Alex Burrell is a GP in Bristol and Associate Editor at BJGP Open.
Yonder: a diverse selection of primary care relevant research stories from beyond the mainstream biomedical literature
Scale
Since 2014, there has been a push for primary care to operate at scale through various means including primary care networks, federations, and super partnerships. While there are some benefits to secondary care organisations becoming larger, it is not clear if the same is true in primary care. This UK study aimed to assess the impact of scale on primary care output.1 Using data from 2022–2024, they found that an average practice provides 4709 appointments per month, of which 51.2% were the same or next day after booking. On average, an increase in one full-time equivalent (FTE) GP was associated with a 5.8% increase in total appointment volumes. This effect was bigger at larger practices but the proportion of appointments within 2 days of booking decreased, suggesting that, while more total appointments are produced, patients have to wait longer on average when the scale of the practice increases. Practices with a higher level of administrative staff were estimated to benefit more from an additional FTE GP, with an extra 243 appointments per month compared to 210 for those with a low level of administrative staff.
Shared care
Shared care agreements (SCAs) are made between a GP and a specialist for the GP to prescribe and monitor specialist medication for patients whose condition is felt to be stable or predictable. They can cover a wide range of medications from DMARDs to mood stabilisers, and are meant to enable patients to receive care in an integrated manner. Some of the medications that need SCAs are high risk and need careful monitoring, but there is no evidence to suggest SCA medications are well managed or monitored. This UK study involving practices in the Northeast of England found that 32.3% of medications being prescribed against SCAs did not have up-to-date monitoring information, most commonly medications prescribed for ADHD.2 Interviews suggested key issues behind this were a lack of patient involvement in setting up SCAs, ambiguities around roles and responsibilities, and uncertainties around whether to continue prescribing in stable patients despite outdated monitoring data.
AI scribes
I have not used AI scribes or ambient voice technology for consultation summaries yet but they have been piloted in my local integrated care board. Evidence remains limited of their clinical utility, safety and validity. This Australian study aimed to compare the quality of clinical documentation between AI scribes and human-generated notes using simulated consultations.3 Four cases from membership exams were used and enacted by an actor and a GP, with notes compared between four commercially available AI scribes and the GPs who performed the mock consultations. Notes were scored by three experienced GPs using a standardised and validated tool (PDQI-9). Human notes scored lower than all four AI scribes on average for every case though this was not statistically significant, and scored lowest in accuracy, thoroughness, usefulness, comprehensibility, synthesis, and internal consistency. Scores in accuracy, thoroughness, succinctness, and freedom from hallucinations varied significantly between AI scribes but there was no significant difference in overall mean score.
Trust in supervision
When supervising anyone in primary care, be that medical students, resident doctors, or allied health professionals, there is an element of trust involved. One component of this is entrustable professional activities (EPAs), those that are ‘the core activities of physicians’ work that are observable and measurable and can be entrusted by a supervisor to a trainee with a certain amount of autonomy’.4 This UK interview study aimed to explore entrustment decision making for GP tutors of final-year medical students.4 The supervisor’s personal approach to and tolerance of risk was a key consideration, as were the professionalism and interpersonal skills of the students they supervised. Patient factors including complexity, acuity, presenting complaint, and behavioural difficulties also impacted decision making. Decisions on what was appropriate to entrust to students varied widely between the small group of participants interviewed, but the students at the university studied only did a 3-week final-year GP placement that was felt to limit meaningful relationship building, an important facilitator of entrustment.
References
1. Zhao T, Meacock R, Sutton M. Scale, skill-mix, and access implications of the production of appointments by primary care practices in England. Health Econ 2025; DOI: 10.1002/hec.70064.
2. Cooper M, Trotter V, Hand A, Nazar H. Specialist medication monitoring and prescribing in primary care: case study of shared care agreements in northern England, UK. BMJ Open Qual 2025; 14(4): e003491.
3. Foo D, Tan J, Hansra A, et al. The great scribe-off: a comparative analysis of AI scribes versus human documentation in simulated general practice consultations. Stud Health Technol Inform 2025; 333: 34–39.
4. Cullen JV, Alberti H. A question of risk: how do undergraduate GP teachers experience entrustment decision-making in primary care? Med Educ Online 2025; 30(1): 2479755.
