Article Text

Download PDFPDF
Decision architecture randomisation: extremely efficient clinical trials that preserve clinician and patient choice?
  1. James Flory1,
  2. Jessica S Ancker2,
  3. Scott Yung Ho Kim3,
  4. Gilad Kuperman1,
  5. Andrew Vickers1
  1. 1 Memorial Sloan Kettering Cancer Center, New York, New York, USA
  2. 2 Vanderbilt University Medical Center, Nashville, Tennessee, USA
  3. 3 Bioethics, National Institutes of Health, Bethesda, Maryland, USA
  1. Correspondence to Dr James Flory, Memorial Sloan Kettering Cancer Center, New York, New York 10065, USA; floryj{at}mskcc.org

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Healthcare is full of choices between standard-of-care approaches where one might be better but we do not know which. Examples include ‘at what threshold should magnesium be supplemented for critically ill patients?’ and ‘which insulin formulation should be started in a hospitalised patient with diabetes?’1–3

Observational studies of such questions can be conducted relatively easily but are prone to biases, especially selection bias, that prevent them from reliably showing causal relationships between treatments and outcomes.4 Randomised controlled trials (RCTs) allow stronger causal inference but are major undertakings, typically costing over US$10 000 per patient.5 Beyond financial cost, traditional RCTs disrupt care, especially by assigning treatment based on random chance rather than clinicians’ and patients’ preference. Even for patients who merely consider trial participation, weighing benefits and risks and making a decision may create substantial burdens and stress.6

Information technology, with most treatment orders now placed through electronic health records (EHRs), has created opportunities to address these challenges by streamlining processes for participant screening, consent, enrolment, randomisation, intervention delivery and outcome ascertainment.1 6 7 For pragmatic trials that compare two standard-of-care interventions without masking treatment assignment, these innovations have lowered costs.8 However, these innovations have not addressed the fact that RCTs require patients and clinicians to prioritise a research study’s needs over patient care by accepting a randomly chosen treatment even when the patient would have selected another, given the option.

Decision architecture randomisation trials

The options that an EHR offers patients and clinicians when treatment and other orders are placed, and the way in which options are presented, are known as ‘decision architecture’, and can influence which option is selected. For example, the first on a list of medications might be more likely to be chosen, or a medication might be less likely to be used if it requires …

View Full Text

Footnotes

  • Contributors JF is a clinical epidemiologist and endocrinologist and is the writer and guarantor of this manuscript. JSA edited the manuscript as an expert in the optimal use of information and information technology to improve decision-making. SYHK edited the manuscript as an expert in bioethics. GK edited the manuscript as an expert in information technology and healthcare. AV edited the manuscript as an expert in clinical trial design.

  • Funding Funding from the Patient Centered Outcomes Research Institute (ME-2022C1-26378) is gratefully acknowledged.

  • Competing interests None declared.

  • Patient and public involvement statement This proposal was discussed by a group of approximately 15 patients and caregivers at a meeting at Memorial Sloan Kettering’s Patient and Family Advisory Council for Quality. This input is gratefully acknowledged and briefly summarised in the main text.

  • Provenance and peer review Not commissioned; externally peer reviewed.