Advanced analytics in primary care: actionable information (ACTION)
Research type
Research Study
Full title
Advanced analytics in primary care: provision of actionable information for quality improvement for antibiotic reduction, care advice and medication safety (ACTION)
IRAS ID
235520
Contact name
Tjeerd van Staa
Contact email
Sponsor organisation
The University of Manchester
Duration of Study in the UK
4 years, 6 months, 31 days
Research summary
Research Summary
Electronic health records (EHRs) provide a wealth of data. By making better use of this data and presenting it in more user-friendly ways to health professionals, the researchers working on this project believe they can make improvements in three key areas: care planning, antibiotic prescribing, and medication safety. Secure online computer tools will be developed that access de-personalised health records from GP practices and other NHS sites in Greater Manchester and Wirral that sign-up to the service. Care planning (PINGR - Performance Improvement plaN GeneratoR). This study aims to the look at how a health informatics tool can be used by general practices to help them make decisions about patient care. The application is a new type of audit software for general practices that is unique in that it suggests bespoke quality improvement action plans, starting with patients with hypertension. Antibiotic prescribing (BRIT - Building Rapid Interventions to reduce antimicrobial resistance & over-prescribing of antibiotics) Antibiotics are used to kill bacteria when we get an infection. At the moment we are facing a crisis in public health. The bacteria are becoming more resistant to the antibiotics and as a result they are becoming less effective. One of the reasons for this is over-prescription. This project, delivered by the Greater Manchester Connected Health Cities is applying a tech savvy solution to help understand and tackle the problem. Medication safety (SMASH - Smart MedicAtion Safety dasHboard). More than 2.7 million medications are prescribed in primary care in England each day. Adverse events arising as a result of prescribing errors (e.g. dangerous drug-drug interactions). A recent study in English general practices identified prescribing and monitoring errors in 5% of prescribed medication items, with one in 550 items containing a severe (potentially life threatening) error. This tool will flag such occurrences to GPs and other NHS staff, such as pharmacists.
Summary of Results
The aim of this study was to improve the health of patients and ensure services could be more joined-up by making better use of the digital information and technology that already exists in our health care system. This study would measure the level of appropriate clinical activities before and after the study period, using data from three independent electronic learning healthcare dashboard systems:
1) PINGR (the Performance Improvement plaN GeneratoR) focused on improved adherence to treatment guidelines in chronic disease management.
2) SMASH (Smart MedicAtion Safety dasHboard) focused on medication safety indicators.
3) BRIT (Building Rapid Interventions to reduce antimicrobial resistance & over-prescribing of antibiotics) focused on improving the effectiveness of antibiotic prescribing.
PINGR and SMASH were already well advanced when this study began, and had been evaluated separately for effectiveness. They were ready to be used in the healthcare system for direct care and so the focus for this project became the development and evaluation of the BRIT project.
The BRIT project itself aimed to reduce the misuse of antibiotics and better target antibiotic prescribing to patients at high risk of infection-related complications. This is a key area of focus for the NHS and was well received by GPs in workshops carried out to design the purpose and functions of the dashboards. 82% of antibiotic prescribing occurs in primary care, yet despite several efforts to reduce prescribing, levels still remained high particularly in the North of England. Dashboards had been developed and were in use by Clinical Commissioning Groups and GP practices, however they lacked the detail to allow practices to really understand where the issues were. The BRIT dashboards would allow for a much more detailed analysis because the data extracted is at the individual patient level data rather than at a practice level. To develop the analysis algorithms, the BRIT research team used data from large national primary care research databases. This allowed the team to understand where the key issues were and to target these for dashboard development.
The data analytics activities in BRIT consisted of benchmarking current practice in primary care, evaluating the levels of suboptimal antibiotic prescribing and identifying opportunities for improvement. Large variability in antibiotic prescribing was observed between practices and within practices: Change points in prescribing did not reflect updates to national guidelines. Prescribing levels within practices were not consistent for different infectious conditions. BRIT also found high levels of prescribing of potentially inappropriate type of antibiotics which were highest for otitis externa (67.3%) and upper respiratory tract infection (38.7%). BRIT found that that over the last 15 years antibiotic prescribing in primary care was not risk-based: patients with very low risk of infection-related hospital admissions were as likely to receive an antibiotic as patients with higher risks. BRIT also evaluated the effectiveness of treating common infections with antibiotics. The findings also indicate that incidental use of antibiotics is effective in reducing infection-related hospital admissions while repeated courses of antibiotics may have limited benefit and be indicative of adverse outcomes. Of 5.1 million antibiotics prescribed in UK primary care, only 14.8% were given to patients without any antibiotic prescribing in the previous three years and 43.6% are for patients who already received 5+ antibiotic prescriptions in the previous three years. These BRIT findings indicate that optimal antibiotic prescribing in primary care is a complex interplay of a patient’s symptoms, age and co-morbidity and previous history of antibiotic use. While incidental use of an antibiotic may reduce the risk of infection-related complications, it may also decrease the effectiveness of the antibiotic for future infections (possibly due to the development of resistance in the patient). The BRIT team published several publications on the analysis of the data, describing some of the key indicators for optimising prescribing in primary care.
The building of the BRIT dashboard infrastructure was complex and required extraction of anonymised data from individual GP practices and the secure transfer of this data into a Trustworthy Research Environment (TRE) for analysis. The summary of this analysis could then be visualised in the dashboards that GPs could access via a secure login. There were significant challenges in ensuring this process was consistent and reliable and required several changes in service providers and rebuilding the infrastructure in an off-site datacentre after the original build at the University of Manchester was no longer viable. At all times the BRIT research team remained conscious of data security and responsive to the needs of GPs. Additional engagement with GPs and practice managers helped to refine the dashboards so that GPs were able to understand their practice level prescribing. In addition, at the request of GPs, individual prescribing dashboards were developed so that GPs could see their own prescribing practice and where they might make improvements. The BRIT team recruited 85 practices to take part in the study, out of a target of 50. Additional practices were recruited recognising the difficulties in extracting data from GP practices who used different electronic health systems, had varying degrees of ageing IT systems, login/ remote access issues and the time requirements needed to remotely set-up the system for data extraction. In all the BRIT team received data on a monthly bases from 55 practices. As the GPs were recruited, and the learning healthcare system effectively established the plan to monitor and encourage use over the next 12 months to measure any change in prescribing activity was put in place. However, in March 2020, this plan was put on hold because of the COVID-19 pandemic. We continued to collect the data until Jun 2021 when the contracts for the extraction service and the TRE provider ended. In March 2021, the BRIT team received some additional funding from Public Health England who had a keen interest in the project and attempted to re-engage GPs by providing them with 6 key dashboard indicators by post and by email, both at a practice level and for each of the individual GPs in the practice. The feedback received was mainly positive and tested a mechanism for behavioural modification techniques required to take the project forward.
It was clear that there had been an impact on prescribing during the pandemic but it would be inappropriate to try to make some inferences as to the effectiveness of the dashboards during this time. However, the pandemic provided the team with some new and exciting opportunities to consider to develop dashboards within the new large national datacentres that had been established to support integrated care across all healthcare settings. This would negate the need to extract data from individual GPs practices and allow all GP practices in the datacentre to have access to prescribing dashboards. The team successfully received additional funding to take the dashboard development further and to use the dashboards to support interventions to impact prescribing, such as behaviour modifications, the inclusion of a knowledge support system to be used at the point of care and an online community of practice to share knowledge and provide support for more complex prescribing issues. All GPs who took part in this project will be invited to take part in the follow-on project. Whilst the aim to test the effectiveness of the system could not be achieved within the time frame and budget for this project, the establishment of the “live” learning healthcare system within primary care was a significant achievement and the BRIT team were able to identify a large number of target areas to optimise prescribing and demonstrate this in dashboards for the follow up studies.REC name
South Central - Oxford A Research Ethics Committee
REC reference
17/SC/0666
Date of REC Opinion
29 Jan 2018
REC opinion
Favourable Opinion