Temperature anomalies undermine the health of reproductive-age women in low- and middle-income countries
Edited by Richard Alba, City University of New York, Graduate Center, New York, NY; received July 7, 2023; accepted January 14, 2024
Significance
This study uses data from 2.5 million women in 59 low- and middle-income countries and matching climate information to examine how temperature and precipitation departures from normal affect women’s reproductive health, nutritional status, and temporary migration. These outcomes are strongly correlated with individuals’ welfare and economic growth in developing countries, and our target population of reproductive-age women has been understudied in the climate change literature to date. The analysis reveals that exposure to hotter-than-usual conditions contributes to temporary migration, underweight, and lack of reproductive autonomy, particularly in rural areas, indicating that ongoing global warming is likely to undermine women’s health and well-being.
Abstract
Climate change is expected to undermine population health and well-being in low- and middle-income countries, but relatively few analyses have directly examined these effects using individual-level data at global scales, particularly for reproductive-age women. To address this lacuna, we harmonize nationally representative data from the Demographic and Health Surveys on reproductive health, body mass index (BMI), and temporary migration from 2.5 million adult women (ages 15 to 49) in approximately 109,000 sites across 59 low- and middle-income countries, which we link to high-resolution climate data. We use this linked dataset to estimate fixed-effect logistic regression models of demographic and health outcomes as a function of climate exposures, woman-level and site-level characteristics, seasonality, and regional time trends, allowing us to plausibly isolate climate effects from other influences on health and migration. Specifically, we measure the effects of recent exposures to temperature and precipitation anomalies on the likelihood of having a live birth in the past year, desire for another child, use of modern contraception, underweight (BMI < 18.5), and temporary migration, and subsequently allow for nonlinearity as well as heterogeneity across education, rural/urban residence, and baseline climate. This analysis reveals that exposures to high temperatures increase live births, reduce desire for another child, increase underweight, and increase temporary migration, particularly in rural areas. The findings represent clear evidence that anthropogenic temperature increases contribute to temporary migration and are a significant threat to women’s health and reproductive autonomy in low- and middle-income countries.
Sign up for PNAS alerts.
Get alerts for new articles, or get an alert when an article is cited.
Climate change is expected to have widespread negative consequences for global population health and well-being, particularly in low- and middle-income countries (1). These countries are predominantly low-latitude and will experience more rapid departure of their climates from the historical envelope (2), including both warming and locally specific changes in flood and drought frequency (3). In many locations, these changes are projected to decrease crop yields (4), increase infectious disease transmission (5), and decrease labor productivity (6). In turn, these impacts are likely to undermine population health (7, 8) and increase population movements (9), with vulnerable groups such as women, children, and the poor likely to be the most affected (10). However, the global links from climate to health and migration have been poorly described due to the near-absence of globally integrated data on social outcomes and climate exposures (11). This is particularly true for reproductive-age women, who are made vulnerable by patriarchal norms and practices but have been the focus of relatively few studies in the climate change literature (12).
Due to data limitations, most studies to date of climatic effects on health and migration in low- and middle-income countries have either focused on single-country contexts (13) or drawn on aggregate data that cannot easily be used to study vulnerable populations (9). However, several recent studies have bridged this gap to examine how climate exposures affect vulnerable populations across large scales while also accounting for the non-random spatial distribution of these exposures (14). These studies have revealed that adverse climate conditions a) undermine child health in Africa (15–17), South Asia (18), and globally (19, 20), b) undermine women’s reproductive health and pregnancy outcomes in Africa (21, 22), and c) can either increase or decrease temporary (23, 24) and lifetime migration (25–27) depending on the context. These studies reinforce the importance of understanding the consequences of climate change for vulnerable populations but are as of yet limited in their geographic scope and the range of outcomes considered.
We advance this literature by examining the consequences of climate exposures for reproductive health, nutritional status, and temporary migration for a sample of 2.5 million adult women across 59 low- and middle-income countries (Fig. 1). We focus on these outcomes because a) they are central to the well-being and human development of women globally, b) they have been consistently measured across many rounds by the Demographic and Health Surveys (DHS), c) they address the core interests of demographers in fertility, migration, and health, and d) previous evidence suggests that they are likely to be influenced by climate exposures (7, 9, 11). These effects are likely to operate via a combination of direct effects on human physiology (28), proximate effects on household livelihoods and labor productivity (4, 6), and distal effects on contextual characteristics such food prices and health services access (29). We cannot directly observe these mechanisms in our data, but a detailed analysis of heterogeneity in the climate effects provides important insights.
Fig. 1.
To implement this, we integrate data from 141 georeferenced nationally representative DHS samples of adult women and link them to community-level measures of temperature and precipitation anomalies from the 24 mo prior to observing the outcome. We then analyze these data using fixed-effect regression models that measure climatic effects on five demographic and health outcomes (i.e., live birth in the past year, desire for another child, use of modern contraception, underweight, and temporary migration) while accounting for the observed characteristics of individuals and their communities, seasonality, all time-invariant characteristics of locations, and regional time trends. By examining multiple outcomes within a unified empirical framework, our aim is to provide a comprehensive characterization of climate effects on women’s health and well-being across the developing world.
Results
The effects of temperature anomalies on health and mobility for the full sample and selected subsamples are presented in Fig. 2 as odds ratios, which can be interpreted as the multiplicative effect of a 1 SD change in temperature or precipitation on the odds of experiencing the outcome. Overall, temperature anomalies increase births, reduce the desire for another child, increase underweight, increase temporary migration, and do not have a statistically significant effect on the use of contraception. For a 1 to 2 SD change in temperature, the magnitudes of these effects are smaller than the well-documented effects of education and rural locations on the same outcomes but, in many cases, become larger for a (currently) rare 3 SD event (Materials and Methods). It is important to keep in mind that temperatures in many low- and middle-income countries are expected to exceed their historical values by several SDs by the end of the century (2).
Fig. 2.
Specifically, we find that the odds of having a birth increase 2.1% with each unit (1 SD) of the temperature anomaly (P < 0.001), and this effect is strongest for the rural (OR = 1.027, P < 0.001), primary-educated (OR = 1.023, P = 0.001), and historically wet subsamples (OR = 1.023, P < 0.001). In contrast, the odds of desiring another child decrease 6.6% with each unit of the temperature anomaly (P < 0.001), with the strongest effects visible in the rural (OR = 0.930, P < 0.001), primary-educated (OR = 0.906, P < 0.001), colder-than-median (OR = 0.885, P < 0.001), and wetter-than-median subsamples (OR = 0.929, P < 0.001). As described below, the negative effects of temperature on desire for another child hold for both women who had a birth in the past year and those who did not, indicating that this effect is not a mechanistic consequence of more births. Instead, it appears that there is a loss of reproductive autonomy in which the desire for children declines but births increase as the temperature anomaly rises, a pattern which is strongest in the rural subsample.
Consistent with this interpretation, the contraceptive response is also complex: The overall effect of temperature on contraceptive use is non-significant, but this masks a combination of positive effects (for the rural, less-than-primary, and historically cold subsamples) and negative effects (for the urban, historically hot or dry, primary-educated, and secondary-educated subsamples). Where increased (decreased) desire for another child translates into less (more) contraceptive use, we would expect the effects on desire and contraception to be in opposite directions. However, for the majority of subsamples (urban, primary-educated, secondary-educated, historically hot, and historically dry), they are in the same direction, suggesting that temperature anomalies undermine fertility regulation for these women. This pattern is particularly visible in historically dry areas, where temperature anomalies significantly increase births (OR = 1.019, P < 0.001), reduce desire (OR = 0.934, P < 0.001), and reduce contraceptive use (OR = 0.954, P < 0.001). Of course, it is also possible that there is heterogeneity in climate effects beyond what we capture here, such that different populations of women experience changes in desire, contraception, and births.
For the smaller (but still quite large) subsamples for which underweight (N = 1,746,001) and migration (N = 855,625) were measured, the effects of temperature on these outcomes are consistent with a dynamic in which heat-related adversity undermines health and increases population movement (Fig. 2). Temperature anomalies increase underweight (OR = 1.048, P < 0.001), especially for women in rural (OR = 1.063, P < 0.001), hot (OR = 1.075, P < 0.001), or wet locations (OR = 1.059, P < 0.001). The temperature effect on temporary migration is also positive (OR = 1.027, P = 0.011), and is largest for women in urban (OR = 1.045, P = 0.007) and hotter-than-median locations (OR = 1.069, P = 0.002). These results help to explain why desire for another child declines with temperature and possibly also why births, nonetheless, increase: Women whose health has declined or who have moved temporarily might be less likely to desire another child but (particularly for migrants) might also be less able to control their fertility.
Precipitation has distinct effects from temperature, decreasing births and desire while increasing contraceptive use, underweight, and temporary migration (Fig. 3). Specifically, precipitation anomalies decrease births (OR = 0.989, P = 0.007), decrease desire for another child (OR = 0.985, P = 0.003), and increase the use of contraception (OR = 1.044, P < 0.001). The effect on births is relatively similar for all subgroups, but the effect on contraception is largest in cold (OR = 1.082, P < 0.001) and wet locations (OR = 1.071, P < 0.001). The effect on desire varies widely, with a large negative effect in colder-than-median locations (OR = 0.933, P < 0.001) and for women with primary education (OR = 0.938, P < 0.001) and a large positive effect in hotter-than-median locations (OR = 1.040, P < 0.001). Underweight (OR = 1.037, P < 0.001) and migration (OR = 1.034, P = 0.001) increase roughly uniformly with precipitation, though with large CI in the case of migration. Nonetheless, the effect on underweight is particularly strong in hot locations (OR = 1.075, P < 0.001) and non-significant in cold locations. These results suggest that wetter-than-usual conditions create adversity (19) to which women are able to respond (unlike temperature) by lowering their fertility, perhaps reflecting the fact that variations in precipitation are more localized or operate through fewer channels than temperature. This finding may also be due to both extreme wet and dry (or hot and cold) conditions having adverse effects, the next possibility that we explore.
Fig. 3.
The results presented in Fig. 4 decompose the climate anomalies into those less than −1 (cold/dry) and greater than 1 (hot/wet) with the intervening “normal” range as the reference category, thus allowing nonlinear climate effects. The results reveal that the relationships between temperature and reproductive health outcomes are primarily driven by the impacts of hotter-than-normal conditions, whereas temperature effects on underweight and migration also arise from the opposite-direction effects of colder-than-normal conditions. Similarly, precipitation effects appear to be mostly the result of dry conditions, but opposite-direction effects from wet conditions are evident for contraception and migration. Additionally, desire for another child declines in both wet (OR = 0.889, P < 0.001) and dry (OR = 0.936, P < 0.001) conditions, consistent with the expectation that both wet and dry conditions can be adverse.
Fig. 4.
The Bottom panel of Fig. 4 allows for additional complexity by examining the overlap of hot/cold and wet/dry conditions. Some of these combinations are quite rare (Materials and Methods), resulting in wide CI and one missing estimate (cold/dry on migration). Nonetheless, this analysis reveals that “double exposures” are particularly important for births, desire, and migration. Births increase under hot and dry conditions (OR = 1.120, P < 0.001); desire decreases under hot and wet conditions (OR = 0.581, P < 0.001) as well as under cold and dry conditions (OR = 0.815, P < 0.001); and migration decreases under cold and wet conditions (OR = 0.643, P < 0.001). These results suggest that attention to the locally specific ways that climate change will unfold (e.g., the magnitude of warming, increases or decreases in precipitation) is needed to accurately develop expectations about climate effects on human populations.
We also report the results of several alternative specifications and two alternative stratifications in SI Appendix, Tables S7–S10. Separately making each of the following changes to the analysis has little impact on the results: excluding countries with only one DHS round, excluding unmarried women, excluding women who moved to the survey location in the prior 24 mo, excluding women who are not usual residents of the sample household, excluding women who had a birth in the past year (for non-birth outcomes), moving the level of the clustering correction from the sample cluster to the CRU pixel or to the province, replacing climate anomalies with raw climate values, and re-measuring all of the effects on the scale of linear probabilities. Only changing the period of climate exposure from 24 to 12 or 36 mo leads to meaningful differences. Notably, contraceptive use significantly increases with 12-mo temperature anomalies but declines with the same measure over 36 mo, while the other effects change little. This suggests that women can increase their contraceptive use in response to heat over the short term, but eventually this response is undermined by the adverse conditions and the effect becomes negative.
Finally, we also present two alternative stratifications in SI Appendix, Table S10, by world region and by the interaction of historically cold/hot and dry/wet conditions (as used above). These stratified analyses reveal that, among the temperature effects, a) the positive effect on births and the negative effect on desire are strongest is Sub-Saharan Africa, b) the positive effect on underweight is strongest in Asia, c) the positive effect on migration is strongest in Latin America as well as hot and dry locations, and d) the temperature effect on contraception varies substantially across world regions as well as detailed climate zones. When the sample is narrowed to Sub-Saharan Africa, the poorest world region and the one where many concerns about climate vulnerability are concentrated (19), the negative effect on desire becomes substantially larger, the contraceptive effect becomes negative, and the migration effect becomes non-significant, suggesting a process in which women are trapped into undesired fertility but cannot leave.
Discussion
Temperatures in most tropical locations are projected to exceed their historical levels by several SD by 2100 (2). In this context, our results (together with previous studies) indicate that ongoing warming has the potential to substantially undermine the health and reproductive autonomy of reproductive-age women in low- and middle-income countries and to increase their temporary migration. Women in rural or hotter-than-median locations appear to be at particular risk along multiple dimensions. However, vulnerability is not confined to these groups, as women in urban or colder-than-median locations experience many of the same or even stronger effects, and education is also not strongly protective. The fact that rural women are disproportionately vulnerable is consistent with an agricultural mechanism that connects climate and health (4), but other mechanisms such as disease transmission and non-agricultural labor productivity are also likely to contribute (5, 6). Precipitation also matters: Like heat, wet conditions appear to be unfavorable in that they increase underweight and migration and decrease the desire for another child. Unlike heat, women are able to respond with higher use of contraception and lower fertility.
These findings hold important lessons for climate vulnerability research. First, a large fraction of such research in low and middle-income countries has continued to focus solely on precipitation exposures or closely related metrics such as soil moisture or vegetation greenness, incorporating temperature only to the extent that it influences moisture. The results presented here, as well as a large body of previous research as cited above, indicate that temperature has independent effects on population well-being that are in many ways more threatening than precipitation extremes. Future studies should consistently address temperature effects and further investigate related metrics (e.g., humidity-adjusted temperatures) in order to create a clearer picture of how population well-being will evolve under climate change (11). Second, we do not find evidence that reproductive-age women in urban locations, in colder locations, or with secondary education are protected from climate shocks, and the temporary migration of urban women is in fact more responsive to temperature exposures than those in rural areas. This finding contributes to existing evidence (5, 16, 18, 22–23, 25–26) that vulnerability to climate exposures is complex and does not necessarily break down in the expected ways along lines of class, urbanicity, and baseline climate. The implication is that ongoing processes of human development and urbanization will not quickly erase vulnerability to climate change, emphasizing the importance of emissions reductions and climate stabilization.
Our findings are limited by the cross-sectional nature of the DHS data, which keeps us from observing the evolution of health and migration over time for particular individuals or communities and may introduce biases associated with past mortality and migration. Additionally, the primary focus of the DHS is on reproductive health, which prevents us from observing climate effects on intermediate outcomes such as income and poverty. However, ongoing initiatives for data collection and harmonization have the opportunity to substantially facilitate and advance this line of inquiry. The IPUMS-DHS project has harmonized 180 DHS samples from Africa and Asia, allowing them to be downloaded as a single, consistent dataset with attached climate data (30). If this project can expand to the full DHS corpus, the work presented here could be updated with much less effort. The World Bank’s LSMS-ISA project has collected internationally comparable, longitudinal, multi-purpose household surveys in eight African countries, allowing the well-being of particular households to be tracked over time following climate exposures (31). Ongoing efforts to aggregate and harmonize data from population censuses and labor force surveys (32), agricultural censuses (33), population surveillance systems (34), and death certificates (7–8) from low- and middle-income countries hold similar promise and should be a high priority for funding agencies and the scholarly community.
Materials and Methods
We use survey data from the DHS, gridded monthly climate data from the Climatic Research Unit (CRU), and first-level administrative boundaries from GADM. The DHS are nationally representative surveys of reproductive-age women (15 to 49 y old) in low- and middle-income countries that collect information on health status and behaviors (35). The DHS data collections use a multistage, clustered sampling strategy and provide survey weights that are incorporated in our analyses. We draw on 141 surveys from 59 countries (SI Appendix, Table S1), including all surveys through the time of preparation for which geolocations were available. We construct an individual-level dataset containing information on fertility behaviors, nutritional status, temporary migration, and socio-demographic controls (SI Appendix, Table S2). Using data from all valid survey rounds, fertility behaviors were captured by whether the woman a) was using modern contraception at the time of the survey (29% of women, N = 2,361,800), b) desired another child at the time of the survey (43% of women, N = 2,299,736), and c) experienced a live birth in the 12 mo preceding the survey (13% of women, N = 2,544,010). Using data on adult height and weight from 126 survey rounds from 56 countries, underweight status was captured by whether the woman’s body mass index was under 18.5 at the time of the survey (16% of women, N = 1,746,001). Women pregnant at the time of interview were excluded from the sample for contraception, desire for another child, and underweight. Using data on mobility collected in 63 survey rounds from 38 countries, temporary migration was defined as being away from home for more than 1 mo in the past 12 mo (12% of women, N = 855,625).
Controls were selected among variables known to influence health and mobility outcomes, available in all survey rounds, and unlikely to have been altered by recent climate exposures. These include age and the square of age, education, marital status, relationship to the household head, children ever born (measured 12 mo before the survey date for the birth outcome), and rural location (SI Appendix, Table S2). Missing values on the controls (<0.3% in all cases) were addressed by including indicators for missingness as predictors in the regression analyses described below.
CRU provides gridded climate data at the 0.5° spatial resolution and monthly time scale, produced as a spatial interpolation of data from over 4,000 global weather stations (36). These data have been widely used in previous analyses of the social impacts of climate (16, 18, 22, 25, 27). Given that DHS survey cluster locations are randomly offset by 0 to 10 km to protect confidentiality (33), we extracted temperature and precipitation from CRU v4.05 as monthly spatial means to 10 km buffers centered on 109,271 geolocations, resulting in 70,941 distinct times series of climate values. Using this location-month dataset, we first defined 24-mo running means and then standardized these values using the overall location-specific historical mean and SD from 1981 to 2020, creating 24-mo z-scores or standardized climate anomalies. These values were then linked to the sample based on location and interview date. These values are plotted by world region in SI Appendix, Fig. S1, both with and without fixed effects removed, demonstrating that there is substantial variation in both temperature and precipitation anomalies within regions across the analytical sample. For births in the past year, we measured these exposures for a 24-mo period beginning 12 mo before the interview date (i.e., months t-12 to t-35), capturing the climate context of the household prior to and/or during the prenatal period. This approach is appropriate given the lag between conception and births and to avoid time-order violations that would result from measuring exposures after a birth occurred.
Climate anomalies have multiple analytical advantages over raw climate values in this setup: a) They can be interpreted as locally meaningful deviations from the expected climate, b) they are (overall) uncorrelated with the historical climate and can be treated as effectively randomized climate exposures (37), and c) they have been shown to be more predictive of social outcomes than raw climate values in multiple contexts (23, 38). Observing climate exposures over 24 mo allowed us to capture lags that have been commonly observed in this literature (39). The historical mean and SD of temperature and precipitation for each location (as defined above using the 24-mo values) are also included in the model as control variables to account for sub-province variation in the historical climate. The sample average temperature and precipitation anomalies are positive (SI Appendix, Table S3), meaning that the years preceding the survey dates were warm and somewhat wet relative to the historical record in the survey locations. Given that the mean survey date is 2011 (SI Appendix, Table S2) and the reference period is 1981 to 2020, this reflects warming at the study locations. As a final data source, we used GADM v3.6 to identify the first-level administrative unit (“province”) of each survey cluster (40).
We analyze this linked dataset using fixed-effect logistic regression models that include a) the temperature and precipitation anomalies described above, b) the socio-demographic and historical climate control variables described above, c) indicators for missingness on the controls as described above, d) fixed effects for the province, e) a region-specific set of fixed effects for interview month of year, f) a region-specific time trend, with regions defined as Latin America, Sub-Saharan Africa, North Africa/Europe, and South/Southeast Asia, g) survey weights, and h) a correction for clustering at the level of the survey cluster (SI Appendix, Table S4). This model can be represented as:
where the log-odds of a value of 1 for woman is a function of province fixed effects (, region-specific month-of-year effects (, region-specific time trends (, standardized climate anomalies that are specific to survey cluster c and time t, and individual and cluster characteristics (. The province fixed effects account for all time-invariant characteristics of provinces, the region-specific month effects account for seasonality in the outcomes in each world region, and the region-specific time trends account for changes in the macro-scale development and economic context. We cannot include fixed effects for the survey cluster or country-specific time trends given the constraints of the data. Consistent with the inclusion of these fixed effects and the use of effectively randomized climate exposures (37), we use causal language to describe the climate effects but also recognize the limitations of using cross-sectional data for this purpose (Discussion).
We subsequently extend this analysis by stratifying the sample by level of education (individual level), rural/urban (cluster level), and historical climate (cluster level) (Figs. 2 and 3). Rural/urban was measured by DHS using country-specific definitions (33). Baseline climate was stratified into hot, cold, wet, and dry at the overall sample median of historical temperature and precipitation (SI Appendix, Table S3). We also allow the climate effects to be nonlinear and interactive by categorizing the anomalies as less than -1 or greater than 1, and then as the interaction of these categories across precipitation and temperature (Fig. 4 and SI Appendix, Table S3). Because of warming over the reference period relative to when samples were collected, the cold/dry and cold/wet conditions are quite rare, affecting only ~1% of the sample, and the former does not occur in the sample for migration (SI Appendix, Table S3).
To check the robustness of our main results, we also a) exclude countries with only one DHS round (SI Appendix, Table S7), b) exclude unmarried women (who are at lower risk for contraceptive use and births) (SI Appendix, Table S7), c) exclude women who reported moving to the survey location in the prior 24 mo (SI Appendix, Table S7), d) exclude women who are not usual residents of the sample household (SI Appendix, Table S7), e) exclude women who had a birth in the past year and then those who did not have a birth in the past year (for non-birth outcomes) (SI Appendix, Table S7), f) move the level of the clustering correction from the sample cluster to the CRU pixel and then to the province (SI Appendix, Table S7), g) replace climate anomalies with raw climate values (SI Appendix, Table S7), h) change the period of climate exposure to 12 mo and then to 36 mo (SI Appendix, Table S7), and i) re-run the main analyses using linear regression, which measures the effects on the alternative scale of linear probabilities (SI Appendix, Tables S8 and S9). Finally, we also re-estimate the main specification using two alternative stratifications: by world region (as defined above) and by the intersection of historically hot/cold and wet/dry conditions as defined above (SI Appendix, Table S10).
Data, Materials, and Software Availability
The underlying data that support these findings are publicly available from the Climatic Research Unit (36), GADM (40), and the Demographic and Health Surveys (DHS) Program (35). The terms of use of the DHS data prohibit redistribution, including of the linked dataset that we analyze here. The complete Stata code that supports our findings is deposited at DOI: https://doi.org/10.17605/OSF.IO/XWDVN.
Acknowledgments
This work was supported by NICHD of the NIH under award numbers R03HD101859, P2CHD050924, and P2CHD041025. The research was also supported by the USDA National Institute of Food and Agriculture and Hatch Appropriations under project #PEN04953 and accession #7006538. We would also like to thank Phillip McDaniel who assisted with extracting the climate data.
Author contributions
C.G. and B.C.T. designed research; performed research; analyzed data; and wrote the paper.
Competing interests
The authors declare no competing interest.
Supporting Information
Appendix 01 (PDF)
- Download
- 923.54 KB
References
1
IPCC, "Climate Change 2022: Impacts, Adaptation, and Vulnerability" in Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change, H.-O. Pörtner et al., Eds. (Cambridge University Press, Cambridge, UK and New York, NY, 2022), 3056 p., doi:https://doi.org/10.1017/9781009325844.
2
D. Frame, M. Joshi, E. Hawkins, L. Harrington, M. de Roiste, Population-based emergence of unfamiliar climates. Nat. Clim. Change 7, 407–411 (2017).
3
N. W. Arnell et al., The global and regional impacts of climate change under representative concentration pathway forcings and shared socioeconomic pathway socioeconomic scenarios. Environ. Res. Lett. 14, 084046 (2019).
4
P. Zhu et al., Warming reduces global agricultural production by decreasing cropping frequency and yields. Nat. Clim. Change 12, 1016–1023 (2022).
5
A. Dimitrova, A. Gershunov, M. C. Levy, T. Benmarhnia, Uncovering social and environmental factors that increase the burden of climate-sensitive diarrheal infections on children. Proc. Natl. Acad. Sci. U.S.A. 120, e2119409120 (2023).
6
C. Z. de Lima et al., Heat stress on agricultural workers exacerbates crop impacts of climate change. Environ. Res. Lett. 16, 044020 (2021).
7
T. Carleton et al., Valuing the global mortality consequences of climate change accounting for adaptation costs and benefits. Quarterly J. Econ. 137, 2037–2105 (2022).
8
A. M. Vicedo-Cabrera et al., The burden of heat-related mortality attributable to recent human-induced climate change. Nat. Clim. Change 11, 492–500 (2021).
9
R. Hoffmann, A. Dimitrova, R. Muttarak, J. C. Cuaresma, J. Peisker, A meta-analysis of country-level studies on environmental change and migration. Nat. Clim. Change 10, 904–912 (2020).
10
G. Formetta, L. Feyen, Empirical evidence of declining global vulnerability to climate-related hazards. Global Environ. Change 57, 101920 (2019).
11
K. Grace, Considering climate in studies of fertility and reproductive health in poor countries. Nat. Clim. Change 7, 479–485 (2017).
12
M. Flatø, R. Muttarak, A. Pelser, Women, weather, and woes: The triangular dynamics of female-headed households, economic vulnerability, and climate variability in South Africa. World Dev. 90, 41–62 (2017).
13
D. J. Kaczan, J. Orgill-Meyer, The impact of climate change on migration: A synthesis of recent empirical insights. Clim. Change 158, 281–300 (2020).
14
T. A. Carleton, S. M. Hsiang, Social and economic impacts of climate. Science 353, aad9837 (2016).
15
S. Blom, A. Ortiz-Bobea, J. Hoddinott, Heat exposure and child nutrition: Evidence from West Africa. J. Environ. Econ. Management 115, 102698 (2022).
16
B. C. Thiede, J. Strube, Climate variability and child nutrition: Findings from sub-Saharan Africa. Global Environ. Change 65, 102192 (2020).
17
R. E. Baker, J. Anttila-Hughes, Characterizing the contribution of high temperatures to child undernourishment in Sub-Saharan Africa. Sci. Rep. 10, 1–10 (2020).
18
K. McMahon, C. Gray, Climate change, social vulnerability and child nutrition in South Asia. Glob. Environ. Change 71, 102414 (2021).
19
M. W. Cooper et al., Mapping the effects of drought on child stunting. Proc. Natl. Acad. Sci. U.S.A. 116, 17219–17224 (2019).
20
K. Le, M. Nguyen, In-utero exposure to rainfall variability and early childhood health. World Dev. 144, 105485 (2021).
21
F. Davenport, A. Dorélien, K. Grace, Investigating the linkages between pregnancy outcomes and climate in sub-Saharan Africa. Popul. Environ. 41, 397–421 (2020).
22
B. C. Thiede, S. Ronnkvist, A. Armao, K. Burka, Climate anomalies and birth rates in sub-Saharan Africa. Clim. Change 171, 5 (2022).
23
V. Mueller, G. Sheriff, X. Dou, C. Gray, Temporary migration and climate variation in eastern Africa. World Dev. 126, 104704 (2020).
24
A. Epstein, O. O. Harris, T. Benmarhnia, C. S. Camlin, S. D. Weiser, Do precipitation anomalies influence short-term mobility in Sub-Saharan Africa? An observational study from 23 countries. BMC Public Health 23, 1–6 (2023).
25
B. Thiede, C. Gray, V. Mueller, Climate variability and inter-provincial migration in South America, 1970–2011. Glob. Environ. Change 41, 228–240 (2016).
26
V. Mueller, C. Gray, D. Hopping, Climate-induced migration and unemployment in middle-income Africa. Glob. Environ. Change 65, 102183 (2020).
27
B. Thiede, H. Randell, C. Gray, The childhood origins of climate-induced mobility and immobility. Popul. Dev. Rev. 48, 767–793 (2022).
28
D. J. Vecellio, Q. Kong, W. L. Kenney, M. Huber, Greatly enhanced risk to humans as a consequence of empirically determined lower moist heat stress tolerance. Proc. Natl. Acad. Sci. U.S.A. 120, e2305427120 (2023).
29
M. Springmann et al., Global and regional health effects of future food production under climate change: A modelling study. The Lancet 387, 1937–1946 (2016).
30
E. H. Boyle, M. King, M. Sobek, IPUMS Demographic and Health Surveys. Version 8 [dataset] (IPUMS and ICF, Minneapolis, MN, 2020), https://doi.org/10.18128/D080.V8. Accessed 7 July 2023.
31
A. Dillon, G. Carletto, S. Gourlay, P. Wollburg, A. Zezza, Agricultural Survey Design: Lessons from the LSMS-ISA and Beyond. World Bank LSMS Guidebook (2021). http://documents.worldbank.org/curated/en/570941632979626559/Agricultural-Survey-Design-Lessons-from-the-LSMS-ISA-and-Beyond.
32
IPUMS, Integrated Public Use Microdata Series, International: Version 7.3 [dataset] (Minnesota Population Center, Minneapolis, MN, 2020). https://doi.org/10.18128/D020.V7.3.
33
S. M. Manson, T. A. Kugler, J. Schroeder, D. Van Riper, S. Ruggles, IPUMS International Historical Geographic Information System: Version 2 [dataset] (IPUMS, Minneapolis, MN, 2020), https://doi.org/10.18128/D120.V2. Accessed 7 July 2023.
34
O. Sankoh, P. Byass, The INDEPTH Network: Filling vital gaps in global epidemiology. Int. J. Epidemiol. 41, 579–588 (2012).
35
ICF, The DHS Program website. Funded by USAID (2023). http://www.dhsprogram.com. Accessed 7 July 2023.
36
I. Harris, T. J. Osborn, P. Jones, D. Lister, Version 4 of the CRU TS monthly high-resolution gridded multivariate climate dataset. Sci. Data 7, 109 (2020).
37
J. Nordkvelle, S. A. Rustad, M. Salmivalli, Identifying the effect of climate variability on communal conflict through randomization. Clim. Change 141, 627–639 (2017).
38
C. Gray, E. Wise, Country-specific effects of climate variability on human migration. Clim. Change 135, 555–568 (2016).
39
R. J. Nawrotzki, J. DeWaard, Climate shocks and the timing of migration from Mexico. Popul. Environ. 38, 72–100 (2016).
40
Global Administrative Areas, GADM database of Global Administrative Areas, Version 3.6. [dataset] (2018). https://www.gadm.org. Accessed 27 July 2023.
Information & Authors
Information
Published in
Classifications
Copyright
Copyright © 2024 the Author(s). Published by PNAS. This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).
Data, Materials, and Software Availability
The underlying data that support these findings are publicly available from the Climatic Research Unit (36), GADM (40), and the Demographic and Health Surveys (DHS) Program (35). The terms of use of the DHS data prohibit redistribution, including of the linked dataset that we analyze here. The complete Stata code that supports our findings is deposited at DOI: https://doi.org/10.17605/OSF.IO/XWDVN.
Submission history
Received: July 7, 2023
Accepted: January 14, 2024
Published online: March 5, 2024
Published in issue: March 12, 2024
Keywords
Acknowledgments
This work was supported by NICHD of the NIH under award numbers R03HD101859, P2CHD050924, and P2CHD041025. The research was also supported by the USDA National Institute of Food and Agriculture and Hatch Appropriations under project #PEN04953 and accession #7006538. We would also like to thank Phillip McDaniel who assisted with extracting the climate data.
Author contributions
C.G. and B.C.T. designed research; performed research; analyzed data; and wrote the paper.
Competing interests
The authors declare no competing interest.
Notes
This article is a PNAS Direct Submission.
Although PNAS asks authors to adhere to United Nations naming conventions for maps (https://www.un.org/geospatial/mapsgeo), our policy is to publish maps as provided by the authors.
Authors
Metrics & Citations
Metrics
Altmetrics
Citations
Cite this article
Temperature anomalies undermine the health of reproductive-age women in low- and middle-income countries, Proc. Natl. Acad. Sci. U.S.A.
121 (11) e2311567121,
https://doi.org/10.1073/pnas.2311567121
(2024).
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.