Objective
We aimed to determine whether ethnicity-specific birthweight distributions more accurately identify newborns at risk for short-term neonatal morbidity associated with small for gestational age (SGA) birth than population-based distributions not stratified on ethnicity.
Study Design
We examined 100,463 singleton term infants born to parents in Washington State between Jan. 1, 2006, and Dec. 31, 2008. Using multivariable logistic regression models, we compared the ability of an ethnicity-specific growth distribution and a population-based growth distribution to predict which infants were at increased risk for Apgar score <7 at 5 minutes, admission to the neonatal intensive care unit, ventilation, extended length of stay in hospital, hypothermia, hypoglycemia, and infection.
Results
Newborns considered SGA by ethnicity-specific weight distributions had the highest rates of each of the adverse outcomes assessed—more than double those of infants only considered SGA by the population-based standards. When controlling for mother’s age, parity, body mass index, education, gestational age, mode of delivery, and marital status, newborns considered SGA by ethnicity-specific birthweight distributions were between 2 and 7 times more likely to suffer from the adverse outcomes listed above than infants who were not SGA. In contrast, newborns considered SGA by population-based birthweight distributions alone were at no higher risk of any adverse outcome except hypothermia (adjusted odds ratio, 2.76; 95% confidence interval, 1.68–4.55) and neonatal intensive care unit admission (adjusted odds ratio, 1.40; 95% confidence interval, 1.18–1.67).
Conclusion
Ethnicity-specific birthweight distributions were significantly better at identifying the infants at higher risk of short-term neonatal morbidity, suggesting that their use could save resources and unnecessary parental anxiety.
For Editors’ Commentary, see Contents
See related editorial, page 397
Conventionally, small for gestational age (SGA) birthweight has been defined as below the lowest 10th percentile of the population adjusted for gestational age. SGA can be a consequence of intrauterine growth restriction, defined as reduced growth during fetal life relative to the genetic growth potential of the fetus. However, infants classified as SGA may be constitutionally small but healthy as opposed to growth restricted. A number of factors influence birthweight, including infant sex, gestational age at delivery and maternal variables such as parity, ethnicity, prepregnancy weight and height. Two recent Canadian studies have focused on the role of ethnicity in fetal growth. The first suggests that birthweight distributions need to be modified for newborns of immigrant mothers and the second indicates that use of ethnicity-specific birthweight distributions improves differentiation between infants who are pathologically small, and thus at risk of adverse neonatal outcomes, from those who are constitutionally small but healthy.
In 2007, Janssen et al reported significant differences in weight, length, and head circumference of infants at birth according to whether their parents were of European, Chinese or South Asian descent. They developed new birthweight distributions for singleton term infants born to healthy mothers of European, Chinese, and South Asian descent. In a subsequent article, we demonstrated that these ethnicity-specific birthweight distributions predicted neonatal morbidity more often than the population-based birthweight distributions in British Columbia. This study was based on 2647 infants born in British Columbia. In the current study, we examine infants born in Washington State between 2006 and 2008 to determine whether the ethnicity-specific birthweight distributions developed by Janssen et al improve the identification of fetuses at risk for neonatal morbidity compared with a population-based growth distribution that is not stratified on ethnicity developed by Kierans.
Materials and Methods
This study uses a large retrospective cohort consisting of all births in Washington State between 2006 and 2008 (the most recent data available at the time of our request). The Washington Birth Certificate data contains information collected in hospital and birth centers about the mother, the father, the pregnancy, and the infant. These data were linked to the Comprehensive Hospital Abstract Reporting System (CHARS), an inpatient discharge database that includes hospitalization information from all Washington State hospitals.
To test the Janssen ethnicity-specific growth distribution, we restricted our sample to singleton infants between 37 and 41 weeks’ gestational age born to parents who both identified their ethnicity as white, Chinese, or South Asian. We were restricted to these ethnicity groups as the Janssen ethnicity-specific standards were developed in a hospital in British Columbia, Canada, where 50% of delivering women are white, approximately 40% are Chinese, and approximately 10% are South Asian. We excluded mothers with known exposure to alcohol or tobacco during pregnancy, diabetes (preexisting or gestational), or hypertensive disorders to minimize confounding by these factors.
We classified newborns into 4 categories; SGA according to the ethnicity-specific distributions developed by Janssen et al ( SGA-J ethnicity specific); SGA according to the Kierans population-based charts ( SGA-K population-based); meeting neither criterion ( not SGA ); and, SGA according to the SGA-K designation but not the Janssen distribution ( SGA-K-only population-based). SGA-K-only differs from SGA-K in that the infants who were also identified as SGA by Janssen’s ethnicity-specific standards (n = 2277) were removed from this group. There were no SGA-J infants that were not also considered SGA-K. The Kierans population-based charts are stratified by infant sex and provide birthweight and head circumferences for percentiles. The Janssen ethnicity-specific distribution is stratified by sex and ethnicity and refers to weight, length, and head circumference. We present results examining infants who are considered SGA according to their birthweight (less than the 10th percentile). We also performed a sensitivity analysis using the 5th percentile of the Kierans population-based distribution to classify infants as SGA-K.
Our primary outcomes were Apgar score <7 at 5 minutes, admission to the neonatal intensive care unit (NICU), ventilation using endotracheal intubation postdelivery (yes, no), ventilation for more than 6 hours, and extended length of stay in hospital ( ≥3 days for vaginal delivery and ≥4 days for cesarean delivery). Using International Classification of Diseases, Ninth Revision (ICD-9) diagnostic codes, we identified infants who suffered from hypothermia (ICD-9 778.3), hypoglycemia (ICD-9 775.6), and infection (ICD-9 771). We were unable to report infant death, hypoxic ischemic encephalopathy, and intraventicular hemorrhage as our data agreement with the Washington Department of Health states that we cannot release data with less than 5 individuals per cell size for privacy reasons. We calculated crude and adjusted odds ratios (aORs) and their 95% confidence intervals (CIs) to estimate risk of outcomes. We adjusted for maternal age, parity, education level, body mass index (BMI), gestational age, mode of delivery (cesarean section vs vaginal), labor induction, and marital status. All analyses were performed using Stata version 11 software (StataCorp, College Station, TX). Women with missing data were excluded from the adjusted logistic regression analyses. We obtained our data from the Washington State Department of Health. Institutional ethics approval was not required because these data are deidentified and publicly available on request.
Results
We started with a sample of 198,439 women who delivered in Washington State between 2006 and 2008. We excluded 9105 (4.6%) mothers because they gave birth to multiples; 43,395 (22%) mothers gave birth preterm or postterm; 12,268 (6.2%) had diabetes; 10,611 (5.3%) had hypertension and 22,597 (11.4%) smoked during pregnancy. Of our final sample of 100,463 included infants, 93,689 were not considered SGA by the population-based or the ethnicity-specific distributions (not SGA), 6774 were defined as SGA by the Kierans population-based distributions (SGA-K) among which 2277 were also defined as SGA by the Janssen ethnicity-specific distribution (SGA-J), thus there were 4497 newborns identified as SGA by Kierans distribution alone (SGA-K-only). Janssen’s ethnicity-specific distribution did not identify any SGA infants in addition to those classified as SGA by Kierans distribution. We also compared infants identified as SGA by the 5th percentile of the Kierans distribution to the Janssen’s ethnicity specific distribution, and our results did not meaningfully change, thus we present the 10th percentile here as this is more clinically relevant.
Although maternal groups were statistically significantly different with respect to parity, age, BMI, and mode of delivery, the absolute differences between groups were small. Mothers of SGA groups were similar in terms of marital status, but mothers of not SGA newborns were less likely to be unmarried ( P < .001). Groups were not significantly different with respect to maternal education and induction of labor. On average, SGA-J infants weighed less than SGA-K-only infants (2537.1 g vs 2825.1 g, P < .001) and both groups were on average lighter than the not SGA group ( P < .001) ( Table 1 ). Head circumference was also significantly different between groups ( P < .001). SGA-K-only infants were more likely to be Chinese or South Asian (12.2%) compared with SGA-J infants (4.8%) ( P < .001). More SGA-K-only infants were female compared with SGA-J infants (51.6% vs 38.4% respectively, P < .001).
| Characteristic | Not SGA n = 93,689 | SGA-K-only (population-based) n = 4497 | SGA-J (ethnicity-specific) n = 2277 a | P value |
|---|---|---|---|---|
| MATERNAL | ||||
| Parity | ||||
| 0 | 37,763 (40.3) | 2513 (55.9) | 1312 (57.6) | |
| 1 | 32,240 (34.4) | 1190 (26.5) | 523 (23.0) | |
| ≥2 | 22,523 (24.0) | 724 (16.1) | 411 (19.4) | < .001 |
| Missing | 1163 (1.2) | 70 (1.6) | 31 (1.4) | |
| Mother’s age, y | 28.9 (5.6) | 28.3 (5.7) | 27.9 (6.0) | < .001 |
| BMI | ||||
| <20 | 8759 (9.4) | 666 (14.8) | 304 (13.4) | |
| 20-25 | 40,384 (43.1) | 2102 (46.7) | 1066 (46.8) | |
| 25-30 | 21,315 (23.8) | 874 (19.4) | 452 (9.9) | |
| >30 | 15,831 (16.9) | 525 (11.7) | 301 (13.2) | < .001 |
| Missing | 6400 (6.8) | 330 (7.3) | 154 (6.8) | |
| Mother’s education | ||||
| High School or less | 17,682 (18.9) | 875 (19.5) | 507 (22.3) | |
| Some postsecondary | 31,356 (33.5) | 1397 (31.5) | 718 (31.5) | |
| Postsecondary | 9915 (10.6) | 552 (12.3) | 211 (9.3) | |
| Postgraduate | 2583 (2.8) | 137 (3.0) | 50 (2.2) | |
| Missing | 304 (0.3) | 15 (0.3) | 5 (0.2) | .061 |
| Marriage status | ||||
| Not married | 14,276 (15.2) | 860 (19.1) | 479 (21.0) | < .001 |
| Missing | 30 (0.03) | 0 (0.0) | 0 (0.0) | |
| Mode of delivery | ||||
| Cesarean section | 24,158 (25.8) | 1097 (24.4) | 634 (27.8) | |
| Vaginal delivery | 69,531 (74.2) | 3400 (75.6) | 1643 (72.2) | .008 |
| Labor induction | 20,339 (21.7) | 992 (22.1) | 536 (23.5) | .100 |
| INFANT | ||||
| Size | ||||
| Mean weight, g (SD) | 3577.1 (410.4) | 2825.1 (178.5) | 2537.1 (349.9) | < .001 |
| Mean head, cm (SD) | 34.5 (1.5) | 33.1 (1.4) | 32.6 (1.7) | < .001 |
| Ethnicity | ||||
| White | 90,292 (96.4) | 3949 (87.8) | 2168 (95.2) | |
| Chinese | 1296 (1.4) | 147 (3.3) | 24 (1.1) | |
| South Asian | 2101 (2.2) | 401 (8.9) | 85 (3.7) | < .001 |
| Sex | ||||
| Male | 47,726 (50.9) | 2177 (48.4) | 1402 (61.6) | |
| Female | 45,963 (49.1) | 2320 (51.6) | 875 (38.4) | < .001 |
| Mean gestational age, wks (SD) | 39.2 (1.1) | 39.4 (1.1) | 39.5 (1.3) | < .001 |
a All SGA-J (ethnicity-specific) newborns were also considered SGA by population-based standards.
SGA-J infants had the highest rates of each of the adverse outcomes assessed—more than double those of infants considered SGA-K-only except for ventilation required after delivery ( Table 2 ). Rates of NICU admission and hypoglycemia, occurred more than 3 times as often in SGA-J infants compared with SGA-K-only infants. Rates of adverse outcomes in the SGA-K-only infants were similar to the rates of the not SGA newborns.
| Variable | Not SGA n = 93,689 | SGA-K-only (population-based) n = 4497 | SGA-J (ethnicity-specific) n = 2277 a |
|---|---|---|---|
| 5 min Apgar <7 | 1344 (1.4) | 75 (1.7) | 84 (3.7) |
| Infection | 498 (0.5) | 36 (0.8) | 40 (1.8) |
| Hypoglycemia | 1004 (1.1) | 55 (1.2) | 112 (4.9) |
| Hypothermia | 126 (0.1) | 18 (0.4) | 25 (1.1) |
| NICU admission | 2124 (2.3) | 151 (3.4) | 255 (11.2) |
| Extended length of stay b | 12,249 (13.1) | 627 (13.9) | 597 (26.2) |
| Ventilation required postdelivery | 2678 (2.9) | 143 (3.2) | 127 (5.6) |
| Ventilation needed for >6 h | 291 (0.3) | 27 (0.6) | 45 (2.0) |
a All SGA-J (ethnicity-specific) newborns were also considered SGA by population-based standards
b Extended length of stay refers to stays of ≥3 days for vaginal delivery and ≥4 days for cesarean delivery.
In a multivariable logistic regression analysis, odds of all adverse outcomes, adjusted for maternal age, parity, prepregnancy BMI, education, mode of delivery, gestational age, labor induction, and marital status were significantly elevated for infants who were considered SGA-J ( Table 3 ). The aORs of a 5 minute Apgar score <7, need for ventilation, and extended length of stay were approximately 2 times higher for SGA-J infants than infants who were not SGA; (aOR, 2.18; 95% CI, 1.69–2.79; aOR, 1.88; 95% CI, 1.55–2.27, and aOR, 2.28; 95% CI, 2.06–2.52, respectively). The odds of infection were 3 times higher among SGA-J infants (aOR, 3.04; 95% CI, 2.17–4.25) and nearly 5 times higher for hypoglycemia (aOR, 4.71; 95% CI, 3.81–5.81). The odds of NICU admission, ventilation for more than 6 hours and hypothermia were more than 5 times higher in SGA-J infants than in infants who were not SGA (aOR, 5.05; 95% CI, 4.37–5.83; aOR, 5.83; 95% CI, 4.14–8.20, and aOR, 6.52; 95% CI, 4.10–10.36, respectively).
| Variable | Not SGA n = 93,689 | SGA-K-only (population-based) n = 4497 | SGA-K (population-based) n = 6774 a | SGA-J (ethnicity-specific) n = 2277 b |
|---|---|---|---|---|
| Unadjusted ORs | ||||
| 5 min Apgar <7 | 1.00 (ref) | 1.12 (0.89–1.42) | 1.65 (1.39–1.95) c | 2.68 (2.12–2.33) c |
| Infection | 1.00 (ref) | 1.41 (1.00–1.70) | 2.12 (1.67–2.71) c | 3.36 (2.42–4.65) c |
| Hypoglycemia | 1.00 (ref) | 1.05 (0.80–1.38) | 2.33 (1.98–2.75) c | 4.70 (3.85–5.74) c |
| Hypothermia | 1.00 (ref) | 2.55 (1.56–4.16) c | 4.74 (3.35–6.71) c | 8.45 (5.48–13.03) c |
| NICU admission | 1.00 (ref) | 1.37 (1.15–1.62) c | 2.75 (2.46–3.07) c | 5.44 (4.74–6.24) c |
| Extended length of stay d | 1.00 (ref) | 1.05 (0.96–1.14) | 1.47 (1.37–1.56) c | 2.36 (2.15–2.59) c |
| Ventilation required postdelivery | 1.00 (ref) | 1.09 (0.92–1.30) | 1.41 (1.24–1.60) c | 1.99 (1.66–2.39) c |
| Ventilation needed for more than 6 h | 1.00 (ref) | 1.30 (0.95–1.80) | 3.45 (2.66–4.47) c | 6.40 (4.66–8.79) c |
| Adjusted ORs | ||||
| n = 86,384 | n = 4116 | n = 6224 | n = 2108 | |
| 5 min Apgar <7 | 1.00 (ref) | 1.09 (0.85–1.40) | 1.49 (1.24–1.79) c | 2.18 (1.69–2.79) c |
| Infection | 1.00 (ref) | 1.34 (0.93–1.93) | 2.03 (1.58–2.62) c | 3.04 (2.17–4.25) c |
| Hypoglycemia | 1.00 (ref) | 1.12 (0.84–1.48) | 2.44 (2.05–2.90) c | 4.71 (3.81–5.81) c |
| Hypothermia | 1.00 (ref) | 2.76 (1.68–4.55) c | 4.67 (3.24–6.75) c | 6.52 (4.10–10.36) c |
| NICU admission | 1.00 (ref) | 1.40 (1.18–1.67) c | 2.77 (2.47–3.10) c | 5.05 (4.37–5.83) c |
| Extended length of stay d | 1.00 (ref) | 1.03 (0.94–1.14) | 1.44 (1.35–1.55) c | 2.28 (2.06–2.52) c |
| Ventilation required postdelivery | 1.00 (ref) | 1.04 (0.87–1.25) | 1.35 (1.18–1.55) c | 1.88 (1.55–2.27) c |
| Ventilation needed for >6 h | 1.00 (ref) | 1.35 (0.97–1.83) | 3.51 (2.65–4.65) c | 5.83 (4.14–8.20) c |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
