Measuring clinical uncertainty and equipoise by applying the agreement study methodology to patient management decisions

Robert Fahed,Behzad Farzin,Tim E. Darsaut,Miguel Chagnon,Jean Raymond

doi:10.1186/s12874-020-01095-8

Abstract

BackgroundClinical uncertainty and equipoise are vague notions that play important roles in contemporary problems of medical care and research, including the design and conduct of pragmatic trials. Our goal was to show how the reliability study methods normally used to assess diagnostic tests can be applied to particular management decisions to measure the degree of uncertainty and equipoise regarding the use of rival management options.MethodsWe first use thrombectomy in acute stroke as an illustrative example of the method we propose. We then review, item by item, how the various design elements of diagnostic reliability studies can be modified in order to measure clinical uncertainty.ResultsThe thrombectomy example shows sufficient disagreement and uncertainty to warrant the conduct of additional randomized trials. The general method we propose is that a sufficient number of diverse individual cases sharing a similar clinical problem and covering a wide spectrum of clinical presentations be assembled into a portfolio that is submitted to a variety of clinicians who routinely manage patients with the clinical problem.DiscussionClinicians are asked to independently choose one of the predefined management options, which are selected from those that would be compared within a randomized trial that would address the clinical dilemma. Intra-rater agreement can be assessed at a later time with a second evaluation. Various professional judgments concerning individual patients can then be compared and analyzed using kappa statistics or similar methods. Interpretation of results can be facilitated by providing examples or by translating the results into clinically meaningful summary sentences.ConclusionsMeasuring the uncertainty regarding management options for clinical problems may reveal substantial disagreement, provide an empirical foundation for the notion of equipoise, and inform or facilitate the design/conduct of clinical trials to address the clinical dilemma.

Highlights

If doctors really knew what to do for their patients with a particular problem, why would anyone conduct a trial? For many authors, including Fried, prospective observational studies of large data bases could solve the difficulty, but the notion of uncertainty resurfaces: this strategy can only study the comparative merits of rival treatments ‘if there is sufficient uncertainty in practice to ensure that similar patients will be managed differently by different physicians’. [11, 12] If ‘clinical uncertainty’, ‘disagreement among expert clinicians’ and ‘equipoise’ are so important to clinical care and research, can they not be subjected to verification and quantification?
Two preliminary remarks are in order: first, it is important to mark the difference with a survey of opinions of preferred treatments: More than surveying whether clinicians agree in principle or in theory, regarding certain types of cases in a generic sense, [18, 37] a reliability study is an empirical investigation that tests the reproducibility of judgments made in practice for a series of real particular cases
The studies we propose are designed to identify and measure clinical uncertainty defined as the repeatability of clinical decisions made by particular clinicians on particular patients

Summary

Introduction

Freedman proposed to ‘call this state of uncertainty about the relative merits of A and B ‘equipoise” He found personal equipoise ‘conceptually odd and ethically irrelevant’, and proposed ‘clinical equipoise’ as a better candidate to justify RCTs, because it ‘places the emphasis in informing the patient on the honest disagreement among expert clinicians’. [6] Fried’s personal care model which emphasizes individualized decisions is one reason for the separation of medical care from clinical research that requires randomized allocation. This problem remains the object of ongoing controversies concerning some recent comparative effectiveness trials [7,8,9].

Objectives

Methods

Results

Conclusion