Abstract
Refugees in the United States have high rates of chronic disease. Both long-term effects of the refugee experience and adjustment to the U.S. health environment may contribute. While there is significant research on health outcomes of newly resettled refugees and long-term mental health experiences of established refugees, there is currently little information about how the combined effects of the refugee experience and the U.S. health environment are related to health practices of refugees in the years and decades after resettlement. We examined cross-sectional survey data for Cambodian refugee and immigrant women 35 to 60 years old (n = 160) from an established refugee community in Lowell, Massachusetts, to examine the potential contributors to health behaviors and outcomes among refugees and immigrants postresettlement. In our representative sample, we found that smoking and betel nut use were very low (4% each). Fewer than 50% of respondents walked for at least 10 minutes on 2 or more days/week. Using World Health Organization standards for overweight/obese for Asians, 73% of respondents were overweight/obese and 56% were obese, indicating increased risk of chronic disease. Depression was also high in this sample (41%). In multivariate models, higher acculturation and age were associated with walking more often; lower education and higher acculturation were related to higher weight; and being divorced/separated or widowed and being older were related to higher risk of depression. The interrelated complex of characteristics, health behaviors, and health outcomes of refugees merits a multifaceted approach to health education and health promotion for long-term refugee health.
The United States has resettled over 3 million refugees since 1975, including between 55,000 and 73,000 annually since 2010 (U.S. Department of Health and Human Services, 2015). U.S. refugees undergo initial health screenings primarily to identify and treat infectious diseases but currently receive little guidance on preventing chronic health conditions (Dookeran, Battaglia, Cochran, & Geltman, 2010). While many refugees arrive malnourished (Rondinelli et al., 2011), evidence suggests that some arrive with high rates of overweight/obesity, hypertension, and cardiovascular disease risk (Culhane-Pera, Moua, DeFor, & Desai, 2009; Dookeran et al., 2010). Despite enormous resilience in surviving trauma and establishing new communities in a foreign environment, refugees have increasing rates of chronic disease in the decades following resettlement (Grigg-Saito, Och, Liang, Toof, & Silka, 2008; Kinzie et al., 2008; Rondinelli et al., 2011; Sorkin, Tan, Hays, Mangione, & Ngo-Metzger, 2008).
Refugee trauma (Boscarino, 2004) and its related depression (Moussavi et al., 2007) can increase risk of chronic disease. Refugee food deprivation is linked to increased risk of overweight/obesity (Peterman et al., 2010). Limited physical activity, smoking, and betel nut use may contribute to overweight/obesity and poor health (Edberg, Cleary, & Vyas, 2010; Nelson & Heischober, 1999), especially in locations with adverse weather conditions (Tovar et al., 2013). Refugee women have elevated risk of poor health due to higher rates of depression; lower literacy, English language skills, and education; lower acculturation; higher prevalence of being widowed; lower social support; and lower reported physical activity than refugee men (Barnes, Harrison, & Heneghan, 2004; Deacon & Sullivan, 2009; Lowell Community Health Center, 2005). These characteristics can increase risk of poor health among immigrants and refugees (Edberg et al., 2010; Luppino et al., 2010).
Substantial research has focused on identifying health behaviors and risk factors of newly arrived refugees (Barnes et al., 2004; Baughan, White-Baughan, Pickwell, Bartlome, & Wong, 1990; Carroll et al., 2007; Dookeran et al., 2010; Renzaho & Burns, 2003; Rondinelli et al., 2011). Such research can be applied to easing early resettlement transitions. For long-resettled refugees and immigrants in the same communities, however, most research has focused on mental health (Marshall, Schell, Elliott, Berthold, & Chun, 2005; Vaage et al., 2010). Information on contributors to refugee health postresettlement and on targeted ways to improve refugee health may help refugee agencies, policy makers, educators, and service providers identify health promoters and stressors and improve community health (Krieger et al., 2002).
Because many Cambodians have lived in the United States since the mid-1980s in large communities, research with Cambodians provides an excellent opportunity to examine health behaviors and outcomes and their potential contributors in an established refugee community. The overall study from which this article is derived was designed based on a conceptual framework detailing the relationship between refugee and immigrant experiences (past food experiences, food beliefs/knowledge, and personal characteristics) and health-related practices and outcomes (dietary practices, food security, weight status, depression, and physical activity) in a large Cambodian community in Massachusetts. The Cambodian community in which this research is based includes both refugee (arrived before the early 1990s) and immigrant (arrived since the early 1990s) women as a single population. In this community, some women are legally considered immigrants because of the date of their arrival in the United States, but they share common experiences of surviving the Khmer Rouge and living in the same cultural environment and thus share many potential pre- and postimmigration disease risk factors. For this reason, this article uses the term refugee in reference to both refugees and immigrants in this community. We have previously detailed how severe and prolonged past food deprivation is related to increased risk of overweight/obesity (Peterman et al., 2010), discussed how personal characteristics are related to current dietary practices (Peterman, Silka, Bermudez, Wilde, & Rogers, 2011), and detailed the relationship between personal characteristics and food security (Peterman, Wilde, Silka, Bermudez, & Rogers, 2013). This article contributes to further understanding refugee health relationships by describing prevalence of and characteristics related to walking, smoking, betel nut use, weight, and depression, and by considering relationships with potential health determinants (acculturation, education, English literacy, income, and marital status) in the same population.
Methods
Study Design and Population
Lowell, Massachusetts, has the second largest Cambodian community in the United States (Grigg-Saito et al., 2008). Our study of Cambodian women 35 to 60 years old (YO) in Lowell was conducted in 2007-2008 by the Cambodian Mutual Assistance Association of Greater Lowell, Inc., and the Lowell Community Health Center in conjunction with the U.S. Centers for Disease Control and Prevention (CDC) Cambodian Community Health 2010 program. The Institutional Review Board of the University of Massachusetts Lowell approved all study procedures, and the Tufts Institutional Review Board accepted all University of Massachusetts Lowell review decisions. The age range was based on the main goal of the study: examining dietary practices among refugee women who survived the rule of the Khmer Rouge.
We constructed a survey using questions from previous surveys in the community (Lowell Community Health Center, 2005) and with focus groups. The survey was translated into Khmer, the primary language in Cambodia; blind back-translated into English; and tested with a convenience sample of 20 women 35 to 60 YO to ensure questions were understandable and translations were accurate. Except for one question in measuring acculturation, testers reported no difficulty with the final survey.
Participants were randomly selected from a list of Cambodian households in Lowell that was derived from a list of Cambodian names from the 2007 Lowell City Census, the 2007 electronic Verizon phone book, a list of registered voters of Cambodian heritage maintained by the University of Massachusetts Lowell, and lists of clients from multiple Cambodian-serving agencies. The final sampling frame included 3,735 nonduplicative households (11,700-25,700 individuals based on average survey household size of 5.0 ± 1.9). At the time of the study, official estimates of Cambodians in Lowell were under 15,000 (U.S. Census Bureau, 2012), while unofficial sources estimated 25,000 (Grigg-Saito et al., 2008). Each household was assigned a number, and Excel’s random number generator was used to select households.
Trained bilingual survey administrators contacted selected households to determine eligibility. They returned to unresponsive households until they made a contact in most cases. Administrators obtained informed consent, read the survey in the preferred language of the participant, and recorded answers. They collected height and weight using standardized equipment (Model 68978 Thinner Digital scale for weight, Seca 214 portable stadiometer for height) and procedures. From the 196 households contacted with eligible women or where there was no in-person contact, 160 women completed surveys (81.6% response rate).
Analysis
Health Behaviors and Outcomes (Dependent Variables)
Health behaviors reported in this article are physical activity, smoking, and betel nut/tobacco use. Physical activity categories were based on how many days/week each respondent reported walking for at least 10 minutes without stopping: <2 days/week (up to 50th percentile) and ≥2 days/week (above 50th percentile); this question was taken from a previous survey in this community (Lowell Community Health Center, 2005). Smoking and betel nut/tobacco use categories were based on current reported and observed practices.
Health outcomes reported in this article are weight and depression. Body mass index (BMI) was calculated from measured height and weight. Categories were determined using BMI cutoffs from CDC and a recommendation from a World Health Organization (WHO) report: ≥23 kg/m2 (overweight/obese WHO), ≥25 kg/m2 (obese WHO, overweight/obese CDC), and ≥30 kg/m2 (obese CDC; CDC, 2015; Inoue et al., 2000). WHO report categories were included because Asians have elevated risk of weight-related chronic disease at low BMIs (Inoue et al., 2000). Depression was measured with the validated Harvard Program in Refugee Trauma’s depression scale and cutoff calculations. Respondents rated how each of 14 statements described themselves over the past week on a scale of 1 to 4, with 1 representing not at all and 4 representing extremely. An example question is, “Feeling hopeless about the future.” Average ranking was calculated for respondents with ≥11 responses. An average score of ≥1.75 was classified as “depressed” (Harvard Program in Refugee Trauma cutoff; Mollica et al., 1992).
Participant Characteristics (Independent Variables)
We assessed relationships between health behaviors/outcomes and demographic participant characteristics that are related to health outcomes in refugee populations. Low education and low literacy are linked to poor health (Edberg et al., 2010; Luppino et al., 2010). In some immigrant populations, social stigma of divorce acts as a barrier to good health (Bauer, Rodriguez, Quiroga, & Flores-Ortiz, 2000). Widowed refugee women have high rates of poor mental health (Sabin, Lopes Cardozo, Nackerud, Kaiser, & Varese, 2003). In the United States, higher acculturation is linked to poor health outcomes, particularly when refugees and other immigrants adopt poor physical activity and dietary practices (Franzen & Smith, 2009; Satia-Abouta, Patterson, Neuhouser, & Elder, 2002).
Education was categorized as ≤1 year of school, some school (including some elementary and/or some high school in Cambodia, Thai refugee camps, or the United States), and high school (HS) graduate. Categories were based on the sporadic and interrupted nature of education in this population. Many respondents had little/no education in Cambodia or Thai refugee camps. Those who arrived in the United States as adolescents enrolled in high school, regardless of the amount of education they had previously, so standard grade levels do not appropriately assess education attained.
Marital status was based on response categories from a previous community survey (Lowell Community Health Center, 2005): single/never married, married/cohabitating, separated/divorced, and widowed. Because only two respondents reported being single/never married, this category was omitted from regression models where odds ratios cannot be estimated for the cell size.
The Psychological Acculturation Scale (PAS; Tropp, Sumru, Coll, Alarcon, & Garcia, 1999) was used to assess acculturation, rather than a proxy for exposure to American culture such as length of time in the United States or English literacy. Respondents ranked how much they identified with Cambodians or Americans for nine questions on a 1- to 5-point scale, with 1 representing identifying only with Cambodians and 5 representing identifying only with Americans. An example question was, “With which group of people do you feel you share most of your beliefs and values?” The original PAS included 10 items, but because one item presented difficulty for translators, testers, and participants, it was omitted from analysis. Average score was calculated for respondents with ≥6 of 9 questions answered. After translation and blind back-translation, this scale was pretested with 15 community members to ensure that the translation was meaningful. This pretesting was in addition to the full survey pretesting. Cronbach’s alpha was .87 for the measure, indicating strong internal consistency in this population. Spearman’s rho correlations between PAS and common measures of acculturation indicated acceptable validity (years living in United States: r = .353, p < .001; age at immigration: r = −.301, p < .001; reported speaking English at home at least occasionally: r = .360, p < .001; reported ability to read English easily: r = .250, p < .001; U.S. high school graduate: r = .242, p < .001).
Descriptive information with health outcomes for English literacy and years living in the United States is presented in Tables 1 and 2. Multivariate models do not include English literacy or years living in the United States to avoid collinearity between acculturation, English literacy, years living in the United States, and education.
Personal Characteristics (Categorical) by Health Behaviors and Outcomes.
Note. BMI = body mass index.
p values shown are for comparison of percentages between categories, chi-square analyses, and for comparison of means (BMI), t tests, and analysis of variance.
Personal Characteristics (Continuous) by Health Behaviors and Outcomes.
Note. BMI = body mass index.
p values shown are for comparison of means between categories, analysis of variance (>2 categories) and t tests (2 categories).
Respondents reported household income from all sources. Income to poverty ratio was calculated using household size and 2007 U.S. Census data (U.S. Department of Health and Human Services, 2008). Age was included as a continuous variable in self-reported YO.
Statistical Analysis
Descriptive statistics for health behaviors and outcomes were calculated overall and by population characteristics. Bivariate relationships were estimated between independent variables and health behaviors and outcomes using t tests and analyses of variance for continuous independent variables and chi-square analyses for categorical independent variables. Multivariate logistic and linear regression models were constructed for each health behavior/outcome and included variables that had a bivariate relationship (p < .10) with the behavior/outcome. Significance was set at p < .05 for all multivariate models. Final interpretation of relationships focuses on results in multivariate models. All analyses were computed using SPSS (Version 20.0, IBM, Chicago, IL).
Results
Sample Characteristics
Results are presented for the 154/160 women with complete data for all measures. Education was low: 25% had ≤1 year of education, 53% had some school, and 21% were HS graduates. Over three fourths (79.2%) were not literate in English. Most respondents (76%) were married or cohabitating, 12% were divorced/separated, and 12% were widowed. Mean acculturation was 2.1 ± 0.7, and 94% of respondents had a score <3 (3 = identifying equally with Americans and Cambodians). Mean income to poverty ratio was 1.14 ± 0.75, indicating very low income. Mean age (46.6 ± 7.5) was constrained by the deliberate sampling of women 35 to 60 YO. Participants had lived in the United States for an average of 18.7 ± 7.7 years (Tables 1 and 2).
About half (49%) of respondents reported walking for at least 10 minutes on ≥2 days/week. Few reported currently smoking (4%) or using betel nut (4%). Mean BMI was 25.9 ± 4.1 kg/m2. About 73% had a BMI ≥ 23 kg/m2 (WHO overweight/obese), 56% had a BMI ≥ 25kg/m2 (WHO obese; CDC overweight/obese), and 16% had a BMI ≥ 30 kg/m2 (CDC obese). Forty-one percent were categorized as depressed.
Simple Relationships
Acculturation, marital status, and age were related to walking ≥2 days/week at p < .10 in bivariate relationships. Education, acculturation, and age were related to weight at p < .10 in bivariate relationships. Marital status, income, and age were related to depression at p < .10 in bivariate relationships (Tables 1 and 2). Thus, the multivariate model for walking included acculturation, marital status, and age; the multivariate model for weight included education, acculturation, and age; and the multivariate model for depression included marital status, income, and age. Because there were no statistical relationships between characteristics and current smoking or betel nut use, no multivariate models were estimated for these behaviors.
Multivariate Relationships
In the multivariate model for physical activity, each 1-point increase on the PAS was associated with a 1.77 increase in the odds of walking for at least 10 minutes on ≥2 days/week (p = .035), and each year older was associated with a 1.07 increase in these odds (p = .013). Marital status had no relationship with physical activity (Table 3).
Multivariate Regression Model for Predictors of Physical Activity (n = 144).
Note. Variables in model: marital status, acculturation, and age.
Compared with HS graduates, those with ≤1 year school had 3.98 higher odds of having a BMI ≥ 23 kg/m2 (p = .028), 4.22 higher odds of having a BMI ≥ 25 kg/m2 (p = .010), and an average BMI that was 2.53 kg/m2 higher (p = .014), all else equal. Compared with HS graduates, those with some school had higher odds of having a BMI ≥ 25 kg/m2 (odds ratio = 3.42, p = .008) and had a BMI that was 1.93 kg/m2 higher (p = .026). In these multivariate models, neither acculturation nor age had a relationship with BMI (Table 4).
Multivariate Regression Models for Predictors of Weight Status (n = 147).
Note. BMI = body mass index; WHO = World Health Organization; CDC = U. S. Centers for Disease Control and Prevention; OR = odds ratio; CI = confidence interval. Variables in model: education, acculturation, and age.
Depression was related to marital status and age in the multivariate model. Women who were divorced/separated had 7.21 higher odds of being depressed (p = .002), and those who were widowed had 4.47 higher odds of being depressed (p = .019), compared to women who were married/cohabitating. For every year older a respondent was, she had 1.08 increased odds of being depressed (p = .010, Table 5), with all else equal.
Multivariate Regression Model for Predictors of Depressed Status.
Note. Variables in model: marital status, income to poverty ratio, and age.
Discussion
The results of this analysis provide some encouraging information about the health behaviors of women ages 35 to 60 YO in this established Cambodian refugee community. The current smoking rate of 4% was slightly lower than rates reported among Cambodian women in a 2009 WHO bulletin (4.5% among women 37-48 YO, 5.8% among women >48 YO; P. N. Singh et al., 2009), less than one fourth of the current U.S. smoking rate for women (Centers for Disease Control and Prevention, 2010), lower than a report of current smoking among all Asian immigrants (6.1%; Gomez, Kelsey, Glaser, Lee, & Sidney, 2004), and much lower than a 2002 survey of the same Cambodian community (11% of women 25+ YO; Lowell Community Health Center, 2005). Betel nut use does remain a concern for a small percentage of community members, as it is linked to oral and esophageal cancers (Nelson & Heischober, 1999).
The results also provide some areas for concern. With fewer than half of respondents reporting that they walked for at least 10 minutes on ≥2 days/week, physical activity appears to be low, although not necessarily lower than other reports of immigrant populations. Gomez et al. (2004) found that only 31.3% of Asian immigrants engaged in low-impact physical activity at least 4 days/month. The low rate of physical activity that we saw may be at least partially due to the climate of Massachusetts (Tovar et al., 2013).
Using WHO report cutoff recommendations for Asians, almost three fourths of the sample was at risk for weight-related chronic disease with a BMI ≥ 23 kg/m2, and more than half was at greatly increased risk with a BMI ≥ 25 kg/m2. Using CDC standards, only slightly more than half would be considered at increased risk (overweight), and a small percentage would be considered at greatly increased risk of disease (obese). Taken with our previous report that women in this sample who experienced severe and prolonged food security had an elevated risk of overweight and obesity (Peterman et al., 2010), this emphasizes the need for health educators and program planners to carefully consider the potential health impact of higher end normal BMI levels (23-25 kg/m2).
Depression was high among respondents, consistent with previous findings in long-resettled Cambodians (Marshall et al., 2005). Our analysis did not show a significant difference in depression by acculturation score. This may have been due to our small sample size and may also have been due to the nearly uniformly traumatic experience shared by this sample that we have previously reported (Peterman et al., 2010). Refugee trauma has long-lasting effects, and these results suggest that additional mental health support for individuals in refugee communities is needed far beyond the initial resettlement period.
The multivariate analyses provide some potential guidance for where to best target education efforts and services. More acculturated respondents were more likely to walk ≥2 days/week, which is similar to results in other Asian populations (Novotny et al., 2012; Wang, Quan, Kanaya, & Fernandez, 2011). However, many migrants decrease physical activity on arrival in the United States (Harrison et al., 2005) and have higher weight with longer stay in the United States (Himmelgreen et al., 2004). We did find that those with higher acculturation were at higher risk of elevated BMI, consistent with many other analyses of refugees and immigrants (Harrison et al., 2005; Perez-Escamilla & Putnik, 2007; G. K. Singh & Siahpush, 2002), although length of stay (often used as a proxy for acculturation) was not related to weight status.
Low education remained a strong predictor of overweight/obesity when controlling for other variables. In the general U.S. population, those with lower education have a higher rate of obesity/overweight (McLaren, 2007), but among Asian immigrants, Bates, Acevedo-Garcia, Alegria, and Krieger (2008) reported that those with higher education had the highest rates of overweight/obesity. Our finding may result from differences in the immigrant experience for refugees versus other Asian immigrants.
The analysis also provides information about community members who are most at risk for depression. Single women, whether divorced/separated or widowed, and older women had high rates of depression, even when controlling for other potential contributors. Because many families are disrupted with one or both adults killed before refugee resettlement (Clinton-Davis & Fassil, 1992), this is likely an issue that will be seen in other refugee groups.
The health issues (physical activity, weight, and depression) and characteristics related to them (income, education, marital status, age, and acculturation) that we discuss in this article are interrelated in refugee communities. Low levels of physical activity combined with high levels of overweight/obesity may result from acculturative changes in dietary practices (Satia-Abouta et al., 2002), U.S. work and social environments, transportation options, time constraints (Himmelgreen et al., 2004), and difficulty in understanding or navigating the U.S. food and health environments because of low education, low acculturation, or past food experiences (Hadley, Patil, & Nahayo, 2010; Peterman et al., 2010). Refugees who face both enormous stress and trauma preresettlement (Mollica, 2006) and postresettlement as minority individuals in a dominant resettlement environment (Marshall et al., 2005) and have relatively low incomes (Belle & Doucet, 2003) are susceptible to depression and other mental and physical health concerns. Widowed or divorced women likely have increased barriers to healthy behaviors and increased risk of poor outcomes because of increased personal responsibility and decreased support.
This article provides information on a limited number of factors in the complicated framework of relationships between the experiences, characteristics, and health behaviors and outcomes in refugee communities. Thus, these results must be addressed jointly with other information about contributors to health (Garber & Nigg, 2012), such as changes in dietary practices (Peterman et al., 2011), current and past access to food (Peterman et al., 2010, 2013), focused attention to mental health (Marshall et al., 2005; Vaage et al., 2010), and use of community strengths (Grigg-Saito et al., 2008).
In addressing refugee health, we suggest an early and sustained approach to chronic disease prevention that focuses on all literacy levels; includes physical activity, healthy dietary choices (supported by our previous work with this population and the high rates of overweight/obesity in this sample), and mental health; and provides special consideration for divorced/separated and widowed women. While our work does not provide comparisons with men in this community, it does link specific gender-related characteristics with health behaviors and outcomes, suggesting that low education and single status of some refugee women may be mechanisms for increased risk of chronic disease.
Strengths and Limitations
This study was conducted through a community agency in partnership with a community health program, which increased community engagement and survey response rates. Additionally, because the sample was randomly drawn from a representative household list and had a high response rate, results can be applied to other women in the population.
Our sample size led to large variance in some measures and proscribed estimating relationships between behaviors and outcomes. Due to the low number of days/week that most women reported walking ≥10 minutes/day, we could not assess variability among characteristics with a higher level of physical activity. Additionally, the physical activity measure was limited to days/week each respondent walked for ≥10 minutes, and thus likely total physical activity could be underestimated. All measures except height and weight were self-reported, and thus are subject to participant bias and misreporting. Because this study was cross-sectional, associations cannot be assumed to be causal, and we are not able to assess changes in behavior over time. We did not collect data on chronic health conditions or social support; future work with larger sample sizes should address these important issues.
Conclusion
Individuals in refugee communities in the United States overcome enormous obstacles in their home countries and on resettlement. If unaddressed, long-term complications of the refugee experience, such as low education, stress and trauma, and low income on resettlement may contribute to poor health as refugees live in the U.S. health environment. Health educators and program planners and developers can help refugees successfully navigate the postresettlement environment and manage the lasting effects of the refugee experience if they take a long-term, multifaceted approach to addressing past experiences and current environment.
Footnotes
Acknowledgements
The authors thank the community members who participated in this research with their time and generosity. We thank the extraordinary survey administration team: Botom Sokieng, Jeannine Chhoeum, Saman Hing, Chanthyda Hout, Julie Hak, and Sam An Um. We also thank Boroeuth Chen and Timothy Mouth for translations and backtranslations of written material.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded by the Cambodian Community Health program (Centers for Disease Control and Prevention agreement U50/CCU12215), the Blue Cross and Blue Shield of Massachusetts Foundation Catalyst Fund, the Feinstein International Center at Tufts University, Project Bread, and the Cambodian Mutual Assistance Association of Greater Lowell, Inc.