The system of unscheduled care in Scotland: understanding the sources of emergency inpatient admission

The system of unscheduled care in Scotland: understanding the sources of emergency inpatient admission Delivering for Health/Long Term Conditions Programme. Information Services Division Working Paper 3 Version 1.0 29 June 2007

Version history and revisions: Version Date Revision Actioned Approved 1.0 29 June 2007 Final Report Eddie Adie Steve Kendrick Authors: Editorial Support: Steve Kendrick / Joy Tomlinson / Jack Vize Penny Bridger / Eddie Adie Acknowledgements: This analysis was possible because we were provided with aggregated data from a number of different people and organisations. We would like to thank both for information and advice: Stephanie Phillips, Scottish Ambulance Service; Sandra Cormack, NHS 24; Jonathan Todd, NHS Lanarkshire; Jim Marshall and Joan Barr, Glasgow Emergency Medical Services; Catriona Haddow and Laura Dobbie, Information Services Division. Contacts: For further information, or to contact the report s authors, please email dhip@isd.csa.scot.nhs.uk.

Key Points. 1. Most of the partner services involved in local systems of unscheduled care are developing increasingly sophisticated information systems for recording patient activity. 2. However, at present it is virtually impossible to pull together information from these sources to monitor the system as a whole. It is difficult to link-up the 'outputs' of one system with the 'inputs' of other systems in order to trace even the main patient pathways through the system either on an aggregate or a patient basis. 3. By combining information from several sources it has been possible to derive order-ofmagnitude estimates for four of the main patient pathways to emergency inpatient admission for one NHS Board (NHS Lanarkshire). The estimates are: Self-referral to A&E 35% Ambulance (999 etc.) 25% GP referral (in hours) 25% Out-of-Hours Service 15% Recommendations. 1. Work should be done to harmonise information relating to the 'connectors' between the different components of the system of unscheduled care such as NHS24, Scottish Ambulance Service, GP referrals, Out-of-Hours services, A&E Departments and hospital admissions. In other words, definitions of patient flows out of one part of the system should link up with definitions of patient flows into other parts of the system. 2. The SMR01 record should contain a field identifying how the patient arrives for inpatient admission: in particular whether or not the patient was admitted via A&E. 3. A&E information systems should be able to identify how patients arrive at A&E and what their destinations are from A&E. - 1 -

The nature of the exercise. It must be stressed that this paper represents a preliminary exploratory analysis undertaken in a relatively short space of time. There was no guarantee at the outset that it would be possible to derive even broad estimates of the relative size of the main patient pathways to emergency inpatient admission based on existing sources of information. The intention was just as much to 'test the system': to see how far we could get. The message that it is very difficult to come up with estimates of the size of the main patient pathways is just as important as any information about those pathways. Introduction. This paper forms part of a programme of work aimed at understanding recent trends and variations in emergency inpatient admission in Scotland. An earlier paper reports of recent trends in emergency admissions at NHS Board level 1 while a companion paper reports an analysis of variation in rates of emergency inpatient admission at GP practice level 2. Although we were able to analyse such variations on the basis of hospital admission records it quickly became clear that we would be unable to shed any light on variations in the level of GP referral for emergency care. Patients can become an emergency inpatient admission along a variety of pathways only some of which involve a GP. The four most common pathways are: patient attends A&E; patient calls an ambulance; patient is referred for emergency care by a GP 'in-hours'; patient is referred by an out-of-hours service. Only the latter two are likely to involve a GP. Obviously this is an extremely simplified view of the complexity of pathways available especially given the increasing involvement of NHS24. - 2 -

However, at present there are no national, routinely collected sources of information which would allow monitoring of even these four main referral routes into hospital. In this paper we attempt to give an estimate of the relative size of these four main pathways using routinely collected information from several sources. Background This section describes the provision of emergency medical care in Scotland and the changes that have been introduced in recent years. Current systems providing emergency medical care Emergency medical care has been described as NHS care which cannot be reasonably foreseen or planned in advance of contact with the relevant health care professional, or is care which, unavoidably, is out with the core working period of NHS Scotland. It follows that such demand can occur at any time and that services available to meet this demand must be available 24 hours a day. This definition was taken from Building a Health Service fit for the future. 3 Sources of emergency admission Patients can be admitted to hospital as an emergency by one of four main routes. At any hour of day or night patients can attend accident and emergency (A&E) departments for immediate emergency care. Alternatively patients can call an emergency ambulance and this will deliver them to hospital. Within working hours patients can attend their own GP surgery for urgent care or they may receive an emergency home visit. Out with working hours patients can receive emergency care from their local primary care out of hours service. We do not routinely collate information from all these possible referral sources but most of the services involved collect information on the numbers of people they see and refer to hospital. Within working hours care The system providing care within working hours has not changed a great deal over recent years. - 3 -

As well as seeking assistance from their own GP, patients can refer themselves directly to A&E departments or call an emergency ambulance if they believe they require immediate care. Out of Hours Care Over the last twelve years there have been considerable changes in the provision and delivery of primary medical services out of hours. These changes may have impacted on each of the patient flows described above, for example it may be the case that more patients now refer themselves directly to A&E rather than calling their GP out of hours. However, since information about source of patient referral is not collected routinely we do not know if or how referral pathways have changed. Before1995 Before 1995 GPs were wholly responsible for the emergency medical care of their patients 24 hours a day every day of the week. However this became unsustainable. There was increasing demand from the public for out of hours care and GPs were increasingly unhappy with their rising out of hours workload. To help practices deal with this situation, the government gave GPs the powers to delegate provision of out of hours care to a third party. This allowed GPs to decide whether to continue to provide out of hours care as before, to set up a practice rota, to form out of hours co-operatives with other practices or to make use of a deputising service. 4,5 This was a huge change for practitioners and patients. Many doctors joined local cooperatives and saw patients in out of hours centres in preference to home visits. Cooperatives used telephone consultation much more than in the past and co-ops often employed drivers to improve the efficiency and safety of home visits. 6 1995-1998 In 1998 the Scottish Office published a report detailing progress since the introduction of out of hours co-operatives. 6 The Scottish Out of Hours Study Group collected information from Health Boards to describe the out of hours care available. At that time 75% of the Scottish population was covered by a co-operative but there was wide regional variation. Coverage was almost 100% in Greater Glasgow but was only 40% in Lanarkshire, which made greater use of deputising services. - 4 -

The study concluded that overall patients were satisfied with the new services and the quality of most GPs lives had been improved by the introduction of co-operatives. The report authors raised concerns about the level of variation between co-operatives for instance they differed in opening hours, the number of doctors on duty, IT systems and home visiting rates. Introduction of NHS24 In December 2000 the SEHD announced the introduction of NHS24. This new organisation had a very broad remit. Responding to demand for out of hours care was only one of the reasons NHS24 was set up. It was also intended to provide health information, advice about self-care and a sign-posting role directing the public to the most appropriate service. The new organisation was designed to be integrated with existing out of hours services, A&E, GP co-operatives and the Scottish Ambulance Service. NHS 24 was introduced gradually across the Scottish regions between May 2002 and November 2004. 4,5 The introduction of NHS 24 was made more difficult by the introduction of a sizeable change in the GP contract during the rollout phase. General practices were allowed to opt out of their responsibility for providing out of hours care and from 31 st December 2004 this duty fell to NHS Boards. This meant that while NHS 24 was being introduced the out of hours co-operatives it was integrating with were undergoing reorganisation. NHS24 has become a major route to emergency primary care services in the out of hours period. An evaluation was carried out in 2005 and this demonstrated that 90% of calls received by NHS24 originated out of hours. 4 Data Sources In order to estimate the relative magnitude of the main referral pathways we used data from several different sources including the Scottish Ambulance Service, A&E services and the SMR01 hospital admission record as well as data from NHS24 and local out-ofhours services. It was impossible to describe the national picture of referral pathways using existing routinely collected information. However in one Scottish region, A&E services have started to collect information about the source of referral and outcome for each patient who attends the department. This allowed us to make a reasonable estimate of the relative size of the main referral pathways but it was still a challenge. - 5 -

NHS 24 and Out of Hours providers NHS 24 provided us with national information about all the calls they receive and the immediate patient outcome. However after calls are passed onto other services, NHS 24 do not receive feedback reports from out of hours centres detailing whether patients are referred on to hospital for admission. Two local out of hours services provided us with information about the number of calls they receive and the way these are handled. However out of hours primary care services do not routinely receive feedback from hospitals about the outcomes for patients they refer there. This means that although the services could tell us how many patients they referred to hospital over a period of time we had no way of knowing what proportion of these patients were actually admitted. National A&E Survey We have information about patients admitted as an emergency through A&E for the whole of Scotland from the National A&E survey. The primary objective of the survey is to report on A&E waiting times but it also enabled us to collect information about patients who were admitted through A&E departments. This gives us a partial picture of the flow of patients into hospital. However, there are two limitations to the information gathered in the A&E survey. Firstly it does not include all emergency admissions to hospital because a small number of patients are admitted directly to hospital wards and do not pass through A&E. Secondly, the A&E survey does not record whether patients self-presented, were brought by ambulance or if they were referred by a GP. Scottish Ambulance Service The Scottish Ambulance service was able to give us very rich information from the whole of Scotland about the number of ambulances dispatched according to the referral source. However the service does not receive feedback on outcomes for each patient e.g. were they admitted after being delivered to hospital? NHS Lanarkshire A&E data NHS Lanarkshire were able to provide us with information on the source of referral and outcome for all patients attending A&E Departments. In Lanarkshire all A&E patient outcomes are recorded as Acute Assessment, Minor Injury and Illness, Medical Admission, Surgical Admission or Death. - 6 -

Information was provided for the period between July 2005 and June 2006. This has allowed us to pull together information about the referral flows for all admitted patients, whether they self presented to A&E, were delivered by ambulance to A&E, or were referred by a GP. However, we have excluded a small number of admissions from our analysis. A small number of patients in the Acute Assessment flow are briefly admitted to hospital. These patients were not identifiable in the aggregated data and so have not been included in our analysis. SMR01 An SMR01 record is completed for every episode of care in a non-obstetric, nonpsychiatric hospital in the NHS in Scotland. All SMR01 records are collated and processed by the Information Services Division (ISD) of National Services Scotland (NSS). Estimation of the relative size of the main patient pathways A definitive mapping of patient flows within a local system of unscheduled care would be immensely detailed and complex. It would involve different types of flow between a range of agencies and services. In order to make the task manageable we have limited ourselves to making an order of magnitude estimate of the relative size of the following four main pathways to emergency inpatient admission: a) patient self-refers to A&E and is then admitted as an emergency inpatient b) patient is taken to hospital after calling the emergency services (usually via NHS24 or 999 but without involving a GP) c) patient is referred to hospital by a GP in-hours d) patient is referred to hospital by an out-of-hours service In order to make these estimates on the basis of available data, it was necessary to make a series of assumptions and imputations. The steps involved are outlined below. - 7 -

Step 1: Assumption that all emergency inpatient admissions come in via A&E. Most commonly patients admitted as emergency inpatients arrive first at A&E. However this is by no means universal and patients can be admitted directly as emergencies to hospital wards. Information collected in A&E will not include these directly admitted patients. In order to make use of the A&E data from Lanarkshire we had to make the assumption that the majority of emergency admissions in Lanarkshire passed through A&E first. The assumption was confirmed by local knowledge that the great majority of inpatient admissions in Lanarkshire came via A&E. In addition, we compared the number of admissions recorded in Lanarkshire in the National A&E survey with SMR01 data collected over the same week. This comparison confirmed that the vast majority of admissions are admitted through A&E departments in Lanarkshire. In discussion with local colleagues, it was reckoned that something of the order of magnitude of 5% of emergency inpatient admissions are direct admissions i.e do not pass through A&E. We have not attempted to incorporate these admissions into our overall estimates since we do not have data on them. Their probable effect would be to swing the estimates away from 'self-referral to A&E' and towards GP referral by one or two percentage points. A final check was made by summing the medical and surgical admissions recorded in the local A&E data for April. This gave us a figure similar to the number of admissions recorded in the National A&E survey. Step 2 Identifying patients admitted as emergencies after attending A&E. Patients were identified as emergency admissions on the basis of the patient outcome code in the Lanarkshire A&E data. Two of these codes, Medical Admission and Surgical Admission, pointed unequivocally to emergency inpatient admission. As already mentioned a small number of patients with patient outcome code 'assessment' will have been admitted but it was impossible to identify these separately. Step 3: Referral source of patients admitted via A&E. The next step in the process was to examine the referral source of those patients attending A&E in Lanarkshire and subsequently admitted as an emergency. The A&E data has 17 codes for referral source. These were condensed into the three broad categories of self-referral, emergency services or GP referral. - 8 -

However, two categories of referral source, NHS24 and 'other', amounting to around 10% of the total subsequently admitted were excluded from the analysis. On this basis, we found that of the patients admitted as an emergency, 35% self-referred to A&E, 34% presented via other emergency services and 31% were referred by GPs. Step 4. In hours versus out of hours referrals. The next step was to attempt to distinguish between patients referred in hours and patients referred out of hours. The Lanarkshire A&E data contains information about the time of day patients were seen in the department. Patients are recorded as attending between 9am and 5pm, 5pm to midnight or between midnight and 9am. Although these categories do not exactly match the standard definitions used to describe the period of out of hours care, it was felt that they provide a close enough approximation. 9am to 5pm was used to define the in-hours period. On this basis it was possible to divide the 31% of patients identified as being referred by GPs into 19% referred in hours and 12% referred in the out of hours period. Step 5. Identifying patients referred by GP but delivered by ambulance. Some of the patients initially referred by a GP or out-of-hours service will have been delivered by ambulance and this was recorded in the A&E data as 'emergency services' rather than GP referral. The Lanarkshire A&E data recorded 34% of admitted patients as having been delivered to hospital by Emergency Services. The vast bulk of these will have been delivered by ambulance. For the purposes of this exercise we have assumed that the local Lanarkshire A&E staff have recorded Emergency Services as the referral source for all patients delivered to hospital by ambulance no matter who made the original request for the ambulance. We were able to make use of Scottish Ambulance Service data to estimate what proportion of those recorded as 'Emergency Services' were the result of an initial request by a GP or out-of-hours service. In 2006/7, 52% of ambulances were dispatched following 999 calls (self-referral), 7% were dispatched following referral by other emergency services, 12% were dispatched following contact with NHS 24 and 29% were dispatched following GP referral. - 9 -

In addition the SAS data enabled us to divided this 29% into in-hours and out-of-hours requests. 19% of emergency ambulance dispatches were made during working hours as a result of a call from a GP whereas 10% of emergency ambulance dispatches were made during the out of hours period as a result of a GP call. Step 6. Adding in the GP initiated/ambulance delivered subset. As we have seen, the Lanarkshire A&E data suggests that 34% of the patients going on to emergency inpatient admission were brought to A&E by the emergency services. However we have now established as an order of magnitude that 19% of this 34% (around 6.5%) were delivered by ambulance as a result of an in hours request by a GP and 10% of the 34% (3.4%) were delivered by ambulance as a result of an out-of-hours request. The four main flows In Step 4 it was calculated that 19% of inpatient admission were categorised as GP referrals in hours. To this proportion must be added the 6.5% delivered by ambulance to give an overall proportion of around 25% of inpatient admission initially referred by a GP in hours. Similarly in Step 4 it was calculated that 12% of inpatient admissions were categorised as referrals out-of-hours. To this proportion must be added the 3.4% delivered by ambulance to give an overall proportion of around 15% These patients delivered by ambulance following in hours or out of hours referral must be subtracted from the 34% 'emergency services' to give a category which must consist primarily of these brought to A&E by ambulance without GP involvement. This category now comes to 25%. The four main flows are summarised in Figure 1. - 10 -

Figure 1: Referral pathways of Lanarkshire patients admitted to hospital in an emergency (based on data collected between July 2005 June 2006) Routes to emergency inpatient admission: Lanarkshire initial estimate Self-referral 35% Amb 6% Other 19% Ambulance (999 etc) GP (in hours) 25% 25% A&E Emergency inpatient admission Amb 3% Other 12% OOH services 15% Discussion We have estimated the relative size of emergency admission pathways for one Health Board area. Although all emergency services collect information about the patients they care for, combining this information provides only a partial picture of the pathways leading to emergency admission. In Lanarkshire we were able to create a map of referral pathways as the majority of emergency admissions are admitted through A&E and Lanarkshire A&E services collect information about source of referral. As part of the more general programme of work describing emergency admission patterns across Scotland a literature review was carried out. This provided two useful historical 'snapshots' of the relative contribution of GP referral to emergency inpatient admission. Donaghy et al 1998, 7 carried out a major survey of emergency inpatient admissions across nine health board areas. The study included 1881 adult patients admitted urgently to medical wards (i.e. excluding paediatric patients and surgical patients). This showed that an order of magnitude of 60-80% (depending on Health Board) of those medical admissions were the result of a referral to hospital by a GP. These proportions are of course much higher than those produced by the current exercise. - 11 -

The second information source is an A&E survey carried out in Greater Glasgow in 2001. 8 Information was collected in all A&E departments about all attendees over a two-week period. This survey collected information about source of patient referral. The survey found that 39% of all patients subsequently admitted to hospital had originally been referred to hospital by a GP a result very close to our own. The main limitation with the Glasgow study for our purposes is that we do not know what proportion of all emergency inpatient admissions at the time followed a pathway via A&E. If we give more weight to the much larger study by Donaghy et al., this would add support to the conjecture that in the last few years there has been a decline in the proportion of emergency inpatient admissions which are a result of referral by a GP and an increase in the proportion which are the result of self-referral to A&E or of individuals calling an ambulance themselves. In other words, patients are increasingly bypassing the traditional gatekeepers to acute care. National information systems do not currently collect information about referral sources and this has limited our investigation of national referral patterns. The introduction of an additional referral source field in SMR01 would ensure information is available on all admissions in future. We found that the information collected in A&E was essential for the development of our Lanarkshire map of patient flows. If information as rich as this was collected across the country it would mean referral patterns could be monitored through A&E information. The key to successful collection of referral information from A&E departments is to ensure it is routinely collected for all patients. For this to happen it must be as simple as possible for data collectors to record information about the four main referral pathways, self, ambulance, GP in hours or GP out of hours. However in some areas patients referred by GPs can be admitted directly to hospital wards bypassing A&E and so data collection from A&E would be incomplete in these areas. The inclusion of an additional field in SMR01 could capture information from these patients too. - 12 -

Summary This short exploration of current referral pathways gives a snap-shot picture of current activity. Information from service providers demonstrates that emergency care within our health care system is complex and changing. If regular collation of existing data sources could be ensured this would give insight into how our emergency care system is developing and responding to external pressures so facilitating informed planning for the future. - 13 -

References 1. ISD Scotland. Delivering for Health/Long Term Conditions Programme. Working Paper 1. Trends in rates of emergency inpatient admission among older people in Scotland: a comparative analysis at NHS Board level. 2007. 2. ISD Scotland. Delivering for Health/Long Term Conditions Programme. Working Paper 2. The system of unscheduled care in Scotland: variation in the level of emergency inpatient admission by GP practice. 2007. 3. Scottish Executive Health Department. Building a Health Service fit for the future; Volume 2. 2005: 1-256. 4. Heaney D, O'Donnell C, Myles S, Munro J. NHS 24 Evaluation. 2005:1-164. 5. NHS Quality Improvement Scotland. The Provision of Safe and Effective Primary Medical Services Out of Hours- National Overview. 2006: 3-84. 6. GP out of Hours Services Working Group. GP Out of Hours Services Working Group Report. 1998. 7. Donaghy E, Parker S, Cunningham-Burley S, Walker D. Emergency medical admissions in Lothian: An investigation of factors associated with admissions and attitudes to alternatives. 1998: 1-62. 8. Murray S, Bashir K. Survey of A&E departments in Glasgow in 2001. 2001;Personal Communication: 1-8. - 14 -