The intersection of randomized trials, realist and economic evaluations: exploring methodological convergence and impacts of a novel financing intervention
Live presentation on 10th Nov from 16:36-16:39 GMT
Dr Karin Diaconu, Prof Sophie Witter
Introduction
Tuberculosis (TB) is one of the greatest global health challenges of our time. Addressing this disease burden, particularly drug-resistant tuberculosis (DR-TB), requires novel systems-based approaches. As part of the Results4TB project in Georgia, researchers, policy-makers and healthcare professionals co-designed a novel performance-based financing intervention intended to strengthen TB related integrated care management. The intervention aims to reduce loss-to-follow-up among drug-sensitive TB cases, thereby curbing the potential development and transmission of DR-TB. Based on co-designed intervention programme theory, we have integrated three evaluation methods, including a cluster randomized clinical trial and realist and economic evaluation.
Methods
This presentation discusses the integration of the three evaluations in practice and presents preliminary results relating to the interventions’ implementation and effects. Overall, 8 public and private clinics have been randomized to take part in the intervention; 8 similar facilities serve as control sites. The primary outcome measure is the difference in the loss-to-follow-up proportion among drug-sensitive TB cases between intervention arms. In a sub-sample of intervention and control facilities, a realist evaluation is conducted concurrently, exploring the actual intervention, the underlying mechanisms and context conditions that explain the observed outcomes. Similarly, a health economic evaluation collects data on patient, service and health systems costs of the intervention.
Results
We present how the three evaluation methods have been integrated based on the development of the programme theory and reflect on the challenges and opportunities of employing such a design, including around coordination of data collection and analysis. Further, we discuss preliminary results of the evaluations at mid-point of the evaluation cycle i.e. 12 months. Overall, the intervention appears to be of low- to moderate-effect in achieving intended outcomes (i.e. relative effects under 20%). However, when disaggregating results and applying a realist perspective to the analysis, more distinct patterns emerge relating to the effects of underlying mechanisms and context conditions on intervention outcomes.
Conclusion
We offer reflections on how to design, use and integrate clinical trial and realist evaluation methods, including on the role of causal mapping in identifying useful points for evaluation convergence.
