Effects of armed conflict on child health and development: A systematic review

Background Armed conflicts affect more than one in 10 children globally. While there is a large literature on mental health, the effects of armed conflict on children’s physical health and development are not well understood. This systematic review summarizes the current and past knowledge on the effects of armed conflict on child health and development. Methods A systematic review was performed with searches in major and regional databases for papers published 1 January 1945 to 25 April 2017. Included studies provided data on physical and/or developmental outcomes associated with armed conflict in children under 18 years. Data were extracted on health outcomes, displacement, social isolation, experience of violence, orphan status, and access to basic needs. The review is registered with PROSPERO: CRD42017036425. Findings Among 17,679 publications screened, 155 were eligible for inclusion. Nearly half of the 131 quantitative studies were case reports, chart or registry reviews, and one-third were cross-sectional studies. Additionally, 18 qualitative and 6 mixed-methods studies were included. The papers describe mortality, injuries, illnesses, environmental exposures, limitations in access to health care and education, and the experience of violence, including torture and sexual violence. Studies also described conflict-related social changes affecting child health. The geographical coverage of the literature is limited. Data on the effects of conflict on child development are scarce. Interpretation The available data document the pervasive effect of conflict as a form of violence against children and a negative social determinant of child health. There is an urgent need for research on the mechanisms by which conflict affects child health and development and the relationship between physical health, mental health, and social conditions. Particular priority should be given to studies on child development, the long term effects of exposure to conflict, and protective and mitigating factors against the harmful effects of armed conflict on children.


Interpretation
The available data document the pervasive effect of conflict as a form of violence against children and a negative social determinant of child health. There is an urgent need for research on the mechanisms by which conflict affects child health and development and the relationship between physical health, mental health, and social conditions. Particular priority PLOS

Introduction
Conflicts force children and families to leave their homes to seek safety within national borders (internal displacement) and across international borders-nearly two-thirds of the 28 million forcibly displaced children are internally displaced. [6] During flight, children may become separated from their families and are more vulnerable to infections, psychological trauma, and exploitation. [7,8] Experiences of trauma affect children's mental health, as well as that of their caregivers. Poor mental health of caregivers may negatively affect children's physical and mental health, as well as their educational attainment and life opportunities. [8,9] The destruction of educational and economic infrastructure creates conditions of poverty, which may last for generations. Economic and political sanctions deepen this poverty and have detrimental effects on child health and nutrition. [10] Little is known about the impact of armed conflict on children's physical health and development-even estimates of the number of children killed by conflict are lacking. [11][12][13][14] Research has focused primarily on the mental health effects of armed conflict on children and on downstream effects such as displacement. [9,[15][16][17][18][19][20] We undertook a systematic review of the evidence of the impact of armed conflict on children's physical health and child development. Where available, risk factors, mitigating factors, and protective factors were abstracted.

Search strategy and selection criteria
Searches were undertaken in PubMed, Web of Science, CINAHL, EMBASE, Latin American and Caribbean Health Science (LILACS), IndMED, Africa-Wide Information, Open Grey and the New York Academy of Medicine Grey Literature Report from 1 January 1945 to the search date. The initial searches were performed 8-12 June 2015. The PubMed and EMBASE searches were updated on the 24 and 25 April 2017, respectively. The review is registered with PROS-PERO: CRD42017036425.
Our intention was to perform a systematic review and meta-analysis of available data on the physical health and developmental effects of armed conflict on children. During the searches, it became clear that the varied focus, heterogeneous design, and variation in reporting of outcomes by published studies would not support this type of review. The aim of our review was therefore shifted to describe published studies on the effects of armed conflict on child health and development.
Search terms included multiple variants of "child" and "war." Terms for physical health and child development were not used, as inclusion of these terms narrowed the search results and missed relevant papers known to the authors. The search terms used are provided in the web appendix. Medical Subject Headings terms were used when available, and snowball and hand searching was used to identify additional studies.
Screening and full text review was conducted by Ayesha Kadir (AK) and Sherry Shenoda (SS) for all publications using Covidence, [21] an electronic organisational tool for systematic reviews. Inclusion criteria included study population, setting of past or current armed conflict or a region where refugees/asylum-seekers are staying, and exposure of the study population to armed conflict. Armed conflict was defined according to the Uppsala Conflict Data Programme (UCDP)/Peace Research Institute Oslo (PRIO) criteria (Fig 1). [5] The UCDP database was used to identify conflicts meeting criteria for inclusion. [22] We included original research studies that provided data on children ages 0-18 years. Outcomes included physical or developmental morbidity associated with exposure to armed conflict, exposure to violence, and access to basic needs, including health care and education. Studies on mental and behavioural health were excluded unless they also provided data on physical health or child development. Additional exclusion criteria included review papers, studies published prior to 1945, and studies with a median date of data collection earlier than 1940. Studies on terrorism were excluded, as terrorist incidents are not universally associated with armed conflict. The Palestinian-Israeli conflict was considered to be an armed conflict. Studies providing data exclusively on nutrition, perinatal mortality, birth weight, breast and infant feeding, and immunization coverage were excluded; while the evidence remains limited on the scale and nature of the impact of armed conflict on these indicators, child nutrition and maternal and newborn health are broadly recognised as carrying high risk in conflict settings. [23,24] However, if these data were presented together with other child health and development outcomes, then data for all reported child health and development outcomes were extracted. Studies on the effects of exposure to the atomic bomb were excluded, as there are existing reviews on this subject. Post war studies were included if they provided associations of the outcomes with armed conflict. No restrictions were made for sex, geographic location, language, or study design.
The risk of bias was assessed at the study ad outcome levels for each individual study based on the data source, study population, sampling strategy, data collection and analysis methods, and any special characteristics of the population. Studies that were deemed to have unsound or invalid methods were excluded. Given the challenges in obtaining data in conflict settings, studies from single facilities, studies using only facility-based data, and case reports were included. Data from studies meeting inclusion criteria were abstracted onto a data extraction form, including time period, study country and sub-region, identified conflict, study design, reference population, type of exposure, health outcomes, access to basic needs, mortality, and associations between exposures and outcomes. Where available, data were abstracted for protective and mitigating factors on child health outcomes. When possible, authors were contacted for missing data. In the case of queries or differences, an agreement was negotiated between the reviewers. of physical injuries affecting all organ systems, broadly classified as penetrating injuries, blunt trauma, crush injuries and burns. Injuries were attributed to shelling, explosions, collapsing buildings, gunshots, and motor vehicle crashes.
Among injured children who reach health facilities, penetrating injuries are most common. [25][26][27][28][29][30][31][32][33][34][35][36][37][38][39] Penetrating head injury is the most frequent form of head injury among children treated in military combat facilities, accounting for 60-75% of all head injuries and carrying the highest mortality risk. [28,30,32] This pattern of head trauma differs markedly from that observed in peaceful settings, where blunt head trauma predominates. It is important to note that the admission criteria for combat support hospitals, access to military facility care, and care seeking behaviours of people living in combat zones are likely to influence the findings in military studies; Spinella et al documented that a child with a severe head injury had sought care at five other hospitals before presenting to a military facility. [40] Reported mortality from trauma ranges from 2. Three-fold increase in infant and child mortality after start of conflict. Four-fold and five-fold increases in age-adjusted mortality from injuries and diarrhoea, respectively. Regional differences in child mortality were maintained or exacerbated after onset of conflict. 6 Avogo and Agadjanian, [86] 2010 Angola Migrants to Luanda 719 Increased mortality among children whose families were displaced due to war. This effect was strongest during the first year after migration.
7 Barisić et al, [139] 1999 Former Yugoslavia Children with nerve injuries 27 Peripheral nerve injuries in children due to war involved multiple nerves, were located proximally on upper extremities, had complete or severe nerve damage, delayed reinnervation and poor spontaneous recovery outcomes. These patterns differed from children with peripheral nerve injury due to accidents, who primarily had single, partial peripheral nerve injuries that were located on distal extremities. 8 Barnes et al, [140] 2007 Iraq US high school students 121 Children of deployed military personnel had significantly higher BMI than non-deployed and civilian counterparts. Children of both deployed-and non-deployed military personnel had a higher mean HR than children of civilians. 9 Bertani et al, [141]  The adolescents describe social isolation from the mother's ethnic group, as well as being assaulted, shot at and threatened with rape. They also described taking on the role of carer for incapacitated mothers.

Feldman et al,[106] 2013
Israel-Palestine War-exposed and non-exposed children 1.5-5 years 232 War-exposed cohort had lower baseline cortisol levels and less reactivity to stress than the non-exposed cohort. Children's baseline cortisol levels were independently related to maternal baseline cortisol lower maternal reciprocity, and greater maternal psychopathology. Mortality estimates ranged 20-120/10,000/day for unaccompanied children and 100-800/10,000/day unaccompanied infants. High rate of death attributed to diarrhoea. . 53 Green, [51] 2007 multiple Case report 3 3 child victims of torture, reported child labour, slow insertion of a knife into the child's thigh to extract information from the parents, and witnessing torture, including witnessing a parent tortured to death. The children reported recurrent nightmares and school absenteeism. 54 Greene et al, [157] 2014 Iraq Case report 1 3 year old girl with blast injuries to the right arm and chest, who required highly specialised thoracic surgery and was hospitalised for 16 days. 55 Grein et al, [158] 2003 Angola Refugees in 4 camps 6,599 18% of the population was U5. U5 mortality was four times above baseline. Main causes of child death were malnutrition, fever and malaria. Children accounted for one-fourth of deaths related to war violence, and 55% of disappearances. During the war, the rates of child homicide and suicide using weapons more than tripled and unintentional child deaths with weapons increased more than 6-fold compared to pre-war period. After the war, these rates gradually returned to pre-war levels.
(Continued ) Traumatic war exposure in female guardian independently predicted child-reported experiences of abuse in the family. Partner violence between guardians and PTSD-symptoms in male guardians were the major proximal risk factors for childreported victimization, suggesting that war exposure and subsequent trauma may be a mediating factor in violence against children.
(Continued ) Men who were evacuated to foster care in Sweden at age <4 years had mortality risk 1.3 times higher than their counterparts who were not evacuated. There were no other significant mortality differences based on gender, age at time of evacuation, or between evacuation-status discordant siblings. 113 Schiff et al, [171] 2006 Israel-Palestine 7th-10th graders in Herzeliya, Israel 1,150 1/3 respondents were in the proximity during an attack and 40% knew someone who was injured (psychological proximity). Physical and psychological proximity to attacks were significantly associated with alcohol consumption, when controlling for PTSD and depressive symptoms. 114 Schiff et al, [172] 2007 Israel-Palestine Jewish 10th and 11th graders in Haifa 960 Close physical exposure to armed conflict predicted higher levels of alcohol consumption, binge drinking among drinkers, and cannabis use. 115 Schiff et al, [173] 2012 Israel-Palestine Jewish and Arab Israeli 7th-11th graders

4,151
The youth reported high rates of exposure to war events. Cumulative exposure to war events was significantly associated with alcohol and drug consumption and involvement in school violence.

Schlecht et al,[105] 2013
Uganda Displaced Ugandan and Congolese refugee youth 133 Armed conflict resulted in breakdown of traditional community social structure and associated protective marriage practices. Displacement was also associated with social isolation and barriers in access to education. These social changes and challenges were associated with earlier sexual debut without involvement or knowledge of parents/caregivers, teen pregnancy, sexual exploitation of girls, transactional sex. 117 Shemyakina, [92] 2011 Tajikistan Households with school age children across Tajikistan 6,160 Based on 2 surveys (one representative at national level and one at regional and urban/rural level). Children 8-15 years old in conflict-affected area were less likely to attend school. Damage to household dwelling negatively associated with the enrolment of girls. Nationally, men and women of school age during the war were less likely to complete nine grades of schooling compared to their pre-war counterparts. 118  Children accounted for >16% of civilian deaths in non-state controlled areas and >23% of civilian deaths in governmentcontrolled areas. The risk of death from different combat activities varied by location, however, children in all areas were more likely than men to die from air bombardments, shells, ground level explosives, and chemical weapons. 852 children were killed by execution, including execution after torture. 9 Hemat et al, [182] 2017 Afghanistan Trauma patients at single facility 35,647 Paediatric trauma patients at Kunduz Trauma Centre Jan 2014-June 2015: Children accounted for 50% of patients registered in the emergency department and 41% of operated patients. 10 Khamaysi et al, [183] 2015 Syria Case report 5 Report on 5 patients with traumatic bile leaks from war injuries, including two children. More than a third report stigma toward their child from the community and 2/3 reported often seeing their assailant and/or remembering the sexual assault when looking at the child. Stigma and maternal mental health disorder was associated with negative parenting attitudes. Family and community acceptance were associated with adaptive parenting attitudes. 16 Stark et al, [65] 2015 Uganda Congolese and Somali refugees in Kampala

>215
175 In-depth interviews, 40 key-informant interviews and 51 focus group discussions. Children reported discrimination in schools and teachers encouraging xenophobia Conversely, some reported reduced school fees and accommodations made for prayer. Children reported social marginalization in the community, barriers in access to sanitation, assault, and lack of access to health care and legal and protective services. 17 Sullivan et al, [98]   mortality when compared with older children and adults. [40,41] No studies provided data on mortality after transfer or discharge from health facilities. While numerous case studies report children with traumatic amputations, only one study examined the prevalence of disability among war-injured children. This single facility retrospective chart review of 94 children with war-related injuries sustained during the wars in Croatia and Bosnian and Herzegovina found that nearly 40% of the children suffered permanent disability. [43] No studies describe the incidence or prevalence of childhood disability associated with armed conflict or its long term effects on health, development, or life opportunities.
Two studies document the effects of chemical or biological weapons on children. Momeni and colleagues describe the clinical manifestations of mustard gas exposure in a group of children during the Iran-Iraq war.
[46] The paper highlights the difference in presentation when compared with adults, including earlier onset of symptoms, more frequent pulmonary and gastrointestinal symptoms, and predominant face and neck symptoms. Guha-Sapir and colleagues found that children in the ongoing conflict in Syria are twice as likely to die from chemical weapon attacks as adults (OR 2.11, 95% CI: 1.69-2.63). [47] Studies documenting the torture of children report a variety of torture methods that children experienced and/or witnessed. These children suffer physical injuries, a variety of somatic complaints, enuresis, constipation, sleep disorders, and psychological disorders.[47-56] A follow-up study on children who were tortured or whose parents were tortured in Chile found that both the prevalence and frequency of somatic symptoms increased over time after resettlement in Denmark. [55] Indirect health effects Diseases. Exposure to armed conflict is associated with a higher burden of infectious, communicable, and noncommunicable diseases in children. A global study of disabilityadjusted life years (DALYs) associated with civil war found significantly reduced DALYs in children under 14 years for all disease categories, with the most severe reductions in the under 5 (U5) age group. [57] Diseases such as malaria, [58] diarrhoea, acute respiratory infections, and fever [59] are more common and carry higher mortality. [60] A nationwide cross-sectional study in Iraq after the first Persian Gulf War found that age-adjusted mortality due to diarrhoea rose from 2.1 pre-war to 11.9 per 1000 person-years after onset of war. [60] Pregnancy and birth in conflict zones are also higher risk. A study in Tuzla Canton, Bosnia, found significantly fewer live births, increased preterm delivery, and low birth weight during wartime, which normalised again after the war. [61] Displacement due to conflict is associated with crowding, limited access to water and sanitation, and increased risk of infectious and communicable diseases. [62][63][64][65] Environmental exposures. Studies on the effects of combat-related environmental exposures have identified an increased prevalence of birth defects and cancer, as well as ongoing risk of injury due to unexploded ordnance (UXO). The incidence of structural heart defects in Kuwaiti infants rose significantly after the first Persian Gulf War to levels exceeding the international incidence. [66] While the mechanism for this jump in prevalence cannot be identified by the study design, the authors suggest that war-related environmental pollution may play a role. This hypothesis is supported by a large study of birth defects in children of US Gulf War Veterans using active case ascertainment. [67] Conversely, an earlier review of military health records suggests no increased incidence of birth defects among the infants of US Gulf War Veterans. [68] A study in Iraq found a significantly increased rate of leukaemia, particularly amongst younger children. [69] Similarly, Wen and colleagues found significantly increased risk of AML in children whose fathers reported having served in Vietnam or Cambodia. [70] AML risk was further elevated if fathers reported two or more tours.
Landmines and unexploded ordinance (UXO) remain a significant risk for children during conflict and long after combat has ended. [71][72][73][74][75][76][77][78][79][80] The burden of injury due to UXO varies by conflict, however children account for approximately half (42-55%) of injuries from UXO in Afghanistan, Nepal, Eritrea, and Iran. [71,73,74,77,80] The pattern of injuries in children is similar across settings; children more often suffer upper body injuries sustained while playing, going to school, or tending animals. [71-74, 78, 79] Access to basic needs. Whilst the burden of disease increases due to conflict, access to health care becomes more difficult. Children in areas affected by conflict are less likely to receive vaccinations [81,82] and conflicts may contribute to outbreaks of vaccine preventable diseases such as the polio outbreaks during the conflicts in Nigeria and Angola [83,84] and a rubella outbreak in Bosnia and Herzegovina after hostilities had ended. [85] Combat may also hamper vaccination campaigns during disease outbreaks. [84] Several studies have reported that pregnant women displaced due to war were less likely to receive prenatal care or deliver with the assistance of a skilled birth attendant. [61,82,86,87] Studies from the former Yugoslavia describe drastic reductions in the number of health personnel [88] and conditions where children with chronic conditions had less frequent access to medical care and experienced worsening of their condition. [89] Conflict also prevents children from going to school and lowers their overall educational attainment. A large time-series cross-sectional study examining the impact of war in sub-Saharan Africa  found that armed conflict significantly reduced school attendance for both boys and girls, and with an inverse correlation between military expenditure and school attendance. [90] Similar findings have also been described in Palestine, [91] Tajikistan [92] and Colombia. [93] The Colombian study findings suggested this may have a long-term effectadults in areas affected by conflict had lower educational outcomes.
Risk of abuse, neglect, and exposure to secondary violence. Children whose caregivers have been exposed to armed conflict are at increased risk for child abuse and neglect. Studies from Timor-l'Este, Uganda, and in Syrian refugees in Lebanon found that caregivers who suffer from stress or have a mental health disorder related to their exposure to armed conflict have higher rates of child-reported and caregiver-reported child abuse. [87,94,95] The Ugandan study findings suggest that war exposure and subsequent trauma are mediating factors for violence against children. Studies among U.S. military personnel have documented increased rates of physical abuse and neglect in the children of veterans, both during the period of deployment and after return. [96,97] A large study of Californian children in civilian public schools found that the children of US military personnel have higher rates of experiencing violence in school and are more likely to carry a weapon. [98] Social changes. Important changes in societal structure, norms, and roles take place in populations affected by conflict. Children may assume adult responsibilities, including providing for their families and caring for ill or disabled parents. [99][100][101] Several studies described changes in sexual behaviour, with earlier sexual debut and child marriage. [102][103][104][105] Displacement and separation from family may place children at increased risk for exploitation, high risk sexual behaviour, sexually-transmitted illness, and teen pregnancy. [104] Two studies from Uganda describe particular barriers in access to information and health care for adolescents that further compound their health risks. [104,105] Toxic stress and child development. While numerous studies document the multiple kinds of war-related violence that children witness, no studies examine the effect of these exposures on child motor and psychosocial development. Two studies examined the influence of war-related trauma on children's stress physiology. Feldman et al. found that war-exposed children had altered cortisol and salivary amylase response to stress. [106] The same study also found that children's baseline cortisol levels were independently related to maternal baseline cortisol, mother-child relationship, and maternal mental health. Similarly, the Helsinki Birth Cohort Study described altered stress physiology in children who were separated from their parents for a period during WWII. [107] Two other studies on the same birth cohort found that separated girls had significantly earlier onset of menarche than non-separated girls, and that children separated from their parents in early childhood had lower scores on intelligence testing, respectively. [108,109] Children associated with armed forces or armed groups ("child soldiers"). Studies from Sierra Leone, the Democratic Republic of the Congo (DRC), and El Salvador document the extreme kinds of violence that children experience during association with armed forces or groups, including being forced to watch and take part in killing, cannibalism, rape, child marriage, and sexual slavery. [52,99,[110][111][112][113][114] Three studies also describe the process of indoctrinating children and methods used to control and isolate them. The children in these studies describe regular physical and psychological abuse, torture, and the normalisation of violence. [52,113,115] In the words of one child: "After some time, [the violence] became part of me". [113] Several papers describe unwanted teenage pregnancies, gynaecological problems, and sexually transmitted illnesses among child soldiers. [99,111,112] A case report on three former child soldiers in the DRC describes a combined total of 403 medical complaints and 275 physical findings in multiple organ systems. [52] The sheer number and diversity of complaints and somatic findings in these three youth provides insight into the extensive harm wrought by association with armed groups.
Sexual violence. The witnessing, experience, and perpetration of sexual violence and sexual exploitation are documented by numerous papers in this review, nine of which dealt exclusively with these issues. [56,100,104,105,[115][116][117][118][119] All included studies of children associated with armed groups document sexual violence of and by these children. However, the studies that focus exclusively on sexual violence and exploitation, suggest the problem is far more widespread. These studies describe children being: a) threatened with rape, b) raped, including gang rape, c) forced to watch rapes, including rapes of family members, and d) forced to rape others and engage in sexual slavery. They also document "survival-", or transactional sex to obtain basic needs, and early marriage. Documented physical findings include traumatic genital injuries, vesicovaginal fistulae, and pregnancy. Two studies describe the experiences of children conceived through sexual violence, including social isolation and exposure to violence and threats from within their communities. [100,118] A survey of women raising children conceived through conflict-associated rape reported that 66.1% often saw their assailant and/ or remembered the sexual assault when they looked at their child. [118] Mortality More than one-third (n = 64) of studies included mortality as a main outcome. The heterogeneity of design and lack of denominators in much of the reported data preclude the pooling of data. However, numerous studies report significant increases in infant and child mortality during periods of armed conflict, in comparison to peacetime or when compared with peaceful parts of the same country. [21,60,82,[120][121][122][123][124][125][126][127] Most of these studies report on under 5 years (U5) mortality.
A pooled study of 37 datasets from 1985-2001 found that the relative risk of dying during periods of conflict was higher for older children (� 5 years) when compared to those U5. [122] In some countries, U5 mortality actually decreased during the period of conflict. The absolute U5 mortality remained higher than the � 5 years mortality independent of the presence of conflict, suggesting that additional factors unrelated to conflict were contributing to U5 mortality in these populations. [122] This study suggests that there are nuanced and contextspecific factors that lead to conflict-associated mortality during different periods of childhood.
Further examples in support of the importance of context on mortality outcomes is seen in studies by Ascherio and Nielsen that found maternal education is protective against child mortality in conflict settings. [60,126] Avogo et al. found displacement of children due to war carries a higher mortality risk than displacement for other reasons, with the highest risk of death during the year after migration. [86] The study suggests that forced migration due to war exposes children to health risks after resettlement to which other migrant children are less vulnerable. Mortality studies by Hicks [53] and Guha-Sapir[47] document the direct deaths of children due to combat activities in Iraq and Syria, respectively. In addition to deaths from bombs and gunshot wounds, both studies document the torture and execution of children.

Geographic distribution of the literature
There is a notable trend in the geography of published studies. During the period 1946-2016, there were 280 distinct armed conflicts. [4] Papers providing data on child physical health and development are focused on a few specific conflicts and on particular regions of the world. There is a remarkable dearth of data on major past conflicts, such as the Korean War and the Chinese Civil War, and more recently the conflicts in Yemen, Myanmar, Guatemala, Mexico, the Central African Republic, and Pakistan. There is also a notable lack of data on protracted conflicts, such as those in Kashmir and in Xinjiang province, China.  Effects of armed conflict on child health and development these conflicts, and 60% of published studies focus on just six conflict zones. Closer examination of geographical reporting patterns reveals a conflict-specific focus on exposures and outcomes. For example, seven of the nine studies examining sexual violence against children were conducted in the DRC and Uganda. [104,105,[115][116][117][118][119] All four studies from Sierra Leone focus on child soldiers. Of the 45 studies examining child health in Afghanistan and Iraq, half are case studies and chart reviews from military combat facilities.

Discussion
This is the first systematic review of the global impact of armed conflict on child physical health and development. The review's findings reveal the breadth of the problem, its pervasive and sustained impact on child health, and important nuances in the way conflict affects child health. Further, it describes how these effects differ based on the direct and indirect nature of the impact, characteristics of the individual conflict (e.g., types of combat, weapons, displacement, relief efforts), and contextual factors associated with the geographic location of the conflicts. The broad view given by the available evidence is quite disturbing and therefore warrants reporting in this first paper on our systematic review. Further analysis of subsets of the data will be presented in future publications.
It is important to reiterate that this review deliberately excluded papers reporting exclusively on child health outcomes that are known to carry high risk in conflict settings, including Effects of armed conflict on child health and development nutrition, perinatal mortality, infant feeding, and immunization coverage. Most papers that reported on perinatal mortality, birth weight, and infant feeding also reported on other child health outcomes, and were thus included in this review. However, the relative paucity of papers in our review providing data on these topics suggests that the evidence for the effect of armed conflict on maternal and newborn health is lacking. This finding is consistent with other recent reviews on the subject. [128,129] Additionally, there is a large literature on the effects of forced displacement on child mental health, [19,20] much of which was not identified by the searches. Finally, studies that were published as books were not included in the review; this is likely to have precluded the examination of a number studies, particularly those undertaken before 1990.
The grey literature database searches retrieved only 382 citations, none of which met inclusion criteria. This review is therefore limited to the peer-reviewed literature; this is an important limitation because traditional public health research in conflict settings is limited by security concerns, population movements, the destruction of public health infrastructure, and the disruption of routine data collection. [13] Furthermore, most of the studies rely on health facility data and are thus likely have selection bias, as they report on those patients who were able to access care in a facility where sufficiently complete records were maintained and were accessible to the researchers. Much of the available information about the health of populations living outside of refugee or IDP camps in areas of active conflict is not reported in the peerreviewed literature and is therefore missing from this review. Among studies providing data from areas experiencing active conflict, the majority were retrospective chart reviews. Of the 11 population-based studies in settings with active conflict, 10 used a cross-sectional and/or qualitative design. [60,64,81,106,126,[130][131][132][133][134] While there were no marked differences in the findings from these studies compared with studies in refugees in neighbouring regions, the design of the studies make it difficult to draw further conclusions.
Through its pervasive harmful effects on children, armed conflict is a negative social determinant of child health. Numerous studies have documented that adversity during childhood can alter the architecture of the brain and neuroendocrine function, leading to alterations in learning, behaviour, and physiology, in turn predisposing the developing child to maladaptive behaviours and ill health throughout the life course. [135] The findings on stress physiology reported in this review are in keeping with previous studies on toxic stress and childhood adversity, and therefore have significant implications for generations of children growing up amid armed conflict.
For each individual child exposed to armed conflict, the way in which this exposure affects the child's health is likely to be determined by a number of factors including genetic predisposition, physical health, mental health, development, behaviour, caregiver physical and mental health, forced displacement, and social arrangements. The influence of caregiver mental health on the physical and mental health of conflict-affected [9] and forcibly displaced children [8] is well-described. The findings of Feldman et al [106] suggest that in war-affected children, caregiver attachment and mental health may play a mediating role, and is therefore a potential area for intervention to mitigate the effects of armed conflict on children.
Numerous studies in this review report on physical health, mental health, and social conditions in the same population, suggesting they are interrelated. Other studies describe alterations in social arrangements due to conflict, such as disruption of communities and barriers in access to education, which affect child and adolescent health and wellbeing. The relationship between physical health, mental health, and social conditions in conflict-affected children is an area in need of further research, and that may provide insight into aggravating factors, mitigating factors, and ways to promote good health and resilience.
The overlap of different typologies of violence against children in the context of armed conflict is an important finding that merits further attention. In addition to risks for being injured, tortured, and/or a witness to or participant in combat, children with direct and indirect exposure to conflict are also at increased risk for other forms of violence, including abuse and neglect, [87,[94][95][96][97] community and school violence, [98] and domestic violence. [94,95] Furthermore, Nelson et al describe an increase in civilian-perpetrated rape of children during a period of conflict in the DRC. [117] While these forms of violence against children are extensively described in grey literature reports, the association of these forms of violence with armed conflict points to an important need for child protection after the exposure to conflict has ended.
There is a compelling need for further research to improve our understanding of the medium and long term effects of the exposure to armed conflict as a form of violence against children. Areas in need of further study include global child development, children with chronic diseases, children with disabilities, and adolescent health. Of particular importance is the need to improve our understanding of the relationship between physical and mental health and social conditions in conflict-affected children and factors that protect and mitigate the harmful effects of armed conflict (Fig 6). In addition to informing the development of evidence-based interventions to treat and mitigate the harmful effects of conflict on children, such knowledge would improve our understanding of how to advance the health and wellbeing of adults who have experienced armed conflict and other adversities during childhood and adolescence. Such information would be of great use for peace research and conflict resolution.
The studies included in this review demonstrate that it is possible to study the ways that armed conflict affects child health and development using both quantitative and qualitative methods. However, the vast majority of the literature is descriptive because traditional epidemiological research is difficult to undertake in settings with insecurity and disrupted health systems. A study published after completion of this systematic review brings attention to a potential way to improve the study of conflict and child health and highlights an important role for humanitarian organizations working in settings where routine data is no longer available. Meiqari et al [136] describe paediatric data collected by Médecins Sans Frontières (MSF) in Tal-Abyad and Kobani, Syria. In spite of challenges, which included population movements and abrupt closing and opening of clinical services due to the shifting front lines of the conflict, the authors describe the epidemiology of a large cohort of children who received care in MSF facilities for periods between 2013-2016, including 27,742 children U5 seen in outpatient clinics, 4672 children under 18 years admitted for inpatient treatment, and a measles epidemic response. Meiqari and colleagues provide important data about child health epidemiology in the study region, as well as information on care-seeking behaviours in a conflict zone notorious for attacks on health care facilities. [137] By supporting colleagues working in conflict zones and developing epidemiological methods to better study health in low-security settings, we can improve the ability to conduct research that can meaningfully improve the care and outcomes of children affected by armed conflict.

Conclusion
This systematic review documents the pervasive effect of armed conflict as a form of violence against children and negative social determinant of child health. The studies serve as a record of the continuing occurrence of the six grave violations of children's rights, which include the killing and maiming of children; recruitment or use of children as soldiers; sexual violence against children; abduction of children; attacks against schools or hospitals; and the denial of humanitarian access for children. There is an urgent need to improve our research on the mechanisms by which conflict affects child health and development and the relationship between physical health, mental health, and social conditions. Priority should be given to studies on child development, the long term effects of exposure to conflict, and protective and mitigating factors against the harmful effects of conflict on children. Collaboration with partners across sectors and incorporating a child rights perspective into research can improve both our understanding of the effect of conflict on child health as well as our response to their needs.