Article Text

Telephone triage of young adults with chest pain: population analysis of NHS24 calls in Scottish unscheduled care
  1. Peter Hodgins1,
  2. Megan McMinn1,
  3. Matthew James Reed1,2,
  4. Stewart William Mercer1,
  5. Bruce Guthrie1
  1. 1Centre for Population Health Sciences, The University of Edinburgh Usher Institute of Population Health Sciences and Informatics, Edinburgh, UK
  2. 2Emergency Medicine Research Group Edinburgh (EMERGE), Edinburgh Royal Infirmary, Edinburgh, UK
  1. Correspondence to Professor Bruce Guthrie, The University of Edinburgh Usher Institute of Population Health Sciences and Informatics, Edinburgh EH8 9AG, UK; Bruce.Guthrie{at}ed.ac.uk

Abstract

Background Telephone triage is increasingly used to manage unscheduled care demand. Younger adults are frequent users, and commonly call with chest pain. We compared pathways of care in younger adults calling with chest pain, and associations of patient characteristics and telephone triage recommendation with hospital admission.

Methods A retrospective study of all triage calls with chest pain to NHS24 advice line by people aged 15–34 years between 1 January 2015 and 31 December 2017 where chest pain was recorded as the call reason. Recommended outcome and subsequent use of services were determined using the continuous urgent care pathways (CUPs) database which records single episodes of care spanning multiple services. We determined the number of services involved, the proportion of patients with inpatient admission, those with an admission for an ‘acute-and-serious’ diagnosis, and the association between the triage call recommendation and these outcomes.

Results There were 102 822 CUPs identified, with 1251 different combinations of services. The most common pathway was an NHS24 call then attendance at a primary care out-of-hours (PCOOH) centre, accounting for 38 643 (37.6%) CUPs. 9060 (8.8%) CUPs ended with hospital admission, 3030 (3.0%) the result of an ‘acute-and-serious’ diagnosis. 8453 (8.2%) were given ‘self-care’ advice and not referred further, while 46.9% ended at PCOOH and 15.2% at ED. ‘Asthma, unspecified’ was the most frequent ‘acute-and-serious’ diagnosis. Compared with people given self-care advice, referral to other services had increased odds of inpatient admission (adjusted OR (aOR) for ambulance called 28.7, 95% CI 22.6 to 36.3; for 1-hour in-home general practitioner (GP) visit arranged aOR 36.8, 95% CI 23.2 to 58.5) and for admission with an ‘acute-and-serious’ diagnosis (aOR ambulance called 23.9, 95% CI 16.2 to 35.4; aOR 1-hour GP visit 48.3, 95% CI 25.5 to 91.6).

Conclusion Chest pain triage by NHS24 appears safe, but care pathways can involve multiple service contacts. While acuity assigned to the call is strongly related to the odds of hospital admission and odds of an ‘acute-and-serious’ diagnosis, ‘overtriage’ means few patients are directed to self-care advice.

  • chest - non trauma
  • emergency care systems
  • primary care
  • triage
  • urgent care
  • access to care

Data availability statement

Data may be obtained from a third party and are not publicly available. Unable to share data. Access requires permission from the Public Benefit and Privacy Panel for Health and Social Care (HSC-PBPP).

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Key messages

What is already known on this subject

  • The unscheduled and emergency care system in the UK consists of a number of different interlinked services. The use of all services is increasing.

  • The recommended access to unscheduled care in the UK is via a single telephone point of access, NHS111 (NHS24 in Scotland), which triages access to other services. Evidence as to the effectiveness of this form of triage is lacking.

  • Chest pain is a common reason for young adults to contact NHS24.

What this study adds

  • In this retrospective study of calls to NHS24 in Scotland, 1 in 40 calls are by young people aged 15–34 years with chest pain.

  • NHS24 triage of undifferentiated chest pain in young adults identified 8.2% as safe to ‘self-care’, and not requiring onward referral. The 8.8% of contacts ended with an inpatient hospital admission.

  • Unscheduled care pathways and service interactions are complex, with 1251 different pathways identified in this study.

Introduction

Use of unscheduled care is increasing worldwide, leading to pressure on healthcare systems and concerns about safety in crowded EDs.1 2 A potential solution is the use of telephone advice lines, whereby trained call handlers, aided by clinical decision computer software, provide advice and triage those looking to access unscheduled care. Whether this non-clinical, algorithmic assessment of symptoms reliably discriminates severity and need for intervention is uncertain.3 4

In the UK, unscheduled care refers to any unplanned health contact with the NHS,5 and includes urgent and emergency care. One suggested cause for increased use of unscheduled care is that easier access has led to more people making contact.2 6 Increased awareness among the public of serious conditions may also contribute, in the context of patient difficulty in knowing when symptoms are ‘urgent’ (not life-threatening but too serious to wait for a routine appointment).7–9

NHS111 (NHS24 in Scotland) is a key gateway by which urgent care services are accessed in the UK, with open access by dialling 111. It offers both health advice and triage.9 NHS24 calls are taken by trained call handlers supported by a Clinical Decision Support System (NHS Pathways). Based on what the patient says, the call handler’s choice of symptom triggers an NHS Pathways algorithm prompting further questions, in turn leading to advice on appropriate disposition. The call handler is not focused on formulating a differential diagnosis but instead on identifying key symptoms, the algorithm determines the nature and type of subsequent questions.

The term ‘triage’ originated in the battlefield, where the wounded were prioritised for treatment based on the likelihood of death if not dealt with promptly.10 In unscheduled care, triage is about identifying symptoms requiring a time-critical intervention. Chest pain is often difficult to triage, most causes are not serious, but a minority of people have conditions requiring rapid treatment to minimise the risk of death or significant morbidity. Understanding NHS24 management and outcomes for callers with chest pain is important to evaluate safety and effectiveness of unscheduled care telephone triage.

A prior analysis of calls made to NHS24 in 2011 found people aged 15–34 years old were the largest number of contacts of any age group (26.9% of the total). Chest pain was the third most common reason to call with over 25 000 calls (after abdominal and dental problems).11 In this group, the total number of deaths from cardiovascular–respiratory causes was 84, compared with 3870 deaths in the 65–74 age group.3 Chest pain is therefore a common concern in younger people with high workload implications, yet few, but still serious, critical events.12 13

This study aimed to assess the effectiveness of telephone triage in the UK in identifying patients needing urgent attention—by examining pathways of care and hospital admission outcomes in a young adult population calling NHS24 with chest pain.

Methods

Study design and data sources

The design was a retrospective, population study using the Scotland-wide Urgent Care Datamart (UCD) which links data from NHS24, the Scottish Ambulance Service, primary care out of hours (PCOOH), EDs, acute hospital admissions (Scottish Morbidity Records 01 (SMR01)), mental health admissions (SMR04) and National Records Scotland death registration. The Community Health Index (the NHS Scotland unique patient identifier) number was used to link all datasets. Linkage was carried out by NHS Scotland Information Services Division. Data on contacts are electronically collected by all services and are complete.14 A continuous urgent care pathway (CUP) is a series of unscheduled care contacts for a single person.14 CUPs terminate when there is no further contact with any unscheduled care service for 24 hours. Each emergency and urgent care service is assigned a code letter: N=NHS24, O=PCOOH, S=Scottish Ambulance Service, E=ED, A=acute hospital emergency admission, M=mental health non-respite admission. CUP names are the chronological sequence of contacts. For example, CUP NSEA represents an NHS24 call, followed by Scottish Ambulance Service call, ED attendance and finally acute hospital emergency admission (figure 1).

Figure 1

Continuous urgent care pathways (CUPs) by (A) NHS24 planned initial disposition; (B) Final disposition.

Setting and patients

The cohort consisted of all callers aged 15–34 years to NHS24 between 1 January 2015 and 31 December 2017 with ‘chest pain’ recorded as the call reason. Call reasons do not discriminate between types of chest pain, including traumatic and non-traumatic causes. Demographic information, NHS24 call recommendation (the action recommended to the patient or taken by NHS24), the CUP pathway, discharge diagnosis if admitted (categorised using IInternational classification of diseases 10th revision [ICD-10]) and mortality data were extracted for all eligible patients.

Outcomes

Mortality in the 7 days after the ending of each CUP was examined, including cause of death. For patients given self-care advice, the number subsequently admitted over 7 days was examined, including discharge diagnoses. Two outcomes were examined: (1) acute inpatient admission, and (2) acute inpatient admission with an ‘acute-and-serious’ disease defined as discharge diagnosis of a condition requiring urgent treatment. Both were examined in association with NHS24 call recommendation (online supplemental appendices 1 and 2).

Independent variables

Other variables examined were age, sex, socioeconomic status (defined by the Scottish Index of Multiple Deprivation (SIMD) grouped into quintiles), urban/rurality (defined by the Scottish Executive Urban Rural Classification (SEURC) grouped into primary city/urban, accessible, remote and very remote, where the latter three categories are defined by drive time to an urban area of <30 min, 30–60 min, >60 min, respectively).

NHS24 call recommendation (the action recommended to the patient or taken by NHS24) was defined based on the records whether the caller was discharged with advice or referred to another service and, for some referral services, the time within which the caller should be seen. Recommendation was categorised as self-care advice; emergency ambulance; self-transport to the ED; general practitioner (GP) home visit within 1 hour, 2 hours or 4 hours; PCOOH attendance within 1 hour, 2 hours or 4 hours; or patient to make a scheduled appointment later.

Data analysis

Data were managed and analysed using Excel, R and SPSS v 24. Demographics, mortality and cause of death were described using summary statistics. Pathways between initial NHS24 contact and the final outcomes (discharge or admission) were described. For the binary outcome variables (admitted, and admitted with an ‘acute-and-serious diagnosis’), logistic regression modelling examined associations between outcomes and NHS24 initial disposition, gender, age group, SIMD and SEURC.

Patient and public involvement

No patient or public involvement.

Results

Demographics of callers

Between 2015 and 2017, 102 822 CUPs were initiated by calls where chest pain was the recorded call reason, in callers aged 15–34 years (table 1), constituting 10.4% of 991 036 calls to NHS24 in this age group, and 2.5% of all 4 033 935 calls across all age groups. A total of 36.9% of these were male, 63.1% female. A total of 35.1% of calls were made by people living in the most deprived quintile of postcodes, with a stepwise decrease to 10.6% from people living in the least deprived quintile.

Table 1

Characteristics of young adult continuous urgent care pathways (CUPs) starting with an NHS24 call for chest pain

NHS24 call outcome recommendations

In total, 26 151 (25.4%) of calls were considered not requiring an urgent care need, with 17 698 (17.2%) advised to attend an in-hours service the next working day and 8453 (8.2%) advised self-care (table 2 and figure 2). Most commonly recommended was referral to PCOOH with 53 619 calls (52.1%). A total of 6091 (5.9%) calls ended with advice to self-transport to ED, and 11 758 (11.4%) were advised to call an emergency ambulance.

Table 2

OR and adjusted OR for all admissions

Figure 2

NHS24 Disposition (reference category is ‘self-care’). Associations between NHS24 planned initial disposition and hospital admission.

The most common pathway within a CUP was an NHS24 call followed by attendance at PCOOH (NO), 37.6% of all CUPs (table 1). Second most frequent was a single call to NHS24 (N—23.7%) with no further attendance, and third an NHS24 call followed by an ED attendance (NE—6.5%). A total of 1254 different CUP combinations were identified, the longest consisting of 62 service contacts (table 1). A total of 9060 calls (8.8%) were part of a CUP ending in an acute hospital admission, with 3080 (3.0%) having an acute-and-serious diagnosis. Of those advised to seek care at another service within 24 hours, 18.7% did not. Of those who were advised to self-care, 10.6% attended another service within 24 hours. Among callers advised to attend ED, 19.9% did not attend (figure 1 and online supplemental appendix 3). Self-care advice was recommended for 8453 (8.2%) of CUPs, and 73 (0.9%) of these patients were subsequently admitted to hospital within the duration of the CUP.

In both univariate and adjusted analysis, all NHS24 recommendations were strongly associated with increased odds of admission compared with self-care advice (table 2 and figure 2). In adjusted analysis, NHS24 referral to Scottish Ambulance Service was associated with higher odds of admission (OR 28.7, 95% CI 22.6 to 36.3) compared with self-care advice, than those told to self-transport to the ED (OR 17.8, 95% CI 14.0 to 22.7). Referral for a GP home visit was associated with higher odds of admission than those referred to the PCOOH centre, with increased odds for more urgent referrals (compared with self-care advice, OR 36.8, 95% CI 23.2 to 58.5) for home-visit in 1 hour vs 36.8 (95% CI 26.3 to 48.9) for 2 hours, vs 19.0 (95% CI 14.1 to 25.5) for 4 hours; OR 18.5 (95% CI 14.5 to 23.6) for PCOOH attendance in 1 hour, vs 15.4 (95% CI 12.1 to 19.5) for 2 hours, vs 7.9 (95% CI 6.2 to 10.0) for 4 hours). Compared with self-care advice, scheduled appointments were associated with the lowest odds of admission to hospital (OR 5.5, 95% CI 4.3 to 7.0). Weak but statistically significant associations were noted between odds of admission and increasing age (OR 1.3, 95% CI 1.2 to 1.4) and increasing socioeconomic deprivation (OR 1.3, 95% CI 1.2 to 1.4) (table 2).

Clinical outcomes

For the 9060 CUPs that ended with acute inpatient admission, 834 different ICD-10 codes were recorded across 18 ICD-10 chapters (data relating to ICD-10 codes available on request). The most frequently assigned ICD-10 code was ‘chest pain, unspecified’ in 7.2% of admissions, followed by ‘asthma, unspecified’ in 7.1% and ‘other chest pain’ in 6.3%. Of the 3080 CUPs ending in admission with an ‘acute-and-serious’ diagnosis, the most frequently assigned ICD-10 code was ‘asthma, unspecified’ accounting for 653 (21.2%), whereas an acute myocardial infarction was diagnosed in 0.1% and cardiac arrhythmia in 0.4%.

Five hundred eighty patients were admitted to critical care: 95 (0.9%) to the intensive care unit, 311 (3.1%) to the high dependency unit and 174 (1.7%) to coronary care (online supplemental appendix 1). The most frequently coded diagnosis was ‘type 1 diabetes mellitus with ketoacidosis’ accounting for 99 cases followed by ‘asthma, unspecified’ accounting for 50 cases (online supplemental appendix 1).

In total, 99 (0.1%) patients who were given self-care advice were admitted in the 7 days following the ending of their initial CUP, 27 (0.3%) of these with an ‘acute-and-serious’ diagnosis. Although, compared with self-care advice, all other NHS24 recommendations had higher odds of admission (table 3 and figure 2). In univariate and adjusted analysis where the outcome was an admission with an ‘acute-and-serious’ diagnosis, attendance by ambulance was associated with higher odds of such a diagnosis than referred to self-transport to the ED (OR 23.9 (16.2 to 35.4) vs OR 14.7 (9.8 to 22.1)) (table 3 and figure 2). Home visits were associated with the greatest odds of serious diagnosis. In addition, there was a positive association with the level of urgency assigned (1-hour home visit OR 48.3 (25.5 to 91.6) vs 2-hour home visit OR 39.5 (24.5 to 63.6) vs 4-hour home visit OR 14.4 (8.7 to 23.7)). A positive association between urgency linked with a PCOOH attendance and odds of serious diagnosis was also observed (1-hour PCOOH attendance OR 20.3 (13.6 to 30.4) vs 2-hour PCOOH attendance OR 17.3 (11.7 to 25.8) vs 4-hour PCOOH attendance OR 6.4 (4.3 to 6.5)). There was an association between SIMD and odds of admission, with those in the most deprived quintile being most likely to be admitted to hospital. The question as to whether remoteness of the caller, in terms of drive time to reach an urban centre, has an influence on the odd of admission or, odds of a serious diagnosis, was examined but no association was found.

Table 3

OR and adjusted OR for a final diagnosis of acute-and-serious disease

Twenty-four (0.03%) patients died within 7 days of the end of a chest pain CUP (online supplemental appendix 2). No deaths occurred in patients who had been given self-care advice by NHS24 and two occurred in those advised to attend a scheduled service.

Discussion

Chest pain represents a symptom of significant concern for young people, accounting for 102 822 calls to NHS24 over a 3-year period (1 in 40 of all calls). Nearly 9% of these calls are part of a series of unscheduled care contacts that end with an inpatient admission, although many admissions end with a non-specific diagnosis. Calls were more common from young adults living in more deprived areas, with those living in more deprived areas more likely to be admitted to hospital than those living in more affluent areas.

NHS24 triage to self-care was effective in identifying people at very low risk of admission (although only 8.2% of calls were triaged to this recommendation). A further 17.2% were referred to in-hours care reducing pressure on other unscheduled care services. There were clear gradients in the odds of admission in relation to the urgency of other NHS24 recommendations, in terms of the service referred to, and the recommended time to respond for GP home visits and PCOOH attendances. Although two deaths within 7 days of first contact were observed in the 17 698 people given advice to attend scheduled services, none was observed in the 8543 given self-care advice. Finally, the wide range of diagnoses associated with an initial presenting symptom of chest pain highlights the challenges faced in first-contact triage of undifferentiated symptoms.

The key strength of the study is that the UCD allows for a comprehensive analysis of linked unscheduled care attendances for the entire Scottish population. Limitations include that we cannot be certain that linked episodes in a CUP are actually for the same problem, since CUPs are defined as sequential episodes over a short period of time—a patient advised to ‘self-care’ who subsequently re-attends within 24 hours, with or without the same symptom, will still be included. Information pertaining to the content of CUPs was complete but there were missing data with respect to NHS24 call recommendation. A further weakness is that CUPs are for episodes linked over a 24-hour period, so a related call with a longer gap will not be linked. In addition, the selection of calls was determined by the call handler’s interpretation of the patient’s symptoms triggering the use of the chest pain algorithm, which may mean that some ‘chest pain’ presentations are not included if another algorithm was used. Finally, we have limited information about the caller’s background health status, so it is unclear the extent to which admissions are for new chest pain or relate to an underlying illness presenting acutely.

The overall number of callers with chest pain is consistent with previous research, as is the finding that the majority of young adult callers with chest pain are women.2 Work on referral outcomes in all age groups in NHS111 in England has found similar proportions requiring ambulance but a higher proportion referred to self-transport to ED (12.8% vs 5.9% in this study).12 Asthma accounted for 7.9% (718) of admissions in the period studied, a small minority of approximately 1500 acute annual admissions with asthma in Scotland for the same age group in 2016.13 An acute myocardial infarction was diagnosed in 0.1% of admitted patients and cardiac arrhythmia in 0.4%. Data from surveys of the general population in the USA have estimated the prevalence of myocardial infarction in the 20–39 age group at 0.3% for both men and women.15 Studies of primary care patients with undifferentiated chest pain have estimated coronary artery disease to be the underlying aetiology in 2%–3%.16 17

Even in this young age group, a significant proportion of those presenting to NHS24 telephone triage are considered in need of hospital admission by a GP or ED doctor, or have an ‘acute-and-serious’ final diagnosis. Assessment by NHS24 call handlers appears safe and reliable when assessing the urgency of undifferentiated chest pain presentations in younger patients. However, safety relies on rapid access to other services, both PCOOH and ED, whose workload is affected by NHS24 triage. NHS24 triage for further urgent assessment had a sensitivity of 89.8% for admission in the same CUP, with a positive predictive value (PPV) of 10.9% (online supplemental appendix 3). Research to improve PPV without compromising sensitivity is important. A deeper understanding of care pathways that involve multiple service contacts is required, as is an understanding of pathways where there is divergence from NHS24 advice. This knowledge may help reduce service contact, thereby minimising duplication of assessment and social contact, a key benefit in the current COVID-19 pandemic.

The pandemic may have a long-lasting impact on how we access medical care, with longer term effort to minimise crowding in waiting rooms. Within the NHS, the ‘Think 111’ campaign is encouraging the public to use 111 as a single point of access for all unscheduled care services. The question of whether it triages as effectively across different age groups and for different presenting symptoms is an area for future research, as is the impact it has on safety and patient flow to other services. Comparing the cohort who phone NHS24 with those who self-present to ED and to PCOOH would allow further examination as to whether initial telephone triage could be a permanent solution for those attempting to access unscheduled care, irrespective of service, thereby minimising the total number of social contacts.

Whether all 9060 inpatient admissions in this cohort are actually required is a topic for future research, since there may be opportunities for more effective triage in PCOOH and the ED. Furthermore, the underlying health issues of those presenting with chest pain require closer examination, particularly those triaged to GP home visit who had the highest probability of admission. Further study may allow for preventative measures and additional support for those with underlying health issues that necessitate a home visit in the first instance.

Conclusion

This study found that chest pain in young adults accounts for a significant number of contacts with NHS24, resulting in a complex variety of different care pathways, and that nearly 10% of such contacts ended with an admission to hospital. NHS24 triage appeared effective, in that the level of acuity assigned to the outcome of the initial NHS24 contact was very strongly related to the odds of admission to hospital, and the odds of the patient having a diagnosis requiring timely medical intervention.

Data availability statement

Data may be obtained from a third party and are not publicly available. Unable to share data. Access requires permission from the Public Benefit and Privacy Panel for Health and Social Care (HSC-PBPP).

Ethics statements

Patient consent for publication

Ethics approval

The NHS Scotland Public Benefit and Privacy Panel for Health and Social Care approved the project (reference 1617/0307).

Acknowledgments

NHS Scotland and Public Health Scotland for providing the Urgent Care Datamart. PH was funded by NHS Education for Scotland as a GP Clinical Academic Fellow. The funder had no role in the conception, design, or conduct of the research, or the decision to submit for publication.

References

Supplementary materials

Footnotes

  • Handling editor Edward Carlton

  • Twitter @PeteHodgins, @mattreed73

  • Contributors All authors contributed to the writing of this paper. BG is the guarantor.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.