Socioeconomic development predicts a weaker contraceptive effect of breastfeeding
Edited by R. Michael Roberts, University of Missouri, Columbia, MO, and approved May 20, 2021 (received for review December 9, 2020)
Commentary
July 30, 2021
Significance
Breastfeeding suppresses postpartum fecundity (the capacity for reproduction). This mechanism is critically important to control fertility (actual reproduction) when contraception is uncommon. Whether dependence of the contraceptive effect of breastfeeding on environmental conditions is strong enough to have a significant effect on fertility remains unclear. Analyzing 2.7 million births in 84 low- and middle-income countries over the past four decades, we find a dramatic weakening of the breastfeeding–postpartum amenorrhea relationship that correlates with improved living standards. These results suggest that, in the absence of contraception, the effect of breastfeeding on fertility depends on the level of socioeconomic development.
Abstract
The contraceptive effect of breastfeeding remains essential to controlling fertility in many developing regions of the world. The extent to which this negative effect of breastfeeding on ovarian activity is sensitive to ecological conditions, notably maternal energetic status, has remained controversial. We assess the relationship between breastfeeding duration and postpartum amenorrhea (the absence of menstruation following a birth) in 17 World Fertility Surveys and 284 Demographic Health Surveys conducted between 1975 and 2019 in 84 low- and middle-income countries. We then analyze the resumption of menses in women during unsupplemented lactation. We find that a sharp weakening of the breastfeeding–postpartum amenorrhea relationship has globally occurred over the time period analyzed. The slope of the breastfeeding–postpartum amenorrhea relationship is negatively associated with development: higher values of the Human Development Index, urbanization, access to electricity, easier access to water, and education are predictive of a weaker association between breastfeeding and postpartum amenorrhea. Low parity also predicts shorter postpartum amenorrhea. The association between exclusive breastfeeding and maintenance of amenorrhea in the early postpartum period is also found in rapid decline in Asia and in moderate decline in sub-Saharan Africa. These findings indicate that the effect of breastfeeding on ovarian function is partly mediated by external factors that likely include negative maternal energy balance and support the notion that prolonged breastfeeding significantly helps control fertility only under harsh environmental conditions.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Future world population growth and the many challenges it brings are expected to primarily depend on fertility trends (1). Fertility is itself determined by both behavioral factors, such as contraceptive use and physiological factors. In particular, it has long been established that lactation delays the resumption of ovarian function following birth (2–4). When contraception remains uncommon, as is still the case in many developing countries, postpartum suppression of ovarian activity by breastfeeding remains critical to control fertility (5).
Using published estimates on 48 pre-1980 populations, Bongaarts and Potter found a tight link between mean (or median) durations of breastfeeding and amenorrhea (6), best summarized by the “Bongaarts–Potter function” (BPF thereafter):where PPA and BF are the mean (or median) durations of postpartum amenorrhea and breastfeeding, both in months. Specifically, this function explained 96% of the sample’s variance on the log scale. This function enabled imputation of amenorrhea duration when missing (6–8) and has been considered strong evidence in that breastfeeding duration alone largely outweighs any other determinant of postpartum amenorrhea (9–11).
Since the BPF was proposed in the early 1980s, physiological studies of the link between energy homeostasis and reproduction have accumulated (12). Several hormones signaling nutritional status, such as leptin and ghrelin, have been shown to modulate the activity of the hypothalamic neurons that control the reproductive axis (13). Direct action of insulin on follicular growth has also been suggested to be an important link between maternal metabolism and reproduction (14). In rats, resumption of ovarian activity is significantly delayed by chronic food restriction during lactation: time to first postpartum proestrus moved from 17 to 29 d when food intake was reduced to 50%, the amount consumed ad libitum (15). While most investigations in humans focused on smaller variations in food intake, a seminal intervention study in rural Gambia found that high energy supplementation during pregnancy and lactation induced significantly faster resumption of ovarian activity, with, for example, an almost twofold increase in estradiol plasma concentration at postpartum weeks 19 to 30 (16, 17).
Many low- and middle-income countries (LMICs) have experienced rapid economic growth and health improvements since the 1980s (18). These changes are expected to improve maternal energetic status in the postpartum period: increased real wages improve food intake, the implementation of labor-saving technologies decreases energy expenditure in agricultural work and household chores, and improved sanitation reduces the prevalence of diarrheal diseases. The few recent studies of specific communities have indeed found unexpectedly short durations of postpartum amenorrhea (19–21), suggesting that the breastfeeding–postpartum amenorrhea relationship might not be captured by a single function such as the BPF.
Using data on 2.7 million births from 301 demographic surveys conducted in 84 LMICs since 1975 (SI Appendix, Fig. S1), we test two predictions derived from the “maternal energetic status” hypothesis: first, that the breastfeeding–postpartum amenorrhea relationship has weakened over time in LMICs and second, that this relationship is statistically modulated by measures of social and economic development such as the Human Development Index (HDI). These patterns are not incompatible with two other competing explanations, namely a swifter transition to residual breastfeeding and changes in “nursing intensity,” since high nursing frequency might help maintain ovarian function suppressed throughout lactation (22, 23). We therefore additionally test whether the association between exclusive breastfeeding and postpartum amenorrhea has weakened over time. Since cities are known to be forerunners of demographic and social change (7, 24), we assume any hidden changes in breastfeeding practice would diffuse more slowly in rural than in urban populations and inspect the evolution of the exclusive breastfeeding–postpartum amenorrhea relationship in rural regions.
Results
Accuracy of BPF Prediction and Country Trajectories across Time.
The BPF systematically overestimates the duration of postpartum amenorrhea except in sub-Saharan Africa (SSA) (Fig. 1). The median deviation from the BPF (BPF prediction − observed duration of amenorrhea) is 1.6 mo (interquartile range [IQR]: 0.1 to 3.5) overall. In SSA surveys, the median deviation is 0.4 mo (IQR: −0.5 to 1.7). By contrast, in Asian surveys, the median deviation is 7.0 mo (IQR: 3.8 to 10.0). The same pattern is observed in all sensitivity analyses (SI Appendix, Fig. S2).
Fig. 1.
Fig. 1 also shows the trajectory of selected countries across successive surveys. For instance, in Bangladesh, while breastfeeding duration has remained stable, the mean duration of amenorrhea has steadily declined, moving from 15.2 mo (95% CI 14.6 to 15.8) in 1975 to 7.7 mo (95% CI 7.3 to 8.0) in 2017. In some other countries such as the Philippines, the duration of breastfeeding has increased without notable changes in postpartum amenorrhea duration (see SI Appendix, Figs. S3 and S4 for detailed breastfeeding and amenorrhea schedules).
Breastfeeding–Development–Postpartum Amenorrhea Relationship.
We assess the role played by development in these deviations from the BPF by estimating a metaregression of mean duration of postpartum amenorrhea. Given the patterns revealed by Fig. 1, we analyze separately Asia-Pacific, SSA, and the rest of the world. The breastfeeding–postpartum amenorrhea relationship is found statistically modulated by the HDI (SI Appendix, Table S1 and Fig. S5), for example, in SSA, where each 0.1 increase in HDI value is associated with a 0.07 (95% CI 0.02 to 0.13) reduction in the slope of the breastfeeding–postpartum amenorrhea relationship (baseline slope at HDI = 0.2: 0.56 [95% CI 0.39 to 0.74]).
To further investigate the role of development, we focus on groups of births defined by characteristics hypothesized as relevant for maternal energetic status (e.g., access to electricity; see Methods and SI Appendix, Fig. S6). Adding these “standard of livings” characteristics to the metaregression’s predictor, we again find a strong negative modulation of the breastfeeding–postpartum amenorrhea relationship (Table 1). For example, in energetic status groups from the Asia-Pacific region, each 0.1 increase in HDI reduces the slope of the breastfeeding–postpartum amenorrhea relationship by 0.06 (95% CI 0.03 to 0.09); access to electricity is found everywhere significantly associated with a reduced slope and/or intercept. Interestingly, the model leaves unexplained a clear SSA–Asia-Pacific difference in the duration of postpartum amenorrhea: ceteris paribus, amenorrhea duration is longer in SSA (see Fig. 2, Left).
Table 1.
| Sub-Saharan Africa | Asia-Pacific | Other regions | |
|---|---|---|---|
| Duration of breastfeeding | 0.50 (0.42 to 0.57) | 0.39 (0.28 to 0.50) | 0.70 (0.51 to 0.89) |
| Living standards variables – Interaction with breastfeeding | |||
| Standardized HDI* | −0.05 (−0.07 to −0.02) | −0.06 (−0.09 to −0.03) | −0.03 (−0.07 to 0.00) |
| Urban residence | −0.08 (−0.15 to −0.02) | 0.04 (0.00 to 0.08) | −0.10 (−0.16 to −0.04) |
| Household has electricity | −0.06 (−0.13 to 0.00) | −0.01 (−0.06 to 0.03) | −0.01 (−0.08 to 0.06) |
| Water source 30 min away or less | −0.04 (−0.09 to 0.00) | −0.02 (−0.08 to 0.04) | 0.00 (−0.13 to 0.12) |
| Maternal education | 0.05 (−0.01 to 0.10) | −0.05 (−0.10 to 0.00) | −0.09 (−0.18 to −0.01) |
| Combined modulation effect† | −0.19 (−0.27 to −0.12) | −0.10 (−0.17 to −0.03) | −0.24 (−0.38 to −0.09) |
| Parity – interaction with breastfeeding | |||
| 1 | −0.05 (−0.11 to 0.01) | −0.04 (−0.08 to 0.00) | −0.07 (−0.13 to −0.01) |
| 2 or 3 | Reference | Reference | Reference |
| 4 or more | −0.04 (−0.09 to 0.01) | 0.06 (0.01 to 0.11) | 0.04 (−0.03 to 0.11) |
| Living standards variables – Main effects | |||
| Standardized HDI* | 0.54 (0.09 to 0.99) | 1.61 (0.95 to 2.27) | 0.00 (−0.55 to 0.55) |
| Urban residence | 0.65 (−0.49 to 1.82) | −1.36 (−2.31 to −0.43) | 1.32 (0.32 to 2.31) |
| Household has electricity | −0.18 (−1.43 to 1.07) | −0.96 (−2.01 to 0.10) | −1.04 (−2.28 to 0.19) |
| Water 30 min away or less | 0.55 (−0.39 to 1.52) | −0.01 (−1.35 to 1.36) | 0.02 (−2.24 to 2.39) |
| Maternal education | −2.09 (−3.21 to −0.96) | 0.88 (−0.19 to 2.02) | 1.64 (0.07 to 3.28) |
| Combined main effect† | −0.53 (−2.01 to 0.94) | 0.16 (−1.54 to 1.84) | 1.93 (−0.73 to 4.67) |
| Parity – main effects | |||
| 1 | −0.69 (−1.81 to 0.44) | −1.06 (−1.99 to −0.13) | −0.32 (−1.29 to 0.64) |
| 2 or 3 | Reference | Reference | Reference |
| 4 or more | 2.05 (1.02 to 3.09) | −0.16 (−1.31 to 0.97) | 0.01 (−1.21 to 1.27) |
| Exclusive breastfeeding | −0.05 (−0.12 to 0.01) | 0.28 (0.17 to 0.38) | 0.33 (0.20 to 0.45) |
| Intercept | 4.81 (1.83 to 7.30) | 0.54 (−3.14 to 4.11) | −0.36 (−3.89 to 3.40) |
Posterior mean (95% CI). See SI Appendix, Fig. S7 for the spline functions of calendar time. In SSA, easier access to water is associated with a 0.04 (95% CI 0.00 to 0.09) reduction in slope and a 0.55 mo (95% CI −0.39 to 1.52) increase in intercept of the breastfeeding–postpartum amenorrhea relationship. Please note that on the range of breastfeeding durations effectively found in SSA (mostly >15 mo), easier access to water is therefore associated with a shorter duration of postpartum amenorrhea (since 15 × 0.04 > 0.55).
*
Defined as 10 × (HDI − 0.2).
†
Assuming a 0.1 change in absolute value of HDI, the posterior mean for the combined modulation effect is simply the sum of the coefficients for individual living standards variables.
Fig. 2.
Consistent with the maternal depletion hypothesis, higher parity is also found associated with longer postpartum amenorrhea. For instance, for a breastfeeding duration of 20 mo, a birth of order four or above is expected to increase amenorrhea duration by similar amounts in Asia (1.0 mo [95% CI 0.7 to 1.3]), SSA (1.2 mo [95% CI 1.0 to 1.4]), and the “other” regions (0.9 mo [95% CI 0.5 to 1.2]) compared to a birth of order two or three.
Breastfeeding–Development–Total Fertility Rate Relationship.
To explore the demographic significance of these findings, we simulate the postpartum infecundability index Ci for different values of our standard of livings variables and fixed realistic values for the other predictors. In all regions, we find a large effect of development on Ci (Fig. 2, Right). For example, in Asia, an HDI of 0.3, rural residence, maternal absence of education, no electricity, and a distant water source predict a Ci value of 0.65 (95% CI 0.63 to 0.68) for a breastfeeding duration of 30 mo. By contrast, an HDI of 0.5 combined with urban residence, maternal education, electricity, and a nearby water source corresponds to a Ci value of 0.73 (95% CI 0.70 to 0.75). This change in Ci in turn leads to a 12% total fertility rate (TFR) increase (95% CI 9 to 15). For SSA, the corresponding model prediction is a 24% (95% CI 20 to 27) TFR increase.
Exclusive Breastfeeding–Postpartum Amenorrhea Relationship.
Studying the same groups of countries across surveys (SI Appendix, Tables S2 and S3), we find a rapid weakening of the association between exclusive breastfeeding and early postpartum maintenance of amenorrhea (Fig. 3). In Asian countries, an exclusively breastfeeding woman on average spent 157 d (95% CI 153 to 160) amenorrheic in the first 6 mo after childbirth in the early 1990s but only 129 d (95% CI 127 to 130) in the latest period (P value = 3.3 × 10−54). We find a more modest, albeit distinct, decline in SSA countries: time spent amenorrheic was 159 d (95% CI 157 to 161) in the latest period versus 172 d (95% CI 169 to 175) in the early 1990s (P value = 4.5 × 10−13). The same result is found when the analysis is restricted to rural regions (SI Appendix, Fig. S8) or when India is removed from the set of countries analyzed (SI Appendix, Fig. S9).
Fig. 3.
Taken together, our results therefore favor improved maternal energetic status as the most likely of the competing hypotheses explaining the weakening of the breastfeeding–postpartum amenorrhea relationship observed in the past four decades.
Discussion
The energy demands of milk production, about 500 kcal a day for exclusive breastfeeding, are high enough to require a major shift in energy homeostasis (25, 26). Pregnancy is yet another energetically demanding period of a woman’s life. The suppression of ovarian function during lactation thus prevents a situation that would compromise both mother’s and offspring’s survival and is therefore considered adaptive (27–29). Since meeting the metabolic cost of reproduction is more challenging under harsh environmental conditions, life history theory also helps explain why ovarian function appears so sensitive to energetic constraints as repeatedly found by reproductive ecology studies (30, 31), clinically with functional hypothalamic amenorrhea (32) and historically in famines (33).
Accordingly, energetics has been suggested by anthropologists as critical to explaining low fertility in contemporary foraging societies (34, 35) or an event such as the Neolithic demographic transition (36). By contrast, most demographers have long assumed that only “a difference in fertility of more than a few percent [can] be expected between poorly and well-nourished women in developing countries” (37). For a long time, intensive breastfeeding and poor energetic status were simply too correlated to disentangle their effects on ovarian function.
While the potential abandonment of breastfeeding in LMICs long remained a matter of concern (38), it became clear in the early 1990s that there was not much evidence for such a trend (39). Thanks to information campaigns on its health benefits, breastfeeding has even been increasing since then (40). At the same time, many LMICs have rapidly developed, thus creating situations of both extended breastfeeding and improved nutritional status. A pioneering study conducted in intensely breastfeeding, well-nourished women of an Argentinian indigenous group for instance found a mean duration of postpartum amenorrhea of only 10 mo (19).
We showed here that the breastfeeding–postpartum amenorrhea relationship at the population level is much more sensitive to environmental conditions than previously thought. Our results therefore resolve the longstanding apparent contradiction between demographic evidence and reproductive ecology findings (41). To the best of our knowledge, all previously published cases of mean durations of postpartum amenorrhea above 20 mo either suffered from methodological limitations (e.g., relying on implausible reports) or pertained to deprived populations where prolonged lactation combines with other severe energetic constraints.
Perhaps the most puzzling finding of the study is the unexplained difference between SSA and Asia. The HDI is not specifically centered on energetics but rather a summary measure of socioeconomic development. Defining energetic status groups aimed at moving closer to energetics, but some potentially relevant variables (e.g., the level of mechanization) remain elusive and might drive the unexplained difference. Another speculative explanation relates to social organization. Women of SSA have traditionally been economically independent and have assumed an unusually high share of agricultural work, while by contrast, many South Asian communities have long been characterized by customary female seclusion and limited labor force participation (42–44).
The data we use have exceptional geographic coverage, enable the regular follow up of the same populations across decades, and offer the possibility of methodological standardization, thus removing important barriers faced by previous research. They nevertheless suffer from several limitations. Light, sporadic bleeding can occur in the early postpartum and may be confused with the return of menses. Babies given only minimal amounts of supplementary food are not considered exclusively breastfed. The uncertainty on date of birth translates into uncertainty on age. Yet, we have no reason to believe these limitations could explain our results, simultaneously observed in countries that differ in many important respects (e.g., start date of the demographic transition, current fertility level, type of contraceptives used, degree of urbanization, ethnicity, religious beliefs). Another limitation, the lack of detailed information on women’s nutritional history and physical workload, prevented us from further testing our hypothesis on the aforementioned Asia–SSA difference.
To conclude, the most important prediction resulting from this work is an accelerated weakening of the breastfeeding–postpartum amenorrhea relationship in SSA with further advances in development. While the slower pace of the African fertility decline has had other causes (45, 46), our results assuredly call for a continued diffusion of contraception.
Methods
Data.
The Demographic and Health Surveys (DHS) are nationally representative household surveys conducted in LMICs that include a questionnaire for (married) women aged 15 to 49 y old. The DHS are intended to be regularly rerun on the same countries, typically every 5 y. A section of the questionnaire contains detailed questions on the births within the last 5 or, for few DHS, 3 y. The mother is asked whether she still breastfeeds the child and whether her menstrual period has returned. The World Fertility Surveys (WFS) program, the forerunner to the DHS program, conducted about 40 surveys in developing countries between 1974 and 1983, a fraction of which investigated postpartum amenorrhea (9).
We pool together 284 DHS and 17 WFS covering the period 1975 to 2019 for 84 LMICs (SI Appendix, Fig. S1). In order to exploit the large sample size of the four Indian DHS, the six standard zones of India are analyzed separately, yielding 321 “pseudosurveys.”
The interview of 1,816,174 women provided information on breastfeeding and postpartum amenorrhea for 2,663,496 children born in the 5 y preceding survey. We exclude from the main analysis all births to women taking hormonal contraception at the time of interview, since this may have interfered with the normal pattern of menses resumption.
Amenorrhea Status.
Though implausible, some women interviewed shortly after childbirth declare menstruation has already resumed. For all computations, we assume all women less than 1 mo postpartum are still amenorrheic. If the mother reported being pregnant at the time of interview, the last period of postpartum amenorrhea is considered over whatever the answer to the question “Has your menstrual period returned?”
Exclusive Breastfeeding Status.
Individual questions relate to the foods given to each child in the 24 h preceding the interview (“Was the child given food X?”). When the answer is no to all of these questions, the child is considered exclusively breastfed. We reconstruct individual-level exclusive breastfeeding status and then recompute median durations of exclusive breastfeeding following the DHS program’s methodology (47) and check that they match routine estimates given by the DHS program (SI Appendix, Fig. S10).
Energetic Status Groups.
We define groups of births based on parity (one; two or three; four or more), residence (urban/rural), maternal access to electricity (yes/no), maternal access to water (source for water 30 min or less away round trip: yes/no), and maternal education (no education/primary or more). Defining such energetic status groups is possible for 254 of our 321 surveys. We restrict the analysis to the 4,031 groups with an (unweighted) sample size of 70 individuals or more. We estimate mean durations of amenorrhea, breastfeeding, and exclusive breastfeeding for each of these groups as described below.
Statistical Analysis.
Estimation of mean durations.
Retrospectively reported durations are known to be unreliable in the WFS and DHS because of recall bias and because of heaping on multiples of 3 mo. For the estimation of mean durations, we therefore predict current status data (e.g., answer to the question “Are you still amenorrheic?”) using a spline function of child’s age. Writing π(a) as the probability of still being amenorrheic a months after childbirth, we fit the logistic regression with s as a monotonically decreasing spline. The mean duration of amenorrhea is then estimated as .
To get unbiased estimates of population means, normalized sampling weights are introduced in the model’s log likelihood. We estimate SEs using bootstrap resampling.
Breastfeeding–postpartum amenorrhea relationship.
We estimate a Bayesian hierarchical metaregression model to assess the breastfeeding (BF)–postpartum amenorrhea (PPA) relationship while accounting for measurement error on breastfeeding and exclusive breastfeeding (EXC) durations. Writing as our estimates for survey s and as their (estimated) SEs, we assume the measurement error models: and a standard linear model linking true durations:
We define HDI* as 10 × (HDI − 0.2) for the interpretability of regression coefficients. (respectively, is the effect of an additional month of breastfeeding (respectively, exclusive breastfeeding) on amenorrhea duration at HDI = 0.20, and is the coefficient for the HDI* − breastfeeding interaction. Xs is a vector of covariates, β is a vector of coefficients, and σ is the SE of the true relationship’s noise. Xs includes EXC, the mean parity in the sample, and a spline function of calendar time (with six uniformly spaced knots) controlling for unobserved smoothly varying factors. The prior distributions for PPA, BF, and EXC are chosen so as to approximately match the observed distributions of and . Weakly informative priors are set on the other coefficients.
The same analysis is then repeated on energetic status groups. Group characteristics are added to the linear predictor of PPA. The total effect of development (defined here as moving from rural to urban residence + maternal education + access to electricity and easier access to water + a 0.1 increase in HDI) on the breastfeeding–postpartum amenorrhea relationship can therefore be summarized by both (slope change) and (intercept change).
Effect on the TFR.
We assess the effect of development on the TFR by translating our results on PPA in terms of postpartum infecundability index Ci that in its simplest form (neglecting age structure and postpartum abstinence) is equal to 20/(18.5 + PPA) (6). We set the following values for the other predictors: EXC = 2.5 mo; survey year = 1985; mean parity = 2.5. We generate breastfeeding–postpartum amenorrhea relationships from the posterior distribution of the previous model and then translate them into breastfeeding–Ci relationships. We repeat the same exercise for the energetic status groups analysis (same EXC and survey year values, parity set to “2 to 3”).
Time spent amenorrheic by exclusively breastfeeding women.
For each group of surveys listed in SI Appendix, Tables S2 and S3, we adjust a logistic regression of current amenorrhea status on a spline function of child’s age restricted to women exclusively breastfeeding and not taking hormonal contraception. We then compute the mean time spent amenorrheic in the first 6 mo postpartum as and use bootstrap resampling for SE estimation.
The extraction of WFS data used the wfs Stata command (48). The metaregression model was written in Stan (49). All other analyses were performed in R (50) using the package rdhs for the extraction of DHS data (51), the package scam for the estimation of shape-constrained additive models (52), and the package survey for bootstrap resampling (53).
Data Availability
DHS data are freely available at https://www.dhsprogram.com for authorized users. All computer codes needed to replicate the analysis are available in GitHub at https://github.com/nptodd/BAR.
Acknowledgments
We thank Félix Balazard, Pierre Bougnères, Peter Ellison, Christophe Z. Guilmoto, Boris Hejblum, Chia Liu, Mikko Myrskylä, and Daniel Schneider for comments and suggestions on earlier versions of this work.
Supporting Information
Appendix (PDF)
- Download
- 20.47 MB
References
1
S. E. Vollset et al., Fertility, mortality, migration, and population scenarios for 195 countries and territories from 2017 to 2100: A forecasting analysis for the global burden of disease study. Lancet 396, 1285–1306 (2020).
2
T. J. Cronin, Influence of lactation upon ovulation. Lancet 2, 422–424 (1968).
3
E. J. Salber, M. Feinleib, B. Macmahon, The duration of postpartum amenorrhea. Am. J. Epidemiol. 82, 347–358 (1965).
4
L. Henry, Some data on natural fertility. Eugen. Q. 8, 81–91 (1961).
5
J. E. Finlay, I. Mejía-Guevara, Y. Akachi, Inequality in total fertility rates and the proximate determinants of fertility in 21 sub-Saharan African countries. PLoS One 13, e0203344 (2018).
6
J. Bongaarts, R. E. Potter, Fertility, Biology, and Behavior: An Analysis of the Proximate Determinants (Academic Press, 1983).
7
S. Singh, J. B. Casterline, J. G. Cleland, The proximate determinants of fertility - Sub-national variations. Popul. Stud. 39, 113–135 (1985).
8
R. J. Lesthaeghe, C. Vanderhoeft, S. Gaisie, G. Delaine, “Regional variation in the components of child-spacing: The role of women’s education” in Reproduction and Social Organization in Sub-Saharan Africa, R. J. Lesthaeghe, Ed. (University of California Press, 1989), pp. 122–166.
9
H. Leridon, B. Ferry, “Biological and traditional restraints on fertility” in Reproductive Change in Developing Countries, J. Cleland, J. Hobcraft, Eds. (Oxford University Press, 1985), pp. 139–164.
10
J. W. Wood, Dynamics of Human Reproduction: Biology, Biometry, Demography (Aldine de Gruyter, 1994).
11
H. Léridon, “Biological and social factors” in Demography: Analysis and Synthesis: A Treatise in Population Studies, G. Caselli, J. Vallin, G. Wunsch, Eds. (Academic Press, 2001), vol. 2, pp. 413–422.
12
M. Manfredi-Lozano, J. Roa, M. Tena-Sempere, Connecting metabolism and gonadal function: Novel central neuropeptide pathways involved in the metabolic control of puberty and fertility. Front. Neuroendocrinol. 48, 37–49 (2018).
13
J. W. Hill, C. F. Elias, Neuroanatomical framework of the metabolic control of reproduction. Physiol. Rev. 98, 2349–2380 (2018).
14
P. T. Ellison, Endocrinology, energetics, and human life history: A synthetic model. Horm. Behav. 91, 97–106 (2017).
15
M. K. McGuire, W. R. Butler, K. M. Rasmussen, Chronic food restriction amplifies the effect of lactation on the duration of postpartum anestrus in rats. J. Nutr. 122, 1726–1730 (1992).
16
P. Lunn et al., Maternal nutrition and lactational amenorrhoea. Lancet 1, 1428–1429 (1981).
17
P. G. Lunn, S. Austin, A. M. Prentice, R. G. Whitehead, The effect of improved nutrition on plasma prolactin concentrations and postpartum infertility in lactating Gambian women. Am. J. Clin. Nutr. 39, 227–235 (1984).
18
S. Jāhāna, Human Development Report 2016: Human Development for Everyone (United Nations Publications, 2016).
19
C. Valeggia, P. T. Ellison, Lactational amenorrhoea in well-nourished Toba women of Formosa, Argentina. J. Biosoc. Sci. 36, 573–595 (2004).
20
L. Rosetta, C. G. N. Mascie-Taylor, Factors in the regulation of fertility in deprived populations. Ann. Hum. Biol. 36, 642–652 (2009).
21
C. M. Worthman, C. L. Jenkins, J. F. Stallings, D. Lai, Attenuation of nursing-related ovarian suppression and high fertility in well-nourished, intensively breast-feeding Amele women of lowland Papua New Guinea. J. Biosoc. Sci. 25, 425–443 (1993).
22
A. S. McNeilly, Lactational control of reproduction. Reprod. Fertil. Dev. 13, 583–590 (2001).
23
M. Konner, C. Worthman, Nursing frequency, gonadal function, and birth spacing among !Kung hunter-gatherers. Science 207, 788–791 (1980).
24
M. Lerch, Fertility decline in urban and rural areas of developing countries. Popul. Dev. Rev. 45, 301–320 (2018).
25
N. F. Butte, J. C. King, Energy requirements during pregnancy and lactation. Public Health Nutr. 8 (7A), 1010–1027 (2005).
26
S. M. Anderson, P. S. MacLean, J. L. McManaman, M. C. Neville, “Lactation and its hormonal control” in Knobil and Neill’s Physiology of Reproduction, T. M. Plant, A. J. Zeleznik, Eds. (Academic Press, San Diego, ed. 4, 2015), chap. 46, pp. 2055–2105.
27
P. T. Ellison, On Fertile Ground: A Natural History of Human Reproduction (Harvard University Press, 2001).
28
C. Valeggia, P. T. Ellison, Interactions between metabolic and reproductive functions in the resumption of postpartum fecundity. Am. J. Hum. Biol. 21, 559–566 (2009).
29
G. Jasienska, R. G. Bribiescas, A. S. Furberg, S. Helle, A. Núñez-de la Mora, Human reproduction and health: An evolutionary perspective. Lancet 390, 510–520 (2017).
30
V. J. Vitzthum, H. Spielvogel, J. Thornburg, Interpopulational differences in progesterone levels during conception and implantation in humans. Proc. Natl. Acad. Sci. U.S.A. 101, 1443–1448 (2004).
31
G. Jasienska, I. Thune, P. T. Ellison, Fatness at birth predicts adult susceptibility to ovarian suppression: An empirical test of the predictive adaptive response hypothesis. Proc. Natl. Acad. Sci. U.S.A. 103, 12759–12762 (2006).
32
C. M. Gordon, Clinical practice: Functional hypothalamic amenorrhea. N. Engl. J. Med. 363, 365–371 (2010).
33
E. Le Roy Ladurie, “Famine amenorrhoea (seventeenth-twentieth centuries)” in Biology of Man in History, R. Forster, O. Ranum, Eds. (Johns Hopkins University Press, Baltimore, MD, 1975), pp. 163–178.
34
R. L. Kelly, The Lifeways of Hunter-Gatherers: The Foraging Spectrum (Cambridge University Press, 2013).
35
K. Hill, A. M. Hurtado, Ache Life History: The Ecology and Demography of a Foraging People (Aldine de Gruyter, New York, 1996).
36
J. P. Bocquet-Appel, When the world’s population took off: The springboard of the neolithic demographic transition. Science 333, 560–561 (2011).
37
J. Bongaarts, Does malnutrition affect fecundity? A summary of evidence. Science 208, 564–569 (1980).
38
J. Knodel, Breast-feeding and population growth. Science 198, 1111–1115 (1977).
39
J. Trussell, L. Grummer-Strawn, G. Rodriguez, M. Vanlandingham, Trends and differentials in breastfeeding behaviour: Evidence from the WFS and DHS. Popul. Stud. 46, 285–307 (1992).
40
C. G. Victora et al.; Lancet Breastfeeding Series Group, Breastfeeding in the 21st century: Epidemiology, mechanisms, and lifelong effect. Lancet 387, 475–490 (2016).
41
J. W. Wood, Maternal nutrition and reproduction: Why demographers and physiologists disagree about a fundamental relationship. Ann. N. Y. Acad. Sci. 709, 101–116 (1994).
42
E. Boserup, Woman’s Role in Economic Development (Earthscan, 1970).
43
J. C. Caldwell, P. Caldwell, High fertility in sub-Saharan Africa. Sci. Am. 262, 118–125 (1990).
44
S. Klasen, What explains uneven female labor force participation levels and trends in developing countries? World Bank Res. Obs. 34, 161–197 (2019).
45
E. Kebede, A. Goujon, W. Lutz, Stalls in Africa’s fertility decline partly result from disruptions in female education. Proc. Natl. Acad. Sci. U.S.A. 116, 2891–2896 (2019).
46
J. Bongaarts, J. Casterline, Fertility transition: Is sub-Saharan Africa different? Popul. Dev. Rev. 38 (suppl. 1), 153–168 (2013).
47
T. N. Croft, A. M. J. Marshall, C. K. Allen, Guide to DHS Statistics (ICF, Rockville, MD, 2018).
48
G. Rodriguez, T. Croft, Extracting and reshaping world fertility survey data in stata. Demogr. Res. 36, 893–904 (2017).
49
Stan Development Team, Stan Modeling Language Users Guide and Reference Manual 2.26 (2021).
50
R Core Team, R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, Vienna, Austria, 2019).
51
O. J. Watson, J. Eaton, rdhs: API client and dataset management for the demographic and health survey (DHS) Data. R package version 0.6.3. (2019).
52
N. Pya, SCAM: Shape constrained additive models. R package version 1.2-10. (2021).
53
T. Lumley, Survey: Analysis of complex survey samples. R package version 3.35-1. (2019).
Information & Authors
Information
Published in
Classifications
Copyright
Copyright © 2021 the Author(s). Published by PNAS. This open access article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).
Data Availability
DHS data are freely available at https://www.dhsprogram.com for authorized users. All computer codes needed to replicate the analysis are available in GitHub at https://github.com/nptodd/BAR.
Submission history
Published online: July 12, 2021
Published in issue: July 20, 2021
Keywords
Acknowledgments
We thank Félix Balazard, Pierre Bougnères, Peter Ellison, Christophe Z. Guilmoto, Boris Hejblum, Chia Liu, Mikko Myrskylä, and Daniel Schneider for comments and suggestions on earlier versions of this work.
Notes
This article is a PNAS Direct Submission.
See online for related content such as Commentaries.
Authors
Competing Interests
The authors declare no competing interest.
Metrics & Citations
Metrics
Altmetrics
Citations
Cite this article
Socioeconomic development predicts a weaker contraceptive effect of breastfeeding, Proc. Natl. Acad. Sci. U.S.A.
118 (29) e2025348118,
https://doi.org/10.1073/pnas.2025348118
(2021).
Copied!
Copying failed.
Export the article citation data by selecting a format from the list below and clicking Export.
Cited by
Loading...
View Options
View options
PDF format
Download this article as a PDF file
DOWNLOAD PDFLogin options
Check if you have access through your login credentials or your institution to get full access on this article.
Personal login Institutional LoginRecommend to a librarian
Recommend PNAS to a LibrarianPurchase options
Purchase this article to access the full text.