BACKGROUND: Few researchers have evaluated neonatal mortality in the combined presence of preterm birth (PTB) and small-for-gestational age (SGA) birth weight. None differentiated between infants with and without anomalies, considered births starting at 23 weeks’ gestation, or defined SGA at a more pathologic cutpoint less than the fifth percentile.
METHODS: We completed a population-based cohort study within the province of Ontario, Canada, from 2002 to 2015. Included were 1 676 110 singleton hospital live births of 23 to 42 weeks’ gestation. Modified Poisson regression compared rates and relative risks of neonatal mortality among those with (1) preterm birth at 23 to 36 weeks’ gestation and concomitant severe small for gestational age (PTB-SGA), (2) PTB at 23 to 36 weeks’ gestation without severe SGA, (3) term birth with severe SGA, and each relative to (4) neither. Relative risks were adjusted for maternal age and stratified by several demographic variables.
RESULTS: Relative to a neonatal mortality rate of 0.6 per 1000 term infants without severe SGA, the rate was 2.8 per 1000 among term births with severe SGA (adjusted relative risk [aRR] 4.6; 95% confidence interval [CI] 4.0–5.4), 22.9 per 1000 for PTB without severe SGA (aRR 38.3; 95% CI 35.4–41.4) and 60.0 per 1000 for PTB-SGA (aRR 96.7; 95% CI 85.4–109.5). Stratification by demographic factors showed a persistence of this pattern of neonatal death. Restricting the sample to births at ≥24 weeks’ gestation, or newborns without a congenital or chromosomal anomaly, also demonstrated the same pattern.
CONCLUSIONS: Methods to detect or prevent PTB or SGA should focus on PTB-SGA, which serves as a useful perinatal surveillance indicator.
- aRR —
- adjusted relative risk
- CI —
- confidence interval
- ICD-10 —
- International Classification of Diseases, 10th Revision
- ICES —
- Institute for Clinical Evaluative Sciences
- IUGR —
- intrauterine growth restriction
- PAF —
- population attributable fraction
- PTB —
- preterm birth
- PTB-SGA —
- preterm birth and small for gestational age
- SGA —
- small for gestational age
- RR —
- relative risk
What’s Known on This Subject:
Infants of preterm birth (PTB) experience higher mortality, as do infants born small for gestational age (SGA). A pregnancy affected by the combination of preterm birth and small for gestational age (PTB-SGA) typically has more underlying placental dysfunction and worse neonatal outcomes.
What This Study Adds:
PTB-SGA confers a high risk of neonatal death beyond that of PTB alone. Methods to detect or prevent PTB or small for gestational age can be optimized by focusing on PTB-SGA given its ease of measurement and its prediction of neonatal death.
Of all infant deaths within the first year of life, ∼75% occur in the neonatal period (ie, from birth to 27 days) largely because of prematurity, congenital anomalies, and birth asphyxia.1 Not only do infants of preterm birth (PTB), <37 weeks’ gestation, experience higher neonatal mortality but so do those who were born small for gestational age (SGA). A pregnancy affected by the combination of preterm birth and small for gestational age (PTB-SGA) typically has more underlying placental dysfunction2,3 with greater adverse sequelae for the newborn infant, including a much higher risk of neonatal death.4,5
There are several limitations of previous published research on the relation between PTB-SGA and neonatal mortality.4–6 One study was restricted to nulliparous women;5 none differentiated between infants with and without a congenital or chromosomal anomaly (the former being more prone to SGA and PTB);7 all considered births from 24 weeks’ gestation onward, whereas survival is certainly possible among infants born at 23 weeks’ gestation;8 none considered maternal ethnicity, a factor that can influence both fetal growth and the risk of neonatal death;9 SGA was typically defined at <10th percentile weight for gestational age, whereas, as in stillbirth, neonatal death is even more likely to occur at a more pathologic cut-point <5th percentile;10–12 and no study has been recently completed in high-income countries, such as Canada, the United States, or the United Kingdom.
In an effort to overcome some of the limitations of previous studies, we completed a population-based retrospective cohort study in Ontario, Canada, where prenatal and obstetrical care is covered under a universal, provincial health insurance plan; foreign-born mothers contribute to 35% of all births; and data are available for all hospital births. We determined the risk of neonatal death <28 days in relation to PTB with severe SGA, PTB or severe SGA alone, or neither. We further stratified and restricted the population sample on important factors, including parity, maternal world region of origin, rural residence, income level, smoking status, type of preterm delivery, any newborn anomaly, and infant sex.3–9
Procedures and Participants
We included all live singletons born within a hospital in Ontario between April 1, 2002, and March 3, 2015. All information was obtained from the Canadian Institute for Health Information’s Discharge Abstract Database, which captures administrative, clinical, and demographic information on hospital discharges, including deaths and interfacility transfers. When possible, we deterministically linked maternal and newborn records on the basis of matching maternal and newborn chart numbers. The sample was restricted to births between 23 and 42 weeks’ gestation. Infant sex, birth weight, and gestational age were available in the Discharge Abstract Database. Of Ontarian women, ∼95% undergo prenatal ultrasonography before 20 weeks’ gestation, enhancing the accuracy of gestational age as determined at birth.13
Information on maternal smoking in pregnancy was available for births between April 1, 2006, and March 31, 2012, through a linkage with the Better Outcomes Registry and Network Ontario Niday database. Statistics Canada census data were used to classify income quintile and rural status. Maternal world region of birth was obtained from the Ontario portion of the federal Immigration, Refugees and Citizenship Canada Permanent Resident database, which has details on country of birth.
All data sets were linked by using unique, encoded identifiers and analyzed at the Institute for Clinical Evaluative Sciences (ICES). Approval was granted by the research ethics board of Sunnybrook Health Sciences Centre.
The main exposure was 1 of 4 states: (1) PTB-SGA with PTB at 23 to 36 weeks’ gestation and concomitant severe SGA less than the fifth percentile for sex and gestational age; (2) PTB at 23 to 36 weeks’ gestation without severe SGA; (3) term birth with severe SGA; or (4) neither (the referent). As a secondary exposure, we further subcategorized PTB to those at 23 to 31 weeks’ gestation and 32 to 36 weeks’ gestation (additional analysis 1). The sex- and gestational age-specific birth weight percentile cutoffs used to define severe SGA for the main model were based on a published reference standard for all live births in Ontario, Canada;9 however, in an additional analysis, customized curves for maternal world region of origin were used, as described below.
The study outcome was neonatal death by 28 days of birth. Neonatal death was determined in a hierarchical manner, starting with a death occurring within 28 days of the index birth hospitalization. Next, if a newborn was discharged from the hospital before 28 days of age, then the ICES Registered Persons Database was used to capture all other neonatal deaths.
We graphically presented the neonatal mortality rate before 28 days, per 1000 live births (95% confidence interval [CI]), at each gestational week, and contrasting infants with and without severe SGA.
For all other analyses, we used modified Poisson regression with a robust error variance to generate relative risks (RRs) (ie, rate ratios) for neonatal death. Generalized estimating equations with an exchangeable correlation structure were used to account for the possibility of >1 delivery per woman. RRs were further adjusted for maternal age (<20, 20–24, 25–29, 30–34, 35–39, and ≥40 years or unknown).
The main model (examining the relation between the main exposure and outcome) was further stratified by parity, maternal world region of origin, maternal rural versus urban residence, residential income quintile (Q1 [lowest], Q3 [middle], or Q5 [highest]), and maternal smoking status in pregnancy (current versus none) (additional analysis 2). Because birth weight, including the cutpoint value for SGA less than the fifth percentile, may differ by ethnicity,9,14 we reran the stratification by maternal world region of origin using customized curves for maternal ethnicity developed in Canada9 (additional analysis 3).
The main model was further restricted to births from 24 to 42 weeks’ gestation (additional analysis 4), births without a recognized congenital or chromosomal anomaly (additional analysis 5), and by spontaneous versus provider-initiated birth without a recognized congenital or chromosomal anomaly (additional analysis 6). For PTBs, provider-initiated birth was based on 2 criteria: (1) the absence of a diagnosis of either preterm spontaneous labor with preterm delivery (International Classification of Diseases, 10th Revision [ICD-10] diagnostic code O601), preterm labor with term delivery (ICD-10 O602), premature rupture of membranes (ICD-10 O42), or delayed delivery after spontaneous or unspecified rupture of membranes (ICD-10 O756); and (2) the presence of a cesarean delivery (Canadian Classification of Health Interventions code 5MD60) and/or induction of labor (Canadian Classification of Health Interventions code 5AC30). For term births, provider-initiated birth was identified by (1) induction of labor, (2) cesarean section after failed induction of labor, or (3) cesarean section without an ICD-10 labor code of O62, O63, O64, O66, or O71.1 (with the exception of cesarean section after failed induction of labor). Because neonatal care has changed over time, we examined the main model in the 2002–2008 and 2009–2015 era (additional analysis 7). Because neonatal mortality may differ by infant sex, we also examined the main model by male and female newborns (additional analysis 8).
Although stratification by or restriction on several of the aforementioned variables was 1 approach to examining the robustness of the effect size in the main model, we also reran the main model and adjusted for variables that might change between pregnancies, namely, maternal age, parity (0, 1, ≥2, or missing), income quintile, and infant sex (additional analysis 9).
For the age-adjusted main model, we calculated the population attributable fraction (PAF) for neonatal mortality among all births from 23 to 42 weeks’ gestation as well as those without a recognized congenital or chromosomal anomaly using formula 4 by Rockhill et al15 (additional analysis 10).
Statistical analyses were performed by using SAS 9.4 (SAS Institute Inc, Cary, NC).
In Ontario, from April 1, 2002, to March 3, 2015, there were 1 731 118 singleton hospital live births. We excluded 55 008 births (3.2%), as outlined in Supplemental Fig 6. Hence, the final cohort was comprised of 1 676 110 births (Table 1).
There was a decline in neonatal mortality with advancing gestational age (Fig 1). Between 23 and 28 weeks’ gestation, the neonatal mortality rate was substantially higher among newborns with severe SGA than those without severe SGA (Fig 1). For example, at 24 weeks’ gestation, the respective rates were 755.1 per 1000 (95% CI 634.7–875.5) and 455.0 per 1000 (95% CI 420.9–489.0), respectively. Between 29 and 32 weeks’ gestation, the rates were no longer significantly different and regained significance at appreciably lower rates from 33 to 41 weeks’ gestation (Supplemental Fig 7 shows a magnified version of Fig 1 starting at 29 weeks’ gestation).
Relative to term infants without severe SGA (0.6 per 1000), the neonatal mortality rate was 2.8 per 1000 among term births with severe SGA (adjusted relative risk [aRR] 4.6; 95% CI 4.0–5.4), 22.9 per 1000 for PTBs without severe SGA (aRR 38.3; 95% CI 35.4–41.4), and 60.0 per 1000 in the presence of PTB-SGA (aRR 96.7; 95% CI 85.4–109.5; Table 2, upper blue). In additional analysis 1, considering PTB at more discrete levels, the risk of neonatal death was highest in the combined presence of PTB at 23 to 31 weeks’ gestation and severe SGA (aRR 471.6; 95% CI 416.1–534.4; Table 2, lower red).
On stratifying the main model by demographic factors, the pattern of neonatal death persisted in relation to the 4 combinations of PTB and severe SGA (see additional analysis 2; Fig 2). However, the absolute rates differed considerably. For example, PTB infants with severe SGA and mothers from the Caribbean or Sub-Saharan Africa had a mortality rate of ∼83 per 1000, whereas those whose mothers were born in South Asia had a rate of ∼31 per 1000 (Fig 2). Notably, infants of South Asian mothers showed the least contrast in the RR of neonatal death in the presence of PTB-SGA (aRR 48.2; 95% CI 29.7–78.4) and that of PTB without severe SGA (aRR 34.5; 95% CI 25.8–46.1) (Fig 2). The contrasting risk of neonatal death among PTB-SGA newborns and those with PTB alone was seen among nonsmokers but not smokers (Fig 2). For additional analysis 3, redefining severe SGA at less than the fifth percentile by using a customized world region–specific curve, this gap became more apparent, with an aRR of 85.6 (95% CI 55.8–131.3) in the presence of PTB-SGA (Fig 3, black circles) and an aRR of 29.0 (95% CI 22.0–38.2) in PTB without severe SGA (Fig 3, red squares). A similar significant reconfiguration of the aRR was seen for infants of immigrant mothers from the Middle East and North Africa (Fig 3).
Restricting the main model to births from 24 to 42 weeks’ gestation (see additional analysis 4) showed an unchanged pattern for neonatal mortality (Fig 4, upper). Limiting the population sample to births without a recognized congenital or chromosomal anomaly also demonstrated the same pattern but with a slight attenuation in the absolute risk (see additional analysis 5; Fig 4, middle). In the presence (or absence) of severe SGA, spontaneous preterm newborns were at higher risk of death than provider-initiated preterm newborns (see additional analysis 6; Fig 4, lower). Neonatal mortality tended to be higher between 2002 and 2008 than between 2009 and 2015 (see additional analysis 7; Fig 5, upper). For PTB-SGA newborns, the respective mortality rates were 65.8 and 54.1 per 1000, although the aRRs were persistently elevated relative to term infants without SGA. Boys with PTB-SGA had a higher mortality rate (62.9 per 1000) than girls with PTB-SGA (56.4 per 1000) (see additional analysis 8; Fig 5, lower). However, when compared with their same-sex counterparts born at term and without SGA, the aRRs followed a similar pattern to that in the main model.
On adding maternal age, parity, income quintile, and infant sex to the main model as covariates, the main findings were largely unchanged (see additional analysis 9; Supplemental Table 3, blue column).
After excluding live births affected by a congenital or chromosomal anomaly, the PAF for neonatal deaths was 8.1% for PTB-SGA and 60.3% for PTB without severe SGA (see additional analysis 10; Supplemental Table 4).
In this population-based study of nearly 1.7 million births, neonatal mortality was shown to be 100 times higher among infants of PTB with severe SGA than those born at term with a birth weight greater than the fifth percentile, representing an excess of 59 neonatal deaths per 1000 live births. This risk pattern was especially prominent before 29 weeks’ gestation and was evident across various demographic groups and newborns without a congenital or chromosomal anomaly. The neonatal mortality rate was highest among PTB-SGA infants whose mothers originated from the Caribbean and Sub-Saharan Africa, those born spontaneously (versus provider initiated), and among male newborns. The mortality rate among PTB-SGA newborns declined from 66 per 1000 between 2002 and 2008 to 54 per 1000 in between 2009 and 2015. At a population level, the PAF for neonatal death was ∼8% for PTB-SGA infants excluding those with a congenital or chromosomal anomaly.
Strengths and Limitations
This study included all live births in the most populated province of Ontario, Canada, comprising a large array of immigrant groups9,16,17 within a universal health care system. Although we did not include home births herein, the neonatal mortality rate for home births is the same as that of our reference group.18,19 Therefore, omitting home births would not have altered our findings. Although most Ontarian women undergo prenatal ultrasonography before 20 weeks’ gestation, enabling an accurate estimation of gestational age,13 a fetus affected by early-onset intrauterine growth restriction (IUGR) might be mislabeled as younger than the true gestational age.
The fifth percentile cutoff used to define severe SGA was chosen a priori. PTB with severe SGA affected just 0.3% of all births and 5.2% of all PTBs. Using a higher cutpoint (eg, the 10th percentile) to define SGA would have increased the prevalence of infants affected by PTB with SGA but at a cost. First, SGA less than the fifth percentile reflects a degree of smallness that is more likely to be pathologic rather than constitutional.11,12 Second, newborns of mothers from South Asia, for example, tend to have lower birth weights and are twice as likely to be classified as SGA <10th percentile on a Canadian curve,10 making the risk of neonatal death appear lower than when using an ethnicity-specific, customized curve.9 Moreover, at a stricter cutpoint of the fifth percentile on a standard Canadian birth weight curve, there was only an incremental increase in the risk of neonatal death for those with SGA-PTB over those with PTB alone for mothers from South Asia, whereas a significant difference was evident for infants of Canadian-born mothers (Fig 2). Only on applying customized world region–specific curves, as recommended by the World Health Organization,14 was there a significant difference in the risk of neonatal death for South Asians (Fig 3). We did not define severe SGA using The International Fetal and Newborn Growth Consortium for the 21st Century standard.20 When The International Fetal and Newborn Growth Consortium for the 21st Century standard was compared with a customized birth weight standard within a New Zealand multiethnic cohort like ours, the former identified fewer infants as SGA and missed some at risk for adverse neonatal outcomes.21 Although we could have defined severe SGA at less than the third percentile, we know of no evidence that this cutpoint is more pathologically reflective of IUGR than the fifth percentile,10 and doing so would have reduced the sample size available in the assessment of PTB-SGA.
In the current study, we considered newborns at 23 weeks’ gestation and onward and an additional analysis starting at 24 weeks’ gestation. Stillborn fetuses were excluded from the current cohort. Because severe SGA is a major risk factor for stillbirth,11,22 our exclusion of these births may have downplayed the true influence of severe SGA, introducing a form of survival bias among those who were live born. Although debated by some experts,23 a future study might consider a perinatal mortality indicator comprising both stillbirths and neonatal deaths, assuming that the gestational age and weight at the occurrence of the stillbirth is accurately captured. Finally, we did not evaluate parental height or weight, maternal nutrition, or the presence of maternal prepregnancy hypertension or diabetes mellitus, each of which may influence fetal growth, nor did we possess reliable data on the causes of death among newborn fatalities.
Possible Explanations and Implications for Clinicians and Policy Makers
Although severe SGA at term accounts for 47 per 1000 births, its PAF for neonatal death was only 2% among newborns without a documented anomaly. Accordingly, a generic approach to SGA <5th or 10th percentile without consideration of timing at birth may conflate the causes of and solutions for reducing neonatal mortality.5 Pregnancies affected by IUGR are more likely to have preterm onset of preeclampsia, often necessitating provider-initiated preterm delivery; in fact, preeclampsia is the leading reason for provider-initiated PTB.24,25 Such pregnancies are known to exhibit more severe placental insufficiency and histologic placental vascular lesions.26 However, there is emerging evidence that some pregnancies resulting in spontaneous PTB are also affected by underlying placental insufficiency.27,28 In our study, neonatal mortality was higher among infants with severe SGA and spontaneous PTB (46.5 per 1000) than among those with severe SGA and provider-initiated PTB (27.7 per 1000).
Placental vascular disease heightens the risk of IUGR, stillbirth, and neonatal death.29 Emerging studies now offer insight about risk factors that distinguish PTB-SGA from PTB or severe SGA alone and their relation to not only mortality but neonatal morbidity as well.29 An histologic evaluation of placental tissue at birth and consideration of whether the PTB is provider-initiated or spontaneous29 should aid the development of more focused screening or therapeutic interventions.
Herein, among PTB-SGA births from 24 weeks’ gestation onward, neonatal mortality was lower (43.8 per 1000) compared with published rates for Asia (106.5 per 1000), Africa (70.2 per 1000), and Latin America (75.5 per 1000).4 For PTB without severe SGA, the rates are somewhat lower, at 16.1, 23.5, 64.2, and 53.0 per 1000 live births, respectively.4 In the absence of an anomaly at birth, we showed that PTB-SGA occurs in 2.8 per 1000 live births with a PAF for neonatal death of 8%. In contrast, PTB without SGA affects 55 per 1000 live births with a PAF of 60%. Clearly, from a global public health perspective, reducing the incidence of PTB remains a priority. However, dovetailed to that strategy should be an effort to reduce the incidence of severe SGA (ie, pathologic IUGR) given its heightened contribution to neonatal death in Canada and elsewhere.
What strategies exist for identifying a pregnant woman at high risk of PTB-SGA, and in whom should surveillance be increased and/or preventive therapy initiated? In early pregnancy, a simple approach has been developed by using clinical risk factors to identify women at high risk for preeclampsia,30 but a comparable approach is lacking for severe SGA or PTB. In the prediction of early-onset IUGR, first-trimester abnormal uterine artery flow velocity waveform has a sensitivity of 39% (95% CI 26%–54%) and a specificity of 93% (95% CI 91%–95%).31 Novel blood markers, such as circulating placental RNA, are now being tested within cohort studies.32 Low-dose aspirin should be offered to a woman identified in early pregnancy as being at high risk for PTB-SGA or PTB alone. There is level 1 evidence that aspirin modestly reduces the risk of spontaneous PTB,33 IUGR, and maternal preeclampsia.34 Observational data for all of Canada suggest a massive reduction in maternal and newborn length of stay, and related costs, with aspirin prophylaxis.25 Later in pregnancy, among women identified at higher risk of or in whom IUGR is suspected, sonographic measurement of fetal growth and Doppler assessment of the umbilicocerebral or cerebroplacental ratio might aid in deciding the likelihood of PTB35 and both the timing and location of delivery.36
A neonate born with severe SGA (less than the fifth percentile) at 32 to 36 weeks’ gestation has a risk of dying that is 4 times higher than a preterm counterpart without severe SGA, whereas for those born at 23 to 31 weeks’ gestation, that risk is doubled. We acknowledge that among PTB infants, no percentile clearly discriminates between an IUGR infant who will or will not experience early morbidity or death.10 However, for a neonatologist, such information would be useful for prognostication in terms of parental counseling as well as decision-making about ongoing efforts at resuscitation.37 It should also be determined if modifications are required in the postnatal treatment of the PTB-SGA infant. Such information would certainly be enhanced by knowing the specific reasons for neonatal death in these infants. Our data also show a higher mortality in PTB-SGA infants who are boys, have Caribbean or African maternal ancestry, and who are born spontaneously. Again, such information might influence the development of protocols for newborn resuscitation or postnatal care.
Compared with PTB alone, PTB-SGA represents an additional 37 neonatal deaths per 1000 live births. New methods to detect or prevent PTB or SGA (especially aimed at reducing neonatal mortality and morbidity) might be optimized by focusing on PTB-SGA. PTB-SGA should be considered as a perinatal surveillance indicator in high- and low-income countries given its ease of measurement and its prediction of neonatal death. Data are needed about long-term child outcomes among PTB-SGA newborns.
This study was completed at ICES, which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care. The opinions, results, and conclusions reported in this article are those of the authors and are independent from the funding sources. No endorsement by ICES or the Ontario Ministry of Health and Long-Term Care is intended or should be inferred.
- Accepted September 1, 2017.
- Address correspondence to Joel G. Ray, MD, MSc, Department of Obstetrics and Gynecology, St Michael’s Hospital, 30 Bond Street, Toronto, Ontario M5B 1W8, Canada. E-mail:
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: Supported by a grant from the Canadian Institutes of Health Research.
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
- Canadian Perinatal Surveillance System
- Katz J,
- Lee AC,
- Kozuki N, et al; CHERG Small-for-Gestational-Age-Preterm Birth Working Group
- Khoury MJ,
- Erickson JD,
- Cordero JF,
- McCarthy BJ
- Urquia ML,
- Berger H,
- Ray JG; Canadian Curves Consortium
- Ray JG,
- Vermeulen MJ,
- Schull MJ,
- Singh G,
- Shah R,
- Redelmeier DA
- Ota E,
- Ganchimeg T,
- Morisaki N, et al; WHO Multi-Country Survey on Maternal and Newborn Health Research Network
- Janssen PA,
- Saxell L,
- Page LA,
- Klein MC,
- Liston RM,
- Lee SK
- Hutton EK,
- Cappelletti A,
- Reitsma AH, et al
- Villar J,
- Cheikh Ismail L,
- Victora CG, et al; International Fetal and Newborn Growth Consortium for the 21st Century (INTERGROWTH-21st)
- Bartsch E,
- Medcalf KE,
- Park AL,
- Ray JG; High Risk of Pre-eclampsia Identification Group
- Copyright © 2017 by the American Academy of Pediatrics