Evaluating Diuretics in Normal Care (EVIDENCE): Protocol of a cluster randomised controlled equivalence trial of prescribing policy to compare the effectiveness of thiazide-type diuretics in hypertension

Introduction Healthcare systems must use treatments that are effective and safe. Regulators licensed many currently used older medications before introducing the stringent evidential requirements imposed on modern treatments. Also, there has been little encouragement to carry out within-class, head-to-head comparisons of licensed medicines. For commonly prescribed drugs, even small differences in effectiveness or safety could have significant public health implications. However, conventional clinical trials that randomise individual subjects are costly and unwieldy. Such trials are also often criticised as having low external validity. We describe an approach to rapidly generate externally valid evidence of comparative safety and effectiveness using the example of two widely used diuretics for the management of hypertension. Methods and Analysis The EVIDENCE (Evaluating Diuretics in Normal Care) study has a prospective, cluster-randomised, open-label, blinded end-point design. By randomising prescribing policy in primary care practices, the study compares the safety and effectiveness of commonly used diuretics in treating hypertension. Participating practices are randomised 1:1 to a policy of prescribing either indapamide or bendroflumethiazide when clinically indicated. Suitable patients who are not already taking the policy diuretic are switched accordingly. All patients taking the study medications are written to explaining the rationale for changing the prescribing policy and notifying them they can opt-out of any switch. The prescribing policies’ effectiveness and safety will be compared using rates of major adverse cardiovascular events (hospitalisation with myocardial infarction, heart failure or stroke or cardiovascular death), routinely collected in national healthcare administrative datasets. The study will seek to recruit 250 practices to provide a study population of approximately 50,000 individuals with a mean follow-up time of 2 years. The primary analysis will test for equivalence with a 30% margin in a per-protocol cohort. Ethics and Dissemination EVIDENCE has been approved by the East of Scotland Research Ethics Service (17/ES/0016, current approved protocol version 4, 28th September 2019). The results will be disseminated widely in peer review journals, guideline committees, National Health Service (NHS) organisations and patient groups. Trial registration number ISRCTN 46635087; registered pre-results, 11/08/2017. Strengths and limitations of this study design A cluster randomisation design maximises generalisability of results to UK NHS primary care. Study interventions with minimal impact on existing NHS workflows should encourage recruitment. Development of electronic study search tools and routinely collected data facilitates participation by remote and rural practices. One-off policy interventions may have a limited long-term effect on prescribing behaviour.