To assess the influence of inadequate birth spacing on birth timing distribution across gestation.
To assess the influence of inadequate birth spacing on birth timing distribution across gestation.
Population-based retrospective cohort study using vital statistics birth records.
Singleton, non-anomalous live births ≥20 weeks to multiparous mothers, 2006–2011.
Birth frequency at each gestational week was compared following short IPIs of <6, 6–12 and 12–18 months versus referent group, normal IPI ≥18 months.
Frequency of birth at each gestational week; preterm <37 weeks; <39 and ≥40 weeks.
Of 454 716 births, 87% followed a normal IPI ≥18 months, 10.7% had IPI 12–18 months and 2.2% with IPI <12 months. The risk of delivery <39 weeks was higher following short IPI <12 months, adjOR (odds ratio) 2.78 (95% CI 2.64, 2.93). 53.3% of women delivered before the 39th week after IPI <12 months compared with 37.5% of women with normal IPI, P < 0.001. Likewise, birth at ≥40 weeks was decreased (16.9%) following short IPI <12 months compared to normal IPI, 23.2%, adjOR 0.67 (95% CI 0.64, 0.71). This resulted in a shift of the frequency distribution curve of birth by week of gestation to the left for pregnancies following a short IPI <12 months and 12–18 months compared to, birth spacing ≥18 months.
While short IPI is a known risk factor for preterm birth, our data show that inadequate birth spacing is associated with decreased gestational age for all births. Pregnancies following short IPIs have a higher frequency of birth at all weeks of gestation prior to 39 and fewer births ≥40 weeks, resulting in overall shortened pregnancy duration.
Inadequate birth spacing, or short interpregnancy intervals (IPI), has been associated with an increased risk of preterm birth. This influence has been reported to be more significant in high risk women. We have previously published findings from vital statistics data reporting the increased risk of early preterm birth <35 weeks with short IPIs, with the risk highest among women with a prior preterm birth.
We have also published several studies regarding the influence of other factors on preterm birth risk, such as black race. In our prior analyses, we found that black race not only increases the risk of preterm birth but is also associated with a reduced likelihood of post-term birth. In essence, the influence of black race on birth timing is a shift of the birth timing frequency distribution curve to the left with more births occurring at all earlier weeks of gestation in black compared with white mothers.
The objective of this study is to describe the influence of short, interpregnancy intervals on birth timing distribution. Based on observations from these previously published studies, we hypothesise that short IPIs influence the frequency of birth timing across the duration of pregnancy, at all gestational ages. We believe that short IPIs not only increase the risk of preterm birth but also reduce the likelihood of post-term birth, and shift the frequency distribution curve of birth timing to the left, resulting in an overall shorter gestational length. We suspect that the influence of short IPI may be more pronounced where there are the shortest intervals between pregnancies.
We performed a population-based retrospective cohort study of all live births in the state of Ohio during a 6-year period (2006–2011) using vital statistics birth records from the Ohio Department of Health.
The primary exposure variable for this study was short interpregnancy interval (IPI). We defined short IPI as time from the immediate preceding birth to subsequent birth of the index pregnancy of <6 months, 6 to <12 months, and 12 to <18 months. [Correction added on 20 August 2014 after first online publication: in the second sentence, the description of the method for calculation of interpregnancy interval was corrected.] We assessed the frequency of birth at each week of gestational age in women with short IPIs compared with a referent group of normal IPI comprising births to women following an IPI ≥18 months. Our analyses were limited to singleton live births occurring at ≥20 weeks with a recorded interpregnancy interval. Few pregnancies (n = 507, 0.1%) had a short interpregnancy interval <6 months, which did not allow for meaningful analysis of this subgroup. Therefore, short IPI was redefined into two groups as <12 and 12 to <18 months for the purposes of all analyses reported in this study.
The primary outcome was the frequency of birth at each week of gestational age for each short IPI group compared with the referent group of IPI ≥18 months. Frequency distribution curves displaying the cumulative percent of births delivered by each week of gestational age and the rate of birth at each week of gestational age for each IPI group were generated. Secondary outcomes were frequency and risk of preterm delivery <37 weeks, birth prior to 39 weeks, and birth ≥40 weeks of gestation. Baseline maternal demographic, behavioural, socioeconomic, prenatal and delivery characteristics were compared between IPI groups. The frequency of maternal characteristics and birth timing outcomes was also stratified by maternal race and compared between births to black versus white mothers.
The ‘best obstetric estimate’ of gestational age (OWGEST), defined as the estimate of gestational age determined by all perinatal factors and assessments such as ultrasound, was used for gestational age comparisons in this study. The best obstetric estimate (OWGEST) variable has been reported to be the most accurate estimate of gestational age from Ohio birth records.
The total number of non-anomalous live births in the state during the study period was 892 733. We excluded multiple gestations (n = 32 282), births <20 weeks (n = 565) and >44 weeks (n = 39), and births to women with missing age (n = 566) or erroneous appearing maternal age ≥55 years (n = 11). There were 342 243 (38.3%) births to primiparous mothers and 46 775 (5.2%) births to multiparous mothers with missing information on interpregnancy interval that were also excluded, as well as 14 584 births (1.6%) with missing information on parity. An additional 952 births (0.1%) were excluded due to missing data on the main covariates used in adjusted models. Analyses were then limited to 454 716 births to multiparous mothers with a recorded interpregnancy interval, representing 51% of the initial study cohort. There was minimal missing data, 2% or less, for pregnancy characteristics listed in Table 1 and outcomes of interest including gestation age at delivery, gestational hypertension, gestational diabetes, small for gestational age, and mode of delivery. Body mass index (BMI) and number of prenatal care visits had 10% missing data. Information regarding the specific details of the immediate preceding birth such as gestational age at birth, live birth or stillbirth, was not available in the data source used for this study. However, whether the mother ever had a prior preterm birth or pregnancy loss (miscarriage or stillbirth) is reported and is shown in Table 1.
|Short IPI <12 months n = 9808 (%)||Short IPI 12 to <18, months n = 48 788 (%)||Referent IPI, ≥18 months n = 396 120 (%)|
|Age, years||25.5 (5.5)||26.0 (5.3)||29.0 (5.4)|
|White||71.5 (7015)||76.5 (37 310)||80.6 (319 122)|
|Black||25.6 (2510)||20.5 (10 022)||15.9 (62 944)|
|Social behaviours & socioeconomic factors|
|Married||44.4 (4358)||54.0 (26 327)||64.9 (256 971)|
|≤High school education||61.6 (6047)||56.0 (14 536)||41.2 (102 666)|
|Medicaid||56.5 (5541)||48.5 (23 656)||35.3 (139 710)|
|Private insurance||28.2 (2762)||34.2 (16 685)||49.8 (197 227)|
|Tobacco use||27.1 (2667)||23.5 (11 483)||18.8 (74 396)|
|Limited (≤5 visits)||18.5 (1813)||14.9 (7261)||8.0 (31 865)|
|Parity (median [IQR])||3 (2, 4)||3 (2, 4)||3 (2, 4)|
|Prior preterm birth||8.7 (848)||6.5 (3 169)||5.2 (20 503)|
|Number of prior pregnancy losses (median [IQR])||0 (0.1)||0 (0, 1)||0 (0, 1)|
|Prior pregnancy loss||27.5 (2697)||27.7 (13 523)||32.5 (128 897)|
|Pre-pregnancy BMI||27.2 (6.9)||26.9 (6.8)||26.6 (6.8)|
|Obese (BMI ≥ 30)||29.2 (2479)||27.0 (11 610)||25.1 (89 116)|
Comparisons of dichotomous variables were performed with chi-square test and continuous variables were compared using ANOVA. Multivariate logistic regression estimated the adjusted odds for dichotomous birth timing comparisons associated with short interpregnancy intervals after adjustment for cigarette smoking, maternal age, race and prior preterm birth. Covariates included in the final regression model were selected based on differences noted in univariate comparisons, biologic plausibility was based on prior publications regarding preterm birth risk factors and parsimony within the model. The previous preterm birth was not adjusted in the primary model because it may be considered an intermediate covariate on the causal pathway. Adjusting for prior preterm birth may therefore attenuate true effect sizes by over-adjusting and drive the results closer to the null. We performed a second adjusted analysis including prior preterm birth as a covariate to demonstrate its effect on the association between short IPI and birth timing.
Analyses were performed using SAS version 9.3, SAS Institute Inc. Cary, NC, USA. Comparisons were considered statistically significant if the probability value was <0.05 or the 95% confidence interval did not include the null value 1.0.
There were 454 716 singleton live births included in this analysis. The reference group of births following a normal IPI (≥18 months) constituted 87.1% of births to multiparous mothers during the study period (n = 396 120). The remaining births had shorter IPIs: 12–18 months (n = 48 788, 10.7%), and <12 months was 2.2% (n = 9808). Preterm birth <37 weeks constituted 8.2% of all births in the study cohort, with 1.1% of births occurring at <32 weeks and 0.6% at 20–28 weeks.
Births included in this analysis primarily comprised two racial groups: 78% were white, 19% were black, and only 3% were other races. Black mothers more frequently had short IPIs compared with non-black mothers, <12 months IPI (3.3 versus, 1.9%, P < 0.01), and 12 to <18 months IPI (13.2 versus 10.1%, P < 0.01). Likewise, the rate of preterm birth was higher in black mothers with a short IPI of <12 months (26.4 versus 8.7%) and 12 to <18 months (12.2 versus 9.3%) compared with non-black mothers, P-values <0.01. Women with optimal IPI of ≥18 months had the lowest rates of preterm birth; however, black women had more preterm births (11.3%) than non-black (6.8%), despite optimal birth spacing, P < 0.01. Other maternal characteristics associated with short IPIs were low education level, limited prenatal care, tobacco use, and prior preterm birth (Table 1). Maternal characteristics of multiparous women with a recorded IPI and those with missing data on IPI were compared. Multiparous women with missing data on IPI tended to be of lower socioeconomic status (more with Medicaid insurance, higher number with less than high school education), had fewer prenatal visits and more were of black race. Cases with missing data on IPI only constituted 5.2% of the total source population. Women who had short birth spacing were not more likely to have experienced a prior pregnancy loss than were women with a normal IPI. Mothers delivering a singleton newborn following a short IPI were more likely to have pregnancy complications of gestational diabetes, gestational hypertension, and small for gestational age (defined as birthweight <10th percentile for gestational age at birth) (Table 2).
|Short IPI <12 months n = 9808 (%)||Short IPI 12 to <18 months n = 48 788 (%)||Referent IPI ≥18 months n = 396 120 (%)|
|Gestational hypertension||3.1 (304)||2.6 (1269)||3.2 (12 676)|
|Gestational diabetes||5.0 (493)||4.3 (2106)||5.7 (22 682)|
|SGA||9.8 (961)||8.9 (4342)||7.7 (30 501)|
|Route of delivery|
|Vaginal||74.3 (7287)||73.1 (35 664)||70.7 (280 057)|
|Caesarean||25.6 (2507)||26.7 (13 048)||29.1 (115 373)|
|Gestational age at birth, median (IQR)||38 (37, 39)||39 (38, 39)||39 (38, 39)|
The frequency of birth at each week of gestational age <39 weeks was higher in women who had a short IPI (<12 months) than women with normal birth spacing. Following a short IPI <12 months, 53.3% of women had delivered before the 39th week of pregnancy compared with 37.5% of women with a normal IPI, P < 0.001. However, the most frequent gestational age of birth was the same for all three IPI groups, 39 weeks: 29.8% of births occurred during the 39th week for <12 month IPI, 21.7% for 12 to <18 month IPI, and 39.3% of births after normal IPI ≥18 months delivered during the 39th week of gestation, P < 0.001. Birth after the estimated due date ≥40 weeks occurred less often in women with short IPI <12 months (16.9%), P < 0.001 and 12 to <18 months (21.8%), P < 0.37 compared with births following a normal IPI (23.1%). This resulted in a shift of the distribution curve of cumulative births by week of gestation to the left for pregnancies following a short IPI <12 months compared with pregnancies with birth spacing ≥18 months (Figure 1). Figure 2 demonstrates the frequency of birth at each week of gestational age with <12 month IPI compared with ≥18 month IPIs. Similarly, the distribution curves of births to women with short IPIs 12–18 months was shifted to the left compared with those with longer IPIs, but to a lesser degree than that of shorter IPIs <12 months (Figure not shown). Post-term births at 42 weeks and beyond were uncommon in all three IPI groups, constituting <0.5% of births overall.
The rate of preterm birth (PTB) <37 weeks was higher in women with short IPI <12 (20.1%) and 12 to <18 months (10.2%), compared with those with an optimal IPI ≥18 months (7.7%), P < 0.001. The risk of PTB <37 weeks for short IPI of <12 months was increased; adjOR (odd ratio) 2.78 (95% CI 2.64, 2.93) as was the risk with short IPI of 12–18 months, adjOR 1.32 (95% CI 1.27, 1.36), even after adjustment for important coexisting risk factors for preterm birth. The likelihood of delivery after the estimated due date, delivery ≥40 weeks of gestation, was lower for women with a short IPI <12 months, adjOR 0.67 (95% CI 0.64, 0.71); and 12–18 months, adjOR 0.91 (95% CI 0.89, 0.93), compared with births following an IPI of >18 months. Addition of prior preterm birth to the model had a minimal influence on the observed effects (Table 3).
|Short IPI <12 months n = 9808||Short IPI 12 to <18 months n = 48 788||Referent IPI ≥18 months n = 396 120|
|Delivery <37 weeks, n (%)||1969 (20.1)||4998 (10.2)||30 405 (7.7)|
|Crude OR (95% CI)||3.02 (2.87, 3.18)||1.37 (1.33, 1.42)||Referent|
|Adjusted OR (95% CI)a||2.78 (2.64, 2.93)||1.32 (1.28, 1.36)||Referent|
|Adjusted OR (95% CI)b||2.68 (2.54, 2.82)||1.29 (1.25, 1.33)||Referent|
|Delivery ≥ 40 weeks, n (%)||1661 (16.9)||10 639 (21.8)||91 831 (23.2)|
|Crude OR (95% CI)||0.68 (0.64, 0.71)||0.92 (0.90, 0.95)||Referent|
|Adjusted OR (95% CI)a||0.67 (0.64, 0.71)||0.91 (0.89, 0.93)||Referent|
|Adjusted OR (95% CI)b||0.69 (0.65, 0.73)||0.92 (0.90, 0.94)||Referent|
Inadequate birth spacing with short interpregnancy intervals is an important risk factor for preterm birth. The overall rate of preterm birth in the USA is 11.7% but is substantially higher in women with inadequate birth spacing.[1, 9] Short interpregnancy intervals are also associated with a variety of other adverse pregnancy outcomes, including uterine rupture with trial of labour after caesarean, birth defects, childhood behavioural conditions, and even maternal death. Despite the knowledge of pregnancy risks attributable to inadequate birth spacing, over one third (35%) of pregnancies occur <18 months following a preceding birth, with a preponderance of those short IPIs in women with other high risk factors for preterm birth. In fact, one of the highest risk groups for preterm birth, women of black race, more frequently have insufficient birth spacing, which further increases their risk of delivering prematurely. Due to the increased risk of perinatal complications with inadequate birth spacing, the Healthy People 2020 objectives call for a 10% reduction in the frequency of pregnancies that occur within 18 months of a previous birth.
Prior studies have reported an increase risk of preterm birth following short birth spacing;[1, 9, 14-17]; however, we are unaware of any prior published studies describing its influence on birth timing at early term, full term and post-term gestational ages. Early term births at 37 and 38 weeks are known also to negatively influence infant health, with the best newborn outcomes occurring when birth occurs at 39 completed weeks of gestation.[18, 19] We hypothesised that due to a variety of nutritional and inflammatory stressors following inadequate birth spacing, short IPIs would result in more births at all preterm and early term gestational ages and fewer at full term and beyond 40 weeks of gestation. In this large population-based cohort study, we found that pregnancies following short interpregnancy intervals are shorter overall, resulting in a higher frequency of birth at all gestational ages prior to 39 weeks and lower frequency of births at 40 weeks and beyond, essentially shifting the frequency distribution curve to the left (Figures 1 and 2). We found that the gestational age with the peak frequency of births was the same for each group, 39 weeks, regardless of IPI duration: 29.8% for IPI <12 months, 36.3% IPI 12 to <18 months, and 39.3% following the referent IPI ≥18 months.
There are a number of limitations of studies using vital statistics data for research. Some data variables including maternal demographic information and gestational age at birth appear to be very reliably recorded in the USA birth certificate.[6, 20] However, data on pregnancy complications and medical comorbid conditions are under-reported, which may limit the ability to adjust thoroughly for confounding risk factors for prematurity. In addition, currently there is no consistent approach for reporting the indication for birth on the USA birth certificate and we were therefore unable to stratify preterm births into spontaneous versus medically indicated. Although the outcome variable for this study (gestational age at birth) may be reliably recorded with a relatively narrow margin of error, the accuracy of recorded birth interval may be missing or possibly inaccurately reported, depending on the method used to ascertain the information. Because data on IPI entered on the US birth certificate is obtained by birth certificate registrars either from review of the medical record or from personal recall from the patient, if this data is not easily available it may be missing and under-reported on the birth certificate. Other studies using unique maternal identifying information directly to link more than one birth to the same mother may have more complete data on birth interval.[1, 22] This study used a data set without maternal identifying information, and a linking number was not available for this purpose. IPI for this study was classified as recorded directly on the birth certificate of the current birth as ‘interval’, which is calculated from the date of the last live birth. This variable does not take into account the time since a previous miscarriage or stillbirth, although the birth interval following those outcomes may not have the same significance of risk as short IPI following a live birth. Therefore, use of the US birth certificate variable ‘interval’ for this study may have resulted in a lower number of pregnancies available for analysis with a known IPI. However, those with a recorded IPI were likely accurately recorded, as they would have been generated by direct extraction from the medical record or recall of the patient, neither of which is likely to be differentially misclassified. There were some births with missing data on IPI, and the characteristics of those births did differ somewhat from those with a recorded IPI. Multiparous women with missing data on IPI tended to be of lower socioeconomic status (more with Medicaid insurance, higher number with less than high school education), had fewer prenatal visits and more were of black race. But because, those with missing data on IPI only constituted 5.2% of the total source population, we feel that these differences should not impact the generalisability of our main findings.
Several hypotheses have been proposed to explain how short birth intervals lead to increased risk of pregnancy complications. The nutritional depletion hypothesis proposes that inadequate birth spacing does not allow the mother sufficient time to replete her nutrient stores prior to her next pregnancy, putting the subsequent pregnancy at risk.[23-26] Folate deficiency, in particular, has been proposed as a contributing factor in the aetiology of pregnancy complications as it is an important substrate for DNA synthesis, methylation and repair. Folate and other micronutrients are particularly important during pregnancy, as rapid cell division occurs throughout gestation. Deficiencies in micronutrients needed for cell proliferation, fetal development, and proper function of the placenta may lead to dysfunction of the maternal–fetal interface, leading to either spontaneous preterm birth or complications related to placental dysfunction and subsequent medially indicated preterm birth. These effects may be most robust in pregnancies already at high risk because of other preterm birth risk factors, such as prior preterm birth, underweight, poor maternal weight gain, and chronic medical conditions that may exert a concomitant or synergistic influence with nutrient depletion.
Another proposed hypothesis regarding the aetiologic influence of short IPI on shortened pregnancy duration is related to the intrauterine inflammatory milieu. Histologic inflammation of the placenta has been associated with increased recurrence of inflammation in the subsequent pregnancy. Histologic placental inflammation is a common finding in spontaneous preterm births, even those without clinical evidence of chorioamnionitis.[28, 29] Short IPIs may provide insufficient time for uterine involution and healing of underlying intrauterine inflammation or endometritis. This may predispose the next pregnancy to increased risk of spontaneous preterm birth or a shorter gestational length due to complications leading to indicated preterm birth, i.e. from fetal growth restriction or placental abruption. Regardless of the specific dominant underlying stimulus, the pathophysiologic mechanisms that lead to increased risk of early delivery likely affect some particularly vulnerable subgroups of women to a greater degree than others. This may in part explain the increased frequency of all births earlier than 39 weeks of gestation, not affecting the peak incidence of birth at 39 weeks in the healthiest pregnancies, which are less likely to be significantly influenced by physiologic stressors associated with nutritional depletion and intrauterine inflammation. The lower number of births at gestational ages beyond 39 weeks may be attributed to the higher number delivered prior to 39 weeks, and fewer remaining after the peak at 39 weeks. Alternatively, all births may be influenced less by a shorter pregnancy duration but those shifted from 39 weeks to birth at earlier weeks of gestation are replaced at 39 weeks by those who would otherwise have delivered at a later gestational age.
Short intervals between pregnancies result in decreased pregnancy length with more deliveries at all gestational ages prior to 39 weeks, and fewer at 40 weeks and beyond. The effect of limited birth spacing on shortened pregnancy duration is similar to that found with other important preterm birth risk factors, such as black race. However, this finding has significant potential clinical impact on preterm birth prevention, as, birth spacing is a modifiable risk factor. All women should be counselled on the importance of optimal birth spacing of 18 or more months, with particular attention paid to those pregnancies with concomitant risk factors for shortened gestational length. Improvements in optimal birth spacing could result in overall reduction in preterm birth across the world, especially when focused on high risk women in whom short interpregnancy intervals occur most frequently.[12, 30]
None of the authors has a conflict of interest.
ED developed the original idea for the study, coordinated the analysis plan, and wrote the manuscript. SE managed the data set including data cleaning and preparation, performed data analyses and assisted with interpretation of study findings. LM assisted with study design and interpretation of findings. SE and LM contributed to writing and revising the manuscript.
A protocol for this study was approved and a de-identified data set provided by the state of Ohio, USA, Department of Health. This study was exempt from review by the Institutional Review Board at the University of Cincinnati, Cincinnati, Ohio, USA.
This work was supported by the Perinatal Institute, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA, and March of Dimes Grant 22-FY13-543 for the March of Dimes Prematurity Research Center Ohio Collaborative.
All of the analyses, interpretations and conclusions that were derived from the data source and included in this article are those of the authors and not the Ohio Department of Health. Access to de-identified Ohio birth certificate data was provided by the Ohio Department of Health.