Growth Trajectories of Refugee and Nonrefugee Children in the United States
BACKGROUND AND OBJECTIVES: Limited data examine longitudinal nutrition outcomes of refugee children after United States resettlement. Among refugee children, our aims were to (1) assess the changes in weight-based nutritional status between baseline (0–3 months) and 10-24 months after arrival and (2) compare the BMI (BMIz) or weight-for-length z score (WFLz) trajectories to nonrefugee children for up to 36 months after arrival.
METHODS: We conducted a retrospective study of refugees aged 0-16 years from Washington and Pennsylvania and compared them with an age and sex-matched nonrefugee low-income sample from Washington. Data included anthropometric measurements from the initial screening medical visit and subsequent primary care visits. Multilevel linear mixed-effects regression models evaluated the change in BMIz or WFLz trajectory.
RESULTS: The study included 512 refugee and 1175 nonrefugee children. The unadjusted prevalence of overweight/obesity increased from 8.9% to 20% (P < .001) for 2- to 16-year-old refugees from baseline to 10-24 months. Refugees (2–16 years old) had a steeper increase in their BMIz per 12 months compared with nonrefugees (coefficient 0.18 vs 0.03; P < .001). Refugees <2 years old had a less steep increase in their WFLz per 12 months compared with nonrefugees (coefficient 0.12 vs 0.36, P = .002).
CONCLUSIONS: Older refugee children exhibited a higher risk of obesity than nonrefugees, whereas refugees <2 years old exhibited a slower increase in their risk of obesity than nonrefugee children. All age groups experienced increasing obesity prevalence. Targeted and culturally tailored obesity prevention interventions may mitigate health and nutrition inequities among refugee children.
- CDC —
- Centers for Disease Control and Prevention
- WHO —
- World Health Organization
What’s Known on This Subject:
Refugee children are at risk for obesity after US resettlement. It is unknown how their change in nutritional status differs compared with a reference group of nonrefugee US children.
What This Study Adds:
Refugee children (2–16 years old) had a greater increase in their BMI z score after US resettlement compared with low-income, nonrefugee US children, and differences existed among refugee children based on their country of origin.
Currently there are ∼60 million people displaced worldwide due to war and persecution, and half are children.1 Children comprise 30% to 50% of the ∼60 000 refugees annually resettled in >40 states in the United States.2–4 Refugee children arrive in the United States after a variety of migration experiences that influence their growth and well-being. Before US resettlement, refugee children may experience recurrent stressors including trauma, family separation, food insecurity, malnutrition, obesity, infectious disease, and environmental toxins.5–9 Medical screening shortly after US resettlement focuses on detecting related health concerns.7,10–13 However, there is a dearth of evidence establishing the long-term health and nutrition outcomes of refugee children after US resettlement.
Childhood nutritional status has important implications for cognitive development, economic opportunity, mental health, adult weight status, and chronic disease risk.14–16 Refugee children arrive in the United States with variable nutritional status based on their preresettlement experiences.9,12,17,18 Although several studies have reported an increase in the prevalence of obesity and obesity-related diseases among adult refugees and an increase in their BMI trajectories after US resettlement, the data among child refugees is limited.19,20 Two small studies demonstrated a trend toward increasing obesity prevalence among refugee children.21,22 However, obesity risk has not been examined in a large sample of refugee children.
Additionally, analyses of growth trajectories by country of origin among refugee children is lacking, and such studies are necessary to identify groups most likely to benefit from culturally and linguistically appropriate nutrition interventions, that is, among children from the current largest resettlement countries to the United States.2,3 Furthermore, no studies have compared the longitudinal growth of refugee children to an age- and sex-matched, low-income cohort of US children. Such comparisons would quantify the extent to which exposure to the US obesogenic environment places newly resettled refugee children at risk for obesity.
The primary aims of our study were to (1) assess the change in weight-based nutritional status of refugee children at baseline (0–3 months after arrival) and time 2 (10–24 months after arrival) and (2) compare the BMI or weight-for-length trajectories of refugee and nonrefugee US children for up to 36 months.
Study Population and Design
Refugee children aged 0 to 16 years were identified from 3 primary care sites. Children from site 1 arrived in King County, Washington, from January 1, 2012, to December 31, 2014; had their initial domestic medical screening visit at the Seattle-King County Public Health clinic; and subsequently initiated primary care at HealthPoint, a Federally Qualified Health Center in King County. The data collected from Seattle-King County Public Health was linked with the electronic medical record data from HealthPoint (n = 422).
Children from site 2 included refugees who arrived in Philadelphia, Pennsylvania, from January 1, 2008, to December 31, 2012, and had their initial domestic medical screening visit and subsequent primary care at Jefferson Family Medicine Associates at Thomas Jefferson University (n = 165).
Children from site 3 included refugee children who arrived in Philadelphia, Pennsylvania, from October 1, 2010, to December 31, 2012, and had their initial domestic medical screening visit and subsequent primary care at the Children’s Hospital of Philadelphia Refugee Health Clinic (n = 78).
The nonrefugee cohort included children initiating care from HealthPoint from 2008 to 2012. In total, HealthPoint provided care to 32 950 children during that time period, 86% of whom were insured by Medicaid. Low-income children in the United States are at risk for malnutrition, stunting, and overweight/obesity,23 and previous research describes socioeconomic status as a key confounder in weight outcomes.24,25 Therefore, we sought to compare the refugee children to nonrefugee children who were low income and received care during the same time period. English-speaking children were assumed to be nonrefugees and were matched to the refugee cohort by age, sex, and year of care initiation at a frequency ratio of ∼2:1. Children whose preferred language was not English were excluded from the nonrefugee cohort.
Inclusion and Exclusion Criteria
Children were included in the study if they were between 0 and 16 years old at the time of their initial domestic medical screening visit. Children were excluded if (1) their visit dates were implausible (eg, domestic medical screening visit preceded their overseas medical visit date or their subsequent follow-up preceded their domestic medical screening visit), (2) their country of origin was unknown, (3) if they did not have plausible height and weight measurements available during their first 3 months after resettlement, and (4) if they did not have at least 1 subsequent height and weight measurement available.
Age, sex, and country of origin were ascertained from the initial domestic medical screening visit.
For refugee children, the time since care initiation at the primary care clinic was based on months elapsed since arrival in the United States, up to 36 months. For nonrefugee children, it was based on the time since their first visit initiating care at the primary care clinic.
Weight and height/length measurements were obtained at the visits by medical professionals based on clinic protocols as a routine component of the physical examination.5,26,27 Following World Health Organization (WHO) and Centers for Disease Control and Prevention (CDC) guidelines for measuring anthropometry, standing height was measured for children ≥2 years old and supine length for children <2 years old.28
Children were categorized into 1 of 3 weight-based nutritional status categories: wasting, healthy weight, or overweight/obese (Table 1). Children were also categorized into the stunting category based on their age- and sex-appropriate height- or length-for-age z score (Table 1). The nutritional status categories are based on anthropometric definitions from the WHO 2006 growth standard definitions for children <2 years of age and from the CDC 2000 definitions for children ≥2 years old, which are the recommended definitions by the CDC for children in the United States.29 Additionally, the WHO 2006 growth standards for children ≥2 years old were included for reference (Supplementary Table 4). Following CDC guidelines, implausible growth measurements were removed, that is weight-for-length, BMI, or height/length-for-age z scores |≥5|.
This study was approved by the Institutional Review Boards of Seattle Children’s Hospital, Children’s Hospital of Philadelphia, Thomas Jefferson University, and the HealthPoint Research Council, and it was deemed exempt by Washington State.
Stata 13.1 (StataCorp LP, College Station, TX) and the program zanthro30 were used to calculate the height- or length-for-age, weight-for-length and BMI z scores.31 We used χ2 and Fisher’s exact tests for unadjusted comparisons of the nutritional status categories at baseline (0–3 months) and time 2 (10–24 months). The nutritional status category at baseline was based on their earliest available measurement, and time 2 was based on their last available measurement.
Our primary research question was whether the refugee and nonrefugee children demonstrated differential patterns in their BMI or weight-for-length z score trajectories for up to 36 months after care initiation. We assessed repeated BMI z scores (for 2–16 years old) and weight-for-length z scores (for <2 years old) within study subjects. To account for the within-subject correlations, we applied linear mixed effects regression models for our main analysis. We included group category (refugee vs nonrefugee), time (months since care initiation at each weight status assessment), and group-by-time interaction as primary predictors of interest in the models. The group-by-time interaction term estimates and tests whether incremental changes over time in BMI or weight-for-length z score were different across groups. In all models, we adjusted for a priori selected covariates: child baseline age, baseline BMI/weight-for-length z score, and sex. All models included a subject-specific random effect to account for correlations due to repeated assessments and a site-specific random effect to account for correlation by clinic site. In addition, we conducted subgroup analyses stratified by (1) baseline nutritional status category and (2) country of origin for the 4 most common countries.
A total of 512 refugee children and 1175 non-refugee children were included. The proportion of female children was similar between the cohorts (47.1 and 47.6%), whereas children in the refugee cohort were slightly older (mean: 6.8 years vs 6.4 years), and contributed fewer months in follow-up to the cohort (median: 3.8 months vs 5.9 months, Table 2). Children from 4 countries made up 86% of the refugee cohort: Bhutan (23.8%), Burma (21.7%), Iraq (18.0%), and Somalia (22.7%).
Prevalence of Nutrition Status Categories
A total of 214 refugee children and 610 nonrefugee children aged 2 to 16 years had BMI and height-for-age z scores at both baseline and time 2. Among refugee children, the prevalence of overweight/obesity increased from 8.9% to 21% for the refugee group (P < .001). The prevalence of overweight/obesity increased from 38.9% to 45.7% (P = .02) in the nonrefugee cohort. When the WHO anthropometric definitions (Supplementary Table 4) were applied to refugee children aged 2 to 16 years, the prevalence of overweight/obesity increased from 14.4% to 26.1% (P < .001).
Forty-five refugee children and 117 nonrefugee children <2 years old had weight-for-length and length-for-age z scores at both baseline and time 2. The prevalence of obesity increased from 0% to 16% (P = .003) for the refugee cohort and from 1.7% to 19.7% (P < .001) for the nonrefugee cohort. In the refugee and nonrefugee cohorts, the prevalence of wasting and stunting were not significantly different between baseline and time 2 for 2- 16-year-olds and <2-year-olds.
Growth Trajectory Models
The repeated measures of all 2- to 16-year-old from both cohorts (refugees: n = 439, nonrefugees: n = 985) were included in an adjusted linear mixed effects regression model, in which the refugee cohort had a steeper increase in their BMI z score per 12 months compared with the nonrefugee cohort (coefficient 0.18 vs 0.03; P < .001; Table 3; Fig 1). Similarly using the WHO definitions for children aged 2 to 16 years, the refugee cohort had a steeper increase in their BMI z score per 12 months compared with nonrefugee children (coefficient 0.16 vs 0.02; P < .001; Supplementary Table 5). Among those with healthy weights at baseline, refugee children had a greater increase in their BMI z score per 12 months than nonrefugee children (coefficient 0.17 vs 0.09, P = .02; Table 3; Fig 2). Children that were stunted at baseline had a steeper increase in their BMI z score per 12 months than nonrefugee children (coefficient 0.31 vs −0.33, P < .001; Table 3; Fig 2). Examining trajectories by country of origin, children aged 2 to 16 years from Burma and Somalia had steeper increases in their BMI z scores per 12 months than nonrefugee children (coefficient 0.45 and 0.15, respectively), whereas there was no difference comparing children from Bhutan (0.02) and Iraq (0.07).
In an adjusted linear mixed-effects regression model of children aged <2 years (refugees: n = 73, nonrefugees: n = 190), refugee children had a less steep increase in their weight-for-length z score per 12 months compared with the nonrefugee children (coefficient 0.12 vs 0.36, P = .002). Children <2 years old were not evaluated by nutrition status category or country of origin because of the sample size.
To our knowledge, this is the first study that examines the longitudinal change in nutritional status of refugee children after US resettlement compared with an age- and sex-matched nonrefugee group of primarily low-income children. Refugee children aged 2 to 16 years had an overall greater increase in their BMI z score than non-refugee children after care initiation. Those refugee children with a healthy weight or stunting at baseline had the greatest increases in BMI z score compared with nonrefugee children. Among refugee children, the prevalence of overweight and obesity more than doubled within 10 to 24 months from US arrival. By comparison, refugees <2 years old showed a smaller increase in their weight-for-length z score compared with nonrefugees, although their prevalence of obesity did increase over the study period. Thus, the risk for obesity appears to steadily increase among refugee children as they reside in the United States. Children aged ≥2 years are at considerably greater risk compared with nonrefugee children to become overweight and obese. Early intervention to prevent childhood obesity is critical due to the health implications of childhood obesity32–35 and risk for subsequent adult obesity associated chronic disease16 in this vulnerable population.
This increase in overweight and obesity among 2- to 16-year-old refugees is consistent with a study among 2- to 18-year-old refugee children resettled in Rhode Island,21 which reported a doubling of the prevalence of overweight and obesity from 17% at arrival to 35% 3 years later. A study in Minnesota (n = 69) among 0- to 18-year-old refugee children from sub-Saharan Africa reported that children who were initially underweight gained weight, and 2% of healthy weight children moved into the overweight category (BMI >85th percentile) within 1 year after resettlement.22 Our results differ from those of the Minnesota study, possibly due to the study time period and/or differences in countries of origin, follow-up duration, or sample size. Given our findings and those of others, obesity prevention should be an important area of focus for refugee children early in their resettlement experience.
In contrast to older refugee children, the young refugee children (<2 years old) had a significantly lower increase in weight-for-length z score compared with nonrefugee children, although both demonstrated an increase in obesity prevalence (16% and 19.7%, respectively). One possible explanation for this result may be a higher rate and longer duration of breastfeeding among refugee children. Immigrant populations have higher breastfeeding initiation rates and duration than women born in the United States,36 and breastfed infants tend to grow more rapidly in the first 2 months of life and less rapidly for the subsequent 3 to 12 months compared with non-breastfed infants.37 Therefore, higher rates of breastfeeding among refugee infants may be contributing to the smaller incremental change in weight-for-length z score observed in this cohort, however, this and other potential explanations require further study.
When the refugee child’s country of origin was taken into consideration, children (aged 2 to 16 years) from Burma (coefficient 0.45) and Somalia (coefficient 0.15) had higher gains in BMI z score compared with the nonrefugee children (coefficient 0.03; P < .001 and P < .02, respectively), whereas there was no difference comparing children from Bhutan (0.02) and Iraq (0.07) to the nonrefugee children. These results may capture differences in experiences after resettlement in the US obesogenic food environment.38 Refugees from different countries of origin may have unique ethnocultural experiences post-US resettlement that may be more or less protective. Although the differences in BMI z score between the Bhutanese and Iraqi children and the nonrefugee children were not significantly different, all children demonstrated an increase in their BMI z score rather than no change, which would be the ideal. Furthermore, these findings highlight the importance of presenting results by country of origin11,12 to tailor surveillance and interventions based on the needs of refugee populations.
Refugee children aged 2 to 16 years who were stunted at arrival had the largest increases in their BMI z score. Although children’s stunting status is defined by their low height-for-age, this increase in BMI z score signifies an increase in weight disproportionate to their gain in height. Stunted children living in countries undergoing a nutrition transition (ie, a shift in dietary consumption coinciding with economic changes and urbanization) have an increased risk of obesity, metabolic syndrome and cardiovascular disease in adulthood.39,40 The coexistence of stunting and overweight has been described in many populations worldwide,14 and is important to consider among individuals immediately transitioning into the US nutrition environment through migration.
Children in both the refugee and nonrefugee samples who were wasted had an increase in their BMI z scores, and children who were wasted and overweight/obese had a decline in their BMI z score. Despite the improvement in BMI z scores for refugee children who were wasted at arrival, the overall prevalence of wasting did not change from baseline to time 2. This may be because children are gaining weight but do not reach their desired age and sex standards during the study time period or because we did not have a sufficient sample size to detect the difference. Future studies on wasted refugee children may evaluate wasting recovery over a longer time period. Additionally, the prevalence of stunting did not significantly change from baseline to time 2. There is an increased global focus on stunting since the scale up of existing nutrition-specific interventions to prevent malnutrition are estimated to reduce stunting by only 20%.41 Larger and longer longitudinal studies of stunted refugee children may provide insight into approaches to support stunting recovery, which may influence malnutrition treatment in other global contexts.
This study has several limitations. The nonrefugee cohort of children may include refugee children who arrived before 2012 whose families’ preferred language was English. We believe that this potential misclassification would represent a small group of children and would bias results toward the null hypotheses. Second, the birthdate of some of the refugee children may have been unknown to the family and therefore listed as January 1 of a specific year (6.6%). This limited birthdate information could misclassify the child’s age and influence his or her nutritional status category. Refugee children were frequency-matched to nonrefugee children from Washington state only (and not from Pennsylvania) due to feasibility. There may be differences in the nutritional context of Washington and Pennsylvania, however, the inclusion of the clinic site in the models accounts for correlation of subjects within study sites. On the basis of the available data, the nonrefugee children were categorized by race/ethnicity and the refugee children by their ethnic country of origin; therefore, we were not able to match the 2 groups by race/ethnicity. Subdivision of children into narrower age categories was limited by our sample size and supports the need for a larger study to evaluate child growth among refugee children. Data on breastfeeding and other dietary factors were not available in this analysis and may influence child weight status.
Refugee children aged 2 to 16 years resettled in the United States have a steeper BMI increase than their nonrefugee US peers, which differs by country of origin. In contrast, refugee children <2 years old had a less steep weight-for-length increase than their nonrefugee US peers, although the obesity prevalence increased for both groups. Altogether these patterns suggest that protective mechanisms of obesity among refugee children wane over time so that refugee children are at an elevated risk of obesity. Targeted and culturally tailored interventions are necessary to mitigate health and nutrition inequities among refugee children.
We thank Ron Uno, Evan Oakes, MD, MPH, Liza Perpuse, MD, and the HealthPoint Research Council for their collaboration. We also thank Megan Fesinmeyer, PhD, for her statistical support early in the project.
- Accepted August 26, 2016.
- Address correspondence to Elizabeth E. Dawson-Hahn MD, MPH, Seattle Children’s Research Institute, M/S CW8-6, PO Box 5371, Seattle, WA 98145. E-mail:
FINANCIAL DISCLOSURE: Dr Dawson-Hahn received an honorarium from the Migration Policy Institute to pay the publishing fees for an unrestricted paper on refugee children.
FUNDING: Dr Dawson-Hahn’s time and effort was funded by the Ruth L. Kirschstein National Research Service Award (T32HP10002). Dr Dawson-Hahn received grant funding for this project from the Seattle Children’s Hospital Center for Diversity and Health Equity Small Grant Funding, and the Seattle Children’s Research Institute, Center for Child Health Behavior and Development Biostatistics, Epidemiology, Econometrics and Programming (BEEP) Core’s Stimulus Funding. Dr Yun’s time and effort was funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (grant 1K23HD082312) and the Agency for Healthcare Research and Quality (grant K12HS021706-02). Funded by the National Institutes of Health (NIH).
POTENTIAL CONFLICT OF INTEREST: Dr Yun is on the board of the Pennsylvania Immigration and Citizenship Coalition.
- United Nations High Commissioner for Refugees
- Martin DC,
- Yankay JE
- Martin DC,
- Yankay JE
- ↵Summary of Refugee Admissions as of October 31, 2015. Washington, DC: US Department of State, Bureau of Population, Refugees and Migration; 2015.
- Mitchell T,
- Jentes E,
- Ortega L, et al
- Dawson-Hahn EE,
- Greenberg SL,
- Domachowske JB,
- Olson BG
- World Health Organization
- Heney JH,
- Dimock CC,
- Friedman JF,
- Lewis C
- Centers for Disease Control and Prevention
- Olson CK,
- Naughton MP,
- Ortega LS
- Centers for Disease Control and Prevention
- World Health Organization
- Vidmar S,
- Carlin J,
- Hesketh K,
- Cole T
- Singh GK,
- Kogan MD,
- Dee DL
- DeBoer MD,
- Lima AA,
- Oría RB, et al
- Popkin BM,
- Richards MK,
- Montiero CA
- Bhutta ZA,
- Das JK,
- Rizvi A, et al; Lancet Nutrition Interventions Review Group; Maternal and Child Nutrition Study Group
- Copyright © 2016 by the American Academy of Pediatrics