Original Investigation | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence
in Low- and Middle-Income Countries
Pin Wang, PhD; Lingzhi Chu, PhD; Jie Ban, PhD; Ernest O. Asare, PhD; Muthusamy Sivakami, PhD; Alexandra Restrepo Henao, PhD;
Lucy Chimoyi, PhD; Tami P. Sullivan, PhD; Kai Chen, PhD
Abstract
IMPORTANCE Recent evidence has shown that elevated temperatures are associated with a higher
risk of intimate partner violence (IPV), a serious violation of human rights. However, it remains
unclear whether drought also heightens the risk and whether drought and extreme heat jointly
affect IPV.
Key Points
Question Are drought conditions or
drought with extreme heat conditions
associated with an increased risk of
intimate partner violence (IPV) in lowand middle-income countries?
OBJECTIVE To investigate the interactive association between drought, extreme heat, and IPV in 42
Findings In this cross-sectional study
low- and middle-income countries (LMICs).
using global survey data for 494 471
women, 12-month drought was
DESIGN, SETTING, AND PARTICIPANTS This cross-sectional study included women answering
associated with an estimated 7%
multicluster surveys in sub-Saharan Africa, South and Southeast Asia, Latin America, and the
increase in IPV risk. We found a negative
Caribbean between 2003 and 2020. The analysis was conducted from January to July 2024.
and significant interaction between
drought and extreme heat.
EXPOSURES Drought was measured by the standardized precipitation evapotranspiration index at
a timescale of 1, 3, 6, or 12 months. Extreme heat days were defined as days with maximum
temperature above the 90th, 92.5th, 95th, or 97.5th percentile of the cluster-specific distribution.
The respective numbers of extreme heat days and drought months were calculated to evaluate the
interaction between drought and extreme heat.
Meaning These findings underscore
the major global health implications of
pressing climate change–related threats
to women’s health, and thus the urgency
to implement actionable interventions
and climate adaptation measures to
MAIN OUTCOMES AND MEASURES IPV against women experienced in the past 12 months before
achieve gender equality.
the survey, including emotional, physical, and sexual violence.
RESULTS Among a total of 494 471 women (mean [SD] age, 32.1 [8.5] years) surveyed, the
prevalence for any, emotional, physical, and sexual IPV was 28.3% (139 901 women), 16.8% (83 100
women), 19.5% (96 531 women), and 7.4% (36 804 women), respectively. Overall drought
+ Supplemental content
Author affiliations and article information are
listed at the end of this article.
exposures at all timescales were associated with increased IPV risk, with the largest effect size at the
12-month scale (relative risk [RR], 1.07 [95% CI, 1.06-1.09]), where mild (RR, 1.08 [95% CI, 1.06-1.09])
and severe (RR, 1.06 [95% CI, 1.04-1.08]) drought were also associated with any IPV. When
stratifying violence IPV by type, the largest effect sizes were observed for overall drought with
emotional violence at the 1-month scale and with physical violence at the 12-month scale. When
estimating relative excess risk due to interaction in 2 ways, there was a negative and significant
interaction between drought and extreme heat, defined using the threshold of the 97.5th percentile
of the distribution.
CONCLUSIONS AND RELEVANCE In this cross-sectional study of 42 LMICs, drought conditions with
various durations were associated with increased IPV risk. Given the increasing extreme events under
climate change, there is a pressing need for enhanced initiatives to prevent domestic violence.
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818
Open Access. This is an open access article distributed under the terms of the CC-BY License.
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
1/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
Introduction
Intimate partner violence (IPV) against women, characterized as violence perpetrated by a husband
or male intimate partner, is one of the most common forms of gender-based violence, hindering the
achievement of gender equality through United Nations’ Sustainable Development Goal 5.2 (ie,
eliminating all forms of violence against all women and girls in public and private spheres).1,2 IPV
poses a significant global public health risk, infringing upon human rights and causing short-term and
long-term physical and mental health consequences, such as injuries, sexually transmitted infections,
substance use, depression, and even suicide and homicide death.2-5 Estimates of the global lifetime
IPV prevalence among women aged 15 years and above in 20132 and 20225 showed a subtle decrease
from 30% in 2010 to 27% in 2018, with the highest prevalence in low- and middle-income countries
(LMICs) in Oceania, central sub-Saharan Africa, Andean Latin America, and South Asia.2,5
Human society has been experiencing the immense impact of climate change and its induced
extreme events with unprecedented frequency and intensity. For decades, scientists have been
seeking to uncover the potential association between unfavorable climatic conditions, especially hot
weather, and interpersonal violence.6-8 In particular, evidence in studies from 2020 and 20239-11 has
shown that higher temperatures increased the IPV risk in both high- and low-income settings. For
example, a 2023 study11 found that every 1 °C increase in the annual mean temperature was
associated with a 4.5% increase in the risk in 3 South Asian countries.
Extreme events can heighten the risk of gender-based violence through multiple mechanisms,
including deteriorating housing conditions, food insecurity, financial hardship, postdisaster trauma,
gender and cultural norms, law enforcement misconduct, the absence of privacy during
displacement, health care and social service suspension, and civil wars.12-14 However, limited
epidemiological evidence exists.12,15 Despite all using the Demographic and Health Surveys (DHS)16
across LMICs, inconsistent findings have been reported on the associations between dry conditions
and IPV, including either significantly positive17-19 or nonsignificant20-22 associations in sub-Saharan
Africa,17,20,21 India,18,22 and Peru.19 Except for 1 study with no specification,22 most of the current
research used negative rainfall shocks as the indicator of dry conditions,17-21 which does not take into
account evapotranspiration that transfer water content from the land surface to the atmosphere via
evaporation and transpiration. In addition, there has been more evidence on temperature
exposure,9-11 as opposed to extreme heat events. We only identified 1 study in Spain demonstrating
increased risks of femicides, police reports, and helpline calls due to IPV associated with heat
extremes over 34 °C.23
With low precipitation as the primary force, drought is a multifaceted environmental
phenomenon shaped by various climatological factors influencing the balance between water supply
and demand.24 Among them, temperatures play an imperative role in evapotranspiration, making it
an indispensable parameter when calculating drought.24 Current evidence solely dependent on
limited precipitation suggests a necessity for studies on more complex drought measures. Moreover,
high temperatures may further exacerbate drought conditions by intensifying evapotranspiration.
However, evidence remains lacking regarding the combined impact of extreme heat and drought on
IPV. Accordingly, this study aimed to (1) examine the association between drought at various
timescales and IPV in LMICs and (2) explore the potential joint association of drought and extreme
heat on IPV.
Methods
Analysis was conducted from January to July 2024 following the Strengthening the Reporting of
Observational Studies in Epidemiology (STROBE) reporting guideline for cross-sectional studies. The
Yale institutional review board determined this study as not human participant research; thus, ethics
approval for this study was not required.
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
2/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
Violence and Environmental Data
We extracted IPV occurrences from 2003 through 2020 from the DHS program, a nationally
representative household survey system conducted in over 90 LMICs worldwide since 1984.16 DHS
adopts a 2-stage sampling approach: (1) selecting survey clusters stratified by geographic region and
urban or rural residence; and (2) within each cluster, 20 to 30 households are randomly sampled for
interviews.25 Each woman was queried about whether she had experienced violence from her
husband or household partner in the past 12 months before the survey, including emotional, physical,
and sexual violence (eTable 1 in Supplement 1). In the analysis, we used a binary indicator for any IPV
(emotional, physical, and sexual IPV combined) or 1 of these subtypes. To allow for comparison
across surveys and countries, we computed a wealth index from multiple household attributes using
a principal component analysis, as described in our previous work.26
We measured drought by the standardized precipitation evapotranspiration index (SPEI), which
describes the difference between water supply, measured by precipitation, and water demand,
measured by potential evapotranspiration.24 Water demand was calculated using a series of monthly
gridded meteorological variables at a resolution of 0.1° (approximately 9 km), including maximum
and minimum temperature, dewpoint temperature, wind speed, incoming solar radiation, and air
pressure, obtained from the ERA5-Land dataset.27 We calculated the SPEI at timescales of 1, 3, 6, and
12 months (hereafter, SPEI-1, SPEI-3, SPEI-6, and SPEI-12) to capture both the short-term and longterm impact of meteorological conditions on drought status. It is important to note that the derived
SPEI is a monthly metric representing the cumulative impacts of climatic conditions over a previous
period. For instance, we calculated SPEI-12 for a given month by integrating the climatic variables for
that month and 11 previous months to sum the water balance during the entire 12 months. We
described the detailed calculation methods elsewhere.26 We defined drought severity according to
the classification method by the Federal Office of Meteorology and Climatology MeteoSwiss (overall
drought, SPEI ⱕ −0.5; mild drought: −1.3 < SPEI ⱕ −0.5; severe drought, SPEI ⱕ −1.3).26
We then linked the gridded SPEI from 2002 through 2019 with each participant based on the
geographic location of survey clusters and the interview date from 2003 through 2020. Because the
reporting window from the woman spanned 12 months, making the exact date of IPV occurrence
unknown, we used the SPEI in the 12th month before the survey to ensure that IPV occurred after
potential drought exposure.
The extreme heat indicator was represented by the total number of extreme heat days during
previous months (ie, 1, 3, 6, or 12) using thresholds of the 90th, 92.5th, 95th, or 97.5th percentile of
the survey cluster–specific maximum temperature distribution between 2002 and 2019. Similar to
drought assessment, we used the 12th month before the survey as the ending month for extreme
heat assessment to ensure temporality. For example, we calculated the total number of 12-month
extreme heat days spanning from 23 to 12 months before the survey. The calculation timeframes for
drought and extreme heat indicators are illustrated in eFigure 1 in Supplement 1.
Statistical Analysis
We employed log-binomial generalized linear mixed models by regressing the binary indicator for IPV
on the categorical drought indicator, with nested random intercepts for country and survey cluster,
allowing for variations across countries and clusters. Log-binomial regression was selected because
the outcome was common. The covariates in the models included partner’s age and education, area
of residence, and quintiles of the computed wealth index. Natural cubic splines of average mean
temperature and survey year with 3 degrees of freedom were also included to adjust for the
association with temperature and long-term trend. The average mean temperature was calculated
during the same calculation period for the SPEI. For example, for the SPEI-12 model, we included the
average temperature between the 23rd and 12th months before the survey. We then estimated the
association with any IPV by drought severity and the association with overall drought (mild and
severe drought combined) by IPV type.
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
3/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
Next, we examined the potential lag of the association of drought at various timescales with any
IPV, employing log-binomial generalized additive models together with distributed lag nonlinear
models.28 We used 11 months as the maximum lag period, with linear and natural spline functions
with 3 degrees of freedom for the predictor and lag matrices, respectively. The detailed methods for
the lagged association are described in eAppendix 1 in Supplement 1.
We examined potential effect modification by incorporating an interaction term between
drought and each of the following variables: woman’s age (median age under 31 years; 31 years or
older), her partner’s age (median age under 37 years; 37 years and older), education for both the
respondent and her partner (primary and below; secondary and above), area of residence (urban;
rural), wealth index (first and second quintile; third, fourth, and fifth quintile), if the partner drinks
alcohol (never; ever), if the partner gets drunk (never; ever), and if the woman and her partner
worked in the past 12 months (no; yes). We used the P value of the interaction term to assess the
statistical significance of subgroup differences.
We applied 2 methods to explore the interactive association of drought and extreme heat with
IPV during the 12-month period from 23 to 12 months before the interview. First, we separately
calculated the total number of drought months at various timescales and the total number of
extreme heat days above various thresholds and tested their individual associations with any IPV.
Then we included an interaction term in the model and calculated the additive interaction
represented by the relative excess risk due to interaction (RERI). Second, we computed the total
number of extreme heat days in drought months, extreme heat days in nondrought months,
non–extreme heat days in drought months, and non–extreme heat days in nondrought months
during the same period. Due to the perfect collinearity of these 4 metrics (their sum equals 12
months), we included the first 3 metrics in the model, where non–extreme heat days in nondrought
months served as a reference for the other 3 metrics. We then calculated RERI using coefficients from
the model. The detailed calculation method is described in eAppendix 2 in Supplement 1.
We tested model robustness using multiple sensitivity analyses (eAppendix 3 in Supplement 1).
All data analyses were completed using R statistical software version 4.3.1 (R Project for Statistical
Computing). All results are reported as the relative risk (RR) associated with exposure to drought
and/or extreme heat compared with nonexposure to drought and/or extreme heat, along with 95%
CIs. A 2-sided P < .05 was considered significant.
Results
A total of 494 471 women from 73 surveys in 42 countries were included in our analysis (age below
31 years at time of survey response, 233 451 [47.2%]) (eTable 2 in Supplement 1). The overall
prevalence for any IPV across all participants was 28.3% (139 901 women) (Table), and the
prevalence for emotional, physical, and sexual IPV was 16.8% (83 100 women), 19.5% (96 531
women), and 7.4% (36 804 women), respectively. Democratic Republic of the Congo exhibits the
highest prevalence of any (4460 of 8540 [52.2%]), physical (3300 of 8540 [38.6%]), and sexual
(2021 of 8540 [23.7%]) IPV, whereas Liberia has the highest prevalence of emotional IPV (2151 of
6244 [34.4%]) (Figure 1; eFigure 2 in Supplement 1). Democratic Republic of the Congo also
experienced the greatest drought events at all timescales during the study period, whereas the
Philippines experienced the least (eFigure 3 in Supplement 1). With the increase in timescale, there
was an increase in the variance of the number of drought events experienced across all studied
regions (eFigure 4 in Supplement 1). The number of extreme heat days decreased with the increase
in extreme heat intensity, and the number of women exposed to drought or extreme heat decreased
with the increase in drought timescale or extreme heat intensity, respectively (eTables 3 and 4 in
Supplement 1). During the 23rd to 12th months before the survey, more women experienced both
overall drought and extreme heat than a single type of event (eTables 5 and 6 in Supplement 1).
Exposure to overall drought at any timescale was positively associated with the risk of IPV, with
the highest RR at a timescale of 12 months (RR, 1.07; 95% CI, 1.06-1.09) (Figure 2A). Mild and severe
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
4/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
drought manifested different patterns, with the association with mild drought also peaking at the
12-month scale (RR, 1.08; 95% CI, 1.06-1.09), whereas SPEI-3 severe drought showed the largest
effect size for the association with IPV (RR, 1.10; 95% CI, 1.08-1.12) (Figure 2A). When analyzing
different types of IPV separately, effect sizes were larger for associations estimated at a shorter
timescale for emotional (SPEI-1) and sexual (SPEI-3) IPV. Conversely, we found a null association
between physical violence and SPEI-1 or SPEI-3 drought, with the highest RR for SPEI-12 drought
(Figure 2B).
The duration of the association with drought decreased with the increase in the timescale.
Specifically, the positive and significant association of SPEI-1, SPEI-3, SPEI-6, and SPEI-12 drought
persisted for 12, 9, 8, and 3 months, respectively (eFigure 5 in Supplement 1).
Table. Study Population Characteristics
Individuals, No. (%) (N = 494 471)
SPEI-12 drought exposure in 12 mos
before surveya
Characteristic
Missing
Yes
No
Yes
No
Intimate partner
violence prevalence
(%)b
Total
0
155 351
311 599
139 901
354 570
28.3
76 387 (49.2)
145 669 (46.7)
70 125 (50.1)
163 326 (46.1)
30.0
78 964 (50.8)
165 930 (53.3)
69 776 (49.9)
191 244 (53.9)
26.7
65 949 (47.0)
134 236 (49.1)
61 997 (50.6)
149 197 (47.4)
29.4
74 501 (53.0)
138 927 (50.9)
60 536 (49.4)
165 810 (52.6)
26.7
No education
52 691 (33.9)
80 919 (26.0)
38 424 (27.5)
99 318 (28.0)
27.9
Primary
47 151 (30.4)
105 998 (34.0)
52 375 (37.4)
108 700 (30.7)
32.5
43 604 (28.1)
98 192 (31.5)
41 579 (29.7)
110 995 (31.3)
27.3
11 898 (7.7)
26 473 (8.5)
7514 (5.4)
35 541 (10.0)
17.5
No education
40 741 (27.7)
58 338 (19.7)
27 985 (20.9)
74 318 (22.1)
27.4
Primary
40 216 (27.3)
94 432 (31.8)
46 226 (34.5)
95 184 (28.3)
32.7
49 681 (33.8)
112 694 (38.0)
49 434 (36.9)
124 346 (37.0)
28.4
16 468 (11.2)
31 179 (10.5)
10 153 (7.6)
42 345 (12.6)
19.3
53 722 (34.6)
120 996 (38.8)
52 430 (37.5)
138 005 (38.9)
27.5
101 629 (65.4)
190 603 (61.2)
87 471 (62.5)
216 565 (61.1)
28.8
First (lowest)
35 191 (23.7)
53 737 (19.0)
31 565 (24.1)
60 291 (18.5)
34.4
Second
30 359 (20.5)
56 533 (20.0)
28 578 (21.8)
61 430 (18.9)
31.8
31 079 (21.0)
56 822 (20.1)
26 376 (20.2)
65 887 (20.2)
28.6
Fourth
29 616 (20.0)
55 100 (19.5)
23 515 (18.0)
67 364 (20.7)
25.9
Fifth (highest)
21 960 (14.8)
61 031 (21.5)
20 788 (15.9)
70 552 (21.7)
22.8
94 002 (61.9)
172 870 (56.6)
60 683 (44.5)
218 780 (62.8)
21.7
57 817 (38.1)
132 629 (43.4)
75 744 (55.5)
129 485 (37.2)
36.9
53 235 (34.3)
116 692 (37.5)
43 778 (31.3)
135 556 (38.2)
24.4
102 062 (65.7)
194 866 (62.5)
96 090 (68.7)
218 948 (61.8)
30.5
Intimate partner violence occurrence
Respondent age, y
<31
≥31
0
Partner age, y
<37
≥37
56 931 (11.5)
Respondent education
Secondary
25 (<0.1)
Higher
Partner education
Secondary
24 480 (5.0)
Higher
Residence
Urban
Rural
0
Wealth quintile
Middle
38 125 (7.7)
Partner alcohol drinking
Never
Ever
9779 (2.0)
Respondent worked last year
No
Yes
99 (<0.1)
Partner worked last year
No
Yes
20 683 (4.2)
2720 (1.8)
5975 (2.0)
2687 (2.0)
6456 (1.9)
29.4
145 300 (98.2)
293 315 (98.0)
132 450 (98.0)
332 195 (98.1)
28.5
Abbreviation: SPEI, standardized precipitation evapotranspiration index.
a
Drought exposure for 27 521 participants was missing. SPEI-12 denotes the
standardized precipitation evapotranspiration index at a 12-month timescale.
b
Prevalence was calculated based on the occurrences of intimate partner violence
against women during the past 12 months before the survey.
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
5/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
We performed the effect modification analysis and interaction test between drought and
extreme heat only for SPEI-12 drought given its largest estimate in the main model. We noted
statistically significantly larger RRs for the association among individuals with higher education,
wealthier households, and employment in the year before the survey, compared with their
Figure 1. Prevalence of Intimate Partner Violence Against Women During the Past 12 Months Before the Survey, 2003-2020
10 20 30 40 50
Prevalence, %
Figure 2. Associations Between Risk of Intimate Partner Violence and Exposure to Drought Represented
by the Standardized Precipitation and Evapotranspiration Index (SPEI) at Various Timescales
A Timescale by drought severity
Mild drought
1.12
1.12
1.12
1.08
1.04
Relative risk
1.16
1.00
1.08
1.04
1.00
1-mo
3-mo
6-mo
12-mo
1-mo
1.04
3-mo
6-mo
12-mo
1-mo
3-mo
SPEI
6-mo
12-mo
SPEI
Timescale by type of violence
Emotional violence
Physical violence
Sexual violence
1.16
1.16
1.12
1.12
1.12
1.08
1.04
1.00
Relative risk
1.16
Relative risk
Relative risk
1.08
1.00
SPEI
B
Severe drought
1.16
Relative risk
Relative risk
All drought
1.16
1.08
1.04
1.00
1-mo
3-mo
6-mo
12-mo
SPEI
1.04
1.00
1-mo
3-mo
6-mo
SPEI
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
1.08
12-mo
1-mo
3-mo
6-mo
12-mo
SPEI
August 20, 2025
6/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
counterparts with lower education, lower household wealth status, and unemployment status
(Figure 3). We found no significant effect modification by other variables.
Year-round exposure to drought months and extreme heat days were both significantly and
positively associated with a higher risk of IPV, with larger estimates for SPEI-1 drought and extreme
heat with the strongest intensity (97.5th percentile) (eFigure 6 in Supplement 1). Surprisingly, we
found that exposure to SPEI-12 drought significantly decreased the negative association of extreme
heat with the highest intensity (97.5th percentile), and we did not observe a significant difference
between extreme heat estimates for other extreme heat definitions (Figure 4). Exposure to extreme
heat at various intensities in either nondrought months or drought months substantially increased
the risk, compared with nonexposure to extreme heat (Figure 4). Different types of calculated RERI
gradually decreased with the increase in extreme heat threshold (eTable 7 in Supplement 1). We also
found a negative and significant additive interaction between SPEI-12 drought and extreme heat
with the 97.5th percentile as the threshold (eTable 7 in Supplement 1).
Our estimate was found to be robust, except for a nonmeaningful increase in the association
with total rainfall adjusted. The application of sampling weight also slightly inflated the estimate with
a logistic regression model structure (eTable 8 in Supplement 1).
Figure 3. Associations Between Risk of Intimate Partner Violence and Exposure to Drought at a Timescale of 12 Months Stratified by Demographic, Socioeconomic,
and Behavioral Characteristics
a
a
a
a
a
1.25
Relative risk
1.20
1.15
1.10
1.05
1.00
<31
≥31
Woman's
age, y
a
<31
≥37
Partner's
age, y
Lower
Higher Lower Higher Urban
Woman's
education
Partner's
education
Rural
Residence
Lower
Higher Never
Wealth
status
Ever
Never
Alcohol
use
Ever
Partner
drunk
No
Yes
Woman
worked
No
Yes
Partner
worked
Statistically significant pairwise differences (P < .05).
Figure 4. Interactive Associations Between Extreme Heat Days With Various Thresholds and Drought
at a Timescale of 12 Months on Intimate Partner Violence
a
a
a
a
a
a
a
a
1.0060
1.0050
Relative risk
1.0040
a
Non–extreme heat days in drought month
Extreme heat days in nondrought month
Extreme heat days in drought month
1.0030
1.0020
1.0010
1.0000
90th
92.5th
95th
97.5th
Extreme heat definition
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
a
Statistically significant pairwise differences (P < .05).
August 20, 2025
7/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
Discussion
This cross-sectional study found a positive and consistent association between IPV and drought
exposure at various timescales in LMICs, with the association estimated to persist for months.
Furthermore, drought alleviated the negative impact of extreme heat with the highest intensity.
Previous evidence on the association between drought and IPV has been based on a simple
indicator of water supply deficiencies such as limited precipitation.17-22 This study furthers our
knowledge of the influence of drought, a complex multifaceted climate condition, considering
multiple timescales and meteorological conditions on the risk of common types of IPV. We revealed
a positive association not only for medium- and long-term drought (SPEI-6 and SPEI-12), coinciding
with previous studies in Africa, Asia, and Latin America,17-19 but also for short-term drought at a
timescale of 1 or 3 months, although the mechanism of varying associations by drought severity and
by IPV type needs further investigation.
Our understanding of the potential pathways through which drought influences IPV remains
quite limited. Agricultural deterioration, such as crop failure and livestock loss due to drought, and
the subsequent household financial crisis, could introduce family primary income earners (more
likely to be men in LMICs) to a series of mental disorders, such as poverty-related stress, anxiety, and
depression,29 increasing the chance of IPV. Trauma exposure among natural disaster survivors can
also potentially exacerbate IPV risks.13 The financial dependence on men already causes
compromised empowerment of women, and drought conditions could further worsen their
autonomy through reduced female employment and income, potentially enhancing men’s marital
control.19 Furthermore, the unemployment status also increases the time women spend with their
abusive partners. In addition, the deteriorating societal environment further weakens protection for
women, and therefore enables a higher possibility of gender-based violence.12
To our knowledge, this study was the first to discover a time lag in the association between
drought and a health outcome. The extended periods show potentially indirect pathways of drought
exposure affecting IPV. Additionally, shorter lags at longer timescales indicate a greater imperative for
governmental and social alleviative interventions to be implemented during severe and persistent
droughts, mitigating the impact of the drought on IPV. In contrast, shorter drought periods might not
prompt the same level of response or awareness, potentially leading to a latent adverse effect. This
differentiation highlights the importance of timely interventions to address IPV risks associated with
varying drought durations. The mechanism behind the unexpectedly larger effect sizes for associations
among more educated individuals and wealthier households remain unclear. However, more educated
individuals are more likely to report any form of IPV, while others will only report the most extreme
cases of violence. Moreover, it is possible that men who were better educated or provided more financial support to their families manifested increased tendencies toward controlling behaviors and experienced heightened financial strain induced by poverty exacerbated by drought. The reproducibility of
the effect modification by these baseline characteristics needs further investigation.
There is minimal evidence of the joint impact of drought and extreme heat on human health. A
2024 study30 also employed RERI and estimated a negative compounding effect of drought and
heatwave on child mental health outcomes. We found a similar interactive effect for extreme heat
with the strongest intensity. The underlying mechanism is unclear. As extreme heat intensified, we
observed a declining difference in the associations with extreme heat days in drought and
nondrought months. This indicates a weakening modifying effect of drought on extreme heat and an
increasingly predominant role of extreme heat in their combined impact as temperatures approach
the high-end extreme. With the increase in compound drought and extreme heat events in a
warming world,31 their interactive effect on health warrants further exploration given the anticipated
characteristic shift in frequency, intensity, and duration. To our knowledge, this is the first study
exploring the compound influence of drought and extreme heat on IPV, 2 extreme events made more
frequent by climate change. The negative interaction we estimated indicated a more effective
intervention targeting IPV if we mitigate these 2 environmental hazards sequentially rather than
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
8/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
jointly. More studies are warranted to examine the interactive effect of extreme heat with
various thresholds.
Limitations
Several limitations should be acknowledged. First, despite the use of the most strict criteria to ensure
temporality, we were unable to conduct a more precise exposure assessment due to the unknown
timing of IPV occurrence during the 12-month period. Second, IPV was self-reported through
interviews and thus is susceptible to inaccurate recall, causing outcome misclassification. Third, the
unequal temporal scales of drought (in months) and extreme heat (in days) hindered us from
defining compound days in which drought and extreme heat co-occurred at a finer scale. Fourth, we
stratified the analysis by 10 variables and tested the interaction by various drought and extreme heat
variables, which may increase the likelihood of chance association due to multiple comparisons. We
consider the effect modification and interaction analysis as exploratory and in need of further
investigation. Fifth, our study was also prone to exposure misclassification (whether individuals were
actually exposed to drought or extreme heat) because data on adaptive measures and individual
mobility were unavailable.
Conclusions
In conclusion, we estimated an increased risk of different types of IPV associated with drought
conditions with various severities and durations. With foreseeable mounting individual and
concurrent extreme events due to climate change,32 more effective efforts, such as emotional
regulation through intervention, should be undertaken to prevent domestic violence.
ARTICLE INFORMATION
Accepted for Publication: June 24, 2025.
Published: August 20, 2025. doi:10.1001/jamanetworkopen.2025.27818
Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2025 Wang P
et al. JAMA Network Open.
Corresponding Author: Pin Wang, PhD, School of Public Health Building, 4200 Valley Dr, College Park, MD 20742
(pinwang@umd.edu).
Author Affiliations: Department of Global, Environmental, and Occupational Health, School of Public Health,
University of Maryland, College Park (Wang); Department of Environmental Health Sciences, Yale School of Public
Health, Yale University, New Haven, Connecticut (Wang, Chu, Ban, Chen); Yale Center on Climate Change and
Health, Yale School of Public Health, Yale University, New Haven, Connecticut (Wang, Chu, Ban, Chen);
Department of Epidemiology of Microbial Diseases, Yale School of Public Health, New Haven, Connecticut (Asare);
School of Health Systems Studies, Tata Institute of Social Sciences, Mumbai, India (Sivakami); Epidemiology Group,
National School of Public Health, University of Antioquia, Medellín, Colombia (Henao); Implementation Research
Division, The Aurum Institute, Johannesburg, South Africa (Chimoyi); Department of Psychiatry, Yale University
School of Medicine, New Haven, Connecticut (Sullivan).
Author Contributions: Dr Wang had full access to all of the data in the study and takes responsibility for the
integrity of the data and the accuracy of the data analysis.
Concept and design: Wang, Sivakami, Chen.
Acquisition, analysis, or interpretation of data: Wang, Chu, Ban, Asare, Restrepo-Henao, Chimoyi, Sullivan, Chen.
Drafting of the manuscript: Wang, Sivakami.
Critical review of the manuscript for important intellectual content: All authors.
Statistical analysis: Wang, Chu.
Obtained funding: Chen.
Administrative, technical, or material support: Wang, Chen.
Supervision: Wang, Sivakami, Chen.
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
9/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
Conflict of Interest Disclosures: None reported.
Data Sharing Statement: See Supplement 2.
Additional Contributions: The authors would like to thank the Demographic and Health Surveys Program, ICF
International for providing the data used in the analysis.
REFERENCES
1. World Health Organization. Violence against women prevalence estimates, 2018: global, regional and national
prevalence estimates for intimate partner violence against women and global and regional prevalence estimates
for non-partner sexual violence against women. March 9, 2021. Accessed January 29, 2025. https://www.who.int/
publications/i/item/9789240022256
2. Devries KM, Mak JY, García-Moreno C, et al. Global health—the global prevalence of intimate partner violence
against women. Science. 2013;340(6140):1527-1528. doi:10.1126/science.1240937
3. Campbell JC. Health consequences of intimate partner violence. Lancet. 2002;359(9314):1331-1336. doi:10.
1016/S0140-6736(02)08336-8
4. Bacchus LJ, Ranganathan M, Watts C, Devries K. Recent intimate partner violence against women and health:
a systematic review and meta-analysis of cohort studies. BMJ Open. 2018;8(7):e019995. doi:10.1136/bmjopen2017-019995
5. Sardinha L, Maheu-Giroux M, Stöckl H, Meyer SR, García-Moreno C. Global, regional, and national prevalence
estimates of physical or sexual, or both, intimate partner violence against women in 2018. Lancet. 2022;399
(10327):803-813. doi:10.1016/S0140-6736(21)02664-7
6. Carlsmith JM, Anderson CA. Ambient temperature and the occurrence of collective violence: a new analysis.
J Pers Soc Psychol. 1979;37(3):337-344. doi:10.1037/0022-3514.37.3.337
7. Hsiang SM, Burke M, Miguel E. Quantifying the influence of climate on human conflict. Science. 2013;341(6151):
1235367. doi:10.1126/science.1235367
8. Mahendran R, Xu R, Li S, Guo Y. Interpersonal violence associated with hot weather. Lancet Planet Health.
2021;5(9):e571-e572. doi:10.1016/S2542-5196(21)00210-2
9. Henke A, Hsu LC. The gender wage gap, weather, and intimate partner violence. Rev Econ Househ. 2020;18(2):
413-429. doi:10.1007/s11150-020-09483-1
10. Nguyen M. Temperature and intimate partner violence. Scott J Polit Econ. 2023;71(2):197-218. doi:10.1111/
sjpe.12365
11. Zhu Y, He C, Bell M, et al. Association of ambient temperature with the prevalence of intimate partner violence
among partnered women in low- and middle-income south Asian countries. JAMA Psychiatry. 2023;80(9):
952-961. doi:10.1001/jamapsychiatry.2023.1958
12. Thurston AM, Stöckl H, Ranganathan M. Natural hazards, disasters and violence against women and girls:
a global mixed-methods systematic review. BMJ Glob Health. 2021;6(4):e004377. doi:10.1136/bmjgh-2020004377
13. National Academies of Sciences, Engineering, and Medicine. Essential Health Care Services Addressing
Intimate Partner Violence. National Academies; 2024.
14. Nabaggala MS, Reddy T, Manda S. Effects of rural-urban residence and education on intimate partner violence
among women in Sub-Saharan Africa: a meta-analysis of health survey data. BMC Womens Health. 2021;21(1):149.
doi:10.1186/s12905-021-01286-5
15. van Daalen KR, Kallesøe SS, Davey F, et al. Extreme events and gender-based violence: a mixed-methods
systematic review. Lancet Planet Health. 2022;6(6):e504-e523. doi:10.1016/S2542-5196(22)00088-2
16. Demographic and Health Surveys Program. November 8, 2021. Accessed February 16, 2024. https://
dhsprogram.com/
17. Epstein A, Bendavid E, Nash D, Charlebois ED, Weiser SD. Drought and intimate partner violence towards
women in 19 countries in sub-Saharan Africa during 2011-2018: a population-based study. PLoS Med. 2020;17(3):
e1003064. doi:10.1371/journal.pmed.1003064
18. Dehingia N, McDougal L, Silverman JG, et al. Climate and gender: association between droughts and intimate
partner violence in India. Am J Epidemiol. 2024;193(4):636-645. doi:10.1093/aje/kwad222
19. Díaz JJ, Saldarriaga V. A drop of love? Rainfall shocks and spousal abuse: evidence from rural Peru. J Health
Econ. 2023;89:102739. doi:10.1016/j.jhealeco.2023.102739
20. Cools S, Flato M, Kotsadam A. Rainfall shocks and intimate partner violence in sub-Saharan Africa. J Peace
Res. 2020;57(3):377-390. doi:10.1177/0022343319880252
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
10/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
21. Cooper M, Sandler A, Vitellozzi S, et al. Re-examining the effects of drought on intimate-partner violence. PLoS
One. 2021;16(7):e0254346. doi:10.1371/journal.pone.0254346
22. Rai A, Sharma AJ, Subramanyam MA. Droughts, cyclones, and intimate partner violence: a disastrous mix for
Indian women. Int J Disaster Risk Reduct. 2021;53:102023. doi:10.1016/j.ijdrr.2020.102023
23. Sanz-Barbero B, Linares C, Vives-Cases C, González JL, López-Ossorio JJ, Díaz J. Heat wave and the risk of
intimate partner violence. Sci Total Environ. 2018;644:413-419. doi:10.1016/j.scitotenv.2018.06.368
24. Vicente-Serrano SM, Begueria S, Lopez-Moreno JI. A multiscalar drought index sensitive to global warming:
the standardized precipitation evapotranspiration index. J Clim. 2010;23(7):1696-1718. doi:10.1175/
2009JCLI2909.1
25. Corsi DJ, Neuman M, Finlay JE, Subramanian SV. Demographic and health surveys: a profile. Int J Epidemiol.
2012;41(6):1602-1613. doi:10.1093/ije/dys184
26. Wang P, Asare E, Pitzer VE, Dubrow R, Chen K. Associations between long-term drought and diarrhea among
children under five in low- and middle-income countries. Nat Commun. 2022;13(1):3661. doi:10.1038/s41467-02231291-7
27. Muñoz-Sabater J, Dutra E, Agustí-Panareda A, et al. ERA5-Land: a state-of-the-art global reanalysis dataset for
land applications. Earth Syst Sci Data. 2021;13(9):4349-4383. doi:10.5194/essd-13-4349-2021
28. Gasparrini A, Armstrong B, Kenward MG. Distributed lag non-linear models. Stat Med. 2010;29(21):
2224-2234. doi:10.1002/sim.3940
29. Vins H, Bell J, Saha S, Hess JJ. The mental health outcomes of drought: a systematic review and causal process
diagram. Int J Environ Res Public Health. 2015;12(10):13251-13275. doi:10.3390/ijerph121013251
30. Sewell K, Paul S, De Polt K, et al. Impacts of compounding drought and heatwave events on child mental
health: insights from a spatial clustering analysis. Discov Ment Health. 2024;4(1):1. doi:10.1007/s44192-02300055-0
31. Tripathy KP, Mukherjee S, Mishra AK, Mann ME, Williams AP. Climate change will accelerate the high-end risk
of compound drought and heatwave events. Proc Natl Acad Sci U S A. 2023;120(28):e2219825120. doi:10.1073/
pnas.2219825120
32. Mazdiyasni O, AghaKouchak A. Substantial increase in concurrent droughts and heatwaves in the United
States. Proc Natl Acad Sci U S A. 2015;112(37):11484-11489. doi:10.1073/pnas.1422945112
SUPPLEMENT 1.
eAppendix 1. Model Specification of Lagged Association Evaluation
eAppendix 2. Calculation of Relative Excess Risk Due to Interaction (RERI)
eAppendix 3. Sensitivity Analysis Specification
eTable 1. Definitions of Different Types of Intimate Partner Violence
eTable 2. Included Countries, Survey Years, and Number of Participants
eTable 3. Descriptive Statistics for the Number of Overall Drought Months and Number of Extreme Heat Days
During the 23rd to 12th Months Before the Survey
eTable 4. Number of Women (Among All 494 471 Women) Exposed to Overall Drought at Various Timescales or
Extreme Heat With Various Definitions
eTable 5. Number of Women (Among All 494 471 Women) Exposed to Extreme Heat With Various Definitions by
Exposure to Overall Drought at a Timescale of 1 or 3 Months During the 23rd to 12th Months Before the Survey
eTable 6. Number of Women (Among All 494 471 Women) Exposed to Extreme Heat With Various Definitions by
Exposure to Overall Drought at a Timescale of 6 or 12 Months During the 23rd to 12th Months Before the Survey
eTable 7. Relative Excess Risk Due to Interaction Between Overall Drought and Extreme Heat
eTable 8. Comparison Between Multiple Sensitivity Analyses and Main Analysis Using Overall Drought at a
Timescale of 12 Months as the Exposure Variable
eFigure 1. Timeframes to Calculate the Drought Indicator (SPEI) at a Timescale of 12 Months in the 12th Month
Before the Survey and Extreme Heat Indicator During the 23rd to 12th Months Before the Survey
eFigure 2. Global Prevalence for Emotional, Physical, and Sexual Intimate Partner Violence (Emotional Violence
Data in Bangladesh and Colombia Were Missing)
eFigure 3. Country-Specific Distributions of the Number of Drought Months (Overall Drought, Including Mild and
Severe Drought) Across All 9-km Grids, Sorted by Area-Weighted Mean Value (Diamonds) During 2002-2019
eFigure 4. Distribution of Total Number of Drought Months at Different Timescales Across All 9-km Grids During
2002–2019
eFigure 5. Lagged Associations Between Overall Drought at Different Timescales and Any Intimate Partner
Violence
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
11/12
JAMA Network Open | Environmental Health
Drought, Extreme Heat, and Intimate Partner Violence in Low- and Middle-Income Countries
eFigure 6. The Association Between Intimate Partner Violence and Number of Drought Months by Timescale and
Number of Extreme Heat Days by Threshold During the 23rd to 12th Months Before the Survey
SUPPLEMENT 2.
Data Sharing Statement
JAMA Network Open. 2025;8(8):e2527818. doi:10.1001/jamanetworkopen.2025.27818 (Reprinted)
Downloaded from jamanetwork.com by guest on 12/16/2025
August 20, 2025
12/12