer have focused on late-stage treatment strategies. Tumour characteristics, apart from stage at diagnosis, were not accounted for in most models.

PROSPERO REGISTRATION NUMBER

CRD42024504250.