Abstract
Background:
Multifetal gestation could be associated with higher long-term maternal mortality because it increases the risk of pregnancy complications such as preeclampsia and preterm birth, which are in turn linked to postpartum cardiovascular risk.
Objectives:
We examined whether spontaneously conceived multifetal versus singleton gestation was associated with long-term maternal mortality in a racially diverse U.S. cohort.
Methods:
We ascertained vital status as of 2016 via linkage to the National Death Index and Social Security Death Master File of 44,174 mothers from the Collaborative Perinatal Project (CPP; 1959-66). Cox proportional hazards models with maternal age as the time scale assessed associations between history of spontaneous multifetal gestation (in last CPP observed pregnancy or prior pregnancy) and all-cause and cardiovascular mortality, adjusted for demographics, smoking status, and preexisting medical conditions. We calculated hazard ratios (HR) for all-cause and cause-specific mortality over the study period and until age 50, 60, and 70 (premature mortality).
Results:
Of eligible participants, 1,672 (3.8%) had a history of multifetal gestation. Participants with versus without a history of multifetal gestation were older, more likely to have a preexisting condition, and more likely to smoke. By 2016, 51% of participants with and 38% of participants without a history of multifetal gestation had died (unadjusted all-cause HR=1.14, 95% confidence interval (CI): 1.07, 1.23). After adjustment for smoking and preexisting conditions, history of multifetal gestation was not associated with all-cause (adjusted HR=1.00, 95% CI: 0.93, 1.08) or cardiovascular mortality (adjusted HR=0.99, 95% CI: 0.87, 1.11) over the study period. However, history of multifetal gestation was associated with an 11% lower risk of premature all-cause mortality (adjusted HR=0.89, 95% CI: 0.82, 0.96).
Conclusions:
In a cohort with over 50 years of follow-up, history of multifetal gestation was not associated with all-cause mortality, but may be associated with lower risk of premature mortality.
Keywords: multiple pregnancy, maternal mortality, premature mortality
Background
Multiple gestations (twins, triplets, and higher order) comprise approximately 3.3% of all yearly births in the United States.1 Compared to singletons, multifetal gestations may impose additional maternal risks, including greater cardiovascular demands, higher rates of nutritional deficiencies, and elevated risk of postpartum hemorrhage.2–4 Multifetal pregnancies are also more likely than singleton pregnancies to be affected by preeclampsia, preterm birth, and other obstetric complications,5,6 which have been linked to elevated risk for poor long-term maternal cardiovascular health.7–10 On the other hand, successful delivery of spontaneous dizygotic twin pregnancy has been hypothesized to reflect a slight maternal fitness advantage (e.g., reflected in higher fertility),11–13 in which case spontaneous twin pregnancy may indicate an underlying protective phenotype. The long-term maternal health effects of multifetal gestation have received limited attention to date.
Five prior epidemiologic studies of long-term maternal postpartum cardiovascular health suggest that carrying a multifetal gestation does not confer additional risk compared to a singleton pregnancy.14–18 However, the prior literature incompletely describes the potential relationship between multifetal gestation and long-term maternal health risks. For example, inference from the prior studies was hindered by incomplete outcome ascertainment (i.e., <5% of study participants, and in some cases <1%, developing outcomes of interest): most participants were in their 50s or younger at the end of follow-up in four of these studies,14–17 and the fifth cohort restricted to mortality before age 70.18 The previous studies also focused largely or entirely on non-Hispanic White populations, leading to questionable generalizability to the US context, which is characterized by greater racial and ethnic diversity. Finally, all prior studies focused on cardiovascular outcomes (including mortality), capturing only some of the potential health effects of multifetal gestation, which may additionally include risk of breast, endometrial, and cervical cancers.19–21
We used the Collaborative Perinatal Project (CPP) cohort, a historical racially diverse cohort of spontaneously conceived pregnancies from 1959 to 1966, to evaluate the extent to which multifetal versus singleton gestation history was associated with all-cause and cause-specific mortality during approximately 50 years of follow-up.
Methods
The CPP was a prospective multisite pregnancy cohort comprised of 48,197 pregnant participants recruited during their first prenatal visit in 1959-1966 at 12 study centers located in 10 U.S. states. At enrollment, participants had a physical examination and an interview with trained study personnel at which they provided sociodemographic, medical, and obstetrical history information. After enrollment, participants were followed across pregnancy and through labor and delivery. Participants could contribute multiple pregnancies to the CPP cohort (n=59,391 total pregnancies). Because the cohort was recruited before the advent of assisted reproductive therapies, all pregnancies were spontaneously conceived. Full details of the cohort have been previously described.22–24
As part of the CPP Mortality Linkage Study, maternal participants in the CPP cohort were linked to recorded vital status as of 2016. To determine vital status, death records were used from the 2017 National Death Index (NDI) and Social Security Administration Death Master File (SSDMF).25 The NDI is a nationwide registry of all deaths and causes of death based on the International Classification of Diseases (ICD), 9th and 10th revisions26,27 that uses a probabilistic matching algorithm to link death reports to individuals. NDI registrations began in 1979 and continue to the present. We additionally used the SSDMF (1962-2016), which records deaths but not causes of death. The SSDMF matches deaths to individuals using social security numbers or names and dates of birth. Because the SSDMF records began in 1962, using this database in addition to the NDI ensured that deaths between 1962 and 1979 were captured. Only 3% of deaths occurred before 1979.25
We excluded participants who died during delivery or in the immediate postpartum period (n=46) and participants whose vital status we could not ascertain (n=1,723),25 leaving 46,428 participants for analysis. The CPP began in the 1950s before the advent of institutional review boards, but maternal participants gave general informed consent when recruited. Institutional review board approval for the mortality linkage study was obtained from the Eunice Kennedy Shriver National Institute of Child Health and Human Development and the Emmes Corporation, which performed abstraction from the CPP records and facilitated the linkages.
Exposure
The primary exposure was a history of multifetal gestation (i.e., having had a twin, triplet, or quadruplet gestation in the current pregnancy or in a previous pregnancy). We defined the history of multifetal gestation as a binary variable (ever versus never) based on the woman’s last CPP pregnancy and self-reported obstetric history.
Outcomes
The primary outcome was all-cause mortality. We additionally analyzed specific causes of death when sample size permitted (n > 10 in analyses for participants with and without history of multifetal gestation): cancer, cardiovascular disease, diabetes, respiratory disease, infection, Alzheimer’s disease and dementia, and kidney disease (definitions for each in this cohort have been published previously).28 The remaining causes of death, including suicide and accidents, were grouped into the “Other” category. Participants without a recorded death by December 31, 2016 were considered to be alive. To maintain participant deidentification, date of death was rounded to the nearest year.
Covariates
Covariates included self-identified race (White, African American, Puerto Rican, Other), marital status (married or common-law married, single, other), family income (≤$1999, $2000-3999 [reference category], $4000-5999, $6000-7999, $8000-9999, ≥$10000/year), maternal pre-pregnancy body mass index (BMI; kg/m2, continuous), parity at the time of the last CPP pregnancy (nulliparous, primiparous, multiparous), smoking status (nonsmoker, <1 pack/day, ≥1 pack/day), study site (Boston, Buffalo, New Orleans, NY/Columbia, Baltimore, Virginia, Minnesota, NY/New York Medical College, Oregon, Pennsylvania, Providence, Tennessee), pregestational diabetes at the time of the last CPP pregnancy (no, yes), and chronic hypertension at the time of the last CPP pregnancy (no, yes) (eFigure S1). For categorical covariates, the reference category in regression models is listed first unless otherwise noted.
Statistical Analysis
For all analyses, we used Cox proportional hazards with maternal age as the time scale to calculate adjusted hazard ratios (HR) and 95% confidence intervals (CI). Follow-up for each woman began at her age of delivery of the last CPP pregnancy. We performed an unadjusted analysis and an adjusted analysis, which was adjusted for pre-pregnancy BMI, race/ethnicity, study site, smoking, marital status, income, parity, pregestational diabetes, and chronic hypertension. Covariates were selected based on CPP design factors and literature indicating that that each is causally related to risk of multifetal gestation, earlier death, or both.29–33
Due to the long follow-up time, many participants were elderly by 2016. To focus on premature mortality, we also calculated the hazard of all-cause mortality before age 50, 60, and 70 years (in three separate models). To limit residual confounding by differences in parity (i.e., because each pregnancy provides another opportunity for a multifetal gestation, history of multifetal gestation is strongly associated with parity), we repeated analyses restricted to multiparous participants (70% of participants). Finally, we stratified analyses by race (White, African American) to evaluate variation by self-identified racial group membership. Analyses were conducted using SAS version 9.4 (SAS Institute, Inc., Cary, North Carolina).
Missing data
Data were missing in 0% to 15.5% of covariates (Table 1). We imputed missing covariate, exposure, and cause of death data using multivariable imputation by chained equations (MICE) in 10 datasets. Effect estimates from imputed datasets were combined using Rubin’s rules.34 We imputed the cause of death for participants whose vital status was recorded as deceased but who were missing cause of death (n=1979). However, we did not impute missing values of vital status. Of participants with missing vital status but recorded obstetric history, 3.1% had a history of multifetal gestation.
Table 1.
Demographic characteristics of the study population by history of multifetal gestation before imputation, Collaborative Perinatal Project (n=46,428)
| All participantsa | No history of multifetal gestation | History of multifetal gestation | |
|---|---|---|---|
|
|
|||
| n = 46428 | n = 42502 | n = 1672 | |
| Years of follow-up, median (IQR) | 52 (45, 54) | 52 (45, 54) | 51 (39, 54) |
| Age (years), median (IQR) | 23 (20, 28) | 23 (20, 28) | 28 (24, 33) |
| Race/ethnicity, n (%) | |||
| White | 21472 (46.3) | 19366 (45.6) | 711 (42.5) |
| African American | 21037 (45.3) | 19523 (45.9) | 859 (51.4) |
| Puerto Rican | 3651 (7.9) | 3547 (8.4) | 78 (4.7) |
| Other | 268 (0.6) | 66 (0.2) | 24 (1.4) |
| Parity, n (%) | |||
| 0 | 13473 (29.8) | 12748 (30.2) | 124 (7.5) |
| 1 | 10290 (22.8) | 9785 (23.2) | 186 (11.2) |
| ≥2 | 21476 (47.5) | 19743 (46.7) | 1353 (81.4) |
| Missing | 1189 | 226 | 9 |
| Marital status, n (%) | |||
| Single | 10324 (24.0) | 10051 (24.1) | 259 (21.6) |
| Married/common-law | 32708 (76.0) | 31619 (75.9) | 942 (78.4) |
| Missing | 3396 | 832 | 471 |
| Pre-pregnancy BMI (kg/m2), median (IQR) | 21.8 (19.9, 24.6) | 21.8 (19.9, 24.5) | 23.32 (20.7, 27.3) |
| Missing | 7192 | 4481 | 596 |
| Smoking status, n (%) | |||
| None | 23966 (53.5) | 22464 (53.6) | 815 (49.5) |
| <1 pack per day | 14147 (31.6) | 13243 (31.6) | 511 (31.1) |
| ≥ 1 pack per day | 6722 (15.0) | 6221 (14.8) | 319 (19.4) |
| Missing | 1593 | 574 | 27 |
| Income, n (%) | |||
| ≤ $1999 | 5927 (14.2) | 5598 (14.3) | 222 (14.4) |
| $2000-3999 | 18429 (44.2) | 17268 (44.0) | 740 (47.9) |
| $4000-5999 | 10173 (24.4) | 9538 (24.3) | 376 (24.4) |
| $6000-7999 | 4440 (10.7) | 4209 (10.7) | 133 (8.6) |
| $8000-9999 | 1583 (3.8) | 1522 (3.9) | 42 (2.7) |
| ≥ $10000 | 1127 (2.7) | 1081 (2.8) | 31 (2.0) |
| Missing | 4749 | 3286 | 128 |
| Education, n (%) | |||
| < High school | 8054 (18.3) | 7455 (18.0) | 410 (25.4) |
| Some high school | 17286 (39.3) | 16223 (39.3) | 659 (40.9) |
| High school graduate | 13239 (30.1) | 12522 (30.3) | 424 (26.3) |
| ≥ Some College | 5442 (12.4) | 5117 (12.4) | 424 (26.3) |
| Missing | 2407 | 1185 | 60 |
| Chronic hypertension, n (%) | 1774 (3.9) | 1635 (3.9) | 118 (7.1) |
| Missing | 636 | 16 | 0 |
| Diabetes mellitus, n (%) | 737 (1.6) | 696 (1.6) | 34 (2.0) |
| Missing | 643 | 20 | 0 |
Exposure columns do not sum to the ‘All participants’ column due to missing values of obstetric history in the unimputed data (n=2,254 (4.9%))
Results
Of 44,174 participants with information about a history of multifetal gestation, 1,672 (3.8%) had ever had a multifetal gestation. Compared to participants with no history of multifetal gestation, participants with a history of multifetal gestation were older (median 28 versus 23 years), more likely to identify as African American (51.4% versus 45.9%), less likely to be nulliparous (7.5% versus 30.2%), more likely to have at least some college education (26.3% versus 12.4%), more likely to smoke ≥1 pack of cigarettes per day (19.4% versus 14.8%), and more likely to have chronic hypertension (7.1% versus 3.9%) (Table 1). Participants with and without a history of multifetal gestation had similar marital status, income, and prevalence of pregestational diabetes mellitus. Imputation did not affect the distribution of covariates in the analytic population (eTable S1).
By the end of follow-up, 51.6% of participants with a history of multifetal gestation and 38.9% of participants without a history of multifetal gestation had died. However, in the adjusted models, participants with a history of multifetal gestation did not have an increased risk for all-cause mortality over the study period (HR = 1.00, 95% CI: 0.93, 1.08), compared to participants without a history of multifetal gestation (Table 2). Similarly, history of multifetal gestation was not associated with increased risk of any specific causes of death during follow-up, including death from cardiovascular causes (HR: 0.99, 95% CI: 0.87, 1.11), diabetes (HR: 0.99, 95% CI: 0.64, 1.23), or cancer (HR: 1.00, 95% CI: 0.93, 1.08), though the risk of death from respiratory causes was elevated among participants with a history of multifetal gestation versus no history (HR: 1.24, 95% CI: 0.98, 1.58) (Table 2). Effect estimates from all-cause and cause-specific mortality models restricted to multiparous participants (eTable S2) and effect estimates from models stratified by race (eTable S3) were similar to estimates from the primary analysis in the full population.
Table 2.
Hazard of all-cause and cause-specific mortality over the full study period, for participants with a history of multifetal gestation (n=1672) compared to participants without a history of multifetal gestation (n=42502)
| Causes of death | No history of multifetal gestation | History of multifetal gestation | Hazard ratios | |||
|---|---|---|---|---|---|---|
| % of deaths | Median (IQR) follow-up time (years) | % of deaths | Median (IQR) follow-up time (years) | Unadjusted HR (95%CI) | Adjusted HR (95%CI) | |
| All causes | 38.5 | 41.0 (31.0, 47.0) | 51.1 | 38.0 (29.0, 45.0) | 1.14 (1.07, 1.23) | 1.00 (0.93, 1.08) |
| Cancer | 12.2 | 41.0 (32.0, 48.0) | 15.2 | 40.0 (30.5, 48.0) | 1.13 (0.99, 1.28) | 1.05 (0.92, 1.19) |
| Cardiovascular | 11.6 | 40.0 (32.0 47.0) | 16.6 | 38.0 (32.0, 44.0) | 1.20 (1.07, 1.35) | 0.99 (0.87, 1.11) |
| Diabetes | 1.8 | 44.0 (36.0, 50.0) | 2.6 | 42.0 (35.0, 48.0) | 1.27 (0.92, 1.74) | 0.89 (0.64, 1.23) |
| Respiratory | 2.8 | 40.0 (31.0, 47.0) | 4.7 | 40.0 (31.0, 45.0) | 1.43 (1.13, 1.82) | 1.24 (0.98, 1.58) |
| Infection | 1.4 | 50.0 (47.0, 52.0) | 1.6 | 51.0 (48.0, 53.0) | 1.04 (0.67, 1.63) | 0.89 (0.57, 1.41) |
| Dementia | 1.3 | 44.0 (38.0, 49.0) | 1.8 | 44.0 (36.0, 48.0) | 0.81 (0.56, 1.18) | 0.82 (0.56, 1.19) |
| Kidney Disease | 1.0 | 39.0 (26.0, 47.0) | 1.1 | 39.0 (28.0, 47.0) | 1.14 (0.70, 1.86) | 0.96 (0.59, 1.57) |
| Other | 6.5 | 41.0 (31.0, 47.0) | 7.4 | 38.0 (29.0, 45.0) | 1.02 (0.85, 1.23) | 0.94 (0.77, 1.13) |
Adjusted model is adjusted for race/ethnicity, site, pre-pregnancy BMI, smoking, marital status, income, parity, pregestational diabetes, chronic hypertension. Effect estimates are based on imputed data. Median follow-up time is calculated among individuals who died of each cause.
CI: confidence interval; HR: hazard ratio; IQR: interquartile range
In models of premature all-cause mortality, 4.6% of participants with history of multifetal gestation and 5.2% of participants without a history of multifetal gestation had died by age 50; 12.6% of participants with history of multifetal gestation and 12.1% of participants without a history of multifetal gestation had died by age 60; and 26.3% of participants with history of multifetal gestation and 23.6% of participants without a history of multifetal gestation had died by age 70 years. In adjusted models of death by age 50, participants with versus without a history of multifetal gestation had somewhat elevated hazard of all-cause mortality (HR: 1.15, 95% CI: 0.97, 1.36). In adjusted models of death by age 60, participants with and without a history of multifetal gestation had similar hazard of all-cause mortality (HR: 1.01, 95% CI: 0.92, 1.11). However, participants with versus without a history of multifetal gestation had a 11% lower risk of all-cause mortality by age 70 years (HR: 0.89, 95%CI: 0.82, 0.96)) (Table 3). Effect estimates from models of all-cause mortality by age 50, by age 60, and by age 70 years restricted to multiparous participants were similar to effect estimates from models in the full population (eTable S4).
Table 3.
Hazard of all-cause mortality by age 50, 60, and 70 years (3 models), for participants with a history of multifetal gestation (n=1672) compared to participants without a history of multifetal gestation (n=42502)
| No history of multifetal gestation | History of multifetal gestation | Hazard ratios | ||||
|---|---|---|---|---|---|---|
| % of deaths | Median (IQR) follow-up time (years) | % of deaths | Median (IQR) follow-up time (years) | Unadjusted HR (95%CI) | Adjusted HR (95%CI) | |
| By age 50 | 5.2 | 19.0 (12.0, 25.0) | 4.6 | 18.0 (12.0, 24.0) | 0.90 (0.75, 1.07) | 1.15 (0.97, 1.36) |
| By age 60 | 12.1 | 27.0 (20.0, 34.0) | 12.6 | 26.0 (19.0, 31.0) | 0.81 (0.74, 0.90) | 1.01 (0.92, 1.11) |
| By age 70 | 23.6 | 39.0 (29.0, 49.0) | 26.3 | 32.0 (26.0, 40.0) | 0.74 (0.69, 0.79) | 0.89 (0.82, 0.96) |
Adjusted model is adjusted for race/ethnicity, site, pre-pregnancy BMI, smoking, marital status, income, parity, pregestational diabetes, chronic hypertension. Effect estimates are based on imputed data. Median follow-up time is calculated among all individuals in each time horizon.
CI: confidence interval; HR: hazard ratio; IQR: interquartile range
Comment
Principal findings
In this historical pregnancy cohort of 44,174 pregnant participants, having a history of multifetal gestation was not associated with all-cause or cause-specific mortality over approximately 50 years following the last CPP delivery. History of multifetal gestation was not associated with risk of mortality before age 50 or before age 60 years. However, participants with a history of multifetal gestation had 11% lower risk of premature mortality before age 70 years than participants without a history of multifetal gestation. Findings provide reassurance that, despite a known higher risk of pregnancy complications in multifetal versus singleton gestations, having a spontaneously conceived multifetal gestation does not increase the risk of premature maternal death over the long term after pregnancy.
Strengths of the study
Our analysis had multiple strengths, including excellent confounding control for both clinical and demographic variables and use of a diverse U.S. population that improves generalizability to the modern U.S. context. We were also able to leverage very long-term follow-up through vital status linkage to ascertain all-cause and cause-specific mortality. Because of its historical nature, all pregnancies in the CPP represent conceptions without the use of assisted reproductive technologies; the increased prevalence of assisted reproductive therapy, especially for multifetal pregnancy, complicates investigations of long-term maternal health effects in modern obstetric cohorts. Finally, findings were robust to a sensitivity analysis restricting to multiparous participants.
Limitations of the data
Our analysis also had limitations. We used self-reported obstetric history to determine the history of multifetal gestation, which may be subject to misclassification; however, we expect that participants will accurately report whether they ever had a multifetal pregnancy.35 Similarly, the cohort did not collect details about the fetal number of prior pregnancies beyond “multifetal” or “singleton,” so we could not exclude triplet or quadruplet pregnancies. However, we expect the number of spontaneously conceived triplet and quadruplet pregnancies to be extremely small (of the last CPP pregnancies included in this analysis, 6 were triplet or quadruplet pregnancies, representing ~1% of multifetal births) and their inclusion is unlikely to influence results. In addition, because we did not have details about pregnancy complications in prior pregnancies, we could not test for effect modification by pregnancy complications. Also, we did not have information about pregnancies occurring after the last CPP pregnancy. Some participants without a history of multifetal gestation during CPP may have a subsequent multifetal pregnancy, but this is likely to affect a small number of participants given the rarity of spontaneous multifetal gestation, and this misclassification would likely bias findings towards the null. Finally, we did not have access to health outcomes other than mortality, so we were not able to assess intermediate health-related endpoints that may affect quality of life after multifetal pregnancy.
Interpretation
Our finding of no difference in long-term all-cause mortality risk among participants with and without a history of multifetal pregnancy is broadly consistent with prior studies of cardiovascular mortality after twin pregnancy. The associations of twin versus singleton pregnancy vary across prior studies, with positive, inverse, and null results, though confidence intervals are often wide. For example, Hiersch and colleagues reported a slight (HR: 1.09, 95% CI: 0.92, 1.28) increased risk of all-cause mortality over a median 13 years of follow-up among participants who had a first twin pregnancy without a hypertensive disorder, compared to a first singleton pregnancy without a hypertensive disorder.16 On the other hand, van Baar and colleagues used registry data in the Netherlands to show that a first twin versus first singleton pregnancy was not associated with cardiovascular mortality over a median of 11 years of follow-up (HR: 1.01, 95% CI: 0.68, 1.52),17 though they reported only 24 cardiovascular deaths among women with twin pregnancy, limiting power to make comparisons. Finally, a study using registry data from Norway and comparing to a reference group of women with three singleton pregnancies reported both positive and inverse findings. Specifically, Basnet and colleagues reported that women with a first twin pregnancy and continued reproduction had about 24% lower risk of cardiovascular mortality before age 70 (HR: 0.76, 95% CI: 0.48, 1.19), while women with a twin pregnancy after a singleton pregnancy had 49% higher risk (HR: 1.49, 95% CI: 1.22, 1.81).18 It is difficult to directly compare these prior studies to ours because of varying reference groups; our study also relied on a more diverse U.S. population of pregnant participants. However, taking the findings together, previous research does not suggest a large survival benefit or deficit after multifetal gestation. Our findings therefore build on previous studies to provide further evidence that multifetal gestation does not increase the long-term risk of maternal mortality.
Our finding of reduced risk of all-cause mortality before age 70 for participants with a history of multifetal gestation was unexpected, as twin pregnancy has not been found to confer either a survival disadvantage or survival benefit in prior studies of cardiovascular mortality,17,18 and some evidence supports an increased risk of cancer.19–21 Consistent with our results, two previous studies of cardiovascular morbidity suggest a slight benefit of multifetal gestation: history of twin versus singleton delivery was associated with 15% lower odds of cardiovascular hospitalization over approximately 11 years of follow-up in an Israeli cohort;14 and among Swedish women without preeclampsia, those with a multifetal pregnancy had 17% lower risk of later compound cardiovascular disease compared to women with a singleton pregnancy.15 Unlike most prior research14,16–18 on the long-term maternal cardiovascular health effects of multifetal gestation, we adjusted for smoking status and BMI, which may have allowed us to reduce bias in our estimate of the effect of interest. Prior studies also focused on cardiovascular morbidity and mortality rather than all-cause mortality, which may explain the different pattern of findings; our sample size did not permit investigation of specific causes of death for analyses restricted to mortality before age 70. We did not observe a lower risk of all-cause mortality among participants with a history of multifetal gestation in analyses restricting to deaths before age 50 or 60 years; in fact, mortality before age 50 appeared somewhat elevated for participants with a history of multifetal gestation. These findings may be due to limited power attributable to a smaller number of very early deaths. We were not able to evaluate cause-specific mortality in premature mortality models, so findings should be treated with caution. It is also possible that successful multifetal pregnancy is a marker for access to good medical care or health consciousness (e.g., diet, exercise) in this historic cohort, which we could not control for in the CPP data. Future research should investigate potential reduced risk of premature mortality associated with multifetal gestation to rule out the possibility of a spurious finding and elucidate mechanisms, and to ascertain the extent to which mortality risks observed in this historic cohort are also seen in a more contemporary pregnancy cohort reflecting current U.S. population-level risk and protective factors (e.g., lower parity, lower smoking rates, modern obstetric care and improved treatments for maternal cardiometabolic diseases developing after pregnancy).
Evolutionary biology and anthropology researchers have hypothesized that a human maternal fitness benefit may be associated with spontaneous multifetal pregnancy.11–13 Historical demographic data in the Utah Population Database, representing a very high-parity population of women born in the 1800s, suggests that among women who survived to age 50, those who ever had twins had a postmenopausal survival benefit compared to those who never had twins.11 Similarly, a population in rural Ghana studied between 1950-1974 indicates higher fertility and shorter inter-birth intervals for mothers of twins versus singletons.13 Although these studies may have limited public health relevance in a context with access to modern contraception, obstetric care, and fertility treatment, they suggest a modest biological phenotypic advantage among individuals who conceive without the use of fertility treatments and successfully carry multifetal pregnancies, which may provide a potential mechanism explaining the reduced deleterious effect of hypertensive disorders of pregnancy on long-term maternal health observed in twin versus singleton gestations.15
Unexpectedly, we reported that participants with versus without a history of multifetal gestation had 24% greater risk of death from respiratory causes. This finding may be due to uncontrolled confounding from smoking; consistent with previous reports,32 smoking was associated with a history of multifetal gestation in our population. Though we controlled for smoking status, there may be uncontrolled variation among participants who smoked more than a pack of cigarettes a day during pregnancy (who were 19% of participants with a history of multifetal gestation, versus 15% of participants without a history of multifetal gestation). Adjustment for smoking, chronic hypertension, and pre-pregnancy diabetes (as well as additional demographic factors) strongly attenuated estimates of the hazard of all-cause and cause-specific mortality (including respiratory mortality) associated with multifetal gestation in this cohort. This finding may also be due to chance; to our knowledge, multifetal gestation has not been previously linked to risk of long-term maternal pulmonary disease.
Conclusions
In this large and racially diverse historical cohort, we found that a history of multifetal gestation was not associated with all-cause or cause-specific mortality over more than 50 years of follow-up, though it was associated with a reduced risk of all-cause mortality before age 70. History of multifetal gestation conceived without use of assisted reproductive therapy may either be harmless or be associated with a small maternal long-term survival benefit; mechanisms underlying a potential survival benefit are unknown. Future investigation into mechanisms may be warranted, including describing variation between spontaneous and assisted reproductive therapy multifetal pregnancies.
Supplementary Material
Synopsis.
Study question
We examined whether spontaneously conceived multifetal versus singleton gestation was associated with long-term maternal mortality.
What is already known
Compared to singletons, multifetal gestations may impose additional risks, including higher rates of pregnancy complications. Five prior epidemiologic studies suggest that multifetal gestation does not confer additional long-term maternal mortality risk compared to a singleton pregnancy. However, these previous studies were limited by incomplete outcome ascertainment or racially homogeneous study populations that do not reflect the U.S. context.
What this study adds
In a racially diverse U.S. cohort with over 50 years of follow-up, history of multifetal gestation was not associated with all-cause mortality, but may be associated with lower risk of premature mortality before age 70.
Acknowledgements
This research was supported, in part, by the Division of Population Health Research, Division of Intramural Research, Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD), National Institutes of Health (Contract Numbers: HHSN275200800002I/27500013). RS, EY, JLM, JLG, and KLG have contributed to this work as part of their official duties as employees of the United States Federal Government. SDM was supported, in part, National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) of the National Institutes of Health (K12AR084219). Intramural investigators at NICHD designed and implemented the study. Funding sources approved the manuscript for publication but had no role in the analysis and interpretation of the data.
Katherine Grantz, Yan Qiao and Rajeshwari Sundaram had full access to all the data. Yan Qiao and Rajeshwari Sundaram take responsibility for the integrity of the data and the accuracy of the data analysis. Yan Qiao conducted the data analysis for this manuscript under the supervision of Rajeshwari Sundaram. The revised manuscript received additional programming support from Anokhi Cifuentes, under the supervision of Rajeshwari Sundaram.
References
- 1.National Vital Statistics Reports; Vol 70 no 2 Births: Final Data for 2019 (National Center for Vital Statistics; ) (2021). [PubMed] [Google Scholar]
- 2.Meah VL, Khurana R, Hornberger LK, Steinback CD, Davenport MH. Cardiovascular function during triplet pregnancy. Journal of Applied Physiology. 2021;130:1286–1292. [DOI] [PubMed] [Google Scholar]
- 3.Kametas NA, McAuliffe F, Krampl E, Chambers J, Nicolaides KH. Maternal cardiac function in twin pregnancy. Obstetrics & Gynecology. 2003;102(4):806–815. [DOI] [PubMed] [Google Scholar]
- 4.Young BC, Wylie BJ. Effects of twin gestation on maternal morbidity. Seminars in Perinataology. 2012;36:162–168. [DOI] [PubMed] [Google Scholar]
- 5.Rao A, Sairam S, Shehata H. Obstetric complications of twin pregnancies. Best Practice & Research Clinical Obstetrics & Gynaecology. 2004;18(4):557–576. [DOI] [PubMed] [Google Scholar]
- 6.Wen SWW, Demissie K, Yang Q, Walker MC. Maternal morbidity and obstetric complications in triplet pregnancies and quadruplet and higher-order multiple pregnancies. American Journal of Obstetrics & Gynecology. 2004;191(1):254–258. [DOI] [PubMed] [Google Scholar]
- 7.Mosca L, Benjamin EJ, Berra K, et al. Effectiveness-based guidelines for the prevention of cardiovascular disease in women -- 2011 update. J Am Coll Cardiol. 2012;57(12):1404–1423. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Behrens I, Basit S, Melbye M, et al. Risk of post-pregnancy hypertension in women with a history of hypertensive disorders of pregnancy: nationwide cohort study. BMJ. 2017;358:j3078. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Tanz LJ, Stuart JJ, Williams PL, et al. Preterm delivery and maternal cardiovascular disease in young and middle-aged adult women. Circulation. 2017;135(6):578–589. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Grandi SM, Filion KB, Yoon S, et al. Cardiovascular disease-related morbidity and mortality in women with a history of pregnancy complications. Circulation. 2019;139(8):1069–1079. [DOI] [PubMed] [Google Scholar]
- 11.Robson SL, Smith KR. Twinning in humans: maternal heterogeneity in reproduction and survival. Proceeding of the Royal Society B - Biological Sciences. 2011;278:3755–3761. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Bhalotra S, Clarke D. Twin birth and maternal condition. The Review of Economics and Statistics. 2019;101(5):853–864. [Google Scholar]
- 13.Sear R, Shanley D, McGregor IA, Mace R. The fitness of twin mothers: evidence from rural Gambia. Journal of Evolutionary Biology. 2001;14:433–443. [Google Scholar]
- 14.Okby R, Shoham-Vardi I, Sergienko R, Sheiner E. Twin pregnancy: Is it a risk factor for long-term cardiovascular disease? Journal of Maternal-Fetal and Neonatal Medicine. 2016;29(10):1626–1630. [DOI] [PubMed] [Google Scholar]
- 15.Bergman I, Nordlöf-Callbo P, Wickström AK, et al. Multi-fetal pregnancy, preeclampsia, and long-term cardiovascular disease. Hypertension. 2020;76:167–175. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hiersch L, Ray JG, Barrett J, et al. Maternal cardiovascular disease after twin pregnancies complicated by hypertensive disorders of pregnancy: a population-based cohort study. CMAJ. 2021;20(193):E1448–1458. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.van Baar PM, Welters SM, Ravelli ACJ, de Boer MA, de Groot CJM. Cardiovascular mortality risk a decade after twin and singleton pregnancies complicated by hypertensive disorders of pregnancy. Pregnancy Hypertension. 2022;28:9–14. [DOI] [PubMed] [Google Scholar]
- 18.Basnet P, Skjærven R, Sørbye LM, et al. Long-term cardiovascular mortality in women with twin pregnancies by lifetime reproductive history. Paediatric and Perinatal Epidemiology. 2023;37(1):19–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Neale RE, Purdie DM, Murphy MFG, Mineau GP, Bishop T, Whiteman DC. Twinning and the incidence of breast and gynecological cancers. Cancer Causes & Control. 2004;15(8):829–835. [DOI] [PubMed] [Google Scholar]
- 20.Albrektsen G, Heuch I, Thoresen S, Kvåle G. Twin births, sex of children and maternal risk of endometrial cancer: a cohort study in Norway. Acta Obstet Gynecol Scand. 2008;87(11):1123–1128. [DOI] [PubMed] [Google Scholar]
- 21.Veisi P, Nikouei M, Cheraghi M, Shahgheibi S, Moradi Y. The association between the multiple birth and breast cancer incidence: an update of a systematic review and meta-analysis from 1983-2022. Arch Public Health. 2023;81(1):76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.The Collaborative Perinatal Study of the National Institute of Neurological Diseases and Stroke: The Women and their Pregnancies. W.B. Saunders Company; 1972. [Google Scholar]
- 23.Klebanoff MA. The Collaborative Perinatal Project: a 50-year retrospective. Paediatric and Perinatal Epidemiology. 2008;23:2–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Hardy JB. The Collaborative Perinatal Project: lessons and legacy. Annals of Epidemiology. 2003;13(5):301–311. [DOI] [PubMed] [Google Scholar]
- 25.Pollack AZ, Hinkle SN, Liu D, et al. Vital status ascertainment for a historic diverse cohort of U.S. women. Epidemiology. 2020;31(2):310–316. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (World Health Organization; ) (1992). [PubMed] [Google Scholar]
- 27.Manual of the International Classification of Diseases, Injuries, and Causes of Death, based on recommendations of the Ninth Revision Conference, 1975 (World Health Organization; ) (1977). [Google Scholar]
- 28.Hinkle SN, Schisterman EF, Liu D, et al. Pregnancy complications and long-term mortality in a diverse cohort. Circulation. 2023;147(13):1014–1025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.VanderWeele TJ. Principles of confounder selection. European Journal of Epidemiology. 2019;34:211–219. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Reddy UM, Branum AM, Klebanoff MA. Relationship of maternal body mass index and height to twinning. Obstet Gynecol. 2005;105(3):593–597. [DOI] [PubMed] [Google Scholar]
- 31.Oleszczuk JJ, Cervantes A, Kiely JL, Keith DM, Keith LG. Maternal race/ethnicity and twinning rates in the United States, 1989–1991. J Reprod Med. 2001;46(6):550–557. [PubMed] [Google Scholar]
- 32.Hoekstra C, Willemsen G, van Beijsterveldt CEMT, Lambalk CB, Montgomery GW, Boomsma DI. Body composition, smoking, and spontaneous dizygotic twinning. Fertility and Sterility. 2010;93(3):885–893. [DOI] [PubMed] [Google Scholar]
- 33.Fellman J, Eriksson AW. On the standardisation of the twinning rate. Twin Research. 2002;5(1):19–29. [DOI] [PubMed] [Google Scholar]
- 34.Rubin DB. Multiple imputation after 18+ years. Journal of the American Statistical Association. 1996;91(434):473–489. [Google Scholar]
- 35.Chin HB, Baird DD, McConnaughey DR, Weinberg CR, Wilcox AJ, Jukic AM. Long-term recall of pregnancy-related events. Epidemiology. 2017;28(4):575–579. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.