Engineering practice variation through provider agreement: a cluster-randomized feasibility trial

Abstract

PurposeMinimal-risk randomized trials that can be embedded in practice could facilitate learning health-care systems. A cluster-randomized design was proposed to compare treatment strategies by assigning clusters (eg, providers) to “favor” a particular drug, with providers retaining autonomy for specific patients. Patient informed consent might be waived, broadening inclusion. However, it is not known if providers will adhere to the assignment or whether institutional review boards will waive consent. We evaluated the feasibility of this trial design.Subjects and methodsAgreeable providers were randomized to “favor” either hydrochlorothiazide or chlorthalidone when starting patients on thiazide-type therapy for hypertension. The assignment applied when the provider had already decided to start a thiazide, and providers could deviate from the strategy as needed. Prescriptions were aggregated to produce a provider strategy-adherence rate.ResultsAll four institutional review boards waived documentation of patient consent. Providers (n=18) followed their assigned strategy for most of their new thiazide prescriptions (n=138 patients). In the “favor hydrochlorothiazide” group, there was 99% adherence to that strategy. In the “favor chlorthalidone” group, chlorthalidone comprised 77% of new thiazide starts, up from 1% in the pre-study period. When the assigned strategy was followed, dosing in the recommended range was 48% for hydrochlorothiazide (25–50 mg/day) and 100% for chlorthalidone (12.5–25.0 mg/day). Providers were motivated to participate by a desire to contribute to a comparative effectiveness study. A study promotional mug, provider information letter, and interactions with the site investigator were identified as most helpful in reminding providers of their study drug strategy.ConclusionProviders prescribed according to an assigned drug-choice strategy most of the time for the purpose of a comparative effectiveness study. This simple design could facilitate research participation and behavior change in non-research clinicians. Waiver of patient consent can broaden the representation of patients, providers, and settings.

Cluster randomised trials of prescribing policy: an ethical approach to generating drug safety evidence? A discussion of the ethical application of a new research method

“…This approach to comparative effectiveness evaluation was suggested by Sabin et al in their discussion of the ethics of cluster randomised trials [15]. The method was tested by McCarren et al in their report on the feasibility of using cluster randomisation to compare chlorthalidone and hydrochlorthalidone [16], randomly assigning health-care providers to favouring one agent over the other for any new prescriptions of thiazide diuretics. That study was granted a waiver of individual consent to broaden inclusion.…”