Background
Armed conflict has spread throughout Syria, with warring parties, including government forces, non-state armed groups, and terrorist groups, fighting to expand their areas of influence and gain control of strategic roads, settlements and critical resources. In the sixth year of the conflict, violence in Syria has “neither abated nor diminished in brutality,” according to the United Nations Security Council [
1,
2].
The humanitarian consequences of this conflict are dramatic. According to United Nations (UN) and regional partners, Syria is the largest displacement crisis globally [
3]. Entering the fourth year of the conflict, around the time of our assessments, an estimated 8.8 million people had been displaced, almost half the population of Syria [
4]. Of them, more than a quarter have been forced to cross international borders and seek refuge in neighboring countries [
4]. An estimated one in five registered refugees are children under 5 years of age [
5].
Previous complex emergencies have demonstrated that conflict and displacement often result in a deterioration of the nutritional status of populations. In the context of prolonged conflict, affected populations often become dependent on humanitarian assistance for food and as a result experience reduced dietary diversity and meal frequency [
6‐
8]. Separation of children from caretakers as a result of the conflict and displacement, also often affects infant and young child feeding (IYCF) practices [
9,
10]. Deterioration of water, sanitation, and health infrastructure, as well as access to key services can compound the effects of crisis on nutrition [
11].
Potential for nutrition insecurity was of particular concern in Syria as assessments of the nutritional status of children prior to the conflict suggested some baseline deficiencies. According to the 2009 Syrian Family Health Survey (SFHS) and the 2006 Multiple Indicator Cluster Survey (MICS), levels of total wasting in Syria were greater than 8 %, poor by World Health Organization (WHO) Crisis Classifications [
12‐
14]. Levels of stunting were greater than 20 % in both assessments.
The scale of the displacement from Syria has further contributed to concerns about the nutritional status of displaced persons. Of the more than two million Syrian refugees displaced in 2013-14, the majority were hosted in Lebanon, Jordan, Turkey, and Iraq [
4]. The displacement, in absolute terms and relative to the host population, has overwhelmed some of these neighboring countries. For example, Lebanon, a country of just over four million people, registered its millionth Syrian refugee shortly after the survey in Lebanon concluded [
15].
While nutrition partners cautioned of a deteriorating situation in Syria, limited data existed at the time of our surveys. Better information was needed to inform the ongoing response—estimated at $30 million in nutrition programming and $1.1 billion in food and agriculture activities for 2014 [
16]. This need prompted calls for rigorous and representative assessments to confirm the nutritional situation of Syrian refugees [
17].
In this context, nutrition surveys were organized in Jordan, Lebanon and Iraq to generate evidence as to the nutritional status of Syrian refugees with a primary objective of assessing the level of acute malnutrition among women and children. Additional indicators including those related to anemia, access to health care, morbidity, IYCF practices, and water and sanitation helped contextualize these findings. Surveys in Lebanon, Jordan and Iraq also allowed us to compare whether the nutritional status of refugees depended on the country where refugees settled or the type of settlement (in refugee camps or integrated with host population).
Methods
Study design and sample size
Data used in this analysis were obtained from cross-sectional surveys of Syrian refugees from seven sites in three countries: Lebanon, Jordan and Iraq. Data were obtained from the United Nations Children’s Fund (UNICEF) in Lebanon and Iraq and the Office of the United Nations High Commissioner for Refugees (UNHCR) in Jordan. In Lebanon, independent survey samples were selected in the four geographical regions of the country: North, South, Beirut/Mount Lebanon and Bekaa. Since there were no established refugee camps in Lebanon, all surveyed refugees resided in host communities. In Jordan, two independent surveys were conducted: among refugees residing in Za’atri refugee camp and among refugees residing among host communities (outside of camps) nationwide. In Iraq, the survey was conducted among refugees residing in Domiz refugee camp in the Kurdistan region of Iraq. Data collection took place between October 2 and November 30, 2013 in Lebanon, April 12 and May 1, 2014 in Jordan and May 20 and 31, 2013 in Iraq.
For surveys in Jordan and Lebanon, sample sizes were calculated based on estimated prevalence of global acute malnutrition, required precision, and design effects which were informed by results of previous cross-sectional nutrition surveys—a September 2012 survey in Lebanon and a November 2012 survey in Jordan [
18,
19]. Adjusting for non-response rate and household demographics, 340, 270, 330 and 260 households were randomly selected in the North, South, Beirut/Mount Lebanon and Bekaa regions of Lebanon, respectively. Similarly in Jordan, 337 and 515 households were selected in Za’atari Camp and in host community, respectively. Parameters used for estimating sample size in Lebanon and Jordan are provided in Additional file
1. In Iraq, a 30 cluster by 30 children survey design was used [
20].
A list of registered refugees and persons awaiting registration maintained by the UNHCR was used as a sampling frame for all surveys, since this database included the most updated information on recently arrived refugees. In all surveys, the UNHCR database was used to select clusters randomly, proportional to the number of refugees residing in each administrative area or camp section. In non-camp settings, households within each cluster were randomly selected from the UNHCR database and called in advance of the field visit to confirm their availability, willingness to participate, and current address. In Jordan, 83 % of households in the database had functional phone numbers. In both of the refugee camps (Domiz in Iraq and Za’atari in Jordan), households were systematically selected; sampling intervals were calculated using household listings by block obtained from community leaders or local health volunteers.
Measurements
The information on household, children and women was collected by teams that included both nationals and Syrian refugees. Teams composed of at least three members each (an interviewer, anthropometry measurer(s), and a hemoglobin (Hb) technician where Hb was measured) received between 2 and 7 days of training, including a field test. All surveys collected data on basic demographics, anthropometry, recent illness, breastfeeding practices, and household level indicators related to water and sanitation. The surveys in Jordan and Lebanon also collected data on Hb concentration, food security and access to antenatal care (ANC). Household level indicators were excluded from the analysis of Iraq data, as the Domiz sample only included households with children under 5 years of age, and therefore was not a representative sample of all households in the camp.
Hb concentration was measured in children aged 6 to 59 months and non-pregnant women of reproductive age (15–49 years). Pregnancy status was self-reported. Hemoglobin was measured in capillary blood using a HemoCue Hb 301+ photometer following a standard procedure [
21]. Anemia in children was classified according to WHO definitions as severe (hemoglobin < 7 g/dL), moderate (7 g/dL ≤ hemoglobin < 10 g/dL), or mild (10 g/dL ≤ hemoglobin < 11 g/dL) [
22]. In non-pregnant women, anemia was classified as severe (hemoglobin < 8 g/dL), moderate (8 g/dL ≤ hemoglobin < 11 g/dL), or mild (11 g/dL ≤ hemoglobin < 12 g/dL) [
22].
Anthropometry indicators in children aged 6 to 59 months were measured following standard procedures [
23]. Weight was measured with a digital Seca scale, accurate to 0.1 kg. Height or recumbent length was measured with a wooden ShorrBoard, accurate to 1 mm. Children’s age was ascertained from UNHCR registration cards or, if unavailable, from immunization cards, birth certificates, or caregiver recall. Presence of bilateral pitting edema was assessed by survey teams; all suspect cases were verified by a supervisor. Acute malnutrition (based on weight-for-height z-score and the presence of edema), stunting (based on height-for-age z-score), and underweight (based on weight-for-age z-score) were defined and classified as moderate (–3 ≤ z-score < –2) and severe (z-score < –3) according to WHO definitions [
24]. Extreme outliers, defined using a flexible exclusion range criterion (±4 z-scores from the observed mean) were excluded from the analysis [
25,
26]. Surveys assessed nutritional status of all women of reproductive age (15 to 49 years), classifying women with a mid-upper arm circumference (MUAC) less than 23.0 cm as malnourished [
27].
Additional indicators were ascertained by self-report. Diarrhea was defined as three or more loose or watery stools in a 24-h period during the preceding 2 weeks. Current breastfeeding included children put to breast during the previous 24-h period. In Lebanon and Iraq both indicators were assessed among all children 0–59 months; in Jordan children 0–5 months were not assessed for diarrhea. Access to antenatal care and iron-folic acid supplements was assessed in all pregnant women (15–49 years). A refugee household living outside of a refugee camp was considered to be hosted if they were residing in the same dwelling as citizens of the host country at the time of the assessment. Survey instruments were modified in each country to meet the information needs of the ongoing response; indicators not collected consistently across settings were excluded.
In Jordan, data were collected on mobile phones with an Android OS and uploaded daily. Anthropometric measurements were recorded on both the phone and paper allowing supervisors to verify data entry. In Lebanon and Iraq, all data were recorded on paper questionnaires.
Statistical analysis
For all surveys, ENA for SMART software 2011 (Version: Aug. 4, 2014) [
28] was used to generate standardized z-scores based on the 2006 WHO Growth Standards and analyze anthropometry data [
24]. Additional variables were analyzed with EPI INFO 3.5.3 software, SPSS version 17.0, and STATA/IC 13.1. Ninety-five percent confidence intervals were calculated adjusting for the complex survey design. Significance was tested adjusting for clustering using the procedure described by Donner et al. [
29] A
p-value <0.05 was considered statistically significant.
Discussion
Data presented here suggest that despite the magnitude of the crisis, Syrian refugees have not experienced elevated rates of acute malnutrition. Among both women and children, the prevalence of acute malnutrition was relatively low. Global acute malnutrition prevalence among children aged 6 to 59 months was less than the WHO threshold of “acceptable” (<5 %) in all settings [
13]. Prevalence of acute malnutrition among women, as measured by mid-upper arm circumference, also indicated relatively low levels of wasting (range 3.5–6.5 %).
Taken together, these data provide a more comprehensive picture of acute malnutrition among Syrian refugees affected by the conflict. Data on the nutrition status of displaced persons within Syria remains sparse given security and access restraints. One of the few representative surveys, a survey of children 6 to 59 months of age in Idleb governorate in June 2014, found that the prevalence of global acute malnutrition was 1.1 %, consistent with the low prevalence we observed among refugee populations [
30]. The levels of acute malnutrition observed, however are notably lower than the prevalence reported in pre-conflict national assessments of children in Syria—9.3 % according to the 2009 SFHS and 8.6 % according to the 2006 MICS [
12,
14]. The prevalence estimates observed in our assessments are more similar to national estimates from neighboring countries now hosting refugees. Demographic Health Survey (DHS) data from Jordan in 2012 estimated a prevalence of 2.4 % for total wasting and 0.6 % for severe wasting [
31]. A 2010 national micronutrient survey in Jordan similarly found the prevalence of total wasting among children 12 to 59 months to be 3.5 % [
32]. Estimated prevalence among long-term Palestinian refugees in Lebanon in 2011 also suggest wasting to be below crisis levels (4.7 %) [
33]. Prevalence of wasting in Iraq documented in the 2011 MICS was slightly higher (7.4 %) [
34]. These findings together suggest that at the time of the assessments, contrary to media reports and popular opinion, Syrian refugee children were not experiencing crisis levels of acute malnutrition but rather had levels comparable to the host populations among which they were displaced.
Whereas for acute malnutrition a change in prevalence from pre-crisis was observed, levels of stunting among refugees in most of our assessments were similar to estimates of stunting in Syria pre-crisis. This distinction is consistent with the fact that stunting is a measure of chronic malnutrition and is less affected by acute changes in food security. Stunting was estimated to be 23.0 % in the 2009 SFHS and 22.4 % in the 2006 MICS [
12,
14]. These pre-crisis levels are very similar to that estimated in North, South and Bekaa regions of Lebanon (range 20.1–21.0 %), and not markedly different from stunting prevalence in Za’atri refugee camp (16.7 %). Interestingly, the prevalence of stunting in our assessment of non-camp based refugees in Jordan was lower than in other contexts surveyed (10.5 %); this prevalence is more similar to the prevalence of stunting estimated in Jordan in the 2012 DHS (7.7 %) than estimates from Syria pre-conflict [
31]. Prevalence of stunting in our assessment of refugees in Beirut (14.1 %) was more similar to that estimated among Palestinian refugees in Lebanon (13.3 %) [
33]. This may reflect a degree of integration of non-camp refugees with the host populations in Jordan and Beirut, Lebanon.
Levels of anemia recorded were high in all assessed settings, but only a problem of major public health significance in the Za’atari refugee camp in Jordan, according to WHO classification [
22]. In Za’atari, prevalence of anemia was 48.4 % among children aged 6 to 59 months and 44.8 % among non-pregnant women. High prevalence of anemia among refugees has been documented in many camp-based refugee populations. A study of Bhutanese refugee children in Nepal documented a prevalence of 43.3 % in 2007, prior to introduction of micronutrient supplementation [
35]. A study of refugees in long-term African refugee camps found even higher prevalence estimates among children in camps in Kenya (61.3 %), Uganda (72.9 %), and Ethiopia (62.9 %) [
8]. High prevalence of anemia among refugees has been attributed in part to the reliance on food rations in camps, an explanation consistent with our finding that prevalence among the camp based refugees in Za’atri was higher than in non-camp based refugees in Jordan and in Lebanon [
7,
8,
35]. In our assessments of non-camp based refugee populations, anemia levels were more consistent with the host populations. Among refugee children in Jordan living in the host population, total anemia prevalence in our survey was slightly lower (26.1 %) than that documented in the Jordanian population among children under five in the 2012 Jordan DHS (32.4 %) [
31]. Similarly, the prevalence of anemia among refugee women residing in host population in Jordan in our survey (31.3 %) was also just below the estimated prevalence from the 2012 Jordan DHS (33.5 %).
For both women and children, prevalence of severe anemia was low, 1.1 % or less in all surveys. In Lebanon, no cases of severe anemia were identified among children. Supplementation with micronutrient powder has been shown in a randomized trial to have no significant effect on hemoglobin levels among children with hemoglobin levels ≥10.0 g/dL [
36]. These findings have subsequently been confirmed in studies of Bhutanese refugee children [
35]. As such, micronutrient powder supplementation may have negligible benefits in populations with a low prevalence of children with hemoglobin <10.0 g/dL, as is the case among refugees in Lebanon and Jordan. Instead, interventions that help ensure a more diversified, micronutrient rich diet may better address the micronutrient deficiencies in the refugee populations than powders, particularly given the rich diversity of food available in markets outside the camps.
Self-reported data provided additional information to contextualize the nutrition findings. With respect to antenatal care, all six assessments in Jordan and Lebanon found that fewer than 55 % of pregnant women 15–49 years of age were enrolled in ANC. Pre-crisis, Syria reported high utilization of reproductive health services. A reported 85.3 % of pregnant women received ANC services one or more times during pregnancy in Syria; 93.0 % of deliveries were attended by a skilled professional [
12]. Declines in antenatal care coverage may in part be related to cost. Antenatal care in Syria was free or inexpensive relative to services in Jordan and Lebanon [
37‐
39]. A cost barrier may also explain the higher enrollment in Za’atri refugee camp, where antenatal services are free and relatively easy to access compared to services out of camp in Jordan. Previous research from Lebanon suggests that unavailability of a reproductive health clinician is also a common barrier to receiving ANC [
40]. The observation from our assessments that some women evidently buy iron-folate supplements but do not go for ANC visits may support this finding.
While these data on antenatal care provide some indication that access to reproductive care could be improved, we caution against over interpretation of these findings. Several studies have reported higher utilization of ANC services than we document. In Lebanon, a 2012 clinic based assessment found that among Syrian refugees, 73 % of pregnant women reported at least one antenatal care visit [
40]. A non-random, convenience sample of refugees in Lebanon estimated that 83 % of pregnant refugee women attended at least one ANC visit [
41]. In Jordan, a nationally representative surveys in 2014, estimated that 82 % of Syrian refugees had made at least one ANC visit while pregnant [
37]. Further analysis to explore whether these differences in coverage estimates are attributable to variations in selection of respondents, or whether they reflect true differences in health seeking behaviors, may be useful for scaling up reproductive health services in the region.
Prevalence of diarrhea in children was assessed in all surveys. Two week prevalence of diarrhea was highest in North Lebanon (33.3 %) and Iraq’s Domiz refugee camp (32.8 %). These data suggest an increase in diarrhea prevalence, compared to that reported among children 0–59 in Syria pre-crisis (8.1 %) or the Iraqi population (14.8 %) [
14,
34]. Notably, the prevalence of diarrhea in these settings is also high relative to that reported from other refugee camps, including Dadaab camps in Kenya (13.5 %),Tongorara camp in Zimbabwe (21.4 %), and Mai-Ain camp in Ethiopia (29.7 %) [
42‐
44].
This high prevalence is particularly notable given reported access to improved sanitation and safe water. In Lebanon and out of camp in Jordan, nearly all refugees had access to a private toilet; fewer than 2 % of refugees in out of camp settings used public toilets. Even in Za’atri camp, less than a third of households reported using public toilets. For comparison, a global review of 90 refugee camps in 2005 found that on average latrines were shared by 27 people (range 5 to 1124 persons per latrine) [
45]. Global analysis have shown an association between prevalence of diarrhea and shared sanitation facilities [
46]. For out of camp refugees, the two most common primary sources of drinking water were running water and bottled water. Running water is an improved source of water as classified by the WHO, as is bottled water when another source of water is available for cooking and personal hygiene [
47]. In Za’atri, the primary source of water was water tankers, an unimproved source of drinking water [
47].
In all settings, reported prevalence of diarrhea was higher among children 0–23 months than children 24–59 months of age. In South and Bekaa regions of Lebanon, prevalence among younger children was more than twice that of older children. The finding of higher diarrhea prevalence among younger children is consistent with previous literature. A 2012 systematic review of data from 72 studies concluded that while rates varied by country, in all regions incidence rates of diarrhea were lower among children aged 24–59 months than among children aged 0–23 months [
48]. Data from a case-control study of rotavirus among refugees in Jordan, similarly found rotavirus prevalence was 2.4 times higher in children less than 24 months of age [
49].
Current breastfeeding was assessed as a measure of infant and young child feeding practices. Among children under 6 months of age, current breastfeeding was lowest in Domiz refugee camp in Iraq. In Domiz, as well as Za’atri, we document a drop-off in current breastfeeding during the second year of life (12–23 months), such that the proportion breastfeeding in both camps is notably lower than that reported from out of camp refugees in Jordan and Lebanon. These differences may be related to pre-conflict factors, such as the place of origin in Syria and related cultural practices, or may be related to differences in the conditions in the camp that affect breastfeeding practices.
Results reported in this manuscript are subject to several limitations. First, these surveys were initially designed in each country to best inform ongoing emergency response activities. As a result, different survey instruments were used in each country. Results presented here are derived from questions that were reasonably similar across the settings. Standardization of additional questions, particularly questions related to water, sanitation and immunization would have allowed for more comparisons. Second, the assessment in Iraq used a different sampling method than either Jordan or Lebanon assessments. The Iraq survey was designed to be representative of households with children under 5 years of age, not of all households. This methodology did not allow for comparison on several household level indicators with other surveys. Third, self-reported indicators, such as feeding practices, diarrhea morbidity and access to antenatal care, may be subject to recall bias. Finally, no surveys were organized in Egypt or Turkey. Around the time of our surveys, approximately 23 % of Syrian refugees were hosted in Turkey and 6 % were hosted in Egypt. [
4] Additionally, while the majority of Iraqi refugees have settled in urban areas, no assessments of non-camp refugees in Iraq were conducted [
50]. However, given the similarities in the nutritional status of refugees across the settings assessed, we suggest it is likely that acute malnutrition among the refugees in Turkey, Egypt and non-camp Iraq is similar to those included in our assessments.
Conclusions
Taken together, the seven surveys of Syrian refugees displaced to Iraq, Jordan, and Lebanon suggest that acute malnutrition was not a significant problem, however refugees from Syria, particularly those in Za’atri refugee camp, experienced elevated prevalence of anemia. In addition to these key findings, we highlight several additional indicators showing similar prevalence across settings and that are important triggers for public health action— low coverage of antenatal care relative to pre-conflict; high prevalence of diarrhea particularly among younger children 0–23 months of age; and low prevalence of continued breastfeeding after 12 months of age.
Given that high prevalence of acute malnutrition is commonly associated with complex humanitarian emergencies, the initial focus of nutrition programming for Syrian refugees focused on treatment of acute malnutrition. However, the findings of these surveys suggest that a drive to scale up treatment of acute malnutrition may not always be the best allocation of scarce resources. While quality treatment of malnourished children at existing health centers remains a priority, these data encouraged a shift in focus of response activities.
Based on the results of these surveys, nutrition partners on the ground prioritized interventions to address micronutrient deficiencies through dietary diversity advocacy and support of food fortification. In Za’atri camp, UN partners transitioned from ration distributions to cash vouchers, a change which allows refugees to shop in markets which offered a wider variety of food, including fresh produce and meat, than that included in the rations. Promotion of appropriate IYCF practices were prioritized to encourage breastfeeding and appropriate introduction of complementary foods. Nutrition partners also recommended inter-sectoral collaboration to promote quality water and sanitation. Given little indication of a quick resolution to the conflict, it was recommended to continue monitoring the situation through periodic nutrition surveys and analysis of facility based surveillance.
Acknowledgements
We acknowledge the contributions of the Syrian, Jordanian, Lebanese and Iraqi interviewers as well as the Syrian families who participated in this study. Additional gratitude is due to: United Nations High Commissioner for Refugees (Douglas Jayasekaran, Ann Burton, Caroline Wilkinson, Ellen Andresen, Aye Aye Than); United Nations Children’s Fund (Azzeddine Zeroual, Mohammad Amiri); World Food Program (Dorte Jessen); Mediar (Gabriele Faender); Jordanian MoH (Mohammed Tarawneh); International Orthodox Christian Charities (Sabeen Abdulsater, Marya Al Nawakil, Dima Ousta, Nanor Karaguezian); Republic of Lebanon MoPH; Duhok DoH; Kurdistan Regional Government; Jordan Health Aid Society; World Health Organization; and United Nations Population Fund.