Article Text

Download PDFPDF

Predictors of frequent attenders of emergency department at an acute general hospital in Singapore
  1. P Paul1,
  2. B H Heng1,
  3. E Seow2,
  4. J Molina1,
  5. S Y Tay2
  1. 1Health Services and Outcomes Research (HSOR), National Healthcare Group HQ, Singapore
  2. 2Emergency Department, Tan Tock Seng Hospital, Singapore
  1. Correspondence to P Paul, Health Services and Outcomes Research (HSOR), National Healthcare Group HQ, 6 Commonwealth Lane, Level 6 GMTI Building, Singapore 149547, Singapore; pradeep_paul_g_gunapal{at}


Objective To determine factors associated with frequent emergency department (ED) attendance at an acute general hospital in Singapore.

Method Patients who attended the ED from 1 January to 31 December 2006 without prior attendance in the preceding 12 months (index attendance) were tracked for 12 months. Variables included in the analysis were age, gender, race, date and time of attendance, patient acuity category scale, mode of arrival, distance to ED and diagnosis based on ICD-9CM code. Frequent attenders were patients who attended the ED ≥5 times for any diagnosis within 12 months.

Results A total of 82 172 patients in the study cohort accounted for a total of 117 868 visits within 12 months, of which 35 696 (30.3%) were repeat attendances. A total of 1595 patients (1.9%) were frequent attenders responsible for 8% of all repeat attendances. Stepwise logistic regression analysis found patients aged 75+ years, male, non-Chinese ethnic groups, Sunday and Monday, time of the attendance from 16:00 to midnight, distance to ED, chronic obstructive pulmonary disease, heart failure and acute respiratory infections to be significantly associated with frequent attendances.

Conclusion With the ageing population and their complex healthcare needs, elderly patients with chronic medical conditions are expected to make up an increasing proportion of the workload of ED in the future. A systems approach and a disease and case management approach in collaboration with primary care providers are interventions recommended to stem this.

  • Emergency department
  • re-attendance
  • predictors
  • frequent use
  • emergency care systems
  • epidemiology
  • research
  • statistics

Statistics from


Globally, hospital emergency department (ED) attendances have increased significantly over the past decade. Several factors have been implicated in the surge, including the ageing population, its accessibility, urgency of the complaint, perceived need for investigations in a hospital setting and inappropriate attendances.1 EDs are increasingly becoming a popular choice for primary care, which is inappropriate and causes a serious threat to healthcare quality and increases the cost of ED care.2 In Singapore, EDs are facing a similar increasing trend in the number of attendances, frequent overcrowding3 and boarding. Boarding in turn may result in delays in care, ambulance diversion, longer hospital stays, medical errors, increased patient mortality, financial losses to the hospital and medical negligence claims.4 Some studies have reported an increase in repeat attendances among certain subgroups of patients, which are often judged as non-urgent but account for a substantial proportion of the total number of attendances to these facilities5–9 and result in prolonged waiting times.8–12

The Singapore healthcare system comprises public and private healthcare with services including primary care, hospital care and intermediate and long-term care. Of the seven public hospitals, five are general hospitals that provide multidisciplinary acute inpatient and specialist outpatient services and a 24 h emergency department. Tan Tock Seng Hospital (TTSH) is a 1200-bed hospital, the second largest acute care general hospital in Singapore, with a geographical catchment of about 1–1.5 million people in the northern and north-eastern part of the island city, and has the highest ED attendances among the five acute public hospitals in Singapore. Not much is known about the extent of re-attendances and its contribution to the increasing workload. The aim of this study is to describe the extent and characteristics of ED re-attendances and to identify factors associated with re-attendances. The results will also serve as a baseline for evaluation of future interventions.


A review of all patients who attended TTSH ED between 1 January 2005 and 31 December 2007 was carried out using data from an existing administrative ED database (EDWeb). Permission to conduct the study was granted by the head of TTSH ED. Variables for analysis included demographic characteristics (age, gender, ethnic group and nationality), day and time of ED attendance, patient acuity category scale (PACS), mode of arrival, distance to the ED, primary diagnosis based on codes using International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9CM) and disposition of patients. Age and primary diagnoses (Appendix 1) were grouped into broader categories for analysis. Patients were classified as P1, P2, P3 and P4 using the PACS, where P1 cases are most acutely ill and need immediate attention, P2 cases are acutely ill with severe symptoms, and P3 and P4 are minor emergency and less acutely ill patients. Distance to ED was included in the analysis to check if proximity to ED influenced re-attendance. Shortest residential distance to the hospital was derived using Drivetime application of Mapinfo Professional Version 8.0.

The outcome of interest was ED re-attendance, defined as the second ED attendance of a patient for any diagnosis which occurred within 12 months of the initial (index) ED attendance. An index attendance was the first ED attendance of a patient defined as one that did not have a prior attendance in the preceding 12 months. The definition of a ‘frequent attender’ was adapted from other studies13 as a patient who attended the ED ≥5 times for any diagnosis. The study included index patients from 1 January to 31 December 2006 (henceforth referred to as the study cohort) who were tracked for re-attendances for 12 months up to 31 December 2007.

Frequency and percentage distributions were used to describe the study cohort. The χ2 test was used for significant testing of the categorical variables. Stepwise logistic regression analysis was carried out to determine independent factors significantly associated with frequent attendances. To avoid overfitting the model, only variables found significant on univariate analysis were included in the regression model. All statistical analyses were performed using SPSS Version 15 and tests were conducted at the 5% level of significance. The ORs and corresponding 95% CIs are reported where applicable.


Overall, there was a total of 444 558 attendances by 305 761 patients to TTSH ED during the period 2005–7 with a mean annual rate of increase in ED attendances of 6.5%. The study cohort comprised 82 172 patients with index attendances from 1 January to 31 December 2006. Figure 1 shows that 74.9% of these patients attended the ED only once during the 12-month follow-up period. The remaining 25.1% of patients who made ≥2 attendances were responsible for 56 316 (47.8%) of the total 117 868 attendances.

Figure 1

Emergency department (ED) re-attendances by study cohort.

Frequent attenders (with ≥5 attendances) comprised 7.8% of patients with re-attendance but were responsible for 26.4% of all repeat attendances (table 1). The rate of re-attendance increased significantly with age, where over 40% of patients aged ≥75 years had at least one repeat attendance (table 2). Generally, men had a higher re-attendance rate than women, the gender difference being greatest among frequent attenders aged <25 years (figures 2 and 3). It was also significantly higher among acutely ill patients (P1 and P2), those staying nearer the hospital and those brought in by ambulance. Table 3 shows that patients with the highest re-attendance rate were those with heart failure (53.9%), chronic obstructive pulmonary disease (COPD) (46.1%) and diabetes (41.3%); the re-attendances accounting for 75.7%, 68.6% and 62%, respectively, of all attendances in these diagnostic groups. The median interval to re-attendance was 128 days. However, a quarter of re-attendances occurred within the first month of the index attendance (table 4). Of these, about half were within 6 days, which were mainly due to acute conditions such as infections and trauma including head injuries and fractures (table 3). Almost 40% of re-attendances occurred at 6 months and later, and these were mainly due to (1) chronic conditions such as COPD, hypertensive disease, heart failure, ischaemic heart disease and diabetes; (2) respiratory infections such as pneumonia and acute respiratory infections; and (3) foreign bodies on the eye, in the pharynx and larynx and superficial injury.

Table 1

Distribution of repeat attenders by frequency of attendances

Table 2

Description of study cohort and univariate analysis

Figure 2

Age-specific and gender-specific rate of non-frequent attendances.

Figure 3

Age-specific and gender-specific rate of frequent attendances.

Table 3

Emergency department (ED) re-attendances by primary diagnosis and interval between index and repeat attendance

Table 4

Distribution of re-attendances

Stepwise logistic regression analysis found that age ≥75 years, male gender, non-Chinese ethnic groups, attending the ED on Sunday and Monday and from 18:00 to midnight, and travelling >2 km to the ED to be significantly associated with higher odds of a frequent attendance. COPD, heart failure and acute respiratory infections were the leading diagnoses associated with frequent attendances (table 5).

Table 5

Independent factors significantly* associated with non-frequent attenders (2–4 attendances) and frequent attenders (≥5 attendances)


This study affirms the increasing attendances at TTSH ED from 2005 to 2007, similar to that worldwide.5 7 9–,12 27 Population ageing has been cited as one of the reasons for the disproportionate ED attendances by elderly subjects,15 contributing a substantial and increasing proportion of ED workload in developed countries.16 17 In Singapore, 7% of the population is currently aged ≥65 years and by 2030 this will increase to 19%.18 During the period 2005–7 the population increased 8%,19 but there was an overall increase of 13% in ED attendances in TTSH. In this study, patients aged ≥65 years comprised 18% of all ED patients, an increase from 12.4% to 12.8% reported in previous studies in 1996 and 2001.14 20 While it is true that the increase in ED attendances can be accounted for by the ageing population, one in three attendances by the study cohort were repeat attendances.

Repeat attenders use a disproportionately large share of ED resources,15–18 their consumption is wasteful and preventable. The complex health needs of elderly patients aged ≥75 years with associated chronic conditions such as heart failure and COPD is likely to incur an exponential increase in the workload of ED staff in the coming years. Case management programmes for ‘difficult’ patients with medical and social problems, which may involve multidisciplinary team and individualised care plans appropriate to the patient, geriatric assessment involving the community for ongoing care and follow-up and providing transitional homes have been reported to reduce unplanned visits significantly in other countries.21–24 Recognising the increasing burden of chronic diseases and the importance of keeping patients with chronic diseases optimally managed out of hospital, in October 2006 the Singapore Ministry of Health implemented the Chronic Disease Management Programme covering diabetes mellitus, hypertension, lipid disorders, stroke, asthma and COPD. The programme engages the support of general practitioners to provide systematic evidence-based care and has allowed the use of Medisave (a portion of contribution towards the central provident fund, a compulsory savings scheme in Singapore) to reduce the out-of-pocket cash payments for outpatient bills.

The high repeat attendance rate within 6 days among patients with infections and trauma/accidents cannot be ignored. These may be due both to patients wishing to be managed by the same department that had managed them initially and the ED routinely holding planned review clinics. The extent of this is not insignificant, and such attendances will only add to the already heavily burdened ED. Formal links between the ED and primary care physicians can make way for these patients to be reviewed more appropriately in primary care. However, the government primary care polyclinics—the first port of call for patients—operate from 09:00 to 16:30, which explains the contribution to the higher ED attendances beyond these operating hours.

Some of the proven methods used in other countries to deal with increasing workload at the ED, such as co-locating a short stay observation ward and ‘fast tracking’ patients with specific conditions, have already been implemented. In the systems dynamics of emergency care, the interconnectedness of TTSH ED with inpatient and ambulatory services and the healthcare system outside the hospital,25 including primary care, cannot be overemphasised. A disease management programme started about a decade ago which targeted patients with asthma has contributed significantly to the decrease in ED re-attendances by patients with asthma. This national multi-pronged programme focuses on improving asthma control in the community by identifying all high-risk patients at the hospital, preventing exacerbations by promoting self-management and inhaled corticosteroid therapy, and a strong primary care focus—a systems approach.26 COPD and pneumonia programmes have also been launched more recently, and the impact of these programmes on repeat ED utilisation will be determined in due course.

Appendix I Diagnosis groups (ICD-9CM)

1Signs and symptoms of ill defined conditions780–789
2Mental disorders290–319
3Chronic obstructive pulmonary disease and allied conditions490–496
4Pneumonia, organism unspecified480–486
6Urinary tract infection599
7Heart failure428
8Gastritis and duodenitis535
9Infections of skin and subcutaneous tissue680–686
10Ischaemic heart disease410–414
11Diseases of the musculoskeletal system and connective tissue710–739
12Diseases of the ear and mastoid process380–389
13Hypertensive disease401–405
15Acute respiratory infections460–466
16Foreign body in pharynx and larynx933
18Ill-defined intestinal infections9
19Acute appendicitis540
21Viral and Chlamydia infection in condition classified elsewhere79
22Intracranial injury of other and unspecified nature854
23Sprains and strains of joint and adjacent muscles840–848
24Open wound of upper limb880–887
25Open wound of lower limb890–897
26Contusion with intact skin surface920–924
27Superficial injury910–919
28Open wound of head, neck and trunk870–879
29Foreign body on external eye930
30Open wound of the head873
31Disorders of conjunctiva372



  • Competing interests None.

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

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.

Linked Articles

  • Primary survey
    Steve Goodacre