Reproductive health (RH) indicators are used by the United Nations High Commissioner for Refugees (UNHCR) and its health partners to ensure resources are correctly targeted to those who need them, respond quickly to public health problems, monitor trends, and evaluate the effectiveness of interventions and service coverage [1]. The RH indicators collected are essential to describe the burden of RH among refugee women of reproductive age. The global distribution of RH indicators can show differences in prevalence between countries or regions that will help inform policy or advocacy but may have very little impact on camp and refugee specific lifesaving interventions. Without timely, and accurate data refugee women could be at increased risk of mortality and morbidity. In general more emphasis needs to be placed on the tools and resources that are needed to collect data to ensure the indicators are of high quality [2].

In January 2013, there were 15.4 million refugees worldwide--approximately 48 % of these refugees were women and girls [3, 4]. As shown in a recent global evaluation of reproductive health (RH) in humanitarian settings from 2012 to 2014, refugee women and children are the most vulnerable to the consequences of displacement [5]. A positive finding in the study showed the number of health proposals with an RH component increased by an average of 10 % per year from 2002 to 2013 [6]. Other positive findings showed significant progress in maternal and newborn health as evidenced by an increase in funding and program evaluations, an increase in funding of gender-based violence (GBV) programs, improvements in mother-to-child-transmission programs and family planning commodities. In contrast, disparities in emergency obstetric care and newborn care services remained unchanged [7] and clinical components of the Minimum Initial Service Package (MISP) were lacking in three settings well beyond the emergency phase [8]. Post-abortion care is still behind compared to other RH interventions due to lack of funding and systematic evaluations. In addition GBV programs showed a lack of prevention efforts against sexual violence and staff trained in the clinical management of rape. Other problem areas found were the inconsistent availability of antiretroviral for HIV and the lack of prevention, testing and treatment of other sexually transmitted infections (STIs). Lastly, an assessment of family planning programs showed a gap in funding and commodities such as intrauterine devices (IUDs), permanent methods and emergency contraception [9]. As humanitarian actors gain experience to solve RH issues, this information will be useful to improve services to populations affected by a crisis.

Since 2006, UNHCR and its health partners have been using a unified health information system (HIS) to monitor refugee public health and HIV programs in camps and urban settings. At the start of 2010, HIS provided services to 1.5 million refugees operating in 18 countries, 85 refugee camps, and through 24 different partners [10]. Past published studies have used the HIS data to evaluate camp nutrition programs, the utilization of outpatient services, the burden of malaria, and under five morbidity and mortality [11-14]. Important contributions of these papers included recommendations to modify nutritional indicators, and the accuracy of population indictor estimates (number of women of reproductive age and children under five). In addition, there were policy and advocacy issues identified such as the need to support equitable and higher quality malaria eradication programs in refugee and host populations. Lastly, through these studies awareness was turned towards the importance of the HIS unified systems utility to analyze and disseminate health information.

To monitor RH services in camps, the HIS uses RH indicators to measure and determine progress in achieving the UNHCR predetermined standards. This study was part of the Interagency Working Group in Reproductive Health Global Evaluation of RH in humanitarian settings. The purpose of this study was to conduct a retrospective review of selected HIS RH indicators of refugees by country to examine trends over time and assess if the indicators meet the UNHCR standards.

This analysis used data from RH indicators obtained from UNHCR’s HIS Microsoft Access™ (Redmond, WA) database over a seven-year period (2007 through 2013). Data selected for this analysis were abstracted from the HIS Access database by the HIS supervisor at UNHCR headquarters in Geneva and converted into Excel spreadsheets. The variables included continent, name of country, name of camp, and date of report, and the numerator and denominator values to calculate the indicator. Analysis was conducted from November 2013-July 2014. The US Centers for Disease Control and Prevention (CDC) conducted the data analysis and determined the study was surveillance activity and not human subject research. All analysis was done using SAS software, Version 9.3 of the SAS Institute, Inc.^© (Cary, NC) [15].

Using HIS data taken from operating refugee camp health facilities from 2006 to 2013, we reviewed HIS data from a total of 23 countries. A country was included in the analysis if it had at least one acceptable camp with monthly reporting data from no later than 2008 and had no more than two unacceptable camps.

Camps that did not meet the inclusion criteria were not included in the analysis. Data analysis started in 2007 or 2008, depending on the availability of the monthly reports.

Fifty-seven RH indicators are included in the HIS system. For this analysis, seven indicators were chosen after screening the RH indicators for usability, plausibility, and relevancy of the reported indicator data. Usability was defined as those indicators identified as high priority by the Inter-agency Working Group on Reproductive Health in Crises [16]. For plausibility, if more than 10 % of the monthly estimates had proportions above 1 prior to cleaning, the data were considered implausible and that indicator was not recommended for analysis. To avoid redundancy, relevancy was determined based on HIS indicators that have been previously analyzed and published. After reviewing the three criteria, the final selection of seven indicators was determined by consensus of RH senior staff at UNHCR and CDC epidemiologists.

The graphs below show the remaining five indicators with the mean and the UNHCR target values for each indicator. The text describes significant trends.

Proportion of births attended by a skilled birth attendant

The plots in Fig. 3 show the yearly average of the monthly camp births attended by a skilled birth attendant for each country along with the UNHCR target number of 100 % coverage. Eight of the 3,566 (0.22 %) skilled attendant birth monthly values were outliers and imputed prior to creating the indicator. The percentage was above 100 % for this indicator in 270 out of 3566 (7.6 %) of the monthly reports. On average most (8 of 10) countries did not meet the target of 100 % for all reporting years. The overall test for a linear trend showed a significant increase over time (p < 0.001). The percent of births attended by a skilled birth attendant increased significantly in 6 of 10 countries: Bangladesh (p < 0.001), Kenya (p = 0.0005), Djibouti (p < 0.0001), Tanzania (p < 0.0001), Uganda (p = 0.0051) and Yemen (p = 0.010). Nepal experienced no change, and met the target indicator 100 % of the time over the study period. There was no significant linear trend in 3 of 10 countries: Chad, Thailand, and Zambia.

Proportion of live births performed by caesarian section

Figure 4 shows the country-level results for the average percentage of live births performed by caesarian section by year. Nine of the 3566 (0.25 %) monthly values for this variable were imputed before analysis. The WHO recommends to have between 5 % and 15 % of live births delivered through caesarian section. The average monthly caesarian section rates within refugee camps were below the recommended minimum for nearly half of countries every year (4 of 10 countries), and an additional 2 countries only met the minimum once over the study period, Djibouti and Yemen, respectively. However, there was a significant linear trend over time (p < 0.001), driven by two countries Nepal and Tanzania, with a larger increase in the percentage of caesarian section deliveries. Seven of 10 countries had a significant increase in proportion of live births performed by caesarian section over the study period: Bangladesh (p < 0.0001), Djibouti (p < 0.0001), Kenya (p = 0.033), Nepal (p < 0.0001), Thailand (p = 0.029), Tanzania (p < 0.0001) and Uganda (p = 0.007). Zambia had a significant decrease in percent of caesarian section births over the study period (p = 0.038). Chad and Yemen did not have any significant change.

Percentage of ANC mothers who were screened for syphilis during pregnancy

The yearly average of monthly camp syphilis screening by country is presented in Fig. 5. Nine of the 3,566 (0.25 %) monthly camp values for the number of women who received syphilis screening were imputed before the screening indicator was created. There were 731 out of 3,566 (20.5 %) camp monthly reports equal to zero for the antenatal syphilis screening, and 359 out of 3,566 (10.1 %) of the monthly report proportions were over 1. According to the UNHCR target, 100 % of women who come for antenatal care should be screened for syphilis. Only 10 % of all countries for all reporting years reached the standard. There was a significant increase in the yearly percentage of women screened within the refugee camps in this analysis (p < 0.001). Five out of 10 countries significantly increased the percentage of ANC mothers screened for syphilis: Bangladesh (p = 0.008), Chad (p < 0.0001), Djibouti (p < 0.0001), Nepal (p < 0.0001), and Thailand (p = 0.003). Two countries experienced a significant decrease over the study period, Tanzania (p < 0.0001) and Zambia (p = 0.007). The remaining 3 did not have a significant change in syphilis screening.

Rate of condom distribution among the population

The average yearly rate of camp condom distribution per person per month can be found in Fig. 6. Six variables were used to create the numerator for the condom distribution indicator, and 491 values were imputed of 21,396 (2.3 %). In the denominator, 22 of 3566 (0.62 %) of the total population values were imputed prior to the creation of the condom distribution indicator. There were 658 out of 3,566 (18.5 %) monthly camp values equal to zero for the condom distribution indicator, meaning that during those months there were either no condoms available for distribution and/or no condoms were distributed. Overall, the trend for condom distribution was not significant (p = 0.109). There was a significant increase in the rate of condoms distributed in 2 of 10 countries: Bangladesh (p < 0.0001) and Nepal (p < 0.0001). Three of 10 countries experienced a significant decrease in the rate of condom distribution over time: Djibouti (p < 0.0001), Tanzania (p < 0.0001) and Zambia (p < 0.010), and 5 of 10 had no significant change over time: Chad, Kenya, Thailand, Uganda and Yemen.

Proportion of reported rape survivors who receive PEP within 72 h of an incident occurring

Figure 7 displays the proportion of rape survivors who were given PEP within 72 hours of a rape incident. Eight of 3566 (0.22 %) monthly values for women who received PEP and eight of 3566 (0.22 %) of the women who reported rape were imputed before the PEP indicator was created. There were 2971 missing values of 3566 (83.3 %) camp monthly reports for the PEP indicator because there were no reported rapes within camps during those months. Only 10 % of all camps for all reporting years reached the standard, but there was a significant increase in the percentage of women receiving PEP globally (p < 0.001). One out of 10 countries had a significant increase over the study period, Uganda (p = 0.0001). Six out of 10 countries experienced no significant change over time: Bangladesh, Chad, Kenya, Nepal, Thailand, and Yemen.

This analysis provides an overview of important RH trends over the seven-year study period. For some indicators (births attended by a skilled birth attendant, caesarian section, syphilis screening, and PEP), improvement from 2007 is clear. For other indicators (MMR, NNMR, and condom distribution), more information is needed to explain current trends and why or if improvement is lacking. In several instances standards were not met. Regardless of improvements made in each country for each indicator studied, this analysis demonstrates that more needs to be done to ensure women in refugee camp settings are receiving high quality RH care, as this will decrease morbidity and mortality within the study population. It is often thought that analysis from the HIS is complete, reliable, and of high quality. This analysis demonstrates a need for improved reporting, as many countries were dropped from the study. To close this gap UNHCR and its health partners need to improve strategies and programs to derive maximum benefit from the HIS.

We suspect that HIS is not capturing some of the maternal and neonatal deaths of the camp populations rather than over reporting live births or consistently achieving very low maternal and neonatal mortality in these refugee camp settings. Several previous studies on maternal and infant mortality in refugee camps discussed underreporting of maternal and neonatal mortality [19]; two recent studies that examine the extent of underreporting of neonatal deaths have been completed in camps in Tanzania and Chad [Idowu R, Morof D, Blanton C, Tappis H, Cornier N, Tomczyk B. Using capture-recapture methods and verbal autopsy to understand the incidence of neonatal mortality, stillbirths and live births in UNHCR refugee camps in Chad 2013. Unpublished report.]. Both neonatal and maternal mortality can fluctuate within refugee camps, particularly when influxes of refugees due to an acute emergency occur during a protracted setting. An outbreak such as Hepatitis E that occurred in Dadaab, Kenya disproportionally affected pregnant women [20]. The maternal death audit system is a UNHCR strategy that has been implemented successfully in Dadaab and includes community sensitization to report deaths that occur at home [21]. Other interventions were also included such as improvement of infrastructure, transport, supplies, skilled staffing and mother incentives. This strategy could be replicated in other camps.

Ensuring SBA at delivery is efficacious in contributing to the reduction in maternal and perinatal mortality and helping to reach the post 2015 MDGs 4 and 5 targets [22, 23]. Overall, the significant increase across countries in this study is encouraging. Some refugee camps in this study had a low use of SBAs; this may be due to a lack of SBAs at the facility level. Additionally, the proportions may have exceeded 100 % in some instances because the live births were inaccurately recoded in the monthly reports or because host community women came to deliver and were misclassified as refugees, but more information is needed to determine the root of the inaccuracies. An important consideration for maternity wards at all camps is that they are staffed, 24 h a day, with a professional midwife capable of responding to common obstetric emergencies. It is also important that UNHCR and its partners provide refresher training and supportive supervision, as needed.

Caesarian section was introduced in emergency obstetric care as a lifesaving procedure both for the mother and baby. Overall, there is a positive trend toward meeting the UNHCR standard of caesarean section rates. It is known that the global picture indicates an uneven distribution of caesarian section that shows underuse in low income settings and adequate or even unnecessary use in middle and high income settings [24], and our analysis shows an uneven distribution depending upon the country. The findings from Nepal are counterintuitive since it showed caesarian section rates that would reflect a high income setting. Two studies have shown an inverse association at population level between caesarian section rates and maternal and infant mortality in low income countries where large sectors of the population lack access to basic obstetric care [25, 26], making this indicator an important morbidity and mortality measure. In refugee camps where health care is provided and access to emergency obstetric care may be disproportionately available, this study indicates there are still gains to be made in maternal and infant mortality by increasing access to and use of improved birth technologies, including cesarean delivery. In addition, concerted actions need to be taken to offer timely caesarian section to women who need it and to advocate for a rationale use of caesarian section in camps with a surplus. In Chad and Zambia where caesarian section rates were low more detailed field assessments would help to contextualize the issue and determine the best course of action. Lastly, other important contributing factors that may increase caesarian rates such as previous caesarian sections, and maternal or fetal causes if captured by the HIS could help to interpret this indicator.

Several factors could decrease the syphilis screening rate in refugee camps. For instance, syphilis screening although a routine part of ANC in refugee settings may be missed due to a lack of supplies, equipment and trained staff. Broader ANC may also be lacking, which indirectly leads to women not being screened for syphilis as regularly as they should be. Finally, health care providers may not be prepared in syphilis prevention, and how to prevent re-infection during pregnancy by promoting condom use. Commodities may be in short supply for testing and laboratories require appropriately trained staff for testing [27, 28]. Improvements in UNHCR syphilis screening programs have included implementation of a decentralized program of syphilis screening involving nurses trained in education, counselling and the provision of on-site testing using the Rapid Plasma Reagin test and partner tracing [29].

Condom distribution was inadequate for the majority of camps (6 of 9 camps were less than 50 % in 2013). When looking more directly at the data, condom distribution was sporadic and indicated months with high condom distribution and months with very low to no condom distribution. This indicator may not provide distinct value as a measurement because distribution does not necessarily equate to use, especially where the product is given away free of charge. Refugee populations may also have a varying proportion of children and/or females, making comparisons across refugee countries difficult without adjusting for the number of people who do not need condoms [30]. Logistical difficulties in obtaining and delivering of supplies due to camp location do occur are major obstacles.

Refugee women who have experienced sexual violence should be referred for health services as soon as possible after the incident. The large number of missing data in this analysis points to the fact that either very few women reported a rape in refugee settings or the rapes were reported but not recorded in the HIS, or women are not willing to report a rape as a majority of the monthly reports did not indicate any rapes. The number of rapes reported in each country fall far below global statistics on sexual violence [31]. Legal reforms, protection policy and high quality services available to rape victims have been influential in increasing the likelihood that women will report. Therefore, a multi-sectoral approach is needed in each refugee setting in order to improve services.

There were a number of limitations in this analysis [32]. Underreporting, lack of representativeness, lack of timeliness, and inconsistency of case definitions are four of the most common limitations of many surveillance systems . The HIS has been implemented since 2006 and the quality and completeness of data is known to be somewhat variable during the first months of using the system and may be variable depending on conditions in the individual camps and availability of human resources. A number of countries were excluded from this analysis because the data was too variable. Camps from countries that were not included in the analysis had a higher variance of reporting variability. Some camps rarely reported and other camps within the same country reported fairly regularly. Another limitation is that data quality may be influenced by a number of factors that we did not measure such as newly opened camps versus long term camps, size of camps, availability of RH services and staffing. The inclusion criteria were designed to limit the amount of poor quality data, but they do not ensure that all the monthly reports for 56 camps were of high quality. It also should be noted that we present the average monthly camp estimates by year within a country, but there is variation within a camp and between camps within a country. Sensitive subjects, such as sexual violence may not be reported accurately.

The UNHCR HIS in this study was limited to refugee camps and is facility-based. It may be biased because populations that do not seek care are excluded (survivors of sexual violence, certain RH patients, and deaths occurring outside of health facilities). It is recognized that in refugee camp settings, women may have better access to quality RH care than is available in their country of origin. The Global Evaluation of RH Services for Refugees and Internally Displaced People, conducted in 2004 found that internally displaced persons had worse access to RH services than refugees [33]. Thus we may anticipate that the results may suggest more positive findings than we might find among the surrounding host population or internally displaced persons [34]. Ideally it would have been potentially helpful to have information on how the data were collected in each health facility in each refugee camp in order to improve our understanding of the HIS RH data from this analysis to provide context to the results.

Many of the refugee RH indicators have been improving over time and most are better than those of host countries, but more can be done to improve the interventions to meet the UNHCR standards. More information is needed on the RH data collection cycle to determine why there was uneven distribution of indicators between countries. In general comprehensive, consistent and comparable statistics on RH provides an opportunity to assess progress towards indicator standards. Despite some improvements over time, this analysis confirms that most countries did not meet standards and that there were differences in RH indicators between countries and across years. Consequently, the HIS periodic monitoring of key reproductive health indicators at the camp level should continue.