Ethical and feasibility challenges in scaling up to a randomised trial of a complex intervention tackling problematic primary care prescribing: a consensus process

Problem

Cumulative meta-analysis of ‘Audit and Feedback’ (A&F) trials have shown no improvement in effect sizes, suggesting a lack of learning on how to improve effectiveness. The Campaign to Reduce Opioid Prescribing (CROP) provided 316 practices in West Yorkshire, UK (population 2.24 million) with bimonthly evidence-based enhanced A&F reports on their opioid prescribing for one year. The CROP intervention reduced predicted opioid prescription spending by £900,000 in West Yorkshire.We explored feasibility and ethical challenges of scaling up the CROP intervention for a primary care randomised controlled trial to reduce harmful opioid prescribing whilst adding significantly to the wider evidence base on A&F.

Approach

A structured Delphi consensus panel of A&F experts, primary care medicines optimisation leads and members of our existing Patient and Public Involvement Panel identified solutions to the ethical and feasibility issues of scaling up the CROP intervention to a trial, drawing upon a state-of-the-science summary of recommendations, on-going research and medicines optimisation expertise.The consensus process had two rounds and participants were presented with a range of trial feasibility issues to be considered (e.g. randomisation, ethical and governance issues). Participants rated characteristics for each recommendation on a 1-9 scale, where scores of ‘1’ indicates the strongest disagreement and scores of ‘9’ indicate strongest agreement.

Findings

Consensus was reached for 36 key conclusions. An opt-out approach to practice consent for a trial was the most acceptable approach to recruitment, however waiving consent completely was considered the most ideal. Randomisation at practice or primary care network level was preferred to higher level randomisation (e.g. by Clinical Commissioning Group). Participants rated detailed prescribing information highly, such as high-risk sub-groups of patients and excluding those not targeted by the feedback, such as those with palliative care patients, rather than total number of prescriptions.

Consequences

Waiving consent is the ideal option for primary care A&F studies using aggregated practice data, as obtaining consent may undermine results and create a greater burden than the A&F intervention, however this may be complicated by data protection. Randomisation at primary care network level is acceptable and may reduce the risk of trial arm contamination, although there are inherent risks of organisational instability which may threaten future randomisation. A key factor to the successful planning and delivery of a primary care A&F intervention trial will be how well ethical and feasibility issues relating to trial consent (waive consent, practice or Clinical Commissioning Group consent), unit of randomisation (practices, Clinical Commissioning Groups or regional), and the source of primary care data (nationally gathered databases or electronic health records) are resolved. We will present the panel's 36 key conclusions for scaling up a primary care intervention into a randomised controlled trial.