Introduction

Since the passage of the 1980 Refugee Act, 3.2 million refugees have been admitted to the United States (U.S.) [1]. Internationally, refugees are defined as “people who have fled their countries to escape conflict, violence, or persecution and have sought safety in another country” [2]. The U.S. definition of refugees is “s a person who is unable or unwilling to return to their country of nationality (or country of last habitual residence, if stateless) because of persecution or a well-founded fear of persecution on account of race, religion, nationality, membership in a particular social group, or political opinion” [3]. Refugees have not yet entered the U.S., while asylees are defined as those who meet the definition of a refugee but either currently reside in the U.S. or are seeking admission at a point of entry [3]. Refugees come from across the globe, and the highest number of refugees relocating in the U.S. in the last ten years came from Burma, Iraq, the Democratic Republic of Congo, Bhutan and Somalia [3].

Refugees and asylees (hereafter, “refugees”) arriving to the U.S. experience a high burden of both communicable and non-communicable diseases, including diabetes, hypertension, tuberculosis, and hepatitis B [4, 5]. While refugees experience higher rates of communicable diseases than the host country populations [4], rates of non-communicable diseases are similar between refugees and host populations [5]. Refugees, however, experience an increase in the prevalence of non-communicable diseases, rather than a decrease, as time in the country of arrival increases [6]. In the U.S., one study found that for each additional year post-resettlement, refugees had an estimated 12% increased odds of diabetes mellitus and 7% increased odds of hypertension [6]. The contributing factors to this decline in health status include reductions in healthy food intake and exercise, as well as limited health literacy and healthcare access [6]. Thus, addressing both communicable and non-communicable disease is equally important in this population.

Addressing this burden of disease for refugees can be overwhelming, considering the wide range of health conditions deserving of attention and the heterogeneity of cultures within the group of people cast under the umbrella designation of “refugee.” Additionally, the prevalence of diseases among refugees changes in response to overseas interventions, and the geographic makeup of refugees entering the U.S. each year depending on global events and U.S. policy [7].

There is a potential to improve health outcomes for refugees through systematic interventions; however, the effectiveness of prior interventions is poorly understood. Identifying interventions that consistently improve health outcomes across various settings and populations is important given the diverse cultural factors and health practices of refugees, as well as the broad range of settings where refugees are resettled, and care is received. Importantly, since refugee health interventions are often culturally tailored, it is important to understand whether effective interventions among a specific refugee group are can also be effective in different refugee populations. Moreover, disseminating the results of effective interventions is vital as refugees are resettled in numerous towns, cities, and communities across the country whose respective refugee resettlement organizations, medical centers, and other aid organizations may not be able to communicate with one another easily. Without a critical synthesis of findings, efforts to improve refugee health are sporadic and may be poorly informed.

We conducted a systematic review to identify patient-level healthcare interventions for refugee populations within the U.S. to (1) identify prior effective interventions and (2) understand where gaps exist in the development of interventions to improve the delivery of healthcare for refugees. To our knowledge, this is the first review addressing this specific topic. Our review allows for an understanding of existing health interventions for refugee populations and can inform new intervention development and their dissemination and scalability. Based on this information, researchers, resettlement agencies, policymakers, and other relevant stakeholders will be able to develop more effective interventions to improve health care for refugees within the U.S.

Methods

We conducted a systematic review according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [8] protocol to identify all empirical studies of interventions targeted at improving refugee health outcomes within the United States.

Eligibility Criteria

Studies were included if they met the following criteria: included a population of refugees of any age, included an intervention aimed at improving health, included an evaluation of the intervention’s outcomes, and were conducted in the U.S. Studies were limited to those conducted in the U.S. since healthcare interventions are influenced by the U.S.’s complex multi-payer healthcare system, as well as federal and state policies, existing integration and support services, and funding allocation which are unique from other countries. Studies were excluded if they did not meet inclusion criteria or if the primary outcome was mental health, as several systematic reviews of interventions targeting mental health outcomes have been previously published [9]. Studies were also excluded if the intervention was not evaluated or there were no outcome data; they only evaluated provider and clinic outcomes, only included financial outcomes (cost-effectiveness, etc.), evaluated outcomes from mixed populations (i.e., refugees and immigrants) that did not include a subgroup analysis of the refugee/asylee population, or were case studies.

Search Strategy and Data Sources

Four databases were searched on January 14, 2023: Medline via PubMed, Web of Science, Embase, and CINAHL. Search terms included a combination of keywords and database-specific subject headings for concepts around (a) refugees and (b) access and/or social determinants of health. Searches were limited to studies published in 2000 to search date, written in English, and conducted in the U.S. The full search strategies can be found in Supplement 1.

Citations were exported from each database and de-duplicated in EndNote (EndNote 20, Clarivate, Philadelphia, PA). The resulting 4,788 citations were clustered via DoCTER Topic Extraction (https://www.icf-docter.com/) into major themes in the results. Of these, citations in three clusters, mental health, law, and migration, were exported into Covidence (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia. Available at www.covidence.org.) and pre-screened by a single reviewer (AHD) for exclusion. Citations retained from this screening step were combined with the remaining references for 3,278 total citations, which were again imported into Covidence.

Study Selection

Titles and abstracts were screened by two authors (LB, PC, CN, MO, and NS) using the above-described inclusion and exclusion criteria. If there was disagreement on whether to include a study, it was either brought to the group for discussion until a consensus was reached or screened by a third independent reviewer. Full-text articles were reviewed by a single author (LB, CN, MO, and NS). This was deemed appropriate since all authors were familiar with the inclusion and exclusion criteria after the lengthy title and abstract review process. If a reviewer was unsure whether to include a full-text article, it was again brought to the group for discussion. The source review PRISMA diagram can be found in Fig. 1.

Fig. 1
figure 1

PRISMA review diagram

Full size image

Quality Appraisal

After reviewing full texts for inclusion and exclusion, all included articles were appraised for quality using the Mixed Methods Appraisal Tool (MMAT) [10]. The MMAT was chosen (1) to accommodate the wide range of study designs included in the review and (2) because it has demonstrated content validity [11] and adequate reliability [12]. A single reviewer appraised studies for quality, and if the reviewer was uncertain about the appraisal, the study was brought to the group for discussion. Also, all studies that the initial reviewer felt to have a high risk of bias were brought to the group for discussion. Studies deemed to have a high risk of bias were excluded (n = 8). In accordance with guidance from the authors of the MMAT [13], studies were awarded one point for a “yes” response to any of the five criteria, which differ based on study design, for an overall score of 0–5. Mixed methods studies were awarded the lowest score of the study components.

Data Extraction and Analysis

Data was extracted by a single reviewer (LB, MO, CN) using a custom extraction template in Covidence. Information regarding study design, setting, sample characteristics (age, gender, country of origin, time in the U.S.), study duration, intervention target (i.e., healthcare access, infectious disease, oral health, etc.), intervention description, outcomes, and outcome effectiveness were extracted. Due to the substantial heterogeneity of study design, intervention type, and outcomes sought, we conducted a narrative synthesis of study characteristics and findings. The analysis of studies is organized by intervention modality.

Results

Study Characteristics

Thirty-seven studies (Table 1) were included in this review, encompassing a broad range of study designs. Included were 11 cohort studies, 9 non-randomized experimental studies, 5 qualitative studies, 4 randomized controlled trials, 3 retrospective studies, 2 quality improvement projects, 2 mixed-methods studies, and 1 cross-sectional study. Most studies (n = 30) targeted an adult population; six included children and their parents, and a single study included an older adult population. The sample size also varied widely, from 14 to 3007 participants. Study participants came from 58 different countries, with the most common countries of origin being Bhutan (n = 13), Myanmar/Burma (n = 11), Somalia (n = 9), Iraq (n = 9), Afghanistan (n = 7), Nepal (n = 6), Eritrea (n = 6), Democratic Republic of Congo (n = 6), Syria (n = 6), Sudan (n = 5), Ethiopia (n = 5), Vietnam (n = 4), Burundi (n = 4), Cambodia (n = 4), and Uganda (n = 4).

Table 1 Study characteristics
Full size table

Study Quality

Included studies met at least two of the five quality criteria, with most studies meeting four (n = 12) or five (n = 13) of the quality criteria. Nine studies met three of five quality criteria, and four met two of five. The results of the study quality appraisal can be found in Supplement 2.

Interventions and Outcomes

A broad range of health conditions were targeted via three main intervention modalities: healthcare provision/management, resource provision, and education. Healthcare provision and management interventions encompassed interventions that provided tailored or modified healthcare. Resource provision interventions consisted of those that provided resources outside of healthcare. Several interventions offered both healthcare and resource provision. Education interventions sought to increase knowledge or awareness of a certain health condition or health behavior. Additionally, two main intervention strategies were identified: cultural/linguistic tailoring of intervention materials and the integration of community members or leaders for intervention delivery. A summary of interventions and their associated outcomes can be found in Table 2.

Table 2 Summary of interventions and outcomes
Full size table

Infectious Disease

Seven studies offered healthcare or resource provision to improve infectious disease health outcomes, six of which targeted tuberculosis [14,15,16,17,18,19] and one which targeted hepatitis B virus [20] treatment. Interventions included enhanced physician follow-up after treatment initiation [15], refugee-specific TB clinics [14, 19], in-home care delivery [16, 18], culturally tailored education [16, 20], streamlining of the treatment referral process [17], and transportation assistance [14, 17, 20]. All seven studies measured some form of treatment completion, including the rate of loss to follow-up [15], treatment completion [14, 16, 18], and attendance to care [17, 19, 20], and all studies reported a significant improvement in their respective outcomes.

Women’s Health

Six studies leveraged healthcare or resource provision [21,22,23,24,25,26] and one used education [27] to improve women’s health outcomes. Three studies targeted breast cancer screening uptake [21,22,23], one targeted cervical cancer screening [27], two focused on childbirth interventions [24, 25], and one focused on general women’s health [26]. To improve breast cancer screening (mammography) rates, patient navigator programs were consistently successful at improving screening rates among refugee women [21,22,23]. The single educational intervention targeting cervical cancer screening knowledge resulted in increased knowledge and awareness, but screening rates were not measured [27].

Interventions for childbirth support had mixed effectiveness. A culturally tailored pregnancy support program that included transportation provision had positive results for some outcomes (labor induction, exclusive breastfeeding intentions) but no difference in others (cesarean delivery, gestational age, low birth weight infants) [24]. A student-run birth companion program found no difference in either maternal or newborn outcomes [25].

Finally, an intervention providing transportation through a clinic-assisted rideshare application to gynecological appointments found participants who used the rideshare had a significantly lower rate of no-shows than those who did not [26].

General Health

Four studies implemented healthcare provision interventions to improve general health outcomes. Interventions included a patient-centered medical home (PCMH) [28], a patient-navigator program [29], telehealth during the COVID-19 pandemic [30], and a home healthcare program for chronic disease management [31]. All found improvements in at least some of their outcomes of interest which included a decrease in emergency room visits [28], rates of no-shows and language barriers [29], rates of telemedicine use [30], and medication management [31].

Two quantitative education interventions targeted general health through either a small group health promotion program [32] or learning circles [33] and reported improvements in general health [32], energy levels [32], quality of life [33], and access to resources [33]. Three qualitative studies reported on general health themes after education interventions [34,35,36]. Two health promotion programs resulted in participants reporting that they could share health knowledge with their community [34] and identify tools for health promotion [35]. The final qualitative study assessed a storytelling intervention for enhancing farming safety and found that evaluating effectiveness was challenging because participants avoided answering questions directly and continued to endorse unsafe practices [36].

Diet and/or Exercise

Among the five studies targeting diet and exercise, four integrated community members or leaders [34, 37,38,39], and one did not [32], in various education interventions. The studies that targeted an increase in education or knowledge all reported positive outcomes [34, 37,38,39]. The studies that measured behavior outcomes reported mixed results. A health promotion program found improvements in exercise among participants [32], while an education program targeting a decrease in rice consumption found no change after the intervention [39]. Finally, a family cooking intervention found that refugee participants reported an increase in cooking and eating together among both children and parents [38].

Health Literacy

The three studies targeting health literacy interventions all reported on knowledge outcomes [32, 40, 41]. Two studies sought to improve pharmacy navigation, and both reported an increase in knowledge and awareness of how to use pharmacy services [40, 41]. Another study sought to increase confidence in understanding one’s illness and discussing concerns with a doctor and found that significantly fewer participants reported no confidence in performing these tasks after a health promotion intervention [32].

Oral Health

Three studies provided an oral health education intervention [42,43,44]. Two studies found no significant change in knowledge [42, 43] or oral health-related quality of life [42]. The remaining study found a significant improvement in oral health literacy scores among participants who received an oral health empowerment intervention [44].

Diabetes

Two education interventions targeted diabetes management using different education modalities, and both studies included hemoglobin A1C (A1C) as an outcome of interest. A group visit program at a community health center [45] and a video education intervention [46] both found no decrease in A1C. One diabetes intervention leveraged a pharmacist-run clinic, which resulted in most participating refugees had medication changes made, and the average A1C decreased from 10.1 to 8.3% [47].

Family Health

Only one educational intervention targeting family health was identified. This study implemented a training program to provide refugee parents with tools and information about parenting in the U.S., and reported that participants experienced increased family communication, satisfaction, and self-efficacy [48].

Substance Use

Two educational interventions targeted substance use; betel nut use [49] and tobacco use [50]. First, educational brochure administration regarding betel nut use resulted in more refugees endorsing betel nut as harmful and cancer-causing, but the study did not report on betel nut use behaviors [49]. A tobacco cessation program resulted in refugee cessation rates that were higher than rates among non-Hispanic White participants in the program at all time points [50].

Discussion

This review identifies interventional studies targeting health outcomes among refugees in the U.S. While the group of studies is relatively small, the types of interventions and reported outcomes are remarkably varied. Interventions fell into two broad categories, education and healthcare or resource provision. Four of the 37 studies reported that the interventions were ineffective, while five reported partial effectiveness, meaning the intervention improved some of the outcomes but not others. In one study, there was no comparison group to measure effectiveness. Interventions were effective in the remaining 27 studies for all reported outcomes.

Interventions fell into three broad categories: healthcare provision/management, resource provision, and education, with quite a bit of diversity within those categories. Within education, for example, offerings included a written handout [49], storytelling [36], and experiential coursework [35], among others. The interventions around resource provision and healthcare provision/management were also diverse but had some shared elements. These included transportation [14, 20, 24, 26], care at home [16, 31] and care coordination or navigation [15, 17, 20,21,22,23, 28, 29].

Considering the wide range of health problems that refugee community members may encounter, it can be difficult to select and design interventions that are a priority for the population of interest. For this reason, community members should be involved early in the process of selecting the health problem of greatest importance and designing the accompanying intervention. Community members provide meaningful insights into local health issues and their involvement ensures that research questions are relevant and sensitive to the community’s needs and priorities. While several studies in this review involved community members in the implementation of their intervention, only a few [26, 29, 36, 44, 48] describe if and how community members were involved in the selection and design of the intervention. Future research should make every effort to include members of it’s target population in every stage of study design from conception to execution. This ensures that interventions are first and foremost meaningful to the target population. Additionally, this approach is likely to enhance recruitment and retention efforts, and increase effectiveness of the intervention.

Among this diverse group of interventions, an extensive range of outcomes were reported, which makes synthesizing them infeasible. Most reported outcomes were process measures such as attendance rate, knowledge, or self-reported health outcomes instead of tangible health or behavior outcomes. For example, Shi et al. offered an educational brochure about betel nut use, and while they report that more people recognized betel nuts as cancer-causing, this study provides no insight into whether betel nut use patterns changed [49]. Conversely, Higgins et al. reported on A1C values before and after a diabetes pharmacy intervention [47], which allows for a tangible understanding of short-term health outcomes, though the long-term sustainability of this change is unclear. Although the reasons for not reporting health or behavior outcomes were often not explained in the reports papers, potential reasons could be methodological challenges (i.e., measuring is impractical or infeasible) or those health / behavioral outcomes could not be observed within the study period. To achieve more robust and meaningful outcome data in relation to interventions, researchers can employ design strategies such as extending study periods, integrating standard-of-care clinical data, or utilizing portable technology, such as wearable sensors or smart home assistants that measure the behaviors of interest. These strategies may allow for the collection of outcomes that reflect participants’ health status and may also help assess the sustainability of interventions through long-term, longitudinal data collection.

Along with a wide range of study outcomes, studies reported using various measurement tools to capture outcomes. This was needed in part because refugees are a highly heterogeneous group of people who speak a wide range of languages and possess different levels of literacy and numeracy. While all studies attempted to meet the language and literacy needs of their population, this was not always possible. For example, Goldsmith et al. did not have formal interpreters available for their pharmacy education intervention, so most participants used “peer interpreters” [40]. These interpreters offered over-the-shoulder support during survey administration, presenting a serious risk of bias in their responses. Prescott et al. also administered a paper survey to a group with varying levels of literacy where some people completed it independently, and others used an interpreter to read answers to participants, potentially biasing the results [41].

For studies that did use translated measures, the quality of the translation varied. The COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) offer a rigorous checklist for assessing the quality of a measure, including the translation methodology [51]. Key steps include forward translation by native speakers of the target language with knowledge of the disease state and backward translation by native speakers of the original language who are naive to the disease state. While most studies described their translation process, almost none described it in adequate detail to assess if these conditions were met, and no studies cited the COSMIN checklist. It is important to consider that measures validated in English must also undergo validation and reliability testing in their translated language and target population before they can be regarded as valid and reliable. Encouragingly, some studies acknowledged their lack of validity testing due to understandable feasibility limitations [48] and recognized it as an important area of future research.

It is challenging to evaluate which components contributed to the outcomes for multi-component interventions, such as those that both modified healthcare provision and offered transportation assistance. In resource-limited settings, where offering several interventions may not be feasible, it is essential to understand which components of an intervention are most impactful. Again, while some studies acknowledge their inability to disaggregate the results by intervention modality [32], no studies have implemented a dismantling study. This type of work is important for dissemination and identifying opportunities to refine further for more effective interventions.

The included studies heavily focus on health care for adults, while children’s health was largely ignored. Only one study involved children [38], and one focused on parents [48]. Refugee children face unique challenges in terms of health conditions, healthcare utilization, and health management at home. In comparison to the general child population, they face higher rates of anemia, elevated lead blood levels, growth abnormalities, and poor oral health [52,53,54,55,56,57,58]. Considering that approximately a third of refugees entering the U.S. are children under 14 years [59], the work regarding refugee children is disproportionately low and deserves more attention.

Most healthcare interventions identified in this review were delivered in person and few leveraged digital technologies. Only one study [30] leveraged a remote intervention in the form of telemedicine during the COVID-19 pandemic. Digital health interventions can potentially support further engagement in patients’ health management and shared decision-making. Several systematic reviews find that interventions leveraging remote health monitoring and personnel support effectively improve disease management, such as hypertension management [60] and heart failure [61], in underserved populations. Remote monitoring, treatment, and support may be worth exploring among refugee populations, and must consider access to technology, especially smartphones, the technological literacy and the language needs of the target population. Also, several studies included in this review leveraged community members for care delivery, which was recognized as valuable to participants in both qualitative and quantitative studies. Digital health interventions may not be able to provide the same level of relatedness and effectiveness, which is an additional consideration for refugee populations.

Conclusion

Refugees in the U.S. face a broad range of health challenges and are forced to navigate a complex, costly healthcare system to receive care. This review demonstrates that while some health conditions, namely tuberculosis, have been addressed with large-scale, sustained interventions, other conditions, namely general health and women’s health, have been addressed through piecemeal, short-term interventions. The evaluation of interventions often focuses on knowledge or satisfaction rather than health or behavior change outcomes. Further work is needed to understand the best strategies for developing and maintaining sustainable interventions that meet the needs of the diverse population of refugees entering the U.S. each year. The success of these interventions must be rigorously measured to capture outcomes most directly associated with improved health.