A Bayesian framework for health economic evaluation in studies with missing data.

Alexina J Mason,Manuel Gomes,Richard Grieve,James R Carpenter

doi:10.1002/hec.3793

Alexina J Mason, Manuel Gomes + Show 2 more

Open Access

https://doi.org/10.1002/hec.3793

Copy DOI

Abstract

Health economics studies with missing data are increasingly using approaches such as multiple imputation that assume that the data are “missing at random.” This assumption is often questionable, as—even given the observed data—the probability that data are missing may reflect the true, unobserved outcomes, such as the patients' true health status. In these cases, methodological guidelines recommend sensitivity analyses to recognise data may be “missing not at random” (MNAR), and call for the development of practical, accessible approaches for exploring the robustness of conclusions to MNAR assumptions.Little attention has been paid to the problem that data may be MNAR in health economics in general and in cost‐effectiveness analyses (CEA) in particular. In this paper, we propose a Bayesian framework for CEA where outcome or cost data are missing. Our framework includes a practical, accessible approach to sensitivity analysis that allows the analyst to draw on expert opinion.We illustrate the framework in a CEA comparing an endovascular strategy with open repair for patients with ruptured abdominal aortic aneurysm, and provide software tools to implement this approach.

Highlights

Health economic evaluations use evidence from multiple sources including meta‐analyses, single randomised controlled trials (RCTs), and observational studies
This study found that the emergency endovascular strategy (eEVAR) strategy was cost‐effective, due to improvements in average quality of life (QoL), but these data were missing for 20% (3 months follow‐up) and 24% of patients (12 months)
This paper proposes and illustrates a flexible Bayesian framework for conducting sensitivity analysis to potential departures from missing at random” (MAR), allowing more realistic assumptions to be made about missing data in health economic evaluation

Summary

Introduction

Health economic evaluations use evidence from multiple sources including meta‐analyses, single randomised controlled trials (RCTs), and observational studies. Multiple Imputation has been widely recommended for handling missing data (Briggs, Clark, Wolstenholme, & Clarke, 2003; Faria, Gomes, Epstein, & White, 2014; Gomes, Diaz‐Ordaz, Grieve, & Kenward, 2013; Hunter et al, 2015; Hughes et al, 2016), but typically assumes that data are “missing at random” (MAR).. The MAR assumption can be expressed in two ways (Carpenter & Kenward, 2013, Ch. 1). The probability that data (e.g., on health outcomes) are observed is independent of their true value, conditional on the fully observed data (selection perspective). A second perspective and one we use in this paper is that given the fully observed data, the distribution of potentially missing data does not depend on whether those data are observed (pattern‐mixture perspective)

Objectives

Findings

Conclusion