Table 1 Characteristics of included studies.

Chubb et al.²⁴ (UK)

To review training, to assess Black, Asian, and minority ethnic babies

67 midwives and students on placement

NA

NR

Study design: Descriptive

Not able to assess long term impact of training. Not all staff trained, but high turnout at 79%

The training was well evaluated but more training and research is required to improve safety of Black, Asian, and ethnic minority families

Inclusion criteria: 67/85 midwives and students at Yeovil District NHS Foundation Trust who completed training

Data Collection: Anonymous pre and post training surveys

Analysis: descriptive statistics and thematic analysis of qualitative data

Gillette et al.²⁵ (USA)

To evaluate the association between maternal race, 5-min Apgar score, and mortality

6,809,653 infants

Normal Apgar score (7–10) 98.8%; Intermediate Apgar score (4–6) 0.9% and low Apgar score (0–3) 0.3%

Non-Hispanic White—52.8%

Study design: Population based cohort study

Missing data within the data set for some variables as the study used routinely collected data. Maternal race only was used as there was a high proportion of missing data for paternal race.

Low Apgar scores are poorer at discriminating risk of mortality in Black and non-Hispanic non- Asian neonates than White infants

Hispanic—23.7%

Inclusion criteria: Singleton infants, 37 + 0–44 + 6 weeks gestation, without congenital malformations, born to mothers over 15 years of age

Infant gender: Female 49.1%

Non-Hispanic Black—13.8%

Data Collection: From linked live birth and infant death datasets.

Non-Hispanic Asian—6.6%

Analysis: Logistic regression adjusted for GA, sex, maternal BMI, education, age, parity, and smoking status and stratified by race

Non-Hispanic other—3.1%

According to self-reported maternal race

Li et al.⁷ (USA)

To evaluate if 5-min Apgar remains pertinent in practice and to assess the Apgar in predicting infant survival

25,168,052 singletons

24–25 wks 0.18%;

Non-Hispanic White 18,095,334 (66.4%)

Study design: Retrospective linked data

Infants born in very severe condition may have been reported as a stillbirth. Proportion of each Apgar score for each ethnicity NR

The Apgar score predicts neonatal and post-neonatal adverse outcomes in term and preterm infants, and is applicable to twins and in various race/ethnic groups

768,305 twins

26–27 wks 0.25%;

Non-Hispanic Black 4,540,838 (16.6%)

Exclusion criteria: Triplets or higher (70,387); BW <500 g or NK (84,177); GA < 24 wks or >44 wks (490,214), 5-min Apgar 0, >10 or NR (8,637,941).

28–29 wks 0.34%;

Hispanic 3,300,185 (12.1%)

Data Collection: From linked live birth and infant death datasets.

30–31 wks 0.56%;

Other 1,334,801 (4.9%)

Analysis: SAS v9.2. Chi-Square, Kaplan-Meier curve, Cox proportional hazard-model adjusted for maternal education, marital status, time AN care started, and smoking during pregnancy. Non-Hispanic White referent

32–33 wks 1.13%;

According to maternal race

34–36 wks 7.38%;

37–41 wks 86.95%;

41–44 wks 3.22%

Mihoko Doyle, et al.²² (USA)

To assess the predictive validity of 1-min Apgar scores on infant mortality looking at race/ethnic-specific variation.

6,544,004 neonates

White: <1500 g 0.6%; ≥1550 g & GA ≤ 36 wks 6.7%; ≥1550 g & GA > 36 wks 92.7%

5,100,942 White, 1,252,870 Blacks, 190,192 MA

Study design: Retrospective linked data

Apgar score involves some subjectivity. Relatively small number of Mexican Americans included in study

Apgar is a strong predictor of infant survival including within race/ethnic groups

Black: <1500 g 2.0%; ≥1550 g & GA ≤;36 wks 15.5%; ≥1550 g & GA > 36 wks 82.5%

NR how attributed ethnicity/race

Exclusion criteria: Multiple births; BW <500 g; <22 wks GA

MA: <1500 g 0.7%; ≥1550 g & GA ≤ 36 wks 8.8%; ≥1550 g & GA > 36 wks 90.5%

Data Collection: 1989–1991 NCHS Linked Birth/Infant Death files for the USA 1-min Apgar scores. Classification as BW<1500 g, or >1500 g with GA ≤ 36 wks or GA>36 wks

Apgar (0–3): White 1.7%, Black 3.1%, MA 1.8%; Apgar (4–6): White 5.9%, Black 7.3%, MA 5.8%; Apgar (7–10): White 92.4%, Black 89.6%, MA 92.4%

Analysis: Logistic regression models, with race/ethnic models estimated

Serunian & Broman²⁰ (USA)

To investigate relationship of 1-min Apgar scores to 8-month Bayley mental and motor score

350 / 391 children had Apgar and Bayley scores

1-min Apgar:

86 White, 89 Black, 75 mixed (Black & Portuguese)

Study design: Prospective cohort

NR

Children with abnormal development at 8 months had significantly lower 1 min Apgar scores

White 0–3 (3.5%); 4–6 (16.3%); 7–10 (80.2%)

Racial designation based on maternal report

Inclusion criteria: Participants selected from the Providence Child Development Study. Selected according to race/ethnicity

Black 0–3 (9.0%); 4–6 (14.6%); 7–10 (76.4%)

Data Collection: 1 min Apgar score. Bayley mental and motor development form at 8-months, as well as Infant Behavior Profile and presence or absence of physical abnormalities.

Mixed (Black & Portuguese) 0–3 (6.7%); 4–6 (9.3%); 7–10 (84.0%)

Analysis: Chi square tests

Shankaran et al.²³ (USA)

To evaluate neurodevelop-mental sequelae among extremely high-risk infants (low birthweight, preterm and low Apgar)

304 infants survived to hospital discharge of whom 246 were seen at 18–22 months

GA: 23.6 ± 0.7 wks

Initial cohort Black: 182/304 (59.9%)

Study design: Prospective cohort

Not all surviving infants followed-up. Loss to follow-up may result in serious ascertainment bias, although clinical and demographic characteristics of those not evaluated were similar

High-risk infants (low birthweight, preterm, and low Apgar) are at high risk of morbidity and mortality

BW (mean): 639.3 ± 63.6 g

Followed up 146/246 (59.3%)

Inclusion criteria: 304 infants surviving to hospital discharge from 12 NNUs participating in previous study

A center-based study therefore potential for referral bias

5-min Apgar <3: 24.4%

NR how attributed ethnicity/race

Data Collection: Follow up at 18–22 months corrected age. Amiel-Tison neurologic examination, Bayley Mental Developmental Index and Psychomotor Developmental Index; medical and social history including parental education, occupation, household composition, income level

Maternal age: 26.7 ± 6.9 yrs

Analysis: Mantel-Haenszel odds ratios

Infant gender: Male 44.7%

Wolf et al.²¹ (Zimbabwe)

To document neurological condition of African neonates with a low Apgar score

165 babies in Zimbabwe compared to asphyxiated infants in the Netherlands (n = 94) and Grenada, Caribbean (n = 11)

In Zimbabwe: Primiparous 51.5%; Infant gender girl 43.6%; preterm 15.7%; congenital malformations 7%, SGA 12%

Only country of origin given:

Study design: Descriptive studies x3

Control groups not strictly comparable due to different Apgar cut off points in the different countries.

Neonatal morbidity was higher in Zimbabwe than in the comparison groups from the Netherlands and Caribbean

BW: Zimbabwe 2846 ± 702 g; Netherlands 2879 ± 978 g.

Zimbabwe; Netherlands or Grenada - Caribbean.

Inclusion criteria: Babies with an Apgar score of ≤5 at 5 min admitted to NNU between 1 July 1991 and 30 June 1992

Unable to ascertain impact of ethnicity vs different care practices between countries

GA: Zimbabwe 38.5 ± 2.9 wks; Netherlands 38.4 ± 3.6 wks.

Data Collection: Comprehensive neurological examination adapted from Prechtl when sufficiently stable to tolerate the examination, but not before the third day after term delivery and when reached corrected term age if born preterm. The two reference groups defined asphyxia as an Apgar ≤ 6 at 3 min. Examinations performed by the same investigator.

Age mother: Zimbabwe 24 ± 6.2 yrs; Netherlands 25 ± 4.7 yrs.

Analysis: Chi square tests, Pearson’s r and ANOVA.

Dawson et al.²⁶ (Australia)

To explore whether the pinkness of an infant’s tongue provided a useful indication that supplemental oxygen was required when pulse oximetry is unavailable

68 neonates (271 paired assessment)

GA: 38 ± 2 wks

Results given by Caucasian/non-Caucasian, proportions in each group not given.

Study design: Prospective observational study

Pulse oximeters not accurate when arterial O₂ < 70%, so impacted by neonate low O₂ sats in first few mins of life.

Tongue color was a specific but insensitive sign that SpO₂ was <70%. When the tongue was pink, it was likely the infant had SpO₂ > 70% and supplemental oxygen not required

Mean BW: 3214 ± 545 g

Inclusion criteria: Convenience sample of infants delivered by CS between Aug and Nov 2012. Excluded if GA < 28 wks, parents did not speak English or pulse oximeter alarm messages.

Only studied infants born by CS

Female n = 36 (53%)

Data Collection: Simultaneous SpO₂ and visual assessment of tongue color when pulse oximeter applied and at 2,3,4,5,6,7 and 10 min after birth. 38 midwives & 7 pediatric trainees carried out assessments

Type of anesthesia: general (n = 2); epidural/spinal (n = 66)

Analysis: Sensitivity, specificity, PPV, NPV, and positive and negative likelihood ratios of tongue not pink to detect SpO₂ < 70%. AUC calculated.

Goldman et al. (1973)⁹

To determine the relationship of clinical assessment of skin and mucous membrane color of neonates to arterial oxygen saturation

93 neonates (182 instances of assessment)

Birthweight: <1500 n = 20; 1500–2500 n = 52; >2500 g n = 21

Skin pigmentation assessed by 3 point scale - fair/medium/dark. Repeated if the infant was re-studied as pigmentation can darken in neonatal period.

Study design: Prospective cohort

NR

Trunk and ears were the least sensitive areas and hands, nailbeds, and around the mouth the most sensitive areas to detect cyanosis

29 were ill - respiratory distress syndrome (n = 21), pneumonia (n = 1), massive aspiration (n = 2), congenital heart disease (n = 3), seizures after birth anoxia (n = 2). The rest were “well”

Dark skinned infants (as judged by at least 2 of the 3 observers) n = 27

Inclusion criteria: Infants PN age <2 wks. Excluded if axillary temp <36 °C, bruising or ecchymoses to trunk, face or hands, or if Hb <13 or >23 gm/dl.

Data Collection: Simultaneous clinical assessment and determination of arterial pO₂ and pH. Assessments by 2 physicians & a nurse in brightly lit room. Cyanosis judged present or absent in 6 areas: lips, ears, trunk, nailbed, hands, region around the mouth. Blood taken after observation.

Analysis: Chi square with Yates correction.

Vesoulis et al.¹⁶ (USA)

To determine whether oxygen saturation is overestimated using pulse oximetry for Black patients.

294 infants

Mean GA: 25.8 ± 2.1 wks (Black 25.6 ± 1.9 wks vs White 25.9 ± 2.1 wks p = 0.09)

124 Black (42%), 170 White (58%)

Study design: Retrospective cohort study

Syringe examined for air prior to analysis.

Modest but consistent difference in SpO₂ error between Black and White infants, with increased incidence of occult hypoxemia in Black infants

Mean BW: 845 ± 265 grams, (Black 805 ± 260 g vs White 875 ± 268 g p = 0.02.)

Classified according to parental identification on birth certificates.

Inclusion criteria: Preterm infants with GA < 32 wks, BW < 1500 g, and admitted to NNU between 2012 and 2019 with valid vital sign data and at least one arterial blood gas.

SpO₂ sensor is rotated every 12 h, but position not routinely charted. Positions other than right upper extremity are post-ductal with potential mismatch between pre- & post-ductal measurements.

Black Female: 64 (52%)

Data Collection: Pulse oximetry with simultaneous arterial blood gas. During the study period alarm limits were set between 88 and 96% until infant reached 35 wks postmenstrual age, then alarm set to 88%. SpO₂ data 30 s before to 30 s after arterial blood gas averaged

Binary race classification used based on self-reported race, further differentiation of skin tone /melanin content would be beneficial in future studies.

White female: 74 (44%)

Analysis: R statistical package. Univariate comparison using nonparametric methods (Fisher’s Exact or Mann–Whitney U test). Pearson correlation coefficient, linear regression, and non-linear regression

Received AN steroids: Black 70%; White 74% (p = 0.49)