Case Study — Meadowbrook Primary Care Network

Module 4, Unit 1 | Healthcare case study

By the end of this case, you will be able to:

• Identify the real business problem beneath competing stakeholder priorities
• Map the current workflow and mark where an AI or automation proposal might responsibly intervene
• Analyse likely human impact, including who may benefit, who may be negatively affected and who needs to be consulted
• Prepare a proposal, workflow diagram and leadership presentation without building a working technical solution

Scenario snapshot. Six GP surgeries, 78,000 patients, three pressures on the leadership team, and an IT lead pushing a flashy AI demo at every network meeting.

Back to the five scenarios

Your Brief

You have been asked to advise leadership on whether and how AI or automation should be used in this organisation. Your recommendation should be specific enough to act on, but restrained enough to be credible.

You are not expected to build a working solution. Use the case to prepare a one-page Opportunity Brief, a workflow diagram or process map, and a 7–10 minute presentation followed by approximately 10 minutes of Q&A.

Organisation Snapshot

Meadowbrook is a Primary Care Network (PCN) covering six GP surgeries across a mixed urban/rural area of South Yorkshire. It serves around 78,000 registered patients and employs approximately 340 staff: 40 GPs, 65 nurses and healthcare assistants, 35 clinical pharmacists and social prescribers, and around 200 admin, reception and managerial staff. It operates as a federation with shared back-office services, but each surgery retains its own clinical leadership.

Business Context

The PCN is under pressure on three fronts. The Integrated Care Board (ICB) has introduced new targets on chronic disease reviews. Patient complaints have risen 22% year-on-year, driven mostly by appointment access and prescription delays. Reception staff turnover is running at 38% annually, with exit interviews citing burnout and abusive patient interactions. The Practice Manager has secured a small modernisation budget and has been told by the Clinical Director to 'show the ICB we are doing something with AI' before the next CQC inspection in eight months.

Process A — Referral letter drafting

When a GP decides a patient needs onward referral (e.g. to dermatology, gastroenterology, mental health), they dictate the referral letter into a digital dictation system. A team of medical secretaries types up the dictation, formats it to the receiving Trust's requirements, attaches the patient's relevant test results and history, and submits via the NHS e-Referral Service.

Pain points

• Current backlog averages 11 days from dictation to submission; the internal target is 3.
• Approximately 14% of referrals are rejected by receiving Trusts for missing information (past medical history, current medications, recent investigations) and require rework.
• Two of the six surgeries have lost their senior secretary in the last year and are running on agency cover.
• GPs report dictating the same patient context repeatedly across consecutive letters.

Process B — Repeat prescription requests

Patients submit repeat prescription requests through the NHS App, the practice website form, by phoning reception, by dropping a paper slip into the surgery, or by emailing the practice. Reception triages the request, the clinical pharmacy team reviews medication appropriateness, and a GP signs off. Prescriptions are then sent electronically to the patient's nominated pharmacy.

Pain points

• Average turnaround is 72 hours; NHS England target is 48.
• Patients can't see the status of their request, so many submit duplicates across multiple channels.
• Reception staff estimate they spend 40% of their shift on prescription queries.
• When a request is queried by the pharmacist, communication back to the patient often fails — voicemails, missed calls, lost slips.

Voices Around the Table

These voices are evidence. They do not all agree, and that disagreement is part of the case.

Practice Manager (lead sponsor): "We need to be seen to be 'doing AI' — the ICB is watching, and the CQC will ask. I want a quick win we can demonstrate."

Senior GP and Clinical Director: "I spend ninety minutes a day signing scripts I have already approved twice. Fix that, not the patient-facing app."

Lead Receptionist (15 years' service): "If you give the public a chatbot before fixing repeat prescriptions, you'll just generate more angry phone calls landing on my desk."

Secretary Supervisor: "The Trust keeps bouncing our referrals for missing past medical history. We need a tool that prompts the GP at the point of dictation, not another patient-facing gadget."

Patient Participation Group Chair: "My mum is 82. She cannot use the NHS App. Please don't make her ring a robot to get her blood pressure tablets."

ICB Digital Lead (external stakeholder): "We want PCNs to align with the national AI strategy. Show us governance, not just enthusiasm."

Proposals Already on the Table

Before your team was brought in, others in the organisation had already proposed ideas. Some may be useful. Some may be distractions. Evaluate them without getting captured by the loudest pitch.

Practical Realities You Should Know

Meadowbrook's six surgeries all use SystmOne as the clinical record system. The PCN pharmacy team uses PharmOutcomes. Community services that the PCN refers into use EMIS. On paper all three systems offer data export. In practice, field naming, SNOMED coding granularity, and use of free-text notes differ noticeably between surgeries: the same clinical question can produce different answers depending on which surgery's data is extracted and how. The IT lead has not yet audited this variation.

Constraints

• Year 1 budget for tooling: £25,000
• Must comply with the NHS Data Security and Protection Toolkit
• No patient-identifiable data may leave the UK
• CQC inspection due in 8 months
• Any solution touching clinical decision-making triggers DCB0129 clinical safety case work
• Patient demographic includes significant proportion of over-75s and digitally excluded users

Your Team Task

• Define the real business problem in measurable terms.
• Map the current workflow and mark where your proposal would intervene.
• Decide whether AI or automation is viable, and whether a lower-tech process change should come first.
• Explain likely human impact: who benefits, who may be negatively affected, and what support or consultation is needed.
• Name the risks, data limits, ethical concerns and governance requirements.
• Prepare your one-page brief, workflow diagram and 7–10 minute presentation.

Remember: This is a proposal exercise. You do not need to build a chatbot, workflow automation, dashboard, wireframe or prototype. A clear workflow diagram is enough if it helps the leadership panel understand your recommendation.

KSB evidence focus

• K22: Collaborative working principles to explore AI and automation solutions and implement prototypes, pilots or proof of concepts.
• S22: Present and communicate information including the translation of technical concepts into accessible materials to support clear dialogue with stakeholders.
• S23: Work with others to achieve agreed outcomes or outputs. Provide evidence-based analysis and insight to leaders on the likely human impacts of automation projects.
• S3: Undertake analysis to identify if automation is viable.
• S6: Review and complete workflow and process mapping.