doi: 10.1111/ppe.

12360 1

Apgar Score Components at 5 Minutes:

Risks and Prediction of Neonatal Mortality
Sven Cnattingius,a Mikael Norman,b Fredrik Granath,a Gunnar Petersson,a Olof Stephansson,a,c Thomas Frisella
a
Clinical Epidemiology Unit, Department of Medicine Solna,
b
Division of Pediatrics, Department of Clinical Science, Intervention and Technology, Karolinska Institutet, Stockholm, Sweden
c
Division of Epidemiology, School of Public Health, University of California, Berkeley, CA

Abstract
Background: The Apgar score consists of five components: heart rate, respiratory effort, muscle tone, reflex
irritability, and colour. Although the Apgar score has been used for 60 years, the specific contribution of the
Apgar score components with respect to risks and prediction of neonatal mortality remains unknown. Likewise,
the value of reduced scores (including less than five Apgar score components) has rarely been investigated.
Methods: In a population-based cohort study of 148 765 liveborn singleton infants in Sweden 2008–2013, we
investigated components of Apgar score at 5 min with respect to relative risks and prediction (using ROC curves,
sensitivity, and positive predictive values) of neonatal mortality.
Results: Reduced values (0–1) of heart rate, respiratory effort, and colour were independently associated with
increased relative risks of neonatal mortality. For the full Apgar score, the sensitivity and positive predictive
values of neonatal mortality (cut-off ≤3) were by gestational age: ≤31 weeks: 56.1% and 49.2%; 32–36 weeks: 25.0%
and 18.2%; and ≥37 weeks: 35.2% and 9.3%, respectively. When only heart rate and respiratory effort were
considered (range 0–4; cut-off ≤2), corresponding values were 66.7% and 34.9%; 37.5% and 13.0%; and 46.3% and
7.6%, respectively.
Conclusions: A reduced Apgar score has generally the same predictability of neonatal mortality as the full Apgar
score. The full Apgar score or reduced scores may be better predictors of neonatal mortality in very preterm
infants (≤31 weeks) than in infants with longer gestations.

Keywords: Apgar score, neonatal mortality, prediction, risk.

The Apgar score, developed by the US anaesthesiolo-
gist Virginia Apgar in the early 1950’s,1 is universally
used to assess new-born’s health. Initially, the Apgar
score at 1 min was used to assess the need of immedi-
ate resuscitation. Later, Apgar score at 5 min was
shown to be a better predictor of neonatal survival
than Apgar score at 1 min.2
With the introduction of modern neonatal intensive
care, the value of the Apgar score was questioned,
especially for infants requiring resuscitation at birth.
Low Apgar score in preterm infants may reflect physi-
ologic immaturity rather than poor condition at birth,
and a low Apgar score may also be caused by mater-
nal drugs, infections, neurological diseases, and
Correspondence:
Sven Cnattingius, Clinical Epidemiology Unit, Department of
Medicine Solna, Karolinska Institutet, Karolinska University
Hospital Solna, Stockholm, Sweden.
E-mail: sven.cnattingius@ki.se
congenital anomalies.3,4 Still, large and more recent
studies find that low Apgar score at 5 min is associ-
ated with substantially increased risks of neonatal
and infant mortality, both in preterm and term
infants.5–7 Compared with normal Apgar score at
5 min (7–10), low Apgar scores (0–3 or 4–6) are associ-
ated with substantially higher relative risks of neona-
tal mortality in term than in preterm infants.
However, as neonatal mortality rates increase with
decreasing gestational age, the absolute neonatal mor-
tality rate difference between infants with low and
normal Apgar scores increases with decreasing gesta-
tional age.6
The Apgar score consists of five components: heart
rate, respiratory effort, muscle tone, reflex irritability,
and colour. Each component is given the same weight,
and ranges from 0 to 2; thus, Apgar score ranges from
0 to 10. These components may not be of equal impor-
tance,8,9 and heart rate and respiratory effort may be

© 2017 John Wiley & Sons Ltd

Paediatric and Perinatal Epidemiology, 2017, , –
2 S. Cnattingius et al.
difficult to assess due to medical interventions.
Although the Apgar score has been used for more
than half a century, we are unaware of studies investi-
gating the specific contribution of the Apgar score
components with respect to mortality risks.
In this population-based study, we investigated
each component of the Apgar score at 5 min individu-
ally and its incremental contribution as part of the
Apgar score with respect to risks and prediction of
neonatal mortality in preterm and term infants.
Methods
Study population
In Stockholm and Gotland counties, Sweden (includ-
ing around 2.1 million inhabitants), a standardized
electronic medical record system, covering all antena-
tal, obstetric, and neonatal care, is available. Starting
at the first antenatal visit, data are prospectively
recorded by midwives and physicians. Original infor-
mation from antenatal, obstetric, and neonatal records
was abstracted to construct the population-based
Stockholm-Gotland Obstetric Data Base. Using the
infant’s person-unique national registration number,10
information about neonatal death was retrieved from
the nation-wide Swedish Cause of Death Swedish
Register.
From 2008 to 2013, there were 149 800 live singleton
births recorded in the Stockholm-Gotland Obstetric
Data base. After excluding infants with no informa-
tion on Apgar score at 5 min (n = 761) or gestational
age (n = 274), the study population included 148 765
infants. We wanted to investigate the value of Apgar
score and Apgar score components for overall neona-
tal mortality, and therefore refrained from other
exclusions, like excluding infants with congenital
anomalies or chromosomal aberrations, which are
generally diagnosed later in the neonatal period.
Neonatal death was defined as death within the first
27 completed days of life (0–27 days). This study was
approved by the Research Ethics Committe at
Karolinska Institutet, Stockholm (No. 2009/275-31
and No. 2012/365-32).
Exposures
Apgar score at 5 min and its components heart rate,
respiratory effort, muscle tone, reflex irritability, and
colour were main exposures. Estimation of gestational
age was based on the following hierarchy: date of
embryo transfer (3.9%), ultrasound assessment (gen-
erally before 18 gestational weeks; 94.3%), and the last
menstrual period (1.8%). Infants were stratified into
being born very preterm (≤31 completed weeks),
moderately preterm (32–36 weeks), and at term
(≥37 weeks). Definitions and categorizations of covari-
ates considered are provided in Table S1.
Statistical analysis
Neonatal mortality rates were calculated for Apgar
score and each component of Apgar score, for all
births and stratified by gestational age. To assess the
association between separate Apgar score compo-
nents (using Apgar score as a continuous variable),
polychoric correlations were calculated, which are
interpreted as the correlation in the continuous traits
underlying the ordinal categories of the Apgar score
components.11 The incremental prediction of neonatal
mortality from each component was estimated from
multivariable regression models, where all compo-
nents were modelled in a joint model. To enable calcu-
lation of relative risk (RR) and 95% confidence
interval (CI), we fit Poisson regression models (with
empirical, robust standard errors) based on the
methods of generalized estimating equations.
In supplementary analyses, we extended the multi-
variable model of neonatal mortality to include: (i)
gestational length, birthweight-for-gestational age,
and mode of delivery; and (ii) maternal age, parity,
family situation, maternal smoking, body mass index,
height, mother’s country of birth, infertility problems,
hypertensive diseases, and year of delivery. Due to
the low number of neonatal deaths in strata of covari-
ates and the strong association between gestational
age and neonatal mortality, these models were run
separately. Categorizations and parameterizations of
continuous covariates were decided based on good-
ness of fit in univariate models (Table S1).
To estimate the prediction of neonatal mortality
when using all available values (i.e. from 0 to 10) on
Apgar score and reduced component scores (after
excluding one, two, or three Apgar score compo-
nents), we plotted receiver operating characteristic
(ROC) curves and calculated the area under curve
(AUC) from logistic regression models. We decided a
priori to comply with the original score giving each
component the same weight, and to include heart rate
and respiratory effort in all models. We also
© 2017 John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2017, , –
 Apgar score components and neonatal mortality 3

calculated the sensitivity and positive predictive val-
ues of cut-off values for the full Apgar score and
reduced component scores with respect to neonatal
mortality.
Results
Apgar score and components of Apgar score and
mortality risks
Overall, reduced values of Apgar score components
(0–1) at 5 min ranged from 0.2% (heart rate) to 9.1%
(colour) (Table 1). The prevalence of reduced values
of Apgar score components increased with decreasing
gestational age. In very preterm infants, reduced com-
ponent values ranged from 11.2% (reduced heart rate)
to 57.7% (reduced muscle tone).
Sixty-five very preterm infants (≤31 weeks), 33
moderately preterm infants (32–36 weeks), and 204
term infants (≥37 weeks) had a 5 min Apgar score ≤3
(Table 1). Within these groups, an Apgar score of 0 or
1 was more common in very preterm infants (n = 31;
48%) than in moderately preterm (n = 11; 33%) or
term infants (n = 45; 22%). In infants with absent
heart rate (value = 0), 97% (31 of 32 infants) had
Apgar score ≤3, whereas the corresponding rate for
infants with 0 values in other Apgar score compo-
nents ranged from 68% (respiratory effort) to 43%
(muscle tone) (data not shown in table).
Of 148 765 singleton infants, 135 (0.91 per 1000)
died during the neonatal period (Table 2). In very pre-
term infants, neonatal mortality rates were 492.3 per
1000 in infants with Apgar score 0–3 and 17.9 per 1000
in infants with Apgar score 7–10 (absolute rate differ-
ence in neonatal mortality was 474.4 per 1000). In term
infants, corresponding neonatal mortality rates were
93.1 and 0.17 per 1000, providing a substantially lower
absolute rate difference (92.9 per 1000). However,
compared with infants with Apgar score 7–10, a very
low (0–3) Apgar score was associated with a

Table 1. Prevalence of Apgar Score and its components at 5 min

All ≤31 weeks 32–36 weeks ≥37 weeks

(n = 148 765) (n = 828) (n = 5518) (n = 142 419)

No. (%) No. (%) No. (%) No. (%)

Apgar score at 5 min

0–3 302 (0.2) 65 (7.9) 33 (0.6) 204 (0.1)
4–6 958 (0.6) 147 (17.8) 124 (2.2) 687 (0.5)
7–10 147 505 (99.2) 616 (74.4) 5361 (97.2) 141 528 (99.4)
Heart rate
0 32 (0.02) 11 (1.3) 4 (0.1) 17 (0.01)
1 311 (0.2) 82 (9.9) 49 (0.9) 180 (0.1)
2 148 422 (99.8) 735 (88.8) 5465 (99.0) 142 222 (99.9)
Respiratory effort
0 384 (0.3) 77 (9.3) 45 (0.8) 262 (0.2)
1 2264 (1.5) 232 (28.0) 278 (5.0) 1754 (1.2)
2 146 117 (98.2) 519 (62.7) 5195 (94.1) 140 403 (98.6)
Muscle tone
0 642 (0.4) 94 (11.4) 63 (1.1) 485 (0.3)
1 6307 (4.2) 383 (46.3) 651 (11.8) 5273 (3.7)
2 141 816 (95.3) 351 (42.4) 4804 (87.1) 136 661 (96.0)
Reflex irritability
0 617 (0.4) 94 (11.4) 64 (1.2) 459 (0.3)
1 3101 (2.1) 284 (34.3) 339 (6.1) 2478 (1.7)
2 145 047 (97.5) 450 (54.3) 5115 (92.7) 139 482 (97.9)
Colour
0 240 (0.2) 47 (5.7) 27 (0.5) 166 (0.1)
1 13 230 (8.9) 371 (44.8) 1064 (19.3) 11 795 (8.3)
2 135 295 (90.9) 410 (49.5) 4427 (80.2) 130 458 (91.6)

Percentages may not sum up to 100% because of rounding.

© 2017 John Wiley & Sons Ltd

Paediatric and Perinatal Epidemiology, 2017, , –
 4

Table 2. Apgar score and its components at 5 min and risks of neonatal mortality

All ≤31 weeks 32–36 weeks ≥37 weeks

S. Cnattingius et al.

No.a (Rate)b RR (95% CI)c No.a (Rate)b RR (95% CI)c No.a (Rate)b RR (95% CI)c No.a (Rate)b RR (95% CI)c

Apgar score at 5 min

0–3 57 (188.7) 580.0 (401.8, 837.2) 32 (492.3) 27.6 (14.6, 52.1) 6 (181.8) 75.0 (30.3, 185.3) 19 (93.1) 549.2 (305.7, 986.8)
4–6 30 (31.3) 96.2 (61.3, 151.2) 14 (95.2) 5.3 (2.5, 11.5) 5 (40.3) 16.6 (6.0, 45.9) 11 (16.0) 94.4 (46.4, 192.0)
7–10 48 (0.33) 1.0 (Reference) 11 (17.9) 1.0 (Reference) 13 (2.4) 1.0 (Reference) 24 (0.17) 1.0 (Reference)
Heart rate
0 20 (625.0) 7.0 (4.1, 12.1) 10 (909.1) 3.0 (1.3, 6.7) 2 (500.0) 6.3 (2.0, 19.7) 8 (470.6) 8.1 (3.0, 22.2)
1 38 (122.2) 2.9 (1.8, 4.6) 23 (280.5) 1.9 (0.9, 3.8) 5 (102.0) 2.4 (0.8, 7.3) 10 (55.6) 2.2 (0.9, 5.7)
2 77 (0.52) 1.0 (Reference) 24 (32.7) 1.0 (Reference) 17 (3.1) 1.0 (Reference) 36 (0.25) 1.0 (Reference)
Respiratory effort
0 59 (153.6) 9.0 (3.7, 22.2) 30 (389.6) 3.7 (1.1, 13.1) 8 (177.8) 6.5 (0.7, 56.5) 21 (80.2) 11.1 (3.0, 40.8)
1 36 (15.9) 5.8 (2.9, 11.6) 20 (86.2) 3.1 (1.2, 8.3) 4 (14.4) 1.7 (0.4, 7.0) 12 (6.8) 6.1 (2.5, 14.6)
2 40 (0.27) 1.0 (Reference) 7 (13.5) 1.0 (Reference) 12 (2.3) 1.0 (Reference) 21 (0.15) 1.0 (Reference)
Muscle tone
0 64 (99.7) 4.1 (0.9, 17.9) 34 (361.7) 2.2 (0.5, 9.8) 8 (127.0) 8.6 (0.6, 121.1) 22 (45.4) 0.9 (0.1, 9.4)
1 38 (6.0) 2.5 (0.7, 8.5) 18 (47.0) 1.1 (0.3, 3.4) 8 (12.3) 3.9 (0.9, 17.6) 12 (2.3) 0.7 (0.1, 4.4)
2 33 (0.23) 1.0 (Reference) 5 (14.2) 1.0 (Reference) 8 (1.7) 1.0 (Reference) 20 (0.15) 1.0 (Reference)
Reflex irritability
0 63 (102.1) 3.7 (0.9, 15.3) 32 (340.4) 1.8 (0.4, 7.6) 8 (125.0) 1.2 (0.0, 33.2) 23 (50.1) 12.0 (1.2, 122.5)
1 36 (11.6) 4.5 (1.4, 14.8) 19 (66.9) 2.3 (0.9, 6.3) 5 (14.7) 1.4 (0.2, 8.5) 12 (4.8) 10.0 (1.6, 63.1)
2 36 (0.25) 1.0 (Reference) 6 (13.3) 1.0 (Reference) 11 (2.2) 1.0 (Reference) 19 (0.14) 1.0 (Reference)
Colour
0 45 (187.5) 3.5 (1.4, 8.7) 26 (553.2) 2.4 (0.7, 8.6) 3 (111.1) 0.8 (0.1, 5.5) 16 (96.4) 4.4 (1.1, 17.7)
1 55 (4.2) 2.2 (1.1, 4.5) 23 (62.0) 1.2 (0.5, 3.1) 12 (11.3) 1.6 (0.4, 5.9) 20 (1.7) 2.5 (0.9, 6.9)
2 35 (0.26) 1.0 (Reference) 8 (19.5) 1.0 (Reference) 9 (2.0) 1.0 (Reference) 18 (0.14) 1.0 (Reference)

RR denotes relative risk; CI denotes confidence interval.

a
Number of neonatal deaths.
b
Rate per 1000 births.
c
Relative risks of each component are adjusted for effects of the other components.

© 2017 John Wiley & Sons Ltd

Paediatric and Perinatal Epidemiology, 2017, , –

substantially larger relative risk of neonatal death in
term infants than in very preterm infants (Table 2).
Regardless of gestational age, reduced heart rate
was associated with the highest neonatal mortality
rates. However, compared with a component value of
2, reduced values of all Apgar score components (0 or
1) were associated with substantially increased neona-
tal mortality rates (Table 2). However, the Apgar
score components were highly correlated (poly-
choric correlations ranged from 0.65 (colour – reflex
irritability) to 0.94 (reflex irritability – muscle tone),
Table 3). When Apgar score components were mutu-
ally adjusted for each other, reduced heart rate, respi-
ratory effort, and colour were associated with
increased neonatal mortality risks. Absent heart rate
and absent respiratory effort (values = 0) provided
the highest relative risks (Table 2).
In very preterm infants (57 neonatal deaths) and in
term infants (54 neonatal deaths), absent heart rate
and reduced respiratory effort (value 0–1) were asso-
ciated with increased neonatal mortality risks. In term
infants, reduced reflex irritability was also associated
with increased neonatal mortality risk. In moderately
preterm infants (24 neonatal deaths), only absent heart
rate conferred an increased risk of neonatal mortality
(Table 2).
Apgar score is often assessed in a stressful situation,
when it is difficult to account for other factors. We
therefore adjusted for birth or maternal characteristics
in supplementary analyses only. Overall neonatal
mortality risk patterns remained, although point esti-
mates were changed (Table S2).
Prediction of neonatal mortality using two to five
components
The AUC of the full Apgar score, reflecting the pre-
dictability of neonatal mortality using all values of
Table 3. Polychoric correlations between Apgar score components at 5 min
Heart rate Respiratory effort
Heart rate 1.00 0.87
Respiratory effort 1.00
Muscle tone
Reflex irritability
Colour
© 2017 John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2017, , –
Apgar score components and neonatal mortality 5
Apgar score, was 0.90 (Table 4, ROC curves in Fig-
ure S1). This can be interpreted as, when using infor-
mation on all Apgar score values, a randomly selected
case of neonatal mortality has a lower Apgar score
than a randomly selected non-case 90% of the time.
When we excluded two of the components muscle
tone, reflex irritability, or colour, there were minor,
although statistically significant reductions in AUC.
In analyses stratified by gestational age, the 5 min
Apgar score was, if anything, a stronger predictor of
neonatal mortality in very preterm infants than in
moderately preterm and term infants (Table 4).
Excluding one or two components did not change the
prediction of neonatal mortality in any gestational age
group. When only including heart rate and respira-
tory effort, the prediction was significantly reduced
compared to the full Apgar score in moderately pre-
term and term infants. Still, the AUC was highest in
very preterm infants.
Of all neonatal deaths, 42.2% had an Apgar score ≤3
(sensitivity), and almost one-fifth of all infants with an
Apgar score ≤3 died neonatally (positive predictive
value [PPV] 18.9%; Table 5). For neonatal mortality,
the sensitivity and the positive predictive value of an
Apgar score ≤3 was substantially higher in very pre-
term infants than in infants with longer gestations.
When we excluded 41 infants born at the border of
viability, i.e. at 22–23 completed weeks (24 neonatal
deaths), the sensitivity and positive predictive value
of a low Apgar score in very preterm infants
decreased (sensitivity 42.4%; positive predictive value
31.8%, not shown in table), but were still higher than
corresponding values of neonatal mortality in moder-
ately preterm and term infants (Table 5).
For the reduced component scores (including four
to two Apgar score components), the cut-off values
were chosen to have the number of infants below the
threshold closest to the number of infants below the
All births
Muscle tone Reflex irritability Colour
0.81 0.84 0.78
0.80 0.84 0.67
1.00 0.94 0.67
1.00 0.65
1.00
6 S. Cnattingius et al.

Table 4. Sums of Apgar score components as predictors of neonatal mortality

All births ≤31 weeks 32–36 weeks ≥37 weeks

AUC (95% CI) AUC (95% CI) AUC (95% CI) AUC (95% CI)

Full Apgar score (range 0–10) 0.90 (0.87, 0.94) 0.87 (0.82, 0.93) 0.83 (0.73, 0.93) 0.85 (0.78, 0.91)
Excluding one of the following components (range 0–8)
Muscle tone 0.90 (0.87, 0.94) 0.87 (0.82, 0.92) 0.81 (0.71, 0.91) 0.85 (0.79, 0.91)
Reflex irritability 0.90 (0.87, 0.93) 0.87 (0.82, 0.92) 0.82 (0.72, 0.92) 0.84 (0.77, 0.91)
Colour 0.89 (0.85, 0.92) 0.87 (0.81, 0.92) 0.81 (0.71, 0.91) 0.83 (0.77, 0.90)
Excluding two of the following components (range 0–6)
Muscle tone & Irritability 0.89 (0.85, 0.92)a 0.86 (0.81, 0.92) 0.79 (0.68, 0.90) 0.84 (0.78, 0.91)
Muscle tone & Colour 0.87 (0.83, 0.91)a 0.86 (0.80, 0.92) 0.78 (0.68, 0.88) 0.82 (0.75, 0.88)
Reflex irritability & Colour 0.88 (0.85, 0.92)a 0.87 (0.81, 0.92) 0.79 (0.68, 0.89) 0.82 (0.76, 0.89)
Excluding the three following components (range 0–4)
Muscle tone, Reflex irritability, & Colour 0.85 (0.81, 0.89)a 0.85 (0.80, 0.91) 0.73 (0.62, 0.84)a 0.80 (0.74, 0.87)a

AUC denotes area under the receiver operating characteristic curve.

a
Significantly reduced AUC compared to full Apgar score (P < 0.05).

cut-off of the full Apgar score (≤3). When one or two
Apgar score components were excluded, there were
negligible changes in sensitivity and positive predic-
tive values of neonatal mortality regardless of gesta-
tional age (Table 5). For these reduced component
scores, the sensitivity and positive predictive values
were also consistently higher in very preterm infants
than in infants with longer gestations. When only
heart rate and respiratory effort were considered
(range 0–4), more infants had a score below the cut-off
(≤2). This probably contributed to increased sensitiv-
ity values and reduced positive predictive values: for
very preterm infants 66.7% and 34.9%, respectively;
for moderately preterm infants 37.5% and 13.0%,
respectively; and for term infants 46.3% and 7.6%,
respectively.
Comment
Principal findings
To our knowledge, this is the first study linking the
Apgar score components at 5 min to risks of neonatal
mortality. Reduced values of heart rate and respira-
tory effort were associated with the highest risks, but
reduced colour was also independently associated
with increased mortality risk. A component score,
including three or four Apgar score components, pre-
dicted neonatal mortality as well as the full Apgar
score. Both the full Apgar score and reduced scores
(including two to four Apgar score components) were,
if anything, better predictors of neonatal mortality in
very preterm infants than in infants with longer
gestations.
Comparison with other studies
The value of Apgar score components may differ
whether the focus is neonatal mortality, short-, or
long-term morbidity. For neonatal mortality, heart
rate and respiratory effort are most important. We
found that almost all infants with absent heart rate
(value = 0) had an Apgar score ≤3. Heart rate is also
the component which is most associated with lower
Apgar scores in preterm infants12 and severe neonatal
morbidity in term infants.13 Muscle tone and reflex
irritability are highly correlated,12,14 and may be more
relevant for neurological outcomes like cerebral
palsy.15 Colour is generally considered to be least
important; removing colour from the Apgar score has
reported to improve16 or not influence12 the correla-
tion between the total Apgar score and umbilical
artery pH. Still, we found that reduced colour was
associated with increased neonatal mortality risk after
adjusting for effects of the other components.
Our results challenge some of the views expressed
by the American Academy of Pediatrics Committee
on Fetus and Newborn and American College of
Obstetricians and Gynecologists Committee on
Obstetrical Practice.3 The first recommendation in this
document reads: “The Apgar score does not predict
individual neonatal mortality or neurologic outcome
and should not be used for that purpose”. We found
that every second very preterm infant (≤31 weeks)

© 2017 John Wiley & Sons Ltd

Paediatric and Perinatal Epidemiology, 2017, , –
 © 2017 John Wiley & Sons Ltd
Table 5. Sensitivity and positive predictive values (PPV) of Apgar score and reduced scores at 5 min for neonatal mortality

All births <31 weeks 32–36 weeks ≥37 weeks

Cut-off level No.a Sensitivityb PPVc No.a Sensitivityb PPVc No.a Sensitivityb PPVc No.a Sensitivityb PPVc

Full Apgar score (range 0–10) ≤3 302 42.2 18.9 65 56.1 49.2 33 25.0 18.2 204 35.2 9.3

Paediatric and Perinatal Epidemiology, 2017, , –

Excluding one component (range 0–8)
Muscle tone ≤3 356 48.1 18.3 75 59.6 45.3 41 29.2 17.1 240 44.4 10.0
Reflex irritability ≤3 346 46.7 18.2 74 57.9 44.6 37 29.2 18.9 235 42.6 9.8
Colour ≤2 297 42.2 19.2 62 54.4 50.0 31 25.0 19.4 204 37.0 9.8
Excluding two components (range 0–6)
Muscle tone and Reflex irritability ≤2 228 38.5 22.8 58 50.9 50.0 27 29.2 25.9 143 29.6 11.2
Muscle tone and Colour ≤2 353 47.4 18.1 73 57.9 45.2 38 29.2 18.4 242 44.4 9.9
Reflex irritability and Colour ≤2 334 45.9 18.6 71 56.1 45.1 34 29.2 20.6 229 42.6 10.0
Excluding three components (range 0–4)
Muscle tone, Reflex irritability, and Colour ≤2 508 53.3 14.2 109 66.7 34.9 69 37.5 13.0 330 46.3 7.6

a
No. denotes number of infants below the cut-off value.
b Number of neonatal deaths with Apgar score below the cut-off  100
Sensitivity: .
All neonatal deaths
c Number of neonatal deaths with Apgar score below the cut-off  100
Positive predictive value: .
All infants below the cut-off
Calculation of sensitivity and positive predictive values are based on the following numbers of births and neonatal deaths: all births: n = 148 765; neonatal deaths n = 135; ≤31 weeks:
n = 828; neonatal deaths n = 57; 32–36 weeks: n = 5518; neonatal deaths n = 24; ≥37 weeks: n = 142 419; neonatal deaths n = 54.
Apgar score components and neonatal mortality
7
8 S. Cnattingius et al.
with an Apgar score ≤3 died neonatally, whereas less
than 2% of those with Apgar score ≥7 died. In term
infants (≥37 weeks), corresponding rates were close to
10% and less than 0.2%, respectively. We think that
these and other results in our study will provide the
clinician with valuable prognostic information for
both preterm and term infants.
Our finding that Apgar score at 5 min may be a bet-
ter predictor of neonatal mortality in very preterm
infants than in infants with longer gestations is
intriguing and several aspects have to be kept in
mind. First, we confirmed previous finding that
Apgar score ≤3 is associated with a substantially
higher relative risk of neonatal mortality in term than
in preterm infants.5,6 This clearly indicates that a low
Apgar score is a severe warning sign in term infants.
However, we also found that the absolute neonatal
mortality rate difference between very preterm infants
with Apgar scores 0–3 and 7–10 was substantially lar-
ger than corresponding rate differences in moderately
preterm and term infants. Second, low Apgar score
and reduced values of the Apgar score components
are more common in very preterm infants than in
infants with longer gestations. This may to a large
extent reflect physiologic immaturity in response to
the stress of being born.4,12 However, a large number
of very preterm infants have experienced deviation
from normal physiology prior to or during labour. For
example, the proportions of pregnancies complicated
by fetal growth restriction and/or early-onset
preeclampsia are higher in very preterm births than in
moderately preterm or term births.17,18 In infants with
Apgar score ≤3, we found that very preterm infants
more often had Apgar scores 0 or 1. Thus, low Apgar
score in very preterm infants may, in addition to
physiologic immaturity, also reflect a high proportion
of severely depressed infants. It is therefore conceiv-
able that pre-existing pathophysiology in very pre-
term infants explains that a score only including
information on heart rate and respiratory effort had
almost the same prediction of neonatal mortality as
the full Apgar score. Our findings support that regu-
lar breathing combined with normal heart rate at
5 min can be used as a marker of “good condition at
birth” in very preterm infants.
Strengths and limitations of the study
Study strengths include the population-based design,
including original information from antenatal,
obstetric, and neonatal records. Information on Apgar
score components and gestational age was almost
complete, and follow-up was complete. Apgar score is
more important for early than for late neonatal mor-
tality (deaths during the first week of life and deaths
during the second to fourth weeks of life, respec-
tively), and least important for post-neonatal
mortality (infant death after the fourth week of life).6
As neonatal mortality is a rare event, our possibilities
to study associations between Apgar score compo-
nents and early or late neonatal mortality were lim-
ited, especially in moderately preterm infants (24
neonatal deaths). Analyses of neonatal mortality were
only stratified by gestational age into three groups
(≤31 weeks, 32–36 weeks, and ≥37 weeks), but we
adjusted for gestational age in completed weeks in
supplementary analyses. Because of power concerns,
we also refrained from studying associations between
Apgar score components and causes of neonatal
death.
Compared with the conventional Apgar score,
information about Apgar score combined with infor-
mation about neonatal interventions was recently
reported to be a better predictor of birth asphyxia and
severe asphyxia-related neonatal morbidity.19–21
However, our aim was to study the risks and prog-
nostic value of Apgar score components at 5 min
regardless of preceding interventions. In preterm
infants born at the border of viability (22–23 com-
pleted weeks), the prevailing recommendation during
the study period in our region was to withhold resus-
citation. When these infants were excluded, an Apgar
score ≤3 was still a better predictor of neonatal mortal-
ity in very preterm infants than in infants with longer
gestations. In an emergency situation, more com-
monly occurring in very preterm infants, it may be
difficult to correctly estimate Apgar score compo-
nents, including heart rate and respiratory effort,
which often is mechanically assisted.22,23 The interob-
server variability in Apgar score has generally been
assessed as poor, especially in very preterm
infants.24,25 Still, we found that the full Apgar score or
a reduced score may be better predictors of neonatal
mortality in very preterm infants than in infants with
longer gestations.
Conclusions
We are unaware of any single measure that is ideal
for predicting severe asphyxia-related neonatal or
© 2017 John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2017, , –

neurodevelopmental outcomes.26 This study provides
strong support that a reduced score, including less
than five Apgar score components, is as valuable as
the full Apgar score to predict neonatal mortality. Our
results also confirm previous assumptions that the
heart rate and respiratory effort components at 5 min
are central for neonatal mortality. Still, our study was
focused on neonatal mortality, and for a general clini-
cal assessment of the immediate needs and the prog-
nosis of the new-born infant, we see no need to revise
the well-established Apgar score. A low Apgar score
can be caused by many other factors than asphyxia,3
and we welcome studies of associations between
Apgar score components and reduced scores with
respect to other outcomes, including neonatal
asphyxia-related morbidity and long-term neurologi-
cal outcomes.27,28 Our finding that the prognostic
value of Apgar score and reduced scores with respect
to neonatal mortality may be larger in very preterm
infants than in moderately preterm and term infants
challenges current view, and may be of clinical
importance.
Acknowledgements
Financial support was provided by an unrestricted
grant from Karolinska Institutet (Distinguished Profes-
sor Award to Cnattingius) and by Stockholm County
Council (ALF projects 20130156, 20140105, and
20150118). The authors report no conflicts of interest.
References
1 Apgar V. A proposal for a new method of evaluation of the
newborn infant. Current Researches in Anesthesia & Analgesia
1953; 32:260–267.
2 Drage JS, Kennedy C, Schwarz BK. The Apgar Score as an
index of neonatal mortality. A Report from the Collaborative
Study of Cerebral Palsy.. Obstetetrics and Gynecology
1964;24:222–230.
3 The Apgar Score. Pediatrics 2015; 136:819–822.
4 Catlin EA, Carpenter MW, Brann BSt, Mayfield SR, Shaul PW,
Goldstein M, et al. The Apgar score revisited: influence of
gestational age. The Journal of Pediatrics 1986;109:865–868.
5 Casey BM, McIntire DD, Leveno KJ. The continuing value of
the Apgar score for the assessment of newborn infants. New
England Journal of Medicine 2001; 344:467–471.
6 Iliodromiti S, Mackay DF, Smith GC, Pell JP, Nelson SM.
Apgar score and the risk of cause-specific infant mortality:
a population-based cohort study. Lancet 2014; 384:
1749–1755.
7 Li F, Wu T, Lei X, Zhang H, Mao M, Zhang J. The apgar
score and infant mortality. PLoS One 2013; 8:e69072.
© 2017 John Wiley & Sons Ltd
Paediatric and Perinatal Epidemiology, 2017, , –
Apgar score components and neonatal mortality 9
8 Apgar V, James LS. Further observations on the newborn
scoring system. American Journal of Diseases in Children 1962;
104:419–428.
9 Crawford JS, Davies P, Pearson JF. Significance of the
individual components of the Apgar score. British Journal of
Anaesthesia 1973; 45:148–158.
10 Ludvigsson JF, Otterblad-Olausson P, Pettersson BU,
Ekbom A. The Swedish personal identity number:
possibilities and pitfalls in healthcare and medical research.
European Journal of Epidemiology 2009; 24:659–667.
11 Olsson U. Maximum likelihood estimation of the polychloric
correlation-coefficient. Psychometrika 1979; 44:443–460.
12 Hegyi T, Carbone T, Anwar M, Ostfeld B, Hiatt M, Koons A,
et al. The apgar score and its components in the preterm
infant. Pediatrics 1998; 101:77–81.
13 Portman RJ, Carter BS, Gaylord MS, Murphy MG, Thieme RE,
Merenstein GB. Predicting neonatal morbidity after perinatal
asphyxia: a scoring system. American Journal of Obstetrics and
Gynecology 1990; 162:174–182.
14 Minami K, Furuhashi M, Miyazaki K, Yoshida K, Kuno N,
Ishikawa K. Would omitting the assessment of heart rate be
harmful to initial neonatal care? Journal of Maternal-Fetal &
Neonatal Medicine 2007; 20:477–480.
15 Walstab JE, Bell RJ, Reddihough DS, Brennecke SP, Bessell CK,
Beischer NA. Factors identified during the neonatal period
associated with risk of cerebral palsy. The Australian & New
Zealand Journal of Obstetrics & Gynaecology 2004; 44:342–346.
16 Marx GF, Mahajan S, Miclat MN. Correlation of biochemical
data with Apgar scores at birth and at one minute. British
Journal of Anaesthesia 1977; 49:831–833.
17 Gilbert WM, Danielsen B. Pregnancy outcomes associated
with intrauterine growth restriction. American Journal of
Obstetrics and Gynecology 2003; 188:1596–1599.
18 Pettit F, Mangos G, Davis G, Henry A, Brown MA. Pre-
eclampsia causes adverse maternal outcomes across the
gestational spectrum. Pregnancy Hypertension 2015; 5:198–204.
19 Dalili H, Nili F, Sheikh M, Hardani AK, Shariat M, Nayeri F.
Comparison of the four proposed Apgar scoring systems in
the assessment of birth asphyxia and adverse early
neurologic outcomes. PLoS One 2015; 10:e0122116.
20 Rudiger M, Braun N, Aranda J, Aguar M, Bergert R,
Bystricka A, et al. Neonatal assessment in the delivery room–
Trial to Evaluate a Specified Type of Apgar (TEST-Apgar).
BMC Pediatrics 2015; 15:18.
21 Dalili H, Sheikh M, Hardani AK, Nili F, Shariat M, Nayeri F.
Comparison of the combined versus conventional Apgar
Scores in predicting adverse neonatal outcomes. PLoS One
2016; 11:e0149464.
22 Kamlin CO, O’Donnell CP, Everest NJ, Davis PG, Morley CJ.
Accuracy of clinical assessment of infant heart rate in the
delivery room. Resuscitation 2006; 71:319–321.
23 Hawkes GA, Hawkes CP, Kenosi M, Demeulemeester J,
Livingstone V, Ryan CA, et al. Auscultate, palpate and tap:
time to re-evaluate. Acta Paediatrica 2016; 105:178–182.
24 O’Donnell CP, Kamlin CO, Davis PG, Carlin JB, Morley CJ.
Interobserver variability of the 5-minute Apgar score. Journal
of Pediatrics 2006; 149:486–489.
25 Rudiger M, Kuster H, Herting E, Berger A, Muller C,
Urlesberger B, et al. Variations of Apgar score of very low
10 S. Cnattingius et al.

birth weight infants in different neonatal intensive care units. web-site:

Acta Paediatrica 2009; 98:1433–1436.
26 Papile LA. The Apgar score in the 21st century. New England Table S1. Definitions and categorizations of covari-
Journal of Medicine 2001; 344:519–520. ates.
27 Moster D, Lie RT, Irgens LM, Bjerkedal T, Markestad T. Table S2. Apgar score components at 5 min and
The association of Apgar score with subsequent
adjusted risks of neonatal mortality.
death and cerebral palsy: a population-based
study in term infants. Journal of Pediatrics 2001; 138:
Figure S1. Receiver operating characteristics (ROC)
798–803.
28 Rei M, Ayres-de-Campos D, Bernardes J. Neurological curves using different combinations of Apgar score
damage arising from intrapartum hypoxia/acidosis. Best components as predictors of neonatal mortality.
Practice & Research Clinical Obstetrics & Gynaecology 2016;
30:79–86.

Supporting Information
Additional Supporting Information may be found in
the online version of this article at the publisher’s

© 2017 John Wiley & Sons Ltd

Paediatric and Perinatal Epidemiology, 2017, , –