Abstract
Risk factors for preterm delivery have been described, but whether risk factors differ in the context of prior preterm delivery history is less understood. We assessed whether known risk factors were different in women with versus without prior preterm delivery using medical records of the first and second singleton deliveries in 25,820 Utah women (2002–2010). Longitudinal transition models with modified Poisson regression calculated adjusted relative risks and 95% confidence intervals, with multiplicative interactions between each preterm risk factor and prior preterm delivery status to explore whether risk factors varied between incident and recurrent preterm delivery at <37 weeks. Fewer second pregnancy factors were associated with recurrent preterm delivery, including alcohol, thyroid disease, and depression. Smoking was associated with increased risk for incident (relative risk (RR) = 1.95, 95% confidence interval (CI): 1.53, 2.49) but not recurrent (RR = 1.09, 95% CI: 0.71, 1.19) preterm delivery, whereas alcohol was associated with an increased risk for recurrent (RR = 2.38, 95% CI: 1.53, 3.71) but not incident (RR = 0.98, 95% CI: 0.67, 1.43; Pinteraction = 0.02 and <0.01) preterm delivery, respectively. Prior term delivery did not necessarily confer protection from known second pregnancy preterm delivery risk factors. In the setting of a prior preterm delivery, many risk factors did not persist. Prior preterm delivery history is important when assessing subsequent preterm delivery risk factors.
Keywords: birth intervals, premature birth, reproductive history
Preterm delivery at <37 weeks of gestation complicated 11.5% of US births in 2012 (1). Prior research has established maternal risk factors for preterm delivery including non-Hispanic black race, lean body mass index (weight (kg)/height (m)2) of <19.8, and smoking (2–5). One of the major risk factors for preterm delivery is a history of prior preterm delivery, which is known to increase the risk of subsequent preterm delivery from 1.5- to 2-fold (6, 7). Yet preterm delivery also occurs in 5%–9% of deliveries for women without a prior preterm delivery, and less attention has been given to this particular group of women (7, 8). Understanding the relationship of risk factors for preterm delivery in a current pregnancy in the context of prior preterm delivery history is important because there may be behavioral or clinical markers that can help predict risk in a second pregnancy. For example, women with a prior term birth may or may not believe that they are susceptible to a risk factor for preterm birth in their second pregnancy and vice versa for women with a prior preterm delivery. Risk and patterns of recurrence for preterm delivery are known to differ on the basis of prior subtype, spontaneous or indicated, so it is important to understand whether risk factors might vary according to the prior preterm delivery subtype (9).
Our objectives were to 1) explore whether risk factors for preterm delivery in a second pregnancy differed on the basis of whether the first pregnancy delivered at term (≥37 weeks of gestation) or preterm (<37 weeks), 2) explore how changes in behavioral risk factors between pregnancies were associated with preterm delivery in the context of prior history, and 3) investigate whether risk factors for recurrent preterm delivery differed after taking prior subtype of preterm delivery, spontaneous or indicated, into account.
METHODS
Study population
The NICHD (Eunice Kennedy Shriver National Institute of Child Health and Human Development) Consecutive Pregnancies Study included 51,086 women with at least 2 consecutive deliveries at ≥20 weeks of gestation in Utah (20 hospitals) from 2002 to 2010. Details of the study have been previously published (9). In brief, detailed information on patient demographics, pregnancy, labor and delivery, and postpartum status was extracted from the antepartum and labor and delivery summary electronic medical records. International Classification of Diseases, Ninth Revision, Clinical Modification, diagnoses from maternal discharge summaries were linked to each delivery. Institutional review board approval was obtained at all participating institutions. The analysis was restricted to nulliparous women with singleton gestations in the first 2 deliveries (n = 27,077). Women with missing information on height (n = 813), prepregnancy body mass index (n = 1,138), marital status (n = 2), smoking (n = 19), drinking (n = 69), and drug use (n = 67) were excluded (note: women could be missing data on more than 1 variable; total missing = 1,257 or 4.6%), leaving 25,820 women for the present analysis.
Outcome
Preterm delivery was defined as delivery at <37 weeks by using the best clinical estimate as recorded in the electronic medical record. Preterm delivery was further characterized by subtype as spontaneous, indicated, and medically unnecessary by using an algorithm previously published by our group (9, 10). In brief, deliveries without an induction or prelabor cesarean delivery were categorized as spontaneous. Preterm premature rupture of membranes was also categorized as spontaneous because of small numbers. Pregnancies with both spontaneous labor or preterm premature rupture of membranes and a pregnancy complication (e.g., fetal growth restriction) were counted only as spontaneous labor. Indicated inductions and prelabor cesarean deliveries for fetal, maternal, or obstetrical indications were identified. Pregnancies with an induction or cesarean delivery without an indication recorded were also considered to be indicated rather than medically unnecessary when the pregnancy had complications (e.g., preeclampsia) as recorded in the medical record. Preterm deliveries that were considered “elective” by the site with no obstetrical, fetal, or maternal conditions reported were categorized as medically unnecessary, as well as inductions or prelabor cesarean deliveries in pregnancies without any other complications.
Predictors
Potential risk factors for preterm delivery at the second pregnancy were explored as previously reported by our group for a similar study using these data and as identified in the Institute of Medicine of the National Academies’ 2007 report from the Committee on Understanding Premature Birth and Assuring Healthy Outcomes (9, 11). Risk factors for preterm delivery included baseline maternal characteristics of age; race/ethnicity (white, black, Hispanic, Asian/Pacific Islander, or other/unknown); height; prepregnancy body mass index; marital status; insurance; smoking, alcohol, or drug use during pregnancy as recorded in the prenatal record (yes or no); chronic diseases including diabetes, hypertension, asthma, thyroid disease, depression, or other (human immunodeficiency virus, heart, renal, or gastrointestinal disease); uterine anomaly as identified by International Classification of Diseases, Ninth Revision, Clinical Modification, diagnoses (code 752.2, doubling of uterus (didelphic uterus), or code 752.3, other anomalies of uterus (bicornuate, unicornis, uterus with only 1 functioning horn)); prior pregnancy loss including miscarriage and terminations calculated as gravidity minus parity >1; and interpregnancy interval based on the number of days between the first delivery and the last menstrual period prior to the second pregnancy. Short stature was defined as <160 cm, and weight status was categorized as underweight (body mass index (BMI), <18.5), normal weight (BMI, 18.5–24.9), overweight (BMI, 25.0–29.9), or obese (BMI, ≥30). Chronic medical conditions were as recorded in the medical record and supplemented with discharge summary data using International Classification of Diseases, Ninth Revision, Clinical Modification, diagnoses.
Statistical analyses
Descriptive characteristics at the second pregnancy according to timing of delivery at the first and second pregnancies were summarized and tested for significance by using χ2 or Fisher's exact tests where indicated (Table 1). The first objective was to examine the risk of incident and recurrent preterm delivery associated with risk factors at the second pregnancy. We used longitudinal transition models with modified Poisson regression to calculate adjusted relative risks and 95% confidence intervals (12, 13). The modeling strategy was to examine the risk of a preterm birth as a linear (on the log scale for relative risks) function of risk factors and the occurrence of a preterm birth in the first pregnancy, as well as the interaction between the 2. The outcome in all analyses was the occurrence of preterm birth in the second pregnancy. The interaction term is an assessment of whether the risk factor changes for incidence versus recurrence for second pregnancy preterm birth. The specific statistical model for each objective is presented in a footnote for Tables 2 and 3 and Appendix Table 1. The reference group for each analysis was specific to each risk factor because we wanted to evaluate the risk of incident or recurrent preterm delivery across levels of each risk factor. The models are interpreted such that a significant interaction between the risk factor and prior preterm delivery status indicates that there is evidence that, for the particular risk factor, the relative risk of preterm delivery is different for incident and recurrent preterm deliveries. The interaction test identifies risk factors that are different between incidence and recurrence, but it does not specifically test whether the risk factor is significant for either incidence or recurrence. The model with interactions terms allows for the estimation of the relative risk for both incident and recurrent preterm deliveries. The advantage of transition models over more traditional multinomial models is that we can directly compare risk factors for incident versus preterm delivery, whereas multinomial models that group women according to their preterm delivery status in both pregnancies compare groups relative to women without preterm delivery in either pregnancy. Thus, a multinomial polychotomous logistic regression model would not satisfy our analytical goals.
Table 1.
Maternal Characteristics at the Second Pregnancy According to Timing of Delivery at the First and Second Pregnancies (n = 25,820), NICHD Consecutive Pregnancies Study, 2002–2010
| Second-Pregnancy Characteristics | First Pregnancy Terma (n = 23,819; 92.2%) |
First Pregnancy Pretermb (n = 2,001; 7.8%) |
P Valuec | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Second Pregnancy Term (n = 22,532) |
Second Pregnancy Preterm (n = 1,287) |
Second Pregnancy Term (n = 1,494) |
Second Pregnancy Preterm (n = 507) |
||||||
| No. | % | No. | % | No. | % | No. | % | ||
| Age, years | <0.01 | ||||||||
| 13–24 | 7,767 | 34.5 | 556 | 43.2 | 589 | 39.4 | 222 | 43.8 | |
| 25–34 | 16,855 | 61.5 | 684 | 53.2 | 839 | 56.2 | 263 | 51.9 | |
| 35–49 | 910 | 4.0 | 47 | 3.7 | 66 | 4.4 | 22 | 4.3 | |
| Race/ethnicity | 0.04 | ||||||||
| Non-Hispanic white | 19,993 | 88.7 | 1,111 | 86.3 | 1,301 | 87.1 | 449 | 88.6 | |
| Non-Hispanic black | 90 | 0.4 | 7 | 0.5 | 7 | 0.5 | 5 | 1.0 | |
| Hispanic | 1,890 | 8.4 | 126 | 9.8 | 140 | 9.4 | 44 | 8.7 | |
| Asian/Pacific Islander | 480 | 2.1 | 40 | 3.1 | 37 | 2.5 | 6 | 1.2 | |
| Other | 79 | 0.4 | 3 | 0.2 | 9 | 0.6 | 3 | 0.6 | |
| Height, cm | <0.01 | ||||||||
| <160 | 4,139 | 18.4 | 287 | 22.3 | 336 | 22.5 | 122 | 24.1 | |
| ≥160 | 18,393 | 81.6 | 1,000 | 77.7 | 1,158 | 77.5 | 385 | 75.9 | |
| Prepregnancy BMId | <0.01 | ||||||||
| Underweight (BMI, <18.5) | 1,157 | 5.1 | 124 | 9.6 | 63 | 4.2 | 31 | 6.1 | |
| Normal weight (BMI, 18.5–24.9) | 13,120 | 58.2 | 706 | 54.9 | 785 | 52.5 | 251 | 49.5 | |
| Overweight (BMI, 25.0–29.9) | 4,940 | 21.9 | 251 | 19.5 | 360 | 24.1 | 129 | 25.4 | |
| Obese (BMI, ≥30) | 3,315 | 14.7 | 206 | 16 | 286 | 19.1 | 96 | 18.9 | |
| Marital status | <0.01 | ||||||||
| Married | 20,360 | 90.4 | 1,084 | 84.2 | 1,281 | 85.7 | 435 | 83.8 | |
| Nonmarried | 2,172 | 9.6 | 203 | 15.8 | 213 | 14.3 | 82 | 16.2 | |
| Insurance | <0.01 | ||||||||
| Private | 16,731 | 74.3 | 898 | 69.8 | 1,015 | 67.9 | 340 | 67.1 | |
| Public | 5,801 | 25.8 | 389 | 30.2 | 479 | 32.1 | 167 | 32.9 | |
| Smoking during pregnancy | <0.01 | ||||||||
| No | 21,982 | 97.6 | 1,196 | 92.9 | 1,436 | 96.1 | 478 | 94.3 | |
| Yes | 550 | 2.4 | 91 | 7.1 | 58 | 3.9 | 29 | 5.7 | |
| Alcohol during pregnancy | <0.01 | ||||||||
| No | 22,194 | 98.5 | 1,258 | 97.8 | 1,480 | 99.1 | 484 | 95.5 | |
| Yes | 338 | 1.5 | 29 | 2.3 | 14 | 0.9 | 23 | 4.5 | |
| Drug use during pregnancy | <0.01 | ||||||||
| No | 22,506 | 99.9 | 1,272 | 98.8 | 1,491 | 99.8 | 502 | 99.0 | |
| Yes | 26 | 0.1 | 15 | 1.2 | 3 | 0.2 | 5 | 1.0 | |
| Diabetes | <0.01 | ||||||||
| None | 22,209 | 98.6 | 1,250 | 97.1 | 1,450 | 97 | 480 | 94.7 | |
| Prepregnancy | 323 | 1.4 | 37 | 2.9 | 44 | 3.0 | 27 | 5.3 | |
| Hypertension | <0.01 | ||||||||
| Normotensive | 22,383 | 99.3 | 1,270 | 98.7 | 1,467 | 98.2 | 499 | 98.4 | |
| Chronic hypertension | 149 | 0.7 | 17 | 1.3 | 27 | 1.8 | 8 | 1.6 | |
| Asthma | <0.01 | ||||||||
| No | 20,715 | 91.9 | 1,137 | 88.3 | 1,354 | 90.6 | 448 | 88.4 | |
| Yes | 1,817 | 8.1 | 150 | 11.7 | 140 | 9.4 | 59 | 11.6 | |
| Thyroid disease | 0.01 | ||||||||
| No | 21,375 | 94.9 | 1,218 | 94.6 | 1,423 | 95.3 | 465 | 91.7 | |
| Yes | 1,157 | 5.1 | 69 | 5.4 | 71 | 4.8 | 42 | 8.3 | |
| Depression | <0.01 | ||||||||
| No | 20,149 | 89.4 | 1,079 | 83.8 | 1,302 | 87.2 | 412 | 81.3 | |
| Yes | 2,383 | 10.6 | 208 | 16.2 | 192 | 12.9 | 95 | 18.7 | |
| Other chronic medical diseasee | <0.01 | ||||||||
| No | 21,771 | 96.6 | 1,222 | 95.0 | 1,426 | 95.5 | 476 | 93.9 | |
| Yes | 761 | 3.4 | 65 | 5.1 | 68 | 4.6 | 31 | 6.1 | |
| Uterine anomaly | <0.01 | ||||||||
| No | 22,497 | 99.8 | 1,281 | 99.5 | 1,482 | 99.2 | 501 | 98.8 | |
| Yes | 35 | 0.2 | 6 | 0.5 | 12 | 0.8 | 6 | 1.2 | |
| Prior miscarriage or termination | 0.05 | ||||||||
| No | 19,338 | 85.8 | 1,071 | 83.2 | 1,272 | 85.1 | 428 | 84.4 | |
| Yes | 3,194 | 14.2 | 216 | 16.8 | 222 | 14.9 | 79 | 15.6 | |
| Interpregnancy interval, months | <0.01 | ||||||||
| <12 | 5,629 | 25.0 | 419 | 32.6 | 470 | 31.5 | 159 | 31.4 | |
| 12–<18 | 6,074 | 27.0 | 315 | 24.5 | 341 | 22.8 | 129 | 25.4 | |
| 18–23 | 4,183 | 18.6 | 205 | 15.9 | 232 | 15.5 | 75 | 14.8 | |
| >23 | 6,646 | 29.5 | 348 | 27.0 | 451 | 30.2 | 144 | 28.4 | |
Abbreviations: BMI, body mass index; NICHD, Eunice Kennedy Shriver National Institute of Child Health and Human Development.
a Term, delivery at ≥37 weeks of gestation.
b Preterm, delivery at <37 weeks of gestation.
c P value for any difference across the 4 groups by using the χ2 or Fisher's exact test. Two-sided P values of <0.05 were considered significant.
d Expressed as weight (kg)/height (m)2.
e Other chronic disease includes heart, renal, and gastrointestinal disease or human immunodeficiency virus.
Table 2.
Adjusted Risk of Recurrent and Incident Preterm Delivery According to Maternal Characteristics at the Second Pregnancy, NICHD Consecutive Pregnancies Study, 2002–2010a
| Second-Pregnancy Characteristics | Incident Pretermb Delivery in the Second Pregnancy |
Recurrent Pretermb Delivery in the Second Pregnancy |
P Valuec | ||
|---|---|---|---|---|---|
| RR | 95% CI | RR | 95% CI | ||
| Age, years | 0.85 | ||||
| 13–24 | 1.22 | 1.08, 1.39 | 1.15 | 0.94, 1.41 | |
| 25–34 | 1.00 | Referent | 1.00 | Referent | |
| 35–49 | 0.97 | 0.72, 1.30 | 1.03 | 0.66, 1.60 | |
| Race/ethnicity | 0.20 | ||||
| Non-Hispanic white | 1.00 | Referent | 1.00 | Referent | |
| Non-Hispanic black | 1.13 | 0.53, 2.38 | 1.59 | 0.64, 3.96 | |
| Hispanic | 1.09 | 0.89, 1.34 | 0.97 | 0.69, 1.36 | |
| Asian/Pacific Islander | 1.39 | 1.01, 1.92 | 0.57 | 0.25, 1.27 | |
| Other | 0.59 | 0.19, 1.83 | 0.95 | 0.30, 3.00 | |
| Height, cm | 0.54 | ||||
| <160 | 1.18 | 1.03, 1.35 | 1.09 | 0.88, 1.35 | |
| ≥160 | 1.00 | Referent | 1.00 | Referent | |
| Prepregnancy BMId | 0.25 | ||||
| Underweight (BMI, <18.5) | 1.79 | 1.48, 2.17 | 1.36 | 0.93, 1.98 | |
| Normal weight (BMI, 18.5–24.9) | 1.00 | Referent | 1.00 | Referent | |
| Overweight (BMI, 25.0–29.9) | 0.90 | 0.78, 1.04 | 1.05 | 0.85, 1.3 | |
| Obese (BMI, ≥30) | 1.02 | 0.87, 1.19 | 0.97 | 0.76, 1.24 | |
| Marital status | 0.13 | ||||
| Married | 1.00 | Referent | 1.00 | Referent | |
| Nonmarried | 1.22 | 1.02, 1.46 | 0.95 | 0.71, 1.26 | |
| Insurance | 0.84 | ||||
| Private | 1.00 | Referent | 1.00 | Referent | |
| Public | 0.93 | 0.81, 1.07 | 0.96 | 0.77, 1.19 | |
| Smoking during pregnancy | 0.02 | ||||
| No | 1.00 | Referent | 1.00 | Referent | |
| Yes | 1.95 | 1.53, 2.49 | 1.09 | 0.71, 1.65 | |
| Alcohol during pregnancy | <0.01 | ||||
| No | 1.00 | Referent | 1.00 | Referent | |
| Yes | 0.98 | 0.67, 1.43 | 2.38 | 1.53, 3.71 | |
| Diabetes | 0.28 | ||||
| None | 1.00 | Referent | 1.00 | Referent | |
| Prepregnancy | 1.85 | 1.33, 2.58 | 1.39 | 0.93, 2.08 | |
| Hypertension | 0.06 | ||||
| Normotensive | 1.00 | Referent | 1.00 | Referent | |
| Chronic hypertension | 1.68 | 1.03, 2.74 | 0.75 | 0.37, 1.54 | |
| Asthma | 0.25 | ||||
| No | 1.00 | Referent | 1.00 | Referent | |
| Yes | 1.33 | 1.12, 1.58 | 1.10 | 0.83, 1.45 | |
| Thyroid disease | 0.09 | ||||
| No | 1.00 | Referent | 1.00 | Referent | |
| Yes | 1.07 | 0.84, 1.37 | 1.53 | 1.11, 2.12 | |
| Depression | 0.86 | ||||
| No | 1.00 | Referent | 1.00 | Referent | |
| Yes | 1.35 | 1.16, 1.58 | 1.32 | 1.04, 1.67 | |
| Other chronic medical diseasee | 0.41 | ||||
| No | 1.00 | Referent | 1.00 | Referent | |
| Yes | 1.44 | 1.12, 1.85 | 1.19 | 0.83, 1.73 | |
| Uterine anomaly | 0.29 | ||||
| No | 1.00 | Referent | 1.00 | Referent | |
| Yes | 2.69 | 1.2, 6.02 | 1.45 | 0.64, 3.27 | |
| Prior miscarriage or termination | 0.35 | ||||
| No | 1.00 | Referent | 1.00 | Referent | |
| Yes | 1.14 | 0.98, 1.32 | 0.99 | 0.78, 1.27 | |
| Interpregnancy interval, months | 0.05 | ||||
| <12 | 1.33 | 1.15, 1.54 | 1.05 | 0.83, 1.32 | |
| 12–<18 | 1.02 | 0.87, 1.19 | 1.16 | 0.91, 1.48 | |
| 18–23 | 1.00 | Referent | 1.00 | Referent | |
| >23 | 0.98 | 0.82, 1.16 | 1.05 | 0.79, 1.39 | |
Abbreviations: BMI, body mass index; CI, confidence interval; NICHD, Eunice Kennedy Shriver National Institute of Child Health and Human Development; PTD, preterm delivery; RR, adjusted relative risk (adjusted for all of the characteristics in the table).
a The equation for this model is as follows: log Prob(PTD2 = 1) = β0 + β1(PTD1) + β2(risk factor X2) + β3(PTD1) × (risk factor X2), where PTD1 and PTD2 are indicators of a preterm birth in the first and second pregnancy, respectively. Risk factor X2 is a risk factor in the second pregnancy. The model shows a single risk factor, but additional risk factors were added by including additional terms for each risk factor and its interaction with the first pregnancies’ preterm delivery status, PTD1.
b Preterm, delivery at <37 weeks of gestation.
c P value represents the interaction between incident and recurrent preterm delivery for the respective characteristic.
d Expressed as weight (kg)/height (m)2.
e Other chronic disease includes heart, renal, and gastrointestinal disease or human immunodeficiency virus.
Table 3.
Adjusted Risk of Recurrent and Incident Preterm Deliveries According to Changes in Risk Behaviors Between the First and Second Pregnancies, NICHD Consecutive Pregnancies Study, 2002–2010a
| First-Second Pregnancy Characteristics | Incident Preterm Deliveryb (n = 1,283) |
Recurrent Preterm Deliveryb (n = 507) |
P Valuec | ||||||
|---|---|---|---|---|---|---|---|---|---|
| No. | % | RRd,e | 95% CI | No. | % | RRd,e | 95% CI | ||
| Smoking during pregnancy | 0.05 | ||||||||
| No-No | 1,182 | 92.1 | 1.00 | Referent | 461 | 90.9 | 1.00 | Referent | |
| No-Yes | 51 | 4.0 | 2.20 | 1.62, 2.98 | 8 | 1.6 | 1.01 | 0.49, 2.07 | |
| Yes-No | 10 | 0.8 | 0.89 | 0.48, 1.67 | 17 | 3.4 | 1.53 | 0.91, 2.56 | |
| Yes-Yes | 40 | 3.1 | 1.70 | 1.21, 2.38 | 21 | 4.1 | 1.21 | 0.74, 1.99 | |
| Alcohol during pregnancy | 0.03 | ||||||||
| No-No | 1,236 | 96.3 | 1.00 | Referent | 469 | 92.5 | 1.00 | Referent | |
| No-Yes | 26 | 2.0 | 1.15 | 0.77, 1.72 | 19 | 3.7 | 2.54 | 1.58, 4.10 | |
| Yes-No | 18 | 1.4 | 0.67 | 0.42, 1.08 | 15 | 3.0 | 0.89 | 0.52, 1.51 | |
| Yes-Yes | 3 | 0.2 | 0.39 | 0.13, 1.22 | 4 | 0.8 | 1.65 | 0.58, 4.69 | |
Abbreviations: CI, confidence interval; NICHD, Eunice Kennedy Shriver National Institute of Child Health and Human Development; RR, adjusted relative risk.
a There were 31 women missing information on smoking status in the first pregnancy and 45 missing information on alcohol (2 missing both) who were excluded from this analysis.
b Preterm, delivery at <37 weeks of gestation.
c P value represents the interaction between recurrent and incident preterm deliveries for the respective change characteristic. Two-sided P values of <0.05 were considered significant.
d Relative risks were calculated for each individual risk factor (smoking or alcohol) with adjustment for maternal age, race, height, prepregnancy body mass index, marital status, insurance, smoking (in the alcohol model), alcohol (in the smoking model), pregestational diabetes, chronic hypertension, asthma, thyroid disease, depression, other chronic medical diseases (includes human immunodeficiency virus and heart, renal, or gastrointestinal disease), prior miscarriage or termination, uterine anomaly, and interpregnancy interval.
e The equation for this model is as follows: log Prob(PTD2=1) = β0 + β1(PTD1) + β2(smoking change categorical variable) + β3(PTD1) × (smoking change categorical variable), where (smoking change categorical variable) reflects changes in risk factors from the first to the second pregnancy. For example, the smoking change categorical variable includes the 4 combinations in pregnancy 1 and pregnancy 2: smoking no-no, no-yes, yes-no, yes-yes. Additional adjustments for second-pregnancy factors were added to the above model.
Our second objective was to explore how changes in behavioral risk factors between pregnancies were associated with preterm delivery in the context of prior history. For the behavioral risk factors of smoking and alcohol that were as recorded in the prenatal record (yes or no), we created 4 new categories for possible combinations in pregnancies 1 and 2: no-no, no-yes, yes-no, yes-yes. We estimated the relative risks of incident and recurrent preterm delivery associated with the change in risk factor categories between the first and second pregnancies with women who were in the no-no group as the reference and adjusted for the previously discussed risk factors. These models were similar to those above except that smoking or alcohol was included as the categorical change variable (refer to the equation in footnote e of Table 3).
The third objective was to investigate whether risk factors for recurrent preterm delivery differed by prior subtype of previous preterm delivery, spontaneous or indicated. We limited this analysis to women with preterm delivery in the first pregnancy, excluding prior preterm deliveries that were medically unnecessary (n = 249), leaving 1,752 women. For these models, the outcome was recurrent preterm delivery in the second pregnancy, while the predictors were preterm delivery subtype (spontaneous or indicated), second pregnancy risk factors, and a multiplicative interaction term between each of the risk factors and prior preterm delivery subtype. Because of the small numbers, race was categorized as non-Hispanic white versus nonwhite for this analysis. Analyses were performed by using SAS, version 9.3, statistical software (SAS Institute, Inc., Cary, North Carolina), and 2-sided P values of <0.05 were considered significant.
RESULTS
Preterm delivery at <37 weeks occurred in 2,001 (7.8%) of first deliveries to the 25,820 women who were nulliparous at study entry. At the second pregnancy, 1,794 (6.9%) were complicated by preterm delivery, of which 1,287 (5.4%) occurred among the 23,819 women with a prior term delivery (i.e., incident preterm delivery), and 507 (25.3%) occurred among the 2,001 women with a prior preterm delivery (i.e., recurrent preterm delivery).
Descriptive characteristics at the second pregnancy according to timing of delivery (preterm or term) at the first and second pregnancies are presented in Table 1. Women with an incident preterm delivery in the second pregnancy were different from women with 2 term deliveries (Table 1). In contrast, women with recurrent preterm delivery had similar maternal demographic characteristics compared with women with a prior preterm delivery who delivered their second pregnancy at term (Table 1), but they were more likely to have risky behaviors and chronic medical problems.
We first explored whether risk factors for preterm delivery in a second pregnancy differed on the basis of whether the first pregnancy delivered at term (≥37 weeks of gestation) or preterm (<37 weeks) and found different risk patterns based on prior history (Table 2). Young maternal age, Asian/Pacific Islander race, short stature, nonmarried status, prepregnancy underweight, smoking, diabetes, hypertension, asthma, depression, other chronic diseases, uterine anomaly, and short pregnancy interval of <12 months were all associated with an increased risk for incident preterm delivery. Fewer factors were associated with recurrent preterm delivery, including alcohol, thyroid disease, and depression. However, the only statistically significant differences for risk factors between second pregnancy incident and recurrent preterm delivery were observed for women who smoked (P = 0.02) or drank alcohol (P < 0.01). Smoking was associated with an increased risk for incident preterm delivery (relative risk (RR) = 1.95, 95% confidence interval (CI): 1.53, 2.49) but not recurrent preterm delivery (RR = 1.09, 95% CI: 0.71, 1.19). However, it is important to note that the underlying baseline risk was higher for recurrent compared with incident preterm delivery, as incident preterm delivery occurred in 14.2% (91/641) of smokers compared with 5.2% (1,196/23,178) of nonsmokers, while recurrent preterm delivery occurred in 33.3% (29/87) of smokers and 25.0% (478/1,914) of nonsmokers (calculated from Table 1). Alcohol use was associated with an increased risk for recurrent (RR = 2.38, 95% CI: 1.53, 3.71) but not incident (RR = 0.98, 95% CI: 0.67, 1.43) preterm delivery. Incident preterm delivery occurred in 7.9% (29/367) of alcohol users compared with 5.4% (1,258/23,452) of alcohol nonusers, while recurrent preterm delivery occurred in 62.2% (23/37) of alcohol users and 24.6% (484/1,964) of alcohol nonusers.
There was a suggestion that the risk for preterm delivery in the second pregnancy differed according to the woman's first pregnancy preterm delivery history for nonmarried status, chronic hypertension, thyroid disease, and interpregnancy interval of <12 months. All were associated with an increased risk of preterm delivery in women with a prior term delivery, except for thyroid disease which was associated with an increased risk of recurrent but not incident preterm delivery (Table 2).
We also explored how changes in the behavioral risk factors of smoking and alcohol use between pregnancies were associated with preterm delivery in the context of prior delivery timing history. When we investigated the changes in smoking status, both initiation and continuation of smoking in the second pregnancy were associated with increased risk of second pregnancy preterm delivery (2.20-fold and 1.70-fold, respectively) in women with a prior term delivery, while women who stopped smoking had no increased risk of incident preterm delivery in the second pregnancy compared with women who did not smoke in either pregnancy (Table 3). Changes in smoking status were not associated with differences in risk of recurrent preterm delivery. When we investigated changes in drinking status, we found no association with differences in risk of incident preterm delivery in the second pregnancy (Table 3). However, women who started drinking had a 2.54-fold (95% CI: 1.58, 4.10) increased risk of recurrent preterm delivery (Table 3).
Finally, because risk factors might vary on the basis of the prior preterm delivery subtype, spontaneous or indicated, we investigated whether risk factors for recurrent preterm delivery differed after taking prior subtype into account. Recurrent preterm delivery was more common among women whose first pregnancy was complicated by a spontaneous (426/1,668; 26.8%) opposed to medically indicated (30/164; 18.3%) preterm delivery (P = 0.04). Alcohol use among women with a prior spontaneous preterm delivery and depression among women with a prior indicated preterm delivery were associated with recurrence (Appendix Table 1).
Thyroid disease was associated with recurrent preterm delivery for both prior subtypes. However, risk factors for preterm delivery in the second pregnancy were not significantly modified by prior preterm delivery subtype, spontaneous versus indicated, although the numbers were small (P values in Appendix Table 1). Recognizing that there were small numbers that limited our power to detect differences, alcohol use may have had a stronger association with recurrent preterm delivery of either type in women with prior spontaneous preterm delivery, depression may have been more associated with recurrent preterm delivery in women with prior indicated preterm delivery, and the association with thyroid disease appeared stronger in the women with prior indicated preterm delivery.
DISCUSSION
Among the first 2 deliveries for women in our study, we found that there were fewer risk factors for preterm delivery at <37 weeks of gestation in the second pregnancy for women with a prior history of preterm compared with term delivery. Importantly, there was the suggestion that risk factors for preterm delivery persisted even among women with a prior term delivery, indicating that these women remained at risk and were not protected from preterm delivery on the basis of their prior term delivery status. However, only alcohol use and possibly maternal thyroid disease and depression were associated with increased risk for recurrent preterm delivery. Thus, in the setting of a prior preterm delivery, many risk factors did not persist.
Our findings are not directly comparable to those in most prior literature as studies have included characteristics that occurred later in pregnancy in addition to baseline characteristics, studied only a few factors, and/or have not directly compared risk factors for incident versus recurrent preterm delivery (3–5, 14, 15). Nonetheless, our findings that smoking and short pregnancy interval were associated with incident preterm delivery are generally consistent with what has been reported in the literature (14–16). Fewer risk factors persisted for recurrent preterm delivery after accounting for the excess risk associated with prior preterm delivery history. Although the relative risks of many factors for recurrent preterm delivery were not increased, the baseline risk of preterm delivery was already so elevated that history of preterm delivery may have trumped many of the established preterm delivery risk factors from further increasing risk (7, 9, 17). This finding may in part be due to the smaller sample size for estimating the relative risk for recurrent versus incident preterm birth. However, in most cases the estimated relative risks were larger in magnitude for incident as compared with recurrent preterm birth, further supporting this point. Alcohol use is associated with increased prostaglandin production as measured in the urine of pregnant women and has been suggested as a biological pathway for increased spontaneous labor (18, 19). This possibility is supported by our finding that alcohol use was associated with recurrent preterm delivery possibly more in women with a prior spontaneous delivery. The finding that maternal thyroid disease and depression may have been more associated with recurrent, but not incident, preterm delivery is novel. Both have been described as risk factors for preterm delivery in general, particularly if untreated, although debate exists as to whether antidepressants themselves are associated with increased risk of preterm delivery (20–22). Our study lacked information on severity or treatment, so these findings warrant future investigation.
Our study is limited by the assessment of smoking and alcohol use as reported in the medical record. However, a previous methodological investigation found that misclassification of alcohol use has minimal bias for associations with an outcome in circumstances where there is underreporting and a large percentage of women are truly nondrinkers, which is more likely to be the case during pregnancy (23). We also lacked enough African-American women to explore whether this race was an independent risk factor in the context of history, as prior literature has reported African-American women to have higher recurrence rates (3). Our results are also exploratory in that we did not postulate biological mechanisms up front. The major strength of our study was the rich clinical data from women with multiple pregnancies allowing a comprehensive investigation of multiple risk factors. Such studies are quite rare as most prior investigations have relied on administrative data with inherent limitations. Although these observed differences in clinical risk patterns are important at a population level, it remains to be seen whether a strategy of obtaining multiple sources of more specific information such as placental pathology, genetics, and other biomarkers can better stratify women at risk for recurrent preterm delivery prior to planning their next pregnancy.
In total, our findings suggest that the association of preterm delivery risk factors for women who already have a history of preterm delivery varies from women without a history of preterm delivery. Women with thyroid disease and/or depression may need to be monitored more closely when they had a prior preterm birth than when they did not, although these study results would need to be confirmed in future studies for this to be widely recommended. Furthermore, there was also a suggestion that risk factors for recurrent preterm delivery such as alcohol use were modified by prior preterm delivery subtype, which may be due to the different etiologies for preterm delivery subtypes that drive these associations. Perhaps equally important was the finding that prior term delivery did not necessarily confer protection from known preterm delivery risk factors in the second pregnancy. It is important to take prior pregnancy delivery timing into account when assessing risk factors for subsequent preterm delivery risk.
ACKNOWLEDGMENTS
Author affiliations: Epidemiology Branch (Katherine L. Grantz, Stefanie N. Hinkle, Pauline Mendola, Lindsey A. Sjaarda), Biostatistics and Bioinformatics Branch (Paul S. Albert), and Glotech, Inc. (Kira Leishear), Division of Intramural Population Health Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development, Rockville, Maryland.
This work was supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (contracts HHSN275200800002I and HHSN27500004).
The results of this study were presented at the 27th Annual Meeting of the Society for Pediatric and Perinatal Epidemiologic Research, Seattle, Washington, June 24, 2014, and at the 47th Annual Meeting of the Society for Epidemiologic Research, Seattle, Washington, June 24–27, 2014.
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Eunice Kennedy Shriver National Institute of Child Health and Human Development.
Conflict of interest: none declared.
APPENDIX
Appendix Table 1.
Risk of Second-Pregnancy Preterm Delivery of Either Subtype by Prior Preterm Delivery Subtype, Spontaneous Versus Indicated, According to Maternal Characteristics at the Second Pregnancy (n = 1,752), NICHD Consecutive Pregnancies Study, 2002–2010a
| Second-Pregnancy Characteristics | Recurrent Preterm With Prior Spontaneous Preterm Deliveryb |
Recurrent Preterm With Prior Indicated Preterm Deliveryb |
P Valuec | ||||||
|---|---|---|---|---|---|---|---|---|---|
| No. | % | RRd | 95% CI | No. | % | RRd | 95% CI | ||
| No. of cases | 426/1,668 | 26.8 | 30/164 | 18.3 | |||||
| Age, years | 0.30 | ||||||||
| 13–24 | 193 | 45.3 | 1.19 | 0.95, 1.48 | 9 | 30.0 | 0.69 | 0.27, 1.78 | |
| 25–34 | 216 | 50.7 | 1.00 | Referent | 18 | 60.0 | 1.00 | Referent | |
| 35–49 | 17 | 4.0 | 1.05 | 0.64, 1.73 | 3 | 10.0 | 2.37 | 0.54, 10.50 | |
| Race/ethnicity | 0.41 | ||||||||
| Non-Hispanic white | 380 | 89.2 | 1.00 | Referent | 26 | 86.7 | 1.00 | Referent | |
| Nonwhite | 46 | 10.8 | 0.83 | 0.59, 1.16 | 4 | 13.3 | 1.55 | 0.38, 6.37 | |
| Height, cm | |||||||||
| <160 | 326 | 76.5 | 1.04 | 0.83, 1.31 | 22 | 73.3 | 0.96 | 0.36, 2.57 | 0.88 |
| ≥160 | 100 | 23.5 | 1.00 | Referent | 8 | 26.7 | 1.00 | Referent | |
| Prepregnancy BMIe | 0.37 | ||||||||
| Underweight (BMI, <18.5) | 30 | 7.0 | 1.42 | 0.96, 2.08 | 0 | 0.0 | |||
| Normal weight (BMI, 18.5–24.9) | 212 | 49.8 | 1.00 | Referent | 14 | 46.7 | 1.00 | Referent | |
| Overweight (BMI, 25.0–29.9) | 109 | 25.6 | 1.11 | 0.88, 1.40 | 7 | 23.3 | 0.97 | 0.32, 2.95 | |
| Obese (BMI, ≥30) | 75 | 17.6 | 1.05 | 0.80, 1.38 | 9 | 30.0 | 0.84 | 0.33, 2.12 | |
| Marital status | 0.78 | ||||||||
| Married | 355 | 83.3 | 1.00 | Referent | 24 | 80.0 | 1.00 | Referent | |
| Nonmarried | 71 | 16.7 | 0.95 | 0.70, 1.28 | 6 | 20.0 | 1.13 | 0.33, 3.85 | |
| Insurance | 0.15 | ||||||||
| Private | 286 | 67.1 | 1.00 | Referent | 15 | 50.0 | 1.00 | Referent | |
| Public | 140 | 32.9 | 0.93 | 0.74, 1.17 | 15 | 50.0 | 1.97 | 0.73, 5.37 | |
| Smoking during pregnancy | 0.11 | ||||||||
| No | 403 | 94.6 | 1.00 | Referent | 26 | 86.7 | 1.00 | Referent | |
| Yes | 23 | 5.4 | 1.00 | 0.62, 1.60 | 4 | 13.3 | 3.41 | 0.86, 13.45 | |
| Alcohol use during pregnancy | 0.19 | ||||||||
| No | 405 | 95.1 | 1.00 | Referent | 29 | 96.7 | 1.00 | Referent | |
| Yes | 21 | 4.9 | 2.55 | 1.60, 4.05 | 1 | 3.3 | 0.25 | 0.01, 10.08 | |
| Diabetes | 0.99 | ||||||||
| None | 406 | 95.3 | 1.00 | Referent | 27 | 90.0 | 1.00 | Referent | |
| Prepregnancy | 20 | 4.7 | 1.36 | 0.86, 2.17 | 3 | 10.0 | 1.38 | 0.28, 6.80 | |
| Hypertension | 0.53 | ||||||||
| Normotensive | 421 | 98.8 | 1.00 | Referent | 28 | 93.3 | 1.00 | Referent | |
| Chronic hypertension | 5 | 1.2 | 0.84 | 0.34, 2.08 | 2 | 6.7 | 1.85 | 0.20, 17.19 | |
| Asthma | 0.52 | ||||||||
| No | 374 | 87.8 | 1.00 | Referent | 27 | 90.0 | 1.00 | Referent | |
| Yes | 52 | 12.2 | 1.14 | 0.84, 1.54 | 3 | 10.0 | 1.81 | 0.48, 6.82 | |
| Thyroid disease | 0.08 | ||||||||
| No | 355 | 83.3 | 1.00 | Referent | 20 | 66.7 | 1.00 | Referent | |
| Yes | 71 | 16.7 | 1.48 | 1.04, 2.11 | 10 | 33.3 | 4.42 | 1.29, 15.15 | |
| Depression | 0.12 | ||||||||
| No | 391 | 91.8 | 1.00 | Referent | 26 | 86.7 | 1.00 | Referent | |
| Yes | 35 | 8.2 | 1.08 | 0.82, 1.41 | 4 | 13.3 | 2.86 | 1.04, 7.89 | |
| Other chronic medical diseasef | 0.52 | ||||||||
| No | 403 | 94.6 | 1.00 | Referent | 27 | 90.0 | 1.00 | Referent | |
| Yes | 23 | 5.4 | 1.12 | 0.73, 1.72 | 3 | 10.0 | 1.99 | 0.40, 9.87 | |
| Prior miscarriage or termination | 0.38 | ||||||||
| No | 365 | 85.7 | 1.00 | Referent | 22 | 73.3 | 1.00 | Referent | |
| Yes | 61 | 14.3 | 0.98 | 0.74, 1.29 | 8 | 26.7 | 1.52 | 0.61, 3.78 | |
| Interpregnancy interval, months | 0.74 | ||||||||
| <12 | 135 | 31.7 | 1.13 | 0.87, 1.46 | 12 | 40.0 | 1.88 | 0.57, 6.18 | |
| 12–<18 | 112 | 26.3 | 1.23 | 0.95, 1.60 | 9 | 30.0 | 2.49 | 0.67, 9.33 | |
| 18–23 | 60 | 14.1 | 1.00 | Referent | 3 | 10.0 | 1.00 | Referent | |
| >23 | 119 | 27.9 | 1.00 | 0.73, 1.36 | 6 | 20.0 | 1.83 | 0.40, 8.30 | |
Abbreviations: BMI, body mass index; CI, confidence interval; NICHD, Eunice Kennedy Shriver National Institute of Child Health and Human Development; PTD, preterm delivery; RR, adjusted relative risk.
a Medically unnecessary preterm deliveries (n = 249) in the first pregnancy were removed from this analysis. There were 1,668 women with spontaneous preterm delivery in the first pregnancy, of which 426 (26.8%) had a recurrent preterm delivery of either subtype, and 164 women with indicated preterm delivery in the first pregnancy, of which 30 (18.3%) had a recurrent preterm delivery of either subtype.
b Preterm, delivery at <37 weeks of gestation.
c P value represents the interaction between recurrent preterm delivery by prior preterm delivery subtypes for the respective characteristic. Two-sided P value of <0.05 was considered significant.
d The equation for this model is as follows: log P(PTD2 = 1) = β0 + β1(PTD1 subtype) + β2(risk factor X2) + β3(PTD1 subtype) × (risk factor X2), where PTD subtype is an indicator of whether the preterm delivery was spontaneous or indicated. We adjusted for additional second-pregnancy risk factors (Table 2) by adding these terms and their interaction with PTD1 to the above model.
e Expressed as weight (kg)/height (m)2.
f Other chronic medical disease includes human immunodeficiency virus and heart, renal, or gastrointestinal disease (uterine anomaly was not included because of small numbers).
REFERENCES
- 1.Hamilton BE, Martin JA, Ventura SJ. Births: preliminary data for 2012. Natl Vital Stat Rep. 2013;623:1–20. [PubMed] [Google Scholar]
- 2.Goldenberg RL, Culhane JF, Iams JD, et al. Epidemiology and causes of preterm birth. Lancet. 2008;3719606:75–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Adams MM, Elam-Evans LD, Wilson HG, et al. Rates of and factors associated with recurrence of preterm delivery. JAMA. 2000;28312:1591–1596. [DOI] [PubMed] [Google Scholar]
- 4.Mercer BM, Macpherson CA, Goldenberg RL, et al. Are women with recurrent spontaneous preterm births different from those without such history? Am J Obstet Gynecol. 2006;1944:1176–1184; discussion 1184–1185. [DOI] [PubMed] [Google Scholar]
- 5.Cnattingius S, Granath F, Petersson G, et al. The influence of gestational age and smoking habits on the risk of subsequent preterm deliveries. N Engl J Med. 1999;34113:943–948. [DOI] [PubMed] [Google Scholar]
- 6.Iams JD, Berghella V. Care for women with prior preterm birth. Am J Obstet Gynecol. 2010;2032:89–100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Mercer BM, Goldenberg RL, Moawad AH, et al. The preterm prediction study: effect of gestational age and cause of preterm birth on subsequent obstetric outcome. National Institute of Child Health and Human Development Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol. 1999;181(5 pt 1):1216–1221. [DOI] [PubMed] [Google Scholar]
- 8.Carr-Hill RA, Hall MH. The repetition of spontaneous preterm labour. Br J Obstet Gynaecol. 1985;929:921–928. [DOI] [PubMed] [Google Scholar]
- 9.Laughon SK, Albert PS, Leishear K, et al. The NICHD Consecutive Pregnancies Study: recurrent preterm delivery by subtype. Am J Obstet Gynecol. 2014;2102:131.e1–131.e8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Laughon SK, Reddy UM, Sun L, et al. Precursors for late preterm birth in singleton gestations. Obstet Gynecol. 2010;1165:1047–1055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Behrman RE, Butler AS, eds. Preterm Birth: Causes, Consequences, and Prevention. Washington, DC: The National Academies Press; 2007. [PubMed] [Google Scholar]
- 12.Diggle PJ, Heagerty PJ, Liang KY, et al. Transition models. In: Analysis of Longitudinal Data. 2nd ed. New York, NY: Oxford University Press; 2002:190–207. [Google Scholar]
- 13.Diggle PJ, Heagerty PJ, Liang KY, et al. Analysis of Longitudinal Data. 2nd ed New York, NY: Oxford University Press; 2002. [Google Scholar]
- 14.Mohsin M, Jalaludin B. Influence of previous pregnancy outcomes and continued smoking on subsequent pregnancy outcomes: an exploratory study in Australia. BJOG. 2008;11511:1428–1435. [DOI] [PubMed] [Google Scholar]
- 15.Simonsen SE, Lyon JL, Stanford JB, et al. Risk factors for recurrent preterm birth in multiparous Utah women: a historical cohort study. BJOG. 2013;1207:863–872. [DOI] [PubMed] [Google Scholar]
- 16.Smith GC, Pell JP, Dobbie R. Interpregnancy interval and risk of preterm birth and neonatal death: retrospective cohort study. BMJ. 2003;3277410:313. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.McManemy J, Cooke E, Amon E, et al. Recurrence risk for preterm delivery. Am J Obstet Gynecol. 2007;1966:576.e1–576.e6; discussion 576.e6–576.e7. [DOI] [PubMed] [Google Scholar]
- 18.Ylikorkala O, Halmesmäki E, Viinikka L. Urinary prostacyclin and thromboxane metabolites in drinking pregnant women and in their infants: relations to the fetal alcohol effects. Obstet Gynecol. 1988;711:61–66. [PubMed] [Google Scholar]
- 19.Albertsen K, Andersen AM, Olsen J, et al. Alcohol consumption during pregnancy and the risk of preterm delivery. Am J Epidemiol. 2004;1592:155–161. [DOI] [PubMed] [Google Scholar]
- 20.Männistö T, Mendola P, Grewal J, et al. Thyroid diseases and adverse pregnancy outcomes in a contemporary US cohort. J Clin Endocrinol Metab. 2013;987:2725–2733. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Casey BM, Dashe JS, Wells CE, et al. Subclinical hypothyroidism and pregnancy outcomes. Obstet Gynecol. 2005;1052:239–245. [DOI] [PubMed] [Google Scholar]
- 22.Stewart DE. Clinical practice. Depression during pregnancy. N Engl J Med. 2011;36517:1605–1611. [DOI] [PubMed] [Google Scholar]
- 23.Verkerk PH. The impact of alcohol misclassification on the relationship between alcohol and pregnancy outcome. Int J Epidemiol. 1992;21(suppl 1):S33–S37. [DOI] [PubMed] [Google Scholar]