Obesity, defined as a body mass index (BMI) greater than 30kg/m2, is a multifactorial disease with severe health and economic consequences. As obesity associated events impact long-term survival, health economic (HE) modelling is commonly used. However, the current set of modelling approaches are very diverse and lack external model validation. The aim of our research was to compare the event simulation and the HE outcomes of different structural approaches.

We performed an external validation of three main structural modelling approaches for estimating obesity-associated events: (i) continuous BMI-related risks; (ii) general risk equations; and (iii) categorical BMI-related risks. The Swedish Obese Subjects (SOS) intervention study was used for validation. Outcomes compared were mortality, cardiovascular events (CVE), and type 2 diabetes (T2D), over time using the long-term data from the SOS study, looking at both the surgery arm and the control arm. Concordance between modelling results and the external validation study was measured by visual examination of the best fitting linear regression lines, R2 coefficients, the root mean squared errors, and F-tests. Furthermore, we compared the HE modelling results, comparing surgery versus control, expressed as cost per quality-adjusted life-year (QALY) gained based on 1,000 Monte Carlo simulation samples.

Mortality was overestimated by all approaches irrespective of the study arm. For CVE an overestimation by all structural approaches was observed for the control arm. The CVE surgery arm was overestimated by the categorical BMI approach and slightly underestimated by the others. For T2D an underestimation was observed for the continuous and the categorial BMI approaches, whereas there was an overestimation by the risk equation approach. Considering different confidence interval limits, the cost per QALY gained are comparable between all structural approaches.

None of the structural approaches provided perfect external event validation results although the risk equation approach showed the smallest deviations compared to the external validation study. The cost per QALY gained resulting from the three approaches were fairly comparable.