PMU43 DISEASE PRIORITISATION MCDA MODEL FOR INFORMING HEALTHCARE PRIORITIES SETTING IN UKRAINE

Abstract

Listen

Translate article

The study aim was to develop the diseases prioritisation model to inform new public health programmes development and healthcare resource allocation decisions in Ukraine. The MCDA (multiple criteria decision analysis ) model accounted for the following 5 criteria: diseases burden expressed in DALY (data from IHME GBD Compare, 2017 year), clinical value based on ASMR criteria, availability of treatment guidelines, public health impact and social solidarity. Disease scores and respective criteria weights were obtained from members of EC (n=12). Methods and model structure were based on published MCDA methodology (ISPOR Task Force 1,2; LSE). The results were presented for 293 diseases and 5 age groups (<5, 5-19, 20-39, 40-64, >65 years). Detailed results were presented as ranked lists of diseases representing 90% of DALY burden. Model estimated top-3 high priority diseases: neonatal preterm death, lower respiratory infections, neonatal sepsis causing 21.9% of death burden in <5 age group; gastritis and duodenitis, acute and myeloid leukemia and encephalitis causing 4.8% of DALY burden in 5-19 years age group; ischemic heart disease, ischemic stroke, major depressive disorder causing 27.3% of DALY burden in >20 years age group. A sensitivity analysis was conducted to explore parametric uncertainty of the the results. MCDA methodology utilising IHME disease burden data inputs can provide useful information for creating a public dialog required to establish healthcare system priorities in countries with limited availability of nationwide epidemiology data. The use of MCDA results should be interpreted with care, as they are subject to bias and subjectivity in assigning the weights and scores in a Delphi panel setting. The scoring criteria have to be methodologically robust, externally validated and assigned by experts trained in MCDA. Sensitivity analysis indicated that resulting disease ranking order was highly sensitive to criteria weights and scores inputs.