Thread by @EdTurnham

Lots of debate recently about triage in GP. Here's a thread with a detailed argument:

@MinalBakhai @SteveLaitner @dave_dlt @dr_musgrave @sib313 @stevekellGP @HelenRSalisbury @H_Atherton @jacobnhaddad @SiliconPractice

The short version:
- there's no such thing as no-triage General Practice
- it's reasonable to argue that traditional model beats current model, for some practices, but:
- tech improvements will make a triage model far superior to traditional model.

Let's not get too hung up on the semantics of 'triage', please. I use it as a shorthand for 'getting a problem dealt with in the most appropriate way, by the most appropriate person, in the right timeframe'.

Part 1. Here's my view of what triage should look like. It's v different to what we have, but mostly could be achieved with a level of tech that is standard in other industries.

Our patient has a medical problem. She goes to her practice's website/app.

She is led through a series of questions that aim to triage her to the most appropriate service (which might be the practice), the most appropriate person within that service (e.g. a nurse), and a level of urgency.

This process might start with her typing her problem/question in free text or through speech, with machine learning/AI used to work out what she means and which questions to ask next.

This triage process acts as a front door for the whole NHS. It could direct her to the out-of-hours GP, a mental health service, a physio, A&E, or her GP.

The algorithm makes use of her medical records. It interprets any new information in the context of her existing conditions. A patient with, say, learning disabilities would be asked simpler questions, with a lower threshold for arranging a consultation.

In some cases the outcome will be that she can manage herself with self-care. No input from a healthcare professional is needed (this happens already with NHS Pathways in 111).

In other cases the algorithm directs her to the right service, providing that service with a fairly detailed history that she has provided in response to a mix of open and closed questions.

If a patient is more complex, or is following up with their GP about a known problem, there might be fewer closed questions and more reliance on a few open questions - like the @askmygp approach.

Now the query is assigned to the appropriate service. Let's say it's the GP practice. The query - including codes such as 'back pain', 'pain on urination' - flows into the medical record.

The process of getting the query to the right person in the practice, maximising continuity of care and maintaining equity of workload - which can currently take several minutes - is mostly automated.

If she has a simple UTI, the query is presented to a nurse practitioner alongside relevant medical history. It takes 1-2 minutes to prescribe antibiotics and send a preset online response including safety-netting.

But let's say a GP is needed. The algorithm works out which GP would be most appropriate, based on who she's spoken to before.

The system also considers GPs' capacity and seeks to spread workload equally. Our patient might be assigned to the 2nd-most-appropriate GP, so that the other GP can concentrate on patients who really need to speak to them.

If the algorithm thinks a F2F appointment is probably appropriate (based on clinical info + the patient's wishes), it allows the patient to book this appointment with that GP, at a time of her choice.

Otherwise, the patient inputs her availability for a phone call, and is given a time slot during which she is likely to receive a call or online response.

The GP is then shown the patient's query, integrated into the medical record. The data from her query, along with the rest of her record, flows into a decision support tool that suggest possible diagnoses and the next data-gathering steps.

The GP has a choice to reply by online message, phone her, click a button to allow her to book a F2F appt, or reassign to someone else.

Throughout this process the patient is automatically kept updated of the progress of her query.

For patients who can't go online, a phone-based triage system will give a similar experience, with a receptionist asking questions prompted by the algorithm.

2. There is no such thing as General Practice without triage. Are there any GPs who take all the bloods, do all the dressings, remove all the sutures? No - receptionists triage these requests to the appropriate member of staff.

Ironically, opposition to 'triage' seems to correlate with support for reception signposting/care-navigation, which is just another form of triage. Why is triage fine when done by a non-clinician, but not when done by a clinician or highly intelligent software?

3. Triage, and dealing with lots remotely, does not mean lower continuity of care. See separate thread:

https://twitter.com/EdTurnham/status/1311423418005360641

Thread continues here:

https://twitter.com/EdTurnham/status/1377738728374951936