Objective
Preterm delivery may predict an increased risk of cardiovascular disease in mothers, providing opportunities for prevention. No study had examined whether gestation length within the term period predicts future CVD, and there are few data segregating spontaneous from medically indicated deliveries.
Study Design
We used proportional hazards models to predict CVD death by gestation length, adjusted for age, education, and delivery year among 688,662 women with births from 1967 through 1998 in the Medical Birth Registry of Norway. Mothers were traced in the National Cause of Death Registry through 2009; there were 2324 CVD deaths.
Results
Compared with women who delivered spontaneously at 39–41 weeks’ gestation, women who spontaneously delivered earlier had higher risks of CVD death. Hazard ratios were 1.9 at 22–31 weeks, 2.2 at 32–34 weeks, 1.6 at 35–36 weeks, and 1.4 at 37–38 weeks. Risks were higher among women with medically indicated deliveries (hazard ratio, 4.8 at 22–31 weeks, 2.7 at 32–34 weeks, 4.3 at 35–36 weeks, and 1.6 at 37–38 weeks compared with spontaneous deliveries at 39–41 weeks). Neither spontaneous nor indicated delivery after 41 weeks was associated with CVD mortality. Risks were highest with recurrent preterm pregnancies, and for women who delivered only one child, especially if that delivery was preterm.
Conclusion
Women who deliver spontaneously before 37 weeks had a 2-fold increased risk of CVD mortality compared with women who had delivered after 38 weeks. Even women with spontaneous deliveries at early term (37–38 weeks) had a 41% elevated risk of CVD death compared with women delivering at 39–41 weeks.
Common pregnancy complications, including preeclampsia, preterm delivery, gestational diabetes, and fetal growth restriction, are associated with elevated risks of future cardiovascular disease (CVD) in mothers. These associations may provide new insights into the development of cardiovascular disease and improve our understanding of the determinants of pregnancy complications.
Women who have delivered preterm have approximately twice the risk of CVD mortality. However, no study has examined the extent to which the association of gestation length with CVD varies within the term period, during which the vast majority of deliveries occur. Furthermore, with 2 exceptions, studies have not distinguished spontaneous from medically indicated deliveries. Finally, no study has reported the association of spontaneous preterm delivery with risk of stroke.
We utilized linked Norwegian birth and mortality registries to quantify the extent to which the gestation length of spontaneous and indicated deliveries is associated with maternal risk of CVD mortality up to 4 decades after birth. We hypothesized that length of gestation would be inversely associated with CVD risk, especially for medically indicated deliveries. We examined recurrent preterm delivery, hypothesizing that women with recurrent preterm deliveries would be at higher CVD risk than women with both preterm and term deliveries. Finally, having observed previously that the CVD risk associated with preeclampsia was especially strong when a preeclamptic first birth was not followed by a later birth, we examined whether mortality patterns varied by whether the preterm birth was the last birth.
Materials and Methods
This study was based on the Medical Birth Registry of Norway (MBRN), a registry of mandatory birth notifications completed by the attending clinician for all live births and stillbirths lasting at least 16 weeks in Norway since 1967. Gestational length was recorded as completed weeks from the last menstrual period because ultrasound dating was not recorded before 1999.
We considered women with a first singleton live birth or stillbirth between 1967 and 1998. Births were linked to mothers by the national identification number. For most analyses, a woman’s first pregnancy was the exposure; to examine recurrent preterm birth, we considered first and second births. Of the 751,651 women with first singleton pregnancies, we excluded 46,261 (6.2%) whose records lacked gestational age and 978 (0.1%) missing birthweight. We excluded 1241 women whose first pregnancies lasted less than 22 weeks and 5604 women whose first pregnancies lasted longer than 44 weeks.
To exclude obviously misclassified gestational ages, we used sex-specific birthweight-for-gestational-age Z-scores based on Norwegian standards to exclude 2671 women in which the birthweights were more than 4 SD higher or lower than the appropriate means. Finally, we excluded 6226 women missing maternal education and 8 women whose deaths lacked dates.
Sensitivity analyses including women missing data on education, using missing indicator terms or assigning high or low education, yielded nearly identical results. For the analysis of second births, we applied the previously cited exclusion criteria for missing and implausible gestational lengths to second births. The study population was 688,662 women for first births and 641,362 women for the analysis of first and second births.
Clinicians reported the start of delivery as spontaneous labor, induced labor, or planned cesarean delivery. In this analysis, induced labor and planned cesarean delivery were considered medically indicated delivery. Labor induced after spontaneous rupture of membranes was considered spontaneous. Augmentation of labor was recorded separately from induction of labor and was not used to define delivery type. If amniotomy was listed as a means of labor induction, the delivery was classified as medically indicated; amniotomy in the context of spontaneous start of delivery was classified as spontaneous delivery. If spontaneous labor or rupture of membranes preceded a planned cesarean delivery by less than 8 hours, the delivery was considered spontaneous; if 8 or more hours passed between spontaneous labor and the time of a planned cesarean delivery, the delivery was classified as a planned cesarean delivery and therefore medically indicated.
Preeclampsia was identified by the eighth revision of the International Classification of Disease (ICD), code 637, for births in 1967–1998 and by ICD-10 codes O14–15 for births after 1998. A test of the accuracy of MBRN preeclampsia reports showed good predictive value: 86% of MBRN registry reports were validated by medical record review. Small for gestational age was defined as below the 10th percentile and large for gestational age as above the 90th percentile of birthweight for gestational age by Norwegian standards.
Women were linked to the National Cause of Death Registry through 2009 via the national identification number. We identified CVD deaths by ICD codes (ICD-8 from 1967 to 1985; ICD-9 for 1986–1995; and ICD-10 for 1996–2009). Deaths caused by coronary heart disease (CHD; 410–414 for ICD-8 and ICD-9; I20–I25 for ICD10) and cerebrovascular disease (430–438 ICD-8 and ICD-9; I60–I69 ICD10) were considered as CVD deaths. We examined cardiovascular mortality as a whole as well as CHD and stroke separately. Forty records had both CHD and stroke listed as the cause of death so that the total number of CVD deaths is less than the sum of the CHD and stroke deaths.
Each participant was followed up from her first delivery between Jan. 1, 1967, and Dec. 31, 1998, until CVD death, other fatality, emigration, or Dec. 31, 2009. We used Kaplan Meier curves to estimate the survival time free of CVD death up to 40 years after first birth; the median follow-up was 24.8 years. We used Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (CIs), using years since first birth as the time axis.
Spontaneous delivery at 39–41 completed weeks was the reference group for the detailed analyses of gestational length and delivery type; however, to allow comparisons to less detailed estimates in the literature, where stated, we also provided statistics that use deliveries from 37 to 44 weeks (term deliveries) as the reference group.
We adjusted for year of delivery (continuous), maternal age (continuous), and maternal education (less than high school, completed high school, attended university) at first birth. Multiplicative interaction was tested by modeling a cross-product interaction term between indication (yes/no) and gestational length categories. As a sensitivity analysis, we restricted to women whose births were uncomplicated by preeclampsia, fetal growth restriction, or macrosomia (less than the 10th or greater than the 90th percentile of birthweight for gestational length, respectively).
We also performed sensitivity analyses to account for unmeasured confounding by smoking and obesity, using the bias correction method described by Lash et al. Statistical analyses were performed with SPSS for Windows (version 19.0; SAS Institute, Cary, NC) and bias correction with the web site developed by Lash et al for that purpose.
This study complies with the Declaration of Helsinki and was approved by the review board of the Medical Birth Registry of Norway, the regional ethics committee, and REK Vest (Norway).
Results
The demographic characteristics of the cohort are classified by length of first pregnancy in Table 1 . Six percent of first births were preterm (<37 weeks). Eighty-five percent of first deliveries were spontaneous (85% of term and 81% of preterm births). As expected from trends in clinical practice, medically indicated deliveries were more prevalent in recent time periods and were more likely in the presence of preeclampsia. Women with medically indicated deliveries were slightly older and more educated (especially if the delivery was preterm).
| Characteristic | Spontaneous deliveries | Medically indicated deliveries | ||||||
|---|---|---|---|---|---|---|---|---|
| Weeks of gestation | 22–31 | 32–34 | 35–36 | 37–44 | 22–31 | 32–34 | 35–36 | 37–44 |
| n (%) | 5488 (0.9) | 8534 (1.5) | 19208 (3.3) | 550604 (94.3) | 1648 (1.6) | 2283 (2.2) | 3820 (3.6) | 97077 (92.6) |
| Maternal age, mean (SD) | 23.6 (5.2) | 23.6 (4.9) | 23.9 (4.7) | 23.9 (4.3) | 25.7 (5.4) | 25.7 (5.3) | 25.6 (5.1) | 25.0 (4.9) |
| Year of mother’s birth, mean (SD) | 1956 (9) | 1957 (9) | 1957 (9) | 1957 (9) | 1961 (9) | 1961 (9) | 1960 (9) | 1958 (9) |
| Year of delivery, mean (SD) | 1980 (9) | 1981 (10) | 1982 (10) | 1982 (9) | 1987 (8) | 1987 (9) | 1986 (9) | 1983 (9) |
| Offspring birthweight (g), mean (SD) | 1218 (534) | 2342 (597) | 2847 (535) | 3485 (467) | 1037 (496) | 1967 (632) | 2636 (637) | 3548 (541) |
| Maternal education, % | ||||||||
| Less than high school | 56.4 | 53.3 | 51.2 | 46.4 | 43.6 | 42.8 | 42.2 | 44.0 |
| High school completed | 21.2 | 22.9 | 23.2 | 23.6 | 25.1 | 25.4 | 26.1 | 23.4 |
| Attended college | 22.4 | 23.8 | 25.6 | 30.0 | 31.4 | 31.8 | 31.7 | 32.5 |
| Married, % | 73.4 | 76.0 | 78.9 | 82.3 | 81.7 | 83.4 | 85.1 | 84.2 |
| Preeclamptic, % | 5.5 | 7.0 | 5.3 | 2.2 | 32.0 | 35.9 | 32.5 | 10.7 |
Over a median 25 years of follow-up, there were 2324 CVD deaths, including 1182 deaths from CHD and 1181 deaths from stroke. Figure 1 shows CVD mortality by gestational length through 4 decades after first birth. The cumulative risk of CVD was less than 3% in any group, consistent with the age of the cohort at the end of follow-up (median, 52; interquartile range, 45–59 years). The lowest CVD mortality was among women with spontaneous term deliveries, followed by indicated term, spontaneous preterm, and medically indicated preterm deliveries. The absolute risks of CVD mortality began to diverge within the first decade after birth and grew wider over 4 decades.
Women whose first delivery was preterm had nearly double the risk of cardiovascular mortality compared with women who delivered at term (37–44 weeks) (HR, 1.9; 95% CI, 1.7–2.2). Spontaneous preterm delivery was associated with an HR of 1.7 (95% CI, 1.5–2.0) and medically indicated preterm delivery with an HR of 3.7 (95% CI, 2.9–4.8), both compared with a spontaneous term delivery.
Figure 2 depicts the hazard ratios for CVD mortality for medically indicated and spontaneous first deliveries of varying gestational lengths, adjusted for maternal age, education, and year of delivery. All gestational lengths shorter than 39–41 weeks, including spontaneous term births at 37–38 weeks, were associated with elevated CVD mortality compared with spontaneous delivery at 39–41 weeks. Women with a history of medically indicated preterm delivery had 2.7- to 4.8-fold higher risks of CVD death, whereas women with a history of spontaneous preterm delivery had 1.6- to 2.2-fold higher risks of CVD death compared with women who had delivered spontaneously at 39-41 weeks ( P = .01 for interaction between gestational length categories and indicated delivery status). Deliveries after 41 completed weeks, whether spontaneous or indicated, predicted the same risk of CVD death as spontaneous deliveries at 39-41 weeks. There was no interaction between type of preterm delivery and maternal age at birth (<30 vs ≥30 years) in predicting CVD mortality ( P = .22).
Women who delivered preterm spontaneously had twice the risk of CHD death (HR, 2.1; 95% CI, 1.7–2.5) compared with women who delivered spontaneously at term (37–44 weeks). Table 2 shows the hazard ratios of CHD death by gestational length. Even early term spontaneous deliveries (37–38 weeks) predicted 49% (25–78%) higher CHD mortality compared with later term spontaneous delivery (39–41 weeks) ( Table 2 ). Although the HRs for medically indicated delivery were higher than those for spontaneous delivery, the statistical test of interaction was not significant ( P = .30).
| Variable | Women | Coronary heart disease | Stroke | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Weeks of gestation | Deaths, n (%) | Age-adjusted HR | Multivariate HR a | 95% CI | Deaths, n (%) | Age-adjusted HR | Multivariate HR a | 95% CI | |
| Spontaneous deliveries | |||||||||
| 22-31 | 5488 | 23 (0.42) | 2.5 | 2.3 | 1.5–3.4 | 19 (0.35) | 1.9 | 1.8 | 1.1–2.8 |
| 32-34 | 8534 | 33 (0.39) | 2.5 | 2.4 | 1.7–3.4 | 29 (0.34) | 2.0 | 1.9 | 1.3–2.8 |
| 35-36 | 19,208 | 62 (0.32) | 2.3 | 2.1 | 1.6–2.7 | 41 (0.21) | 1.3 | 1.3 | 0.9–1.7 |
| 37-38 | 71,111 | 155 (0.22) | 1.6 | 1.5 | 1.3–1.8 | 161 (0.23) | 1.5 | 1.4 | 1.2–1.7 |
| 39-41 | 403,028 | 573 (0.14) | 1.0 | 1.0 | Reference | 640 (0.16) | 1.0 | 1.0 | Reference |
| 42-44 | 76,465 | 104 (0.14) | 1.0 | 1.0 | 0.8–1.2 | 116 (0.15) | 1.0 | 1.0 | 0.8–1.2 |
| Medically indicated deliveries | |||||||||
| 22-31 | 1648 | 7 (0.42) | 5.9 | 4.7 | 2.2–9.8 | 9 (0.55) | 5.9 | 5.4 | 2.8–10.4 |
| 32-34 | 2283 | 8 (0.35) | 4.4 | 3.4 | 1.7–6.9 | 5 (0.22) | 2.2 | 1.9 | 0.8–4.7 |
| 35-36 | 3820 | 25 (0.65) | 7.7 | 6.2 | 4.2–9.3 | 13 (0.34) | 3.3 | 2.9 | 1.7–5.1 |
| 37-38 | 12,256 | 28 (0.23) | 2.8 | 2.3 | 1.6–3.4 | 14 (0.11) | 1.1 | 1.0 | 0.6–1.7 |
| 39-41 | 53,094 | 112 (0.21) | 1.8 | 1.6 | 1.3–2.0 | 84 (0.16) | 1.2 | 1.1 | 0.9–1.4 |
| 42-44 | 31,727 | 52 (0.16) | 1.1 | 1.1 | 0.8–1.5 | 50 (0.16) | 1.0 | 1.0 | 0.7–1.3 |
a Adjusted for maternal age at delivery, year of delivery, and maternal education (less than high school, completed high school, attended university).
Associations of gestational length with stroke mortality were less consistent, although generally elevated for both spontaneous and medically indicated preterm births ( Table 2 ). Overall, spontaneous preterm delivery was associated with an HR of 1.5 (95% CI, 1.2–1.8) and medically indicated preterm delivery with an HR of 3.0 (95% CI, 2.0–4.3) compared with spontaneous delivery at 37–41 weeks. Women who delivered spontaneously at early term (37–38 weeks) had a 41% (19–68%) increased risk of stroke mortality compared with women delivering later in the term period (39–41 weeks). However, there was no evidence that medically indicated delivery predicted future stroke risk across the range of term deliveries. The association of gestation length with stroke was stronger for medically indicated than for spontaneous deliveries (interaction P = .01).
Table 3 shows CVD mortality risk by preterm status in first and second births. Compared with women who bore at least 2 term infants, women who bore only 1 child had increased risk of CVD mortality, whether their first and only child was born at term (HR, 2.0; 95% CI, 1.8–2.3) or preterm (HR, 4.2; 95% CI, 3.4–5.1). Women with 2 consecutive preterm pregnancies had a 3-fold increase in CVD risk (HR, 3.3; 95% CI, 2.4–4.5). Women who had 2 deliveries and delivered preterm in only 1 of them had intermediate risks of CVD. The order of the term and preterm births did not seem to matter. There was no apparent interaction between these categories and maternal age at first birth (<30 years vs ≥30 years) in predicting CVD mortality ( P = .19).
