OBJECTIVES: This review aims to critically appraise the methodology and quality of cost-effectiveness analyses (CEAs) of continuous glucose measurement (CGM) in type 1 diabetes (T1D).

METHODS: Searches were conducted in MEDLINE, Embase, Web of Science, Cochrane Library, and EconLit from January 2000. Model- and empirical data-based studies evaluating any CGM system in T1D reporting an incremental cost-effectiveness ratio (ICER) were included. Two authors independently conducted study selection and data extraction. Data extraction focused on study design, inputs, validation, and uncertainty. Methodological quality was assessed using the Philips, CHEC, and AdViSHE checklists. The review was reported conform PRISMA guidance.

RESULTS: From 1,736 articles screened for eligibility, 34 were included. Most studies (32) were model-based CEAs. Only 2 studies were empirical-data-based CEAs. 24 studies used the CORE diabetes model and one study applied the Sheffield model. The remaining 7 models were newly developed, though most were variants of each other. The effects of CGM ran via reductions in HbA1c (n=30), hypoglycaemic events (n=28), and fear of hypoglycaemia (n=23). All but one model-based studies assumed a lifetime duration of the effect size. In total 14 different combinations of CGM and insulin administration were evaluated. The empirical-data-based CEAs used a within-study time horizon of 6-12 months and all concluded that CGM is cost-effective/saving. All but 4 model-based studies concluded CGM to be cost-effective. In total 25 studies were sponsored by the device producer. Most studies tested parameter uncertainty, though few did a full probabilistic sensitivity analysis. Structural uncertainty testing was limited.

CONCLUSIONS: A major concern of current CEAs is the limited testing of structural uncertainty, while most analyses were based on the same model or a variant of that model, which implies results should be interpreted with care. Additionally, there is need for long-term data to justify the assumption of a lifelong effect size.