OBJECTIVES: Modelling cardiovascular (CV) outcomes in type 2 diabetes mellitus (T2DM) typically employs the extrapolation of short-term changes in surrogate markers and estimation of their long-term impact via predictive risk equations. The evidence base from cardiovascular outcomes trials (CVOTs) in T2DM is growing rapidly, due to FDA requirements for new treatments to demonstrate CV safety versus placebo as part of standard care. This new generation of CVOTs may require a new approach for associated cost-effectiveness models in T2DM.

METHODS: A targeted literature review was conducted to identify opportunities for future modelling of CVOT evidence from approaches taken in other therapy areas. Current NICE clinical and public health guidelines for CV conditions were reviewed to identify methods employed in de novo economic models to assess the impact of comparator interventions on CV outcomes.

RESULTS: Across 22 guidelines for treatment of hypertension, lipid modification, myocardial infarction (MI), stroke and other CV conditions, a total of 21 cost-effectiveness models were identified that explicitly modelled at least one CV outcome: MI, stroke, angina, revascularisation, peripheral arterial disease (PAD), heart failure and/or CV mortality. The majority of evaluations utilised lifetime (n=19) cohort-level (n=20) Markov (n=16) modelling approaches; the only patient-level evaluation utilised time-to-event simulation. Intervention-specific relative risks derived from meta-analyses were commonly applied to baseline risks of CV events, obtained from clinical trials and observational studies, including audit data. Surrogate markers were rarely modelled and the use of published risk equations limited to the estimation of baseline CV risk: Framingham (n=4), QRisk2 (n=1) and UKPDS (n=1) in the only T2DM-specific evaluation.

CONCLUSIONS: When modelling the outcomes of CVOTs in T2DM patients, alternative modelling methods may be more appropriate than typical T2DM approaches; a more suitable framework, consistent with the approach taken in CV modelling, may be to assess the comparative evidence via the application of relative risks.