Changing Outcomes Through Achievement Emails in COPD Using Routine Healthcare Audits

Audit and feedback (A\&F) is defined by Ivers et al. as the "provi(sion of) a recipient with a summary of their performance over a specified period of time" and has long history of use within a diverse area of research and business to promote clinician and organisational behaviour change. In healthcare, A\&F has been used either alone, or as part of a composite quality improvement project (QIP), to improve health practitioner practice based on guidelines or targets. A 2012 Cochrane review systematically reviewed 140 papers on A\&F spanning several decades and produced a meta-analysis showing an absolute risk difference (RD) in practitioners changing behaviour to comply with the preferred practice of 4.3% (interquartile range 0.5% to 16%). Their investigation emphasised that A\&F is particularly effective when it is targeted specifically at prescribing practices (mean adj. RD 13.1% verses diabetes management 0.5%) and where baseline compliance is low.

Meta-regression of the Cochrane data showed that the effect size of A\&F has remained largely unchanged over the last two decades of study, leading the authors to postulate that there is no longer reasonable clinical equipoise to question if A\&F is effective. However, they also conclude that studies aiming to isolate how A\&F can be utilised most effectively by comparing different A\&F variations are few and far between and that this should be the target of future A\&F research.

A\&F has previously been used successfully within the West Yorkshire Clinical Commissioning Group (CCG) - now replaced by the NHS West Yorkshire Integrated Care Board (ICB) - to effect change on primary care prescribing practices. The ASPIRE cluster Randomised Controlled Trial (cRCT) of A\&F in this region produced a reduction in risky prescribing (odds ratio 0.82), while the CROP quasi-experimental controlled interrupted time series analysis of A\&F correlated with 15,000 fewer patients being inappropriately prescribed opioids over a period of a year compared to controls (35,36). Both studies utilised a similar A\&F design, utilising a theoretical structure based around the Theoretical Domains Framework (TDF) that is ideal for determining the possible perceptions of interventions by its intended targets. They also operated the Behaviour Change Taxonomy (BCT) to describe their employed behavioural change techniques and their intended effects.

The combined Long-Acting Beta Agonist / Inhaled CorticoSteroid (LABA+ICS) inhaler category has been used to treat COPD patients who present with asthmatic features of partially reversible lung function in the past. The most recent Global Initiative for Chronic Obstructive Lung disease (GOLD) guidelines move away from this however, by actively suggesting their use should be discouraged in favour of LABA-Long-Acting-Muscarinic-Antagonist-ICS (LABA-LAMA-ICS) triple inhalers. This is based on two separated studies suggesting LABA-LAMA-ICS inhalers reduced exacerbations of COPD in a subset of patients compared to LABA-ICS. In 2018, an observational study in the UK suggested that LABA/ICS usage in the COPD population was approximately 20%.

Precision feedback is a concept that suggests that aligning feedback messaging to individual, objective performance and goals may improve the performance of audit and feedback to create change. Here we propose to use the opportunity provided by a pre-planned QIP into COPD prescribing in West Yorkshire to undertake a randomised controlled trial to address this question.

The study design is a Randomised Controlled Trial, which is taking place inside a pre-arranged quality improvement project (QIP) that will involve sending digital A\&F feedback reports to all GP practices within West Yorkshire. The intervention element of this RCT will focus on the email that accompanies this digital A\&F feedback reports, where primary care practices within the NHS West Yorkshire ICB region would be randomised to receive their digital A\&F feedback reports alongside either personalised achievement emails, or generic emails only. All participants would take part in the pre-arranged QIP with or without this proposed research going ahead as it has already been agreed by the regional team. Therefore, randomisation will only affect the email that accompanies the A\&F report. Once randomised, each practice will receive 7 A\&F reports over the course of a year alongside their allocated intervention email format.

The personalised email wording will change based on the achievement of the practice and the content in the feedback reports, which is being distributed as part of the pre-existing QIP, rather than this research study. The recommendations in bullet points within the personalised emails suggesting how to improve care will be informed as relevant succinct summaries of the key guidance contained with the feedback reports, which will be based on either national guidelines or respected published literature. This will be written by the lead researcher for this project (who is the author of the QIP feedback reports themselves) and will be reviewed by the same group that reviews the QIP feedback report content before publication. The control email will be the same text each time for all practices within that arm and will contain no suggestions for improving practice. Examples of these intervention email types have been uploaded with this submission.

The outcome of this study will be a comparison between the proportion of patients with COPD on ICS/LABA inhalers within this primary care region. Secondary outcomes would look at the other markers of COPD prescribing selected for change in the QIP by the expert panel, as detailed in A11. Features that may influence the impact of the A\&F intervention, including practice size, deprivation index, and respiratory disease burden will also be collected. Where available and when a formal data sharing agreement in the form of an organising information document has been signed, NHS Digital commissioning data for care of COPD patients that has been aggregated and pseudonymised at practice-level will also be supplied for analysis by West Yorkshire ICB.

Primary Care practices within NHS West Yorkshire ICB are already required to gain consent from patients to share patient data for quality improvement purposes. West Yorkshire ICB has a pre-arranged data sharing agreement with each individual GP practice for audit and feedback work, which includes permission to share the data with the University of Leeds. The pre-arranged, separate quality improvement project to improve COPD outcomes in the region through audit and feedback will involve the sharing of relevant patient data between primary care practices and the ICB. As part of this quality improvement project, the ICB will anonymise all patient level data, then aggregate this into a practice level data-point. Standard practice for this quality improvement project would then be for the ICB to pseudonymise this practice level data and share this data with the University of Leeds to aid with analysis. The only differences that this Randomised Controlled Trial adds to this process is that pseudonyms will be pre-arranged by the ICB for each practice. The research team will then randomise (using block randomisation via a simple computerised Random Number Generator) each practice pseudonym to receive their audit and feedback report alongside either a personalised achievement letter email, or a generic email. The research team will remain blinded to the identity of which practice has been assigned which pseudonym. However, clearly it is impossible for the ICB and the practices themselves to be blinded to the assignment status of each practice due to the nature of the intervention. The ICB will retain the process of sending out reports by email and post, as would be standard practice in such a quality improvement project.