Rattananon et al.
Predictors of Mortality among Inter-Hospital
Transferred Patients in a Middle-Income Country:
a Retrospective Cohort Study
Parin Rattananon, M.D.*, Isara Yenyuwadee, M.D.**, Tanchanok Dheeradilok, M.D.**, Parichaya Boonsoong,
M.D.*, Nintita Sripaiboonkij Tokanit, Dr.PH***, Salinthip Chimdist, M.D.*, Tawin Siwanuwatn, M.D.*,
Sidsadeeya Yuwapoom, M.D.*, Paibul Suriyawongpaisal, M.D.****
*Prachuap Khiri Khan Hospital, Ministry of Public Health, Prachuap Khiri Khan 77000, Tailand, **Tap Sakae Hospital, Ministry of Public Health,
Prachuap Khiri Khan 77000, Tailand, ***Ramathibodi Comprehensive Cancer Center, Ramathibodi Hospital, Mahidol University, Bangkok 10400,
Tailand, ****Department of Community Medicine, Faculty of Medicine, Ramathibodi Hospital, Mahidol University, Bangkok 10400, Tailand.
ABSTRACT
Objective: To identify predictors for hospital mortality among inter-hospital transferred patients in low-resource
settings of rural hospitals in Tailand.
Methods: We conducted a retrospective cohort study of patients transferred from emergency room(ER) of a community
hospital to its designated tertiary care hospital in a western province of Tailand. During March 2018 and February 2019,
medical records of 412 patients were reviewed and extracted for potential predictor variables and outcomes. We defined
deaths within 72 hrs afer a transfer as primary outcome and overall hospital mortality as secondary outcome. Multivariate
logistic regression analysis was performed to identify predictors of the outcomes adjusted for potential confounders.
Results: Out of 412 patients, a total of 37 patients (9.0%) died during the stay in receiving hospital and 18 (4.4%) of
them died within 72 hrs afer transfer. Top ten primary diagnostic categories included road traffic injuries (19.7%),
acute appendicitis (9.7%), and acute myocardial infarction (5.1%). Univariate analysis revealed early mortality (<72
hrs) was associated with NEWS2, Emergency Severity Index (ESI), cardiac arrest prior to transfer, use of vasoactive
agents, endotracheal intubation and admitting service. Using multiple logistic regression model adjusted for the
predictors identified by univariate analysis, we found early mortality was independently associated with NEWS2 ≥
9 (compared to NEWS2 0-6) with OR= 17.51(95%CI 3.16-97.00) and vasoactive medication use (OR= 5.46, 95%CI
1.39-21.46). Similarly, overall mortality was also independently associated with NEWS2 ≥ 9(OR= 4.76, 95%CI
1.31-17.36) and vasoactive medication use (OR= 7.51,95%CI 2.76 -20.45).
Conclusion: Tis study identified predictors of early (<72 hrs) hospital mortality and overall hospital mortality
among ER patients transferred from a rural community hospital to its designated tertiary care hospital in Tailand,
a middle-income country with universal healthcare coverage. Te findings might be helpful to inform decision-
making dealing with the inter-hospital transfer of ER patients in resource-poor rural settings with similar case-mix.
Keywords: Patient transfer; critical illness; prognosis; mortality (Siriraj Med J 2021; 73: 312-321)
INTRODUCTION
and coordination from varied healthcare providers.1 Te
Inter-hospital transfer(IHT) is considered a complex
transitional process is vulnerable for discontinuity error,
and challenging practice, requiring multiple resources
combining with restricted resources outside hospital
Corresponding author: Parin Rattananon
E-mail: parin.rt@gmail.com
Received 15 October 2020 Revised 1 April 2021 Accepted 2 April 2021
312
Volume 73, No.5: 2021 Siriraj Medical Journal
Original Article SMJ
settings during transport, IHT patients are at risk of
Setting
adverse events and unsatisfied outcomes.2
Our study involved ER patients transferred from
Additional to the systemic threats, growing evidence
a community hospital to its designated tertiary care
demonstrated higher acute severity, a longer length of
hospital in a western province of Tailand, a middle-
stay, higher hospital mortality and higher resources use
income country with universal healthcare coverage.
in IHT patients when compared to non-IHT cases.3-7
Te community hospital is a 60-bed public hospital
Tese undesirable outcomes of IHT patients could be
(No intensive care bed) staffed with 1 pediatrician, 7
due to heterogeneity among IHT patients depending
general practitioner physicians, 5 pharmacists, and 54
on the diagnosis, presenting a nuanced assessment of
nurses. Four ambulances equipped with an oxygen tank,
this complex care transition.8 Variability in transfer
suction, blood pressure monitor, and a defibrillator. are
practices means ambiguity and subjectivity in decision
available for IHT and Emergency Medical Services. At
making between transferring physicians and receiving
ER of the community hospital, there are 1 physician, 3
physicians.9,10 Standardization of the care processes is
ER nurses, and 2 assistant nurses for each 8-hour shif.
considered a means to minimize the variability, which
Te estimated nurse-to-patient ratio in the ER is 1 to 9.
is amenable to improving the quality of care among IHT
Te estimated annual number of IHT patients from ER
patients.11
and inpatient care are 750. Te receiving hospital is a
According to earlier studies, prognostic factors for
278-bed (12 intensive care beds) tertiary hospital staffed
early death (<72 hrs ) included male gender, summer
with 4 internists, 1 gastroenterologist, 1 nephrologist, 4
season, admitting service, diagnostic related group
general surgeons, 2 neurosurgeons, 3 orthopedic surgeons,
level, Charlson Comorbidity Score, insurance type,
2 ophthalmologists, 3 obstetricians, and 2 pediatricians.
and major diagnostic category. For overall hospital
Te distance between the two hospitals is 43 kilometers,
mortality, prognostic factors included length of stay,
with an average ground transport time of 30 minutes.
medical complication, distance traveled, insurance type,
When a transfer decision is determined, a primary care
and major diagnostic category.5,6,8 Application of such
doctor will contact the transfer operation center in the
knowledge in overcrowded emergency room (ER) settings
receiving hospital. Afer receiving the referral request,
is a challenge.
the center, operated by registered nurses, will notice the
As a result several triage systems have been proposed
specialist and present all the patient information. Te
and were found to be significantly related with admission
teleconsultant will be provided for initial management.
rate and medical resource consumption.,4,5 According to
If the referral request is accepted, the patient will be
previous reports, triage systems such as Acute Physiology
transported to the emergency department (ED) of the
and Chronic Health Evaluation (APACHE) or Sequential
tertiary hospital, where the patient’s conditions are
Organ Failure Assessment (SOFA) were frequently applied
reevaluated before a decision for hospitalization. ER
to estimate disease severity in IHT patients.4,5,12,13 However,
patients deemed a need for IHT are accompanied by
some parameters (e.g., arterial oxygenation and blood
an ambulance staffed with a nurse and a nurse assistant.
pH) in these scoring systems may not be available at ER
As there is no clinician accompanies the ambulance,
of rural community hospital settings where resources
the emergency patient needs to be stabilized enough
are limited.
before transfer.
In Tailand, many hospitals, especially in rural areas,
have no standardized decision-support and communication
Study design
tool during patient transfer. Even in a similar patient,
A retrospective cohort study was conducted during
management decisions may differ as there is variation in
March 2018 and February 2019. We included adult
clinical practices among physicians. Tis study intends to
patients aged 16 or above who were transferred from
identify predictors of IHT patients using basic parameters,
ER of the transferring hospital and hospitalized at the
which are generally available at ER of rural community
tertiary care hospital. We excluded obstetric patients,
hospitals in Tailand. Te expected findings might be
pediatric patients, IHT patients not hospitalized at the
useful to facilitate patient care during IHT.
receiving hospital and patients with incomplete data.
Patients with multiple transfers were considered the
MATERIALS AND METHODS
same episode.
This study was approved by the Office for
Te authors, working independently in two teams,
Research Ethics Committee of Hua Hin Hospital, Prachuap
reviewed all the extracted data from electronic and/or
Khiri Khan, Tailand (RECHHH145/2019).
paper-based medical records using a standard data form.
Volume 73, No.5: 2021 Siriraj Medical Journal
313
Rattananon et al.
Te first team, working as primary care doctor in the
Data analysis
community hospital, documented patients’ characteristics
Data analysis was conducted using STATA statistical
consisting of demographics, health insurance status,
sofware version 14. Continuous and categorical variables
primary diagnosis categories based on the International
were presented as means with standard deviation (SD)
Statistical Classification of Diseases and Related Health
and as frequencies with percentages, respectively. To
Problems (ICD-10), underlying diseases, past medical
identify potential predictors, patient characteristics of
history, physiological parameters and severity categories
those with or without the outcomes were compared using
according to the Emergency Severity Index (ESI). Te
Student’s t-test for continuous variables and Chi-square
ESI is a five-level triage scale, ranging from level 5 (Non-
test for categorical variables.
urgent) to ESI level 1 (Resuscitative), based on patient
Multivariate logistic regression models using
acuity and resource needs.14 Te ESI system has been
backward stepwise regression for variables selection
used primarily in Tailand for triaging ER patients.15
were developed to identify predictors of the outcomes.
National Early Warning Score 2 (NEWS2) for each
Parameters associated with a p-value below 0.25 were
patient was calculated from the physiological parameters
included in the initial model. Highly related parameters
on arrival at the ER to represent acute severity index of
were removed to diminish multicollinearity. Least significant
IHT patients. Tis aggregated scoring system is built from
factors were deleted one by one according to a backward
six basic parameters including respiratory rate, oxygen
elimination algorithm until reaching the final models.
saturation, temperature, systolic blood pressure, heart
Te receiver operating characteristic curve (ROC) was
rate, and level of consciousness.16 Underlying diseases
developed with a calculated area under the curve(AUC)
and past medical history were reviewed and calculated
to inform model performance. P-values (p) less than
into the Charlson’s comorbidity score.17 Apart from
0.05 were considered as statistically significant.
those variables, the following were also included: events
before the transfer (cardiac arrest, use of vasoactive drugs,
RESULTS
and endotracheal intubation); transfer time in minutes
Tere were 519 patients transferred from ER of
(starting from a patient’s arrival at the transferring hospital
the community hospital to the designated receiving
until admission at the receiving hospital). Te second
hospital during the study period (Fig 1). Afer applying
team, working as a general practitioner at the receiving
the inclusion and exclusion criteria, 412 patients were
hospital, extracted patient outcomes from electronic
entered into the study. Among them, 11 patients revisited
health records, consisting of diagnosis based on ICD-
ER of the transferring hospital and were re-hospitalized
10, length of stay, and discharge status. Within 72-hour
to the tertiary hospital twice, and 3 more patients faced
mortality afer IHT was considered primary outcome
these experiences for three times. Tirty-seven patients
and overall hospital mortality as secondary outcome.
(9.0%) died upon discharge, half of them died within
Fig 1. Flow diagram of included and excluded patients.
314
Volume 73, No.5: 2021 Siriraj Medical Journal
Original Article SMJ
three days afer a transfer). Tirty-eight patients were
Te predictors discovered from our study allow
discharged home or transferred back to the community
healthcare providers to estimate the severity of the ER
hospital or transferred to a higher-level hospital within
patients who might need transfer to other hospitals
72 hrs of the admission.
capable of providing definitive care. Scoring systems
Out of 412 patients, a total of 37 patients (9.0%) died
such as NEWS2 provided a standardized tool for clinical
during the stay in receiving hospital and 18 (4.4%) of
monitoring and assessment. By combining physiological
them died within 72 hrs afer transfer (Table 1). Table 2
variables into scores, it reduces variation in assessing
demonstrates top ten primary diagnostic categories
patient status among healthcare professionals. Several
including road traffic injuries (19.7%), acute appendicitis
triage systems, including ESI, have been developed
(9.7%), and acute myocardial infarction (5.1%). Univariate
for use in the ER. However, they are not designed to
analysis (Table 1) reveals early mortality (<72 hrs) was
detect deterioration in patients.21 NEWS can further risk
associated with NEWS2, Emergency Severity Index
stratifying patients within higher ESI risk categories, both
(ESI), cardiac arrest prior to transfer, use of vasoactive
for death and need for admission.22 Patients with a high
agents, endotracheal intubation and admitting service.
NEWS score have not only been identified as being at
For overall mortality, univariate analysis identified age
risk of a poor outcome but have already physiologically
and Charlson’s co-morbidity score as predictors in
deteriorated to the extent where urgent medical review
addition to those for early mortality. Using multiple
and intervention is required. With a common scoring
logistic regression model adjusted for the predictors
system between facilities, it also functions as a standard
identified by univariate analysis (Table 3), we found early
language in communication on patient’s clinical acuity.23
mortality was independently associated with NEWS2 ≥ 9
Out of 412 transfer patients (mean age 53) from
(compared to NEWS2 0-6) with OR= 17.51(95%CI 3.16
the transferring hospital to the receiving hospital (43
- 97.00) and use of vasoactive medication (OR= 5.46,
km apart), 9.0% died upon discharge with a half died
95%CI 1.39-21.46). Similarly, overall mortality was also
within 72-h afer the transfer. We could not identify other
independently associated with NEWS2 ≥ 9(OR= 4.76,
studies in a similar setting both in high-income countries
95%CI 1.31 - 17.36) and use of vasoactive medications
(HICs) and low-middle income countries (LMICs) for
(OR= 7.51,95%CI 2.76 - 20.45) (Table 4). Performance
mortality comparison. Our overall-mortality figure is,
of the multivariate models were validated with AUC
at most, one-third of the reported figures from several
0.91 (95% CI 0.82-0.99) for the first model (Table 3) and
other studies dealing with intensive care patients.12,24
0.88 (95% CI 0.83-0.94) for the second model (Table 4).
Tis indicates our patients were in much less critical
conditions than those in other studies. Finally, similar
DISCUSSION
to findings from other studies7,8, the patients’ profiles of
Applying multiple logistic regression analysis to
our study were heterogeneous (Table 2).
the cohort data (N=412), we were able to identify two
In our study, we found no association between
independent predictors for early mortality: NEWS2
transfer time and patient mortality, which is compatible
score ≥ 9 (OR: 17.51; 95% CI 3.16-97.00, p=0.001) and
with previous similar studies.12,13 As suggested from many
vasoactive agent use (OR 5.46; 95% CI 1.39-21.46, p=0.015).
guidelines for the interfacility transport, our finding also
NEWS2 is used internationally as an early warning
supports a “stabilize and shif” approach rather than a
score for triaging in ER and monitoring hospitalized
“scoop and run” strategy.25-27 However, even though
patients. From the Royal College of Physicians report, the
there is no significant relationship between transfer times
aggregated score of 7 or more is defined as a threshold
and hospital mortality, some studies have demonstrated
for emergency response, and patient transfer to a higher
the benefit of appropriate, timely referrals in lessening
setting facility should be considered.16 Our findings
complications, length of stay, and morbidity of IHT
are comparable with previous studies that reported
patients.28,29 Additionally, certain diseases such as ST-
high acute severity index and events such as cardiac
segment elevation myocardial infarction or expanding
arrest, mechanical ventilation, and vasoactive drug use
intracranial hematoma, are considered as time-sensitive
as mortality predictor in IHT patients.12,13,18 With ROC
emergency conditions.30,31 Delays to definite treatment in
0.91(95% CI 0.82-0.99), our model performs as high as
such diseases could result in lethal outcomes. We conclude
that of other studies in HICs and LMICs, although the
that, in general, critically ill patients should be resuscitated
results, in this regard, may not be directly comparable
until achieving possibly maximum stabilization by the
given different sets of predictors and study settings.19,20
referring hospital before the interhospital transport
without unnecessary delays.
Volume 73, No.5: 2021 Siriraj Medical Journal
315
Rattananon et al.
TABLE 1. Patient characteristics and admitting service categorized by the outcome status.
Within 72 hrs
Overall
Variables
All patients
Alive
Dead
Alive
In-hospital
p-value
p-value
(n = 412)
(n = 394)
(n = 18)
(n = 375)
Death (n = 37)
Patient characteristics
Age, mean years (±SD)
53 (±20)
53 (±20)
59 (±20)
0.18
52 (±19)
64 (±19)
<0.001
Gender, male, n (%)
245 (59.5)
235 (59.6)
10 (55.6)
0.73
220 (58.7)
25 (67.6)
0.293
Health insurance status, n (%)
0.733
0.204
Universal Coverage
268 (65.1)
254 (64.5)
14 (77.8)
235 (62.7)
30 (81.1)
Compulsory Motor Insurance
78 (18.9)
75 (19.0)
3 (16.7)
77 (20.5)
4 (10.8)
Social Security Scheme
17 (4.1)
17 (4.3)
0 (0.0)
17 (4.5)
0 (0.0)
CSMBS
42 (10.2)
41 (10.4)
1 (5.6)
39 (10.4)
3 (8.1)
Out-of-pocket
7 (1.7)
7 (1.8)
0 (0.0)
7 (1.9)
0 (0.0)
Transfer time, mean minutes (±SD)
226 (±97)
227 (±98)
212 (±74)
0.531
226 (±98)
232 (±81)
0.733
Charlson’s co-morbidity score, n (%)
0.533
19 (5.1)
0 (0.0)
0.002
0
149 (36.2)
145 (36.8)
4 (22.2)
144 (38.4)
5 (13.5)
1-2
137 (33.3)
131 (33.3)
6 (33.3)
125 (33.3)
12 (32.4)
3-4
94 (22.8)
88 (22.3)
6 (33.3)
81 (21.6)
13 (35.1)
>4
32 (7.8)
30 (7.6)
2 (11.1)
25 (6.7)
7 (18.9)
NEWS2, mean (±SD)
4 (±4)
3 (±3)
12 (±4)
<0.001
3 (±3)
9 (±4)
<0.001
316
Volume 73, No.3: 2021 Siriraj Medical Journal
Original Article SMJ
TABLE 1. Patient characteristics and admitting service categorized by the outcome status. (Continue)
Within 72 hrs
Overall
Variables
All patients
Alive
Dead
Alive
In-hospital
p-value
p-value
(n = 412)
(n = 394)
(n = 18)
(n = 375)
Death (n = 37)
The ESI (Level of urgency), n (%)
<0.001
<0.001
1 (Resuscitative)
35 (8.5)
27 (6.9)
8 (44.4)
22 (5.9)
13 (35.1)
2 (Emergent)
101 (24.5)
95 (24.1)
6 (33.3)
86 (22.9)
15 (40.5)
3 (Urgent)
161 (39.1)
157 (39.9)
4 (22.2)
153 (40.8)
8 (21.6)
4 (Less urgent)
111 (26.9)
111(28.2)
0 (0.0)
110 (29.3)
1 (2.7)
5 (Non-urgent)
4 (1.0)
4 (1.0)
0 (0.0)
4 (1.1)
0 (0.0)
Cardiac arrest prior to transfer, yes (%)
8 (1.9)
2 (0.5)
6 (33.3)
<0.001
1 (0.3)
7 (18.9)
<0.001
Any vasoactive agent, yes (%)
32 (7.8)
20 (5.1)
12 (66.7)
<0.001
13 (3.5)
17 (46.0)
<0.001
Endotracheal intubation prior to transfer, yes (%)
68 (16.5)
55 (14.0)
13 (72.2)
<0.001
46 (12.3)
22 (59.5)
<0.001
Admitting service
Inpatient department, n (%)
0.003
<0.001
Internal Medicine
136 (33.0)
123 (31.2)
13 (72.2)
108 (28.8)
27 (73.0)
General Surgery
161 (39.1)
159 (40.4)
2 (11.1)
96 (25.6)
5 (13.5)
Neurosurgery
49 (11.9)
46 (11.7)
3 (16.7)
74 (19.7)
4 (10.8)
Orthopedic
47 (11.4)
47 (11.9)
0 (0.0)
42 (11.2)
1 (2.7)
Others*
19 (4.6)
19 (4.8)
0 (0.0)
36 (9.6)
0 (0.0)
Abbreviations: CSMBS, Civil Servant Medical Benefit Scheme; ESI, Emergency Severity Index; ETT, Endotracheal tube; NEWS2, National Early Warning Score 2; SD, Standard deviation.
* Others include Gynecology, Ophthalmology, and Otorhinolaryngology
Volume 73, No.3: 2021 Siriraj Medical Journal 317
Rattananon et al.
Supplementary Table 1. Characteristics of study patients according to mortality status within the same admission
afer transfer.
All patients
In-hospital
Variables
Alive (n = 375)
p-Value
(n = 412)
Death (n = 37)
Age, mean years (±SD)
53 (±20)
52 (±19)
64 (±19)
<0.001
Gender, male (%)
245 (59.5)
220 (58.7)
25 (67.6)
0.293
Health Insurance status, n (%)
0.234
Universal Coverage
268 (65.1)
238 (63.5)
30 (81.1)
Compulsory Motor Insurance
78 (18.9)
74 (19.7)
4 (10.8)
Social Security Scheme
17 (4.1)
17 (4.5)
0 (0.0)
CSMBS
42 (10.2)
39 (10.4)
3 (8.1)
Out-of-pocket
7 (1.7)
7 (1.9)
0 (0.0)
Transfer time, mean minutes (±SD)
226 (±97)
226 (±98)
232 (±81)
0.733
Inpatient department, n (%)
<0.001
Internal Medicine
136 (33.0)
109 (29.1)
27 (73.0)
General Surgery
161 (39.1)
154 (41.1)
7 (18.9)
Neurosurgery
49 (11.9)
46 (12.3)
3 (8.1)
Orthropedic
47 (11.4)
47 (12.5)
0 (0.0)
Others*
19 (4.6)
19 (5.1)
0 (0.0)
Charlson’s co-morbidity score, n (%)
0.002
0
149 (36.2)
144 (38.4)
5 (13.5)
1-2
137 (33.3)
125 (33.3)
12 (32.4)
3-4
94 (22.8)
81 (21.6)
13 (35.1)
>4
32 (7.8)
25 (6.7)
7 (18.9)
NEWS2, mean (±SD)
4 (±4)
3 (±3)
10 (±4)
<0.001
ESI scores (Level of urgency), n (%)
<0.001
1 (Resuscitative)
35 (8.5)
22 (5.9)
13 (35.1)
2 (Emergent)
101 (24.5)
86 (22.9)
15 (40.5)
3 (Urgent)
161 (39.1)
153 (40.8)
8 (21.6)
4 (Less urgent)
111 (26.9)
110 (29.3)
1 (2.7)
5 (Non-urgent)
4 (1.0)
4 (1.1)
0 (0.0)
Cardiac arrest prior to transfer, yes (%)
8 (1.9)
1 (0.3)
7 (18.9)
<0.001
Any vasoactive agent, yes (%)
32 (7.8)
14 (3.7)
18 (48.7)
<0.001
ETT insertion prior to transfer, yes (%)
68 (16.5)
46 (12.3)
22 (59.5)
<0.001
Abbreviations: CSMBS, Civil Servant Medical Benefit Scheme; ESI, Emergency Severity Index; ETT, Endotracheal tube; NEWS2, National
Early Warning Score 2; SD, Standard deviation.
* Others include Gynecology, Ophthalmology, and Otorhinolaryngology.
318
Volume 73, No.5: 2021 Siriraj Medical Journal
Original Article SMJ
TABLE 2. Most common primary diagnoses according to ICD-10.
Early mortality*, n (%)
Overall mortality, n (%)
Primary diagnostic categories with ICD-10
Alive
Dead
Alive
Death
All patients (n = 412)
(n = 394)
(n = 18)
(n = 375)
(n = 37)
C15-C26 Malignant neoplasms of digestive organs
9 (2.3)
0 (0.0)
7 (1.9)
2 (5.4)
(n=9, 2.2%)
I21 Acute myocardial infarction
19 (4.8)
2 (11.1)
16 (4.3)
5 (13.5)
(n=21, 5.1%)
I61 Intracerebral haemorrhage
20 (5.1)
0 (0.0)
20 (5.3)
0 (0.0)
(n=20, 4.9%)
I63 Cerebral infarction
16 (4.1)
0 (0.0)
16 (4.3)
0 (0.0)
(n=16, 3.9%)
J12-J18 Pneumonia
13 (3.3)
2 (11.1)
11 (2.9)
4 (10.8)
(n=15, 3.6%)
K27 Gastric ulcer with perforation
9 (2.3)
0 (0.0)
9 (2.4)
0 (0.0)
(n=9, 2.2%)
K35 Acute appendicitis
40 (10.2)
0 (0.0)
40 (10.7)
0 (0.0)
(n=40, 9.7%)
K92.2 Gastrointestinal haemorrhage, unspecified
17 (4.3)
0 (0.0)
17 (4.5)
0 (0.0)
(n=17, 4.1%)
S72 Fracture of femur
11 (2.8)
0 (0.0)
11 (2.9)
0 (0.0)
(n=11, 2.7%)
V01-V99 Road traffic injuries
78 (19.8)
3 (16.7)
77 (20.5)
4 (10.8)
(n=81, 19.7%)
Other diagnoses
162 (41.1)
11 (61.1)
151 (40.3)
22 (59.5)
(n=173, 42.0%)
ICD-10, the International Statistical Classification of Diseases and Related Health Problems.
* Defined as death within 72 hrs afer an inter-hospital transfer
TABLE 3. Multivariate logistic regression analysis of factors associated with early mortality (< 72 hrs) (n = 412).
Variables
OR
95% CI
p
NEWS2
7-8 vs. 0-6
6.61
0.77-56.62
0.085
≥ 9 vs. 0-6
17.51
3.16-97.00
0.001
Cardiac arrest prior to transfer
5.37
0.79-36.54
0.086
Vasoactive agent use
Yes vs. No
5.46
1.39-21.46
0.015
Abbreviations: NEWS2, National Early Warning Score 2; OR, Odds ratio; p, p-value
Volume 73, No.5: 2021 Siriraj Medical Journal
319
Rattananon et al.
TABLE 4. Multivariate logistic regression analysis of factors associated with overall mortality (n = 412).
Variables
OR
95% CI
p
NEWS2
7-8 vs. 0-6
1.49
0.32-6.84
0.608
≥ 9 vs. 0-6
4.76
1.31-17.36
0.018
Age
1.02
1.00-1.05
0.076
Endotracheal intubation prior to transfer
2.28
0.73-7.17
0.158
Vasoactive agent use
Yes vs. No
7.51
2.76-20.45
<0.001
Abbreviations: NEWS2, National Early Warning Score 2; OR, Odds ratio; p, p-value
Another interesting finding from our study is an
CONCLUSION
apparent degree of unplanned ER revisits and re-transfers.
To our best knowledge, our study may be the first
Tese events may be explained either by the nature and
demonstrating outcome predictors of inter-hospital
severity of individual diseases or inappropriate post-
transfer patients in Tailand and low- and middle-income
discharge follow-up care. Because most patients would
countries. We managed to identify predictors of hospital
receive follow-up care afer discharge at their transferring
mortality for transfer patients from a rural hospital ER
hospital, appropriateness of discharge communication
to a receiving hospital i.e., high NEWS2 scores and use
about a follow-up plan from the receiving hospital could
of vasoactive agents. Tese factors could be used to
improve the quality of care at the transferring hospital.32
standardize rationale and clinical care processes in ER
Future studies should explore deeper to clarify the causes
patients transferred from rural community hospitals to
of repeated transfers in our area.
other hospitals capable of providing definitive care. With
Our present study has three potential limitations
NEWS2 included among the predictors, we were able
which need consideration. Firstly, this study was conducted
to suggest using NEWS2 as a value-added tool to better
in a single hospital in a rural area of Tailand and its
monitoring of the patients’ status during the transfer
designated tertiary care hospital. Patient characteristics
and facilitate a mutual agreement between clinicians.
and performance in transfer practices may be different
from other hospital settings. For this reason, external
ACKNOWLEDGMENTS
validity is uncertain, so results from this research should
Te authors want to express gratitude to Dr. Supaporn
be carefully examined before application. Secondly, the
Pamonsut, a hospital director at Tap Sakae Hospital,
number of included patients in the retrospective cohort
Prachuap Khiri Khan, Tailand, for providing us valued
may not be large enough, as indicated by wide confidence
comments. We also would like to thank Mr. Stephen Pinder,
intervals. With a small sample size, the power of tests
a Medical Education/ Medical English specialist at the
may not be sufficient to detect a statistically significant
Department of Clinical Epidemiology and Biostatistics,
association in some clinically relevant parameters. Lastly,
Ramathibodi Hospital, Mahidol University, for proof
we have not accounted for adverse incidents during
reading and grammar correction of our manuscript.
inter-hospital transport as a predictor variable in our
study due to inaccessible data and/or unavailability of
Data availability: Te datasets used to support the findings
data. Those unexpected events are common during
of this study are available upon request.
transport and could greatly influence the outcomes in
critically ill patients.33 Hence, further studies are needed
Conflicts of interest: Te authors state that they have
to explore this key area of healthcare with complexity,
no Conflict of Interest (COI).
which is understudied, especially in LMICs.
Funding statement: Tis research received no external
funding.
320
Volume 73, No.5: 2021 Siriraj Medical Journal
Original Article SMJ
REFERENCES
18.
Durairaj L, Will JG, Torner JC, Doebbeling BN. Prognostic factors
1.
Kiss T, Bolke A, Spieth PM. Interhospital transfer of critically
for mortality following interhospital transfers to the medical
ill patients. Minerva Anestesiol 2017;83:1101-8.
intensive care unit of a tertiary referral center. Crit Care Med
2.
Eiding H, Kongsgaard UE, Braarud AC. Interhospital transport
2003;31:1981-6.
of critically ill patients: experiences and challenges, a qualitative
19.
Haniffa R, Isaam I, De Silva AP, Dondorp AM, De Keizer NF.
study. Scand J Trauma Resusc Emerg Med 2019;27:27.
Performance of critical care prognostic scoring systems in
3.
Sokol-Hessner L, White AA, Davis KF, Herzig SJ, Hohmann
low and middle-income countries: a systematic review. Crit
SF. Interhospital transfer patients discharged by academic
Care 2018;22:18.
hospitalists and general internists: Characteristics and outcomes.
20.
Knaus WA, Zimmerman JE, Wagner DP, Draper EA, Lawrence
J Hosp Med 2016;11:245-50.
DE. APACHE-acute physiology and chronic health evaluation: a
4.
Golestanian E, Scruggs JE, Gangnon RE, Mak RP, Wood KE.
physiologically based classification system. Crit Care Med 1981;
Effect of interhospital transfer on resource utilization and
9:591-7.
outcomes at a tertiary care referral center. Crit Care Med 2007;
21.
Ludikhuize J, Smorenburg SM, de Rooij SE, de Jonge E.
35:1470-6.
Identification of deteriorating patients on general wards;
5.
Hill AD, Vingilis E, Martin CM, Hartford K, Speechley KN.
measurement of vital parameters and potential effectiveness
Interhospital transfer of critically ill patients: demographic
of the Modified Early Warning Score. J Crit Care 2012;27:424.
and outcomes comparison with nontransferred intensive care
e7-13.
unit patients. J Crit Care 2007;22:290-5.
22.
Alam N, Vegting IL, Houben E, van Berkel B, Vaughan L,
6.
Hernandez-Boussard T, Davies S, McDonald K, Wang NE.
Kramer MHH, et al. Exploring the performance of the National
Interhospital Facility Transfers in the United States: A Nationwide
Early Warning Score (NEWS) in a European emergency
Outcomes Study. J Patient Saf 2017;13:187-91.
department. Resuscitation 2015;90:111-5.
7.
Mueller SK, Zheng J, Orav J, Schnipper JL. Interhospital Transfer
23.
Brangan E, Banks J, Brant H, Pullyblank A, Le Roux H, Redwood
and Receipt of Specialty Procedures. J Hosp Med 2018;13:
S. Using the National Early Warning Score (NEWS) outside
383-7.
acute hospital settings: a qualitative study of staff experiences
8.
Mueller S, Zheng J, Orav EJ, Schnipper JL. Inter-hospital
in the West of England. BMJ Open 2018;8:e022528-e022528.
transfer and patient outcomes: a retrospective cohort study.
24.
Riviello ED, Kiviri W, Fowler RA, Mueller A, Novack V,
BMJ Qual Saf 2019;28:e1.
Banner-Goodspeed VM, et al. Predicting Mortality in Low-
9.
Bosk EA, Veinot T, Iwashyna TJ. Which patients and where:
Income Country ICUs: Te Rwanda Mortality Probability
a qualitative study of patient transfers from community hospitals.
Model (R-MPM). PLoS One 2016;11:e0155858.
Med Care 2011;49:592-8.
25.
Warren J, Fromm RE, Jr., Orr RA, Rotello LC, Horst HM,
10.
Wagner J, Iwashyna TJ, Kahn JM. Reasons underlying interhospital
American College of Critical Care M. Guidelines for the inter-
transfers to an academic medical intensive care unit. J Crit
and intrahospital transport of critically ill patients. Crit Care
Care 2013;28:202-8.
Med 2004;32:256-62.
11.
Mueller SK, Zheng J, Orav EJ, Schnipper JL. Rates, Predictors
26.
Sethi D, Subramanian S. When place and time matter: How
and Variability of Interhospital Transfers: A National Evaluation.
to conduct safe inter-hospital transfer of patients. Saudi J
J Hosp Med 2017;12:435-42.
Anaesth 2014;8:104-13.
12.
Strauch U, Bergmans DC, Winkens B, Roekaerts PM. Short-
27.
Kulshrestha A, Singh J. Inter-hospital and intra-hospital patient
term outcomes and mortality afer interhospital intensive
transfer: Recent concepts. Indian J Anaesth 2016;60:451-7.
care transportation: an observational prospective cohort study
28.
Duke GJ, Green JV. Outcome of critically ill patients undergoing
of 368 consecutive transports with a mobile intensive care
interhospital transfer. Med J Aust 2001;174:122-5.
unit. BMJ Open 2015;5:e006801.
29.
Catalano AR, Winn HR, Gordon E, Frontera JA. Impact of
13.
Patel JJ, Kurman J, Al-Ghandour E, Tandra K, Mawari S, Graf
interhospital transfer on complications and outcome afer
J, et al. Predictors of 24-h mortality afer inter-hospital transfer to
intracranial hemorrhage. Neurocrit Care 2012;17:324-33.
a tertiary medical intensive care unit. J Intensive Care Soc
30.
Scholz KH, Maier SKG, Maier LS, Lengerfelder B, Jacobshagen
2018;19:319-25.
C, Jung J, et al. Impact of treatment delay on mortality in ST-
14.
Gilboy N, Tanabe T, Travers D, Rosenau AM. Emergency Severity
segment elevation myocardial infarction (STEMI) patients
Index (ESI): A triage tool for emergency department. Rockville,
presenting with and without haemodynamic instability: results
MD: Agency for Healthcare Research and Quality, 2011.
from the German prospective, multicentre FITT-STEMI trial.
15.
Wachiradilok P, Sethasathien A, Sirisamutr T, Chaiyasit S.
Eur Heart J 2018;39:1065-74.
A National Survey of Emergency Departments Triage Systems in
31.
Seelig JM, Becker DP, Miller JD, Greenberg RP, Ward JD,
Tailand. Prehospital and Disaster Medicine. 2017;32(S1):S233-
Choi SC. Traumatic acute subdural hematoma: major mortality
S233.
reduction in comatose patients treated within four hours. N
16.
Physicians RCo. National Early Warning Score (NEWS) 2.
Engl J Med 1981;304:1511-8.
Standardising the assessment of acute-illness severity in the
32.
Organization WH. Management of health facilities: Referral
NHS. Updated report of a working party. RCP London, 2017.
systems. World Health Organization, 2014.
17.
Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new
33.
Gray A, Gill S, Airey M, Williams R. Descriptive epidemiology
method of classifying prognostic comorbidity in longitudinal
of adult critical care transfers from the emergency department.
studies: development and validation. J Chronic Dis 1987;40:373-
Emerg Med J 2003;20:242-6.
83.
Volume 73, No.5: 2021 Siriraj Medical Journal
321
