Article Text

Download PDFPDF

Characteristics of frequent attenders in an emergency department: analysis of 1-year attendance data
  1. L Moore1,2,
  2. A Deehan1,
  3. P Seed1,
  4. R Jones1
  1. 1
    Department of General Practice and Primary Care, King’s College London, UK
  2. 2
    Clinical Governance Department, Southwark PCT, London, UK
  1. Dr L Moore, Department of General Practice and Primary Care, 5 Lambeth Walk, London SE11 6SP, UK; lynda.moore{at}southwarkpct.nhs.uk

Abstract

Background: There is a significant literature examining the reasons why patients attend emergency departments frequently. This body of work suggests that sociodemographic characteristics are important in understanding why patients re-attend, but it does not provide a definition of what frequent attendance means. This paper aims to identify personal and attendance factors associated with frequent attendance at an emergency department.

Methods: One-year emergency department attendance data from a south-east London teaching hospital (2006–7) were analysed. The dataset was analysed at two levels: the individual patient level and the attendance level. Frequencies and cross-tabulations were produced to describe the dataset. Confidence intervals were calculated for both patient and attendance level data.

Results: 82 812 patients made 117 187 attendances to the emergency department during 1 year. Each patient made an average of 1.4 attendances; 46% were repeat attendances. The analysis demonstrated differences in the personal and attendance profile of patients who attended the emergency department more frequently during the study period. A change in the patient profile first appeared at the fourth attendance and the change became more pronounced as attendances increased. Frequent attenders were more likely to be men (50.5% of single attendances; 69.5% of ⩾10 attendances), older (single attendance, mean age 32 years; ⩾10 attendances, mean age 45.6 years), to attend outside daytime hours (51.4% of single attendances; 69.2% of ⩾10 attendances) and to be triaged into the more serious categories (36.1% of single attendances; 54.3% of ⩾10 attendances).

Conclusion: Where local services are being designed to divert frequent attenders, existing data sources can be a rich source of information to inform service design. For example, this analysis identifies older men at their fourth or more attendance as a potentially important group when examining frequent attendance at this particular hospital. It also identified a potential need for services outside normal surgery hours, although frequent attenders tend to be triaged into the more urgent categories.

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.

Emergency departments were established to meet the needs of patients with either acute injuries or sudden onset of illness. In 2006–7, A&E departments and minor injury units in England reported nearly 19 million attendances and Department of Health figures demonstrate year on year increases in attendance.1 2 As part of the Darzi review of the NHS, the regional report for London included detailed modelling of future healthcare needs and reported growth in the city’s A&E activity to increase by 21% over an 11-year period.3 The same report suggests that many patients who attend emergency departments could receive their care in more appropriate settings.3

Many studies have attempted to understand why patients return to emergency departments within a set time period.4 5 Some research suggests that socioeconomic factors are important indicators of re-attendance at emergency departments,4 610 while others found social issues such as housing status, substance misuse and psychiatric problems to be associated with re-attendance4 6 7 and concluded that early identification of these patients and their improved management would result in fewer re-attendances.4 6 11 However, the evidence to date is inconclusive. Some research in North America has demonstrated the effectiveness of the implementation of case management on reducing re-attendance.12 13 Alternatively, an Australian study found that emergency department case management appeared to increase the use of the emergency department.14

However, one of the major flaws in the literature is the lack of a meaningful and consistent definition of re-attendance.15 For example, Byrne and colleagues4 categorised patients attending the emergency department >4 times as frequent attenders while a New Zealand study interpreted frequent attendance as >10 visits.5 Recognising these inconsistencies, one British study aimed to develop a definition of frequent use. The study, by Locker and colleagues,15 identified a group of patients who present repeatedly due to non-random events and proposed that the term “frequent user” be defined as any patient who makes >4 attendances per year. This lack of consistency makes it difficult for planners to develop services tailored to the cohort of “frequent attending” patients.

This study analysed 1-year emergency attendance data from a large city hospital. It aims to identify personal and attendance factors which are associated with the number of times patients attend an emergency department in order to verify the definition of “frequent user” of ⩾4 attendances by Locker et al, Byrne et al and Murphy et al.4 10 15

METHODS

Description of the dataset

The data analysed are routinely received by a primary care trust from the emergency department of its main provider acute hospital. The hospital is a teaching hospital serving primarily the local population of two primary care trusts, covering a multi-ethnic community population of 550 000 with high levels of deprivation.16 17 One-year emergency department attendance data (2006–7) were provided for this research. Personal and attendance level variables were analysed including patient gender and age, triage category and time and date of arrival. Data were checked and cleaned for any inconsistencies. Complete data were available for most variables except the diagnosis categories. Some variables were recoded to enable more meaningful analysis (eg, 15 triage categories were recoded into 5).

For the purposes of this study, variation in attendance was analysed by dividing the attending population into categories based on the number of times individuals attended the department during the study period. As discussed above, there are many definitions of frequent attenders. The analysis is therefore focused on the number of times individuals attended the department to ascertain whether patient characteristics changed as the number of attendances increased. The literature points to 4 attendances being an important cut-off point, as are both ⩾5 and ⩾10 attendances.15 Consequently, for analysis in this paper, the attendance categories are as follows: single (or 1) attendance, 2 attendances, 3 attendances, 4 attendances, 5–9 attendances and ⩾10 attendances.

Data analysis

Analysis was performed using Stata Survey Commands V. 9.2 (StataCorp, College Station, Texas, USA). The dataset was analysed at two levels, first at the individual patient level and the second at the attendance level. Frequencies and cross-tabulations were produced to describe the dataset. All results reported were statistically significant except where otherwise stated. Confidence intervals (95% CI) were calculated where appropriate for both patient and attendance level data. Where confidence intervals did not overlap, differences were regarded as significant. Attendance level data were adjusted for repeated observations.

RESULTS

Description of individual patients

During the research period 82 812 individuals attended the emergency department on average 1.4 times, resulting in 117 187 attendances. Their mean age was 32 years (range 0–100); 50.4% were male and 49.6% were female.

Of the 82 812 individuals who attended the emergency department, 76% did not return during the study period (table 1). Over 13 000 individuals (16%) attended a second time, 5% had two more return visits and the remainder returned at least another three times. While only 3.5% of individuals visited the emergency department on five or more occasions during the research period, this accounted for 8.8% of all attendances.

Table 1 Category of attendance frequency by individual patients and number of attendances

Description of attendances

Attendances were spread fairly evenly across the week with the highest proportion attending on a Monday (15.5%), and almost half of attendances were between 09.00 and 16.59 h. Almost one-quarter of patients (24.8%) arrived at the emergency department by ambulance; however, the majority (61.3%) arrived either by public or private transport (table 2). Triage is a method of ensuring that patients are seen in order of clinical need rather than order of attendance.18 As a result, triage category is a good indicator of the immediacy of need for medical care. Patients are assessed generally by a triage nurse and categorised into five colour bandings (table 2).

Table 2 Arrival, triage category and outcome

A small proportion of all attendances were triaged as red or orange and deemed to need immediate or very urgent attention (<3%). A little over one-third were categorised as yellow and needed urgent attention. Most attendances were categorised as green “standard” implying that, while they needed care, it was not needed immediately. A comparatively small proportion (1%) were categorised as blue, implying that presentation was non-urgent.

The patient’s presenting problem is recorded using the Manchester Triage system18 list with a few categories added for local use. The 49 categories were combined for the purposes of this study into the following complaint groups to simplify analysis: unwell (25.1%), limb (16.9%), injury (14.7%), gastrointestinal symptoms (9.7%), chest/respiratory (7.2%), other complaint (7.8%), ENT (6.4%), pain (4.9%), pregnancy/PV bleeding (2.8%), psychosocial (2.6%) and chronic disease (1.8%). The patient outcomes or disposal categories were also combined from 38 categories for this study to simplify analysis into: discharged (no follow-up) (45%), admitted (20.5%), own GP (19%), outpatients (6.3%), did not wait (4.1%), emergency follow-up clinic (4.1%), community (0.3%) and died (0.1%).

Comparing patients by number of attendances

The following analysis examines the number of times patients attended the emergency department by attendance categories grouped as single (or 1) attendance, 2 attendances, 3 attendances, 4 attendances, 5–9 attendances and ⩾10 attendances as described in the Methods section.

Gender and age

With regard to the personal characteristics of the patients, significant patterns emerged in the dataset (table 3). The mean age rose with the number of times an individual attended the emergency department. For patients attending 1–3 times the mean age was around 32 years; for those with 4 attendances the mean age increased slightly and continued increasing at 5–9 attendances (36.3 years; CI 34.9 to 37.7) and ⩾10 attendances (45.6 years; CI 42.8 to 48.3).

Table 3 Personal characteristics (age and gender) by category of attendance frequency (patient level data, n = 82 812)

This pattern was similar for gender. The proportion of men rose with the number of times an individual attended the emergency department. The proportion of men attending 1–3 times was about 50%; at 4 attendances 51% were male, with a slightly higher proportion of those attending 5–9 times being men (53.5%; CI 50.7% to 56.2%) and more than two-thirds of those who attended ⩾10 times (69.5%; CI 62.1% to 76.3%) were men.

Triage category

The triage category assigned to an individual on arrival is an indication of their clinical need. Very small numbers of patients are triaged as needing immediate attention (triage 1) or very urgent attention (triage 2). Nonetheless, larger proportions of individuals are triaged into these categories if they attend more frequently. Those triaged as needing urgent attention (triage 3) show a similar pattern. The reverse is true of those triaged as needing standard attention (triage 4). There was no particular pattern for those triaged as non-urgent (triage 5), although those attending ⩾10 times were twice as likely to be in this category as those with a single attendance (table 4).

Table 4 Triage category by category of attendance frequency (attendance level data, n = 117 187)

Combining immediate attention, very urgent attention and urgent attention (triage categories 1, 2 and 3) provided an overview of the more urgent categories. A little over one-third of patients who attended only once (36.1%) were triaged into one of these three categories. The proportion categorised as needing urgent attention increased as the number of attendances increased. Nearly half of those attending 4 times (46.0%; CI 44.3% to 47.7%) were triaged into the urgent categories. For those attending 5–9 times this proportion increased to just over one-half (50.6%; CI 49.0% to 52.3%) and increased again to 54.3% (CI 50.0% to 58.6%) for those attending ⩾10 times.

Time of attendance

Attendance by day of the week was similar across attendance categories at around 14% per day.

Time attendance data in hourly bands was recoded in three bands of 01.00–08.59 h, 09.00–16.59 h and 17.00–00.59 h. The times of attendance across the bands were similar for those attending 1–3 times with approximately half the attendances in day time between 09.00 h and 16.59 h and half outside these hours. As with the analysis of age and gender variables, there was a small increase in the proportion of patients attending outside daytime hours at 4 attendances (53.1%) and larger increases at 5–9 attendances (57.6%) and at ⩾10 attendances (63.2%; table 5).

Table 5 Time bands, out-of-hours attendance and personal characteristics by category of attendance frequency (attendance level data, n = 117 187)

Mean age was calculated for those attending outside daytime hours (09.00–16.59 h) by category of attendance. For single attendances outside daytime hours the mean age was 30.7 years. Again, during outside hours the more frequently an individual attended the older they tended to be (1 attendance, 51 years; 2 attendances, 50 years; 3 attendances, 50 years; 4 attendances, 51 years; 5–9 attendances, 55 years; and ⩾10 attendances, 69 years). Those who attended <4 times had an average age similar to the average age of the population but those who attended >4 times were generally noticeably older.

The proportion of men was also calculated for those attending outside daytime hours by category of attendance. The proportion of men was higher the more frequently patients attended. The pattern for gender and outside daytime hours followed the same pattern reported earlier with a small increase from around 50% at 1–3 attendances to 52.6% at 4 attendances to nearly 70% at ⩾10 attendances. However, the relationship between gender, age and time of day were not found to be statistically significant.

An important variable to examine in terms of the patterns of attendance times is “out of general practice hours” (OOH). This covers a wider timeframe and takes into account the hours that most general practices are closed before 08.00 h and after 18.00 h Monday to Friday and all hours Saturday and Sunday. Nearly two-thirds of all patients attended during these hours. Slightly higher proportions attended during OOH, the more frequent they attended the emergency department (table 5). Those attending OOH on average were aged 31.8 years and, again, the more often they came the older they were. The gender breakdown for OOH followed the same pattern as for outside daytime hours. Again this was not found to be statistically significant.

DISCUSSION

This study aimed to establish characteristics which could be associated with “frequent attendance”. Like similar studies, it showed that older male patients tend to have higher levels of attendance.3 10 It also dispels the myth—in this local population at least—that these patients are time wasters, as the more often they attended the more likely they were to be triaged into the more serious categories. Some evidence is presented that frequent attendance could be related to out of working hours and out of general practice opening hours attendance. The data also indicate that the fourth attendance is an important one as, thereafter, the more often patients attend, the more the characteristics outlined above are present.

A Swedish study in 2001 found that frequent users of emergency departments were higher users of other healthcare facilities and had a higher than expected mortality rate.19 Darzi3 has suggested that frequent attenders at emergency departments could receive their care in more appropriate settings. Other commentators have suggested that services in emergency departments should be made more appropriate to the patient20 and extended to meet requirements.20 Evidently, these patients could be a potential saving for the NHS if they could be diverted successfully into the community. Personal and Social Care Unit Costs for England21 costed an emergency service attendance at between £94 and £126. In comparison, a GP surgery consultation was £32 and £10 for a consultation with a community nurse. However, there is only mixed evidence of the success of such diversionary programmes. Some studies have demonstrated reduced re-attendance at the emergency department following implementation of a case management programme.12 13 However, an Australian study on re-attendance after five visits found the opposite. This could be because most of the patients (73%) presented with substance misuse or psychosocial issues.3 14

The alternative to community diversion is to employ GPs in the emergency department. Work by Dale and colleagues in the 1990s suggested that employing GPs in emergency departments resulted in reduced rates of investigations, referrals and prescriptions22 and corresponding reduced costs without any detrimental effects on patient outcome.23

Another important finding—especially in the light of the Secretary of State for Health’s recent letter to GPs encouraging them to open for longer hours—is that the proportion of patients attending OOH increases as attendance increases. This may suggest that there is a need for more OOH community services. However, the categorisation of more frequently attending patients into the more urgent triage categories may well suggest the opposite.

This analysis describes a rich dataset to which all emergency departments will have access. The literature does not offer a credible definition of frequent attendance—possibly because such a definition will vary locally—but, like this study, more often than not it found that 4 attendances is an important cut-off point to consider. These data suggest that the fourth visit may be the appropriate time for intervention in our emergency department, if one is to be made. In other places this may not be the case. Nonetheless, such data analysis will inform the point in the patient journey through the emergency department at which diversion should be attempted and with whom.

CONCLUSION

This study demonstrates that emergency department data are a potentially rich source of information for service planners. Older men attending for the fourth time are the potential focus for any intervention in the emergency department studied.

Acknowledgments

The authors thank the hospital and Southwark Primary Care Trust for their support in this work.

REFERENCES

Footnotes

  • Funding: LM undertook this work as a research fellow seconded to the Department of General Practice and Primary Care at King’s College London. This research fellowship was funded by Guy’s and St Thomas’ Charity through the Primary and Community Care Research Support Programme.

  • Competing interests: None.

  • Ethics approval: The Chair of the local research ethics committee considered this work to be an audit of patients and as a result the project did not need to be formally submitted to the Research Committee.