This study used a descriptive, cross-sectional design to explore the prevalence and predictors of food insecurity among chronically ill Romani adults in Jordan. Data were collected from four governorates—Amman, Irbid, Madaba, and Al-Mafraq—chosen because they host sizable Romani populations and capture different urban–rural contexts. Amman, the capital, is a highly urbanized city with advanced infrastructure and wide access to services. Madaba, by contrast, is a smaller urban center, while Irbid, the country’s second-largest city, combines both urban and peri-urban features. Al-Mafraq, in turn, is more remote and underserved compared to other regions, with large rural areas and limited service availability. Together, these regions provide a diverse backdrop of socioeconomic and healthcare conditions, allowing us to examine how geography shapes disparities in food security.

The target population consisted of self-identified Romani adults (≥18 years) residing in the selected regions with a confirmed or self-reported diagnosis of at least one chronic illness lasting 1 year or more, requiring ongoing medical care or limiting daily activities, such as hypertension, type 2 diabetes, cardiovascular diseases, cerebrovascular diseases, chronic respiratory conditions, or neurological disorders. Exclusion criteria included severe cognitive or psychiatric impairments (e.g., schizophrenia, dementia) that could interfere with informed consent or the reliability of survey responses. Exclusions were based on self-reported physician diagnoses and cognitive status assessed by trained researchers using the 6-Item Cognitive Impairment Test (6CIT) during recruitment. This validated brief tool evaluates orientation, attention, and short-term memory, with a maximum score of 28. Participants scoring ≥10 were excluded based on established thresholds indicating moderate-to-severe cognitive impairment (18).

Given the absence of an official Romani registry, convenience sampling was utilized. Community engagement strategies, including collaboration with recognized Romani leaders, facilitated participant recruitment and access to settlements. Each district leader assisted in identifying eligible participants and in assigning interpreters or facilitators as needed. The overrepresentation of Amman-based participants (53.9%) mirrors the spatial distribution of Jordan’s Romani population, who predominantly cluster in the capital’s informal settlements due to socioeconomic push-pull factors (7). While this limits granular geographic comparisons, our analysis accounts for regional variation via stratified models and confirms robustness through sensitivity testing (see Results).

A priori sample size estimation was conducted using G*Power 3.1 software. Based on a binary logistic regression model, with an alpha level of 0.05, power of 80%, and an anticipated odds ratio (OR) of 1.5, the minimum required sample size was calculated to be 308 participants. To accommodate potential non-responses or missing data, a 10% buffer was added, resulting in a target of 350 participants. After excluding three incomplete surveys, the final analytical sample comprised 347 individuals.

Data were collected between January and February 2021 via structured, face-to-face interviews administered by the principal investigator and a trained nursing graduate student. Interviews were conducted in private settings—typically within participants’ homes or tents—to ensure confidentiality and comfort. A culturally adapted Arabic-language questionnaire was used, developed based on validated tools and pretested among 50 Romani individuals to ensure clarity, reliability, and cultural appropriateness.

Before participation, voluntary verbal informed consent was obtained, in accordance with ethical protocols for populations with high illiteracy. No identifying information was collected. Interviews lasted approximately 15–20 min and were conducted with interpreter support as needed.

All instruments were translated and culturally adapted into Arabic following the World Health Organization’s recommended protocol for adapting and translating research instruments (21). The process involved forward translation by bilingual experts, expert panel review, back-translation, and cognitive pretesting with the target population to ensure conceptual and functional equivalence.

The study received ethical clearance from the Research Ethics Committee affiliated with the University of Jordan. The study complied with the principles of the Declaration of Helsinki and national regulations. Given the high illiteracy rate, verbal informed consent was obtained with comprehensive oral explanations of the study’s purpose, risks, and participant rights. Privacy, anonymity, and confidentiality were strictly maintained throughout the study.

All data were processed using IBM SPSS Statistics version 26. Summary statistics—including proportions, averages, and standard deviations—were calculated to describe participant profiles. Relationships between food insecurity and key variables were initially examined through chi-square tests and independent t-tests. Binary logistic regression analysis was conducted to identify independent predictors of food insecurity. Variables with p < 0.10 in bivariate analyses were entered into the multivariable model. Odds ratios (ORs) with 95% confidence intervals (CIs) were reported. Multicollinearity was assessed using variance inflation factor (VIF), and model fit was evaluated using the Hosmer–Lemeshow test. A two-tailed p-value of <0.05 was considered statistically significant.

A total of 347 Romani adults with chronic illnesses participated in the study. The sample was nearly gender-balanced, with 52.4% men and 47.6% women participants. The mean age was 48.2 years (SD = 13.5), with the majority (59.7%) falling within the 41–60 age group. The majority of individuals reported being currently married (89.1%), with over half residing in the capital city, Amman (53.9%).

Household sizes were generally large, with an average of 6.8 members (SD = 3.4); 30% reported living in households with more than eight individuals. Educational attainment was strikingly low, with 99.7% reporting illiteracy. Unemployment was high, with only 18% reporting current employment (Table 1).

The most frequently reported primary chronic illnesses were type 2 diabetes (38.4%) and elevated blood pressure (36.8%), followed by stroke-related disorders (7.4%) and neurological impairments (6.0%). Smaller proportions reported heart disease (3.9%), lung disease (3.0%), or other chronic conditions (4.6%). On average, participants had been living with their chronic condition for 5.1 years (SD = 4.6), and 21.6% had lived with it for more than 6 years. Over half (55.0%) lacked health insurance coverage. Most participants were non-smokers (64.0%) and reported no alcohol consumption (98.8%). Regarding self-perceived health, 14.1% rated their general health as poor or fair, and 38.3% rated their mental health as poor or fair. In the past 12 months, 75.2% reported utilizing healthcare services, yet 81.6% experienced unmet healthcare needs, often due to cost, availability, or other access barriers. Perceived access to healthcare was generally poor, with 81.0% reporting difficulty obtaining needed services. Most participants (63.5%) were living with two or more chronic conditions, underscoring the complex health needs of this population (Table 2).

As shown in Figure 1, 90.5% of participants experienced some level of food insecurity. Specifically, 75.6% reported low food security, while an additional 14.9% reported very low food security. Only 5.7% were identified as food secure, and 3.7% as marginally food secure. This indicates that nearly all participants lived with inadequate access to nutritious and sufficient food, with a substantial proportion experiencing severe deprivation.

Results from the binary logistic regression analysis are presented in Table 3. Several factors emerged as significant predictors of food insecurity.

Other variables — including sex, age, employment status, mental health, health insurance, healthcare utilization, household size, smoking status chronic disease duration and unmet healthcare needs — were not statistically significantly associated with food insecurity in the bivariate logistic regression analyses. The overall logistic model was statistically significant overall [χ² (19, N = 341) = 136.459, p < 0.001] and explained 46.1% of the variance in food insecurity status (Nagelkerke R²). The model correctly classified 79.1% of cases, indicating strong predictive accuracy.

This study provides compelling evidence of the disproportionate burden of food insecurity among chronically ill Romani adults in Jordan, revealing a crisis that far exceeds national norms. While food insecurity in the general Jordanian population has been estimated between 13.9 and 23.1% during the COVID-19 pandemic (24, 25), the rate among our study participants (75.6% low food security and 14.9% very low) represents a stark disparity that demands urgent public health attention.

These findings are consistent with global literature documenting the heightened vulnerability of marginalized ethnic communities, including Roma populations, to food insecurity (5, 26, 27). These findings align with studies from across Europe, which reveal that Roma communities often face much higher levels of food insecurity and poorer diets. This stems from a mix of economic hardship and systemic barriers, as seen in countries like Spain, Romania, and Hungary, as well as in broader EU research (27–31).

These findings are consistent with the well-documented bi-directional relationship between chronic illness and food insecurity, whereby poor dietary intake and instability in food access can exacerbate disease progression, while illness itself can diminish income-generating capacity and strain household resources (10, 32, 33).

Geographic disparities also emerged as a key social determinant. Participants living in Amman and Madaba (urban areas with relatively better infrastructure and access to services) had significantly lower odds of food insecurity compared to those in Al-Mafraq, a more remote and underserved region. These findings mirror patterns observed in other LMICs, where rural and peri-urban residents face heightened vulnerability due to geographic isolation, fragile employment markets, and weak social protection networks (34).

Marital status further influenced outcomes. Divorced or widowed participants were more likely to experience food insecurity, likely due to diminished household income, caregiving burdens, and loss of social support. This trend aligns with evidence that female-headed and single-parent households are particularly susceptible to food deprivation across both high-income and low-income settings (35, 36).

It is important to view these findings through the lens of intersectional structural disadvantage. The Romani community in Jordan faces multi-layered barriers, including legal invisibility, discrimination, informal employment, and stigma factors that collectively undermine household stability and food security. Although gender, age, and household size did not show significant associations in our model, this may reflect sample homogeneity and unmeasured protective social dynamics, such as intergenerational caregiving and food-sharing practices within extended families.

Further, although detailed income data could not be reliably collected due to the informal and undocumented nature of economic activity in Romani communities, proxy indicators such as employment status and education provide insight into economic vulnerability. Nonetheless, we acknowledge this as a limitation and highlight the need for culturally sensitive tools to better capture economic precarity in marginalized populations.

Together, these results suggest that improving food security for Romani populations with chronic illness will require multi-sectoral approaches that address both immediate nutritional needs and the broader social, health, and economic exclusion they face.

Illiteracy was prevalent among participants 99.7%, far exceeding Jordan’s national illiteracy rate of 4.5% (37). Illiteracy restricts access to healthcare information and effective chronic disease management, compounding food insecurity. These findings align with global research showing educational barriers among Roma communities due to poverty and systemic exclusion (7, 38).

Employment challenges were equally pronounced, with 82% of working-age participants unemployed compared to Jordan’s national unemployment rate of 21.4% (39). Roma employment is often informal and precarious, including activities like scrap collection and street vending. Logistic regression revealed no significant correlation between employment and food security (p = 0.413) which may reflect not only the unstable nature of Roma employment but also the buffering effect of extended kinship structures, where shared caregiving, pooled informal income, and collective resource management help mitigate food insecurity risks despite individual unemployment (40). Addressing these disparities requires vocational training, literacy programs, and anti-discrimination policies to create pathways to stable employment.

Difficulty in accessing healthcare services was another key predictor. Participants reporting poor access to outpatient care had significantly higher odds of food insecurity, reinforcing the interconnectedness of healthcare and nutrition. Healthcare barriers—including lack of insurance, transportation challenges, and cost—may divert resources away from food expenditures or delay disease management, exacerbating both malnutrition and morbidity (41). Integrated health and nutrition support services, particularly in underserved communities, are critical for breaking this cycle.

The findings of this study have direct implications for public health planning in Jordan and similar LMICs. First, routine screening for food insecurity should be integrated into chronic disease management, particularly in clinics serving vulnerable populations. Second, culturally tailored interventions—such as mobile food pantries, targeted cash transfers, or community kitchens—should be implemented in Romani settlements, taking into account linguistic, cultural, and geographic barriers. Third, broader structural reforms are necessary to tackle the underlying causes of food insecurity among ethnic minorities. These include expanding access to universal health coverage, improving infrastructure in informal settlements, and providing vocational training programs aimed at improving employment opportunities. Intersectoral collaboration—linking health, nutrition, education, and social protection services—is essential for achieving sustainable progress.

The relationship between food insecurity and chronic illness observed in this study is best understood through the lens of Syndemic Theory, which conceptualizes the clustering and mutual reinforcement of disease and social disadvantage in structurally vulnerable populations (42). Rather than existing as isolated challenges, food insecurity and chronic disease co-occur in a dynamic interaction shaped by broader structural forces—such as poverty, marginalization, and limited access to care. This framework is particularly applicable to Romani communities in Jordan, who experience a unique intersection of ethnic discrimination, legal invisibility, informal labor conditions, and exclusion from health and social services. Chronic illness reduces individuals’ ability to work and increases out-of-pocket health expenditures, further deepening food insecurity. Conversely, ongoing food deprivation can lead to poor nutrition, compromised immunity, and disease progression, reinforcing a syndemic loop.

Additionally, the Social Determinants of Health (SDH) framework emphasizes the role of upstream societal and policy-level factors—such as geography, marital status, and education—in shaping individual health and nutritional outcomes (43). For instance, the heightened food insecurity observed among participants living in Al-Mafraq or those who were widowed/divorced reflects structural barriers beyond personal choice. Integrating these frameworks provides a more explanatory understanding of how food insecurity among chronically ill Romani adults is not only a reflection of individual health status but also of the social systems in which they are embedded.

While prior research has established links between variables such as health insurance coverage and household size with food insecurity risk, our analysis found no statistically significant associations between these factors and food insecurity status among Romani adults with chronic illness. This divergence warrants careful contextual and methodological interpretation.

One possible explanation lies in the cultural and communal dynamics of the Romani population. Unlike nuclear family models often assumed in food insecurity research, Romani households frequently exhibit extended kinship structures characterized by shared caregiving responsibilities, pooled informal income, and collective resource management (27, 44). In such contexts, larger household size may not correspond to increased economic burden. Instead, it may provide resilience against food insecurity through mutual support and resource-sharing practices, such as communal cooking and joint financial decision-making. These patterns have been observed in Roma communities across Eastern Europe and the Middle East, where intra-family networks mitigate the effects of economic hardship in the absence of formal social protections (27).

Similarly, the lack of association between health insurance and food security could reflect discrepancies between coverage and functional access. While approximately half the participants reported having some form of insurance, informal probing during data collection revealed that many were either unaware of how to use their benefits or enrolled in schemes offering only minimal coverage (e.g., emergency services only). These findings suggest that possession of health insurance in marginalized populations does not necessarily equate to improved access or financial protection, thereby limiting its expected buffering effect on food security. Similar mismatches between perceived and actual insurance utility have been reported in other disadvantaged and under-resourced communities (10, 45).

From a methodological standpoint, it is also important to consider the relative homogeneity of our sample. All participants were Romani adults with chronic illness, a subgroup already characterized by overlapping vulnerabilities. This may have minimized between-group variability and obscured associations that could be more detectable in broader or more diverse samples (a phenomenon well documented in prior studies focused on narrowly defined vulnerable populations) (27). Moreover, while our sample was adequate for estimating prevalence, it may have lacked sufficient statistical power to detect small to moderate effect sizes in regression models—particularly for variables with complex or context-dependent influence.

Together, these findings emphasize the importance of culturally informed analysis and the risks of overgeneralizing from models developed in different sociocultural or economic contexts. Future research should aim to integrate ethnographic insights, functional measures of social protection, and culturally adapted indicators to more accurately capture food insecurity risks in Romani and other marginalized populations.

This study offers valuable insights into an under-researched population using validated tools and rigorous statistical methods. The involvement of community leaders and interpreters enhanced participant engagement and data validity. However, several limitations must be acknowledged. The use of a convenience sample limits generalizability beyond the study sites. Self-reported data may be subject to recall and social desirability bias. Furthermore, the cross-sectional design precludes causal inference. Future research should consider longitudinal or mixed-method approaches to explore temporal relationships and lived experiences of food insecurity. Moreover, limitations related to economic measurement necessitate future research that incorporate validated, culturally sensitive tools to assess economic status and livelihood insecurity, particularly in populations with informal or undocumented income sources.

Finally, although covering four governorates enhances geographic representation, the large differences between these regions may introduce variation that is difficult to isolate in the analysis, especially in the absence of a cluster design or representative sampling, therefore we recommend that future research employ stratified or multistage sampling to address this issue more systematically.

Further studies are warranted to investigate the long-term health and nutritional consequences of food insecurity among Romani and other marginalized communities. Qualitative research can provide deeper insights into cultural norms, coping strategies, and barriers to food access. Comparative studies involving other ethnic minorities or refugee populations in the region could broaden the understanding of intersectional vulnerabilities. Evaluations of pilot interventions—such as integrated health-nutrition clinics or mobile aid services—would also be valuable in guiding policy.