Article Text

Download PDFPDF

Patients who leave emergency departments without being seen: literature review and English data analysis
  1. A J Clarey,
  2. M W Cooke
  1. Warwick Medical School, Coventry, UK
  1. Correspondence to Professor Matthew W Cooke, Warwick Medical School, Coventry CV4 7AL, UK; m.w.cooke{at}warwick.ac.uk

Abstract

Objectives The purpose of this review was to determine the rate of those that leave the emergency department (ED) without being seen and their reasons, to clarify if such behaviour poses a health risk, to analyse the impact initiatives have made on the leave without being seen (LWBS) rate, and to discuss the implications of using it as a national performance indicator within the NHS.

Methods A combination of data sources was reviewed: a ‘realistic’ literature review, analysis of hospital episode statistics data from England and a local NHS trust audit.

Major Findings LWBS rates vary across the world, from 15% to 0.36%. Also initiatives to reduce LWBS rates demonstrated mixed outcomes, with reductions in the rate by as much as 96%, while others were ineffective. The most common reason quoted for LWBS was long waiting times and there were few data to suggest LWBS posed a risk to patient health.

Conclusions LWBS is an issue experienced in many countries that has responded in a varying manner to many initiatives in attempts to reduce it; however, it is clearly associated with the waiting times experienced in ED and therefore working within a packet of performance measures it would assess the effect of waiting times from another perspective.

  • Burns
  • clinical care
  • comparative system research
  • despatch
  • doctors in PHC
  • effectiveness
  • intermediate care
  • major incidents
  • operational
  • prehospital care
  • research

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.

Patients who leave without being seen (LWBS) reflect a group of patients whose expectations have not been fulfilled. Exploring the reasons for them leaving may allow these expectations to be met. It will also help to quantify any potential clinical risk from the absence or delay in care they receive. This measure now forms one component of the new set of clinical quality indicators for English emergency departments (ED).1 This paper aims to increase understanding of the extent of the problem, the causes of LWBS, the impact of this decision on the patient and healthcare provision and potential ways of reducing the numbers. It also aims to discuss the utility of using this as a measure of quality of care.

A multi-methods approach is being used. Analysis of national-level routinely collected data will quantify the extent of the problem and give some indication of causes. Analysis of one NHS organisation's locally collected data will give more detail, while accepting the limitations of data generalisation. Finally, a literature review considers the latest research on LWBS patients; assessing the varying rates in different ED, the patient and hospital factors associated with LWBS patients (including acuity) and the consequences of such a patient group. In addition, the review will look at various solutions trialled in ED and the effect they may have on LWBS rates.

Methods

English national data

Routinely collected data were obtained from ED hospital episode statistics (HES) for the period of 2008–9 relating to the number of patients that leave ED without receiving treatment; it considers the primary diagnosis of such patients. Also it enables trust comparison and the calculation of a national average. These data do, however, have significant quality and coverage issues.

Local data

Routinely collected data from one NHS trust ED information system was obtained from 1 June 2010 to 14 June 2010 relating to LWBS patients; considering their presenting complaint, mode of arrival, their age, gender and re-attend status.

Literature review

A realistic review approach was taken for the review process2 as this was considered more appropriate for policy-making research than a systematic review. This consists of four stages: (1) mapping the required knowledge area; (2) purposive searching for evidence; (3) iterative approach; (4) focusing results on what is required by policy-makers. The required knowledge area was informed by one author (MWC) conferring with (and leading) the core team developing the national clinical indicators in England, and determining their need to understand LWBS using the parameters stated in the aims. An electronic search was conducted of the biographical database Medline and the Cochrane Library for systematic reviews in June 2011. The search used the following words and phrases: ‘leave/left without being seen’, ‘leave/left without being treated’, ‘leave/left without receiving treatment’, ‘leave/left before being seen’ and ‘LWBS’ in combination with the keyword ‘emergency’. This narrow search was verified by checking bibliographies for other terms and no other terms were found to be necessary. In addition to this the bibliographies of all included articles were reviewed for additional articles that matched the search criteria.

For inclusion, papers had to be written in English and either provide original data on patient LWBS rates in ED or determine patient or organisational concepts for LWBS rate improvement. The abstracts of the papers were reviewed by one author (AJC) to determine whether they fulfilled the inclusion criteria.

Results

The search method produced a total of 36 papers of which, after reading the abstracts and titles, 29 were considered suitable for inclusion in this study; 26 containing original LWBS data and three only discussing causative factors or initiatives for improvement. All of these were then reviewed thoroughly according to the principles of realistic review and are detailed in the literature review below.

There was a great diversity in the LWBS rates reviewed from 15% to as low as 0.84%, and the additional papers discussed various attempts to reduce LWBS rates, concepts for further improvement or a review of predicted causative factors.

National data

These data showed a national LWBS rate average across 167 English NHS trusts of 3.26% LWBS, which equates to 449 924 patients annually. Similar to the data found in the current literature, there was significant diversity between hospital trusts; ranging from 14.5% to 0%.3 As mentioned above the quality of these data is poor: it is recognised that some organisations only submit part of their dataset to HES. Therefore, although an average may be relatively acceptable, statistically a comparison between trusts would be ineffective. These data also clarified the arrival mode of ED patients; 27.1% of LWBS patients used the emergency ambulance service (999) to access the ED, and LWBS patients accounted for 3.6% of all patient arrivals to ED by an emergency ambulance.

Local NHS data

These data showed that in the study ED, with a census of 114 367 new patients per annum, there was a 4.32% LWBS rate across a 1-week data-assessment period. Within this group, 12.7% re-attended within 7 days of their initial presentation to the same ED; however, these data cannot demonstrate the presentation of the LWBS patients at other healthcare-providing facilities. In concordance with the national data, 22.5% of LWBS patients used an emergency ambulance to arrive at this ED. Sixty-two per cent of the LWBS patients were men, which does not correlate with the current literature, showing no significant gender skew in this patient category. The most common LWBS presentation was musculoskeletal injuries (24.5%) but with relative diversity across other categories of presenting complaints. This hospital does not use triage categorisation and so in this respect is not comparable with current literature.

Literature review

LWBS rates

The majority of research has been conducted in the USA, with rates varying from 15%4 to 0.84%,5 indicating that within one healthcare provision system a variety of factors, both patient and hospital, can influence the rate of patients that LWBS. As demonstrated in the studies listed below, the LWBS rate can vary quite significantly between and within countries (table 1).

Table 1

The LWBS results from the papers incorporated into the literature review

Patient-associated factors

Several papers demonstrated that patients within specific sociodemographic groups were significantly more likely to LWBS. Several papers showed that the highest proportion of LWBS patients were young; under 29, 18–24 and 15–44 years were all identified as high-risk groups.6 ,9 ,20 In addition, other high-risk groups were those in a low socioeconomic status group or without private health insurance10; 11 ,21 the latter risk factor is generally indicative of those in lower socioeconomic status groups; however, it is somewhat redundant when considering the current NHS ED provision. Being non-English speaking further increases the probability of a patient LWBS.20 There is an association between previous LWBS episodes and recurrent episodes; 17% of LWBS patients have LWBS in the previous year and there is an OR of 48.02 if there were four or more LWBS episodes in the previous year.9

Hospital-associated factors

Patients frequently list excessive waiting times as the foremost reason for LWBS; between 44.8% and 86% of interviewed patients listed this as the primary reason for choosing not to stay for treatment.4 ,12 In addition to this there is a correlation between LWBS rates increasing and ED overcrowding,22 peak ED times,10 the lead physician not being trained in emergency medicine,13 the facility being a large (>30 000 census) hospital,14 teaching or trauma centre21 and the arrival of trauma patients.15 These factors may act directly on a patient's choice to LWBS; however, all are likely to increase the waiting time, which is a significant contributor to LWBS, as stated above. Temporal variance also exists; weekends show increased LWBS rates, as do public holidays; again this may be indirectly correlated with waiting times.7 ,16

Risks and severity of LWBS

There is significant correlation between LWBS rates and triage levels; as the triage level decreases then the proportion of patients that LWBS increases. Research shows a dose–response relationship between LWBS and triage; with 0.1% of highest level patients and 15.2% of lowest level patients LWBS.9 Additional research shows there is a 58.3 times increased risk of LWBS between emergency and non-urgent triage levels.20 The varying triage systems used make comparison difficult; however, the proportion of patients in higher categories does vary; some demonstrate as few as 4% are in the urgent category,6 while others show 49% of LWBS patients required urgent treatment or treatment within three hours.4 Although some adverse events have occurred with LWBS patients (single event recorded in a study) there is little evidence that seriously ill patients LWBS and therefore experience significant adverse incidents.12 A recent study in Canada demonstrated no significant adverse incidents occurred in an initial 7-day period to those that LWBS, and instead demonstrated a link between long waiting times and poorer clinical outcome.19 Data imply that a significant proportion of patients that LWBS from an ED do attend other healthcare services; 56.9% attend their general practitioner and 22% return to the ED.6 ,14 Eleven per cent of LWBS patients have required hospital admission.17 Of greater concern is the 15.3% who stated they would not return to the same ED again.18

Interventions and their effects

A variety of interventions has been trialled in attempts to improve LWBS rates as well as additional ED quality indicators (table 2). A variety of fast-track units has been implemented in ED; these manage non-urgent patients separately from the main ED, and aim to provide rapid treatment that enables streamlining of the patients admitted to the main ED. Such fast-track units have confirmed efficacy, with reductions in LWBS rates and additional indicator improvements such as decreased waiting times and return rates.23–25 Other effective interventions have included the use of an acute care unit, which halved the LWBS rate, also the use of a physician in triage and the initiation of ancillary investigations and treatment at triage all reduced the LWBS rates.26–28 A large-scale and combined human resources and facilities restructuring showed a significant LWBS rate reduction by improving waiting times.29 A similar restructuring in a Spanish study, however, did not demonstrate an LWBS rate improvement, irrespective of other quality indicator improvements.30 Other staffing alterations have also failed to reduce the LWBS rate; including increasing the number of emergency medicine attending physicians.15 In follow-up studies patients suggested that improved communication of waiting times/delays and alterations that would reduce waiting times would have prevented them LWBS.5 ,6

Table 2

The details of included papers that researched interventions to improve LWBS rates

Limitations

The national HES data are of variable quality; it is recognised that many ED patients are not included. Coding is undertaken by the clinical staff who may be untrained and so the reliability of coding as LWBS is not known. This latter issue also applies to the local study. The cross-referencing and similarities between the three data sources aim to triangulate data and thus reduce the effects of poor data quality.

The literature review relates to several different healthcare systems and international applicability is unknown.

Conclusions

What is the extent of the problem?

The use of the mixed methods approach has been able to confirm that rates of LWBS of 3–8% are recorded in the UK with marked variation, although this may be a reporting issue with HES. However, international evidence supports this finding of wide variation. It is clear that although this is a common issue for many ED there are examples in which the number of LWBS patients is extremely low. There is evidently great diversity between NHS trusts in England and in ED around the world; some English trusts recording 0% and ED in Hong Kong noting a rate of only 0.36%.21 The diversity in LWBS proportions could be due to the challenges and differences in data collection and definition for this category of patient; however, it may also be indicative of local policies, variation in waiting times or cultural differences that may be influential in tackling this issue.

There are two aspects to consider: the hospital factors that promote high LWBS rates and patient factors that increase their probability of LWBS.

As one would expect, excessive waiting times are the predominant causative factor listed by patients for choosing to LWBS; factors that further promote higher LWBS rates include periods of ED overcrowding, lack of department management by an emergency medicine trained physician and temporal factors including seasonal and holiday-associated variance. The consequence of these additional factors is an increase in waiting times, which inevitably raises LWBS rates. The fact that waiting times are so important in determining LWBS means that data from overseas may not be applicable in the UK, where total time in the ED is significantly less than in the USA, where most of the data originate.

The literature suggests risk factors that increase a patient's likelihood of LWBS; being in a lower socioeconomic status group, being young and not speaking English (where English is the indigenous language). In addition, there is correlation in patients who have previous episodes of LWBS; suggestive of a pattern in healthcare-seeking behaviour. Although of little relevance for the NHS, data indicate a further increased risk of LWBS if a patient has no health insurance.

Is this a clinical risk?

There is obvious concern for the health of patients who choose to attend an ED yet decide not to remain for thorough assessment and/or treatment; however, the risk of significant clinical incidents in this category appears to be low; only one article reported a single death within 6 days of a patient LWBS. However, the local data did show that one in eight return within 1 week. Patient follow-up studies indicate varying rates of re-attendance to ED and presentations to other healthcare providers. Similarly, there is correlation between triage levels and LWBS rates, which again reduces clinical concern. Perhaps of more serious concern for patient health is the consequence of a LWBS patient's experience being that they may not choose to attend that ED again.

Other implications for the healthcare system

This issue presents several issues to the NHS beyond that of clinical risk: financial implications due to repeated episodes of presentation due to failed management of initial presentation at the ED and reputational damage in trusts with significantly higher LWBS rates. Some LWBS patients are frequent attendees and therefore high users of health care; undertaking case management may be able to reduce the burden they create on the health service.

In addition, there is the previously mentioned issue with regard to ambulance usage by this category of patient. The cross-referencing of the data sources suggests that between a quarter and a third of LWMS patients arrive by ambulance. The impact on the healthcare system is twofold with the potential misuse of the ambulance service, additional costs for these incomplete episodes of illness and also the potential inability of the ambulance service to respond to additional 999 calls while responding to a LWBS patient.

Ways of reducing LWBS

The core principle behind all initiatives trialled to reduce LWBS rates is to reduce the waiting time between arrival/triage and treatment. A variety of ED initiatives has been successful: the creation of fast-track units, allowing triage nurses to provided basic treatment and to initiate investigations and using a physician to triage. The greater effects have been found in schemes that restructured both human resources and facilities within the ED, while approaches that focused on increasing the number of ED medical staff or administrative staff alone were unsuccessful, often attributed to the bottleneck effect as with only one area of improvement the bottleneck shifts rather than resolves. It is likely that the interventions were specific to the departments studied, and therefore introducing specific interventions to reduce LWBS may be ineffective in other departments.

Utility as a quality improvement indicator

LWBS rates are often an illustration of the waiting times of an ED; however, the local culture and patient sociodemographic factors are greatly influential on this figure, and these features go beyond the capacity of an ED to manage.

With the correlation between triage levels and LWBS rates and the extremely low level of significant clinical incidents associated with LWBS patients, then this figure is perhaps more indicative of the use of ED for healthcare needs that could well be managed more appropriately by other aspects of a health service. The higher LWBS rates during out-of-hours periods and the relatively high presentation rate to general practitioners and other community-based healthcare providers after the LWBS episode support this theory. It may therefore be a better indicator of whole systems flows than ED performance, when used in a system with low ED waits. LWBS may also help to triangulate other quality measures, being a measure of waits and experience.

The achievability of a 0% LWBS rate is also questionable; even with sound population education regarding appropriate healthcare-seeking methods and low waiting times there will be a proportion of patients that still LWBS, therefore it should be used as an indicator rather than as a standard in which minimisation is seen as the goal.

When considering the evidence for LWBS rate improvement there would be significant costs; facility alteration and staff level improvement would require a relatively large investment in the ED. Such alterations have the potential to decrease the LWBS rate; however, if the majority of LWBS cases are more suitable for presentation to other aspects of a health service then a broader scheme for urgent healthcare access may have a greater impact on both the system and patient health.

Finally LWBS are relatively difficult to measure, with varying recording mechanisms and the nature of the issue; a national standard regarding the definition of a LWBS patient is now available in England and may help support analysis in the future.

In the USA the national quality forum has included LWBS rates as one of 10 indicators of performance in an attempt to drive forward ED efficiency allowing analysis of patient flow to improve services.31

The lack of UK studies and its adoption as a national quality indicator (with consequent improving quality of HES data) mean that further analysis of the group of patients in the NHS is warranted.

Acknowledgments

We would like to thank Jo Flowers and Sivakumar Anandaciva for help in obtaining the data.

References

Footnotes

  • Competing interests None.

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

Linked Articles

  • Primary survey
    Jonathan Wyatt