Aims: Economic evaluations of diabetes interventions do not usually include analyses on effects and cost of implementation strategies. This leads to optimistic cost-effectiveness estimates. This study reports empirical findings on the cost-effectiveness of two implementation strategies compared with usual hospital outpatient care. It includes both patient-related and intervention-related cost.
Patients and methods: In a clustered-randomized controlled trial design, 13 Dutch general hospitals were randomly assigned to a control group, a professional-directed or a patient-centred implementation programme. Professionals received feedback on baseline data, education and reminders. Patients in the patient-centred group received education and diabetes passports. A validated probabilistic Dutch diabetes model and the UKPDS risk engine are used to compute lifetime disease outcomes and cost in the three groups, including uncertainties.
Results: Glycated haemoglobin (HbA(1c)) at 1 year (the measure used to predict diabetes outcome changes over a lifetime) decreased by 0.2% in the professional-change group and by 0.3% in the patient-centred group, while it increased by 0.2% in the control group. Costs of primary implementation were < 5 Euro per head in both groups, but average lifetime costs of improved care and longer life expectancy rose by 9389 Euro and 9620 Euro, respectively. Life expectancy improved by 0.34 and 0.63 years, and quality-adjusted life years (QALY) by 0.29 and 0.59. Accordingly, the incremental cost per QALY was 32 218 Euro for professional-change care and 16 353 for patient-centred care compared with control, and 881 Euro for patient-centred vs. professional-change care. Uncertainties are presented in acceptability curves: above 65 Euro per annum the patient-directed strategy is most likely the optimum choice.
Conclusion: Both guideline implementation strategies in secondary care are cost-effective compared with current care, by Dutch standards, for these patients. Additional annual costs per patient using patient passports are low. This analysis supports patient involvement in diabetes in the Netherlands, and probably also in other Western European settings.