Primary care
As the first point of consultation for patients who show signs of infection, General Practitioners prescribe three quarters of antibiotics in the UK.
Because they oversee the management of patients’ various health problems, GPs are best placed to judge their need for antibiotic treatment. However, in order to do so effectively, GPs need resources to deal with diagnostic uncertainty and ensure both physicians and patients have the confident to withhold or delay antibiotic treatment when it is not needed.
Reducing antibiotic prescribing in primary case can have a major impact on the total amount of antibiotics prescribed and reduce the population-wide exposure to antibiotics.
What we already know
- In England, more than three quarters of antibiotics are prescribed in primary care with one-third of patients receiving at least one antibiotic prescription annually (Dolk et al. 2018)
- Almost half this prescribing (46%) is targeted at respiratory tract infections, many of which tend to be self-limiting and are likely to improve in the absence of antibiotic treatment (Dolk et al. 2018)
- Other common reasons for antibiotic prescribing in primary care are urinary tract infections (23%) and skin and soft tissue infections (14%) (Dolk et al. 2018)
- There is substantial variation in the use of antibiotics between GP practices, with some prescribing at twice the rate than others even after accounting for the severity of the patient’s illnesses (Stuart et al. 2020)
- Ascertaining the appropriateness of prescribing decisions in primary care is difficult but researchers agree that a substantial proportion of prescribing is inappropriate, with estimates ranging from 8%-23% (Smith et al. 2018)
- The prescribed duration of therapy also often exceeds the recommended length of treatment stated in national guidance (Pouwels et al. 2019)
- The use of antibiotics is particularly high in young children and the elderly, which are commonly perceived at an increased risk of complications and thus treated conservatively (Dolk et al. 2018)
- Comorbid patients are another patient group that receives a disproportionate amount of antibiotics in primary care, leading to a situation in which more than 50% of all primary-care antibiotics are concentrated in only 10% of patients (Shallcross et al. 2017)
- Antibiotic prescribing rates have been declining in English primary care since 2014, possibly due to increased antibiotic stewardship measures (Public Health England 2019).
- Most existing interventions target prescribing for self-limiting infections in otherwise healthy populations and little evidence exists on how to optimise prescribing in vulnerable patients at the highest risk of resistant infections
What questions did we ask?
Electronic health records
We analysed individual-level EHRs from the Clinical Practice Research Datalink (CPRD), a large, pseudonymized, retrospective database of primary care records from the UK. CPRD includes data for 4.4 million actively registered patients (roughly 7% of the UK population). This database contains information on symptoms, prescriptions, diagnoses, referrals to specialist care, and diagnostic tests. We used it to investigate the following questions:
- What is the rate of antibiotic prescribing in healthy patients compared to patients perceived at a high risk of infection (e.g. elderly patients, or patients with lung disease)?
- What are the drivers of antibiotic prescribing in those patient groups with particularly high antibiotic prescribing rates?
- Are patients that do not receive immediate antibiotic prescribing at higher risk of negative outcomes?
Interviews
The aim of the interviews was to get a better understanding antibiotic prescribing decisions, views on the appropriate use of antibiotics in primary care and antimicrobial resistance.
The questions asked were underpinned by the theoretical domains framework (Atkins et al, 2017) to identify the individual, social and environmental factors likely to influence the prescribing or not prescribing of antibiotics.
Interviews took place in 2 high and 2 low prescribing GP practices in the Midlands and in London. Participants included GPs and nurse prescribers.
Observations
A researcher conducted non-participant observation in four GP practices in two regions (East Midlands and London). Two high and two low prescribing practices were selected using Public Health England’s 2016 Fingertips data.
The researcher observed in communal areas of the practice, at team meetings, and in patient consultations, and conducted informal conversations with staff, over a total of 117 hours.
The researcher recorded detailed handwritten field notes while at the research site, and audio recorded details from the field notes, reflections and interpretations. These audio recordings were then transcribed and analysed alongside pictures and policies/guidelines collected from the research site.
Data were analysed using a thematic analysis approach assisted by NVivo software to identify influences on antibiotic stewardship. Analysis was informed by a framework of six universal challenges to improving quality commonly faced by healthcare organisation (Bate et al. 2008).
Design workshops
We held two design workshops in 2 of the 5 NHS primary care practices where we had completed interviews and observations. 11 to 18 staff joined 1 to 2 workshops each lasting up to 2 hours. Staff included practice managers, clinical staff and administrative staff.
Workshop activities aimed to find out more about the pressures and stresses experienced by staff on a daily basis that might impact how able practices are to support antibiotic stewardship. Staff completed worksheets and undertook activities that invited them to describe, reflect, evaluate their experiences or generate new ideas, both as individuals or in small groups. These fed into whole-group discussions.
What did we find?
Electronic health records
- Patients with chronic obstructive lung disease (COPD) are among the most highly prescribed patients and account for 11.5% of all antibiotics handed out in primary care although they make up only 2.6% of the population
- Antibiotic prescribing is elevated by ~60% even in patients with the mildest form of COPD compared to similar patients without COPD
- Patients with mild to moderate COPD might offer an ideal target for interventions such as delayed antibiotic prescribing, as they experience few exacerbations and have a low risk of hospital admission but are nevertheless on average prescribed between one and three antibiotics per year
- Patients with other comorbidities linked to an increased risk of infection (e.g. asthma, diabetes, or cardiovascular disease) all tend to receive increased numbers of antibiotic prescriptions
- Antibiotic prescribing for most comorbidities was already high one year before the formal diagnosis of the comorbidity in primary care, and remained 20-50% percent higher one year after diagnosis with increases temporally correlated with the date of diagnosis
- Antibiotic prescribing for asthma, COPD, diabetes and heart failure increased particularly dramatically in the months immediately prior to the initial diagnosis of comorbidity, potentially due to misdiagnosis of symptoms as respiratory infections
- Careful reanalysis of the risk of sepsis in elderly patients not immediately treated with antibiotics in primary care for suspected urinary tract infection did not find a previously reported higher risk of sepsis linked to delayed/non-prescribing
- The conflicting results highlight the difficulty of estimating reliable treatment effects from routine health data, as patients with and without treatment often tend to differ in many aspects not well recorded in the data (e.g. severity of disease)
Interviews
Many different behaviours were performed by GPs and prescribing nurses when managing infections. Prescribing antibiotics was straightforward when signs and symptoms clearly indicated an infection. Difficulties arose when there was uncertainty about how the infection might develop if left untreated; this situation required safety-netting to ensure the patient understood what to do should their symptoms worsen. Some GPs would issue a delayed prescription; however, these were not followed up to check if used or not. When an antibiotic was not prescribed, it was necessary to explain the reasons why and to cope with any expectations from the patient. Prescribing decisions were often made without or before diagnostic test results were obtained.
The identified key themes fell within 7 theoretical behavioural domains: Memory, Attention & Decision-Making; Beliefs about Capabilities, Beliefs about Consequences, Social Influence, Environmental Context & Resources, Behaviour Regulation and Social/Professional Role & Identity.
Prescribing decisions were based on the autonomy of the individual to make their own clinical judgements. Key enablers were increased confidence in the diagnosis from the use of clinical scoring tools, e.g. FeverPAIN, Centors Criteria and diagnostic tests, e.g. urine dips and use of guidelines to select the appropriate antibiotic. Less confidence in diagnosing symptoms for high risk patients with comorbidities including COPD and older patients was reported. Also important was a practice wide prescribing policy which included audits, feedback sessions and regular clinical meetings.
Key barriers were lack of team-working and communication within the practice, ability to cope with patients’ demands and from parents seeking antibiotics for children when withholding antibiotics. Environmental barriers included availability of appointments within the practice, easy accessibility of alternative healthcare providers, such as out-of-hours clinics, and socio-economic factors related to the location of the practice.
Beyond clinical needs, the benefits of prescribing included patients getting better quicker and it is a fast and easy solution when time is limited, however, the risks are it can fulfil patient expectations and increase antibiotic resistance. The risks of not prescribing was the increased likelihood of deterioration of symptoms and hospital admissions or patients going elsewhere to seek antibiotics.
Low prescribing practices placed more importance on team-working as part of stewardship and were aware of the practices’ prescribing status. Others felt antibiotic stewardship was equally important in other healthcare settings and in non-human use too.
Observations
Collective ethos toward antibiotic prescribing Individual GPs varied in engagement with stewardships; a collective practice ethos was vital for effective stewardship and lower prescribing. Creating a collective ethos required local leadership; communication and coordination between GPs; and space and time for meetings, training and feedback. GP turnover and high reliance on locums presented challenges to developing this ethos. Higher levels of antibiotic prescribing in out of hours services was seen as undermining GP practice Antimicrobial Stewardship efforts.
Systems, tools, and performance monitoring Technology, including decision tools, alert systems and templates were used to assist decision-making and optimise prescribing but could be burdensome and intrusive. Quality of data for performance monitoring was problematic, and feedback lacking in some practices. Targets and incentives could drive practice activity but had little impact on individual GP behaviour.
Managing risk and access GPs perceptions of their practice population as a whole as elderly, co-morbid, vulnerable, deprived, homeless influenced their prescribing activity, with antibiotics seen as way of managing risk. Practice access and availability of urgent appointments impacted on GP willingness to use a safety netting approach to avoid antibiotic over-use.
Education and awareness raising Patient education about antibiotic overuse (e.g. through practice leaflets, and media campaigns) was seen as improving awareness and understanding of AMR, but negotiation of patient expectation for treatment and patient education within consultations was challenging.
Design workshops
We found the following daily common stress points made it harder for practices to be able to support good stewardship:
- time pressures to complete consultations and administrative tasks
- barriers to communication, both internally as a team, and with external services and staff
- IT problems
- patient behaviour, including demands for antibiotics
- lack of space or reliable equipment
- challenges in managing staffing and clinics
- insufficient clinical and administrative training
- insufficient patient education around antibiotics and resistance
- few ‘watch and wait’ options or strategies in place of ‘just-in-case’ prescribing.
Team comments also suggested the following would enable good stewardship in primary care:
- access to local and national prescribing and resistance data
- good internal and external communication in supporting stewardship consistency
- strong practice team collective ethos for stewardship and good team work generally
- quick and easy access to good patient information
- good access to training for both clinical and administrative staff
- improved patient education on their infection and on self-care.
Intervention development
We are using what we have learnt and to develop interventions that will address antibiotic resistance in healthcare. This involves:
- reviewing barriers and facilitators of stewardship
- identifying mechanisms of change and behaviour change techniques
- collaborating with practitioners to decide how best to deliver behaviour change techniques in terms of feasibility and acceptability.
PASS Interventions
PASS Interventions Presentation
(PDF, 2.5Mb)
References
Atkins, L., Francis, J., Islam, R., O’Connor, D., Patey, A., Ivers, N., Foy, R., Duncan, E. M., Colquhoun, H., Grimshaw, J. M., Lawton, R., & Michie, S. (2017). A guide to using the Theoretical Domains Framework of behaviour change to investigate implementation problems. Implementation Science, 12(1), 77. DOI: 10.1186/s13012-017-0605-9
Dolk, F. C. K., Pouwels, K. B., Smith, D. R. M., Robotham, J. V, & Smieszek, T. (2018). Antibiotics in primary care in England: which antibiotics are prescribed and for which conditions? Journal of Antimicrobial Chemotherapy, 73(suppl. 2), ii2–ii10. DOI: 10.1093/jac/dkx504
Pouwels K. B., Hopkins, S., Llewelyn M. J., Walker A. S., McNulty C. A. M., Robotham J. V. (2019) Duration of antibiotic treatment for common infections in English primary care: cross sectional analysis and comparison with guidelines. British Medical Journal, 364:l440. DOI: 10.1136/bmj.l440
Public Health England (2019). English surveillance programme for antimicrobial utilisation and resistance (ESPAUR) Report 2018-2019. Available online [Accessed 10 Feb 2020]
Rockenschaub, P., Jhass, A., Freemantle, N., Aryee, A., Rafiq, M., Hayward, A., & Shallcross, L. (2019). Opportunities to reduce antibiotic prescribing for patients with COPD in primary care: a cohort study using electronic health records from the Clinical Practice Research Datalink (CPRD). Journal of Antimicrobial Chemotherapy. DOI: 10.1093/jac/dkz411
Shallcross, L., Beckley, N., Rait, G., Hayward, A., & Petersen, I. (2017). Antibiotic prescribing frequency amongst patients in primary care: a cohort study using electronic health records. Journal of Antimicrobial Chemotherapy, 72(6), 1818–1824. DOI: 10.1093/jac/dkx048
Shallcross, L., Rockenschaub, P., Blackburn, R., Nazareth, I., Freemantle, N., & Hayward, A. (2020). Antibiotic prescribing for lower UTI in elderly patients in primary care and risk of bloodstream infection: A cohort study using electronic health records in England. PLOS Medicine, 17(9), e1003336. DOI: 10.1371/journal.pmed.1003336
Smith, D. R. M., Dolk, F. C. K., Pouwels, K. B., Christie, M., Robotham, J. V, & Smieszek, T. (2018). Defining the appropriateness and inappropriateness of antibiotic prescribing in primary care. Journal of Antimicrobial Chemotherapy, 73(suppl. 2), ii11–ii18. DOI: 10.1093/jac/dkx503
Dolk, F. C. K., Pouwels, K. B., Smith, D. R. M., Robotham, J. V, & Smieszek, T. (2018). Exploring the appropriateness of antibiotic prescribing for common respiratory tract infections in UK primary care. Journal of Antimicrobial Chemotherapy, 75(1), 236–242. DOI: 10.1093/jac/dkz410
