In a non-interventional, epidemiological study, data were collected from medical records and via interviews for 938 subjects from various geographical areas of Greece. HRQoL and health utilities were explored with the EQ-5D-5L. Univariate associations were evaluated with the Mann-Whitney and Kruskal-Wallis tests for continuous and Chi-squared tests for nominal variables, and binary logistic regressions were employed to obtain marginal effects. Employing a split sample approach, various specifications of ordinary least squares regression models were evaluated in terms of goodness of fit, model specification, shrinkage and predictive and discriminative performance, to select the best model for mapping health utilities using the whole dataset.
Overall, the most important factors of impaired HRQoL and health utilities were higher age, female gender, obesity, poor glycemic control and increased duration of insulin treatment. History and increasing concurrence of all complications assessed were associated with exacerbated HRQoL problems, decreased health utilities and diminished health state, although it was not always statistically significant. The highest disutilities were associated with stroke (- 0.082), diabetic retinopathy (- 0.066), diabetic neuropathy (- 0.051) and severe hypoglycemia (- 0.050).
The deleterious impact of comorbidities on insulin-dependent T2DM subjects’ HRQoL has been confirmed and clinicians should adapt the priorities of disease management accordingly. The derived health utility estimates may be valuable for conducting economic evaluations of interventions in the area of T2DM when data are not available.