1 Background

The four main dimensions of food security encompass food availability, economic and physical access to food, food utilization, and stability of the aforementioned over time [1]. Disruption to any of these dimensions can lead to food insecurity, “a household-level economic and social condition of limited or uncertain access to adequate food,” which has significant implications for health and wellbeing [2, 3]. In the U.S., food insecurity disproportionately affects individuals aged 18–34, females, low-income individuals, and those identifying as non-Hispanic Black [4, 5]. Individuals experiencing food insecurity often face other social determinants of health (SDOH) challenges, such as a lack of health insurance and stable housing or the ability to afford medications and dental care [6,7,8]. Food insecurity is associated with a myriad of health issues, including cognitive dysfunction, mental health problems, obesity, diabetes, anemia, asthma, hypertension, chronic kidney disease, and hyperlipidemia [3, 9,10,11]. The COVID-19 pandemic has also exacerbated the issue for certain communities, with 10.5% of households experiencing food insecurity at some point during 2020 [12,13,14].

In spite of programs like the Supplemental Nutrition Assistance Program (SNAP), designed to combat food insecurity and reduce estimated annual healthcare costs, many households still struggle to access nutritious food options [15, 16]. The prevalence of food insecurity and its associations vary considerably based on community and population characteristics, complicating efforts to address the issue [17, 18]. This variability underscores the importance of local needs assessments and food security programs to cater to each community’s unique needs [19]. To date, there has been minimal research into social or health-related risk factors that may be associated with food insecurity among uninsured individuals [20, 21], with even fewer studies focusing on this population in Wisconsin.

In this study, we describe a 6-month needs assessment conducted at the Saturday Clinic for the Uninsured, the student-run free clinic associated with the Medical College of Wisconsin. Supported by a community hospital and staffed by volunteer physicians, pharmacists, and medical and pharmacy students, this clinic exclusively serves patients without health insurance, providing primary and specialty care, including mental health care, and social work services free of charge.

While many studies describe food insecurity associations between one to two SDOH needs or health outcomes [22, 23], we conducted a more comprehensive analysis of how food insecurity may predict association with nine other SDOH needs, including utility assistance, legal concerns, and substance use resources, not previously studied. Additionally, we explored food insecurity associations with eight chronic medical conditions and whether COVID-19 affected patient responses to the SDOH needs assessment. By understanding these diverse associations with food insecurity, we can not only inform the future development of an evidence-based, accessible food assistance program at Saturday Clinic for the Uninsured, but also gain valuable insights into the uninsured population.

2 Methods

2.1 Participants

The study population consists of patients who completed the SDOH survey during their in-person or telehealth visit at Saturday Clinic for the Uninsured in Milwaukee, WI between October 2021 and April 2022. The inclusion and exclusion criteria were based on the patients seen at the clinic, which exclusively serves adult patients (≥ 18 years of age) of all genders, ethnicities, and languages, without health insurance. Clinic volunteers verbally administered the SDOH survey during appointments after obtaining patient consent.

2.2 Data collection & measures

The SDOH survey questions were developed by the research team, clinic social worker, and medical director, based upon previously conducted research [24, 25], observed patient needs, and clinic processes. Participants responded Yes/No when asked if they: had ever skipped medications to save money, were worried about not having stable housing, needed assistance paying utility bills, had questions about legal issues related to immigration, child support, or eviction, were interested in learning about educational or work opportunities, would like resources for substance use, were interested in mental health services, would like information regarding health insurance options, and would like free or low-cost dental health. We refer to these SDOH needs through the following labels: medications, housing, utilities, legal concerns, education or work opportunities, substance abuse resources, mental health resources, health insurance, and dental care.

Food insecurity was assessed using the USDA “Household Food Security Survey Module: Six-Item Short Form,” a reliable and validated scale [26]. The USDA 6-item short form notes raw scores of 0–1 as high or marginal food security, scores of 2–4 as low food security, and scores of 5–6 as very low food security [26]. Regardless of score, we asked all participants if they would like food-related resources, which consisted of zip code-specific food pantries and services, information on Impact 2-1-1 (a local social services helpline), separate virtual visits to facilitate SNAP or Medicaid (BadgerCare) applications, health and diet-related fliers, and other tailored programs depending on eligibility.

We collected participant self-reported demographics, including age (date of birth), sex assigned at birth (male, female, unknown), race (American Indian/Alaskan Native, Asian, Black or African American, Native Hawaiian or other Pacific Islander, White, Unknown, Decline to Specify, and Other Race), Hispanic or Latino ethnicity (Yes, No, or Decline), and zip code (inferred county). We also asked participants if the COVID-19 pandemic affected any of their responses by reviewing the list of SDOH needs assessed.

A select group of medical student volunteers administered the SDOH survey as part of clinic procedures after patients provided verbal consent. These medical students were trained on verbal survey administration, data collection into a REDCap database [27], and documentation into the Electronic Health Record. Additionally, volunteers were trained on selection and distribution of resources based on patients’ social contexts. Clinic volunteers routinely assessed SDOH needs unless the patient completed the survey in the previous 4 weeks. Patients usually had follow-up clinic visits every three months and during these subsequent visits, if they completed the SDOH survey again, we collected information on the usefulness of resources they were previously provided. When necessary for communication with non-English speaking patients, clinic volunteers used certified medical interpreters via a telephone service.

The research team conducted chart review on participants that completed the SDOH survey to determine whether they had the following conditions based on diagnoses noted in the medical record, most recent lab values (if available in the last 6 months), and whether participants were on active medications for these conditions. The chart review was conducted at least one month after initial completion of the survey to ensure that new lab results were included in our data analysis. Lab records older than 6 months were excluded. We sought to standardize and objectively track these chronic medical conditions using multiple sources as EMR data can sometimes contain old or missing diagnoses [28]. The specific criteria for each chronic medical condition include:

2.2.1 Obesity

Diagnostic inclusion criteria included a clinical diagnosis and/or body mass index (BMI) ≥ 30 kg/m2 [29]. We calculated BMI using participant height and weight from the most recent clinic visit.

2.2.2 Hypertension

Diagnostic inclusion criteria included any 1 or more of the following: a clinical diagnosis, an active prescription for anti-hypertensive medications, and/or meeting American Heart Association criteria for hypertension (elevated systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg across 2 office visits) [30].

2.2.3 Type 2 Diabetes

Diagnostic inclusion criteria included a clinical diagnosis, elevated hemoglobin A1c (HbA1c) values ≥ 6.5% (Normal < 5.7%, Prediabetes 5.7% to 6.4%) [31], or an active prescription for a diabetes medication. We did not consider fasting blood glucose values as these are not routinely obtained at the study clinic.

2.2.4 Dyslipidemia

Diagnostic inclusion criteria included a clinical diagnosis, abnormal lipid panel values (total cholesterol > 200 mg/dL, HDL < 40 mg/dL, LDL cholesterol > 130 mg/dL, and triglycerides > 150 mg/dL) [32], and/or active prescriptions for cholesterol or lipid lowering medications.

2.2.5 Chronic kidney disease (CKD)

Diagnostic inclusion criteria included a clinical diagnosis or abnormal values for creatinine (> 1.2 mg/dL) and eGFR (< 60 ml/min) [33].

2.2.6 Gastroesophageal reflux disease (GERD)

Diagnostic inclusion criteria included a recorded diagnosis or GERD-specific medication prescriptions. GERD is a clinical diagnosis.

2.2.7 Anxiety

Diagnostic inclusion criteria included a clinical diagnosis, abnormal Generalized Anxiety Disorder 7-item (GAD-7) scores (score ≥ 9, range 0–21), and/or active prescriptions for anxiolytic medication prescriptions. The GAD-7 questionnaire is widely used to screen for anxiety and has strong internal reliability and external validity [34]. GAD-7 scores range from 0–21 with scores of 0–4 signifying minimal anxiety, 5–9: mild anxiety, 10–14: moderate anxiety, and 15–21: severe anxiety.

2.2.8 Depression

Diagnostic inclusion criteria included a clinical diagnosis, abnormal Patient Health Questionnaire-9 (PHQ-9) scores (score ≥ 9, range 0–27), or anti-depressant medication prescriptions. The PHQ-9 is a widely used questionnaire to screen for depression and has strong internal reliability and external validity [35]. PHQ-9 scores range from 0–27 with scores of 1–4 signifying minimal depression, 5–9: mild depression, 10–14: moderate depression, 15–19: moderately severe depression, and 20–27: severe depression.

2.3 Analysis

To enhance statistical power, we grouped participants with scores of 0–1 as food secure and scores of 2–6 as food insecure like previous studies [19, 36]. We conducted statistical analysis using R 4.1.2. (R Foundation for Statistical Computing, Vienna, Austria) and the gtsummary [37] and MASS packages [38]. After excluding incomplete SDOH surveys, we calculated odds ratios using binomial multivariable logistic regression, with food security status as the independent variable and each SDOH need (Table 2) or chronic medical condition (Table 3) as the dependent variable and adjusting for participant age and sex. A separate model was created for each SDOH need or chronic medical condition and tested for significance (P ≤ 0.05).

Another model of ordered logistic regression was conducted by stratifying participants’ total number of health conditions into 0, 1, 2–3, and 4 + groups with food security status as the independent variable and adjusting for participant sex. To account for the known increase in health conditions with age, we stratified the analysis by age groups: "Under 50" and "50 and Over." Like the other models, significance was set to P ≤ 0.05.

3 Results

Of the 190 unique patients seen at the clinic from October 2021 to April 2022, 157 (82.63%) patients were administered the SDOH survey. We included 135 (71.1%) participants who responded to the complete SDOH survey in our analyses. Most participants completed the survey in-person (n = 132, 97.8%), fell between 50–59 years of age (mean age = 48.44 years ± 14.99 years), were female (n = 86, 64%), and resided in Milwaukee County (n = 124, 91.9%) (Table 1). The race of most participants was Black or African American (n = 55, 40.7%) and 36 participants (26.7%) were Hispanic or Latino (Table 1). Among SDOH needs and chronic medical conditions, most participants wanted resources for dental care (n = 81, 60%) and were diagnosed with hypertension (n = 71, 52.6%) (Table 1).

Table 1 Demographic Characteristics and Screening Outcomes of Patients from October 2021-April 2022 (n = 135)
Full size table

Among those experiencing food insecurity (n = 22, 16.3%), most came from Milwaukee County (n = 21, 95.5%) (Table 1). At the end of the SDOH survey, participants were asked if the COVID-19 pandemic affected their responses, and 24 participants (17.8%) responded “Yes” with most (n = 9, 37.5%) specifying a change in food security. Participants were asked if they would like food-related resources regardless of their food insecurity score and 36 participants (26.7%) requested some resources. Based on social context and interest, 30 (22.2%) participants received information on food pantries and 16 (11.9%) on SNAP (some received both resources). When participants visited the clinic again during this study period (~ 3 months later) and were asked about the usefulness of the resources they were provided, 7 (58.3%) out of 12 respondents found SNAP useful and 12 (66.7%) out of 18 respondents found food pantries useful.

We found that in comparison to participants who were food secure, those experiencing food insecurity had greater odds of needing support for medications (adjusted odds ratio [AOR] = 7.28; 95% Confidence Interval [95% CI] = 2.33–23.2); p-value [P] =  < 0.001, housing (AOR = 9.99; 95% CI = 2.29–48.7; P = 0.002), utilities (AOR = 3.94; 95% CI = 1.07–13.5; P = 0.03), mental health resources (AOR = 4.54; 95% CI = 1.66–12.5; P = 0.003), health insurance (AOR = 2.86; 95% CI = 1.09–8.22; P = 0.04), and dental care (AOR = 3.65; 95% CI = 1.26–13.3; P = 0.03). (Table 2). We identified that food security status did not increase the odds of seeking resources for legal concerns, education or work opportunities, or substance use.

Table 2 Association of SDOH Needs for Participants Without Health Insurance and Experiencing Food Insecurity (n = 135)
Full size table

Participants experiencing food insecurity had higher odds of having anxiety (AOR = 3.26; 95% CI = 1.23–8.38; P = 0.02) and depression (AOR = 2.88; 95% CI = 1.01–7.80; P = 0.04) diagnoses or symptoms but not diagnoses of obesity, hypertension, diabetes, dyslipidemia, CKD, or GERD (Table 3). After stratifying participants by age into two groups, 'Under 50' and '50 and Over', we explored the association between food insecurity, sex, and the categorized number of health conditions (0, 1, 2–3, 4 +). Among individuals under 50 years of age, food insecurity significantly increased the likelihood of having a higher number of health conditions (AOR = 6.35; 95% CI = 1.49–26.99; P = 0.01), indicating a substantial impact of food insecurity on health outcomes in this younger cohort. Conversely, for participants aged 50 and over, food insecurity did not significantly influence the number of health conditions, suggesting the effects of food insecurity on health conditions may diminish with age or be overshadowed by other factors in older populations.

Table 3 Association of Chronic Medical Conditions for Participants Without Health Insurance and Experiencing Food Insecurity (n = 135)
Full size table

4 Discussion

In this 6-month cross-sectional study at a student-run free clinic, we applied the SDOH conceptual framework to evaluate relationships between food security status, various SDOH, and chronic medical conditions. The SDOH framework emphasizes the interconnectedness of various social, economic, and environmental factors that influence health outcomes. Our study also incorporated an exploration of the impact of the COVID-19 pandemic on participant responses to the SDOH needs assessment.

Our findings bolster existing evidence that supports the SDOH theoretical framework, underscoring the interconnectedness of various SDOH domains. Corroborating previous research, we found that low food security correlates with other financially-related SDOH, including the lack of health insurance, medication unaffordability, and housing cost burden [5, 8, 39,40,41]. Notably, our study confirms these associations in patients without insurance, an understudied population, and broadens the scope of these connections to utility assistance, dental care, and mental health services. Similar to past studies, we found a heightened risk of anxiety and depression diagnoses or symptoms among participants with food insecurity [13, 42].

Among the participants who reported that the COVID-19 pandemic affected their responses to the SDOH needs assessment, most noted an effect on their food security status. This highlights the profound influence of large-scale socioeconomic disruptions, such as a pandemic, on individual SDOH needs. The pandemic deepened pre-existing inequities within the workforce, triggering a ripple effect that has exacerbated food insecurity across diverse communities [14, 43]. From April 2021 to April 2022, the urban consumer price index for food in the Midwest region drastically increased, significantly affecting food affordability [44]. Consequently, participants likely faced decreased purchasing power during the study period.

Our stratified analysis offers a nuanced perspective on the relationship between food insecurity and health outcomes, revealing significant age-related disparities. While prior research has identified associations between food insecurity and various chronic medical conditions [9, 11, 45,46,47], our findings indicate that these associations are more pronounced in younger individuals under 50 years of age. Specifically, we observed a substantial increase in the likelihood of accumulating health conditions among younger participants experiencing food insecurity. Conversely, in the older cohort (50 and over), food insecurity did not significantly influence the number of health conditions, suggesting that the impact of food insecurity on chronic disease development may diminish with age or be mitigated by other factors prevalent in older populations.

These findings highlight the complex interplay between food insecurity, age, and health outcomes as contrary to prior research, we found no association between food insecurity and several chronic medical conditions, including obesity, hypertension, diabetes, and dyslipidemia, and CKD [9, 11, 45,46,47]. This lack of correlation may also be due to our study’s temporal context—the pandemic may have recently thrust some participants into food insecurity, making it difficult to observe the potential long-term effects within our study period. Another explanation may be that the role of insurance status is a greater driving factor than food insecurity status in chronic disease development. Both explanations warrant further investigations.

The prevalence of food insecurity in Wisconsin (7.2%) and Milwaukee County (11.8%) are lower than this study population (16.3%) [48]. However, the true prevalence may be even higher, as several participants requested food-related resources despite not screening as food insecure. This discrepancy could stem from the stigma associated with admitting food insecurity or limitations of the assessment tool, the USDA 6-item short form [26], which we had chosen to decrease respondent burden. To capture a more nuanced understanding of food insecurity, future studies could consider employing the more detailed 18-item survey [49].

Our preliminary findings on participant feedback for resources provided suggest that more than half of the participants that returned to clinic found SNAP and food pantry resources useful. We plan to continue collecting and analyzing data on resource usefulness to better tailor these offerings and effectively address the needs of patients served by this study clinic.

5 Limitations

Limitations of this study include the small sample size and cross-sectional design. The large confidence intervals from regression modeling have the potential to overemphasize or underestimate the associations investigated in this study. Another limitation is that the assessment tool we used to identify food insecurity, while validated, may not fully capture the complexity of food insecurity in households with children. Another critical consideration is the temporal context of the COVID-19 pandemic, which may have introduced unique, short-term influences on food insecurity and health outcomes not typically present outside such global crises. Based on these limitations, we interpret our findings with caution and recommend a deeper assessment into these associations in a larger study population of participants without health insurance.

6 New contribution to the literature

Our study underscores the importance of incorporating an SDOH needs assessment as part of standard clinical practice, especially within clinics serving vulnerable patient populations, such as those without insurance. The intricate interplay between food security associated SDOH needs, and chronic medical conditions, especially with respect to younger individuals, revealed by our study highlights the necessity of adopting a comprehensive, multi-faceted approach to patient care. This data provides valuable context to plan interventions such as food distribution campaigns or other financial SDOH interventions tailored for uninsured populations. Healthcare providers caring for patients with food insecurity should also consider the mental health implications, as our study found these individuals to be more susceptible to anxiety and depression diagnoses or symptoms. Follow-up studies on SDOH needs, health outcomes, and resource usefulness would provide valuable insights into food insecurity trends over time for individuals without health insurance and inform the development of robust interventions.