This is a collection of my thoughts about the challenge of improving appointment booking within the urgent and emergency care sector. It has a fairly technical focus and may well lack context for people not so familiar with the field.
I also want to explicitly acknowledge that this post does not examine user needs in detail as it was originally intended to support technical consideration – it doesn’t replace or negate the need for properly understanding the needs of users both professional (within the NHS) and public (users of the NHS).
Out of hours care has always had the concept of appointment booking for patients that need contact with a clinician from an out of hours service – this could be a face-to-face appointment at a local out of hours centre, or an appointment for a telephone consultation with a clinician (although this tended to happen less).
Prior to the introduction of NHS 111, out of hours care was generally provided locally (i.e. a patient would contact an out of hours service local to their area), and so booking of appointments was handled internally within the same organisation or with local arrangements between specific nearby services. In the majority of cases the transactions would all happen within the same IT system.
NHS Direct was a service that operated at national scale, but they did not offer appointments for patients; you would either receive a callback from a nurse, or NHS Direct would pass your details through to your local out of hours service which would then make contact with you.
With the introduction of NHS 111 in 2013, urgent care in England moved to a ‘national service delivered locally’ model. This model made use of national telephony infrastructure to route telephone calls to 111 through to different call centres around the country. In the vast majority of cases, 95% of 111 calls would be routed to a NHS 111 call centre close to the location of the caller using either an approximation of the landline location, or an approximation of a mobile phone using mobile mast triangulation.
Call demand on NHS 111 is incredibly ‘peaky’ – you can see this from the openly available monthly datasets that are published by NHS England. Particularly busy times for urgent & emergency care are long weekends (e.g. bank holidays), Easter, and Christmas / New Year – this is a well understood demand pattern in the NHS.
The challenging aspect of these few ‘super peaks’, however, is that they are several order of magnitudes larger that the regular weekly peaks (which general happen on Saturday mornings around 09:00 to 11:00). In terms of managing and planning for demand, this poses an immense challenge for services for whom simply ‘getting all of your staff in’ isn’t guaranteed to cover it.
An advantage of the national telephony approach is that it allows NHS 111 to distribute incoming calls flexibly across the country – this provides significant contingency for a range of situations (be it a particularly high demand in a particular region, or maybe a significant incident inhibiting the ability of a particular region to handle the local demand).
How does this relate to appointment booking though?
One impact of moving to the national service model is that it introduced a separation between the services handling telephone calls, and the services providing the appointment-based care to patients; in most situations this means separate IT systems are used too.
NHS 111 has been making good use of nationally-standard interoperability since its inception – this allows services to provide safe clinical handover of a patient between services (i.e. an NHS 111 service handing a patient over to an out of hours service so that they can provide further clinical care), however to this date there is no national standard defined for appointment transactions.
The supplier market responded to the gap by building appointment booking interoperability solutions for their customers which augmented the existing nationally-defined clinical handover interoperability – these solutions fill an important gap and several years on these solutions are used fairly widely both within the urgent care sector (e.g. between NHS 111 call centres and other urgent care services) and more recently between urgent care and general practice (GP surgeries).
The existing proprietary solutions are already becoming more widely deployed, and will continue to be deployed more and more as the urgent care system moves towards its aspiration of facilitating appointment booking into services for all patients (see the Next Steps on the Five Year Forward View or my precis of the urgent care specific elements of the FYFV). Given the relatively short timescale of the FYFV aspirations, and the fact that we don’t yet have a broad enough, nationally-defined standard for appointment booking interoperability using proprietary solutions in the short term remains a valid option.
There are some limitations to the existing proprietary solutions, and there is a common list of issues and gaps that service providers have started to identify. Some of the limitations are with the implementations themselves often relating to them being specific to individual IT systems (this is understandable in where solutions have been driven by the competitive market and have grown organically), other issues are related to the fact that there is some coordination needed across different services and regions which cannot realistically be provided by market-led proprietary point solutions.
Implementation of proprietary solutions could also increase the implementation cost in the long-run as it generally involves licensing on usage which attracts a cost to the users. There is also a coordination overhead that comes with the rollout of any proprietary point solutions. Having said that, the ROI is still considered to be favourable for investing in these existing capabilities in order to move interoperability between services on from where it is now.
Existing interfaces generally provide the ability to query slots of a destination service, pick an empty slot for a patient, and book an appointment into that slot. In some cases it is possible to change or cancel that appointment whilst still within the initial booking journey. Some systems have integrated the appointment booking transaction into the wider urgent care workflow, linking the appointment directly to the clinical handover (using the CDA documents that are sent between services). The official Integrated Urgent Care CDA messages have a section that allows systems to embed appointment reference details for the purpose of linking up with appointments within their system. There is a mixture of business rules applied by the various systems involved that guide how appointments can / can’t be booked, and what data items are required to do so. Some suppliers have attempted to build on top of existing proprietary solutions to avoid further divergence but this does mean building on top of existing limitations as well.
The biggest gaps are currently:
a) a lack of clearly defined appointment booking workflow that can be applied consistently across all services
b) a lack of an appointment repository / index that allows services to easily find out if and where a patient already has an appointment booked
c) a lack of business rules around how appointment booking should be handled when happening across regional borders (introduces financial, political, and accountability challenges)
So what comes next?
System suppliers and service providers alike have asked for standardisation of the appointment booking specification for urgent care and emergency care (note specification covers both an API specification and the associated business processes that support it).
There are a number of benefits from standardising the specification:
- it will reduce the implementation cost for both system suppliers and service providers through use of configuration over installation, and less variability related to the IT systems in use;
- it will introduce consistency of workflow across services and therefore ultimately improve consistency of experience for the users of the services (i.e. patients)
What about the existing systems like eRS and GP Connect?
System suppliers have been very clear that in order to realise the benefits of standardising the interoperability for appointment booking it is important that we don’t simply introduce YAAS (yet another appointments system). In particular alignment with the existing GP Connect work is preferred.
The key is in the user needs and the fact that nothing exists currently to meet all of the user needs in urgent and emergency care.
A drastic over-simplification of the use cases in urgent and emergency care appointment booking is:
- An urgent / emergency care service booking an appointment into general practice (a patient’s GP surgery)
- An urgent / emergency care service booking an appointment into another urgent / emergency care service
GP Connect is currently focused on enabling the first of these: booking an appointment into general practice (and specifically only the IT system suppliers that are part of the national GPSoC framework) – it therefore could help to meet some of our user needs in urgent and emergency care, but not all of them.
The e-Referrals Service (eRS) (formerly known as Choose and Book) currently targets a different set of needs: primarily referral from primary care into elective care* services (e.g. your GP referring you to a specialist in a hospital). The sort of workflows that support eRS use cases are quite different to what we need in urgent and emergency care with a key example being the respective definitions of an urgent appointment: in elective care this means 2 weeks, in urgent & emergency care this might mean within 1 hour.
In both cases there are IT systems in urgent and emergency care that do not already have integration with either GP Connect or eRS. This does not mean that neither can be of use, but it does mean that neither can provide what is needed in the current form.
The following notes were made whilst summarising some key characteristics of appointment booking in urgent & emergency care – they are highly likely missing context and are not exhaustive…
Appointment types (e.g. routine, urgent, 1 slot 1 patient, 1 slot many patients, not date/time bound, etc.)
Urgent care appointments are generally specific time-based slots i.e. come to x place at 20:30. In most cases a single patient will be booked into a single as would happen at a GP surgery, but busier or higher acuity services may utilise group slots i.e. come to x place at 16:00 and you will be seen around that time – this helps manage high demand and the general challenge of keeping to specific appointment times in dynamic services such as healthcare.
The urgency of an appointment tends to be denoted by the ‘disposition’ – a term which represents the concept of a timeframe within which the patient needs to be seen (e.g. must be seen within 6 hours) combined with the general type of care the patient requires (e.g. see a clinician, see a dentist, see a pharmacist). When a disposition is derived for a patient the ‘timer is started’ with regards to the timeframe.
As mentioned above, there will be variety in the types of appointment patients require e.g. it might be with a mental health specialist, with a nurse, they might need a clinician who specifically can prescribe medication, they might need to see an emergency dentist.
There is a mixed model of how this is provided: some service providers will supply several types of appointment, whereas other specialist providers might only supply one type (e.g. an emergency dental provider might only supply emergency dental appointments).
Slot management (publishing, synchronisation, ownership, etc.)
All existing models within urgent care use real-time slot queries to get availability of appointments. At the busiest periods even this can result in appointments becoming unavailable between the initial search and confirming it with the patient – this is not unlike the high-demand ticket-booking scenario. Demand on services can fluctuate rapidly, and services maintain the ultimate control over the appointments they are making available; it is not unheard of for a service to remove some availability when they see an influx of patients to prevent running over-capacity.
Ultimately the services providing the slots have ownership over what they do and don’t make available – there are some political complications if the service providers are contracted to provide a certain number of appointments for instance, but this is a separate detail.
Services will sometimes make slots available for certain types of patient, and have to manage their slot allocation carefully so as to avoid using all slots within the first hour of opening – they have an obligation to continue to provide a service during the out of hours period and so they have to apply demand predictions and spread their slot availability out. This can be done by restricting certain slots for patients with certain clinical urgency, and spreading these through the schedule (e.g. reserve a % of slots every hour that can only be booked into for patients with a urgency of < 1 hour). Another method of controlling demand is to use the disposition timeframe to filter the time range for appointments e.g. if a patient needs to be seen within 6 hours, don’t show slots sooner than 2 hours away. Obviously some serious thought has to be put into how this is all filtered and as to how they make sure they are providing an acceptable service for patients.
Who can do what? (professionals and patients)
Patients do not currently have any access to their own appointment information in urgent care, and there is no existing facility for patients to book appointments directly without first speaking to a member of staff within a service such as NHS 111.
The permission to book appointments does not tend to be restricted at an individual user level – at this point in time it is more about inter-organisational agreements i.e. can X NHS 111 call centre appointments into Y urgent treatment centre.
Generally if someone is able to deal with a patient on the phone, they will be able to take the patient through the full journey including booking them an appointment at another service. On a technical level, API authentication tends to happen at the organisational level with the authentication and authorisation of individual users remaining the responsibility of the respective systems – it is rare that Jackie Jones from Townsville NHS 111 call centre is specifically given authorised to book an appointment into Townsville Urgent Treatment Centre.
Slot mapping/semantics (e.g. slot = service, or slot = care role, or slot = name clinician, etc.)
This varies depending on the scheduling models that systems implement, and to a certain extent does it matter? Ultimately, though, slots are made available by services, and may be restricted by the type of care provided and restricted by certain patient criteria i.e. only patients who have received a certain disposition can book this slot.
There is no significant precedence of slots being at named clinician level within urgent care as staff are much more ephemeral than in other services, and encounters with urgent care by definition are not the start of an ongoing long-term relationship between the patient and service.
Managing appointments (booking, cancelling, rescheduling, recording attendance, etc.)
Proprietary solutions currently provide Book and limited Cancel / Reschedule functionality.
Cancel / Reschedule is limited because this can only really be done during the initial call with the patient i.e. if the patient changes their mind before the telephone call is completed.
None of the existing proprietary solutions support Cancel / Reschedule after the initial call has completed, and this is one of the big workflow challenges for us: how do we identify that an appointment has previously been booked for a patient if they are to call back to NHS 111 at some point after their original call. In particular considering the notes above regarding distribution of telephone calls nationally there is no guarantee that a patient will speak to the same call centre, let alone same operator, if they call back to NHS 111.
Any national approach will need to provide interoperability and operational workflows that support these missing bits.
Access to the booking system(s)
Patients do not currently have access to the appointment booking systems in urgent care.
Existing proprietary interoperability is all system-to-system – there are no publicly-available or open APIs for third-parties to easily consume.
Monitoring/Analytics/Reporting (e.g. DNAs, unused slots, etc.)
All handled locally by system, nothing centralised at the moment. Being able to look at how appointments are utilised across the urgent care system would be a good thing to tackle, providing it isn’t at the expense of the patient experience and operational workflows.
From a data perspective, it should be possible to get decent activity information about appointment booking without requiring patient-identifiable data – this will need looking at in more detail.
* Elective care is pre-arranged, non-emergency care, including scheduled operations. It is provided by medical specialists in a hospital or another care setting. You will usually be referred by your GP.