Using MCDA (Multi-Criteria Decision Analysis) to prioritise publicly-funded health care

Abstract

New Zealand, like many other countries, is grappling with the problem of how to allocate limited resources across a range of health and disability support services at a time when demand for health care continues to grow faster than health budgets. It is becoming increasingly important for decision-makers to adopt robust processes for setting priorities so that limited health resources are allocated efficiently, effectively and transparently. In my thesis I use multi-criteria decision analysis (MCDA) to build a framework (at the meso-level of health care funding) which can be used by decision-makers to assist them in priority-setting.

Potential criteria, elicited from six focus groups (including members of the public, private and public health care providers, health professionals and policy makers), are combined with advice from health experts and criteria from comparable studies in the literature to establish six prioritisation criteria: ‘need’, ‘individual benefit’, ‘societal benefit, ‘age’, ‘lifestyle’ and ‘no alternative treatment’. An online decision survey implemented through 1000Minds software (Ombler & Hansen 2012) and the PAPRIKA method (Hansen & Ombler 2008) is used to determine the relative importance of the criteria. According to the results of a ‘test re-test’, the decision survey accurately captures the preferences of respondents.

The results of the decision survey reveal that ‘need’ and ‘individual benefit’ are the most important prioritisation criteria, and though patients are unlikely to be prioritised according to their age or lifestyle (because of discrimination), greater preference is shown for ‘age’ and ‘lifestyle’ compared to ‘societal benefit’ and ‘no alternative treatment’.

Regression analysis (including the application of a fractional multinomial logit model) and cluster analysis are used to determine whether the demographic characteristics of respondents can predict preferences. Several relationships are found. For example, health care workers, respondents on low incomes and Maori place more importance on ‘need’ (relative to the other criteria) compared to respondents who do not work in health care, respondents on middle or high incomes and non-Maori. Though several statistically significant results are found, it appears that overall the variation in preferences is largely due to the idiosyncrasies of respondents and not to particular demographic characteristics.

The criteria weights from the random sample are then brought together with cost and other additional factors in a prioritisation framework. With the aid of a Value for Money (VfM) chart and associated budget allocation table, decision-makers can consider all the prioritisation variables in a transparent and consistent way. The framework can be used as a communication tool, to allocate fixed budgets across a range of services, to keep track of previous decisions or to re-allocate resources when the budget has been cut.

The framework developed in this thesis illustrates how health care can be prioritised at the meso-level of health care funding in New Zealand. Ultimately it is up to the decision-makers to choose which treatments to fund, but if decisions are made explicitly within a transparent and robust framework that includes all relevant considerations (including the preferences of key stakeholders), then there is likely to be more acceptance in the outcome.