Apgar Score and Risk of Neonatal Death Among Preterm Infants

The n e w e ng l a n d j o u r na l of m e dic i n e
Original Article
Apgar Score and Risk of Neonatal Death

among Preterm Infants
Sven Cnattingius, M.D., Ph.D., Stefan Johansson, M.D., Ph.D.,
and Neda Razaz, Ph.D.
A BS T R AC T
BACKGROUND
Gestational age is the major determinant of neonatal death (death within the first From the Division of Clinical Epidemiol-
28 days of life) in preterm infants. The joint effect of gestational age and Apgar ogy, Department of Medicine Solna (S.C.,
S.J., N.R.), and the Department of Clini-
score on the risk of neonatal death is unknown. cal Science and Education (S.J.), Karolin-
ska Institutet, Stockholm. Address reprint
METHODS requests to Dr. Cnattingius at the Division
of Clinical Epidemiology, Department of
Using data from the Swedish Medical Birth Register, we identified 113,300 pre- Medicine Solna, T2, Karolinska University
term infants (22 weeks 0 days to 36 weeks 6 days of gestation) born from 1992 Hospital Solna, SE-171 76 Stockholm,
through 2016. In analyses stratified according to gestational age (22 to 24 weeks, Sweden, or at sven.cnattingius@ki.se.
25 to 27 weeks, 28 to 31 weeks, 32 to 34 weeks, and 35 or 36 weeks), we esti- N Engl J Med 2020;383:49-57.
mated adjusted relative risks of neonatal death and absolute rate differences in DOI: 10.1056/NEJMoa1915075
Copyright © 2020 Massachusetts Medical Society.
neonatal mortality (i.e., the excess number of neonatal deaths per 100 births) ac-
cording to the Apgar scores at 5 and 10 minutes and according to the change in
the Apgar score between 5 minutes and 10 minutes. Scores range from 0 to 10,
with higher scores indicating a better physical condition of the newborn.
RESULTS
There were 1986 neonatal deaths (1.8%). The incidence of neonatal death ranged
from 0.2% (at 36 weeks of gestation) to 76.5% (at 22 weeks of gestation). Lower
Apgar scores were associated with higher relative risks of neonatal death and
greater absolute rate differences in neonatal mortality in all gestational-age strata.
For example, among infants born at 28 to 31 weeks, the adjusted absolute rate
differences according to the 5-minute Apgar score, with those who had a score of
9 or 10 serving as the reference group, were 51.7 (95% confidence interval [CI],
38.1 to 65.4) for a score of 0 or 1, 25.5 (95% CI, 18.3 to 32.8) for a score of 2 or
3, 7.1 (95% CI, 5.1 to 9.1) for a score of 4 to 6, and 1.2 (95% CI, 0.5 to 1.9) for a
score of 7 or 8. An increase in the Apgar score between 5 minutes and 10 minutes
was associated with lower neonatal mortality than a stable Apgar score.
CONCLUSIONS
In this study, Apgar scores at 5 and 10 minutes provided prognostic information
about neonatal survival among preterm infants across gestational-age strata.
(Funded by the Swedish Research Council for Health, Working Life, and Welfare
and Karolinska Institutet.)
n engl j med 383;1 nejm.org July 2, 2020 49

The New England Journal of Medicine
Downloaded from nejm.org on May 17, 2021. For personal use only. No other uses without permission.
Copyright © 2020 Massachusetts Medical Society. All rights reserved.
P
reterm birth (≤36 completed gesta- Nevertheless, a current policy statement of
tional weeks) is the leading cause of neo- the American Academy of Pediatrics includes an
natal death, and the risk of neonatal death overall recommendation that “the Apgar score
increases with the degree of preterm birth.1 does not predict individual neonatal mortality or
However, other factors also influence the risk of neurologic outcome and should not be used for
neonatal death. In a systematic review of studies that purpose,”7 with no specific recommendation
involving very preterm infants (≤31 completed regarding the use of the Apgar score in preterm
gestational weeks), the risk of death was lower infants. Because neonatal mortality increases
among infants with a normal birth weight for substantially with decreasing gestational age,5
gestational age, female infants, infants receiving we hypothesized that absolute rate differences
glucocorticoids prenatally, and infants with high- (i.e., the excess number of neonatal deaths per
er 5-minute Apgar scores.2 100 births) related to Apgar score values may
In the early 1950s, Dr. Virginia Apgar devel- increase with decreasing gestational age in pre-
oped a score to assess the physical condition of term infants. We used a nationwide Swedish
the newborn and the need for resuscitation.3 database to investigate associations between the
Initially, the score was measured at 1 minute, risk of neonatal death and Apgar scores at 5 and
but the Apgar score at 5 minutes was later 10 minutes and the change in the Apgar score
shown to be a better predictor of neonatal sur- from 5 to 10 minutes among preterm infants in
vival.4 The Apgar score consists of five compo- analyses stratified according to gestational age.
nents (heart rate, respiratory effort, muscle tone,
reflex irritability, and color), each given a value Me thods
from 0 to 2. Thus, total scores range from 0 to
10, with higher scores indicating a better physi- Study Population
cal condition. A low Apgar score, commonly de- From January 1, 1992, through December 31,
fined as less than 4 or less than 7, is associated 2016, the Swedish Medical Birth Register includ-
with an increased risk of neonatal death among ed information for 2,534,467 live-born singleton
term infants (≥37 weeks).5 We recently reported infants; information on gestational age was avail-
that, among term infants with 5-minute Apgar able for 2,532,254 of these infants (99.9%). We
scores in the normal range (7 to 10), the risk of identified 125,183 preterm infants (22 weeks
neonatal death was higher among those with a 0 days to 36 weeks 6 days of gestation) and ex-
score of 7 or 8 than among those with a score cluded 2817 births with missing or invalid person-
of 9 or 10.6 unique national registration numbers16 of the
The Apgar score was developed primarily to mother or infant and 9066 infants with major
assess term infants during a time when neonatal congenital malformations (Table S1 in the Sup-
mortality was very high among preterm infants. plementary Appendix, available with the full text
The frequency of low Apgar scores increases with of this article at NEJM.org). The study popula-
decreasing gestational age and may reflect bio- tion included 113,300 nonmalformed, live-born
logic immaturity in preterm infants. Extremely preterm infants.
preterm infants (<28 weeks) and very preterm The nationwide Swedish Medical Birth Register
infants (28 to 31 weeks) in particular may seem includes prospectively recorded information from
less vigorous and be assigned a lower score be- prenatal, delivery, and neonatal care, and the va-
cause of an immature breathing drive and lower lidity of most variables has been assessed as be-
muscle tone. The value of the Apgar score to ing high.17 With the use of the national registra-
assess the condition of the preterm infant has tion numbers of mothers and infants, the Birth
therefore been questioned.7-9 Previous studies Register was linked to the nationwide Swedish
involving preterm infants, which used broad Cause of Death Register18 and the Patient Register,
categories of Apgar score values (0 to 3, 4 to 6, including information on all inpatient care and all
and 7 to 10, or 0 to 3 and 4 to 10), showed in- outpatient hospital care from 2000.19,20 Informa-
creased relative risks of neonatal death with tion about diseases and deaths is coded accord-
decreasing 5-minute Apgar score.5,10 A low Apgar ing to the International Classification of Diseases, 9th
score (0 to 3) has also been associated with a Revision and 10th Revision. We also retrieved infor-
substantially increased risk of death among ex- mation from the yearly updated Education Regis-
tremely preterm infants.11-15 ter21 and from the Total Population Register.22
50 n engl j med 383;1 nejm.org July 2, 2020

All women in Sweden are offered an ultraso- risk of neonatal death and maternal and infant
nographic investigation to estimate gestational characteristics, Apgar scores at 5 and 10 minutes,
age no later than in the early second trimester.23 and gestational age. Results were expressed as
The following hierarchy of estimating gestational crude and adjusted relative risks and absolute
age was used: an ultrasonographic investigation rate differences with 95% confidence intervals.
(82.6% of the mothers), the last menstrual peri- The absolute rate difference represents the ex-
od (10.0%), and a postnatal clinical assessment cess number of neonatal deaths per 100 births
by the midwife or the pediatrician (7.4%). among infants receiving a 5-minute or 10-minute
Apgar score of less than 9, as compared with
Exposures those receiving a score of 9 or 10.
Preterm birth is commonly categorized as ex- Covariates were categorized as described in
tremely preterm (22 to 27 weeks), very preterm Table S2. In multivariable analyses, confounders
(28 to 31 weeks), and moderately preterm (32 to were selected on the basis of their hypothesized
36 weeks). We categorized gestational age as 22 relationship with Apgar score and neonatal mor-
to 24 weeks, 25 to 27 weeks, 28 to 31 weeks, 32 to tality or on the basis of statistical significance
34 weeks, and 35 or 36 weeks. We split the ex- (P<0.10). Adjustments were made for maternal
tremely preterm group into 22 to 24 weeks and age, parity, smoking status, country of birth,
25 to 27 weeks, because infants born at 22 to 24 body-mass index, the presence or absence of
weeks have the highest neonatal mortality. We hypertensive diseases (i.e., essential hypertension
also split the moderately preterm group, since or preeclampsia), mode of delivery, birth weight
practically all infants born at 32 to 34 weeks are for gestational age, hospital level (regional or
admitted to neonatal intensive care units in general hospital), year of birth (continuous), and
Sweden, whereas infants born at 35 or 36 weeks gestational age (in days within each stratum). We
usually stay in the maternity ward. conducted supplementary analysis using multi-
In Sweden, Apgar scores are routinely mea- ple imputation of missing values. We repeated the
sured at 1, 5, and 10 minutes after birth. In pre- Poisson regression analysis with missing values
term births, Apgar scores are most commonly imputed through multiple imputation using
measured by the attending neonatologist or pedia- chained equation.26 Results from 10 multiple-
trician. Information on the Apgar score at 5 and imputation cycles were combined with the use of
10 minutes was available for 98.1% and 92.1% of the MIANALYZE procedure in SAS software. We
all preterm births, respectively. performed supplemental analysis of neonatal
We also obtained information about maternal, mortality, with stratification according to calen-
pregnancy, and birth characteristics (Table S2). dar year of birth. Because of statistical-power
Information about birth weight for gestational concerns, we categorized gestational age into
age was based on the ultrasonography-based three groups and Apgar score into four groups,
Swedish reference curve for normal fetal growth.24 and we provided only crude estimates. We also
analyzed associations between Apgar score val-
Outcome ues and early neonatal mortality. No adjustments
Neonatal death was defined as death within the were made for multiple comparisons. Because
first 27 completed days of life, as recorded in the the 95% confidence intervals around the relative
Cause of Death Register. In supplementary analy- risks and absolute rate differences are not ad-
ses, we also analyzed early neonatal death (death justed for multiple comparisons, these estimates
within the first 6 completed days of life). may not be reproducible.
Statistical Analysis
R e sult s
To investigate dose–response relationships be-
tween Apgar scores and the risk of neonatal Overall Neonatal Mortality
death, the highest Apgar score group (9 or 10, Among 113,300 nonmalformed preterm infants,
or 7 to 10) was, for consistency, used as the refer- there were 1986 neonatal deaths (1.8%). Neona-
ent in all analyses stratified according to gesta- tal mortality increased with decreasing gesta-
tional-age categories. Multivariable log-linear re- tional age, from 0.2% at 36 weeks to 76.5% at 22
gression models with robust variance estimates25 weeks (Table 1). Neonatal mortality also in-
were used to examine associations between the creased with decreasing Apgar score at 5 and 10

Maternal and Infant Characteristics

Table 1. Gestational Age, Apgar Score at 5 Minutes, Neonatal Mortality,
and Adjusted Relative Risks of Neonatal Death.* The prevalence of low Apgar scores (0 to 3 or
4 to 6) at 5 minutes appeared higher among
Infants Neonatal Deaths Adjusted Relative infants of mothers with placental abruption or
Variable (N = 113,300) (N = 1986) Risk (95% CI)†
hypertensive diseases, infants delivered by ce-
no. of deaths sarean section, and infants with a low birth
no. (%) (% of infants) weight for gestational age (<10th percentile)
Gestational age (Table S4). Associations between maternal,
22 wk 226 (0.2) 173 (76.5) 641.9 (492.7–836.4) pregnancy, and birth characteristics and rela-
23 wk 482 (0.4) 228 (47.3) 335.4 (256.5–438.6) tive risks of neonatal death are shown in Table
S5. Infants born from 1992 through 1997 had a
24 wk 704 (0.6) 226 (32.1) 199.6 (151.6–262.7)
higher relative risk of neonatal death than in-
25 wk 976 (0.9) 212 (21.7) 115.5 (86.6–153.9)
fants born later. Low Apgar score values (0 or
26 wk 1,139 (1.0) 159 (14.0) 61.9 (45.4–84.5) 1, 2 or 3, and 4 to 6) at 5 and 10 minutes were
27 wk 1,324 (1.2) 159 (12.0) 55.6 (41.0–75.4) most common among the most immature in-
28 wk 1,616 (1.4) 98 (6.1) 23.5 (16.6–33.3) fants (gestational age, 22 to 24 weeks), and the
29 wk 2,120 (1.9) 87 (4.1) 15.6 (10.9–22.4) percentage of infants with low scores gradually
30 wk 2,628 (2.3) 81 (3.1) 10.9 (7.6–15.8)
decreased with increasing gestational age (Ta-
ble S6).
31 wk 3,488 (3.1) 94 (2.7) 9.7 (6.8–13.8)
32 wk 5,364 (4.7) 64 (1.2) 5.7 (4.0–8.2) Relative Risks of Neonatal Death
33 wk 8,045 (7.1) 81 (1.0) 4.4 (3.1–6.1) In analyses stratified according to gestational age,
34 wk 13,973 (12.3) 90 (0.6) 2.8 (2.0–3.9) relative risks of neonatal death increased with
35 wk 23,894 (21.1) 126 (0.5) 2.5 (1.8–3.3) decreasing Apgar score at 5 minutes (Table 2),
36 wk 47,321 (41.8) 108 (0.2) Reference but the increase appeared markedly higher
5-Min Apgar score
among infants with an older gestational age.
Among infants born at 22 to 24 weeks, those
0 118 (0.1) 82 (69.5) 548.1 (416.1–721.9)
with an Apgar score of 0 or 1 had almost 5 times
1 401 (0.4) 283 (70.6) 422.4 (331.9–537.5) the relative risk of neonatal death as those with
2 413 (0.4) 176 (42.6) 261.7 (200.8–341.2) an Apgar score of 9 or 10. Among infants born
3 547 (0.5) 149 (27.2) 165.5 (124.5–220.0) at 35 or 36 weeks, the corresponding relative
4 750 (0.7) 135 (18.0) 102.7 (76.3–138.3) risk was more than 300 times higher (Table 2).
5 1,152 (1.0) 147 (12.8) 76.2 (57.2–101.5) Similar associations were observed between the
Apgar score at 10 minutes and relative risks of
6 2,540 (2.2) 208 (8.2) 46.6 (35.3–61.5)
neonatal death (Table S7).
7 4,064 (3.6) 165 (4.1) 23.9 (17.9–31.8)
8 7,914 (7.0) 138 (1.7) 10.3 (7.7–13.9) Absolute Rate Differences in Neonatal
9 20,551 (18.1) 146 (0.7) 4.1 (3.1–5.5) Mortality
10 72,694 (64.2) 125 (0.2) Reference The absolute rate difference in neonatal mortal-
Missing data 2,156 (1.9) 232 (10.8) ity from the rate among infants with a 5-minute
Apgar score of 9 or 10 increased substantially
* Percentages may not total 100 because of rounding. with decreasing Apgar scores in all gestational-
† Values were adjusted for maternal age, parity, smoking status, country of birth,
body-mass index, the presence or absence of hypertensive diseases (i.e., essen-
age strata (Fig. 1A). The adjusted rate differences
tial hypertension or preeclampsia), mode of delivery, birth weight for gestational in neonatal mortality according to Apgar score
age, hospital level (regional or general hospital), and year of birth (continuous). were highest among infants born extremely pre-
term (22 to 24 weeks or 25 to 27 weeks) and
minutes, even within the “normal” range of gradually declined with increasing gestational
Apgar score values (7 to 10) (Table 1 and Table age. These associations appeared more pro-
S3). Adjusted relative risks of neonatal death nounced for the Apgar score at 10 minutes
increased substantially with decreasing gesta- (Fig. 1B). Similar results were obtained with
tional age and Apgar scores at 5 and 10 minutes. multiple imputation of missing data (Table S8).

Table 2. Gestational Age, Apgar Score at 5 Minutes, and Relative Risks of Neonatal Death.*
Gestational Age and 5-Minute Crude Relative Risk Adjusted Relative Risk
Apgar Score Infants Neonatal Deaths (95% CI) (95% CI)†
no. of deaths
no./total no. (%) (% of infants)
22–24 Wk
Score of 0 or 1 218/1412 (15.4) 203 (93.1) 6.7 (4.6–9.8) 4.7 (2.7–8.3)
Score of 2 or 3 196/1412 (13.9) 110 (56.1) 4.1 (2.7–6.0) 3.3 (1.9–5.9)
Score of 4 to 6 409/1412 (29.0) 141 (34.5) 2.5 (1.7–3.7) 2.1 (1.2–3.8)
Score of 7 or 8 293/1412 (20.8) 59 (20.1) 1.5 (0.9–2.3) 1.3 (0.7–2.4)
Score of 9 or 10 166/1412 (11.8) 23 (13.9) Reference Reference
Missing data 130/1412 (9.2) 91 (70.0)
25–27 Wk
Score of 0 or 1 72/3439 (2.1) 57 (79.2) 12.8 (9.9–16.5) 8.7 (6.2–12.2)
Score of 2 or 3 178/3439 (5.2) 72 (40.4) 6.5 (4.9–8.7) 5.1 (3.6–7.3)
Score of 4 to 6 764/3439 (22.2) 156 (20.4) 3.3 (2.5–4.3) 2.7 (2.0–3.7)
Score of 7 or 8 1107/3439 (32.2) 118 (10.7) 1.7 (1.3–2.3) 1.4 (1.0–2.0)
Missing data 205/3439 (6.0) 58 (28.3)
28–31 Wk
Score of 0 or 1 76/9852 (0.8) 46 (60.5) 50.6 (37.5–68.3) 50.7 (33.2–77.6)
Score of 2 or 3 190/9852 (1.9) 63 (33.2) 27.7 (20.3–37.9) 26.8 (17.6–40.7)
Score of 4 to 6 988/9852 (10.0) 99 (10.0) 8.4 (6.2–11.3) 8.3 (5.5–12.4)
Score of 7 or 8 2404/9852 (24.4) 50 (2.1) 1.7 (1.2–2.5) 2.2 (1.4–3.5)
Missing data 421/9852 (4.3) 33 (7.8)
32–34 Wk
Score of 0 or 1 69/27,382 (0.3) 22 (31.9) 147.8 (94.4–231.2) 109.4 (63.1–189.5)
Score of 2 or 3 193/27,382 (0.7) 46 (23.8) 110.5 (75.5–161.7) 80.9 (51.7–126.4)
Score of 4 to 6 1095/27,382 (4.0) 59 (5.4) 25.0 (17.1–36.5) 22.3 (14.2–35.2)
Score of 7 or 8 3546/27,382 (13.0) 33 (0.9) 4.3 (2.8–6.7) 3.5 (2.0–6.1)
Score of 9 or 10 21,784/27,382 (79.6) 47 (0.2) Reference Reference
Missing data 695/27,382 (2.5) 28 (4.0)
35 or 36 Wk
Score of 0 or 1 84/71,215 (0.1) 37 (44.0) 450.3 (319.8–634.1) 313.8 (192.5–511.4)
Score of 2 or 3 203/71,215 (0.3) 34 (16.7) 171.2 (115.5–253.8) 114.7 (67.6–194.6)
Score of 4 to 6 1186/71,215 (1.7) 35 (3.0) 30.2 (20.0–45.4) 20.0 (11.9–33.5)
Score of 7 or 8 4628/71,215 (6.5) 43 (0.9) 9.5 (6.5–14.0) 8.6 (5.5–13.5)
Score of 9 or 10 64,409/71,215 (90.4) 63 (0.1) Reference Reference
Missing data 705/71,215 (1.0) 22 (3.1)
* Percentages may not total 100 because of rounding.

† Values were adjusted for maternal age, parity, smoking status, country of birth, body-mass index, the presence or absence of hypertensive
diseases (i.e., essential hypertension or preeclampsia), mode of delivery, birth weight for gestational age, hospital level (regional or general
hospital), year of birth (continuous), and gestational age (in days within each stratum).

A 5-Minute Apgar Score Figure 1. Gestational Age, Apgar Scores at 5 Minutes

Subgroup Adjusted Rate Difference (95% CI) and 10 Minutes, and Adjusted Rate Differences
in Neonatal Mortality.
22–24 Wk of gestation
Score of 0 or 1 64.9 (54.3 to 75.6) Absolute rate differences represent the excess number
Score of 2 or 3 35.9 (23.9 to 48.0) of neonatal deaths per 100 births. Values were adjusted
Score of 4 to 6 16.6 (6.4 to 26.8) for maternal age, parity, smoking status, country of
Score of 7 or 8 2.4 (−7.9 to 12.8) birth, body-mass index, the presence or absence of hyper-
Score of 9 or 10 Reference
25–27 Wk of gestation tensive diseases (i.e., essential hypertension or preeclamp-
Score of 0 or 1 65.5 (54.2 to 76.9) sia), mode of delivery, birth weight for gestational age,
Score of 2 or 3 31.3 (22.4 to 40.1) hospital level (regional or general hospital), year of birth
Score of 4 to 6 12.0 (8.3 to 15.8) (continuous), and gestational age (in days within each
Score of 7 or 8 3.0 (0.2 to 5.8)
Score of 9 or 10 Reference stratum). CI denotes confidence interval.
Score of 0 or 1 51.7 (38.1 to 65.4)
Score of 2 or 3 25.5 (18.3 to 32.8) Supplementary Analyses
Score of 4 to 6 7.1 (5.1 to 9.1)
Score of 7 or 8 1.2 (0.5 to 1.9) In analyses stratified according to year of birth
Score of 9 or 10 Reference and gestational age, relative risks of neonatal
Score of 0 or 1 27.7 (15.3 to 40.1) death according to Apgar score were consistent
Score of 2 or 3 23.9 (17.2 to 30.7) with those in the entire cohort, including the
Score of 4 to 6 4.9 (3.5 to 6.4)
Score of 7 or 8 0.6 (0.2 to 0.9) results for the most recent time interval (2010
Score of 9 or 10 Reference through 2016). Absolute rate differences in neo-
Score of 0 or 1 44.5 (32.6 to 56.3) natal mortality tended to be lower in the later
Score of 2 or 3 16.2 (10.5 to 21.9) birth cohorts than in corresponding analyses of
Score of 4 to 6 2.0 (1.1 to 2.9)
Score of 7 or 8 0.8 (0.5 to 1.0) the entire cohort (Table S9).
Score of 9 or 10 Reference Owing to power concerns, analyses of chang-
−20 0 20 40 60 80 100 es in Apgar score values from 5 to 10 minutes
were stratified into three gestational age groups
B 10-Minute Apgar Score
and four Apgar score groups. Increased Apgar
Subgroup Adjusted Rate Difference (95% CI)
scores from 5 to 10 minutes were associated
Score of 0 or 1 70.2 (62.0 to 78.5)
with lower relative risks of neonatal death than
Score of 2 or 3 59.3 (47.1 to 71.5) scores that remained stable, and the absolute rate
Score of 4 to 6 19.3 (9.3 to 29.2)
Score of 7 or 8 10.1 (2.5 to 17.6)
differences in neonatal mortality were smaller
Score of 9 or 10 Reference (Table 3).
Score of 0 or 1 80.0 (70.4 to 89.6)
Early neonatal death (within the first 6 com-
Score of 2 or 3 54.3 (40.5 to 68.1) pleted days of life) occurred in 1566 infants
Score of 4 to 6 24.3 (17.3 to 31.2)
Score of 7 or 8 6.6 (3.5 to 9.6)
(1.4%) (Table S10). The associations between
Score of 9 or 10 Reference Apgar scores at 5 and 10 minutes and adjusted
Score of 0 or 1 73.8 (59.3 to 88.3)
relative risks of early neonatal death and abso-
Score of 2 or 3 59.9 (43.6 to 76.2) lute rate differences in early neonatal mortality
Score of 4 to 6 16.7 (11.8 to 21.6)
Score of 7 or 8 3.4 (2.1 to 4.7)
appeared similar to or stronger than those in
Score of 9 or 10 Reference corresponding analyses of neonatal mortality.
Score of 0 or 1 48.2 (33.0 to 63.5)
Score of 2 or 3 54.0 (39.4 to 68.6)
Score of 4 to 6 18.8 (13.6 to 24.0) Discussion
Score of 7 or 8 1.5 (0.8 to 2.1)
Score of 9 or 10 Reference In this nationwide study, we found that Apgar
Score of 0 or 1 45.8 (32.3 to 59.3)
scores provided substantial information about the
Score of 2 or 3 36.9 (23.7 to 50.2) risk of neonatal death among preterm infants
Score of 4 to 6 9.6 (5.6 to 13.5)
Score of 7 or 8 1.2 (0.7 to 1.8)
across gestational-age strata. The relative risk of
Score of 9 or 10 Reference neonatal death consistently increased according
−20 0 20 40 60 80 100 to decreasing Apgar score in all gestational-age
strata, but the increases in relative risk were

markedly higher among infants with an older of Apgar score and gestational age than previous
gestational age. The absolute rate differences studies, we were able to show dose–response
(i.e., the excess number of neonatal deaths per associations between Apgar score and risks of
100 births) also increased with decreasing Apgar neonatal death across gestational age. In addi-
score in all gestational-age strata. The greater tion to relative risks, we also estimated absolute
increase in absolute rate differences in neonatal rate differences in neonatal mortality according
mortality according to decreasing Apgar score in to Apgar score, which is of substantial prognostic
extremely preterm infants is explained by their importance. We also included population-based
higher baseline neonatal mortality. An increase information about the Apgar score at 10 minutes,
in the Apgar score from 5 to 10 minutes was, which enabled us to estimate how risks of neo-
even within the normal range (i.e., an increase natal death changed with Apgar score values from
from 7 or 8 to 9 or 10), associated with lower neo- 5 to 10 minutes. The risk of bias was reduced by
natal mortality than scores that remained stable. the virtually complete follow-up and by adjust-
The consistent finding that relative risks of ment for confounders in multivariate analyses.
neonatal death gradually increased with decreas- Limitations of our study include the substan-
ing Apgar score values also raises questions about tial interobserver28 and international29 variability
the commonly used broad categorizations of in assessing the Apgar score. We found a higher
Apgar score values. Taken together, our findings percentage of infants with an Apgar score of
support the hypothesis that Apgar scoring is 9 or 10 than might be expected, especially for
useful in evaluating the health of preterm in- extremely preterm infants. Still, the percentage of
fants across gestational age. infants with available data who had Apgar score
Our findings challenge the view that the Ap- values of 7 to 10 at 5 minutes in our cohort of
gar score may be of limited use in preterm in- preterm infants (94.7%) is in agreement with the
fants.7-9 We acknowledge that the Apgar score percentages in two American studies (95.2% of
reflects the condition of the infant only shortly infants born at 26 to 36 weeks and 96.0% of in-
after birth and may reflect a variety of underly- fants born at 24 to 36 weeks).10,12 Among pre-
ing factors and preceding events. Apgar scoring term infants born at 22 to 31 weeks who were
as such may be more challenging in preterm admitted to 200 neonatal intensive care units in
infants who, owing to immaturity, may appear Europe, 81.5% of those with information on the
less vigorous than term infants.7,27 In agreement Apgar score had a 5-minute score of 7 to 1030;
with other studies, we also found that the per- the corresponding percentage in our study was
centage of infants with low Apgar scores in- 77.8%. We found no published percentages of a
creased with decreasing gestational age.8,9 Sev- 5-minute Apgar score of 9 or 10 in preterm infants.
eral components of the Apgar score — including The percentage in the current study (83.9%) is in
respiratory effort, irritability, muscle tone, and agreement with the corresponding percentage
color — may be influenced by gestational age.7,9 among singleton preterm infants (22 to 36 weeks)
Consequently, it has been argued that a low Ap- born in Norway from 1999 through 2015 (79.4%)
gar score in preterm infants may reflect biologic (Skjaerven R: personal communication).
immaturity rather than fetal depression in an We lacked information about interventions
otherwise healthy infant.7,8 However, to the ex- during the initial stabilization, which may have
tent that physiological response patterns that are influenced Apgar score values, changes in Apgar
reflected by Apgar scores in preterm infants may scores, and neonatal survival.7 The finding that
be a proxy for vulnerability related to immatu- an increase in the Apgar score from 5 minutes
rity, Apgar scores may provide useful prognostic to 10 minutes was associated with lower neona-
information for the survival of preterm infants.10 tal mortality may reflect a better health status at
The findings of this large, nationwide study of 5 minutes in infants whose scores subsequently
neonatal mortality among preterm infants ex- increased than in those whose scores did not in-
tend the findings of previous studies in several crease; still, this may be valuable prognostic in-
ways.5,10-15 By using more granular categorizations formation. Infants with missing information on

Table 3. Gestational Age, Apgar Scores at 5 and 10 Minutes, Adjusted Relative Risks of Neonatal Death, and Adjusted
Rate Differences in Neonatal Mortality.
Gestational Age and Apgar Adjusted Relative Risk Adjusted Rate

Score Infants Neonatal Deaths (95% CI)* Difference (95% CI)*†
no. of deaths
no. (% of infants)
22 to 27 Wk
5-Min score of 0 to 3
10-Min score of 0 to 3 390 347 (89.0) Reference Reference
10-Min score of 4 to 6 155 52 (33.5) 0.5 (0.3 to 0.7) −45.5 (−54.4 to −36.5)
10-Min score of 7 or 8 75 16 (21.3) 0.2 (0.1 to 0.4) −62.8 (−73.6 to −52.0)
10-Min score of 9 or 10 12 0
5-Min score of 7 or 8
10-Min score of 7 or 8 644 107 (16.6) Reference Reference
28 to 31 Wk
32 to 36 Wk
10-Min score of 7 or 8 109 1 (0.9) — —
* Values were adjusted for maternal age, parity, smoking status, country of birth, body-mass index, hypertensive diseases
(i.e., essential hypertension or preeclampsia), mode of delivery, birth weight for gestational age, hospital level (regional
or general hospital), year of birth (continuous), and gestational age (in days within each stratum).
† Absolute rate differences represent the excess number of neonatal deaths per 100 births.

the Apgar score at 5 or 10 minutes (1.9% and umbilical artery. Although both the relatively
7.9% of the population, respectively) had relatively homogeneous population and the equal access
high neonatal mortality. Thus, missing informa- to care in Sweden may favor internal validity, the
tion on the Apgar score in these circumstances results may not be generalizable to more hetero-
may have reflected caregivers who were engaged geneous populations.
in resuscitating very ill neonates. Nevertheless, the In this nationwide study, 5-minute and 10-
overall observed associations were unchanged minute Apgar scores and changes in the score
with multiple imputation of missing data. We between 5 minutes and 10 minutes were associ-
lacked information about values of Apgar score ated with neonatal mortality among preterm
components. Heart rate is likely to be the most infants across gestational ages.
important component for predicting neonatal Supported by a grant (2017-00134) from the Swedish Re-
search Council for Health, Working Life, and Welfare and by
death31 and of a low Apgar score in preterm in- an unrestricted grant ( 2368/10-221) from Karolinska Institutet,
fants.9 We adjusted for a large number of possi- Stockholm (Distinguished Professor Award to Dr. Cnattingius).
ble confounders, but residual confounding by Disclosure forms provided by the authors are available with
the full text of this article at NEJM.org.
unmeasured factors remains possible. We lacked We thank Rolv Skjaerven, Ph.D., of the University of Bergen
information about blood gas analyses from the in Norway for providing additional information to this study.
References
1. Harrison MS, Goldenberg RL. Global moving beyond gestational age. N Engl J (https://www.scb.se/en/f inding-statistics/
burden of prematurity. Semin Fetal Neo- Med 2008;358:1672-81. statistics-by-subject-a rea/education-a nd
natal Med 2016;21:74-9. 12. Lee HC, Subeh M, Gould JB. Low Ap- -research/education-of-t he-population/
2. Medlock S, Ravelli AC, Tamminga P, gar score and mortality in extremely pre- educational-attainment-of-the-population/).
Mol BW, Abu-Hanna A. Prediction of mor- term neonates born in the United States. 22. Ludvigsson JF, Almqvist C, Bonamy
tality in very premature infants: a system- Acta Paediatr 2010;99:1785-9. AK, et al. Registers of the Swedish total
atic review of prediction models. PLoS 13. Weinberger B, Anwar M, Hegyi T, Hiatt population and their use in medical re-
One 2011;6(9):e23441. M, Koons A, Paneth N. Antecedents and search. Eur J Epidemiol 2016;31:125-36.
3. Apgar V. A proposal for a new method neonatal consequences of low Apgar scores 23. Høgberg U, Larsson N. Early dating by
of evaluation of the newborn infant. Curr in preterm newborns: a population study. ultrasound and perinatal outcome: a co-
Res Anesth Analg 1953;32:260-7. Arch Pediatr Adolesc Med 2000;154:294- hort study. Acta Obstet Gynecol Scand
4. Drage JS, Kennedy C, Schwarz BK. 300. 1997;76:907-12.
The Apgar score as an index of neonatal 14. Bartman T, Bapat R, Martin EM, 24. Marsál K, Persson PH, Larsen T, Lilja
mortality: a report from the Collaborative Shepherd EG, Nelin LD, Reber KM. Apgar H, Selbing A, Sultan B. Intrauterine growth
Study of Cerebral Palsy. Obstet Gynecol score at 5 minutes is associated with mor- curves based on ultrasonically estimated
1964;24:222-30. tality in extremely preterm infants even foetal weights. Acta Paediatr 1996;85:843-8.
5. Iliodromiti S, Mackay DF, Smith GCS, after transfer to an all referral NICU. Am 25. Zou G. A modified Poisson regression
Pell JP, Nelson SM. Apgar score and the J Perinatol 2015;32:1268-72. approach to prospective studies with bi-
risk of cause-specific infant mortality: 15. Wyckoff MH, Salhab WA, Heyne RJ, nary data. Am J Epidemiol 2004;159:702-6.
a population-based cohort study. Lancet et al. Outcome of extremely low birth 26. White IR, Royston P, Wood AM. Mul-
2014;384:1749-55. weight infants who received delivery room tiple imputation using chained equations:
6. Razaz N, Cnattingius S, Joseph KS. cardiopulmonary resuscitation. J Pediatr issues and guidance for practice. Stat Med
Association between Apgar scores of 7 to 2012;160(2):239-244.e2. 2011;30:377-99.
9 and neonatal mortality and morbidity: 16. Ludvigsson JF, Otterblad-Olausson P, 27. Papile LA. The Apgar score in the 21st
population based cohort study of term in- Pettersson BU, Ekbom A. The Swedish per- century. N Engl J Med 2001;344:519-20.
fants in Sweden. BMJ 2019;365:l1656. sonal identity number: possibilities and 28. O’Donnell CPF, Kamlin COF, Davis PG,
7. American Academy of Pediatrics Com- pitfalls in healthcare and medical re- Carlin JB, Morley CJ. Interobserver variabil-
mittee on Fetus and Newborn, American search. Eur J Epidemiol 2009;24:659-67. ity of the 5-minute Apgar score. J Pediatr
College of Obstetricians and Gynecolo- 17. Källén B, Källén K. The Swedish Med- 2006;149:486-9.
gists Committee on Obstetric Practice. ical Birth Register: a summary of content 29. Siddiqui A, Cuttini M, Wood R, et al.
The Apgar score. Pediatrics 2015;136:819- and quality. Stockholm:Swedish National Can the Apgar score be used for interna-
22. Board of Health and Welfare, 2003. tional comparisons of newborn health?
8. Catlin EA, Carpenter MW, Brann BS IV, 18. Brooke HL, Talbäck M, Hörnblad J, Paediatr Perinat Epidemiol 2017;31:338-45.
et al. The Apgar score revisited: influence of et al. The Swedish Cause of Death Register. 30. Haumont D, Modi N, Saugstad OD,
gestational age. J Pediatr 1986;109:865-8. Eur J Epidemiol 2017;32:765-73. et al. Evaluating preterm care across Europe
9. Hegyi T, Carbone T, Anwar M, et al. 19. Quality and content in the Swedish using the eNewborn European Network
The Apgar score and its components in the Patient Register. Stockholm:Swedish Na- database. Pediatr Res 2020 January 23
preterm infant. Pediatrics 1998;101:77-81. tional Board of Health and Welfare, 2014. (Epub ahead of print).
10. Casey BM, McIntire DD, Leveno KJ. 20. Ludvigsson JF, Andersson E, Ekbom A, 31. Cnattingius S, Norman M, Granath F,
The continuing value of the Apgar score et al. External review and validation of the Petersson G, Stephansson O, Frisell T.
for the assessment of newborn infants. Swedish national inpatient register. BMC Apgar score components at 5 minutes: risks
N Engl J Med 2001;344:467-71. Public Health 2011;11:450. and prediction of neonatal mortality.
11. Tyson JE, Parikh NA, Langer J, et al. 21. Educational attainment of the popula- Paediatr Perinat Epidemiol 2017;31:328-37.
Intensive care for extreme prematurity — tion. Stockholm:Statistics Sweden, 2020 Copyright © 2020 Massachusetts Medical Society.


Apgar Score and Risk of Neonatal Death Among Preterm Infants

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Apgar Score and Risk of Neonatal Death Among Preterm Infants

Uploaded by

Copyright:

Available Formats

The n e w e ng l a n d j o u r na l of m e dic i n e