Preterm birth can be classified into three clinical subtypes: spontaneous preterm labor with intact fetal membranes, preterm premature rupture of membranes (PROM) before onset of labor, and indicated preterm birth.1 The heterogeneity of preterm birth subtypes that exist in clinical management may also exist, albeit inconclusively, in cause and prognosis.2–6 Preterm birth is a major cause of neonatal death in addition to birth defects.7 Understanding the heterogeneity of preterm birth subtypes in relation to neonatal death can help describe the prognosis of preterm birth and provide related information for clinical decision-making. Because gestational age is an influential determinant of neonatal death, heterogeneity of preterm birth subtypes in relation to neonatal death can present in at least two aspects: 1) distribution of gestational age, and 2) gestational-age-specific neonatal death rate. Two previous studies in Latin America and Sweden that have examined gestational-age-specific neonatal death rate by preterm birth subtypes did not show consistent results.6,8 It is unclear whether preterm PROM and indicated preterm birth have different gestational-age-specific neonatal death risk from spontaneous preterm labor in the contemporary United States. If the heterogeneity of preterm birth subtypes in neonatal death is present, it may also indicate different patterns in risk factors and biologic mechanisms (infection and inflammation, placenta vascular lesions, uterine distension, maternal complications, etc). Therefore, the objective of this study was to investigate the heterogeneity of preterm labor, preterm PROM, and indicated preterm birth in overall and gestational-age-specific neonatal death risk.
MATERIALS AND METHODS
The 2001 U.S. linked birth/infant death (birth cohort, not period linked) data sets were obtained from National Center for Health Statistics (NCHS), Centers for Disease Control and Prevention. The data sets include information from birth certificates (1989 version) from all U.S. live births occurring in 2001 and death certificates if the infant died in the first year of life (either in 2001 or in 2002). The National Center for Health Statistics natality and mortality files were coded according to a uniform standard and underwent a rigorous quality control, review, and edit process before being released. We restricted the data sets to births and infant deaths in the 50 U.S. states and the District of Columbia area. Because multiple births have much higher risk of preterm birth and infant death, they were excluded from the analysis. We also excluded infants with gestational age less than 24 weeks or more than 44 weeks and infants without sufficient information to determine preterm birth subtypes. The study was classified as exempt by Creighton University Institutional Review Board.
Gestational age was determined by NCHS mainly based on dates of last menstrual period (LMP) and delivery. In case LMP was not available or LMP-based gestational age was clearly inconsistent with birth weight, clinical estimate of gestational age was used instead by NCHS.9 Preterm birth was defined as a live birth less than 37 completed weeks of gestation (less than 259 days). Mutually exclusive preterm birth subtypes were defined based on information on the U.S. birth certificate.5,10–12 Preterm PROM was a preterm birth with premature rupture of membranes more than 12 hours before the onset of labor.9 Indicated preterm birth was a preterm birth with 1) induction of labor or 2) cesarean delivery but without the presence of labor (tocolysis; cephalopelvic disproportion; precipitated, prolonged, or dysfunctional labor).13 Those preterm births that were not preterm PROM or indicated preterm birth were classified as idiopathic preterm labor.
Neonatal death was defined as neonates who died before 28 days of life, including early (0–6 days) and late (7–27 days) neonatal death. Neonatal death risk was calculated as the number of neonatal deaths divided by the number of live births. Gestational-age-specific neonatal death risk was defined as the number of neonatal deaths in neonates born at a specific gestational age divided by the number of live births at that gestational age. Underlying causes of infant death were coded according to the Tenth Revision of the International Classification of Diseases (ICD-10) in the death certificate and were reported in 130 causal categories. Because the number of a specific cause of death (chorioamnionitis, bacterial sepsis, etc) was small and underpowered for gestational-age-specific neonatal death risk analysis, we grouped underlying causes of neonatal death into four broad groups: 1) pregnancy complications/preterm birth/low birth weight (LBW) (including maternal complications of pregnancy [P01], complications of placenta, cord, and membrane [P02], and disorders of short gestation and LBW [P07]), 2) respiratory distress syndrome (RDS)/respiratory conditions (P22–P28), 3) birth defects (congenital malformations, deformations, and chromosomal abnormalities [Q00–Q99]), and 4) other causes.
We compared the distribution of demographic and socioeconomic characteristics, birth outcomes, and neonatal death risk by preterm birth subtypes, using χ2 tests. Next, we plotted gestational-age-specific neonatal death risk by the three preterm birth subtypes according to reported preterm PROM, induction of labor, cesarean delivery, and pregnancy and labor complications. We then categorized the gestational age of preterm neonates into four groups for interaction and stratified analysis: 24–27 weeks, 28–31 weeks, 32–33 weeks, and 34–36 weeks.8 Because the data were from the 2001 birth cohort and neonatal death can be modeled as a time-to-event outcome, we used Cox proportional hazard model to estimate hazard ratio (HR) and 95% confidence interval (CI). We compared preterm PROM and indicated preterm birth with preterm labor (referent group) for overall and gestational-age-specific neonatal death risk. A priori covariates included maternal age, race, education, marital status, live birth order, and the sex of the infant, categorized as in Table 1. Information regarding smoking during pregnancy was unavailable for California mothers, but additional adjustment for smoking did not change the estimates, so these results will not be presented. In addition, we explored the risk of early and late neonatal death by preterm birth subtypes to investigate whether the heterogeneity in gestational-age-specific neonatal death risk was confined to first a few days of life. We also analyzed cause-specific neonatal death risk by preterm birth subtypes at different gestational ages (with 32–36 weeks combined because of lower neonatal death risk). This was performed mainly to explore whether causes of neonatal death differ by preterm birth subtypes, which may indicate causal heterogeneity of preterm birth. We used SAS 9.1 (SAS Institute Inc. Cary, NC) for statistical analysis and R 2.7 (R Development Core Team) for graphing.
In the 2001 birth cohort, 99% of infant deaths were linked to birth certificates. In the sample of 3,763,306 singleton live births, the proportion of preterm birth was 10.3% (0.8% preterm PROM, 3.7% indicated preterm birth, and 5.8% preterm labor). Table 1 shows the distribution of maternal and infant characteristics by preterm birth subtypes and term birth. Compared with term births, mothers of preterm births were more likely to be African American, unmarried, and smoke during pregnancy. Among preterm birth subtypes, mothers with preterm PROM and indicated preterm birth tended to be older (aged 35 years or older), but were less likely to be Hispanic. Infants of mothers with preterm PROM were born earlier and lighter than infants of mothers with preterm labor, so were infants of mothers with indicated preterm birth but to a lesser extent. The neonatal death risk was 2.7% for preterm PROM, 1.8% for indicated preterm birth, and 1.1% for preterm labor. Compared with idiopathic preterm labor, preterm PROM had an unadjusted HR of 2.53 (95% CI 2.34–2.74) for neonatal death, whereas indicated preterm birth had unadjusted HR of 1.63 (95% CI 1.54–1.72).
Figure 1 shows gestational-age-specific neonatal death risk by preterm birth subtypes. Preterm PROM and indicated preterm birth had neonatal death risk similar to preterm labor for births at 24–27 weeks, but had higher risk after that. In a Cox model with main effects and interactions between preterm birth subtypes and gestational length (24–27, 28–31, 32–33, and 34–36 weeks), three of six interaction terms were significant (P<.05, Table 2), thus the effects of preterm birth subtypes differed by gestational age.
Table 3 shows the neonatal death risk by preterm birth subtypes at different gestational ages. After covariates adjustment, preterm PROM and indicated preterm birth had higher risk of neonatal death than preterm labor after 28 weeks of gestation. The increased risk from preterm PROM and indicated preterm birth was not limited to early neonatal death in the first 7 days. Unadjusted results were similar to adjusted ones (data not shown).
When causes of neonatal death were separated, preterm birth subtypes had different associations with these causes (Table 4). At 24–27 weeks, indicated preterm births tended to have a higher risk of neonatal death from RDS/respiratory conditions and birth defects. At 28–36 weeks, preterm PROM also had higher risk of neonatal death from these two conditions. Indicated preterm birth had higher risk of neonatal death from all causes at 32–36 weeks (Table 4).
In this study, neonates whose mothers had preterm PROM had the highest risk of neonatal death among preterm birth subtypes, likely due to shorter gestational age and higher gestational-age-specific neonatal death risk at 28–36 weeks. Indicated preterm birth had a consistently higher risk of neonatal death than preterm labor at 28–36 weeks of gestation. The results generally suggest that heterogeneity of preterm birth subtypes on neonatal death risk exist.
Previous studies were not consistent in neonatal death risks by preterm birth subtypes, but large-scale studies tended to have results similar to ours.8,14 A multinational hospital-based study of 1.7 million births in Latin America suggested preterm PROM and indicated preterm birth had higher neonatal death risk than idiopathic preterm labor, but the authors did not report gestational-age-specific results.14 For gestational-age-specific neonatal death risks between preterm PROM and preterm labor, the World Health Organization study found preterm PROM had lower risk.6 It is possible that the World Health Organization study sample was relatively small and more selective (as in the clinical trial), and the study only evaluated neonatal death before hospital discharge (not complete for referrals to other institutions and for deaths after discharge). For indicated preterm birth, the Swedish study of 1 million births suggested it had higher neonatal death risk than spontaneous preterm birth (preterm PROM and preterm labor combined), most pronounced at 34–36 weeks.8 Another small matched study of women with indicated cesarean delivery and women with spontaneous preterm birth at 24–33 weeks of gestation found a twofold risk of neonatal death in the cesarean delivery groups.15 Our study confirmed higher risk of neonatal death in indicated preterm birth, with the risk ratio higher at late preterm birth. We found at any preterm gestational age that neonatal death due to RDS and other respiratory conditions was more common in indicated preterm birth compared with preterm labor, suggesting immature lung development. Indicated preterm birth was associated with higher risk of neonatal death from all causes at 32–36 weeks, which indicates that the underlying causes of indicated preterm birth may be more diverse at this gestational age.
Preterm PROM, indicated preterm birth, and preterm labor share certain causative components, such as infection, inflammation,16,17 and placental vascular lesions,18 but they may differ in the scope and severity of these conditions.19 It has been shown that infection and inflammation are more common in early than in late preterm births,16 and this may be more related to preterm PROM, which has lower mean gestational age. Indicated preterm birth is mostly due to maternal and fetal indications,19 and the underlying diseases or conditions may actually contribute to the neonatal death risk. It has been suggested that “ischemic placental disease,” including preeclampsia, fetal distress, small for gestational age, and placental abruption,20,21 may explain most of the indicated preterm births. Further studies are needed to differentiate possible underlying pathophysiology of preterm subtypes using biomarkers in blood, amniotic fluid, or placenta samples.22
This study has several limitations. First, the preterm birth subtypes may be misclassified. Preterm PROM more than 12 hours was recorded on the birth certificate, thus those with preterm PROM 12 hours or less were not identified. It is unclear whether preterm PROM 12 hours or less had worse neonatal outcomes than preterm PROM more than 12 hours, but one study reported preterm PROM less than 48 hours had higher neonatal death risk at less than 30 weeks of gestation but not thereafter than those with longer latency.23 It is also possible preterm PROM more than 12 hours was underreported, because clinical samples tended to have higher preterm PROM proportion (1–4%),24–26 but the extent of underreporting is unclear in birth certificates. Unless those underreported preterm PROM had much better neonatal outcome, the results from this study are still informative. Indicated preterm birth may include underreported emergency cesarean delivery after abnormal labor. Emergency cesarean delivery, however, accounted for about 3% of all cesarean deliveries.27 Cesarean delivery on maternal request would be uncommon at a preterm gestational age.28 Second, the gestational age estimation may not be entirely accurate. In the vital statistics, there can be a bimodal birth weight distribution for births at 28–31 weeks, which indicates possible misclassification of gestational age.29 This is, however, less obvious after 32 weeks of gestation. Our results for preterm births after 32 weeks should be less affected by this concern. Clinical estimation of gestational age was not reported in California birth certificates; and it may also be subject to error. Therefore, further studies may use early ultrasonography-based gestational age for gestational-age-specific analysis. Third, this study only used data from 2001 births, and the generalizability to births many years away needs to be determined in future studies. It seems that some preterm PROM and preterm labor in previous years may end up with indicated preterm birth later.10 Thus, the relative effect size between indicated preterm birth and preterm labor may vary by time, but no dramatic change is expected in the neighboring years.
Despite the limitations, this study has several strengths. This is the first attempt to investigate the effect of preterm birth subtypes on neonatal mortality using national data in the United States. The birth cohort with almost complete linkage of birth and death certificates confers advantage for this analysis. The underlying causes of death were also explored to indicate possible heterogeneity. The results of this study may stimulate future research into causative factors of preterm birth subtypes and their prevention. Clinicians may consider a specific treatment regimen to reduce a preterm birth subtype (eg, preterm PROM) and prevent subsequent neonatal death. Clinical decision of labor induction and cesarean delivery for indicated preterm birth at a given gestational age needs to take into account the risk of adverse neonatal outcomes. To conclude, although gestational age is a strong predictor of neonatal death in preterm infants, preterm birth subtypes showed some heterogeneity in gestational age and birth weight distribution, risk of neonatal death, and the underlying causes of death.
1. Cunningham FG, Leveno KJ, Bloom SL, Hauth JC, Gilstrap LC III, Wenstrom KD. Williams obstetrics. 22nd ed. New York (NY): McGraw-Hill Medical Publishing Division; 2005.
2. Ananth CV, Vintzileos AM. Epidemiology of preterm birth and its clinical subtypes. J Matern Fetal Neonatal Med 2006;19:773–82.
3. Savitz DA, Blackmore CA, Thorp JM. Epidemiologic characteristics of preterm delivery: etiologic heterogeneity. Am J Obstet Gynecol 1991;164:467–71.
4. Berkowitz GS, Blackmore-Prince C, Lapinski RH, Savitz DA. Risk factors for preterm birth subtypes. Epidemiology 1998;9:279–85.
5. Zhang J, Savitz DA. Preterm birth subtypes among blacks and whites. Epidemiology 1992;3:428–33.
6. Villar J, Abalos E, Carroli G, Giordano D, Wojdyla D, Piaggio G, et al. Heterogeneity of perinatal outcomes in the preterm delivery syndrome. Obstet Gynecol 2004;104:78–87.
7. Callaghan WM, MacDorman MF, Rasmussen SA, Qin C, Lackritz EM. The contribution of preterm birth to infant mortality rates in the United States. Pediatrics 2006;118:1566–73.
8. Morken NH, Kallen K, Jacobsson B. Outcomes of preterm children according to type of delivery onset: a nationwide population-based study. Paediatr Perinat Epidemiol 2007;21:458–64.
9. Martin JA, Hamilton BE, Ventura SJ, Menacker F, Park MM, Sutton PD. Births: final data for 2001. Natl Vital Stat Rep 2002;51:1–102.
10. Ananth CV, Joseph KS, Oyelese Y, Demissie K, Vintzileos AM. Trends in preterm birth and perinatal mortality among singletons: United States, 1989 through 2000. Obstet Gynecol 2005;105:1084–91.
11. Nohr EA, Bech BH, Vaeth M, Rasmussen KM, Henriksen TB, Olsen J. Obesity, gestational weight gain and preterm birth: a study within the Danish National Birth Cohort. Paediatr Perinat Epidemiol 2007;21:5–14.
12. Nohr EA, Vaeth M, Bech BH, Henriksen TB, Cnattingius S, Olsen J. Maternal obesity and neonatal mortality according to subtypes of preterm birth. Obstet Gynecol 2007;110:1083–90.
13. Ananth CV, Getahun D, Peltier MR, Salihu HM, Vintzileos AM. Recurrence of spontaneous versus medically indicated preterm birth. Am J Obstet Gynecol 2006;195:643–50.
14. Barros FC, Vélez Mdel P. Temporal trends of preterm birth subtypes and neonatal outcomes. Obstet Gynecol 2006;107:1035–41.
15. Kurkinen-Raty M, Koivisto M, Jouppila P. Preterm delivery for maternal or fetal indications: maternal morbidity, neonatal outcome and late sequelae in infants. BJOG 2000;107:648–55.
16. Goldenberg RL, Hauth JC, Andrews WW. Intrauterine infection and preterm delivery. N Engl J Med 2000;342:1500–7.
17. Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet 2008;371:75–84.
18. Arias F, Rodriquez L, Rayne SC, Kraus FT. Maternal placental vasculopathy and infection: two distinct subgroups among patients with preterm labor and preterm ruptured membranes. Am J Obstet Gynecol 1993;168:585–91.
19. Behrman RE, Stith Butler A, editors. Institute of Medicine, Committee on Understanding Preterm Birth and Assuring Healthy Outcomes, Board on Health Sciences Policy. Preterm Birth: Causes, Consequences, and Prevention. Washington, DC: The National Academies Press; 2007.
20. Ananth CV, Vintzileos AM. Medically indicated preterm birth: recognizing the importance of the problem. Clin Perinatol 2008;35:53–67.
21. Ananth CV, Vintzileos AM. Maternal–fetal conditions necessitating a medical intervention resulting in preterm birth. Am J Obstet Gynecol 2006;195:1557–63.
22. Himes KP, Simhan HN. Risk of recurrent preterm birth and placental pathology. Obstet Gynecol 2008;112:121–6.
23. Pasquier JC, Bujold E, Rabilloud M, Picaud JC, Ecochard R, Claris O, et al. Effect of latency period after premature rupture of membranes on 2 years infant mortality (DOMINOS study). Eur J Obstet Gynecol Reprod Biol 2007;135:21–7.
24. Ernest JM. Neonatal consequences of preterm PROM. Clin Obstet Gynecol 1998;41:827–31.
25. Lee T, Silver H. Etiology and epidemiology of preterm premature rupture of the membranes. Clin Perinatol 2001;28:721–34.
26. Mercer BM. Preterm premature rupture of the membranes. Obstet Gynecol 2003;101:178–93.
27. Chauhan SP, Magann EF, Scott JR, Scardo JA, Hendrix NW, Martin JN Jr. Cesarean delivery for fetal distress: rate and risk factors. Obstet Gynecol Surv 2003;58:337–50.
28. American College of Obstetricians and Gynecologists. ACOG Committee Opinion No. 394, December 2007. Cesarean delivery on maternal request. Obstet Gynecol 2007;110:1501.
29. Ananth CV. Menstrual versus clinical estimate of gestational age dating in the United States: temporal trends and variability in indices of perinatal outcomes. Paediatr Perinat Epidemiol 2007;21 suppl:22–30.