Accounting for behaviours and context in evaluations of complex health interventions

Abstract

Health care systems across developed countries face a perfect storm of rising demand and constrained funding. Systems have relied so far on short-term fixes but the time for incremental piecemeal solutions is passing. To achieve transformational change and fundamental service redesign, policy makers are resorting to ever more complex interventions. Evaluating their effects is far from trivial. From public health programmes, to integrated and community care services, to electronic health technologies, complex health interventions typically exhibit a large number of components and interactions among them and other parts of the system; involve numerous intricate behaviours by those delivering and receiving the intervention; engage multiple and diverse groups, organisational levels and populations; result in many outcomes, typically with a high degree of variability; and are extensively tailored to local settings and circumstances. Evaluating such interventions is as much about whether they work, as how and why. In this research, I examine the difficulties in using standard economic evaluation methods to assess complex interventions in the outpatient setting, and develop an approach to evaluation which uses methods and techniques that can explicitly address complexity, incorporate preferences and behaviours of patients and carers, and account for wider contextual influences. I apply the suggested approach to the evaluation of teleconsultation in Alentejo, drawing on insights from previous theoretical and empirical research, new econometric and statistical studies, and simulation modelling. The application makes contributions to extant research on behaviour and decision making, and has implications for the evaluation of teleconsultation, as well as for broader discussions of how to assess complex interventions. Complex health interventions have the potential to deliver a revolution in health care, but to achieve it we must be able to identify those that truly work, how and why. It is hoped the approach suggested here will contribute to that objective.