The association between parity and spontaneous preterm birth: a population based study

This study was based on data from the Netherlands Perinatal Registry (PERINED). This database is a population based registry that covers approximately 97% of all deliveries in The Netherlands and contains information on deliveries at ≥22 weeks of gestation and birth weight of ≥500 g. Furthermore, all admissions to the neonatology care unit are registered until 28 days after birth. The perinatal database is obtained by a validated linkage of 3 different registries: the midwifery registry (LVR1), the obstetrics registry (LVR2), and the neonatology registry (LNR) of hospital admissions of new-born infants [13, 14]. It is used primarily for an annual assessment of the quality indicators of obstetric care.

The data in the perinatal registry are anonymous; therefore, ethical approval was not mandatory under Dutch law. The Netherlands Perinatal Registry gave their approval for the use of their data for this study (approval no. 17.34).

We studied singleton first, second, third, fourth and fifth pregnancies (P0 through P4) resulting in delivery between 22 and 43 weeks of gestation in the 5-year period from 2010 through 2014. We excluded multiple pregnancies and pregnancies that were complicated by congenital abnormalities or stillbirth.

Our primary outcome was risk of spontaneous preterm birth < 37 weeks of gestation per parity. Other outcome variables were spontaneous preterm birth < 32 and < 28 weeks. We performed additional analyses for the outcome late spontaneous preterm birth between 34 and 37 weeks to assess pregnancies in women that were not offered additional screening or treatment to prevent recurrent preterm birth. The PERINED registry contains data on whether a delivery started spontaneous (i.e., with spontaneous rupture of the membranes or contractions) or iatrogenic (i.e., planned Caesarean section or induction of labor).

To estimate the effect of parity on spontaneous preterm birth < 37, < 32 and < 28 weeks (and between 34 and 37 weeks), we used a univariate logistic regression model and expressed the effect estimates as odds ratios (OR) and corresponding 95% confidence intervals (CI). We used multivariate logistic regression analysis to adjust for the most common known risk factors for preterm birth that were available in the national perinatal registry that we used for our study. The chosen variables were based on previous studies about risk factors for (spontaneous) preterm birth [6, 15, 16]. First, we adjusted for possible maternal confounders (correction model A) including maternal age (< 20 years, ≥40 years and continuous), non-White ethnicity, low socioeconomic status (SES), and, in multiparous women, a prior preterm birth. Additional analysis were performed to adjust for potentially mediating factors occurring in the pathway between the independent (parity) and dependent (spontaneous preterm birth) variables (correction model B). Correction model B included the maternal confounders as in model A and in addition artificial reproductive techniques (ART), male fetal gender, hypertension, preeclampsia and small for gestational age (SGA) < p10. All variables were extracted from PERINED, including SES which was based on the 4 digit postal code of the woman’s home address. SES was divided into low (< 25%), middle (25–75%) and high (> 75%) status.

In multiparous women (P1 through P4), we used the Cochran-Armitage Trend Test to test for a trend in parity on the incidence of spontaneous preterm birth < 37, < 32 and < 28 weeks (and between 34 and 37 weeks). The data were analyzed with the SAS statistical software package (version 9.3; SAS Institute Inc., Cary, NC).

The proportion of pregnancies per parity plus the maternal and pregnancy characteristics of the parity groups are presented in Table 1. The mean maternal age increased with higher parity from 29.25 years in P0 to 35.03 years in P4 (p < .0001). The percentage of non-white ethnicity increased with higher parity, 17.3% non-White in P0 compared to 39.8% in P4 (p < .0001). Also the percentage of women with a low SES increased with higher parity, 24.9% low-SES in P0 increasing to 34.0% in P4 (p < .0001). Hypertension and preeclampsia occurred more often in nulliparous women while these rates remained relatively stable in multiparous women (Table 1).

The overall incidence of preterm birth < 37 weeks of gestation among singletons without congenital anomalies was 5.4% in The Netherlands during the 5 year study period. Rates of spontaneous and iatrogenic preterm birth < 37 weeks of gestation were 3.8 and 1.7%, respectively (Table 1). The incidence of total, spontaneous and iatrogenic preterm birth stratified for parity are presented in Fig. 1 and Table 1. The highest incidence of spontaneous preterm birth was observed among nulliparous women (P0, 4.9%) and women in their fifth pregnancy (P4, 3.7%) (Fig. 1 and Table 1). In addition, among the 18,170 women in their second, third, fourth or fifth pregnancy who had a preterm birth, 8% (n = 1446 out of 18,170 women) had a prior preterm birth and 92% were new preterm births (Table 1). These percentages per parity were 8.0% for P1, 7.9% for P2, 8.1% for P3 and 7.2% for P4 (Table 1).

Spontaneous preterm birth risks by gestational age were examined for parity. We used women in their second pregnancy (P1) as reference; the results are demonstrated in Table 2 and Fig. 2.

Both nulliparous women and women in their fifth pregnancy had the highest risk for all preterm birth outcomes. Preterm birth risk in nulliparous women slightly increased after adjusting for confounders compared to the unadjusted risk, whereas in women in their fifth pregnancy the risk slightly decreased after correcting for the same confounders (Table 2 and Fig. 2).

For spontaneous preterm birth < 37 weeks we observed the highest risk in nulliparous women (OR 1.95, 95% CI 1.89–2.00) and women in their fifth pregnancy (OR 1.26, 95 CI 1.13–1.41) (Fig. 2a). For spontaneous preterm birth < 32 weeks, nulliparous women had the highest risk (OR 2.15, 95% CI 1.98–2.3) followed by women in their fifth pregnancy (OR 1.72, 95% CI 1.31–2.25) (Fig. 2b). Although we observed an increased risk for spontaneous preterm birth < 32 weeks in women in their fourth pregnancy, no effect was seen after adjusting for confounders (OR 1.15, 95% CI 0.95–1.40). The risk for spontaneous preterm birth < 28 weeks was highest in women in their fifth pregnancy (OR 2.44, 95% CI 1.73–3.45), followed by nulliparous women (OR 2.02, 95% CI 1.78–2.29) and women in their fourth pregnancy (OR 1.38, 95% CI 1.05–1.81) (Fig. 2c). Women in their first pregnancy had the highest risk for a spontaneous preterm birth between 34 and 37 weeks of gestation (OR 1.85, 95% CI 1.79–2.29) followed by women in their fifth pregnancy (OR 1.17, 95% CI 1.03–1.33) (Fig. 2d). We did not observe significant differences between results obtained from model A and model B in all parity groups and outcomes (Table 2).

After exclusion of nulliparous women, we observed an increase in incidence of spontaneous preterm birth < 37 weeks (p = 0.0178), < 32 weeks (p < .0001) and < 28 weeks (p < .0001), with increasing parity in multiparous women (Table 3). This trend was in line with our observation of increasing odds ratio’s in multiparous women for all three outcomes. No trend was seen in spontaneous preterm birth between 34 and 37 weeks (Table 3).

Multiple epidemiologic studies have reported associations of nulliparous women with increased risk of preterm birth [9, 18, 22‐24]. Yet, in many of these studies, parity has been categorized as nulliparous and multiparous, with women with their second pregnancy often grouped in with those of higher-order parity. In our study, we evaluated the effect per parity separately which allowed us to identify the increased risk in both nulliparous women and women with higher parity.

Unfortunately, due to low reporting within the perinatal database, we were not able to correct for smoking during pregnancy and maternal body mass index (BMI) in our analyses. The general incidence of smoking in The Netherlands is 22.4% in the population > 18 years old, 19.2% of all women are smokers [25]. The incidence of smoking in pregnant women in The Netherlands is 7.4% [26]. The general incidence of obesity in The Netherlands is 50.2% in the population > 18 years, of all women 47.2% has obesity (30.4% has moderate obesity and 16.9 has severe obesity) [27]. Smoking and very low or very high maternal BMI are known risk factors for spontaneous preterm birth [28, 29]. This may have influenced our results. Because we corrected for low socio-economic status in our analyses, and it is known that low socio-economic status is strongly correlated to both smoking and maternal obesity, we do not think that this issue of missing adjustment factors has influenced our results to a large degree. In addition to smoking and BMI, we were not able to correct for other potential risk factors that contribute to the risk of preterm birth, such as polyhydramnios, intra-uterine infection, single marital status, short interpregnancy interval (< 6 months) and specific maternal diseases (uterus anomaly, cervical excision procedures, maternal surgery during pregnancy, depression) [6].

Pregnancies ending < 22 weeks were not included in our national database. Although we corrected for a prior preterm birth, which was available in our dataset, we did not have information on multiple occurrence nor severity of the prior preterm birth. Because no longitudinal linked obstetric database was available, pregnancies could not be related to the level of the individual woman in this study. We therefore could not identify women that were included multiple times due to multiple pregnancies between 2010 and 2014 which may have influenced our results.