This initial evaluation of the STARRS-GM approach was undertaken to determine the feasibility of this comprehensive digital approach and particularly the utility of the LungHealth asthma computer-guided consultation (CGC). Health informatics and multiple deprivation index metrics were utilised to select one of the most deprived areas and the primary care networks serving Greater Manchester. This allowed the identification of a PCN with the challenge of excessive SABA use and poor asthma outcomes.

In this PCN the bespoke MIQUEST/SNOMED search tool was used to identify two cohorts of patients for review, the LungHealth asthma guided consultation was then utilised. The results show that the approach is practical. When the 338 patients receiving the guide consultation are considered, the first observation is that the CGC characterised patients grouping them into levels of control (as seen in Table 1) suggesting that use of the MIQUEST/SNOMED tool could be used to correctly prioritise selected patients for review using the guided consultation. At this point we recognise that only a proportion of the population has been evaluated. It is possible that in the whole cohort the tool would prove to be less specific, however, this data gave us enough assurance to continue the project with this search methodology. The consultation was also seen to be adept at identifying issues with care, which may lead to excessive SABA use and poor asthma control and identifying gaps in patient care. In addition to identifying and addressing gaps in their care such as 19% not having written action plans or the poor adherence in 18.5%, use of the CGC also prompts medication changes towards guideline management, though the healthcare professional does make the final decision as described. 44% of those reviewed had medication changes recommended with a step up in 82% and a step down in 18%. Referral for specialist assessment was also suggested in a significant number of patients though it must be noted that the population studies here is a subset of those on asthma register and many patients were selected for review because they were identified as being poorly controlled.

In primary care services, healthcare professionals are faced with the challenge of implementing an increasing number of complex clinical guidelines from different specialties to deliver optimal patient outcomes18. However, despite an emphasis on the importance of guideline-standard care, it is apparent that in conditions such as asthma the strategy of guideline dissemination in the hope of this translating into clinical benefit has yielded limited success. For example, while it is evident that the use of written personalised action plans and patient education leads to a significant reduction in healthcare utilisation, the implementation of this key practice point has been historically low, a finding mirrored here where 19% of patients were lacking a personalised action plan19,20. However, following CGC review, this had been achieved for nearly every patient in this cohort suggesting that the introduction of such intelligent clinical decision support system software into patient pathways may lead to a greater uptake of evidence-based practice, upskilling healthcare professionals and reducing variation in the delivery of care as has been demonstrated previously in the setting of COPD and OSA13,14. The CGC assesses asthma control using a multi-dimensional framework incorporating validated tools such as the ACT, assessment of adherence and physiological indices such as lung function and its algorithms also prompt the operator to consider asthma triggers and suspected occupational factors during review. All this ensures that patients with symptoms of uncontrolled asthma are not missed during a CGC consultation and are highlighted to the operator for further action. The National Review of Asthma Deaths stressed the need for patients to adhere to regular inhaled corticosteroid medication in order to maintain good asthma control and prevent deaths6. The use of the CGC highlighted 5% of patients who were found not to have been prescribed regular inhaled corticosteroid therapy despite the majority of this sub-group having poorly controlled asthma at the time of review. Following CGC review, all but one of these patients were commenced on regular inhaled corticosteroid therapy thus reducing the risk of future harm due to uncontrolled asthma. The finding of excess SABA use in a patient also represents a risk factor for future asthma attacks and national guidance states that the identification of this future risk is an important component in the delivery of personalised asthma care6 (https://www.brit-thoracic.org.uk/news/2019/btssign-british-guideline-on-the-management-of-asthma-2019/). Meeting this requirement is an area integral to CGC functionality as its algorithms alert the operator to those patients who meet guideline thresholds for excess SABA use and inhaled corticosteroid underuse.

Another important deficiency in asthma care that has come under recent scrutiny concerns the failure of healthcare professionals to recognise severe asthma in a timely and appropriate manner and trigger referral for specialist assessment according to guideline-based practice. This is particularly apparent with the advent of biologic therapies21,22,23. The implementation of the CGC resulted in three quarters of the cohort in the “specialist therapies” stage” or at the “additional controller” stage being identified as sub-optimally controlled. The CGC works to prompt specialist referral in such cases while also taking into account other modifiable factors such as adherence and any acute precipitating factors. At the opposite end of the spectrum, there remains a reluctance to de-escalate treatment in asthma where it is safe and clinically appropriate to do so thus risking adverse clinical and health economic consequences, e.g., side effects of high-dose inhaled corticosteroids24. The CGC prompted consideration of de-escalation in most cases where it deemed asthma control to be “good” with the operator actually de-escalating therapy in 37% of these cases. A 6-month prospective Dutch study focusing on severe asthma demonstrated that encouragingly, the use of an internet-based tool incorporating fractional exhaled nitric oxide (FeNO) levels and asthma control questionnaire (ACQ) resulted in a reduction in steroid dose (median cumulative steroid dose was 205 mg lower in the intervention group) without a deterioration in asthma control25. Our evaluation did not utilise FeNO measurements when stepping down therapy on this occasion but did reveal a significant difference between the number of reliever inhalers collected and those actually used. While these data are limited by self-reporting actual inhaler usage, it raises the important issue regarding the health economic impacts of medicines wastage and encourages development of strategies to address this issue26.

The role of clinical decision support software (CDSS) in the assessment of adult asthma in the UK has been described previously in the literature27,28. A Canadian study reported the impact of CDSS software on the uptake of asthma action plans and reported an increase in uptake from 0 to 17.8% and an increase in the proportion undergoing assessment of asthma control with a proportion of patients having therapy escalated compared27. One difference between the CDSS evaluated by these authors and that reported here is that in the latter, assessing asthma control is mandatory in order to complete the consultation. A critique of CDSS applicability in asthma published in 2014 commented that the effectiveness of such technology was found to be limited at the time due to the system’s recommendations not always being followed and a paucity of use28. However, since then, the increasing imbalance between capacity and demand within healthcare systems alongside the challenges posed by the COVID-19 pandemic has created new opportunities for the development and evolution of such digital solutions particularly when systems are fully integrated within the primary care EHR as in the case of the CGC reported here. Importantly, the remote capability of the CGC coupled with direct two-way connectivity to the primary care server enables elective primary care reviews to continue during pandemic conditions as patients may undergo such reviews from home and indeed healthcare professional can also work remotely if required.

This service evaluation carries some limitations in terms of extrapolation to wider clinical practice. All patients undergoing review with the CGC were on the GP asthma register with a primary care diagnosis of Asthma. It is recognised that there are patients on primary care Asthma registers who may not have a true diagnosis of Asthma and this evaluation does not take such a cohort into account29. However, the CGC is currently being further developed to consider important differential diagnoses and the presence of atypical symptoms in patients with a less certain asthma diagnosis. Further studies are required in this area to determine diagnostic validity in this setting.

The importance of appropriate use of and adherence to asthma medications cannot be overemphasised in clinical practice. The implementation of this CGC with the existing linkage to the primary care server and the MIQUEST© tool enables those patients who are deemed at being high risk of adverse asthma outcomes (e.g., excess SABA use and underuse of inhaled corticosteroids) easily to be identified and invited for a structured CGC review. Where poor adherence was addressed by patient education on the benefit of regular medicines, reinforcing self-management, addressing inhaler technique and arranging earlier follow up. However, at present, any benefit of the CGC in adherence assessment may be limited by the subjective account of actual inhaler use. Future clinical pathways may be enhanced further with the application of “e-inhaler” technology in selected “high risk” patients following CGC review and this area requires also detailed prospective study30. The use of FeNO in the assessment and management of asthma is gaining prominence within primary care and while the CGC enables the operator intelligently to interpret FeNO readings during a consultation both diagnostically and to aid therapy de-escalation, this was not evaluated in this preliminary analysis31. The two cohorts evaluated here represent a group in a PCN with a high deprivation index and in addition satisfied the priority of Health Innovation Manchester STARRS-GM project meeting high-risk criteria for adverse asthma outcomes or suitability for therapy de-escalation as opposed to an unselected asthma population. Nevertheless, it is clear this targeted approach is feasible and the scale of changes suggest beneficial outcomes can be envisaged and a roll out to an additional seven PCN’s is currently underway. As this is a preliminary cross-sectional analysis, we describe the management changes but not the clinical consequences of implementing the changes recommended resulting from the CGC review and a further longitudinal evaluation is planned aiming to measure the impact of this pathway in terms of reduction in SABA use, healthcare utilisation and hospitalisation due to asthma including outcomes in the cohort where de-escalation of therapy occurred.

The CGC was used here in a remote fashion by trained respiratory nurses based in primary care, but future service evaluations will involve use by practice nurses. Such an evaluation will also incorporate and define the training needs of practice nurses and General Practitioners in order to gain competency in the use of the CGC in such a pathway. Already available is an on-line training portal and a test site for users to enter test patient. We do recognise some users may require mentorship support in the first 1–2 clinics. Detailed longitudinal studies are also required to measure the health economic impact of such technology in primary care asthma management alongside any clinical benefits.

We have demonstrated that an end-to-end digital service solution is possible from the recognition of PCNs for prioritisation based on deprivation and/or poor asthma outcomes through to the identification of priority patient groups for review where there is the most gain. The introduction of clinical decision support software in the form of a computer-guided consultation when conducting asthma reviews within primary care is feasible. Not only this, but its use leads to management change in the majority of patients reviewed and the increased implementation of guideline-level standard of care, which is integral to improving patient outcomes and reducing health inequality.