Maternal and Child Health Journal

, Volume 16, Issue 4, pp 807–813

Singleton Preterm Birth: Risk Factors and Association with Assisted Reproductive Technology

Authors

    • Division of Reproductive Health, National Center for Chronic Disease Prevention and Health PromotionCenters for Disease Control and Prevention
  • Sherry L. Farr
    • Division of Reproductive Health, National Center for Chronic Disease Prevention and Health PromotionCenters for Disease Control and Prevention
  • Bruce B. Cohen
    • Bureau of Health Information, Statistics, Research and EvaluationMassachusetts Department of Public Health
  • Angela Nannini
    • University of Massachusetts Lowell, Department of NursingCollege of Health and Environment
  • Zi Zhang
    • Center for Health Policy and ResearchUniversity of Massachusetts Medical School
  • John E. Anderson
    • Division of Reproductive Health, National Center for Chronic Disease Prevention and Health PromotionCenters for Disease Control and Prevention
  • Denise J. Jamieson
    • Division of Reproductive Health, National Center for Chronic Disease Prevention and Health PromotionCenters for Disease Control and Prevention
  • Maurizio Macaluso
    • University of Cincinnati College of Medicine
Article

DOI: 10.1007/s10995-011-0787-8

Cite this article as:
Tepper, N.K., Farr, S.L., Cohen, B.B. et al. Matern Child Health J (2012) 16: 807. doi:10.1007/s10995-011-0787-8

Abstract

The objectives of this study were to determine risk factors for early (less than 34 weeks gestation) and late (34–36 weeks gestation) preterm singleton birth, by assisted reproductive technology (ART) status. We linked data from Massachusetts birth records and ART records representing singleton live births from 1997 through 2004. Using multinomial regression models, we assessed risk factors for early and late preterm birth by ART status. From 1997 to 2004 in Massachusetts, among non-ART births, risk factors for early and late preterm birth were similar and included women <15 and ≥35 years of age, those of non-white race or Hispanic ethnicity, those with ≤12 years of education, those with chronic diabetes, those with gestational diabetes, those with gestational hypertension, those who smoked during pregnancy, those who used fertility medications, and those who had not had a previous live birth. Among ART births, risk factors for early and late preterm birth differed and odds of early preterm birth were increased among women with ≤12 years of education while odds of late preterm birth were increased among women with gestational diabetes. Odds of both early and late preterm birth were increased among women of non-white race or Hispanic ethnicity and among women with gestational hypertension. Among non-ART births, increased risk for preterm birth was more strongly related to socioeconomic factors than among ART births. Medical conditions were associated with an increased risk for preterm birth regardless of women’s ART status. Efforts to prevent preterm births should focus on reducing modifiable risk factors.

Keywords

Assisted reproductive technologyEarly preterm birthLate preterm birthRisk factors

Introduction

Preterm birth (before 37 weeks gestation) is one of the most significant contributors to neonatal and infant morbidity and mortality in the United States. One of the objectives of U.S. Healthy People 2010 is to reduce the preterm birth rate to less than 7.6% [1]. However, despite efforts to reduce this rate, it has increased steadily from 1990 to 2006 (from 10.6 to 12.8% overall and from 9.7 to 11.1% among singleton births) [2]. Much of this increase is due to an increase in late preterm births (between 34 and 36 weeks gestation) [2], which now account for approximately 75% of all preterm births [3]. Although born near term, late preterm infants have higher rates of death and complications than term infants [4].

Assisted reproductive technology (ART), infertility treatments in which both eggs and sperm are handled in a laboratory for the purpose of establishing a pregnancy, has been associated with an increased risk for preterm birth [5, 6]. The percentage of infants conceived through ART has been rising steadily and constituted 1% of all infants born in the U.S. in 2006 [7]. The association between ART and preterm birth is found among multiple births, which tend to be born at earlier gestational ages [6]. However, ART has also been shown to be associated with preterm birth among singletons, although the etiology for this association has not been clearly elucidated [8, 9]. The etiology of preterm birth is multifactorial and risk factors for early and late preterm births may differ by ART status. Therefore, using linked data from Massachusetts birth records and Massachusetts ART records from singleton births during 1997–2004, we examined selected risk factors for early and late preterm birth and sought to better characterize differences by ART status.

Materials and Methods

Study Population

The study population consisted of all singleton infants born in Massachusetts from 1997 through 2004, the most recent data available (n = 616,632). Birth records for these infants were merged with Massachusetts live-birth ART records for 1997–2004. Details of this linkage have been described previously [10]. Following all births in Massachusetts, information from birth records is abstracted by hospital staff and sent to the Massachusetts Registry of Vital Records and Statistics. All clinics that perform ART report standardized data for every initiated ART cycle to the Centers for Disease Control and Prevention (CDC), as mandated by the Fertility Clinic Success Rate and Certification Act [11]. For this analysis, investigators linked birth records with live-birth ART records for mothers with a Massachusetts residence in two year intervals on the basis of maternal and infant dates of birth and plurality. Questions regarding duplicate matches were resolved using postal code of residence.

There were a total of 6,660 ART records that linked to a birth certificate record (83% of ART singleton live births). These were defined as ART births. ART records not linked to a birth record (n = 1,323, 17% of ART singleton live births) were excluded. The Massachusetts birth record collects information on whether ART was used to conceive the pregnancy. There were 364 births (0.06% of singleton births) that were excluded because they reported use of ART on the birth record but were not linked to an ART record. All other births were defined as non-ART births. Gestational age at birth was obtained from the clinical estimate in weeks recorded on the birth record. Births were excluded if the gestational age was less than 24 weeks (n = 5,703, 0.9% of singleton births) or greater than 43 weeks (n = 208, 0.03% of singleton births). Births were also excluded if the reported parity was incompatible with the maternal age (if maternal age minus parity was less than 12) (n = 302, 0.05% of singleton births). Births missing information on any risk factor were also excluded (n = 7,017, 1.1% of singleton births). Therefore, a total of 595,168 non-ART and 6,547 ART births were included in univariable and multivariable analyses.

Definitions

Births were categorized as early preterm (<34 weeks), late preterm (34–36 weeks), or term (≥37 weeks). Potential risk factors for preterm birth based on past literature were obtained from birth records. Demographic risk factors of interest included maternal age, race/ethnicity, and education. Maternal age was categorized as <15, 15–19, 20–34, 35–39, and ≥40 years. For non-ART births, maternal race/ethnicity was categorized as non-Hispanic white, non-Hispanic black, Hispanic, Asian/Pacific Islander, and American Indian or other race/ethnicity. For ART births, due to small numbers in some categories, maternal race/ethnicity was re-categorized as either non-Hispanic white or other. For non-ART births, education was categorized as <12 years, 12 years, and >12 years. For ART births, education was categorized as ≤12 years and >12 years.

Medical risk factors of interest included chronic diabetes, gestational diabetes, chronic hypertension, gestational hypertension, smoking before pregnancy, and smoking during pregnancy. Smoking before or during pregnancy was categorized as yes (>0 cigarettes per day) or no (0 cigarettes per day). For non-ART births, use of fertility medications in the current pregnancy was also examined.

Obstetric risk factors of interest included the number of previous live births (0 or ≥1) and having had a prior preterm or small for gestational age (SGA) infant (combined on the birth record and categorized as yes, no prior preterm birth, and no prior live births).

Statistical Analysis

To compare characteristics of non-ART births with those of ART births, the Wilcoxon rank-sum test was used for continuous variables and chi-square tests were used for categorical variables. Multinomial logistic regression was performed to produce odds ratios (ORs) and 95% confidence intervals (CIs) for early and late preterm birth compared to term birth for each risk factor. Models were constructed separately for non-ART and ART births due to the differences in characteristics among these births. Chronic hypertension and previous preterm or SGA infant were excluded from multivariable models because of a large number of records with missing data for these variables. For non-ART births, smoking prior to pregnancy was excluded from multivariable models because of its high correlation with smoking during pregnancy. For ART births, smoking prior to pregnancy, smoking during pregnancy, and preexisting diabetes were excluded from multivariable models due to some categories having less than 10 births. SAS statistical software version 9.1 (SAS Institute, Cary, NC) was used for all analyses. All research based on the data linkage was approved by the institutional review boards at CDC and the Massachusetts Department of Public Health.

Results

During 1997–2004, among non-ART births, 5.8% were born preterm, 4.3% late preterm and 1.5% early preterm. During the same time period, among ART births, 10.1% were born preterm, 7.0% late preterm and 3.1% early preterm. Characteristics of 595,168 non-ART and 6,254 ART births were compared in Table 1. Median maternal age was higher among ART births. Women who used ART were more likely to be white and have more education, chronic diabetes, gestational diabetes, chronic hypertension and gestational hypertension. Women who did not use ART were more likely to smoke before or during pregnancy and to have a previous live birth.
Table 1

Characteristics of women who delivered singleton infants, by use of ART—Massachusetts 1997–2004

Characteristic

Non-ART N (%)

ART N (%)

P-value

Total

595168

6547

 

Median age, year

30

36

<0.0001

Race

 Non-Hispanic white

435333 (73.1)

5913 (90.3)

<0.0001

 Non-Hispanic black

43308 (7.3)

142 (2.2)

<0.0001

 Hispanic

69802 (11.7)

176 (2.7)

<0.0001

 Asian/Pacific Islander

35007 (5.9)

264 (4.0)

<0.0001

 American Indian/Other

11718 (2.0)

52 (0.8)

<0.0001

Education

 <12 years

80706 (13.6)

104 (1.6)

<0.0001

 12 years

143076 (24.0)

853 (13.0)

<0.0001

 >12 years

371386 (62.4)

5590 (85.4)

<0.0001

Chronic diabetes

4434 (0.7)

112 (1.7)

<0.0001

Gestational diabetes

18204 (3.1)

334 (5.1)

<0.0001

Chronic hypertensiona

4633 (1.0)

92 (1.9)

<0.0001

Gestational hypertension

17931 (3.0)

321 (4.9)

<0.0001

Smoked cigarettes before pregnancya

100944 (17.0)

367 (5.6)

<0.0001

Smoked cigarettes during pregnancy

58219 (9.8)

110 (1.7)

<0.0001

≥1 previous live births

330726 (55.6)

2174 (33.2)

<0.0001

ART denotes assisted reproductive technology

aDoes not equal total number of births because of missing values

Among non-ART births, risk factors for early preterm birth were similar to those for late preterm birth (Table 2). In multivariable analyses, odds for both early and late preterm birth were elevated among women <15 and ≥35 years of age, those of non-white race or Hispanic ethnicity, those with ≤12 years of education, those with chronic diabetes, those with gestational diabetes, those with gestational hypertension, those who smoked during pregnancy, those who used fertility medications, and those who had not had a previous live birth. Sensitivity analyses were performed among three subgroups: women aged ≥20, women who did not use fertility medications, and non-Hispanic white women. In these sensitivity analyses, risk factors did not differ (data not shown).
Table 2

Risk factors for early and late preterm births among 595,168 singleton non-ART births—Massachusetts 1997–2004

Risk factors

Term (>37 weeks)

Late preterm (34–36 weeks)

Early preterm (24–33 weeks)

N (%)

N (%)

Multivariable OR (95% CI)

N (%)

Multivariable OR (95% CI)

Total births

560522 (94.2)

25,531 (4.3)

 

9,115 (1.5)

 

Agea

 <15

483 (0.1)

43 (0.2)

1.41 (1.03–1.93)

25 (0.3)

1.71 (1.14–2.58)

 15–19

362280 (6.5)

2174 (8.5)

1.03 (0.97–1.08)

974 (10.7)

1.04 (0.96–1.13)

 20–34

407518 (72.2)

17968 (70.4)

1.00

6157 (67.6)

1.00

 35–39

97140 (17.3)

4353 (17.1)

1.11 (1.07–1.15)

1537 (16.9)

1.27 (1.20–1.34)

 >=40

19101 (3.4)

993 (3.9)

1.24 (1.16–1.32)

422 (4.6)

1.68 (1.52–1.86)

Racea

 Non-Hispanic white

412527 (73.6)

17337 (67.9)

1.00

5469 (60.0)

1.00

 Non-Hispanic black

39174 (7.0)

2678 (10.5)

1.58 (1.51–1.65)

1456 (16.0)

2.64 (2.48–2.80)

 Hispanic

64842 (11.6)

3516 (13.8)

1.23 (1.18–1.28)

1444 (15.8)

1.51 (1.42–1.61)

 Asian/Pacific Islander

33052 (5.9)

1451 (5.7)

1.06 (1.00–1.12)

504 (5.5)

1.16 (1.05–1.27)

 American Indian/Other

10927 (2.0)

549 (2.2)

1.15 (1.05–1.26)

242 (2.7)

1.52 (1.33–1.73)

Educationa

 <12 years

74549 (13.3)

4363 (17.1)

1.24 (1.19–1.29)

1794 (19.7)

1.51 (1.41–1.62)

 12 years

133821 (23.9)

6559 (25.7)

1.10 (1.07–1.14)

2696 (29.6)

1.39 (1.32–1.46)

 >12 years

352152 (62.8)

14609 (57.2)

1.00

4625 (50.7)

1.00

Chronic diabetesa

 Yes

3744 (0.7)

509 (2.0)

2.40 (2.18–2.65)

181 (2.0)

2.36 (2.02–2.75)

 No

556778 (99.3)

25022 (98.0)

1.00

8934 (98.0)

1.00

Gestational diabetesa

 Yes

16485 (2.9)

1311 (5.1)

1.54 (1.45–1.63)

408 (4.5)

1.28 (1.15–1.42)

 No

544037 (97.1)

24220 (94.9)

1.00

8707 (95.5)

1.00

Chronic hypertensiona

 Yes

3871 (0.9)

448 (2.3)

b

314 (4.6)

b

 No

416611 (99.1)

18816 (97.7)

 

6489 (95.4)

 

Gestational hypertensiona

 Yes

15474 (2.8)

1803 (7.1)

2.54 (2.41–2.67)

654 (7.2)

2.53 (2.34–2.75)

 No

545048 (97.2)

23728 (92.9)

1.00

8461 (92.8)

1.00

Smoked cigarettes before pregnancya

 Yes

94110 (16.8)

5116 (20.0)

c

1718 (18.8)

c

 No

466398 (83.2)

20414 (80.0)

 

7395 (81.2)

 

Smoked cigarettes during pregnancya

 Yes

53764 (9.6)

3306 (13.0)

1.40 (1.34–1.46)

1149 (12.6)

1.33 (1.24–1.42)

 No

506758 (90.4)

22225 (87.0)

1.00

7966 (87.4)

1.00

Use of fertility medicationa

 Yes

1834 (0.3)

129 (0.5)

1.53 (1.28–1.83)

42 (0.5)

1.47 (1.08–2.00)

 No

558688 (99.7)

25402 (99.5)

1.00

9073 (99.5)

1.00

Previous live birthsa

 0

246955 (44.1)

12607 (49.4)

1.26 (1.22–1.29)

4880 (53.5)

1.54 (1.47–1.61)

 ≥1

313567 (55.9)

12924 (50.6)

1.00

4235 (46.5)

1.00

Prior preterm or small-for-gestational age infanta

 Yes prior preterm

3029 (0.7)

719 (3.7)

b

248 (3.6)

b

 No prior live births

184969 (44.0)

9435 (49.0)

b

3619 (53.3)

b

 No prior preterm

232388 (55.3)

9096 (47.3)

 

2927 (43.1)

 

ART denotes assisted reproductive technology

aChi-square P-value <0.05 for preterm birth compared with term birth

bNot included in multivariable model due to missing information

cNot included in multivariable model due to strong correlation with smoking during pregnancy

Among ART births, risk factors for early and late preterm birth differed in multivariable models (Table 3). Odds of early preterm birth were increased among women with ≤12 years of education. Odds of late preterm birth were increased among women with gestational diabetes. Odds of both early and late preterm birth were increased among women of non-white race or Hispanic ethnicity and among women with gestational hypertension. When the analysis was restricted to non-Hispanic white women, odds ratio estimates for all other risk factors were similar to the original model (data not shown).
Table 3

Risk factors for early and late preterm births among 6,547 Singleton ART Births—Massachusetts 1997–2004

Risk factors

Term (>37 weeks)

Late preterm (34–36 weeks)

Early preterm (24–33 weeks)

N (%)

N (%)

Multivariable OR (95% CI)

N (%)

Multivariable OR (95% CI)

Total

5886 (89.9)

460 (7.0)

 

201 (3.1)

 

Agea

 20–34

2368 (40.2)

178 (38.7)

1.00

91 (45.3)

1.00

 35–39

2361 (40.1)

188 (40.9)

1.06 (0.85–1.31)

67 (33.3)

0.77 (0.56–1.06)

 >=40

1157 (19.7)

94 (20.4)

1.06 (0.81–1.38)

43 (21.4)

1.01 (0.69–1.46)

Raceb

 Non-Hispanic white

5338 (90.7)

403 (87.6)

1.00

172 (85.6)

1.00

 Other race

548 (9.3)

57 (12.4)

1.37 (1.02-1.83)

29 (14.4)

1.52 (1.01–2.29)

Educationb

 ≤12 years

843 (14.3)

69 (15.0)

1.03 (0.79–1.35)

45 (22.4)

1.62 (1.15–2.29)

 >12 years

5043 (85.7)

391 (85.0)

1.00

156 (77.6)

1.00

Chronic diabetesb

 Yes

88 (1.5)

16 (3.5)

c

8 (4.0)

c

 No

5798 (98.5)

444 (96.5)

 

193 (96.0)

 

Gestational diabetesb

 Yes

282 (4.8)

35 (7.6)

1.51 (1.05–2.19)

17 (8.5)

1.64 (0.98–2.74)

 No

5604 (95.2)

425 (92.4)

1.00

184 (91.5)

1.00

Chronic hypertensionb

 Yes

68 (1.6)

15 (4.5)

d

9 (5.8)

d

 No

4225 (98.4)

319 (95.5)

 

147 (94.2)

 

Gestational hypertensionb

 Yes

261 (4.4)

44 (9.6)

2.23 (1.59–3.12)

16 (8.0)

1.84 (1.09–3.13)

 No

5625 (95.6)

416 (90.4)

1.00

185 (92.0)

1.00

Smoked cigarettes before pregnancy

 Yes

327 (5.6)

31 (6.7)

c

9 (4.5)

c

 No

5559 (94.4)

429 (93.3)

 

192 (95.5)

 

Previous live births

 0

3916 (66.5)

308 (67.0)

e

149 (74.1)

e

 ≥1

1970 (33.5)

152 (33.0)

 

52 (25.9)

 

ART denotes assisted reproductive technology

aNo women <20 years of age had an ART birth

bChi-square P-value < 0.05 for preterm birth compared with term birth

cNot included in multivariable model because of numbers less than 10 in some categories

dNot included in multivariable model because of missing data

eNot associated with outcome in multivariable model

Comment

Our results showed that risk factors for early and late singleton preterm birth were similar among non-ART births. However, risk factors for early and late singleton preterm birth differed among ART births. To our knowledge, this study is the first to examine the differences in risk factors for singleton preterm births between non-ART and ART births in the same population.

The risk for preterm birth is likely to have multiple causes. Among non-ART births, increased risk for preterm birth was related to socioeconomic factors and, in particular, was elevated among women at extremes of age, non-white race, and lower education. Lower socioeconomic status is related to higher rates of unintended pregnancies [12] and unintended pregnancy is associated with preterm birth among some subgroups [1315]. Socioeconomic factors were less strongly associated with the risk of preterm birth among the ART population, most likely due to their higher overall socioeconomic status and the intendedness of pregnancies in that group.

Among the non-ART population, socioeconomic factors were associated with a larger risk of early rather than late preterm birth, consistent with findings from another study [16]. Although the reason for this is unclear, it may be reflective of different etiologic processes underlying early and late preterm births.

Our analysis also showed an increased risk for preterm birth among non-ART births with reported use of fertility medications. Infertility treatments other than ART, such as ovarian hyperstimulation and intrauterine insemination, may increase the risk of preterm birth [17, 18]. Infertility treatment is likely to be underreported on birth certificates, similar to underreporting of ART [19]. The underreporting precludes us from more fully examining the impact of these treatments on the overall rate of preterm birth.

Our finding that medical conditions such as diabetes and hypertension were associated with an increased risk for preterm birth regardless of women’s ART status was also consistent with findings from another study [20]. Although the increasing prevalence of obesity in the U.S. population [21] is likely contributing to corresponding increases in related medical conditions, previous studies of the association between maternal obesity and preterm birth have produced inconsistent results [22, 23]. However, we were unable to assess rates of maternal obesity from these data. We were also unable to determine whether a preterm delivery was induced for medical reasons, although the increase in the frequency of late preterm births has been noted for both spontaneous and medically indicated deliveries [24, 25].

A major strength of this analysis was that our data come from two large population-based datasets that captured almost all births in Massachusetts during the study period. By linking birth certificate records and ART records, those births that resulted from ART were able to be more accurately identified. Although Massachusetts birth certificate records have collected information on the use of ART since 1996, this measure has been found to have a low sensitivity for ART treatment [19]. In addition, a more complete assessment of outcomes among infants conceived through ART was feasible, as outcome information is limited in ART records.

Some limitations should be considered before generalizing these results. First, reliability of data from birth records varies. Certain maternal risk factors, such as smoking, have moderate sensitivity on birth records, whereas others, such as chronic and gestational hypertension, have low sensitivity [26]. Clinical estimate of gestational age from the birth record was used to classify infants as preterm. Reliability of this measure is most likely higher among ART births, since the day of embryo transfer is known. Additionally, some ART births may have been classified as non-ART births if they were not linked to an ART record. However, given the greater than 80% linkage rate, and the large number of non-ART births relative to the number of ART births, any misclassification of ART births would not have substantially affected our results. In addition, records that did not link generally did not differ from linked records, as previously reported [10].

Second, these results represent Massachusetts births and may not be generalizable to the larger U.S. population. Because Massachusetts law requires commercial insurance coverage of ART, the profile of women using ART in Massachusetts may differ from the profile of women using ART in other states. Third, due to the large size of the study population, certain findings may have been statistically significant due to chance. However, the consistency of our findings with results from other studies of risk factors for preterm birth suggests validity of these findings. Finally, small numbers of non-white ART births did not allow sub-group analysis by race, which limits conclusions about preterm birth among higher risk groups such as non-Hispanic blacks [20, 25].

In conclusion, preterm birth rates remain a substantial problem, among both ART and non-ART births. Efforts to reduce preterm birth should focus on socioeconomic and medical risk factors among both ART users and non-ART users. Efforts to reduce the number of unintended pregnancies may help reduce preterm birth rates, particularly among women of low socioeconomic status, as may efforts to prevent and treat medical conditions such as diabetes and hypertension in reproductive-age woman before they conceive.

Acknowledgments

The original ART surveillance system data used for this study were collected by the Society for Assisted Reproductive Technology (SART). This system is jointly supported by SART, the American Society for Reproductive Medicine (ASRM), and the CDC. The authors thank SART and ASRM for their contributions to this work. The authors also gratefully acknowledge Maria Gallo, PhD, Jeffrey Wiener, MS, and William Callaghan, MD, MPH, for their helpful input and expertise.

Copyright information

© Springer Science+Business Media, LLC (outside the USA)  2011