Out-of-Hours and NHS 111 Referrals: Where Remote Triage Fits

There’s a version of the remote triage pitch that every practice principal has now heard at least twice. It goes like this: patients ring at 11pm in pain, your phones are closed, an AI answers, sorts the abscess from the sensitive dentine, and the urgent care rota gets quieter. Some vendor demos even show a dashboard with a “cases resolved out of hours” counter ticking up.

That pitch is wrong, and not in a small way. The rota is set by your commissioner, the UDA targets are contractual, and no software product is going to stand between a swollen face and an emergency appointment. If you buy out of hours dental triage expecting it to shrink the emergency slot count, you will be disappointed within a fortnight and the licence will be cancelled before the first renewal.

Buy it for Monday morning instead. That’s where it pays.

What actually happens between Friday 5pm and Monday 9am

Run the numbers on your own practice for one weekend. A mixed NHS practice with three surgeries and roughly 6,500 active patients will typically log somewhere between 40 and 90 inbound contacts across a Friday-night-to-Monday-morning window, once you count voicemails, website form submissions, emails, and the NHS 111 Directory of Services referrals that land in your inbox as a DOS profile match.

Of those, the genuinely time-critical group is small. Cellulitis with trismus, a bleeding socket that won’t settle, dental trauma with an avulsed permanent incisor, spreading infection with systemic signs. In a caseload of 60 weekend contacts that’s maybe four to seven people, and they need the rota, not a chatbot. NHS England’s own urgent dental care standards treat those presentations as requiring face-to-face assessment within 24 hours, and quite right.

The other 50-odd contacts are the problem. They break down something like this:

Add up rows one, two, three, four, five and six. That’s 76% of the weekend’s contacts arriving unsorted into a Monday morning where your receptionist has 34 voicemails, two clinicians running 15 minutes late from the 8:40 starts, and a phone line that starts ringing at 8:01.

The Monday morning cost, in money

Here’s the calculation that actually matters when you’re signing a licence.

A practice manager triaging 40 weekend voicemails and forms by hand spends, conservatively, four minutes per contact: listen, call back (often twice, because people don’t answer), ask the clinical questions, decide, book or redirect, note it. That’s 160 minutes. At a practice manager’s fully loaded cost of around £22/hour, you’re looking at £59 of labour before 11am, every Monday.

But the labour isn’t the expensive part. The expensive part is what gets booked wrongly under time pressure.

Monday 09:14 — voicemail #17, transcribed
"Hi, it's Dawn Whitworth, I had a filling done Thursday
and it's really aching now, especially at night, keeping
me awake. Can someone ring me back."

Reception decision under pressure: book urgent slot, 11:20 today
Actual clinical need:  post-op sensitivity, day 4, no swelling,
                       responds to analgesia — phone reassurance
                       plus review in 10 days
Cost of the wrong call: one urgent slot consumed, one genuine
                       urgent case pushed to tomorrow

That’s the whole business case in miniature. Your urgent slots aren’t scarce because of demand at 11pm Saturday. They’re scarce because Monday’s triage happens in a rush by someone holding a phone in one hand, and the safe default under rush conditions is always “book them in.” Over-triage is rational behaviour for a receptionist who doesn’t want to be the person who sent away an abscess. It’s also the single largest consumer of your same-day capacity.

Where the tools actually earn their fee

Three categories are worth knowing by name, because they do genuinely different things and vendors blur the distinction.

Asynchronous photo triage. Toothfairy (the UK teledentistry platform, used by a growing number of NHS practices for remote assessment) and Dentulu both run on the same logic: patient submits photos through a guided capture flow plus a structured symptom history, a clinician reviews it asynchronously, a disposition comes back. Toothfairy’s own published figures put clinician review time at around 90 seconds per submitted case. That’s the number to compare against your four minutes of phone tag, and it’s the number that makes the maths work.

Front-desk automation with clinical routing. Dental Focus and Practice Plan’s patient-contact tooling, plus the newer conversational layers bolted onto SOE Exact and Dentally, capture out-of-hours contacts into a structured queue rather than a voicemail box. The clinical sorting is thin, but the structuring is the point. A voicemail gives you 40 seconds of audio. A structured capture gives you pain score, duration, swelling yes/no, trauma yes/no, analgesia taken, last visit date, all sortable.

Symptom-assessment engines. Anything derived from the Scottish Dental Clinical Effectiveness Programme urgent care pathways or the FGDP emergency guidance, delivered as a decision tree. These are deterministic, auditable, and boring, which for triage is a compliment. Toothpic (now operating across several UK corporates) sits somewhere between this and category one.

What none of these do reliably: replace the clinical judgement in the 8% urgent group. Don’t buy them for that.

A worked example: two Mondays at the same practice

Monday, before. 41 weekend contacts. Reception works through them 08:05 to 11:10. Nine urgent slots booked, of which four turn out to be genuinely urgent on examination. Two actual urgent cases from Saturday get callbacks at 10:40 and 10:55, and are offered Tuesday because Monday’s urgent list is already full of loose crowns. The 14% registration enquiries don’t get answered until Wednesday. One complaint logged about the wait for a callback.

Monday, after. 38 weekend contacts, 29 of which came through a structured out-of-hours capture form pushed from the practice website and the Google Business profile. Sunday evening, the on-call associate spends 22 minutes reviewing 29 structured submissions on a phone at home. Dispositions come back: 3 same-day urgent, 6 phone-call-first, 14 routine-book-anytime, 4 not-us (ortho provider, GP, community pharmacy), 2 needs-more-information.

Reception arrives Monday to a pre-sorted list. The three same-day urgent cases are booked by 08:20. Nine remaining voicemails are worked by 09:15. The 22 minutes of clinician time on Sunday cost the practice roughly £18 at an associate’s effective hourly rate, and it bought back about two hours of Monday reception time, six urgent slots that stayed open for people who walked in Monday with real problems, and the two genuine Saturday cases getting seen Monday rather than Tuesday.

Note what didn’t change: the rota. The on-call arrangements were identical both weekends. Nobody was diverted away from urgent care. The gain came entirely from arriving Monday with the sorting already done.

The NHS 111 referral question

This is the bit most vendors get wrong, so it’s worth being precise.

When 111 refers a patient to you via the Directory of Services, you get a referral with a disposition code already attached. The clinician or call handler has already triaged, using NHS Pathways. Your remote triage tool is not a second opinion on that, and treating it as one creates an audit problem: you’d be overriding a national triage system with a commercial product, without a documented clinical rationale.

What remote triage does help with on 111 referrals is the gap between disposition and appointment. A 111 referral with a “dental, urgent, within 24 hours” disposition arriving at 2am Sunday still needs someone to contact the patient, confirm they’re still symptomatic, check they haven’t already been seen elsewhere, and get them into a real slot. Roughly 20–30% of 111 dental referrals in our experience across mixed practices are stale by the time anyone rings: patient went to A&E, symptoms settled, or they were seen by another practice. A structured contact-back that runs before your reception team opens identifies those, and stops you holding slots for people who no longer need them.

Use the tool downstream of 111, never in parallel with it. That distinction will also matter if your ICB ever asks how your urgent access pathway works.

Buying it without buying the wrong thing

Ask the vendor how the output arrives on Monday morning. If the answer is “a dashboard,” push back. Your reception team is not going to log into a fourth system at 8am. The output needs to land in the queue they already work: the SOE Exact task list, the Dentally diary, or at minimum a single sorted email with the clinical summary in the body rather than behind a login.

Pricing sanity check. Per-case pricing in the UK market sits around £4–£9 for asynchronous clinician-reviewed triage, and £180–£450 a month for the capture-and-route layer with no clinician attached. At 150 out-of-hours contacts a month, per-case pricing gets expensive fast, which is why the capture layer plus your own on-call associate doing a 20-minute Sunday review usually beats it on cost. Run that comparison with your actual volume before you sign anything.

Check the indemnity position explicitly. If a clinician in your practice is issuing dispositions on photo submissions, that’s remote clinical practice, and your indemnifier needs to know. Dental Protection and the MDDUS both cover it, but neither covers it silently. The vendor’s own indemnity covers the platform, not your clinician’s decision.

And insist on seeing the escalation path in writing. Every one of these tools needs a hard rule that certain answer combinations bypass the queue entirely and generate an immediate phone contact: facial swelling plus difficulty swallowing, trauma with avulsion, bleeding not controlled by pressure after 20 minutes, temperature above 38°C with dental pain. If a vendor can’t show you that logic as an actual rule set, they’ve built a form, not a triage system.

For the wider framework on how asynchronous assessment fits alongside in-chair examination, including the record-keeping requirements, our Triage and Remote Assessment pillar covers the clinical governance side in more depth.

What to measure at 90 days

Pick three numbers before you start, and measure them the week before go-live so you have a baseline.

Urgent slots consumed by non-urgent presentations, counted on Mondays only. This is your primary metric and the one that justifies the spend. A practice getting this right should see it fall by half within two months.

Time from weekend contact to first patient conversation. Not to appointment, to conversation. If it was 26 hours and it becomes 9, that’s the patient experience improvement, and it’s what you put in your ICB report.

Reception minutes spent on weekend backlog, measured Monday 08:00 to 12:00. Have the practice manager actually time it for a fortnight. It’s tedious and it’s the only way to know whether the tool did anything.

If those three numbers haven’t moved by day 90, the problem is usually distribution rather than the software: patients don’t know the out-of-hours form exists because it’s three clicks deep on the website and not on the voicemail greeting. Fix the greeting, put the link in the SMS appointment reminder footer, and give it another six weeks before you judge it.