Article Text

Original research
Prevalence of childhood exposure to intimate partner violence in low-income and lower-middle-income countries: a systematic review
  1. Berit Kieselbach1,
  2. Melissa Kimber2,
  3. Harriet L MacMillan2,3,
  4. Thomas Perneger4
  1. 1Institute of Global Health, University of Geneva, Geneva, Switzerland
  2. 2Department of Psychiatry and Behavioural Neurosciences, McMaster University, Hamilton, Ontario, Canada
  3. 3Department of Pediatrics, McMaster University, Hamilton, Ontario, Canada
  4. 4Division of Clinical Epidemiology, Geneva University Hospitals, and Faculty of Medicine, University of Geneva, Geneva, Switzerland
  1. Correspondence to Berit Kieselbach; berit.kieselbach{at}


Objective To determine the proportion of children in low-income and lower-middle-income countries exposed to intimate partner violence (IPV).

Design Systematic review.

Data sources PubMed, CINAHL, ERIC, PsycINFO, Web of Science, WHO Global Index Medicus, and Violence and Abuse Abstracts, hand searching of specialised journals from inception until 19 May 2019.

Eligibility criteria for selecting studies Primary quantitative studies that included a measure of self-reported exposure to IPV prior to age 18 and were conducted in low-income and lower-middle-income countries.

Data extraction and synthesis Data were screened, extracted and appraised by two independent reviewers. The prevalence estimates were pooled using a random-effects model. Outcomes included lifetime and past-year prevalence of childhood exposure to IPV. Meta-regression was used to explore heterogeneity. Publication bias was assessed using a funnel plot and Egger’s regression test. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines were followed.

Results Sixty-two studies with a total of 231 512 participants were included. Eighty-five lifetime prevalence estimates and 6 estimates of past-year prevalence were available for synthesis. The average lifetime prevalence of childhood exposure to IPV was 29% (95% CI 26% to 31%). The average past-year prevalence in children was 35% (95% CI 21% to 48%). The lifetime prevalence disaggregated by WHO regions ranged from 21% to 34%. There were no statistical differences in prevalence estimates between samples of men and women.

Conclusion Almost one-third of children in low-income and lower-middle-income countries have been exposed to IPV in their lifetime. There was large heterogeneity between estimates that was not explained by available study and sample characteristics. Our findings indicate that children’s exposure to IPV in low-income and lower-middle-income countries is common and widespread; prevention of this major public health exposure should be a priority.

PROSPERO registration number CRD42019119698.

  • Child protection

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Strengths and limitations of this study

  • This is the first systematic review to assess the prevalence of childhood exposure to intimate partner violence in low-income and lower-middle-income countries.

  • A large number of eligible articles were screened and included in the review.

  • This systematic review responds to a policy-relevant priority identified by stakeholders from low-income and lower-middle-income countries.

  • The lack of consensus around the definitions and measures of exposures makes it challenging to derive a global prevalence figure for low-income and lower-middle-income countries.

  • Only published studies were included in the systematic review. Unpublished data and data from government or non-governmental organization (NGO) reports were not considered.


Intimate partner violence (IPV) is a serious human rights and public health problem globally. Worldwide, one in three women is affected by IPV.1 Such violence often takes place in the presence of children and can have severe and long-lasting impact on children’s health and development.

Childhood exposure to IPV includes either the direct observation or mere awareness without directly seeing or hearing violent acts or abuse between caregivers who are current or former spouses or intimate partners.2 3 Such awareness can include the child seeing some of the immediate consequences or overhearing a conversation about the violent act, experiencing life changes as a consequence of violence (eg, separation from a parent), or intervening directly in an attempt to stop the violent act.4 Childhood exposure to IPV has been associated with a broad range of physical and mental health problems, health risk behaviours and social consequences. These effects vary depending on age and developmental stage of the child at the time of exposure, as well as factors such as duration and severity, and overlap with other types of maltreatment.5 6 The broad range of mental health problems associated with childhood exposure to IPV include increased risk for depression, anxiety, conduct disorder, adjustment problems and post-traumatic stress disorder.7 8 IPV exposure has also been associated with reduced cognitive ability and educational achievement.9 Witnessing IPV in childhood is consistently identified as a risk factor for perpetrating and experiencing IPV in adulthood.10 11 Children exposed to IPV have a higher likelihood of engaging in health risk behaviours including tobacco use, the harmful use of alcohol, other types of substance use and unsafe sex.12 The increased risk for these behaviours partly explains the increased risk for chronic health outcomes including HIV, reproductive health problems and non-communicable diseases, including cardiovascular disease, cancer and diabetes, among others.13 14

Despite the widespread nature of IPV and its severe consequences for children, major gaps remain in understanding the epidemiology of children’s exposure, especially in low-income countries. Much of the literature has focused on high-income countries, which have shown that 8%–25% of children are exposed to IPV in their home.2 To our knowledge, no systematic review has attempted to synthesise existing prevalence studies of childhood exposure to IPV from low-income and lower-middle-income countries. This information is important in identifying risk factors for physical and mental health conditions in children in low-income and lower-middle-income countries, and determining what is needed with regard to policies and service provision.

We conducted a systematic review to address the need for an overview of prevalence estimates of exposure to IPV among children living in low-income and lower-middle-income countries around the world.


Research questions and outcome variables

This systematic review addresses the following research questions: (1) What is the lifetime prevalence of childhood exposure to IPV among children and adults in low-income and lower-middle-income countries? (2) What is the past-year prevalence of exposure to IPV among children in low-income and lower-middle-income countries?

The outcome of interest, childhood exposure to IPV, was defined as direct observation or awareness of violence between caregivers who are current or former spouses or intimate partners.2 3 IPV refers to behaviour by an intimate partner or ex-partner that causes physical, sexual or psychological harm, including physical aggression, sexual coercion, psychological abuse and controlling behaviours.15 To determine lifetime prevalence, relevant studies included data collected from adults, who reported exposure to IPV at any point in their lives up to the age of 18 years, and children, who reported exposure to IPV at any point in their lives up to the time of the survey.

The study followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (online supplemental material 1).

Literature search strategy

A four-step search strategy was applied to identify relevant studies. First, we searched seven electronic databases: PubMed, CINAHL (EBSCOhost), ERIC (ProQuest), PsycINFO (ProQuest), Web of Science, WHO Global Index Medicus and Violence and Abuse Abstracts (EBSCOhost). A search strategy was developed for each database using a combination of free text and controlled vocabulary and was reviewed by a PhD-trained information scientist with extensive experience in systematic review methodology and systematic reviews focused on exposure to various types of interpersonal violence, including childhood exposure to IPV. All papers published before 19 May 2019 were considered. Searches were conducted in English language but no language restrictions were placed on the search results.

The search terms included combinations and iterations of “prevalence”, “childhood”, “intimate partner violence” and “exposure” or “witnessing”. The full search strategy for each database is available in online supplemental material 2. Searches for each database were evaluated against a subsample of 10 papers that were predetermined by the research team to meet the inclusion criteria.16

Database searches were supplemented by hand searches of specialised journals focused on interpersonal violence, which were conducted in May 2019. The journals included Child Abuse & Neglect, Child Maltreatment, and Trauma, Violence & Abuse. Forward and backward citation chaining of included papers was conducted from April 2020 until May 2020 to capture any papers potentially missed by database searches and which may have been published up until the finalisation of this manuscript.

Eligibility criteria

We included primary quantitative studies that measured the prevalence of current and past exposure to IPV prior to the age of 18. Male, female and mixed-sex (Some surveys use biological sex and some surveys use the term gender) samples from low-income and lower-middle-income countries according to World Bank country and lending classification (as of October 2019) were considered.17 Data collected at national or sub-national levels were eligible. Data from both household surveys and school surveys were considered. The minimum cut-point for survey response rate was set at over 60%.

Title and abstract screening, full-text screening and data extraction

Titles and abstracts of all articles identified via the search strategy were screened by one reviewer (BK). A sample (5%) of the total records was screened by a second reviewer (MK) to check the consistency of the application of the inclusion/exclusion criteria. Disagreements were resolved through discussion and involvement of a third reviewer (HM). The interrater reliability was substantial with Cohen’s Kappa k=0.74.

At the second stage, 100 full texts were assessed for eligibility by one reviewer (BK) applying the checklist with inclusion/exclusion criteria. A subset (20%) of the full texts was assessed by a second reviewer (MK). The agreement between the reviewers was substantial with Cohen’s Kappa k=0.74.

A standardised template was created for data extraction. The main variables included study information, characteristics of the sample, study methodology (study type, sampling method, survey item, mode of data collection), and prevalence estimates. Data extraction for all included studies was conducted by one reviewer (BK). Twenty papers underwent independent data extraction by a second reviewer (MK). There was perfect agreement on the extraction of study information, including prevalence estimates, across reviewers.

Quality assessment and assessment of bias

Study quality was assessed during the data extraction process using a standardised risk of bias tool for prevalence studies (box 1) adapted from Hoy et al.18 The nine items cover different aspects of external and internal validity. Two reviewers (BK and MK) classified each of the items describing potential sources of bias into low risk or high risk. A summary score was then calculated by adding all the items rated high risk. A summary score of 0–3 is considered low risk, 4–6 moderate risk, and a score of 7–9 indicates the study is at high risk of bias. Studies with low and moderate risk of bias were included in the systematic review.

Box 1

Risk of bias assessment (adapted from Hoy et al18)

External validity (maximum score=4)

Was the study’s target population a close representation of the national population in relation to relevant variables such as age, sex, occupation, urban/rural population?

(Yes: low risk=0 points; no: high risk=1 point)

Was the sampling frame a true or close representation of the target population (household sample and/or primary school sample)?

(Yes: low risk=0 points; No: high risk=1 points)

Was some form of random selection used to select the sample, or was a census undertaken?

(Yes: low risk=0 points; no: high risk=1 point)

Was the likelihood of non-response bias minimal (response rate ≥75% or explicitly stated that there was no difference between responders and non-responders)?

(Yes: low risk=0 points; no: high risk=1 point)

Internal validity (maximum score=5)

Were data collected directly from the subjects (as opposed to a proxy)?

(Yes: low risk=0 points; no: high risk=1 point)

Was an acceptable case definition used in the study? Where subjects asked whether they witnessed or were aware of physical, sexual or emotional violence between their caregivers?

(Yes: low risk=0 points; no: high risk=1 point)

Was the study instrument that measured the parameter of interest shown to have reliability and validity (item derived from an instrument that had widely been tested for reliability or validity, or explicitly stated that validity has been measured)?

(Yes: low risk=0 points; no: high risk=1 point)

Was the same mode of data collection used for all subjects?

(Yes: low risk=0 points; no: high risk=1 point)

Were the numerator(s) and denominator(s) for the parameter of interest appropriate?

(Yes: low risk=0 points; no: high risk=1 point)

We assessed publication bias using a funnel plot and Egger’s regression test.19

Data synthesis

A meta-analysis was performed to synthesise the lifetime and past-year prevalence of childhood exposure to IPV. Prevalence rates were calculated from raw proportions or percentages reported in the included studies. Pooled prevalence estimates were determined for lifetime and past-year prevalence. The prevalence estimates were disaggregated by gender, wherever this information was available. Studies that did not disaggregate by gender were included in the category ‘mixed samples’. When studies provided different estimates for exposure to physical violence and emotional violence for the same sample, we chose ‘physical violence’, as this was the measurement applied by the majority of the studies. All analyses were done with METAPROP in STATA V.14.0 designed to perform meta-analyses of proportions. The programme computes 95% CIs using the score statistic and the exact binomial method and incorporates the Freeman-Tukey double arcsine transformation of proportions.20 The overall prevalence estimates were pooled based on a random-effects model, which takes into account that observed differences between proportions cannot be entirely attributed to the sampling error and that other factors such as true differences between study populations and methodologic differences can also contribute. Weights were applied according to the inverse of the variance. Given that within-study variance was relatively small and the variance between studies was substantial, the weights were similar across all studies. 95% CIs were calculated around the pooled estimates. To assess the extent of variation between studies, heterogeneity tests with the I2 statistic were performed.

No prespecified stratified analyses were planned for this study. Additional analyses and visual inspection of the data were conducted post hoc, following the observation of the high heterogeneity of the prevalence estimates.

Patient and public involvement

Not applicable. We performed a systematic review on published data.


A total of 6903 records were obtained through database searching and 265 additional records through hand searches. After duplicates were removed, 5556 titles and abstracts were screened for their relevance. This first screening resulted in 100 potentially eligible studies, which were then screened using the full text of the article. After full-text screening, 55 studies were identified for inclusion in the review. Main reasons for exclusion were that several papers were published using data from the same sample or did not provide prevalence estimates. Detailed reasons for exclusion are provided in the PRISMA diagram (figure 1). If several publications drew on data from the same study, the study that provided the most information, was selected. Forward and backward citation chaining of included studies yielded another seven eligible articles, so that a total of 62 studies were included in the review (online supplemental material 3). According to the risk of bias assessment18 eight studies were classified as moderate risk of bias, and 54 studies were classified as low risk of bias. No studies had to be excluded based on the risk of bias assessment. Some of these studies are multicountry studies, or they disaggregated data collection by males and females, so that the total number of available prevalence estimates is 91.

Figure 1

PRISMA flow chart of studies identified, included and excluded. PRISMA, Preferred Reporting Items for Systematic Reviews and Meta-Analyses.

We retrieved studies from 29 low-income and lower-middle income countries with data from 231 512 individuals. Twenty-seven estimates were based on data from representative national surveys and 64 estimates were based on data from sub-national administrative units such as regions or districts. Almost all studies reported applying a form of random sampling (k=57); five studies used convenience samples. The included studies yielded 85 estimates for lifetime prevalence of childhood exposure to IPV and 6 estimates on past-year prevalence. Sixty-eight prevalence estimates were determined from household sample data; 22 prevalence estimates were based on data from school-based samples and one prevalence estimate was based on data collected in public institutions in the community. Most studies measured exposure to physical IPV between caregivers (k=55); seven studies measured exposure to physical and emotional IPV. Twenty-two studies operationalised exposure to physical IPV between caregivers as bidirectional violence, and 45 studies explicitly asked whether IPV was perpetrated by the father against the mother.

The overall random-effects pooled lifetime prevalence of childhood exposure to IPV across all samples (n=85) was 29% (95% CI 26% to 31%) with a high level of heterogeneity across studies (I2=99.67%, p<0.001 ; T2=0.02). Lifetime prevalence estimates ranged from a minimum of 2% to a maximum of 78%, with an IQR from 16% to 37% and a median of 26%. The pooled past-year prevalence (n=6) was 35% (95% CI 21% to 48%) with similarly high levels of heterogeneity (I2=98.3%, p<0.001; T2=0.03). The past-year prevalence estimates spread from 12% to 57%. The IQR reached from 22% to 49% with a median prevalence of 34%.

The lifetime prevalence in studies that involved either male or female samples or provided a gender breakdown (n=76) was 27% (95% CI 23% to 30%) for females and 31% (95% CI 25% to 38%) for males. Minimum and maximum values and quartiles for female, male and mixed samples are shown in figure 2. The past-year prevalence (n=4) was 29% (95% CI 26% to 32%) for females and 28% (95% CI 25% to 31%) for males. The difference between female, male and mixed samples was not statistically significant for lifetime (p=0.39) or for past-year prevalence (p=0.66).

Figure 2

Box-plot of the lifetime prevalence of childhood exposure to IPV disaggregated by gender. IPV, intimate partner violence.

To explore the sources of heterogeneity, sample size, median age of the sample, risk of bias rating, geographical region and data collection method (household, school) were entered into a meta-regression. None of the independent variables was statistically significantly associated with prevalence.

The funnel plot was asymmetric, whereby asymmetry was caused by smaller studies that tended to give results emphasising higher prevalence rates. Egger’s regression test was significant (p=0.03). We applied the trim and fill method to calculate whether potential publication bias had an impact on the pooled prevalence estimates.21 Seven additional studies were imputed, but they did not change the summary estimate.

The global and WHO regional prevalence estimates for childhood exposure to IPV are shown in figure 3. The pooled prevalence in low-income and lower-middle-income countries in the South East Asian Region, based on 23 samples, was 26% (95% CI 21% to 30%), in the African Region (AFRO) 34% (95% CI 27% to 40%) based on 30 samples, in the Region of the Americas 34% (95% CI 19% to 49%) based on seven samples, in the Western Pacific Region 27% (95% CI 20% to 34%) based on 13 samples, in the Eastern Mediterranean Region (EMRO) 21% (95% CI 15% to 26%), based on seven samples, and in low-income and lower-middle-income countries from the European Region (EURO) 21% (95% CI 12% to 29%) based on five samples. The heterogeneity between geographical regions was statistically significant (p=0.04).

Figure 3

Pooled prevalence for childhood exposure (CE) to IPV in low-income and lower-middle income countries disaggregated by who region. IPV, intimate partner violence.


We used meta-analytical methods to explore prevalence estimates of childhood exposure to IPV, which were reported in 62 studies, citing results of 91 samples from low-income and lower-middle-income countries. The average lifetime prevalence was 29% (past-year prevalence: 35%); almost one in three individuals reported being exposed to IPV during their childhood. Based on 2019 population estimates,22 this amounts to 117 million children in low-income and lower-middle-income countries who have experienced exposure to IPV. We found high levels of heterogeneity across studies. Therefore, results need to be interpreted with caution. We cannot assume that the average prevalence we found is universally valid for the countries we studied. The median prevalence of the studies we reviewed was 26%, with an IQR between 16% and 37% for the lifetime prevalence of childhood exposure to IPV.

To our knowledge, this is the first systematic review of prevalence of children’s exposure to IPV in low-income or lower-middle income countries. A review of child maltreatment from high-income countries2 has shown that 8%–25% of children witnessed IPV. A review from high-income and middle-income countries in the Asia Pacific Region23 reported that 10%–39% of children were exposed to IPV. Given the heterogeneity of the estimates that was also found in the studies conducted in high-income countries, it would be premature to draw conclusions about the relationship between the socio-economic status of the country and childhood exposure to IPV. Poorer economies are potentially less able to invest in social welfare programmes and law enforcement tends to be underfunded, which is likely to be associated with higher levels of IPV. Results from several studies show that economic policies that contribute to reductions in household income and increased financial uncertainty are associated with increases in maltreatment.24

Childhood exposure to IPV continues to receive less attention than other forms of violence, although the issue has gained some visibility in recent years. It is a type of violence that is often not included as a focus of researchers and policy-makers who address either violence against children or violence against women. This is also reflected in international agreements. While physical, psychological and sexual abuse of women and physical and sexual abuse of children are explicitly addressed in the targets of the 2030 Agenda for Sustainable Development, which has been adopted by all United Nations Member States in 2015, the international community did not address the fact that these forms of violence are often linked and that violence against women can also have detrimental effects on children.

Statistically significant differences were found between WHO regions. Childhood exposure to IPV was highest in the Americas and the AFRO Region and lowest in low-income countries of the European Region and the EMRO Region. Factors that could explain the variance between regions include true differences in prevalence influenced by culture-specific social or gender-norms that affect the frequency of occurrence of IPV, whether IPV is occurring in front of children or concealed from children, or the social acceptability for children to admit to being exposed to IPV. Since the items assessing exposure to IPV were not validated across cultural settings, differences in the understanding of the semantic content across cultures could also have affected the differences found between WHO regions.

Although we found prevalence estimates from almost half of the countries that are classified as low-income and lower-middle-income countries, prevalence studies seem to be sparse in large parts of Africa, Maghreb, in countries with civil war and conflict, and in countries with small populations. This can only partially be explained by the fact that we only considered papers that were published in certain languages.

Similar to findings from surveys from high-income countries, we did not find statistically different prevalence estimates between male and female samples.25 26 This finding is surprising, as in many societies, especially when traditional gender norms persist, girls tend to spend on average more time at home than boys.27

High heterogeneity seems to be a shared feature of prevalence reviews on children’s exposure to IPV2 ,23 and on other types of violence against children.28–31 The large variance we found is likely associated with common methodological issues related to how prevalence estimates are derived or due to a true variability of exposure to IPV. We did not find that study characteristics such as the sample size, the median age of the sample, the risk of bias rating or the setting in which data was collected could explain the heterogeneity. There are few analyses of how study characteristics influence prevalence in child maltreatment research, and none in the area of childhood exposure to IPV. Meta-analyses in other areas of child maltreatment prevalence research found that less rigorous sampling strategies and smaller sample sizes were associated with higher prevalence estimates.30 31

There are several strengths of this systematic review. It is the first study to synthesise existing prevalence data on childhood exposure to IPV from low-income and lower-middle-income countries. Although measurement issues make it difficult to derive a global prevalence figure, results of our review indicate that children’s exposure to IPV is a very important public health problem across countries.

Research implications of findings

The large between-studies heterogeneity reported here and elsewhere, highlights the importance of further research to identify and address the sources of such large variance. It would be important to establish to what extent the heterogeneity is due to real variations in childhood exposure to IPV and to what extent it is a methodologic artefact.

Future research would thereby benefit from clear definitions of childhood exposure to IPV. Several researchers have stressed the importance of comprehensive measurement of children’s exposure to IPV.32

Although there is some congruence in the measurement instruments used to assess the prevalence of childhood exposure to IPV, there is no gold standard. It remains to be determined whether the various instruments that are applied are comparable. To improve the accuracy and comparability of items that measure childhood exposure to IPV, instruments should at least specify the type of IPV exposure (physical, emotional, sexual), and in what way the child was exposed (eg, as a direct observer, having overheard someone talk about the abuse, having direct involvement, experiencing negative consequences from abuse in the home).

Few surveys use a similar methodology across countries. A global research effort involving systematic approaches to measuring childhood victimisation would provide important epidemiological information that could assist prevention and intervention efforts.

Practice and policy implications

Our findings show that children’s exposure to IPV is widespread in low-income and lower-middle-income countries. Given that childhood exposure to IPV is linked to a broad range of physical and mental health problems, health risk behaviours and social consequences7–14 33 including in low-income countries,34 healthcare and social service providers should consider the impact that IPV has on children, when providing care and services to survivors of IPV.

Services for child and adult survivors of IPV are commonly not delivered in an integrated manner. Policy-makers should invest in the development of integrated interventions for IPV and evaluate whether they lead to better health outcomes for children, particularly in settings with limited human and financial resources.

The study highlights the importance of investing in the primary prevention of IPV. Reducing IPV has the potential to reduce negative health outcomes among children living in households with IPV. Systematically implementing policies to target major risk factors for IPV, such as strengthening access to education for girls and economic empowerment of women has proven to be effective in reducing IPV.35


Given the large heterogeneity across studies, we recommend caution in drawing conclusions about a global estimate for childhood exposure to IPV. The pooled estimate of the random effects model cannot be interpreted as universal true effect; rather it is the average of survey-specific estimates.

The items that were used to measure childhood exposure to IPV varied between studies. In most studies, measures were used without appropriate cross-cultural validation and adaptation such that comparability of prevalence estimates has limitations.

The majority of the study populations were adults aged 18 and older, who were asked about IPV exposure in their own childhood. Research on other types of child maltreatment and family discord suggests that such retrospective data may be subject to recall bias, which can lead to a systematic underestimation of the prevalence.36

There is substantial variability in the tools and a lack of consensus about the domains that should be assessed in risk of bias assessments of prevalence studies.37 Although the inter-rater reliability was high in this study and previous studies,18 we noted possible limitations in the application of Hoy et al’s risk of bias tool. Some dimensions, which can influence bias were not assessed. These include the sample size and the sampling procedure, which were not assessed in sufficient detail. Sampling techniques can still differ largely in terms of their representativeness. Some studies did not report whether a sample was drawn from the entire population of a country or from a subnational administrative entity. Under-reporting of the applied research methods, which is common, can result in certain domains not being assessed, which can lead to a falsely elevated risk of bias rating.

Many of the estimates were collected from studies whose primary purpose was not the measurement of childhood exposure to IPV. We derived the estimates from general health surveys, such as Demographic and Health Surveys, studies on Adverse Childhood Experiences, or from studies that assessed risk factors for other health conditions. If childhood exposure to IPV or child maltreatment was not reported in the abstract or the full text, the study would not have been identified, which could have led to a risk of bias at the review level.


We conclude that the exposure of children to IPV is highly prevalent in low-income and lower-middle-income countries. The pooled prevalence mirrors global estimates of IPV. From a large number of studies, including those performed in lower-income countries, we know that childhood exposure to IPV can lead to severe and long-lasting health and social consequences. Therefore, healthcare and social care providers should be able to recognise child exposure to IPV, provide first line support, including psychosocial support, address associated mental health consequences and link exposed children with other support services to prevent subsequent impairment.

We believe that the lack of consensus in defining and measuring childhood exposure to IPV is contributing to large variations in reported prevalence rates. Better agreement about definitions and the operationalisation of childhood exposure to IPV as well as consistent use of instruments would be an important step in improving measurement and the ability to compare outcomes.

The findings of this study strengthen the case for further efforts to address childhood exposure to IPV including in low-income and lower-middle-income countries. Considering the severe and long-lasting health and social consequences, the health sector, in collaboration with other sectors, has an important role in raising awareness and addressing the consequences of children’s exposure to IPV.

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statements

Patient consent for publication


We would like to thank Jill McTavish for her review and guidance regarding the search strategies and Kathleen Dobson for reviewing the coding.


Supplementary materials


  • Twitter @mel_kimber

  • Contributors BK developed the research questions. All authors made substantial contributions to the research protocol. BK conducted the searches. BK and MK screened the papers for inclusion and extracted the data for the analysis. MK provided detailed guidance throughout the review process. BK performed the data analysis. TP provided advice on research methods and verified the analytical methods. BK took the lead in writing the manuscript. All authors provided critical feedback and helped shape the research, analysis and draft manuscript. TP and HM provided overall guidance and supervision. All authors approved the final version of the manuscript and agree to be accountable for all aspects of the work and ensure that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. BK accepts full responsibility for the finished work, had access to the data, and controlled the decision to publish.

  • Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors HM is supported by the Chedoke Health Chair in Child Psychiatry.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.