PSY34 Assessing Interventions for Adults With Metabolic Syndrome: A Comprehensive Economic Markov Decision Model

Abstract

Metabolic Syndrome (MetS) is defined as a clustering of risk factors for diabetes mellitus (T2DM) and cardiovascular disease (CVD) which puts individuals at increased risk of developing these conditions and consequently leads to a reduction in life expectancy and increased morbidity. A systematic review and network meta-analysis was undertaken to assess the relative clinical effectiveness of a number of lifestyle and pharmacological interventions, both independently and in combination. A second systematic review, and series of meta-analyses, was also undertaken to estimate the increased burden that a MetS diagnosis has on the subsequent risk of T2DM, CVD and all-cause mortality. A fully probabilistic economic Markov decision model was developed in WinBUGS, and which directly included the series of meta-analyses above, in order to assess the cost-effectiveness of the various interventions. The use of both lifestyle and pharmacological interventions in combination was dominated in the incremental cost-effectiveness analysis, with the use of both of them independently producing greater health gain at lower cost. Pharmacological intervention was cost-effective compared to standard care (ICER £3050 with a probability of 0.53 at a threshold value of £20K/QALY), and lifestyle intervention was cost-effective compared to pharmacological (ICER £6933 with a probability of 0.52 at a threshold value of £20K/QALY). A series of sensitivity analyses were also undertaken both with regards to the model inputs/distributions and a number of methodological assumptions, but the results remained largely insensitive to these changes. The use of a lifestyle intervention would appear to be a potentially cost-effective treatment strategy for adults with MetS, however considerable uncertainty surrounds this decision. The use of a comprehensive approach to economic modelling within a WinBUGS framework allowed distributional assumptions to be relaxed, sources of correlation to be appropriately accounted for, and more complex sensitivity analyses to be easily undertaken.