Social Science & Medicine 59 (2004) 2207–2218

Effects of nurse follow-up on emergency room revisits:

a randomized controlled trial
Frances Kam Yuet Wonga,*, Susan Chowa, Katherine Changa,
Albert Leeb, Jiexin Liuc
a
School of Nursing, The Hong Kong Polytechnic University, Hunghom, Hong Kong
b
Department of Community and Family Medicine, The Chinese University of Hong Kong, Hong Kong
c
Institute for Health Policy Research, Robert C. Byrd Health Sciences Center, West Virginia University, USA
Available online 25 May 2004

Abstract

The emergency room (ER) is the gatekeeper of the hospital. It receives clients seeking help from the health care
system, then refers them for necessary further care or, discharges them back into the community. A 1-year randomized
control trial was conducted in an acute general hospital in Hong Kong to see if post-ER nurse follow-up helped to
change health outcome and health care utilization. The intervention group received two follow-up calls from an
experienced ER nurse, within 1–2 days and 3–5 days after ER discharge. The calls’ content and the management
options decisions were protocol driven. A total of 795 patients (intervention group=395, control group=400)
completed the study. Bivariate analysis shows two significantly different variables between the intervention and control
groups, improvement of the condition and ER revisit within 30 days. When other factors are controlled in a
multivariate analysis using logistic regression, the effect of the intervention on re-visits to ER within 30 days is reversed.
Gender, times of attending general practitioner after ER visits, and not considering other doctors are risks factors
related to a higher chance of re-visiting ER within 30 days. Those who have an improved health condition and higher
number of times attending general outpatient after ER visits are associated with a lower 30-day ER revisit rate. Nurse
telephone follow-ups might have sensitized subjects to health care needs. Some subjects tended to use the ER as a
primary care setting and some were doctor-shoppers. A more structured ER transitional model that incoporates nurse
telephone follow-up and better interfacing between private and public health care sector, primary and tertiary care
might help to decrease inappropriate ER use.
r 2004 Elsevier Ltd. All rights reserved.

Keywords: Emergency room use; Nurse follow-up; Health service utilization; Help seeking behavior; ER transitional care; Hong Kong

Introduction
The emergency room (ER) is the gatekeeper of the
hospital. The hospital authority (HA), Hong Kong records
a total of 2,403,090 ER visits each year. The annual
attendance number is equivalent to one-third of
*Corresponding author. Tel.: +852-2766-6419; fax: 852-2364-
9663.
E-mail address: hsfwong@inet.polyu.edu.hk (F.K.Y. Wong).
the total Hong Kong population. Evidence suggests that the
use of ER maybe confined to a small proportion of the total
population in spite of the high attendance rate (The
Harvard Team, 1999). Our survey in a regional hospital
disclosed that 3.8% of patients re-attended ER within 48 h,
but the rate increased to 26% within 1 month (Wong,
Halloran, Ho, Chang, & Leung, 1999). The prevalence of
unscheduled ER visits within 48 h is comparable to
overseas findings. However, the figure for 30-day return
seems to be over-represented in Hong Kong.

0277-9536/$ - see front matter r 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2004.03.028
2208 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218
The causes of unscheduled return visits are numerous,
including patient-related, illness-related, and system-
related factors (Kelly, Chirnside, & Curry, 1993). Patient-
related factors include gender, age and socio-economic
class. The findings for the association between gender and
ER use were inconclusive (Kushel, Perry, Bangsberg,
Clark, & Moss, 2002; Pierce, Kellerman, & Oster, 1990).
Young children were likely to make repeated visits than the
older adult group (Lee et al., 2000; Kushel et al., 2002;
Wong et al., 1999). Those with no insurance or on public
insurance were related to multiple use (Pierce et al., 1990;
McCauley et al., 1998; Zimmerman et al., 1996).
As for illness-related factors, those with respiratory and
gastrointestinal conditions diagnoses have the highest
return rate (Pierce et al., 1990; Zimmerman et al., 1996).
System-related factors include errors of diagnosis or
treatment, or a failure to make adequate follow-up
arrangements (Kelly et al., 1993). One study suggested that
32.3% (out of 407) of the attendance were avoidable if
there were better medical management, prescribed follow-
up, and/or patient education (Keith, Bocka, Kobernick,
Krome, & Ross, 1989).
Measures were introduced in an attempt to reduce the
number of ER visits. Studies have reported positive results
in using telephone-based services to reduce ER visits
(Dale, Crouch, & Lloyd, 1998; O’Brien & Miller, 1990).
Unfortunately, most of the telephone follow-up studies
done in the ER setting are conducted as an evaluation of
the provision of new services without control. Randomized
controlled trial investigating ef-fects of nurse telephone
consultation service in other settings, however, suggest
positive effects (Landefeld, Palmer, Kresevic, Fortinsky, &
Kowal, 1995; Naylor et.al., 1999).
In Hong Kong, the Government has always had a stated
policy that no citizen should be deprived of adequate health
care. The Hong Kong Government subsidizes heavily all
the services (Fung, Tse, & Yeoh, 1999). The patients pay
HK$100 (approximately US$12.8) for a hospital stay,
inclusive of room charge and all health care services
received and HK$55 (approximately US$7) for each
outpatient visit (Hospi-tal Authority, 2003). The ER is free
before November 2002 (Fung et al., 1999). Statistics show
that health care in Hong Kong is available and affordable.
However, the figures of nonurgent users in ER, 57% (Lee
et al., 2000) to 72% (Law & Yip, 2002), suggest that the
efficiency of care access is questionable. A consultancy
team criti-cized that there is a lack of coordination and
cohesion between primary and inpatient care, acute and
commu-nity medicine, and the private and public sectors in
Hong Kong (The Harvard Team, 1999). This study is
therefore launched, using ER attendants as subjects, to see
if an ER transitional care model using nurse follow-up
would help to reduce the use of ER.
A conceptual model
The factors that predispose and enable the use of health
services are multiple. In this study, Andersen’s model is
used as a framework to guide the selection of a range of
explanatory variables for the 30-day revisit of ER. The
model is selected for it embraces the key conditions that
will determine access to health care. The conditions
include predisposing, enabling, and need factors
(Andersen, 1995; Andersen & Davidsen, 2001). The
predisposing factors are the existing conditions that may
relate to health care use. These factors include age, gender,
living place, education, and occupation (Be´land, Lemay,
& Boucher, 1998; Lee et al., 2000; Liu, Sayre, & Carleton,
1999). The enabling conditions refer to the resources that
facilitate or hinder the use of health care. These include
ability to pay, or availability of insurance (Kushel et al.,
2002; Zimmerman et al., 1996), and access of care (Baker,
Stevens, & Brook, 1994; Lee et al., 2000; O’Brien et al.,
1998). In this study, a contextual enabling factor is
introduced to the intervention group, that is the nurse
follow-up, to see if the inclusion of this enabling factor will
affect the use of health services. The need for care
comprises the perceived need and evaluated need.
Perceived needs are the client’s perception of the urgency
of the condition which affects the decision to seek ER care
(Boushy & Dubinsky, 1999). Evaluated needs are the
provider’s assessment of the patient’s condition whether
the patient needs immediate attention or he/she can wait
(Liu et al., 1999; Young, Wagner, Kellermann, Ellis, &
Bouley, 1996).
Methods
Setting and subject recruitment
The study took place in an acute general hospital with
1284 beds providing a range of medical services in Hong
Kong. We calculated the sample size by assuming the rate
of return ER visits within 1 month to be 26% if no
intervention is taken, and that there will be 35%
improvement (i.e. that the re-visit rate will drop to 17%),
using EpiInfo (Centers for Disaease Control & Prevention,
1997). We estimated that a sample size for the study will
be 834 (417 for intervention and 417 for control), with a
dropout rate of 10% for an alpha of 0.05 and a power of
0.85. In order to obtain a representative sample, the study
proportionally sampled the number of hours selected with
each of the utilization dates. The study days were randomly
spread through the year to reduce seasonal effects and the
time sessions covered all 24 h. At the end, 25 Sundays, 21
Mondays, 20 Tuesdays, 20 Wednesdays, 21 Thursdays, 21
Fridays and 22 Saturdays were selected, making a total of
150 days. Six time points of ER visit was randomly
identified for
 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218 2209

each selected day. A computer generated randomized table * cannot provide informed consent (guardians can
was used. Each recruited patient was serially assigned, with provide consent for subjects under 18),
an odd number indicated intervention group and an even * do not live within the hospital service area.
number the control group.

The subject inclusion criteria were:
* with presenting problems related to fever, respiratory or
gastrointestinal condition (note: these three problems
were chosen because our pilot study revealed that these
three conditions were the most commonly presented
problems in return visits in ER),
* speak Cantonese,
* Twenty-nine in the study group declined the follow-up
are alert and oriented when visited ER,
* discharged home,
* are able to be contacted by telephone after discharge.
The exclusion criteria were:
* referred to follow-up visits at ER,
* referred to follow-up visits at general or specialty
outpatient for an appointment within less than 4 weeks,
* discharged to nursing homes,
* dying,
The first patient who met the criteria in the 900 time
points was invited to participate, amounting to 900
potential subjects. Twenty subjects were excluded before
randomization for 19 refused to participate and 1 did not
speak Cantonese. The written consent was obtained after
the subjects showed an initial verbal agreement before
randomization. Amongst the 880 subjects who agreed to
participate in the study, we lost 85 subjects in the process.
intervention and 56 (study=16, control=40) could not be
reached for data collection. At the end of the study, 795
subjects (88.3% out of the 900 potential subjects)
completed the study (please refer to Fig. 1).
Ethical considerations
The study was approved by the Research Ethical
Committees of the study hospital and the university that the
principal investigator was affiliated with. The purpose,
design of the study and the involvement of the subjects
were explained to the potential subjects by a

Fig. 1. Flow chart of subject progress through the phases of randomization.

2210 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218
research nurse stationed in the study hospital. The subjects
were reassured that their care would not be affected by
their decision not to participate in the study. Their personal
identity would be protected for all data would only be
identified by a case number designated by the research
assistant. The patients were given an information sheet
which contained the telephone num-bers of the principal
investigator and the secretary of Research Ethical
Committee if they had any problems about the research.
All participants were asked to sign a consent form and
were informed that they could withdraw from the study any
time during the process.
Study design
The study was a randomized clinical trial to compare the
30-day ER return visits between the control and the
intervention group. The control group received usual post-
discharge care and the intervention group was followed up
by a nurse. Data collected involved instruments to be
described in the data collection section.
Intervention
The intervention included two follow-up calls, one
within 1–2 days and one 3–5 days after ER discharge. The
three study nurses were experienced ER nurses working in
the setting with an average experience of over 10 years.
They had the ER patient documentation with them when
they conducted the telephone follow up. The intervention
protocol was developed based on previous work on
telephone triage and health assessment (Simonsen, 1996;
Briggs, 1997; Grossman, 1999) which content has been
reviewed by experienced ER nurses and physicians. The
protocol contained subsections that addressed specific
symptoms related to the selected conditions, namely fever,
respiratory and gastrointest-inal problems. The telephone
follow-up began with an assessment of the existence
(yes/no) of symptoms (examples of symptoms for the fever
condition are fever over 104 F, decreased urine, seizures).
The second part was a decision on management options
which was underpinned by professional judgement and
objective referral criteria developed based on the reference
mentioned above. The management options included:
(1) telephone follow-up for patients requiring health
advice and reinforcement of health behavior such as
compliance with prescribed treatment; (2) referral to the
community nurse for home visit if physical assessment of
clients’ conditions and/or provision of direct health
instructions are needed; (3) referral to primary care setting
if clients’ conditions require medical follow-up; (4) referral
to ER if clients’ conditions deteriorate.
In order to ensure that the advice and prescription of
management options were valid, safe and reliable, three
measures were taken. First, the study nurses were trained
for the post-ER nurse telephone follow-up. Second, the
intervention protocol was tested in the pilot study using 5%
of the intended sample size (i.e. 40 cases). The audiotapes
of the telephone follow-up were independently reviewed
by an experienced ER physician, a nurse manager and a
research team member (an experienced family physician).
Each reviewer made judgement on the appropriateness of
advice given and decided on an appropriate outcome
management option. The rate of agreement was 97%.
Finally, in the main study, 10% (i.e. 80 cases) of the
subjects were randomly selected for independent review by
an experienced ER nurse manager and a research team
member (an experienced family physician) for auditing
purposes.
Instrument and data collection
The data were collected in the 30-day follow-up
interview after the ER visit, from 1 February 2001 to 31
January 2002. The instrument was validated by an
experienced ER physician, an ER nurse manager, an ER
nurse specialist and a researcher. The information on age,
gender, disease categories, triage category and ER re-visits
were retrieved from patients’ records while other data were
collected from interviews. The interview questions are
quantitative, open-ended questions. Open-ended questions
allow the respondents the opportunity to describe their
unique experiences and thought processes (Shi, 1997). For
example, the interviewer would ask, ‘‘Why did you choose
ER on the day of visit?’’ ‘‘What is your condition now
after visiting the ER?’’ The responses were put into
categories for summary statistics. Test–retest was
performed on eight patients over a period of one week for
the interview items. The rate of coding and response
consistency, computed by summing up all numbers of
identical cases, and divided by total case-item numbers,
was 95% and 83.2%, respectively. For children, the data
were collected from the primary caretakers who were
mainly the parents. Some of the data, including education,
occupation, marital status, income reflected the status of
the parents. A research assistant who was blind to the
patients’ group assignments and had no contact with the
client at the study site telephoned the subjects 30 days after
randomization for interview.
Independent variables
Predisposing characteristics (age, gender, education,
occupation, marital status, living place). Gender and
marital status were coded in 2 categories; living place in 3
categories; age and education in 4 categories, and
occupation in 5 categories.
Enabling resources (income, what to do other than
attending ER, general treatment of minor problems,
 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218
consideration of other doctors, nurse follow-up). Income
was measured on a 3-point scale (1=not adequate for daily
living; 2=just meet daily expenses; 3=more than adequate).
What to do other than attending ER for the index visit was
measured on a 2-point scale (1=nothing can be done,
2=consult GP, alternative therapies). General treatment of
minor problems had 3 categories (1=attending ER; 2=self-
medication; 3=attending general practitioner (GP), general
outpatient clinic (GOPC)). Consideration of other doctors
prior to the index ER visit was coded no (=1) and yes (=2).
Nurse follow-up intervention was coded no (=1) and yes
(=2).
Needs (disease categories, triage, why chose ER on the
day, general reason for choosing ER). The disease category
was coded into febrile conditions, gastrointest-inal
conditions and respiratory conditions. Triage was
categorized as non-urgent(=1), semi-urgent(=2), and
urgent(=3). Patients were interviewed for why they chose
ER on the day (1=convenient; 2=felt need of care, or
referred by other health care providers; 3=felt sick, urgent
with no other choice of care) and general reason for
choosing ER (1=convenient; 2=no improve-ment after
medical consultation; 3=no other help available; 4=urgent
for care).
Dependent variables
Health Outcome (affecting daily life, improvement of
conditions, self-reported health, consumer satisfac-tion).
The health outcomes are clients’ conditions 30 days after
the index ER visit. The affect on daily life was yes (=1), no
(=2). The improvement of conditions was on a 4-point
scale (1=getting worse, 2=no change, 3=improved a bit,
4=recovered). Self-reported health was on a 5-point scale
(1=not good, 2=fair, 3=nothing abnormal, 4=quite good,
5=very good). The satisfac-tion was measured on a 3-point
scale (1=dissatisifed, 2=fairly satisfied, 3=satisfied).
Health care utilization (number of times visiting GP,
GOPC and ER). The data for health care utilization was
obtained from the hospitals’ secondary administrative
sources and through interview. Each subject’s record was
examined in the database for the number of re-visits to the
ER 30 days after the index visit. In the interviews, patients
were asked the number of times they had attended GPs for
the same problem after the index ER visit, the number of
times they had attended GOPC for the same problem after
the index ER visit.
Statistical analysis
This study included both continuous and categorical
2 improved health outcome, and the 30-day ER
variables. w tests were used for categorical data and
2211
independent samples t-tests were applied for continuous
data to assess differences between the two groups using
SPSS 11.0. Variables in the dimensions of predisposing
characteristics, enabling resources, needs, health outcome
and health care utilization were as displayed in Tables 1–
3. To test the specific effects of the potential explanatory
variables on the intervention, these variables were also
subjected to binary logistic regression (backward step-
wise) method in which 30 days ER re-visits was the
dependent variable. Backward method logistic regression
was executed with non-significant variable removed from
the model one at a time, if the variable has the p-value
larger than 0.05. The final model contained all the
statistically signficant variables that predicted a return visit
to ER within 30 days of the index visit.
Results
Table 1 displays the predisposing characteristics of both,
the control and study groups. The Chi-square tests showed
that there were no significant differences among these
baseline predisposing variables. The table revealed that the
subjects were generally well educated with over 60% of the
patients (for children, the patient’s education level was
used) attaining secondary level or higher. There were 63%
and 66% in the control and study group respectively whose
age was 18 or lower. There were more males (52–54%)
than females in both groups. Over 75% of the patients had
a residence in the public housing or private homes, while
the other 25% rented a bed or lived in quarters. Only a very
small percentage, 6% for control and 4% for study, were
unemployed.
Table 2 displays the comparison of the variables in
enabling resources and needs between the study and the
control groups. For the financial status, 60% (control) and
64% (study) said their income adequately met the daily
expenses. Over 50% of both groups said that they could do
nothing except attending ER. However, only 14% (control)
and 12% (study) groups indicated that they would attend
ER for treatment of minor diseases. On the day of index
visit, over 60% of the subjects from both groups had not
considered doctors other than ER. Most of the subjects
attended ER because of a respiratory condition (75–76%),
followed by gastrointestinal condi-tion (21–22%) and
febrile condition (2.8%). Around 4% of the subjects were
classified as urgent, though 21–22% of them felt that their
condition was urgent and warrant ER attendance on the day
of index visit. Eighteen percent (control) and 15% (study)
of the subjects admitted that they chose ER because of
convenience.
Table 3 displays the comparison of the dependent
variables of health outcome and health care utilization
between the study and the control groups. There were two
variables that were found to be significant. They were the
2212 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218

Table 1
Comparison of predisposing characteristics of patients in the study and control groups

Study (n=395) Control (n=400) Chi-square value p value

Educational level 1.278 0.783

No formal education 26 (6.58%) 27 (6.75%)
Primary level 114 (28.86%) 122 (30.50%)
Secondary level 227 (57.47%) 230 (57.50%)
Tertiary level 28 (7.09%) 21 (5.25%)

Age 3.22 0.359

0–18 years old 261 (66.08%) 253 (63.25%)
19–40 years old 67 (16.96%) 65 (16.25%)
41–60 52 (13.16%) 56 (14.00%)
Over 60 years old 15 (3.80%) 26 (6.50%)

Gender 0.204 0.652

Male 206 (52.15%) 215 (53.75%)
Female 189 (47.85%) 185 (46.25%)

Marital status 0.073 0.787

Alone 67 (16.96%) 65 (16.25%)
Married 328 (83.03%) 335 (83.75%)

Living place 0.144 0.988

Quarters or rent a room or bed 93 (23.54%) 98 (24.5%)
Public estate or wooden house 177 (44.81%) 177 (44.25%)
Private room 125 (31.65%) 125 (31.25%)

Occupation 3.89 0.42

Unemployment 17 (4.30%) 24 (6.00%)
Student, housewife, retired 207 (52.41%) 191 (47.75%)
Farmers, no skilled workers 117 (29.62%) 118 (29.5%)
General profession, technician 41 (10.38%) 55 (13.75%)
High profession, or employers or managers 13 (3.29%) 12 (3.00%)

revisits. The results suggested that nurse telephone follow-
up tended to produce positive health outcome, yet at the
same time increased ER utilization. In order to explain
further why the subjects returned to ER repeatedly, we
conducted a logistic regression to explore the predictors for
the 30-day ER return.
Table 4 shows the results of the logistic regression
model. The results showed that gender, study group,
improved health condition, times of attending GOPC with
the same problem, times of attending GP with the same
problem, and considered doctors other than ER were
significantly assoicated with re-visits to ER within 30 days.
The final model had a good fit with value of 766.5 for 2
Log likelihood. The approximately estimated Nagelkerker
2 weak interaction effect.
R is 0.272. Patients who were males were 1.69 times more
likely than females for returned ER visits. Those who did
not consider other doctors also were 1.69 times more likely
to return to second ER visits within 30 days. The study
group who received calls from nurses post ER visit, stood a
1.54 times higher chance to another ER visit within 30
days.
The greater improvement of the condition after the ER
visit, the less likely the patients would return, and the odds
ratio was 0.546. The higher number of times the patients
attended GP for the same problem after ER visits, the more
likely they returned to ER within 30 days. The odds ratio
was 8.18. However those who had higher times of
attendance to GOPC after ER visit for the same problem
had a lower chance of returning to the ER, and the odds
ratio was 0.69.
We examined the interaction effects of the variables in
the model and none of them are statistically significant.
The p-values of all interaction effects are bigger than 0.10,
except for the interaction effect between interven-tion and
improved health outcome (p=0.096) which indicates a very
Discussion
This randomized control trial study examined if nurse
follow-up had effects on health outcome and health care
 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218 2213

Table 2
Comparison of enabling resources and needs of patients in the study and control groups

Study (n=395) Control (n=400) Chi-square value p-value

Enabling resources
Income 1.709 0.426
Not adequate for daily living 141 (35.70%) 159 (39.75%)
Just meet the daily expenses 218 (55.19%) 211 (52.75%)
More than adequate 36 (9.11%) 30 (7.50%)
What to do other than attending ER 0.144 0.705
Nothing can be done 208 (52.66%) 216 (54.00%)
Consult GP, alternative therapies 187 (47.34%) 184 (46.00%)
Treatment of minor diseases
Attending ER 47 (11.9%) 54 (13.50%) 5.2 0.074
Self medication 134 (33.92%) 106 (26.50%)
Attending GOPD, GP 214 (54.18%) 240 (60.00%)
Consider doctors other than ER 0.254 0.614
No 241 (61.01%) 251 (62.75%)
Yes 154 (38.99%) 149 (37.25%)

Needs
Disease group 0.196 0.907
Febrile condition 11 (2.78%) 11 (2.75%)
Gastrointestinal condition 89 (22.53%) 85 (21.25%)
Respiratory condition 295 (74.68%) 304 (76.00%)
Triage 0.571 0.752
Not urgent 141 (35.70%) 134 (33.50%)
Semi-urgent 238 (60.25%) 247 (61.75%)
Urgent 16 (4.05%) 19 (4.75%)
Why choose ER that day 0.413 0.719
Available conveniently 79 (20.00%) 85 (21.25%)
Refer by other health care providers 233 (58.99%) 227 (56.75%)
Feel sick, urgent with no other choice 83 (21.01%) 88 (22.00%)
General reason for choosing ER 3.3 0.344
Convenient 60 (15.19%) 70 (17.50%)
No improvement after medical consultation 44 (11.14%) 53 (13.25%)
No other help available 267 (67.59%) 246 (61.50%)
Feel great urgency 24 (6.08%) 31 (7.75%)

po0.05.

utilization. We expected that the intervention group would
have improved health outcome and reduced ER return
visits. Data shows that perceived improved health status
decreases ER return rate, and this health improvement is
significant in the nurse follow-up group. Paradoxically, the
group with nurse follow-up tends to have a higher chance
of returning to ER. We believe that this phenomenon can
be explained by factors that occur at the individual and
system level.
At the individual level, we suspect that the nurses’ call
have provided the clients with health advice to sustain the
improved health condition yet, at the same time sensitized
their needs to health care. With the sensitiza-tion of needs,
patients might be motivated to actively seek help from
health care professionals. As they would not access to
nurse counseling services in the commu-nity, they turned
to ER for help. This phenomenon is also reported in studies
where hospital readmission were
found to be higher with nurse follow-up (Hovendon &
Newton, 1996; Weinberger, Oddone, & Henderson, 1996).
It is believed that follow-up care provides channels for
patients to voice complaints and for early detection of
patients’ problems that may lead to readmission.
Runciman, Currie, Nicol, Green, and McKay (1996) also
found that elderly people who were followed up by health
visitor after ER discharge used significantly more services
in spite of an increased independence level when compared
with the control group.
Another factor at the individual level that may contribute
to ER return visits is related to patients’ help-seeking
behavior. Findings in this study suggest that some people
tend to doctor shop. The group of patients who visited GPs
after the index ER visit paradoxically were found to be
more likely to use the ER service again for the same
problem. Some studies
2214 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218

Table 3
A comparison of the variables in health care utilization and health outcome between the study and control group

Study (n=395) Control (n=400) Chi-square value (t-test value) p-value

Health outcome
Affect daily life 0.423 0.515
Yes 239 (60.51%) 251 (62.75%)
No 156 (39.49%) 149 (37.25%)
Improvement of condition 9.286 0.026
Getting worse 3 (0.76%) 5 (1.25%)
No change 8 (2.03%) 22 (5.50%)
Improved a bit 102 (25.82%) 117 (29.25%)
Recovered 282 (71.39%) 256 (64.00%)
Self-evaluated health 4.5 0.343
Not good 12 (3.04%) 13 (3.25%)
Fair 82 (20.76%) 88 (22.00%)
Nothing abnormal 99 (25.06%) 122 (30.50%)
Quite good 131 (33.16%) 119 (29.75%)
Very good 71 (17.97%) 58 (14.50%)
Customer satisfaction 0.27 0.603
Unsatisfied 6 (1.52%) 17 (4.25%)
Fairly satisfied 281 (71.14%) 267 (66.75%)
Satisfied 108 (27.34%) 116 (29.00%)

Heath care utilization

Times of attending GP with the same problem 1.03a 0.305
0 334 (84.57%) 347 (86.75%)
1 52 (13.16%) 45 (11.25%)
2 4 (1.01%) 6 (1.50%)
3 4 (1.01%) 2 (0.50%)
4 or above 1 (0.25%) 0 (0%)
Times of attending GOPC with the same problem 1.448a 0.148
0 299 (75.70%) 301 (75.25%)
1 70 (17.72%) 53 (13.25%)
2 17 (4.30%) 30 (7.50%)
3 6 (1.52%) 9 (2.25%)
4 or above 3 (0.76%) 7 (1.75%)
ER revisit within 30 days 4.4 0.036
No 274 (69.37%) 304 (76.00%)
Yes 121 (30.63%) 96 (24.00%)
a t value based on t-test.
po0.05.

found that the frequent users of primary care, i.e. the doctor
shoppers, were found to be the group with psychological
problems (Byrne et al., 2003; Demers, 1995), vague
somatic complaints, chronic conditions, drug or alcohol
dependence (Demers, 1995). In this study, the frequent
visits might be attributable to the relatively large
percentage of children, and parents needed repeated
reassurance of the conditions of their child (Zimmerman et
al., 1996). Another reason for patients to doctorshop
because they may feel dissatisfied with the quality of care
received by GPs as Hong Kong has not yet adopted
mandatory vocational training in family medicine (Lee et
al., 2000). A study examining the reasons for
disenrollement from a primary physician’s practice also
found that the quality of care was the
strongest predictor (Safran, Montgomery, Chang, Mur-phy,
& Rogers, 2001).
At the system level, we speculate that the high ER return
visits is related to access of care and poor interfacing
between primary and tertiary health care sectors. Results
showed that patients who visited GOPC for the same
problem after the index ER visit had a lower chance for
subsequent ER return visits. However, those who visited
GPs tended to return to ER more often. The two findings
were somewhat conflicting, making the relationship
between access of care and reduced ER re-visits uncertain.
The literature also reported mixed findings. Having a
regular source of care does not necessarily lead to the less
use of ER services. Padgett and Brodsky (1992) noted that
46% of
 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218 2215

Table 4
Risk factors for unplanned ER return visit within 30 days of the index visit, logistic regression result

Explanatory variables Beta coefficient S.E. Significance Odds ratio 95% C.I.for EXP(B)

Lower Upper

Gender
Female 1
Male 0.523 0.183 0.004 1.688 1.180 2.415
Group
Control group 1
Study group 0.433 0.183 0.018 1.542 1.078 2.207

Improved health condition 0.606 0.146 0.000 0.546 0.410 0.726

Times of attending GOPC with the same problem 0.375 0.142 0.006 0.688 0.526 0.898

Times of attending GP with the same problem 2.102 0.230 0.000 8.181 5.214 12.836
Considered doctors other than ER
Yes 1
No 0.527 0.193 0.006 1.694 1.160 2.473
Constant 0.054 0.547 0.921 1.056
po0.05.
po0.01.
po0.001.

the ER patients in a study in the Netherlands elected to use
ER rather than their own GP though all persons were
virtually assigned to GPs who offer 24-h services. This
concurred with a Canadian study which revealed that 62%
of subjects used ER for nonurgent medical problems had a
regular physician (Burnett & Grover, 1996). Both studies
suggested that a lack of access to medical care did not
seem to be a factor in patients’ decision to seek help from
ER. ER care was preferred to GP care for its convenience,
comprehensiveness and one-stop service for assessment
and treatment. It was found that it was not affordability, but
the availability and accessbility of comprehensive primary
health care services that caused patients to utilize
emergency services as an alternative to primary health care
services (Burnett & Grover, 1996; Lee et al., 2000). On the
other hand, some studies showed that patients who visited
medical outpatient clinics more were found to use less the
urgent care facilities such as ER (Totten et al., 1998; Okin
et al., 2000). Those who did not have access to providers in
other settings tend to identify ER as their regular source of
ambulatory care and had a lower frequency of use of
physician services (Baker, Stevens, & Brook, 1994). There
is a proportion of people who would use the emergency
department as their only available source of care (Kushel et
al., 2002; O’Brien et al., 1998). Patients who reported ER
as their regular source of care were three times more likely
to have used
the ER more than once in the preceding year (O’Brien et
al., 1998). Many studies in different parts of the world also
reported the ready availability of hospital care coupled with
lack of availability of primary care physicians being main
reasons for inappropriate utiliza-tion of A&E services
(Burnett & Grover, 1996; Walsh, 1995).
Our study has important implications to health care
policy. Policies are implied first by determining what
variables explain utilization (Andersen, 1995). Emer-gency
care and nonurgent ambulatory care are inex-tricably
linked and form the cornerstone of a universal-access
health care system (Gould, 2001). Findings disclosed that
24–30% of patients returned to ER, 25% visited GOPC,
16% visited GP for the same problem after the index ER
visit. These figures suggest that interfacing between public
and private sector, primary and tertiary care is weak in
Hong Kong. ER visits are costly, averaging HK $570
(approximate-ly=US$73.1) per visit. The ER service was
free to Hong Kong citizens till November 2002, and now
each patient had to pay HK$100 (approximately =
US$12.8) per visit (Hospital Authority, 2003). In the
private sector, each consultation visit is HK$150
(approximate-ly=US$19.2) (Gould, 2001), and this will be
out-of-pocket payment by the user. McGhee, Leung, and
Hedley (2001) urged the Hong Kong government to
impose supply side measures in maintaining access to
2216 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218
care in the interests of equity. These measures may include
regulations governing clinical practice, distribu-tion of
resources and reallocation between services to improve
allocative efficiency. In dealing with the seemingly overuse
of ER by nonurgent patients, we suggest the introduction of
a structured ER transitional care model. The rationale of
this proposal is based on our findings which revealed that
nurse follow-up after ER visits improved health outcomes
and improved health outcomes helped to reduce ER re-
visits. However, the nurse follow-up was not sustainable,
maybe because the intervention only involved 2 calls and
there was no structured arrangement for follow-up, for
example, formalized arrangement for GOPC/GP visits if
appro-priate. One study found that, with a more structured
transitional model of care post ER visits, nursing home
admissions at 30 days were lowered but the overall effect
on service use rates was insignificant. The intervention was
more effective for high-risk than low-risk elders (Mion et
al., 2003). Two similar studies were done with
contradictory findings which showed that the interven-tion
group had a higher service utilization rate (McCusker et al.,
2003; Runciman et al., 1996). However, when patients
were stratified into those with GP continued follow-up and
those without, it is interesting to note that the intervention
increased the visits to the GP in the former group and to
ER with the latter group (McCusker et al., 2003). These
studies concur with findings in this study which suggest
that the effects of post-ER follow-up intervention need to
be sustained with a formalized linkage and referral system
to primary care. Results showed that the percentage of
urgent cases is low (4.4%) which indicated that patients
who attended ER did not require emergency care.
However, the subjective need to seek help for these
patients was real. Self-assessed health was found to have a
strong association with use of health services, including
use of general and specialist practitioners (Dunlop, Coyte,
& McIsaac, 2000; Finkelstein, 2001; Jordan, Ong, & Croft
2003; Miilunpalo, Vuori, Oja, Pasanen, & Urponen, 1997)
and use of hospital and emergency room (Gold, Franks, &
Erickson, 1996; Harrison et al., 2002). A qualitative study
reveals that life distress and non availability of other care
access have influenced patients’ need for ER care, a
context that goes beyond professional judgement (Koziol-
McLain, Price, Weiss, Quinn, & Honigma, 2000). It is
therefore important for the ER transitional care model to
include interventions that capture possible risk factors at
the individual as well as system level. At the system level,
more structured interface between primary and tertiary care
should be arranged. At the individual level, the subjective
needs of the patients should be addressed. This includes
measures that attend to not only the somatic complaints,
but the possible underlying psycho-logical and/or social
basis of the problem.
Conclusion
Data shows that perceived improved health status
decrease ER return rate, and this improvement is
significant in the nurse follow-up group. Paradoxically, the
group with nurse follow-up tend to have a higher chance of
returning to ER. Counselling by nurses have improved the
health conditions but the intervention by nurses cannot be
sustained with present model of health care delivery in
Hong Kong, so patients return to ER for help. Health
service planning needs to aim at improving health, but also
ensuring efficient and effective access of quality primary
health care. Access means not only getting to service but
also getting to the right services at the right time to
promote improved health outcomes (Andersen &
Davidson, 2001). The use of ER may be effective when
health improvement and satisfaction are measured, but not
efficient for the level of care required by these frequenters
with a high percentage of ER return visits. The average
cost per ER visit was estimated HK$570 (approximately
US$73.1) so prevention of unnecessary ER re-visit would
have significant impact on cost savings (The Harvard
Team, 1999). A more structured ER transitional care model
is worth introdu-cing to test its effects for the nonurgent
patients. Nurse follow-up is found to be effective but not
sustainable. The ER transitional care model needs to
incorporate better interfacing between primary and tertiary
care, public and private care, in support of the nurse
follow-up call.
Acknowledgements
The authors would like to thank Vivian Wan, Lora Lam,
Vilna Leung, Vera Cheung, for their assistance in the
research project, Dr Robert Krane, Professor Ronald
Andersen and Professor Rance Lee for their valuable
comments on the initial draft of the manuscript and Dr
Tony Chan for his statistical advice. This paper is in
memory of the late ER nurse manager, Ms Ada Leung.
Source of funding of this project is University Grant
Council, Hong Kong (Project code G-YC73, The Hong
Kong Polytechnic University).
References
Andersen, R. M. (1995). Revisiting the behavioral model and
access to medical care: Does it matter? Journal of Health and
Social Behavior, 36, 1–10.
Andersen, R. M., & Davidson, P. L. (2001). In R. M. Andersen, T.
H. Rice, & G. F. Kominski (Eds.), Improving access to care in
America individual and contextual indicators (pp. 3– 30). San
Francisco: Jossey-Boss A Wiley Company.
 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218
Baker, D. W., Stevens, C., & Brook, R. H. (1994). Regular source
of ambulatory care and medical care utilization by patients
presenting to a public hospital emergency depart-ment. Journal
of the American Medical Association, 271, 1909–1912.
Be´land, F., Lemay, A., & Boucher, M. (1998). Patterns of visits
to hospital-based emergency rooms. Social Science &
Medicine, 47, 165–179.
Boushy, D., & Dubinsky, I. (1999). Primary care physician and
patient factors that result in patients seeking emergency care in
a hospital setting: The patient’s perspective. The Journal of
Emergency Medicine, 17, 405–412.
Briggs, J. K. (1997). Telephone triage protocols for nurses.
Philadelphia: Lippincott.
Burnett, M., & Grover, S. (1996). Use of the emergency
department for nonurgent care during regular business hours.
Canadian Medical Association Journal, 154, 1345–1351.
Byrne, M., Murphy, A. W., Plunkett, P. K., McGee, H. M.,
Murray, A., & Bury, G. (2003). Frequent attenders to an
emergency department: A study of primary health care use,
medical profile, and psychosocial characteristics. Annals of
Emergency Medicine, 41, 309–318.
Centers for Diseases Control & Prevention (CDC, USA) and
World Health Organization (1997). EpiInfo 6 Word Proces-
sing, Database and Statistics Program for Public Health
[Computer program]. Geneva, Switzerland: the Author.
Dale, J., Crouch, R., & Lloyd, D. (1998). Primary care: Nurse-led
telephone triage and advice out-of-hours. Nursing Standard,
12, 41–45.
Demers, M. (1995). Frequent users of ambulatory health care in
Quebec: The case of doctor-shoppers. Medical Association,
153, 37–42.
Dunlop, S., Coyte, P., & McIsaac, W. (2000). Socio-economic
status and the utilisation of physicians’ services: Results from
the canadian national population health survey. Social Science
& Medicine, 51, 123–133.
Finkelstein, M. M. (2001). Do factors other than need determine
utilization of physicians’ services in Ontario? Canadian
Medical Association, 165, 565–570.
Fung, H., Tse, N., & Yeoh, E. K. (1999). Health care reform and
societal values. Journal of Medicine and Philosophy, 24, 638–
652.
Gold, M., Franks, P., & Erickson, P. (1996). Assessing the health
of the nation: The predictive validity of a preference-based
measure and self-rated health. Medical Care, 34, 163–177.
Gould, D. B. (2001). The reform of health care funding. Hong
Kong Medical Journal, 7, 150–154.
Grossman, V. G. A. (1999). Quick reference to triage.
Philadelphia: Lippincott.
Harrison, M., Browne, G. B., Roberts, J., Tugwell, P., Gafni, A., &
Graham, I. D. (2002). Quality of life of individuals with heart
failure: A randomized trial of the effectiveness of two models
of hospital-to-home transition. Medical Care, 40, 271–282.
The Harvard Team (1999). Improving Hong Kong’s health care
system: Why and for whom? (main report and special reports)
Hong Kong: SAR Government.
Hospital Authority (2003). Fee and service guide. Retrieved
December 13, 2003 from http://www.ha.org.hk/hesd/nsapi/
2217
Hovendon, J., & Newton, W. P. (1996). Effect of primary care on
hospitalization. The Journal of Family Practice, 43, 534–535.
Jordan, K., Ong, B. N., & Croft, P. (2003). Previous consultation
and self reported health status as predictors of future demand
for primary care. Journal of Epidemiology & Community
Health, 57, 109–113.
Keith, K. D., Bocka, J. J., Kobernick, M. S., Krome, R. L., &
Ross, M. A. (1989). Emergency department revisits. Annals of
Emergency Medicine, 18, 964–968.
Kelly, A., Chirnside, A. M., & Curry, C. H. (1993). An analysis of
unscheduled return visits to an urban emergency department.
New Zealand Medical Journal, 106, 334–335.
Koziol-McLain, J., Price, D., Weiss, B., Quinn, A., & Honig-man,
B. (2000). Seeking care for nonurgent medical conditions in
the emergency department: Through the eyes of the patient.
Journal of Emergency Nursing, 26, 554–563.
Kushel, M. B., Perry, S., Bangsberg, D., Clark, R., & Moss, A.
(2002). Emergency department use among the homeless and
marginally housed: Results from a community-based study.
American Journal of Public Health, 92, 778–784.
Landefeld, C. S., Palmer, R., Kresevic, D., Fortinsky, R., &
Kowal, J. (1995). A randomized trial of care in a hospital
medical unit designed to improve the functional outcomes of
acutely ill older patients. The New England Journal of
Medicine, 332, 1338–1344.
Law, C. K., & Yip, P. S. F. (2002). Acute care service utilization
and the possible impacts of a user-fee policy in Hong Kong.
Hong Kong Medical Journal, 8, 348–353.
Lee, A., Lau, F., Hazlett, C. B., Kam, C., Wong, P., Wong, T.,
& Chow, S. (2000). Factors associated with non-urgent
utilization of Accident and Emergency services: A case-control
study in hong kong. Social Science & Medicine, 51, 1075–
1085.
Liu, T., Sayre, M. R., & Carleton, S. (1999). Emergency medical
care: Types, trends, and factors related to non-urgent visits.
Academic Emergency Medicine, 6, 1147–1152.
McCauley, J., Jenckes, M. W., Saeger-Proctor, C., McGuire, M. J.,
Ford, D. E., & Warwick, A. M. (1998). Frequent fliers:
Managed care patients with multiple emergency room visits.
Abstract Book Association for Health Services Research, 15,
39–40.
McCusker, J., Dendukuri, N., Tousignant, P., Verdon, J., de
Courval, L. P., Belzile, E. (2003). Academic Emergency
Medicine, 10, 233–243.
McGhee, S. M., Leung, G. M., & Hedley, A. J. (2001). Efficiency
is dependent on the control of supply. Hong Kong Medical
Journal, 7, 169–173.
Mion, L. C., Palmer, R. M., Meldon, S. W., Bass, D. M., Singer,
M. E., Payne, S. M. C., Lewicki, L. J., Drew, B. L., Connor, J.
T., Campbell, J. W., & Emerman, C. (2003). Case finding and
referral model for emergency department elders: A randomized
clinical trial. Annals of Emergency Medicine, 41, 57–68.
Miilunpalo, S., Vuori, I., Oja, P., Pasanen, M., & Urponen, H.
(1997). Self-rated health status as a health measure: The
predictive value of self-reported health status on the use of
physician services and on mortality in the working-age
population. Journal of Clinical Epidemiology, 50, 517–528.
Naylor, M., Brooten, D., Campbell, R., Jacobsen, B., Mezey, M.,
Pauly, M., & Schwartz, J. S. (1999). Comprehensive
2218 F.K.Y. Wong et al. / Social Science & Medicine 59 (2004) 2207–2218
discharge planning and home follow-up of hospitalized elders:
A randomized clinical trial. Journal of American Medical
Association, 281, 613–620.
O’Brien, R. P., & Miller, T. L. (1990). Urgent care center pediatric
telephone advice. American Journal of Emergency Medicine,
8, 496–497.
O’Brien, G. M., Stein, M. D., Zierler, S., Shapiro, M., O’Sullivan,
P., & Woolard, R. (1998). Use of the ED as a regular source of
care: Associated factors beyond lack of health insurance.
Annals of Emergency Medicine, 32, 113–114.
Okin, R. L., Boccellari, A., Azocar, F., Shumway, M., O’Brien,
K., Gelb, A., Kohn, M., Harding, P., & Wachsmuth, C. (2000).
The effects of clinical case management on hospital service
use among ED frequent users. American Journal of Emergency
Medicine, 18, 603–608.
Padgett, D. K., & Brodsky, B. (1992). Psychosocial factors
influencing non-urgent use of the emergency room: A review
of the literature and recommendations for research and
improved service delivery. Social Science & Medicine, 35,
1189–1197.
Pierce, J. M., Kellerman, A. L., & Oster, C. (1990). Bounces: An
analysis of short-term return visits to a public hospital
emergency department. Annals of Emergency Medicine, 19,
752–758.
Runciman, P., Currie, C., Nicol, M., Green, L., & McKay, V.
(1996). Discharge of elderly people from an accident and
emergency department: Evaluation of health visitor follow-up.
Journal of Advanced Nursing, 24, 711–718.
Safran, D. G., Montgomery, J. E., Chang, H., Murphy, J., &
Rogers, W. H. (2001). Switching doctor: Predictors of
voluntary disenrollment from primary physician’s practice.
The Journal of Family Practice, 50, 130–136.
Shi, L. (1997). Health services research method. Albany: Delmar
Publishers.
Simonsen, S. M. (1996). Telephone health assessment: guidelines
for practice. St Louis: Mosby.
The Harvard Team (1999). Improving Hong Kong’s health care
system: Why and for whom? (main and special reports). Hong
Kong: SAR Government.
Totten, A. M., Lin, S. L., Wagner, M., Lenz, E., Mundinger, M., &
Garfield, R. (1998). Characteristics and health services
utilization of medicaid ER and urgent care patients who keep
primary care follow-up appointments. Abstract Book
Association for Health Services Research, 15, 298–299.
Walsh, M. (1995). The health belief model and the use of accident
and emergency services by the general public. Journal of
Advanced Nursing, 22, 694–699.
Weinberger, M., Oddone, E. Z., & Henderson, W. G. (1996). Does
increased access to primary care reduce hospital readmissions?
The New England Journal of Medicine, 334, 1441–1446.
Wong, F. K. Y., Halloran, E., Ho, M., Chang, K., Leung, A.
(1999). A study of patients returning to A&E within 48 hours
in a regional hospital in Hong Kong. Unpublished raw data.
Young, G. P., Wagner, M. B., Kellermann, A. L., Ellis, J., &
Bouley, D. (1996). For the 24 hours in the ED study group.
Ambulatory visits to hospital emergency departments: Patterns
and reasons for use. Journal of American Medicine
Association, 276, 460–465.
Zimmerman, D. R., McCarten-Gibbs, K. A., DeNoble, D. H.,
Borger, C., Fleming, J., Hsieh, M., Langer, J. C., &
Breckenridge, M. B. (1996). Repeat pediatric visits to a
general emergency department. Annals of Emergency
Medicine, 28, 467–473.