From Spreadsheets to Script: Experiences From Converting a Scottish Cardiovascular Disease Policy Model into R

Area of Science:

• Health economics
• Computational modeling
• Software development

Background:

• Growing interest in converting spreadsheet-based health economic models to R for enhanced transparency, reproducibility, and efficiency.
• Need for upskilling health economic modellers in R for broader model re-use.

Purpose of the Study:

• To convert the Scottish Cardiovascular Disease (CVD) Policy Model from Excel to R.
• To document lessons learned during the conversion process.
• To assess the feasibility and outcomes of using R for health economic modeling.

Main Methods:

• Replication of an existing Excel-based competing risk state transition cohort model in R by four health economists with varying R experience.
• Development of two R versions of the CVD model: one without specialist health economics packages and one using the 'heemod' package.
• Comparison of R model results with the original Excel model and assessment of computational efficiency.

Main Results:

• Conversion of the Excel CVD model to R is feasible, with the R version requiring less than half the computational time.
• The R model developed without specialist packages successfully reproduced identical results to the Excel version.
• The R model using the 'heemod' package did not achieve identical results, potentially due to model complexity and package limitations.

Conclusions:

• Coding health economic models in R is achievable for health economists, offering significant advantages in transparency and efficiency.
• Using R without specialist packages provides full flexibility and ensures result fidelity, while specialist packages offer convenience for suitable model structures.
• Transparency and replicability are paramount in model programming, regardless of the software or approach chosen for model conversion.