We demonstrated the mobilisation of 7.6 million people’s data to report asthma incidence, prevalence, care process /modifiable factors and outcomes in near-real time on a dashboard, which compared individual GP practices to the average of other practices in the RSC network. The dashboard data illustrates that inhaled relievers to preventers were over one in the study week and almost 40% in RSC did not have an asthma self-management plan from the dashboard inception and asthma review in the last 12 months. About 45% people with asthma did not have influenza vaccination. The dashboard provides a vehicle for A&F, and this paper reports the feasibility of doing this. The A&F topics would be around differences in incidence (Table 1 data), process of care measures (Table 2), and outcomes (Table 3). Thus, the dashboard can provide a helpful insight for busy GP practices of how the care and outcomes for their patients compare with reference standards and other practices in the network.

The asthma pyramid described here in England with a wide base of 520,069 people and a sharp peak of 0.02% hospitalisations is very similar to another cross-sectional study in Scotland30. Annual asthma prevalence in RSC of 6.7% is comparable to 6.5% in QOF in England for 2022–202331. Seasonal flu uptake in RSC of 55.5% in people with asthma is also comparable to 55.0% uptake across clinical risk groups and eligible age-groups in England for 2022–202332. There is huge room for improvement in our study population since the current advice is that people with asthma should have an annual review and asthma self-management plan33,34,35,36, and about 40% of RSC population with asthma did not have one. Since pay-for-performance incentives are not always effective, a dashboard updated frequently with potential for A&F as reported in this study, with perhaps the addition of goals and recommendations35,36, could be another way of motivating HCPs in primary care11. An interdisciplinary approach with HCPs, content and IT experts, to co-develop, review, modify and implement the dashboard was suggested18,37. Our dashboard prototype has A&F features which can provide objective data on current practice and can also compare other HCPs9,10, should HCPs feedback be useful to them. It was reported that implementation of asthma clinical practice guidelines in primary care could be improved by teamwork and assisting HCPs, especially nurses and pharmacists7,38,39,40. Furthermore it was found that asthma education and self-management programs when delivered by an integrated team of clinicians and allied HCPs are both clinically and cost effective41.

Also comparison of practice data with peers in a randomised control trial was found to be highly motivating42. Similar algorithms by the research team have been used for dashboards for atrial fibrillation43, influenza vaccine effectiveness26, and virological surveillance44, which have been peer-reviewed. The same research team also recently reported that their respective dashboards, which are economical and easy to scale-up, had been effective in increasing flu vaccination, medicine optimisation, improving diabetes care and impact on primary care quality45. Thus although this piece of work is not validated for asthma yet, we have evidence from similar exercise in RSC for other disease areas that such a RSC-dashboard has led to improvement.

To monitor and change HCPs behaviour, both to increase accountability and to improve quality of care, A&F is used widely in healthcare by a range of stakeholders, including research funders and health system payers, delivery organisations, professional groups and researchers. Three Cochrane systematic reviews between 2003 and 2012 on A&F interventions reported A&F led to small, but potentially important improvements in care and outcomes46,47,48. Later evidence reported that feedback was most effective when it was delivered by a supervisor or respected colleague, presented frequently, included both specific goals and action-plans in writing, aimed to change the targeted behaviour that needs addressed, focused on a problem where there was substantial scope for improvement and when the recipients were non-physicians35,36. The latest systematic review reported that even after 140 RCTs of A&F, it has remained difficult to identify how to optimise A&F, since of the 32 studies conducted after 2002, feedback was delivered only 19% of times by a supervisor or respected colleague and none of the studies included feedback with both explicit goals and action plans as recommended by previous systematic reviews35,36,48. The 2017 systematic review on electronic A&F, which is more common in recent times due to increased use of EHR, reported unreliable average effects, due to high heterogeneity and medium to high risk of bias in few studies (n = 7)49. Four statistically significant features were identified as independent predictors of improved clinical practice from 70 trials in a systematic review: automatic provision of decision support as part of clinician workflow, provision of recommendations rather than just assessments, provision of decision support at the time and location of decision making and computer based decision support50.

Timing of e-A&F is a big issue in its delivery51. e-A&F often provided by dashboards49,52, provide relevant and timely information via data visualisations of clinical performance summaries to healthcare professionals17,53. A review reported timely or near real-time e-A&F systems were imperative for proactive management of clinical risks, which resulted in increased participation and increased likelihood of reporting favourable outcomes54. Users found promptness of the feedback beneficial, insightful, made the data appear more reliable and performance-representative54. Less prompt feedback was frequently perceived as additional work needed and seemed to have taken place outside of the established workflow54. Absence of prompt feedback resulted in delayed effective action54. Furthermore, a barrier to usage of e-A&F systems was absence of real-time feedback54. Recent evidence suggests that e-A&F system implementation is effective within a highly stretched healthcare system when feedback is provided at near real-time, specific to user roles with an action plan embedded55. Near-real time feedback in dashboard was seen particularly effective during the COVID-19 pandemic to monitor the situation, assist in making clinical decisions and public health policies56. In contrast to near real-time notification, point-of-care notifications were more effective when data were about screening than lifestyle57. This was attributed to clinicians prioritising clinical information or which could be resolved comparatively quickly with less effort57. Alert fatigue from point-of-care notification was found in several studies, which resulted in distraction in workflow, clinician ignored content of message and which might have affected patient safety58. In summary, we conclude that both approaches have merits and limitations, and that these are best regarded as complementary approaches to driving forward quality improvement initiatives. The key strengths of point-of-care decision support capability is the potential for information availability to guide decision making during the process of care. The key limitations include presentation of what is perceived as irrelevant information by busy clinicians, presentation of information at a time that it interrupts workflows and high rates of over-riding of this information with associated risks. In contrast A&F offers the opportunity for assessing trends over time, benchmarking and assessing the impact of quality improvement initiatives. The main challenges with this are the presentation of data outside of the clinical EHR, which is an important barrier to access for busy clinicians and the non-contemporaneous nature of the data, which is something we have sought to overcome.

This study used a large primary care population database to create a comparative dashboard co-designed by multi-disciplinary team, on asthma at GP practice level and for all the other participating RSC practices. Thus, the findings on epidemiology, modifiable factors and outcomes can be generalised in the UK context. The feasibility of the dashboard has largely been technical, automating data flows to produce a dynamic contemporary data display across many practices. Additionally, we have had feedback on how we might develop the dashboard further. Generalisability on the A&F will come from a larger pilot where a larger group of practices are getting involved in A&F and we can look at engagement and change related to A&F. The A&F element would involve educational sessions with the intervention practices to explore how they might improve the processes of care (Table 2) and to provide a forum to discuss the outcomes of care (Table 3) with the goal of reducing adverse outcomes. Generally, the A&F cycle is a three-to-four-month cycle of review, then plan, practice activities, implement and then review again. Furthermore, these are possible to review by looking at the cumulative reporting in Tables 2 and 3 to get an indication of quality over the previous year as well as events in the last week.

The e-A&F is at practice-level. It is therefore not part of EHR. Rather, healthcare professionals have to log into asthma dashboard to view the e-A&F data. The dashboard though available using secure login, has not yet been promoted to HCPs due to lack of resources. Thus, we do not know if it is being used or what would be further useful to HCPs. Although the data existed to be able to categorise patients by mild, moderate or severe asthma, this was not attempted as our objective was to provide a snapshot at aggregated practice-level, as a technical feasibility exercise. The count of asthma self-management plan given and PPV in the dashboard are high since the look back period were long for both. People’s asthma symptoms vary over time and not all deteriorations lead to GP, out-of-hours or hospital visits. Such self-managed exacerbations of asthma will therefore not be recorded in GP records leading to a systematic under-estimate of the true prevalence of asthma exacerbations. Clinicians wanted data for previous week since they thought they might remember the details of an admission or exacerbation the previous week. However, the weakness of this approach is that it does not provide enough data for infrequent events such as asthma admissions. We recognise we may have to display exacerbations and admissions over a longer period.

A systematic review of e-A&F found that feedback displays were often graphical displays of individual practice performance and benchmarking, presented in dashboards49. While the feedback is not a written statement currently to economise on space in the dashboard, the graphs and tables have been kept simple for busy clinicians to quickly interpret. Thus, the smileys are just a quick indicator but not the only feedback. Dashboard algorithms are run in the background when new data are generated to do all calculations for tables, graphs and smileys. Whether tables, graphs and smileys were representing the data have been checked internally, before publishing on the web. The change in the smiley face is programmatic and highly reliable.

Since asthma is highly prevalent, the relatively low rates of asthma exacerbations still translate into a high number of A&E and hospital visits in a week. This may be considered a failure of asthma management and thus a bell-weather of the general management of asthma. This trend can be changed if we investigate the asthma modifiable factors and address them when required. Data on asthma modifiable factors are not routinely available, but insight of those data has the potential to alter asthma severity in people by timely intervention. The asthma dashboard with epidemiology, modifiable factors and severity measures could be an important CDSS tool in this respect.

Building on this foundational work, there is the potential to embark on a collaborative learning model involving a multi-disciplinary team of doctors, nurses, pharmacists, patient and public involvement members, behavioural health scientists, clinical informaticians, to better engage with people with asthma, aided by near-real time information as provided in the dashboard. Together we could develop methods to elicit what information HCPs would find most useful and how HCPs could be further motivated and based on their feedback we can improve the development and implementation of the asthma dashboard in primary care. Given the current system in place in RSC, there is an opportunity to implement A&F cycles in clinical practice and do a mixed method evaluation study of implementation effectiveness, by taking insights from HCPs on the ground interacting with the dashboard prototype and accordingly improve the dashboard iteratively, adopting a phased evaluation strategy52, with both explicit goals and action plans and find out what would aid scalability18,34,35,36,39,40,48. Multiple interventions were found to be more effective in asthma management than single interventions in primary care38. Given the recommendations from systematic reviews, we propose comparing GP practices to the high performing quintile for that parameter and set that as target for improvement, in near-real time.

This technical feasibility study found that an interactive, weekly, dashboard on asthma, with actionable insights for quality improvement, could be created with potential to support national A&F efforts, through a platform that is easily accessible online using primary care data. There are now opportunities to build on this foundational work through national experimental studies of A&F interventions to improve asthma care processes and outcomes.