Presentation (CTI)

OBJECTIVES: Excel remains the dominant platform for health economic models in HTA, often relying on extensive formulas, complex workbook structures, and VBA. Recent advances in “modern Excel”—including dynamic arrays and functions such as LET, LAMBDA, REDUCE, and MAKEARRAY—enable more compact and programmatic model implementations. Paulden (2026) demonstrated cohort Markov models without VBA, while Blissett (2026) extended this to patient-level simulation. These developments provide an opportunity to improve model design and efficiency. Objectives were to understand how to incorporate modern Excel approaches into HTA models to support transparency, reproducibility, and validation.
METHODS: A published cost-effectiveness model was replicated in Excel using a dynamic-array-based approach informed by Paulden (2026). Core components—including transition matrices, Markov traces, cost and QALY calculations, PSA and CEAC were implemented without VBA. Qualitative assessments were made on development effort, common sources of error, and approaches to identifying and resolving issues as well as transparency, ease of validation, reproducibility, and accessibility of intermediate calculations.
RESULTS: The model successfully reproduced key results using a VBA-free dynamic array structure. The approach reduced formula duplication and improved consistency but required greater upfront design effort. Errors were more likely to arise from complex array logic than from cell-referencing issues, though these could be identified through structured checks and use of intermediate variables. Transparency depended on how intermediate results were exposed, and reliance on newer Excel versions introduced compatibility constraints. Reproducibility of PSA required fixed random inputs.
CONCLUSIONS: Modern Excel functions can enhance efficiency and consistency in HTA models but introduce challenges for transparency and accessibility. A hybrid approach using modern functions alongside explicit outputs, documentation, and validation checks offers a pragmatic solution that aligns with HTA requirements. Continued research into Excel integration with AI/Python may further support HTAs but requires careful consideration of implications for auditability and decision-making acceptability.