Comparing the Health and Welfare of Refugees and Non-Refugees at the Outset of the COVID-19 Pandemic: The Results of a Community Needs Assessment

Refugees are a vulnerable population who experience signiﬁcant health disparities. They may also be at disproportionately high risk of adverse outcomes due to the COVID-19 pandemic. This paper presents the results of a community needs assessment to investigate the impact of the pandemic on health and welfare in a refugee relocation community in the United States. A multilingual data collection team made up of refugees surveyed 179 participants (128 refugees vs. 51 non-refugees). Only 55.9% of refugee respondents said they would be able to provide enough food for their family this week, compared with 84.0% of non-refugees (p < 0.01), and this difference was even greater for food next week (29.4% vs. 76.0%, p < 0.01). A non-signiﬁcantly smaller proportion of refugees reported knowing where to go if they were sick (69.1% vs. 81.6%, χ 2 = 2.8, p = 0.10), and being able to get the medicine they need (75.0% vs. 87.8%, p = 0.07), while signiﬁcantly fewer refugees reported feeling safe at home (72.8 vs. 87.8%, χ 2 = 4.5, p = 0.04). Overall, refugees fared worse on nearly every measure. These ﬁndings should motivate further observational research and inform clinicians about the signiﬁcant disparities in social determinants of health that refugees may experience during the pandemic.


Introduction
Resettled refugees are a vulnerable population who are at a significant health disadvantage compared with non-refugee immigrants (Reed & Barbosa, 2016). Before, during, and after displacement and relocation, refugees face significant challenges, including high prevalence of chronic and infectious disease, mental health issues, and unfulfilled basic needs (Abbas, et al., 2018). Further, the traumas and stressors related to initial displacement are frequently compounded by post-displacement stressors, including unemployment, poverty, insecure or inadequate housing, language barriers, social isolation, and anxieties over immigration status (Miller & Rasmussen, 2017;Grace, Bais, & Roth, 2018;Mirza et al., 2014). In countries such as the United States, discriminatory policies, stigma, and the prohibitive cost of care contribute to lack of access to the healthcare needed to prevent and treat health problems that refugees experience (WHO, 2013). Because population density in refugee communities tends to be high, the risk of communicable disease transmission is concomitantly high, and refugees' already tenuous economic welfare and healthcare access may be at particular risk during times of social upheaval related to public health emergencies (Kluge, 2020). These and other challenges indicate that the COVID-19 pandemic will likely exact a disproportionately large toll on refugee populations, resulting in higher infection rates and more severe negative outcomes related to government restrictions on movement and work when compared with non-refugees (Lau, 2020). In response to this situation, recent guidelines from an inter-agency standing committee, including the World Health Organization, the International Organization for Migration, the United Nations High Commissioner for Refugees, and the International Federation of Red Cross and Red Crescent Societies, urge those working with refugees to implement proactive strategies to address the unique vulnerabilities of refugees in the face of the pandemic (IFRC et al., 2020).
Clarkston, Georgia, a refugee relocation community about five miles from the city of Atlanta, has received approximately 17,000 refugees arriving in the US through the resettlement program established by Congress and signed into law by President Carter in 1980 (S. 643). Since that time, newcomers in Clarkston have been supported in their integration into the United States by a collection of non-profits, faith-based groups, social service organizations, and government organizations. Though the exact figures are difficult to determine, it is home to speakers of at least 60 languages from more than 40 countries. The Georgia State University (GSU) Clarkston campus, formerly the campus of Georgia Perimeter College (GPC), has long been a site of refugee service activities. After GPC consolidated with GSU, research collaborations began to evolve around the various needs of the community. A 2019 CDC-funded Prevention Research Center (PRC) grant to GSU to examine social determinants of health promised to accelerate that work. Due to the community-engaged nature of that grant, early work focused on eliciting and prioritizing community needs and establishing a Community Advisory Board to guide the work of the GSU PRC.
When the COVID-19 pandemic struck, we shifted focus to understanding the emerging needs of the community. Clinicians from a local community-based clinic asked that we collect data on the initial impact of COVID-19, in the hope that better understanding of the disease' effect on the social determinants of health in this community would facilitate a more targeted effort to mitigate the likely long-term effects of this pandemic. This paper presents the results of a community needs assessment during the early days of the COVID-19 pandemic that was conducted in response to that input from our community partners. The goal of the needs assessment was both to assess specific needs that could be addressed by local service agencies (health and healthcare access) and to examine how social determinants of health, such as economic factors and household makeup, may be affected during the COVID-19 pandemic. Though our focus was understanding specifically the needs of the refugee community, this broad survey of community residents included both refugee and non-refugee residents of Clarkston. This allowed us to compare responses of refugees to nonrefugees to understand the distinct needs of the refugee population.

Design and Setting
Data were collected between mid-April and mid-May 2020 via two collection mechanisms: phone and web surveys. To maximize recruitment in the absence of inperson interaction, multiple recruitment methods were used, including distribution of an English language web survey link through the PRC's Community Advisory Board (CAB), which is made up of community members, local social service organizations, government representatives, NGOs, faith-based organizations, ethnic community groups, and educators, as well as through informal networks that PRC staff have developed through their work in the community. Additionally, in an effort to reach non-English-speaking community members and those who preferred to be surveyed by phone, a recruitment survey was distributed through both the CAB and informal networks, which asked in what language participants preferred to be surveyed. These phone interviews were conducted by a multilingual data collection team made up of refugees with fluency in Arabic, Somali, Amharic, and Burmese/Karen, who translated the survey items into the participants' preferred language. Though no information was collected on participant country of origin, this set of language skills could facilitate communication with refugee residents from Syria, Iraq, Iran, the region informally recognized as Kurdistan, Sudan, Somalia, Ethiopia, Eritrea, and parts of Kenya and Tanzania, as well as members of the Burmese-Karen ethnic community. IRB approval for this study was provided by the Georgia State University Institutional Review Board.

Subjects
Of 179 respondents, 128 were refugees and 51 were non-refugees (71.5% vs. 28.5%). * Mean household size was larger among refugees than non-refugees (5.3 vs. 3.7, t = 4.1, p < 0.01), and refugee households had a larger mean number of children under the age of 18 (2.5 vs. 1.21, t = 4.1, p < 0.01). One hundred and fifty participants were surveyed via the English language web version of the instrument, while 29 were surveyed by phone using multilingual data collectors.

Measurement
The survey instrument (see Appendix 1) was developed in late March and early April 2020 through an iterative process, in which newly generated items were shared with community stakeholders on the CAB, as well as methodological and subject matter experts. The * N.B. Of the 128 respondents who reported being refugees, 9 also reported being born in the US. While there is no way to be sure why respondents identified themselves as being in these two mutually exclusive categories, they have been included in the analysis for two reasons: first, it is highly plausible that a refugee respondent, concerned about immigration status, would answer that they were born in the United States for reasons of safety, and second, it is also plausible that American-born children of refugees answered yes to both questions. process began by listening to the concerns of clinicians in the community, who then assisted the development of questions related to the major themes that they raised. The final instrument contained questions on household makeup, refugee status, economic security, food security, anxiety and mental health, health and healthcare access, COVID-19, and adherence to social distancing guidelines. The survey contains two sets of paired questions such that the second question in the pair was only asked of participants who responded "yes" to the first question. For example, the question, "Do you receive free school lunches at Clarkston High School or at bus stop drop offs," was only asked of those participants who responded yes to, "Do you have children in school?" Similarly, only those who reported having a laptop, tablet, or computer were asked whether that device was connected to the internet. A separate question was also asked regarding overall access to the internet. While a much more in-depth survey instrument would have conferred some advantages, the brevity of the instrument developed reflects the input of community stakeholders and methodologists that further questioning (for example, regarding country of origin) would have increased the invasiveness of the instrument and placed undue burden on participants.

Statistical Analysis
Data collected by the two mechanisms (web and phone), were merged into a single data set and analyzed using SAS Version 9.4. For all questions, univariate and bivariate descriptive statistics were calculated, with bivariate distributions stratified by refugee status. In the inferential phase, t-tests, chi-square tests, and Fisher's exact tests were used. T-tests were used for continuous outcomes, whereas chi-square and Fisher's exact tests were used for categorical outcomes. Some respondents chose not to answer certain questions, so the denominators used to calculate the percentages reported in the tables are in some cases smaller than the reported overall sample size. The number of missing responses ranges between 2 and 9 total.
Only 55.9% of refugee respondents said that they would be able to provide enough food for their family this week, compared with 84.0% of non-refugees (p < 0.01), and this difference was even greater for food next week (29.4% vs. 76.0%, p < 0.01). Refugees were also more likely to report acquiring food from a foodbank than non-refugees (23.0% vs. 10.0%, χ 2 = 3.9, p < 0.05). Further, a non-significantly greater proportion of refugee respondents reported losing income due to the COVID-19 pandemic than did non-refugee respondents (71.4 vs. 57.1%, χ 2 = 3.3, p = 0.07), and refugee respondents were significantly less likely than non-refugee respondents to say that they would be able to pay bills this month (42.9% vs. 71.4%, χ 2 = 11.6, p < 0.01) or next month (20.6% vs. 63.3%, χ 2 = 29.8, p < 0.01).

Health and Welfare
A non-significantly smaller proportion of refugees reported knowing where to go if they were sick (69.1% vs. 81.6%, χ 2 = 2.8, p = 0.10) and being able to get the medicine they needed (75.0% vs. 87.8%, p = 0.07), while significantly fewer refugees reported feeling safe at home (72.8% vs. 87.8%, χ 2 = 4.5, p = 0.04). Regarding information sources, the least frequent source for both refugees and non-refugees was "doctor or nurse," and refugees were especially likely to not use a doctor or nurse for information compared to non-refugees (18.4% vs. 36.7%, χ 2 = 6.6, p = 0.01). Refugees were significantly less likely to report getting information from the internet (52.8% vs. 77.6%, χ 2 = 9.0, p < 0.01) and were less likely to list avoiding close contact as a method of prevention than non-refugees (85.5% vs. 98.0%, p = 0.02).

Discussion
These findings warrant attention, given that they indicate that refugees' health and welfare may be at greater risk during the pandemic and subsequent lockdown compared to the health and welfare of their nonrefugee counterparts. Further, these findings give an empirical account of the areas in which those risks may be asymmetrically distributed. While some of the differences between refugees and non-refugees reported here likely predate the pandemic, some findings indicate that refugee health and welfare may be more negatively impacted by the COVID-19 pandemic and ensuing social and economic upheaval than that of nonrefugees. For example, the fact that refugees were less likely to list avoiding contact with others and less likely † All references to significance in the reported results refer to statistical significance, not practical significance. Test statistics are reported for chi-square and t-tests. Lack of reported test statistic indicates a Fisher's exact test.  to report getting their information about the virus from a healthcare professional than non-refugees is a cause for significant concern (though in the latter case, a minority of participants in both groups reported doing so). Especially given recent findings linking the proportion of immigrants in a community, as well as household size, to higher COVID-19 rates, contact avoidance is a critical tool for protection against transmission (Figueroa et al, 2020). Further, with the proliferation of misinformation about the virus and its spread, having reliable sources of information would be of considerable value.
There are also reasons to believe that some of the disparities observed in this sample will result in more severe negative health outcomes among refugees, even if they predate the pandemic altogether. It has been demonstrated that refugee households that report one member with acute illness, poor mental health, or poor overall health are more likely to report food insecurity (Logie et al., 2020). Food insecurity, for which the difference in proportion between refugees and nonrefugees is one of the most striking in this study, has been linked with depression in refugees, over and above the malnutrition that is its most direct impact; this phenomenon may at least partially explain why a greater proportion of refugees reported feeling unsafe at home than did non-refugees (Ghattas et al., 2015). Despite a relatively low proportion of respondents indicating that they had symptoms similar to those of symptomatic COVID-19 infections, it is increasingly clear that many infected people show minimal or no symptoms and that asymptomatic individuals can still spread the disease (Furukawa, Brooks, & Sobel, 2020;Ghandi, Yokoe, & Havlir, 2020). This reality, coupled with the larger household size and higher population density of refugee communities, suggests the possibility of elevated transmission risk. Given that a greater portion of refugee respondents also reported not knowing where to go if they were sick and not being able to acquire the medicines that they need, the prospect of elevated COVID-19 incidence in this community is particularly worrisome.

Limitations and Strengths
This study has three major limitations that may impact both generalizability and the possibility of causal attribution: the cross-sectional design, the convenience sampling method, and the limited sample size. A larger sample size, in addition to increasing our inferential models' capacity to detect differences between refugees and non-refugees, would potentially have enabled the use of more sophisticated multivariate modeling frameworks to test more complex hypotheses regarding the associations between survey items. A probability sampling method would have provided some protection against the possibility of sampling bias inherent in convenience sampling, as well as improving external validity. Further, the cross-sectional nature of this study precludes the possibility, for most questions on the survey instrument, of determining with certainty whether and to what extent the observed differences between refugee and non-refugee respondents can be attributed to the pandemic or predate it altogether. Because of the diversity present in the community, a wider range of linguistic skills within the data collection team would have allowed us to survey participants with a broader array of cultural backgrounds, including West Africans from the Democratic Republic of Congo and other ethnic communities from Burma (i.e., Rohingya and Chin).
The difficulty of obtaining these data, especially during a period of widespread social restrictions, should be taken into account when assessing these issues. To our knowledge, no other published study has reported empirical findings comparing unmet needs of refugees and non-refugees during the early months of the COVID-19 pandemic. Overall, we observed a wide range of differences between refugees and nonrefugees, consistent with the contention that refugees are at a broad health disadvantage, and this empirical trend is worthy of note for clinicians and researchers alike.

Conclusions
Overall, the findings of this needs assessment underscore the significant ways in which refugees experience greater health risks than their non-refugee counterparts and raise concerns about the possibility that they may be at elevated risk of negative health outcomes due to the pandemic and subsequent social and economic restrictions. These findings should: 1) influence clinicians and other social service providers who work with refugees to be more cognizant of the elevated risks their refugee clients may experience; 2) induce the development and implementation of intervention strategies to address those risks; and 3) highlight the need for more robust observational research to better understand the impact of the pandemic and lockdown within resettled refugee communities. tion 4.0 International License (CC BY 4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding Source: This publication was supported by Cooperative Agreement number 1U48 DP006393, funded by the Centers for Disease Control and Prevention. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention or the Department of Health and Human Services.