This study explores the perceived consequences of armed conflict on MRH services and outcomes in two post-conflict countries in sub-Saharan Africa. In the paragraphs that follow we discuss our main findings with respect to the existing literature on MRH services and outcomes in conflict, post-conflict, and other crisis settings.
Armed conflict as the cause of limited access to and poor quality of maternal and reproductive health services
While armed conflicts are known to negatively affect access to and quality of health services, including MRH services, the main mechanisms vary from one setting to another. The main channels we observed in our study have also been reported elsewhere. In Nepal, Ghimire and Pun  have reported similar channels as a result of the Maoist insurgency, where health facilities were attacked and destroyed, rendering them inoperable, and some health providers were targeted by armed groups for perceived support of their adversaries. Furthermore, during the war in Bosnia-Herzegovina, health services, especially those supporting maternal and child health were badly disrupted, with about 35% of facilities destroyed or badly damaged . In post-conflict Somaliland, Leather et al.  observed that the conflict also inflicted serious damage on the health infrastructure and caused the death and migration of many health workers. During the Chiapas armed conflict in Mexico, Brentlinger et al.  also reported a wide range of similar channels to those observed in our study. In addition, they found that ambulances were seldom available to transport patients to functional health facilities, and pregnant women who developed complications after sunset were rarely transported because travel at night was very unsafe, with harassment, assault, or extortion by armed persons at roadblocks a common phenomenon. Also, while we found reports of bias in the delivery of health services on ethnic basis at our study site in Burundi, the Colombian study also reported cases of bias, but on political and religious affiliation. Although reports of such practices are not widespread in the public health literature, healthcare delivery in some states emerging from ethnic conflict have also been shown to be segregated on ethnic lines [26-29]. Despite the fact that health providers are expected to be impartial in the provision of healthcare, it is challenging to establish effective mechanisms to enforce this during crisis situations like armed conflicts.
Furthermore, in the conflict in Eastern Myanmar, the targeting of medical staff, buildings, and supplies has also been reported . The targeted killing of health care providers and attacks on health infrastructure during armed conflicts appears to be a growing and worrying tactic in modern day warfare. For example, a 2011 report  published by the International Committee of the Red Cross (ICRC) revealed that violent lethal attacks on patients, health care workers and facilities, and on medical vehicles are widespread in many conflict settings and pose a serious concern to accessing health care in such settings. Another ICRC report  showed that in 22 countries affected by armed conflicts and other emergencies, 921 violent incidents affecting health care were reported over the course of 2012, included 150 killings and 73 kidnappings of health care providers.
In our study, most of these mechanisms were common across the study sites but a few were unique to each study site. Notably, the channel of targeted killing of health providers and favouritism in the delivery of services was reported only in Burundi, while the mechanism of abduction of health providers was reported only in Northern Uganda. These differences appear to reflect aspects of the conflict and the level of coordination of health services within the affected areas. The very strong ethnic character of the conflict in Burundi may have engendered attack of health providers across ethnic lines, as health providers were perceived as key ‘lifesavers’. With such practices common, health providers who continued to provide services would have preferred to offer such services mainly in areas where they felt their safety and security was guaranteed, and that might be within their own ethnic community. In Northern Uganda, since the conflict was largely fought as an insurgency, with rebels constantly on the run, abducting health providers from nearby health facilities to cater to rebel health needs would appear to have been a war strategy.
With a strong likelihood that access to health facilities might be compromised during times of conflict, alternative strategies of delivering health care need to be explored by the relevant public health actors on the ground. These may include, but are not limited to, the provision of mobile health services to affected populations, and negotiation among the various warring parties for emergency ceasefires to enable health services to be provided to affected populations. Furthermore, the use of local community structures and organisations in the provision of basic health care could also be explored. This may include the training and equipping of traditional birth attendants and community health workers.
Armed conflict as a cause of poor maternal and reproductive health outcomes
Findings similar to ours have been reported in other conflict and post-conflict settings. During the Chiapas armed conflict in Mexico, Brentlinger et al.  observed a substantial increase in the maternal and perinatal mortality ratios associated with increased home deliveries, and difficulty in accessing emergency obstetrical care. The authors also found that both home delivery and mortality were associated with higher levels of intra-community division based on political and religious affiliations, and the primary providers of ANC services were traditional birth attendants. In Eastern Myanmar where an armed conflict between various separatist groups and the military regime is ongoing, with very high levels of maternal mortality comparable to those in our study sites, Loyer et al.  associated this to a combination of health system-, conflict- and politically-related factors. They found that lack of access to and availability of maternal healthcare services is the main cause of maternal deaths. They also noted that while conflict-related factors are difficult to link directly to maternal deaths, these factors, especially human rights violations such as forced labour, soldier violence, theft or destruction of food supplies, injuries from landmines, and forced displacement appear to increase the risk of illness and death in pregnancy. Similar observations have been reported in other conflict and post-conflict regions including Sri Lanka [33,34], and Bosnia and Herzegovina . O'Hare and Southall  have equally reported that sub-Saharan African countries that have recently experienced conflict have higher levels of maternal deaths, coupled with lower levels of skilled-attended births. These findings suggest that a key strategy to reduce maternal morbidity and mortality in such settings is to ensure access and availability of maternal health services, including quality emergency obstetric and neonatal care signal functions.
Our observations equally suggest that the protracted nature of the conflict in Northern Uganda has enhanced the spread of HIV/AIDS. The relationship between armed conflict and the transmission and spread of HIV remains complex and contentious. Iqbal and Zorn  found a positive relationship between armed conflict and the prevalence of HIV/AIDS among 43 African countries from 1997 – 2005. On the other hand, Spiegel et al.  have undertaken a systematic review on the prevalence of HIV infection in some conflict-affected and displaced people in seven countries in sub-Saharan Africa and their findings are contrary to those reported by Iqbal and Zorn. To better understanding the effect of conflict on HIV/AIDS prevalence Mock et al.  have proposed a number of contextual factors that may enhance the transmission and spread of the infection. These include increased interaction among military and civilians; increased levels of commercial or casual sex; decreased availability and utilisation of reproductive health and other health services; decreased use of means to prevent HIV transmission; and increased population mixing following large internal population movements among others. The presence of these factors in an area where the prevalence of HIV/AIDS is already high can serve to push up the prevalence in the general population. This seems to be the scenario that existed in Northern Uganda during the over two decades of conflict. A recent study undertaken in Northern Uganda supports our findings. Patel et al.  reported an overall prevalence of 12.8% among young men and women. According to Westerhaus et al. , the major factors that enhanced the spread of HIV/AIDS during the insurgency in Northern Uganda were the mass abduction of children into the rebel Lord’s Resistance Army (LRA), the phenomenon of night commuting, and the existence of IDP camps. Although concerns around HIV/AIDS were not very common among our participants in Burundi, one of the policy makers interviewed was concerned that this is a neglected domain that, if not well addressed, may turn into a serious problem in the future. With a number of contextual factors that could engender the spread of HIV and other STIs in both Northern Uganda and Burundi, including the current refugee crisis in South Sudan, effective interventions need to be put in place to stem the spread of these diseases.
The reported growing concern of SGBV observed in our study confirms earlier findings in Burundi and Uganda. A national survey on attitudes towards intimate partner violence in Uganda found that more than half of the men and about three-quarters of the women expressed attitudes that were supportive of wife beating . The survey also found that half of married women had experienced intimate partner violence while 41% of men reported being perpetrators of intimate partner violence. A recent survey found that 27% of women in Kampala had experienced physical and sexual violence from an intimate partner . The prevalence of SGBV in Burundi for the past years has also been high, a situation largely associated with the influx of returning refugees and displaced persons, the presence of large demobilized ex-combatants, a high prevalence of female-headed households, widespread lack of economic opportunities and a general breakdown in social norms . Furthermore, the moral degradation and normalization of violence that characterised the war has led to a situation where many people perceive sexual violence as acceptable . In another post-conflict setting in Liberia, Allen and Devit  have equally observed very high levels of intimate partner violence including physical abuse, sexual and verbal abuse and economic abuse, with some level of acceptability of these practices within the study population. While the main current perpetrators of SGBV in our study settings are intimate partners or someone within the family, during the conflict, violators also included warring actors. The reported high levels of SGBV in these sites might be linked to a growing knowledge gap between couples, where most educational interventions by NGOs and other policy makers have largely been focused on women and girls, leaving out the men in a disadvantaged or vulnerable position. A number of stakeholders expressed concerns that most interventions and programmes are disproportionately biased towards women. Another possible explanation for the high levels of SGBV might be that the economic empowerment programmes targeting women might have had the unintended effect of creating more assertive women who are the main providers for the household, creating a complex of inferiority within some men who might exercise this through SGBV. This is very much in agreement with a recent report on conflict analysis in Northern Uganda which observed that changing gender relations during and after the conflict have contributed to high levels of domestic and sexual and gender-based violence in the region . While most traditional interventions to reduce SGBV have largely been focused on women’s empowerment through education/training, economic empowerment through micro-finance programmes, and the provision of legal support, it is also important to explore and address some of the root causes of this phenomenon on the side of the perpetrators. Providing interventions that can meaningfully improve the economic and psychological wellbeing of the perpetrators also needs some attention. Failure to explore and address these issues might only go to further entrench the existing pattern of abuse of women and the emergence of other forms of violence against women.
High levels of teenage pregnancy are a commonly reported problem in certain conflict and post-conflict settings. In some refugee camps hosting Burundian refugees in Tanzania, Märta  reported a similar pattern, with the high prevalence associated with low education attainment, breakdown of the culture, poverty, and unstable family relations. In post-conflict Liberia, Kennedy et al.  and Atwood et al.  observed that in-school adolescent girls were not only vulnerable to unplanned pregnancies, but also HIV/AIDS and other STIs; high levels of sexual activity, early sexual debut, and unprotected and risky sexual practices including transactional sex. Within some post-conflict areas in Eastern Uganda, Muhwezi et al.  have observed vulnerability to high-risk sexual behaviour among the population, including transactional sex, early and forced marriage, and sexual predation. They equally observed that these practices were associated with a high concentration of people in camps, where idleness and unemployment were very common. Rujumba and Kwiringira  and Westerhaus et al.  have equally linked the conflict in Northern Uganda to the growing phenomenon of prostitution reported in our study. They associated this with a rise in moral decadence, loss of property and livelihood, and increasing poverty that characterised life in overcrowded IDP camps. The economic hardship that characterised the war in Burundi has equally facilitated and perpetuated the practice of prostitution. With very high levels of violence against women and insecurity in Burundi during the war, some women, especially widows, perceived prostitution and concubinage as a safer option rather than being alone, where they may be more exposed and vulnerable to abuse and violence by unknown delinquents, armed gangs and bandits . These findings strongly point to the long-term impact of conflict on the breakdown of social and cultural norms and practices, where practices that used to be considered culturally unacceptable are becoming the norm. Addressing such issues will require a complex set of interventions that not only address the root causes of such practices, but also the predisposing factors.
The high total fertility rate reported in our study was expected, as Burundi and Uganda currently have among the highest fertility rates in the world. A recent study  also found that sub-Saharan African countries that have recently experienced armed conflict have a higher fertility rate compared to those that have not experienced conflict. Our findings are also supported by a recent study in post-conflict Rwanda that reported an increase in total fertility following the genocide, and this was associated with a strong replacement effect among the general population . A strong replacement effect, coupled with a strong cultural desire for large family size and a low uptake of family planning services could possibly account for the high total fertility within our study settings. The replacement effect may be stronger in ethnic conflicts, especially in contexts where groups are competing for population size as might arguably be the case in Burundi. Data from the Population Reference Bureaub, however, shows a slow and gradual decline in the total fertility rate in Burundi and Uganda from 1970 to 2013, with similar levels of decline across the period. From 1970 to 2013, TFR has only declined from 7.3 to 6.1 and from 7.1 to 5.9 in Burundi and Uganda respectively. This might suggest that the replacement effect is not only limited to ethnic conflicts.
Armed conflict as a route to improved access to health care
On a positive note, we observed that a number of respondents from Northern Uganda felt that staying in a government-recognised IDP camp within the conflict-affected region enhanced their access to basic health services. This was because some of these camps were attached to a health facility and better resourced with support from humanitarian organisations. Some facilities even offered 24-hour services that also facilitated the management of emergencies. While this improved access to health care was limited only to a small segment of the population, especially those relocated from remote rural areas where access to basic healthcare was very poor before the conflict, a huge segment of the population experienced a deterioration in access to basic healthcare during the conflict. Chan and Kim  have also reported similar findings among smaller IDP camps in Kashmir, Pakistan following an earthquake. They observed that residents in smaller unofficial camps had worse health outcomes; had less access to information, medical services and medication; had the largest average family size; and received the least assistance and resources to sustain livelihood compared to residents in smaller official camps. However, among the larger IDP camps, the health outcomes were similar between the residents in the official and unofficial camps. This is the same phenomenon that some of our respondents in Northern Uganda experienced during the armed conflict, where residents in official IDPs camps had better access to health services than in their communities of origin prior to displacement. Relatedly, studies among refugee populations have also reported better health outcomes in camp populations compared to both non-camp populations in the area of refuge, and populations in the area of origin [55-57]. Additionally, Howard et al.  found that among Liberian and Sierra Leonean refugee camps in Guinea, the knowledge and use of contraceptives was much higher than for the populations at large in Liberia, Sierra Leone, or Guinea. These findings highlight the opportunities presented by official and government-recognised settlements for displaced populations as important platforms to facilitate access to health care for large populations, with relatively fewer logistic challenges. Furthermore, it also draws attention to the need to improve the delivery of health care to populations in non-camped and unofficial settlements.
Overall, our main findings and contributions to the research on MRH in post-conflict settings are as follows. Firstly, armed conflict is a major cause of limited access to and poor quality of MRH services and the mechanisms through which this plays out are context-specific and vary from one armed conflict to another. Consequently, initiatives to improve access to and quality of MRH services in conflict and post-conflict settings should consider the main contextual channels contributing towards this phenomenon. Secondly, armed conflict negatively affects a range of MRH services and outcomes at the health system and individual levels. In this regard, interventions or packages of interventions for improving MRH in conflict and post-conflict settings should be developed and implemented, taking these factors into consideration, focusing on the most badly affected services and outcomes. Thirdly, while contextual factors like living in a government-recognised and well-resourced IDP camp can engender the development of poor attitudes and practices at the individual and community levels, including poor and risky sexual practices, it can equally serve as a platform to substantially improve access to basic health services. As such, while ensuring the availability of health facilities within camp settings, it is important for camp planners to explore strategies that will reduce the mediating factors that engender the growth of poor attitudes and practices that may further endanger population health.
Strengths and limitations of study
The effect of armed conflicts on MRH services and outcomes is an issue of global importance, especially as conflict-affected countries are lagging behind in the attainment of the MDGs. However, there is a relative lack of research on the issue. Working with study sites that experienced different forms of armed conflict provides a ‘richer’ and more extensive appreciation of the effects of armed conflict on MRH services and outcomes compared to limiting the study to one study site. Additionally, the choice of healthcare providers, policy makers, and staff of organisations involved in health systems support and strengthening as research participants, complemented by perspectives of MRH clients, provides a more comprehensive picture of how the conflict affected health outcomes and services. Each of these participant groups contributed their unique experiences to the study that could not have been achieved with only one category of participants.
Several limitations were identified. The women participants recruited for the study were living within the catchment areas of some local health centres or had regular weekly access to basic healthcare services through mobile outreach clinics. We were unable to recruit women participants in highly disadvantaged remote areas that were not regularly served with basic health services. As such the perspectives of that group of women are not well captured in our study. In addition, being a qualitative study with non-random sampling of research participants in sites with unique contextual factors, the findings are not necessarily generalisable to other conflict and post-conflict settings. Another limitation is the lack pre-conflict and conflict baseline data to enable us to better compare with the current situation. However, with the perennial problem of lack of reliable data in conflict-affected countries and a general poor state of data collection in our study sites, we could only depend of the perspectives of the research participants. Additionally, considering that the time between our interviews and FGDs and the end of the conflict is over 8 years, combined with the non-random sampling method to recruit the research participants, any assessment of the effects of conflict on MRH services and outcomes may be subject to recall bias.