Secondary Logo

Journal Logo

Contents: Multiple Gestations: Original Research

Delayed Childbearing as a Growing, Previously Unrecognized Contributor to the National Plural Birth Excess

Adashi, Eli Y. MD, MS; Gutman, Roee PhD

Author Information
doi: 10.1097/AOG.0000000000002853

Delayed childbearing, a sociodemographic shift rooted in the latter half of the 20th century, has had a profound effect on contemporary reproduction.1 Commonly defined as pregnancy occurring in women 35 years and older, delayed childbearing has been the subject of growing scholarship.2,3 A recent report of the National Center of Health Statistics documented a ninefold increase in the number of first births to women 35 years and older over the 1972–2012 interval.4 More recent estimates of age-specific birth rates proved confirmatory.5 Corollaries such as infertility and involuntary childlessness remain a source of concern.1 Increased rates of fetal aneuploidy, fetal loss, ectopic gestation, preterm births, and stillbirths have been similarly emphasized.1–3 Increased maternal mortality rates have also been entertained.3

Delayed childbearing has also been established as a risk factor for unassisted (multizygotic) plural births.1–3 The precise mechanism(s) undergirding this phenomenon remains under study.6–14 First reported in 1865, the observation that “twin-bearing increased with the mother's age” has since been extended to higher order plural births.6–14 Note has also been made of the fact that age-dependent birth plurality is more pronounced in black than in white women.7–10 Delayed childbearing thus could play a role in the current national plural birth excess, the primary iatrogenic driver of which is assisted reproduction.15 This possibility has yet to be carefully considered. This communication seeks to establish the annual and the cumulative multiyear contribution of delayed childbearing to the national plural birth excess.

MATERIALS AND METHODS

Approved by the institutional review board of Brown University, this study made use of publicly available national birth data reported by the National Center of Health Statistics.16 No national birth data were reported by National Center of Health Statistics for calendar years 1969 and 1970.16 Rates of twin and of higher order (triplet or greater) plural births were expressed per 1,000 live births in keeping with National Center of Health Statistics convention.17 Maternal age categories (younger than 15, 15–19, 20–24, 25–29, 30–34, 35–39, 40–44, and 45–49 years of age) followed standard National Center of Health Statistics groupings.17 Maternal race designation complied with National Center of Health Statistics convention.16,17 Estimates of intervals are reported as CIs. Assisted reproduction is defined as the practice of ovulation induction, ovulation enhancement, and in vitro fertilization.15 The national plural birth excess represents the sum of the assisted (iatrogenic) and unassisted (delayed childbearing-related) contributions to the national plural birth rates above and beyond the expected unassisted plural birth rate for calendar year 1971. All analyses were performed using the R 3.3.2, Meta, and MGCV software packages. Simulated predictions were generated by the JAGS software (Plummer M. JAGS: a program for analysis of Bayesian graphical models using Gibbs sampling. In: Hornik K, Leisch F, Zeileis A, editors. Proceedings of the 3rd International Workshop on Distributed Statistical Computing (DSC 2003). Vienna, Austria; 2003). A relevant Appendix 1, available online at http://links.lww.com/AOG/B145, has also been provided.

Unassisted plural birth rates by maternal age at birth and maternal race were informed by a national birth cohort (n=71,570,717) corresponding to an 18-year interval (1949–1966) predating the assisted reproduction era. The onset of the assisted reproduction era is defined herein by the 1967 approval of the ovulation-inducing drug clomiphene citrate by the U.S. Food and Drug Administration.18 The total national birth cohort comprised 1,436,134 twin and 19,154 higher order plural births. The white and black maternal race categories accounted for 99% of the total national birth complement. Age- and race-adjusted plural birth rates and the relative risk of unassisted twin, higher order, and total plural births were estimated using a multilevel logistic regression model, which accounted for heterogeneity across years.19

Delayed childbearing trend characterization of the total national birth cohort relied on the variation of maternal age at birth by calendar year for the younger than 30, 30–40, and older than 40 years age categories of white and black women. A comparable analysis was carried out for the plural national birth complement (Appendices 2 and 3, available online at http://links.lww.com/AOG/B145). Delayed childbearing trends were defined by declining plural births in the younger than 30 years of age category coupled with increments in 30–40 years and older than 40 years counterparts. Multinomial logistic regression with the reference category being younger than 30 years was used to examine linear trend (on the logit scale) of maternal age at birth over calendar years.20

Plural birth rate estimates by maternal age and race at birth were informed by a national birth cohort (n=166,817,655) corresponding to a 45-year interval (1971–2016) coincident with the assisted reproduction era.15 The total national birth cohort comprised 4,346,832 twin and 174,821 higher order plural births. The white and black race categories accounted for 89% or greater of the annual total birth complement. The remaining race categories, Native American and Alaska Natives and Asian and Pacific Islanders were not subjected to further analysis in that their limited representation from 1949 to 1966 precluded the derivation of unassisted plural birth risk estimates.

Using the 1949–1966 model, we derived a Bayesian multilevel logistic model to estimate the unassisted plural birth rate by age for black and white women. Estimates of the unassisted national plural birth rates from 1971 to 2016 were derived using the aforementioned logistic model with the observed annual age distribution for white and black women from 1971 to 2016. The expected national plural birth rates are defined as the estimated unassisted national plural birth rates in 1971 for white and black women.

The national plural birth excess rate is defined as the difference between the observed plural birth rate and the expected unassisted plural birth rate. The national plural birth excess attributable to delayed childbearing is defined as the difference between the age-adjusted unassisted plural birth rate and the expected unassisted plural birth rate. The national plural birth excess attributable to assisted reproduction is defined as the difference between the observed plural birth rate and the plural birth rate attributable to delayed childbearing.

Simulated projection through 2025 assumes that trends observed from 1971 to 2016 will have proceeded apace. These trends comprise the age- and race-specific live births and the observed national plural birth rates (Appendix 1, http://links.lww.com/AOG/B145).

RESULTS

The relative risk of unassisted plural births by maternal age before the assisted reproduction era (1949–1966) proved unimodal and race-dependent (Fig. 1). The age-dependent risk increments for unassisted plural births proved more pronounced for black women than for their white counterparts. The risk of unassisted plural births for 35- to 9-year-old black women proved 2.75-fold higher than that of 15- to 19-year-old counterparts (39.81/1,000 to 14.48/1,000; P<.001; 95% CI 2.67–2.83). A 2.47-fold risk increment was estimated for comparably aged white women (28.76/1,000 to 11.63/1,000; P<.001; 95% CI 2.43–2.52). The risk of unassisted plural births for 45- to 49-year-old black women proved 1.39-fold higher than that noted for 15- to 19-year-old counterparts (P<.001; 95% CI 1.16–1.66). A 1.07-fold risk increment was estimated for comparably aged white women (P=.2; CI 0.97–1.18). Qualitatively comparable trends were obtained for unassisted twin and higher order plural births (Table 1).

Fig. 1.
Fig. 1.:
The association of maternal age at birth and of race with the unassisted plural birth rate in the total national birth cohort (n=71,570,717) corresponding to the 1949–1966 interval. White women (A), black women (B).Adashi. Delayed Childbearing and Plural Birth Excess. Obstet Gynecol 2018.
Table 1.
Table 1.:
The Effect of Maternal Age at Birth and of Race on the Relative Risk of Unassisted Twin and Higher Order Plural Births in the Total National Birth Cohort (n=71,570,717) Corresponding to the 1949–1966 Interval

The variation of maternal age at birth by calendar year before the assisted reproduction era (1949–1966) proved statistically significant but of limited clinical significance for both white and black women (Fig. 2). The odds ratios (ORs) per 1-year difference of white women being 30–40 years or older than 40 years as compared with younger than 30 years of age at birth were 0.98 and 0.997 (P<.001), respectively. The corresponding ORs for black women were 0.999 and 1.003 (P<.001), respectively. The proportion of white women younger than 30 years of age at birth increased from 73% to 77%. The corresponding proportion of black women decreased from 80% to 79%. The relative representation of white women 30–40 years of age at birth decreased from 24% to 20%. The corresponding share of black women increased from 18% to 19%. The proportion of white and black women older than 40 years of age at birth stayed constant at 2%. Qualitatively comparable results were obtained for the plural national birth complement (Appendix 2, http://links.lww.com/AOG/B145).

Fig. 2.
Fig. 2.:
The observed distribution pattern of maternal age at birth by calendar year for white (A) and black (B) women in the total national birth cohort (n=71,570,717) corresponding to the 1949–1966 interval. The maternal age at birth categories include younger than 30, 30–40, and older than 40 years of age.Adashi. Delayed Childbearing and Plural Birth Excess. Obstet Gynecol 2018.

The variation of maternal age at birth by calendar year during the assisted reproduction era (1971–2016) was consistent with increasingly prominent delayed childbearing trends for both white and black women (Fig. 3). The OR for 1-year increment of white women being 30–40 years or older than 40 years as compared with younger than 30 years of age at birth was 1.03 and 1.05 (P<.001), respectively. The corresponding ORs for black women were 1.03 (P<.001) and 1.04 (P<.001), respectively. The proportion of white women younger than 30 years of age at birth decreased from 83% to 54%. A more modest decline from 85% to 65% was apparent for comparably aged black women. The fraction of white women 30–40 years of age at birth increased from 16% to 43%. A more limited increase from 14% to 33% was apparent in comparably aged black women. The representation of white women older than 40 years of age at birth increased from 1.2% to 3.0%. An increase from 1.4% to 2.8% was apparent for comparably aged black women. Qualitatively comparable results were obtained for the plural national birth complement (Appendix 3, http://links.lww.com/AOG/B145).

Fig. 3.
Fig. 3.:
The observed distribution pattern of maternal age at birth by calendar year for white (A) and black (B) women in the total national birth cohort (n=166,817,655) corresponding to the 1971–2016 interval. The maternal age at birth categories include younger than 30, 30–40, and older than 40 years of age.Adashi. Delayed Childbearing and Plural Birth Excess. Obstet Gynecol 2018.

The contribution of delayed childbearing to the national plural birth excess during the assisted reproduction era (1971–2016) grew steadily at an annual rate of 0.76 (95% CI 0.65–0.86) and 1.21 (95% CI 1.08–1.35) per 10,000 live births for white and black women, respectively (Fig. 4). In so doing, delayed childbearing gave rise to an estimated cumulative total of 255,964 (95% CI 134,746–375,581) and 66,271 (95% CI 34,099–96,197) unassisted plural births by white and black women, respectively. In 2016 alone, delayed childbearing accounted for 24% (95% CI 15–32%) and 38% (95% CI 28–47%) of the national plural birth excess for white and black women, respectively.

Fig. 4.
Fig. 4.:
The expected, age-adjusted, and observed national plural birth rates for white (A) and black (B) women in the total national birth cohort (n=166,817,655) corresponding to the 1971–2016 interval. The national plural birth excess comprises an unassisted component attributable to delayed childbearing and an assisted component attributable to assisted reproduction.Adashi. Delayed Childbearing and Plural Birth Excess. Obstet Gynecol 2018.

Simulated projection revealed that by 2025 delayed childbearing could account for 46% (95% CI 32–60%) and 40% (95% CI 30–53%) of the national plural birth excess for white and black women, respectively (Fig. 5). The average annual contribution of delayed childbearing to the national plural birth excess could grow at a rate of 1.39 (95% CI 0.03–2.76) and 2.97 (95% CI 1.10–4.84) per 10,000 live births for white and black women, respectively. In so doing, delayed childbearing could give rise to an estimated cumulative total of 112,419 (95% CI 87,806–136,280) and 44,358 (95% CI 37,201–51,350) of unassisted plural births by white and black women, respectively.

Fig. 5.
Fig. 5.:
The expected, age-adjusted, and observed national plural birth rates for white (A) and black (B) women in the total national birth cohort (n=166,817,655) corresponding to the 1971–2016 interval herein extended forward using simulation through 2025. The national plural birth excess comprises an unassisted component attributable to delayed childbearing and an assisted component attributable to assisted reproduction.Adashi. Delayed Childbearing and Plural Birth Excess. Obstet Gynecol 2018.

DISCUSSION

Delayed childbearing, a major social and demographic trend, has upended traditional reproductive patterns.1–4 Late to be noted, delayed childrearing received little mention in the peer-reviewed literature until 1984.21 As noted by Baldwin and Nord,21 “the proportion of first births to women 25 and older rose from 19% to 36%” between 1970 and 1982. The representation of “women still childless at ages 25–34 increased by 56% at about the same time.”21 Now the subject of significant scholarly attention, delayed childbearing is being studied for its demographic and medical consequences.1–3 We reevaluate the risk association between advancing maternal age and unassisted (naturally occurring) plural births with an eye toward estimating the annual as well as cumulative contribution of delayed childbearing to the national plural birth excess.22

The national plural birth excess, largely the product of five decades of fertility promotion, dates back to the 1967 approval of the first ovulation-inducing drug (clomiphene citrate) by the U.S. Food and Drug Administration.18 The introduction of in vitro fertilization in 1981 and the advent of ovulation enhancement since 1987 further contributed to what came to be known as the “multiple gestation epidemic.”23–25 All told, the national twin and higher order plural birth rates increased 1.9- and 6.7-fold, respectively, over the 50 years in question.17 The national higher order plural birth rate declined by as much as 48% since its 1998 high as a result of the modification of in vitro fertilization practice patterns.17 This trend, however, arrested, for the first time, in 2016.17 The increased national twin birth rate has yet to reverse course.17,26

Our observations suggest that delayed childbearing constitutes a growing previously unrecognized contributor to the national plural birth excess. Over a 45-year interval (1971–2016) coincident with the assisted reproduction era, the contribution of delayed childbearing to the national plural birth excess grew annually at a rate of 0.76 and 1.21 per 10,000 live births for white and black women, respectively. In so doing, delayed childbearing gave rise to an estimated cumulative total of 255,964 and 66,271 unassisted plural births by white and black women, respectively. In 2016 alone, delayed childbearing accounted for 24% and 38% of the national plural birth excess for white and black women, respectively. Simulated projection suggests that by 2025, delayed childbearing could account for 46% and 40% of the national plural birth excess for white and black women, respectively. The projected contributions of delayed childbearing (24–46%) and assisted reproduction (76–54%) to the national plural birth excess of white women displayed opposite directionality as a result of the projected decline in the observed national plural birth rate. Only limited differences were noted between the projected contributions of delayed childbearing (38–40%) and assisted reproduction (62–60%) to the national plural birth excess of black women in whom the observed national plural birth rate is projected to increase.

A notable strength of this study is its reliance on large birth cohorts corresponding to time intervals antecedent to (1949–1966) and coincident with (1971–2016) the assisted reproduction era.18,21,23,24 Whereas the former permitted the unencumbered estimation of the age- and race-related risk of unassisted plural birth rates, the latter highlighted the contribution of delayed childbearing to the national plural birth excess. Estimates derived from the former informed the latter. Indeed, estimates of unassisted plural birth rates from 1949 to 1966 informed the expected and age-adjusted distribution of unassisted plural birth rates from 1971 to 2016.

A limitation of this study is the absence of variables other than age and race previously implicated in unassisted birth plurality. Failure to adjust for these variables could have influenced estimates reported here. Special note is made of the absence of parity, the declining trend of which could have led to overestimation of unassisted plural birth rates.12 The absence of heredity (eg, family history) and birth order data is also worthy of note.12 Other missing variables of potential import include maternal height and maternal body mass index.27 The unfeasibility of correcting for coincident decrements in perinatal and maternal mortality resulting from health care improvements and for delayed births attributable to assisted reproduction must also be acknowledged.28 The inability to distinguish unassisted multizygotic plural births from their (age-independent) monozygotic counterparts constitutes yet another shortcoming.7–11 One final limitation is rooted in the presumption that the age at birth trends from 1971 to 2016 will persist through 2025 for simulation purposes. This assumption could well prove erroneous.

The growing previously unrecognized contribution of delayed childbearing to the national plural birth excess requires that it be given greater consideration in future clinical, analytic, and policy deliberations as well as in individual family planning decisions. Thus far, analysis of the national plural birth excess focused primarily on the contribution of assisted reproduction.29 Increased awareness of the concurrent growing effect of delayed childbearing appears warranted.

REFERENCES

1. Schmidt L, Sobotka T, Bentzen JG, Nyboe Andersen A; ESHRE Reproduction and Society Task Force. Demographic and medical consequences of the postponement of parenthood. Hum Reprod Update 2012;18:29–43.
2. Johnson JA, Tough S; SOGC Genetics Committee. Delayed child-bearing. J Obstet Gynaecol Can 2012;34:80–93.
3. Berkowitz GS, Skovron ML, Lapinski RH, Berkowitz RL. Delayed childbearing and the outcome of pregnancy. N Engl J Med 1990;322:659–64.
4. Mathews TJ, Hamilton BE. Mean age of mothers is on the rise: United States, 2000–2014. NCHS Data Brief 2016:1–8.
5. Centers for Disease Control and Prevention, National Center for Health Statistics. Vital Statistics of the United States. Available at: https://www.cdc.gov/nchs/products/vsus.htm. Retrieved February 14, 2018.
6. Duncan JM. On the comparative frequency of twin-bearing in different pregnancies. Edinb Med J 1865;10:928–9.
7. Yerushalmy JS, Sheerar SE. Studies on twins: I. The relation of order of birth and age of parents to the frequency of like-sexed and unlike-sexed twin deliveries. Hum Biol 1940;12:95–113.
8. McArthur N. The frequency of monovular and binovular twin births in Italy, 1949–1950. Acta Genet Med Gemellol (Roma) 1953;2:11–8.
9. Bulmer MG. The effect of parental age, parity and duration of marriage on the twinning rate. Ann Hum Genet 1959;23:454–8.
10. Shipley PW, Wray JA, Hechter HH, Arellano MG, Borhant NO. Frequency of twinning in California. Its relationship to maternal age, parity and race. Am J Epidemiol 1967;85:147–56.
11. Myrianthopoulos NC. An epidemiologic survey of twins in a large, prospectively studied population. Am J Hum Genet 1970;22:611–29.
12. Bortolus R, Parazzini F, Chatenoud L, Benzi G, Bianchi MM, Marini A. The epidemiology of multiple births. Hum Reprod Update 1999;5:179–87.
13. Lambalk CB, De Koning CH, Braat DD. The endocrinology of dizygotic twinning in the human. Mol Cell Endocrinol 1998;145:97–102.
14. Beemsterboer SN, Homburg R, Gorter NA, Schats R, Hompes PG, Lambalk CB. The paradox of declining fertility but increasing twinning rates with advancing maternal age. Hum Reprod 2006;21:1531–2.
15. Kulkarni AD, Jamieson DJ, Jones HW Jr, Kissin DM, Gallo MF, Macaluso M, et al. Fertility treatments and multiple births in the United States. N Engl J Med 2013;369:2218–25.
16. Centers for Disease Control and Prevention, National Center for Health Statistics. Vital statistics rapid release. Quarterly provisional estimates. Natality dash board. Age-specific birth rates. 2017. Available at https://www.cdc.gov/nchs/products/vsrr/natality-dashboard.htm#. Retrieved February 14, 2018.
17. Martin JA, Hamilton BE, Osterman MJ, Driscoll AK, Drake P. Births: final data for 2016. Natl Vital Stat Rep 2018;67:1–55.
18. U.S. Food and Drug Administration. Drugs@FDA. FDA approved drug products. Clomid (clomiphene citrate). 1967. Available at: http://www.accessdata.fda.gov/scripts/cder/drugsatfda/index.cfm?fuseaction=Search.DrugDetails. Retrieved February 14, 2018.
19. DerSimonian R, Laird N. Meta-analysis in clinical trials revisited. Contemp Clin Trials 2015;45:139–45.
20. Agresti A. Categorical data analysis. 2nd ed. Hoboken (NJ): John Wiley & Sons; 2002.
21. Baldwin WH, Nord CW. Delayed childbearing in the U.S.: facts and fictions. Popul Bull 1984;39:1–42.
22. Enders T, Stern C. The frequencies of twins, relative to age of mothers, in American populations. Genetics 1948;33:263–72.
23. Jones HW Jr, Jones GS, Andrews MC, Acosta A, Bundren C, Garcia J, et al. The program for in vitro fertilization at Norfolk. Fertil Steril 1982;38:14–21.
24. Dodson WC, Whitesides DB, Hughes CL Jr, Easley HA III, Haney AF. Superovulation with intrauterine insemination in the treatment of infertility: a possible alternative to gamete intrafallopian transfer and in vitro fertilization. Fertil Steril 1987;48:441–5.
25. Adashi EY, Barri PN, Berkowitz R, Braude P, Bryan E, Carr J, et al. Infertility therapy-associated multiple pregnancies (births): an ongoing epidemic. Reprod Biomed Online 2003;7:515–42.
26. Adashi EY. Seeing double: a nation of twins from sea to shining sea. Am J Obstet Gynecol 2016;214:311–3.
27. Hoekstra C, Willemsen G, van Beijsterveldt CE, Lambalk CB, Montgomery GW, Boomsma DI. Body composition, smoking, and spontaneous dizygotic twinning. Fertil Steril 2010;93:885–93.
28. Barfield WD. Improving systems in perinatal care: quality, not quantity. JAMA 2012;307:1750–1.
29. Practice Committee of American Society for Reproductive Medicine. Multiple gestation associated with infertility therapy: an American Society for Reproductive Medicine Practice Committee opinion. Fertil Steril 2012;97:825–34.

Supplemental Digital Content

© 2018 by the American College of Obstetricians and Gynecologists. Published by Wolters Kluwer Health, Inc. All rights reserved.