Developing Reporting Standards for Population Health Microsimulation: A Scoping Review of Current Practices

Background

Population health simulation models—including microsimulation, agent-based, system dynamics, and Markov models—are essential tools for understanding noncommunicable disease (NCD) burden and evaluating policy interventions. However, inconsistent reporting practices limit the transparency, reproducibility, and credibility of this work. Unlike clinical trials and observational studies, which benefit from established reporting guidelines (CONSORT, STROBE), no comprehensive standards exist for population health simulation models.

The POPCORN initiative (Population Health Modelling Consensus Reporting Network) aims to address this gap by developing the first EQUATOR Network reporting guideline specifically for population health NCD models. Following EQUATOR methodology, guideline development requires three phases: (1) a scoping review to identify current reporting practices and gaps, (2) international Delphi consensus to prioritise reporting items, and (3) pilot testing with modellers and journal editors. This abstract presents findings from the first phase.

Methods

We conducted a scoping review following the Arksey and OMalley framework with Joanna Briggs Institute guidance. Scoping reviews systematically map the evidence base to identify key concepts, gaps, and research needs—making them ideal for informing guideline development. We searched MEDLINE, Embase, Scopus, and Global Health for studies published July–December 2025.

Eligible studies were computational simulation models examining population-level outcomes for eight NCD groups (cardiovascular, cancer, diabetes, respiratory, mental health, neurological, musculoskeletal, injury) or six risk factors (tobacco, diet, physical activity, alcohol, obesity, hypertension). We applied PICOSIM criteria adapted for simulation studies, covering population, intervention/comparator, outcome, study approach, integration of data sources, and model adaptability.

Given the high volume of literature, we implemented AI-assisted screening using large language models with retrieval-augmented generation to support title/abstract review alongside five human reviewers. Data extraction captured reporting elements across model structure, inputs, validation, and outputs.

Results

From 8,474 records, deduplication yielded 6,427 unique citations. Our database search identified 90% of studies from a reference set of 65 known eligible studies. AI-assisted screening achieved 90% sensitivity and 95% specificity compared to human consensus (κ=0.69), missing only 2 of 20 eligible studies in the validation set.

Identifying simulation models proved challenging: key discriminating terms were absent from nearly 10% of titles and abstracts. Preliminary extraction revealed systematic reporting gaps in model specification, parameter uncertainty quantification, and validation procedures—precisely the areas where standardised guidance could improve practice.

Conclusions

This scoping review establishes the evidence base for POPCORN reporting guideline development. Findings confirm substantial variation in how population health models are reported, supporting the need for consensus-based standards. The microsimulation community will play a central role in the upcoming Delphi process, and we welcome collaborators interested in shaping guidelines that serve both modellers and the policy audiences who rely on their work.