Evaluations of reproductive health programs in humanitarian settings: a systematic review

Provision of reproductive health (RH) services is a minimum standard of health care in humanitarian settings; however access to these services is often limited. This systematic review, one component of a global evaluation of RH in humanitarian settings, sought to explore the evidence regarding RH services provided in humanitarian settings and to determine if programs are being evaluated. In addition, the review explored which RH services receive more attention based on program evaluations and descriptive data. Peer-reviewed papers published between 2004 and 2013 were identified via the Ovid MEDLINE database, followed by a PubMed search. Papers on quantitative evaluations of RH programs, including experimental and non-experimental designs that reported outcome data, implemented in conflict and natural disaster settings, were included. Of 5,669 papers identified in the initial search, 36 papers describing 30 programs met inclusion criteria. Twenty-five papers described programs in sub-Saharan Africa, six in Asia, two in Haiti and three reported data from multiple countries. Some RH technical areas were better represented than others: seven papers reported on maternal and newborn health (including two that also covered family planning), six on family planning, three on sexual violence, 20 on HIV and other sexually transmitted infections and two on general RH topics. In comparison to the program evaluation papers identified, three times as many papers were found that reported RH descriptive or prevalence data in humanitarian settings. While data demonstrating the magnitude of the problem are crucial and were previously lacking, the need for RH services and for evaluations to measure their effectiveness is clear. Program evaluation and implementation science should be incorporated into more programs to determine the best ways to serve the RH needs of people affected by conflict or natural disaster. Standard program design should include rigorous program evaluation, and the results must be shared. The papers demonstrated both that RH programs can be implemented in these challenging settings, and that women and men will use RH services when they are of reasonable quality.


Introduction
Increased attention to the reproductive health (RH) needs of people affected by armed conflict or natural disaster began in the mid-1990s with a few key events. The Lancet published an editorial identifying family planning as a complete gap in services for refugees [1]. The groundbreaking report Refugee Women and Reproductive Health Care: Reassessing Priorities highlighted how the health of refugee women fleeing war was further threatened by near absence of reproductive health services [2]. The 1994 International Conference on Population and Development in Cairo specifically recognized the rights of displaced populations to RH [3]. This led to the formation in 1995 of the Inter-Agency Working Group on RH in Crisis (IAWG), a consortium of non-governmental organizations (NGO), donors and United Nations (UN) agencies, to advance RH services in humanitarian settings. In 1999, the IAWG developed the Inter-Agency field manual on reproductive health in humanitarian settings to provide technical and program guidance to field staff [4].
In 2004, the IAWG completed a global evaluation of RH in humanitarian settings at field, agency and global levels. The evaluation found that more RH services were available than a decade earlier, although major gaps remained in most of the technical areas, with gender-based violence as the least developed technical area. Although RH services were somewhat more available for refugees living in camps, they were largely absent for internally displaced (IDP) and non-camp populations [5]. Adolescents were underserved, and safe abortion was not even assessed. The global evaluation identified a need to improve RH data collection to ensure that useful data were collected and properly interpreted, as well as for more rigorous program evaluations.
From 2012-2014, another ten years on, the IAWG conducted a second global evaluation of RH in humanitarian settings. This systematic review, one component of the 2014 global review, sought to explore the evidence regarding RH services provided in humanitarian settings. Are RH programs in these settings being evaluated? Do the programs work? What is the quality of the evaluations? Which RH services receive more programmatic and financial attention based on program evaluations and descriptive data?

Search strategy
This literature review summarized peer-reviewed papers published since the last global evaluation (between 2004 and 2013) that were identified via the Ovid MEDLINE database, followed by a PubMed search to pick up more recent papers not currently indexed. In addition, references for included papers were cross-checked to ensure that all relevant literature was identified and included. A combination of terms describing conflict and natural disasters were used with terms describing RH under the broad categories from the Inter-agency field manual on reproductive health in humanitarian settings of maternal and newborn health, family planning (FP), gender-based violence (GBV), HIV/AIDS and other sexually transmitted infections (STIs), safe abortion and adolescent reproductive health. Searches were limited to papers published in English. This initial search was broad and intended to capture all papers on RH in humanitarian settings. Papers on quantitative evaluations of RH programs, including experimental and non-experimental designs that reported outcome data were included. Descriptive quantitative studies with no specific health intervention identified and no outcomes or outputs reported (e.g., studies that reported only descriptive or baseline data) as well as purely qualitative papers were excluded. Studies were not excluded on the basis of their quality. Other inclusion and exclusion criteria are detailed in Table 1. Papers excluded under these criteria but that reported descriptive or prevalence data were logged to permit comparison of the sectoral spread of evaluation papers (the focus here) and broader prevalence or descriptive papers.

Quality assessment of the papers
The quality of each included study was assessed using criteria from the STROBE checklist for observational studies or the CONSORT checklist for clinical trials [6,7]. Papers were assigned a rating of high, medium or low quality based on the number of met criteria in a list adapted from these checklists.

Results
The search strategy yielded 5,669 papers after duplicates were removed; 5,310 were excluded based on a review of the title. Of the 359 papers for which abstract or fulltext review was conducted, 323 papers were excluded, leaving 36 papers describing 30 programs (Figure 1). Of the 36 papers, 25 described programs in sub-Saharan Africa, six in Asia, two in Haiti and three reported data from multiple countries and continents. Some RH technical areas were better represented than others: seven papers reported on maternal and newborn health (including two that also covered FP), six on FP, three on GBV, 20 on HIV and other STIs and two on general RH topics ( Table 2). None of the papers described safe abortion or post-abortion care programs, and five of the papers described HIV prevention programs targeting adolescents. Only six papers were classified as high quality while the majority was classified as medium quality or low quality. Fewer than half (16) of the papers Papers that reported on other reproductive health topics (e.g., female genital mutilation, forced or early marriage, reproductive cancers)

Types of Papers/ Data
Quantitative evaluations of RH programs or services, including experimental and non-experimental designs that report outcome data Descriptive quantitative papers with no specific health intervention and no outcomes (e.g., reporting only descriptive or baseline data); purely qualitative papers Settings Humanitarian crises in conflict, post-conflict or natural disaster settings in lower or middle income countries Papers in locations that were not affected by armed conflict or natural disaster; that were more than ten years post-conflict; disaster settings in higher income countries

Types of publications
Papers in peer-reviewed journals Letters, editorials, commentaries; grey literature; review papers (although these were screened for references) reported comparison data, either in the form of pre-and post-intervention measures or intervention and comparison groups. Table 3 provides a summary of the included papers.
Of the 323 papers reviewed and excluded, 93 papers reported descriptive or prevalence data on RH in crisis settings. Again, some RH technical areas were better represented than others: 20 papers on maternal and newborn health (including one that also reported on FP and one that also looked at GBV), four on FP, 32 on GBV, 27 on HIV or other STIs (only six of which mentioned other STIs), seven papers on general RH and five on adolescent RH (specifically HIV, GBV or FP) ( Table 2).

Maternal and newborn health
Seven of the 36 papers described evaluations of maternal and newborn health programs, including two programs that also addressed family planning. The papers covered a range of topics including emergency obstetric and newborn care (EmONC), antenatal care (ANC) and the training of traditional birth attendants or community health workers (CHWs) to improve maternal health outcomes.
Two papers described the outcomes of programs to improve EmONC services, the first for Afghan refugees in Pakistan [8] and the second in humanitarian settings in nine countries [9]. Although not all supported facilities met the WHO criteria of fully functional EmONC facilities [10], the papers reported greater availability post-intervention of EmONC services 24 hours a day and subsequent increased use of those services in most facilities. The authors of both papers described challenges in calculating the UN process indicators for EmONC a at baseline [10], primarily due to the absence of key data from delivery registers; however, both reported these indicators at endline.
Other program approaches to improve maternal and newborn health involved training mobile health workers to provide elements of basic EmONC plus blood transfusion and ANC in eastern Burma [11]; seconding refugee health workers to health facilities serving the refugee population and training refugee women to promote RH in the community in Guinea [12]; and training CHWs in Afghanistan to strengthen the link between the community and formal health services [13]. All three papers reported increased use of skilled birth attendants postintervention. The Afghanistan study, however, found that only the presence of a female CHW was associated with increased skilled birth attendance; the association was absent with male CHWs. One paper assessing the effectiveness of baby tents (clean spaces to support mothers to practice healthy infant feeding) established in Haiti found that 70% of babies less than six months old were exclusively breastfed and 10% of non-exclusively breastfed infants moved to exclusive breastfeeding while enrolled [14]. Finally, an evaluation of a home-based lifesaving skills training for traditional midwives in Liberia found that midwives' knowledge improved from pre to post training and remained stable one year later [15].

Family planning (FP)
Six papers described FP programs, including two that also described maternal and newborn health outcomes. Programs used different strategies to improve FP use: providing the full range of FP methods, including longacting and permanent methods, via mobile clinics and strengthening health centers' provision of short-and long-acting FP in northern Uganda [16]; training mobile health workers to provide short-acting methods in eastern Burma [11]; seconding refugee providers to health facilities serving refugees to provide FP and training female CHWs to promote FP use in Guinea [17]; and training CHWs to conduct FP education and provide  short acting methods in Afghanistan [18]. All four papers reported that contraceptive prevalence increased from baseline or was higher than national levels. Additional papers found that the presence of a female CHW was associated with higher FP use in Afghanistan [13], and that contraceptive use was higher among Afghan refugee women in Pakistan who received subsidized health services than among those with access to un-subsidized services [19].

Gender-based violence (GBV)
Although the literature search included broader terms related to GBV, all three included papers focused specifically on care for survivors of rape. Two papers reviewed the effectiveness of psychosocial interventions for survivors. A randomized controlled trial in the Democratic Republic of the Congo (DRC) on the effectiveness of group cognitive processing therapy versus individual support to female survivors of rape found that those who received group psychotherapy showed greater improvement in depression, anxiety and post-traumatic stress disorder (PTSD) symptoms six months after treatment compared to those in the control group [20]. The second paper found that the global functioning of survivors in the Republic of Congo improved following post-rape psychological care, and improvement was maintained one to two years later although high loss to follow up weakened these results [21]. The third paper reviewed the effects of a multi-media training tool for clinical care for rape survivors on the knowledge, attitudes and practices of health providers in four conflict settings [22]. The authors found that although negative attitudes towards survivors did not significantly change, respect for patient rights increased and provider practice improved from pre-training to three months post-training.

HIV and other sexually transmitted infections (STIs)
More papers (20) focused on HIV and other STIs than any other RH component; however only three of these reported on STIs other than HIV. Three papers reported results of retrospective record reviews to evaluate programs to prevent mother to child transmission of HIV (PMTCT), two in northern Uganda and one in a refugee camp in Tanzania. One program found that higher proportions of HIV-positive pregnant women identified in ANC used anti-retroviral prophylaxis in northern Uganda compared with the national average [23]. The other two programs reported high numbers lost to follow-up before completing infant HIV testing at 18 months. In one study, this was primarily due to a lack of understanding of its importance and infant death; incomplete or no ARV prophylaxis, early weaning and prolonged breastfeeding were associated with increased risk of loss to follow-up and infant death [24].
In the final study, more than two-thirds of the HIVinfected women were repatriated to their home country before delivery; among those who delivered in the camp, nevirapine uptake was 98% [25]. Eight papers reported the outcomes of anti-retroviral therapy (ART) programs for HIV-positive adults or children in East Africa, Haiti and globally. Three papers found that ART patients in northern Uganda had mortality rates and adherence comparable to or better than ART patients in stable settings or who were not displaced [26][27][28]. Similarly, a review of the data from 24 ART programs in conflict or post-conflict settings found that patient outcomes were comparable to those in stable settings [29]. Five papers examined the effect of a crisis on ART programs: the post-election violence in Kenya in early 2008 [30][31][32], acute conflict in DRC in 2004 [33] and the earthquake in Haiti in 2010 [34]. Notably, although the papers found higher rates of treatment interruption immediately post-disaster, generally services were quickly re-established and patient attendance and adherence rebounded soon after.
Eight papers reported HIV and/or STI knowledge, attitudes and behavior results following HIV prevention programs. Two papers reported on a group randomized controlled trial to evaluate the impact of an evidencebased HIV prevention intervention on sexual risk behaviors of in-school 6 th graders in Liberia [35,36], and six used post-intervention surveys to assess program effectiveness in four African countries [37][38][39][40][41][42]. All of the papers reported mixed results of their prevention programs regarding some elements of knowledge and behavior change; however, the four that follow reported more positive results. A comparison of pre-and post-intervention survey data in Sierra Leone found that HIV-related knowledge and condom use increased among adolescents [37], commercial sex workers and military personnel [38] following an HIV prevention program including intensive IEC activities and distribution of free condoms. Two papers on refugee camps in Guinea reported that exposure to program peer educators was associated with improved HIV and STI knowledge and changed behavior to prevent HIV [39,40].

General RH
Two papers reported on unique efforts related to reproductive health. A program to improve and measure the quality of RH services at a clinic serving Burmese refugees and migrant workers on the Thailand-Burma border improved the quality of care, and also increased staff skills and motivation to collect and use data to make program decisions [43]. An evaluation of a literacy program that used RH content in Guinea found that refugee women who completed the program reported high knowledge on maternal and newborn health, HIV and STIs; increased use of FP; and a marked increase in feelings of empowerment [44].

Discussion
This review found that some RH programs in crisis settings have been evaluated although most evaluations were medium in quality, suggesting limitations in study design and analysis. Most of the papers reported generally positive results suggesting that these programs are likely well-designed and reasonably well-implemented. The papers demonstrated both that RH programs can be implemented in these challenging settings and that women and men will use RH services when they are of reasonable quality. In comparison to the program evaluation papers identified, three times as many papers were found that reported RH descriptive or prevalence data in humanitarian settings. While data demonstrating the magnitude of the problem are crucial and were previously lacking, the need for RH services and for evaluations to measure their effectiveness is clear [45,46]. It is critical to more directly link research to interventions and increase the evidence base for RH service delivery strategies in humanitarian settings. This includes not only the research but also publication and sharing of results. An increased focus on implementation science is needed to explore how best to improve delivery and use of RH services as well as the use of research to improve practice [47].
Although published articles are not representative of RH programs implemented in humanitarian settings as most programs do not publish their results, they may reflect relative attention, both programmatic and financial, to particular areas. A preponderance of papers reported on HIV/ AIDS programs although few mentioned other STIs. While GBV was under-represented among program evaluations, one-third (32) of the descriptive papers reported prevalence and types of sexual violence perpetrated in humanitarian settings. This suggests that GBV does, in fact, receive attention in research, although perhaps less in programming which when implemented may be only rarely evaluated. FP, on the other hand, was under-represented among both program evaluations and descriptive papers suggesting that FP overall receives less attention than the other RH components. Adolescents often face additional barriers to meeting their RH needs [48], but only four HIV prevention programs targeted adolescents and no papers evaluated adolescent-friendly RH services. No papers mentioned safe abortion which remains virtually unavailable in humanitarian settings [49], nor post-abortion care.
Programs requiring long-term follow-up faced specific challenges introduced by the instability of crisis settings and associated population movements. Some of these challenges, such as brief interruptions to treatment that arose during incidents of crisis, can and should be managed or prevented with planning, as demonstrated in the response to post-election violence in Kenya [30,32] and an upswing in violence in DRC [33]. Training refugee or IDP health workers, who would likely move with their community, may be a potential strategy for ensuring continued access to care for displaced people after they return home. Additional challenges to the implementation of RH programs were identified in the papers. For example, highly trained health workers are needed to provide RH services, and they may require updated competency-based training, particularly for EmONC, longacting and permanent FP and clinical care for survivors of rape. The evaluation of a training tool for providers suggested that although attitudes are challenging to change, care for survivors of rape can be improved [22].
Proven evidenced-based strategies should be adapted and implemented in humanitarian settings. For example, EmONC is crucial to reduce maternal morbidity and mortality, and is thus a component of the minimum standard in humanitarian RH service delivery (the Minimum Initial Service Package) [4]. Yet, only three of the seven maternal and newborn health programs that were evaluated aimed to improve the availability of these critical services. Only one of the evaluated programs improved the availability of long-acting or permanent FP methods; the other programs were generally limited to short-acting methods, despite evidence that a broad choice of methods is an essential component of good FP programming and also associated with increased use [50][51][52]. Although a foundation in social change theory has been shown to be important for behavior change [53], only one of the HIV prevention programs appears to have had such a base [35,36]. Behavior change communication efforts implemented in humanitarian settings should adapt such proven evidence-based strategies. Moreover, it is critical that best practices be shared across the humanitarian and development fields. While the humanitarian field has adapted strategies that have been successful in development settings for many RH components, response to sexual violence is one area where the humanitarian field may be in advance of the development field, and it is crucial that these programs be implemented, rigorously evaluated and published. Further, it would be useful for programs (and journals) to publish results of programs that were unsuccessful so others may learn from those experiences.
Fewer than half of the papers used any kind of comparison, either between pre-and post-measures or between intervention and comparison groups. This is not a call for more randomized controlled trials, however, since randomizing clients is not often appropriate, due to the fundamental principle of client choice in FP and GBV programming [54]. Evaluations using pre-and post-intervention measures or quasi-experimental designs may be appropriate, particularly where a program strategy is implemented in phases and a group that has not yet received the intervention serves as a comparison for a group in an earlier phase of the program. In addition, the challenges to collecting data in humanitarian settings are well-recognized [55,56], and population-based surveys may be particularly challenging in these unstable and insecure settings [57]. Therefore, other rigorous measures of program quality that are feasible to collect should be explored. For example, the UN process indicators of EmONC were developed to monitor interventions proven to reduce maternal mortality without the limitations and expense of a maternal mortality survey by using information available at health facilities [10,58]. What similar practical approximations could be used to measure the success of FP and GBV programs? It is plausible that evaluations of clinical HIV programs were in the majority because program quality could be measured using clinical data (patient adherence and outcomes) that were routinely collected. Challenges to collecting appropriate data have been noted [5,9]; increased effort should be put into routine data collection to ensure that good quality data to measure standard indicators are collected, and shared. This may mean adapting registers to capture data on, for example, obstetric complications or to record new, continuing and switching FP clients.
Limitations of this review include its restriction to quantitative methodologies and to papers published in English, which may have excluded relevant publications. The selected search parameters may have missed papers that did not explicitly refer to conflict or humanitarian settings or natural disasters, or the general RH topics that were searched in the title, abstract or key words. While the included papers may be representative of peer-reviewed published literature, they are not representative of RH programming in humanitarian settings: humanitarian agency staff may not have time to write up results for publication and negative or null findings may be difficult to publish.
Program evaluation and implementation science should be incorporated into programs to determine the best ways to serve the RH needs of people affected by conflict or natural disaster. Standard program design should include rigorous program evaluation [59] and improved routine data collection. The results must be shared so that proven evidence-based strategies for RH are implemented in humanitarian settings. These papers demonstrated both that RH programs can be implemented in these challenging settings, and that women and men will use RH services when they are of reasonable quality.

Competing interests
The author declares that she has no competing interests.

Authors' information
The author is a member of the Steering Committee of the Inter-Agency Working Group on RH in Crises.

Medium
Casey Conflict and Health 2015, 9(Suppl 1):S1 http://www.conflictandhealth.com/content/9/S1/S1 Guinea Refugee health workers seconded to health facilities provided free RH services and trained refugee women as lay health workers Cross-sectional post-intervention multi-stage cluster survey in intervention area of women (n=444) and men (n=445) of reproductive age (Liberian and Sierra Leonean refugees) living in one of 48 refugee camps in Guinea in 1999.
Approval of FP was high, but more than 40% had not discussed FP with partner. Current use of modern FP (17%) was higher than in country of origin (3.9%) or host country (4.1%). Perceived service quality was most important determinant in choice of where to get FP.

Medium
Huber et al (2010) [18] Afghanistan Improved access to FP using CHWs and community-based distribution of short acting methods