ORIGINAL ARTICLE
Comparison of Neonatal Transport Scoring Systems and
Transport-Related Mortality Score for Predicting
Neonatal Mortality Risk
Sumer Sutcuoglu, MD, Tugce Celik, MD, Senem Alkan, MD, Ozkan Ilhan, MD, and Esra Arun Ozer, MD
Objectives: To predict the risk of mortality of neonates, birth weight and
Downloaded from https://journals.lww.com/pec-online by BhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3nPNgA+JUrYDufGC2lWcKi+d4Mf6yWHP4Sgdf0/HzZTk= on 09/04/2020
gestational age were previously used. However, these criteria were consid-
ered inadequate; therefore, various scoring systems have been developed in
the recent years. The aim of the study was to evaluate the performance of
predicting mortality by Mortality Index for Neonatal Transportation
(MINT), Score for Neonatal Acute Physiology-Perinatal Extension II
(SNAP-PE-II), and Transport Related Mortality Score (TREMS).
Methods: All infants transferred to the neonatal intensive care unit be-
tween January 1 and December 31, 2011, were included. The scores of
SNAP-PE-II, MINT, and TREMS of the all cases were calculated. TREMS
is our proposed scoring system and it consists of 5 variables (hypoglyce-
mia, hypoxia, hypercarbia, hypotension, and hypothermia). The scoring
systems, SNAP-PE-II, MINT, and TREMS, were compared in terms of
mortality risk.
Results: A total of 306 newborn infants constituted the study population.
The mean gestational age was 33.1 ± 5 weeks and the mean birth weight
was 2031.2 ± 1018 g, and 183 (59%) babies were male. The sensitivity
of MINT score for predicting mortality was higher than SNAP-PE-II and
TREMS. However, specificity was higher in TREMS score. The negative
predictive value was highest in MINT score, whereas TREMS has the
highest positive predictive value.
Conclusions: The TREMS scoring system is a simple scoring system
with a high specificity for predicting mortality. Further studies with larger
sample size including more centers and newborn infants with diverse clin-
ical problems are needed to assess the validity and reliability of the TREMS
scoring system.
Key Words: illness severity, neonatal mortality, neonatal transport, scoring
systems
(Pediatr Emer Care 2015;31: 113–116)
D espite the reduction in the perinatal mortality rate due to tech-
nological and scientific advances in perinatal and neonatal
medicine, neonatal mortality is still an important issue for
Turkey.1 The first few hours of the newborn after birth are ex-
tremely critical in respect to neonatal mortality as well as long-
term neurodevelopmental outcome. Sophisticated neonatal trans-
port has improved the safety of transporting newborn infants,
but may not substitute for the benefits of in utero transport. How-
ever, neonates may need transfer after delivery.2,3 Therefore, trans-
port environment including distances and characteristics of
referral centers, transport team composition, training, mode of
transport, use of standard protocols, and access to a specialist or
a medical control physician is extremely important to transfer
From the Tepecik Teaching and Research Hospital, Neonatal Intensive Care
Unit, Izmir, Turkey.
Disclosure: The authors declare no conflict of interest.
Reprints: Esra Arun Ozer, MD, Tepecik Teaching and Research Hospital,
Neonatal Intensive Care Unit, Gaziler Cad. Yenisehir, Izmir, Turkey
(e‐mail: esra.arun@gmail.com).
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
ISSN: 0749-5161
Pediatric Emergency Care • Volume 31, Number 2, February 2015
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
the neonates in a safe manner, resulting in a reduced morbidity
and mortality.
To predict the risk of mortality of neonates, birth weight and
gestational age were widely used as the essential indices.4 How-
ever, these indices were considered inadequate for a precise
prediction of the degree of derangements from physiological nor-
mal and they cannot be used as a direct measure of illness severity;
therefore, various scoring systems have been developed in the re-
cent years.2,3 Recent scoring systems are used to predict the short-
and long-term prognosis of the patient, to evaluate the perfor-
mance of different units and mortality rates.5 Various scoring sys-
tems used in the neonatal period consist of the parameters such as
gestational age, birth weight, ethnicity, sex, body temperature,
heart rate, blood pressure, blood gases, respiratory status, and di-
uresis.5,6 Seven variables (Apgar score at 1 minute, birth weight,
presence of a congenital anomaly, and infant's age, serum pH, ar-
terial partial pressure of oxygen, and heart rate at the time of
the call) are used to generate the Mortality Index for Neonatal
Transportation (MINT) score. Score for Neonatal Acute Physiol-
ogy (SNAP-II) includes 6 physiologic items (blood pressure, tem-
perature, PaO2/FiO2, serum pH, seizure, urine output); to this
are added points for birth weight, low Apgar score, and small
for gestational age to create a 9-item SNAP-Perinatal Extension-II
(SNAP-PE-II). Transport Related Mortality Score (TREMS) is
our proposed scoring system and it consists of 5 variables as fol-
lows: hypoglycemia, hypoxia, hypercarbia, hypotension, and hypo-
thermia. Maximum points for MINT, SNAP-PE-II, and TREMS
scores are 40, 47, and 5, respectively. In all scoring systems that
were mentioned, the higher the points, the worse the outcome.
The aim of the study was to evaluate the performance of the
3 scoring tools completed on admission to the neonatal intensive
care unit (NICU) in predicting neonatal mortality.
METHODS
Study Participants
This prospective study was approved by the local ethics com-
mittee of the Ministry of Health, Izmir Tepecik Teaching and Re-
search Hospital. All infants transferred to the NICU between
January 1 and December 31, 2011, were included in the study. In-
fants with inborn error of metabolism, and outpatient referrals
without neonatal transportation were excluded from the study.
All infants were transported with ground vehicles by Neona-
tal Emergency Transport Service of Izmir 112 Emergency Call
Center. The Neonatal Emergency Transport Service team consists
of a physician and a paramedic who had certification of newborn
resuscitation and interhospital transportation. The transport vehi-
cles had equipment for basic newborn care such as an incubator
with a mechanical ventilator, medical air and oxygen, as well as
emergency medicines and resuscitation tools. Standard care
guidelines were followed during the transport. The transport
distance varied between 1 and 90 km at maximum and all infants
www.pec-online.com 113
Sutcuoglu et al
were stabilized before the transport at facilities with pediatricians/
neonatologists.
Demographic and Clinical Data
Gestational age, birth weight, sex, birth center, type of deliv-
ery, multiple gestation, gestation via assisted reproductive tech-
niques, blood gases, Apgar scores, blood glucose levels, blood
pressure, congenital anomalies, and intubation before transporta-
tion were recorded. Intracranial hemorrhage, sepsis, and chro-
nic lung disease (CLD) diagnosed during the follow-up were
also recorded.
Calculation of Scores of MINT, SNAP-PE-II,
and TREMS
Scores of SNAP-PE-II, MINT, and TREMS of the all cases
during the study period were calculated on admission. TREMS
is a proposed scoring system of our study and it consists of 5 var-
iables as follows: hypoglycemia, hypoxia, hypercarbia, hypoten-
sion, and hypothermia; the lowest score point being score “0”,
and the highest point being score “5.” Hypoglycemia was defined
as blood sugar below 45 mg/dL, hypoxia as pulse oximetry mea-
surement of oxygen saturation below 85%, hypercarbia as PCO2
value in venous capillary blood gas analysis greater than or equal
to 55 mm Hg, hypotension as gestational and postnatal age-
dependent blood pressure values below 10th percentiles, and hy-
pothermia as body temperature below 36°C. The scoring systems
of SNAP-PE-II, MINT and TREMS were compared with each
other in terms of mortality risk.
Statistical Analyses
Statistical analyses were performed using SPSS 18.0 soft-
ware. Independent samples t test, χ2 test, and Pearson correlation
test were used and a P value less than 0.05 was considered
statistically significant.
To determine the specificity of the scoring systems, “receiver
operating characteristic” (ROC) curve and the associated “area
under the curve” (AUC) values were calculated. The AUC values
are distributed between 0.5 and 1.0. If a parameter has a value of
0.5, it represents that the parameter has no any determinative
value, whereas a value of 1.0 represents highly determinative
value. To assess model fit, Hosmer-Lemeshow test was used. In
addition, the sensitivity and specificity of the cutoff scores
were investigated.
RESULTS
A total of 306 newborn infants, which were transported to
our institute, Izmir Tepecik Teaching and Research Hospital, from
remote centers between January 1 and December 31, 2011, consti-
tuted the study population. The mean gestational age was 33.1 ±
5 weeks and the mean birth weight was 2031.2 ± 1018 g, and
183 (59%) babies were male. Birth weight of 27 (8.8%) cases
was below 750 g, of 34 (11%) cases was 751 to 1000 g, and of
63 (20.5%) cases was 1001 to 1500 g. Ninety-three newborns
were delivered vaginally, multiple gestational rate was 9.5% (29)
and 8 gestations were achieved by assisted reproductive tech-
niques. Congenital anomalies were present in 52 (17%) cases.
The mean duration of hospitalization was 27.9 ± 29.7 days and
overall mortality rate was 18.3% (56).
The mean scores of MINT, SNAP-PE-II, and TREMS were
6.4 ± 6.3, 8.8 ± 12, and 1.3 ± 1.1, respectively. On admission,
183 (59.8%) cases were hypothermic, 33 (10.7%) cases were hy-
potensive, and 48 (15.6%) were hypoglycemic. Hypoxia was
114 www.pec-online.com
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Pediatric Emergency Care • Volume 31, Number 2, February 2015
detected in 92 (30%) patients and hypercarbia was found in 55
(17.9%) newborns.
The comparison of mortality-related risk factors in survivors
and nonsurvivors was shown in Table 1. Gestational age and birth
weight were statistically significant (P < 0.001) in terms of mortal-
ity; whereas sex, type of delivery, multiple gestation, using
assisted reproductive techniques, CLD, and sepsis were not found
statistically significant (P > 0.05). Mean scores of each system
were significantly higher in non-survivors (P < 0.001).
The correlation between the scoring systems and mortality
was shown in Table 2. There was a strong negative correlation
in MINT and SNAP-PE-II scoring systems (r = −0.64); whereas
there was a medium negative correlation in TREMS (r = −0.50).
However, there was a small negative correlation between gesta-
tional age and birth weight, and mortality (r = −0.25 and
−0.24, respectively).
Sensitivity, specificity, and positive and negative predictive
values of scoring systems for predicting mortality are shown in
Table 3. The sensitivity of MINT score was higher than SNAP-
PE-II and TREM. However, specificity for predicting mortality
was higher in TREMS. The negative predictive value was highest
in MINT score, whereas TREMS has the highest positive predic-
tive value along with SNAP-PE-II.
The ROC analysis was used to assess the accuracy of the
scoring systems for predicting mortality. The AUC calculations
from ROC were 0.92 for MINT, whereas 0.84 for SNAP-PE-II
and TREMS (Fig. 1).
DISCUSSION
We proposed a scoring system for the prediction of neonatal
mortality based on variables that reflect critical illness and stability
and when abnormal are unfavorable conditions for neonatal trans-
port. Although neonatal transport is one of the key components of
neonatal-perinatal care in Turkey, transferred newborn infants in
this study often experience loss of stability or clinical deteriora-
tion, regardless of their characteristics, and this was related to a
higher mortality. Therefore, it is critical to optimize care strategies
during all neonatal transports.7
The parameters used in scoring systems for predicting mor-
tality in neonates are diverse such as gestational age, birth weight,
ethnicity, sex, body temperature, heart rate, blood pressure, blood
gas values, mechanical ventilation status, medications, diuresis,
and other physiological data.5,6,8,9 However, to ensure an appro-
priate and reliable prediction, data should be easily obtained, and
the cost amount should be low. Moreover, the system must be ob-
jective, noninvasive and is expected to be repeated.
The timing of these scores may be at the time of referral, time
of transport team's first assessment or on admission to NICU over
the first 12 hours. Zupancic et al10 collected data for calculation of
SNAP-II and SNAPPE-II during the first 12 hours after NICU ad-
mission. Similar studies collected data during the first 12 hours.6
In our study, we collected data on admission and during the
first 12 hours.
In the present study, the mean gestational age was 33.1 ±
5 weeks and the mean birth weight was 2031.2 ± 1018 g.
Broughton et al9 reported that the median gestational age was
36 (24-43) weeks and the median birth weight was 2782 (520-
6140) g out of 2504 infants evaluated with MINT score. Likewise,
Lee et al11 assessed the Transport Risk Index of Physiologic
Stability (TRIPS) score in 1723 infants and found the mean gesta-
tional age 36 ± 5 weeks and the mean birth weight 2607 ± 1010 g.
Similar to our study, both studies included all consequent
patients regardless of gestational age. However, our study group
© 2015 Wolters Kluwer Health, Inc. All rights reserved.
Pediatric Emergency Care • Volume 31, Number 2, February 2015 TREMS for Predicting Neonatal Mortality Risk

TABLE 1. The Comparison of Mortality-Related Risk Factors in Survivors and Nonsurvivors

Survivors (n = 250) Nonsurvivors (n = 56) P

Gestational age, wk* 33.6 30.4 <0.001
Birth weight, g* 2152 1490 <0.001
Sex, male, % 148 (59) 35 (14) 0.64
Type of delivery (normal/sectio) 74/176 19/37 0.52
Multiple gestation, % 24 (10) 5 (2) 0.87
Assisted reproduction techniques, % 8 (3) 0 (0) 0.17
Congenital anomaly, % 30 (12) 22 (9) 0.001
Intraventricular hemorrhage, % 42 (17) 21 (8) 0.001
CLD, % 31 (12) 6 (2) 0.72
Sepsis, % 68 (27) 16 (6) 0.83
*Data are presented as mean.

consisted of cases with smaller gestational age and lower birth
weight.
There was a statistically significant relationship with mortal-
ity and both gestational age and birth weight as expected
(P < 0.001).6 Also, intraventricular hemorrhage (IVH) was
strongly linked with mortality (P < 0.001). Gestational age and
TRIPS score are associated with severe IVH, and TRIPS score
may predict severe IVH as good as gestational age and SNAP-II
score.9 One of the variables of the scoring systems for predicting
mortality is sex. It is reported that the risk of mortality was higher
in male infants in several studies.9 Although it was statistically in-
significant in our study, mortality rate was as high as 62.5% in
male infants, which was 37.5% in females. The type of delivery,
multiple gestation, assisted reproduction techniques, CLD, and
sepsis were not statistically significant risk factors for mortality
(P > 0.05).
In our study, the relationship between mortality and presence
of congenital anomalies was statistically significant (P < 0.001).
In a study consisted of 2504 infants evaluated for MINT score,
the presence of a congenital abnormality was significantly associ-
ated with death.6 One of the limitations of the present study is that
only the presence or absence of a congenital abnormality was re-
corded at the time of the admission to NICU. Therefore, it cannot
be stated whether weighting for the severity of the abnormality
might have generated a more accurate score.
One of the important data in our study is that a significant
portion of the cases were transferred without confirming the basic
neonatal transport rules such as maintaining normothermia during
transport, supplying sufficient amount of oxygen when necessary,
infusing crucial medications for the homeostasis, and others. Of
306 infants, 183 (59.8%) were hypothermic, 92 (30%) were hyp-
oxic, 55 (17.9%) were hypercarbic, 33 (10.7%) were hypotensive,
and 48 (15.6%) were hypoglycemic. The relationship between all
these parameters and survival was statistically significant
TABLE 2. The Correlation Between the Scoring Systems and
Mortality
Characteristic Parameter r P
Mortality MINT score −0.64 <0.001
SNAP-PE-II −0.54 <0.001
TREMS −0.50 <0.001
Birth weight 0.25 <0.001
Gestational age 0.24 <0.001
(P < 0.001). Similarly, survival and TREMS were directly related.
The most commonly encountered problem was low body temper-
ature in this study. Therefore, there is a need to take an action on
this issue, that is, optimizing care strategies during all neonatal
transports and education of transport teams. Arora et al12 reported
that temperature change, either alone or in combination with other
indices, was responsible for change in TRIPS score (deterioration
or improvement) in 79% of very-low-birth-weight infants. Better
temperature regulation during interhospital transport may de-
crease the chances of deterioration in physiologic status of very-
low-birth-weight infants.12
The discrimination of MINT, TREMS, and SNAP-PE-II
scoring systems in terms of prediction of death was evaluated with
ROC curve (Fig. 1). The ROC curve is a common tool that is used
as a diagnostic test in clinical trials.13,14 A frequently derived sta-
tistic from the ROC curve is the AUC. An AUC of 1.0 represents a
perfect biomarker, whereas an AUC of 0.5 (as would be derived
from the line of no discrimination) indicates a result that is no bet-
ter than expected by random chance. An AUC of 0.75 or above is
generally considered a good biomarker, and an AUC of 0.9 or
above would represent an excellent biomarker.15 In a previous
study, the area under the ROC curve for neonatal death for the
TRIPS was 0.76 and those for the MINT score was 0.80.6 How-
ever, in a recent study conducted in Turkey, the AUC for MINT
score was 0.65.5 We suggest this AUC variation may result from
the differences of transport facilities.
The present study aimed to investigate the value for
predicting mortality of TREMS scoring system, which was devel-
oped with basic conditions of neonatal transport. We also found
that TREMS score was as effective as SNAP-PE-II for predicting
mortality as well as it may be a noninvasive and simple scoring
system where neonatal transport conditions are poor and well-
trained neonatal transport team is insufficient. Although the per-
formance of the predictive value of MINT score is better than
the other scoring systems in our study, we suggest that TREMS
TABLE 3. Evaluation of the Accuracy of the Scoring Systems for
Predicting Mortality
MINT SNAP-PE-II TREMS
Sensitivity, % 96 87 82
Specifity, % 54 70 71
Positive predictive value, % 31 39 38
Negative predictive value, % 98 96 94

© 2015 Wolters Kluwer Health, Inc. All rights reserved. www.pec-online.com 115

Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.

Sutcuoglu et al
FIGURE 1. ROC curves for scoring systems.
is an acceptable and easy-to-use scoring system, especially for
busy units.
In conclusion, TREMS scoring system is a simple scoring
system with a high specificity for predicting mortality. Further
studies with larger sample size including more centers and new-
born infants with diverse clinical problems are needed to assess
the validity and reliability of the TREMS scoring system.
REFERENCES
1. Erdem G. Perinatal mortality in Turkey. Paediatr Perinat Epidemiol. 2003;
17:17–21.
116 www.pec-online.com
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.
Pediatric Emergency Care • Volume 31, Number 2, February 2015
2. Fenton AC, Leslie A, Skeoch CH. Optimising neonatal transfer. Arch Dis
Child Fetal Neonatal Ed. 2004;89:F215–F219.
3. Cornette L. Contemporary neonatal transport: problems and solutions.
Arch Dis Child Fetal Neonatal Ed. 2004;89:F212–F214.
4. Bookshelf. 2 Measurement of Fetal and Infant Maturity. Preterm Birth:
Causes, Consequences, and Prevention. http://www.ncbi.nlm.nih.gov/
books/NBK11382/. Accessed November 4, 2013.
5. Yılmaz Ö, Çalkavur Ş, Olukman Ö, et al. Comparison of different scoring
systems for predicting mortality risk of neonates followed in neonatal
intensive care units: TRANSPORT, SNAP-PE II and MINT Scores. İzmir
Dr Behçet Uz Çocuk Hast Dergisi. 2011;1:42–50.
6. Broughton SJ, Berry A, Jacobe S, et al. The mortality index for neonatal
transportation score: a new mortality prediction model for retrieved
neonates. Pediatrics. 2004;114:e424–e428.
7. Goldsmit G, Rabasa C, Rodríguez S, et al. Risk factors associated to
clinical deterioration during the transport of sick newborn infants. Arch
Argent Pediatr. 2012;110:304–309.
8. Ferrera A, Atakent Y. Neonatal stabilization score. A quantitative method
of auditing medical care in transported newborn weighing less than 1,000 g
at birth. Med Care. 1986;24:179–187.
9. Broughton SJ, Berry A, Jacobe S, et al. An illness severity score and
neonatal mortality in retrieved neonates. Eur J Pediatr. 2004;7:385–389.
10. Zupancic JA, Richardson DK, Horbar JD, et al. Revalidation of the Score
for Neonatal Acute Physiology in the Vermont Oxford Network. Pediatrics.
2007;119:e156–e163.
11. Lee SK, Zupancic JA, Pendray M, et al. Transport risk index of physiologic
stability: a practical system for assessing infant transport care. J Pediatr.
2001;139:220–206.
12. Arora P, Bajaj M, Natarajan G, et al. Impact of interhospital transport
on the physiologic status of very low-birth-weight infants. Am J Perinatol.
2014;31:237–244.
13. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a
fundamental evaluation tool in clinical medicine. Clin Chem. 1993;39:
561–577.
14. Pepe MS. The Statistical Evaluation of Medical Tests for Classification and
Prediction. New York, NY: Oxford University Press; 2003.
15. Nyugen TM, Devarajan P. Biomarkers for the early detection of acute
kidney injury. Pediatr Nephrol. 2008;23:2151–2157.
© 2015 Wolters Kluwer Health, Inc. All rights reserved.