Food insecurity prevalence and risk factors among persons with HIV in a southeastern US clinical care setting

Abstract

Food insecurity (FI) is associated with adverse health outcomes for persons with HIV (PWH). Little is known about FI among PWH in southern or non-urban settings. We examined FI prevalence, risk factors, and access to services in a southeastern HIV clinic. Among PWH in the UNC CFAR HIV Clinical Cohort who were screened for FI as part of HIV care between 2021 and 2022, we estimated unadjusted prevalence ratios (PRs) comparing the probability of reporting FI by demographic and clinical characteristics. The 479 PWH screened for FI were 65% cisgender men, 62% non-Hispanic Black PWH, a median of 54 years old (IQR 41–62), and 93% with an HIV viral load (VL) <200 copies/mL. FI prevalence was 36.3% (95% CI 32.3%−40.9%). Cisgender women and transgender adults were more likely to report FI than cisgender men (PRs 1.24 [95% CI 0.97–1.59] and 2.03 [1.32–3.12], respectively). Compared with White PWH, the PR was 1.71 (1.20–2.42) for Black and 2.44 (1.56–3.82) for Hispanic PWH. The PR was 1.42 (0.98–2.05) for PWH with VL≥200 versus <200 copies/mL. Having no or public versus private health insurance was also associated with FI. PWH with FI had a high prevalence of comorbidities including hypercholesterolemia (49%) and hypertension (48%), though these were not associated with FI. Almost half of PWH with FI were not accessing a food pantry or nutrition assistance program. Identifying FI in PWH is critical as FI is common and may contribute to viral non-suppression, poor comorbidity control, and gender and racial/ethnic health disparities in PWH.

INTRODUCTION

In 2021, 13.5 million households (10.2%) in the United States (US) experienced food insecurity, defined as being unable to acquire enough food for all household members because of insufficient money or other resources (1). Persons with HIV (PWH) bear a particularly high burden of food insecurity in the US, though prevalence estimates from prior studies vary according to study setting and population. In the Veterans Health Administration, 24% of PWH in care from 2002 to 2008 were food insecure (2). Smaller but more recent studies based in Miami and New York City HIV clinics reported prevalence estimates of 26% and 36%, respectively (3, 4). In the Women Interagency HIV Study (WIHS), up to 44% of participants from 2013 to 2015 experienced food insecurity (5–7). Studies targeting vulnerable groups of PWH, such as those who are homeless or have unstable housing, using social services, or with hepatitis C virus co-infection, have generally found higher food insecurity prevalence, ranging 50%–60% (8–13).

Studies from the general population have shown that food insecurity is associated with adverse health outcomes including mental health problems (14, 15), incidence and poor control of cardiometabolic conditions such as diabetes mellitus and hypertension (14, 16, 17), cardiovascular disease (14, 16), and mortality (18). PWH have a high burden of these health conditions (19–23), and food insecurity may contribute to their incidence and progression. Among PWH, food insecurity is further associated with lower antiretroviral therapy (ART) adherence, lower HIV care engagement, lower rates of viral suppression, more transmission risk behaviors, higher levels of inflammation, and higher mortality (24–29). Given its high prevalence and negative impact on both HIV and non-HIV-related health outcomes, addressing food insecurity in PWH is an important priority.

Several important knowledge gaps exist regarding food insecurity in PWH. First, food insecurity among PWH in the US South has not been well-characterized. Prior studies examining food insecurity prevalence among PWH have generally been based in large urban centers in California and the northeastern US (5, 6, 10, 11, 24, 30). Several studies have reported food insecurity prevalence in national samples of women or veterans with HIV (2, 5, 6, 25, 26, 28). Few studies have examined food insecurity among PWH in the South (8, 9, 31), yet this region warrants special attention. In the general population, the states in the US South have a higher prevalence of food insecurity compared to the northeast (1). PWH in the South may have different risk factors for food insecurity, such as residing in rural areas, lacking transportation, and having limited local resources supporting access to food (32). Food insecure PWH in the South may therefore have different demographic and clinical characteristics than those in urban centers in other regions. A detailed understanding of food insecurity burden and risk factors in southern PWH is essential to inform intervention efforts in clinical care.

There is also limited evidence on the clinical conditions associated with food insecurity in PWH. Many studies have investigated immunologic status, virologic suppression, and mental health and substance use problems in PWH with food insecurity (2, 5, 6, 8, 11, 24), but little is known about non-HIV medical comorbidities in food insecure PWH. In particular, cardiometabolic risk factors such as hypertension, hyperlipidemia, and diabetes mellitus may be more common in PWH with food insecurity. Evidence on these conditions in food insecure PWH may help target interventions to effectively address food insecurity and improve clinical outcomes, such as cardiometabolic risk factor control and prevention of cardiovascular and other diseases. Finally, most studies on food insecurity in PWH pre-date the COVID-19 pandemic, which may have affected food insecurity prevalence or the availability of services (1, 33, 34). There is a need to characterize food insecurity among PWH after the onset of the pandemic.

To address the knowledge gaps on food insecurity, particularly in southern populations of PWH, we aimed to investigate food insecurity in an HIV clinic in the southeast after the onset of the COVID-19 pandemic. In this study, we used food insecurity screenings conducted as part of HIV care to characterize the prevalence of food insecurity, demographic and clinical factors associated with food insecurity, and access and barriers to food services in food insecure PWH.

METHODS

Study population

This study was based in the University of North Carolina (UNC) Center for AIDS Research (CFAR) HIV Clinical Cohort (UCHCC). The UCHCC is a prospective clinical cohort including PWH receiving in HIV care at the UNC Infectious Diseases (ID) clinic since 1996, with over 7,000 participants enrolled to date (35). The UNC ID clinic is a Ryan White HIV/AIDS Program (RWHAP) funded clinic providing HIV care to PWH referred from across the state. In brief, the UCHCC collects demographics, medications, diagnoses, and laboratory results from electronic health records (EHR) and via EHR reviews. PWH provide written, informed consent to participate in the UCHCC. Most (>95%) PWH in care in the UNC ID clinic agree to participate.

In August 2021, the UNC ID clinic implemented a food insecurity screening program. As part of this program, a registered dietitian (RD) screened PWH for food insecurity, ascertained resources (e.g., food pantry, Supplemental Nutrition Assistance Program [SNAP]) and barriers to food (e.g., lack of transportation), and provided referrals to government and community-based resources. The RD targeted PWH receiving RWHAP benefits for these food insecurity screenings. In addition, PWH referred to the RD for nutritional counseling were also screened for food insecurity. Screenings were completed in the clinic during HIV care visits, via telemedicine encounters, or over the phone.

All demographic and clinical data came from the UCHCC. Data on food insecurity, income level, and resources and barriers to food access came from clinic records. For this study, we included all UCHCC participants who were screened for food insecurity between August 1, 2021, and September 30, 2022. For PWH who were screened more than once, we included only the first available screen. This study was approved by the UNC Institutional Review Board.

Study measures

The UNC ID clinic intervention ascertained food insecurity using a two-item tool derived from US Department of Agriculture Household Food Security Survey (HFSS-18) (36). The RD read two statements to PWH: “We worried whether our food would run out before we got money to buy more” and “The food that we bought just didn’t last, and we didn’t have money to get more”. PWH were asked whether the statement was “often true”, “sometimes true”, or “never true” for their household in the past month. If a patient responded “often true” or “sometimes true” to either statement, the RD recorded that patient as having food insecurity and further ascertained access and barriers to food and provided referrals (described above).

The patient characteristics of interest in our analyses were gender, race/ethnicity, age, HIV risk group, current and nadir CD4 cell count, HIV RNA viral load (VL), body mass index (BMI), comorbid health conditions (diabetes mellitus, hypertension, hypercholesterolemia), health insurance type, and annual household income as a percentage of the Federal Poverty Level (FPL). All variables were indexed to the date of food insecurity screening. Race/ethnicity was categorized as non-Hispanic Black, non-Hispanic White, Hispanic of any race, or other. HIV risk group was categorized as men who have sex with men (MSM), injection drug use (IDU, including PWH with both MSM and IDU as risk factors), or other risk factors including heterosexual. For current CD4 count, VL, BMI, and income, we used the closest measurement in the year prior and up to food insecurity screening date. BMI was categorized as underweight (<18.5), normal weight (18.5–24.9), overweight (25–29.9), or obese (≥30 kg/m²). PWH with underweight and normal weight were combined in analyses due to the small number of underweight PWH. Comorbid conditions were defined using algorithms validated in large HIV clinical cohorts in prior studies, based on laboratory measurements, medication prescriptions, and International Classification of Diseases Tenth Revision (ICD-10) diagnosis codes (19, 20, 37, 38). Hypertension was defined as meeting both of the following: 1) use of an antihypertensive medication and 2) a clinical diagnosis (ICD-10 code I10). Hypercholesterolemia was defined as meeting at least one of the following: 1) total cholesterol >240 mg/dL or 2) use of a lipid-lowering medication including statins. Diabetes mellitus was defined as meeting at least one of the following: 1) HbA1c ≥6.5%, 2) use of a diabetes mellitus specific medication (e.g., insulin), or 3) use of a diabetes mellitus related medication (e.g., metformin) with a clinical diagnosis (ICD-10 codes E08–E13). For each health condition, PWH meeting the definition prior to or on the screening date were defined as having that condition.

Statistical analysis

We estimated the prevalence of food insecurity as the proportion of PWH with food insecurity divided by all screened PWH. Due to the relatively small sample size, exact confidence intervals (CIs) for prevalence were estimated using the Mid-P method (39). We then estimated prevalence ratios (PRs) and 95% CIs comparing food insecurity prevalence by patient characteristics. CIs for PRs were estimated using the Wald method (40). Because the study objective was to describe the population with food insecurity, and there was no primary exposure variable of interest, we examined only unadjusted estimates. We did not fit multivariable models adjusting for confounding, which would obscure differences of interest between patient groups. PWH who were missing CD4 count (n=53, 11%), VL (n=10, 2%), nadir CD4 count (n=4, <1%), BMI (n=40, 8%), or income level (n=85, 18%) were excluded only from analyses using that variable. All P values were two-sided, with a pre-specific alpha of 0.05. Analyses were conducted in SAS v9.4 (SAS Inc., Cary, NC).

RESULTS

Study sample

We examined 479 PWH screened for food insecurity, who were 65% cisgender men, 62% non-Hispanic Black, and a median age of 54 years (interquartile range [IQR] 41, 62) on the date of food insecurity screening (Table 1). The median CD4 count was 664 cells/μL (IQR 450, 861), median nadir CD4 count 217 cells/μL (60, 386), and 93% of PWH had a VL <200 copies/mL. The BMI distribution was 46% obese, 30% overweight, 22% normal weight, and 2% underweight. Comorbid health conditions were common, with 48% having a history of hypercholesterolemia, 46% hypertension, and 25% diabetes mellitus. Nineteen percent of PWH were uninsured. FPL was assessed for 394 PWH; of these, 55% had an FPL ≤100% and 36% had an FPL between 101% and 200%.

Table 1.

Characteristics of 479 persons with HIV who were screened for food insecurity at the University of North Carolina (UNC) between August 1, 2021, and September 30, 2022.

Characteristic	N (%) or median (IQR)
Gender
Cisgender man	313 (65%)
Cisgender woman	154 (32%)
Transgender person	12 (3%)
Race/ethnicity
Non-Hispanic Black	298 (62%)
Non-Hispanic White	125 (26%)
Hispanic, any race	30 (6%)
Other	26 (5%)
Age, years	54 (41, 62)
HIV risk group
MSM	220 (46%)
IDU	42 (9%)
Heterosexual/othera	217 (45%)
Years in HIV care at UNC	12 (5, 20)
Current CD4 count, cells/μL	664 (450, 861)
Nadir CD4 count, cells/μL	217 (60, 386)
HIV RNA <200 copies/mL	437 (93%)
Body mass index, kg/m2
<18.5	7 (2%)
18.5–24.9	98 (22%)
25–29.9	132 (30%)
≥30	202 (46%)
Diabetes mellitus	118 (25%)
Hypertension	218 (46%)
Hypercholesterolemia	231 (48%)
Insurance type
None	93 (19%)
Medicaid	83 (17%)
Medicare	161 (34%)
Private	139 (29%)
Other	3 (1%)
Annual household income as percentage of federal poverty level
≤100%	217 (55%)
101%–200%	140 (36%)
201%–300%	34 (9%)
>300%	3 (1%)

Abbreviations: IDU, injection drug use; IQR, interquartile range; MSM, men who have sex with men.

^aThis category includes 200 persons whose risk factor was heterosexual transmission, 8 receipts of blood products, 2 health care exposure, 2 perinatal transmission, and 5 unknown.

Food insecurity prevalence and risk factors

Of the 479 screened PWH, 174 screened positive for food insecurity, a prevalence of 36.3% (95% CI 32.3%, 40.9%). Food insecurity prevalence was highest for cisgender women (40.9% [33.3%, 48.8%]), transgender adults (66.7% [37.7%, 88.4%], non-Hispanic Black PWH (39.6% [34.2%, 45.2%]), Hispanic PWH of any race (56.7% [38.7%, 73.4%]), PWH with unsuppressed VL (50.0% [33.1%, 66.9%]), and those with no insurance (50.5% [40.4%, 60.6%]) or with Medicaid (48.2% [37.6%, 58.9%]) (Table 2).

Table 2.

Food insecurity prevalence and prevalence ratios among 479 persons with HIV screened at the University of North Carolina between August 1, 2021, and September 30, 2022.

Characteristic	Prevalence (95% CI)a	PR (95% CI)b
Gender
Cisgender man	32.9% (27.9%, 38.3%)	1 (ref.)
Cisgender woman	40.9% (33.3%, 48.8%)	1.24 (0.97,1.59)
Transgender	66.7% (37.7%, 88.4%)	2.03 (1.32,3.12)
Race/ethnicity
Non-Hispanic Black	39.6% (34.2%, 45.2%)	1.71 (1.20,2.42)
Non-Hispanic White	23.2% (16.4%, 31.2%)	1 (ref.)
Hispanic, any race	56.7% (38.7%, 73.4%)	2.44 (1.56,3.82)
Other	38.5% (21.5%, 57.9%)	1.66 (0.93,2.97)
Age, years
<30	23.5% (11.6%, 39.8%)	1 (ref.)
30–39	43.2% (32.3%, 54.7%)	1.84 (0.95,3.56)
40–49	40.0% (30.0%, 50.7%)	1.70 (0.88,3.29)
50–59	36.8% (28.7%, 45.5%)	1.56 (0.82,2.99)
≥60	33.5% (26.6%, 41.1%)	1.43 (0.75,2.71)
HIV risk group
MSM	30.5% (24.6%, 36.8%)	1 (ref.)
IDU	31.0% (18.4%, 46.0%)	1.02 (0.62,1.67)
Heterosexual/other	43.3% (36.8%, 50.0%)	1.42 (1.11,1.83)
Current CD4 count, cells/μL
<350	43.8% (32.0%, 56.0%)	1.23 (0.89,1.69)
350–500	43.3% (31.8%, 55.3%)	1.22 (0.89,1.66)
>500	35.6% (30.3%, 41.2%)	1 (ref.)
Nadir CD4 count, cells/μL
<200	36.0% (29.9%, 42.4%)	0.98 (0.77,1.24)
≥200	36.8% (31.0%, 42.9%)	1 (ref.)
HIV RNA viral load, copies/mL
<200	35.2% (30.9%, 39.8%)	1 (ref.)
≥200	50.0% (33.1%, 66.9%)	1.42 (0.98,2.05)
Body mass index, kg/m2
<25	45.7% (36.4%, 55.3%)	1 (ref.)
25–29.9	43.2% (34.9%, 51.7%)	0.94 (0.71,1.26)
≥30	29.2% (23.2%, 35.8%)	0.64 (0.47,0.86)
Diabetes mellitus
Yes	39.8% (31.3%, 48.9%)	1.13 (0.87,1.47)
No	35.2% (30.4%, 40.2%)	1 (ref.)
Hypertension
Yes	38.1% (31.8%, 44.7%)	1.09 (0.86,1.38)
No	34.9% (29.3%, 40.8%)	1 (ref.)
Hypercholesterolemia
Yes	36.8% (30.8%, 43.2%)	1.03 (0.81,1.30)
No	35.9% (30.1%, 42.0%)	1 (ref.)
Insurance type
None	50.5% (40.4%, 60.6%)	2.87 (1.91,4.32)
Medicaid	48.2% (37.6%, 58.9%)	2.74 (1.80,4.17)
Medicare	38.5% (31.2%, 46.2%)	2.19 (1.46,3.28)
Private or other	17.6% (12.0%, 24.5%)	1 (ref.)
Annual household income as percentage of federal poverty level
≤100%	45.2% (38.6%, 51.8%)	2.09 (1.11,3.92)
101%–200%	40.7% (32.8%, 49.0%)	1.88 (0.99,3.59)
≥201	21.6% (10.6%, 37.0%)	1 (ref.)

Abbreviations: CI, confidence interval; PR, prevalence ratio.

^aCIs estimated using the Mid-P method.

^bCIs were estimated using the Wald method.

Food insecurity prevalence differed by patient characteristics (Table 2). Compared with cisgender men, the PR was 1.24 (95% CI 0.97, 1.59) for cisgender women, and 2.03 (1.32, 3.12) for transgender adults. Compared with white PWH, the PR was 1.71 (1.20, 2.42) for Black and 2.44 (1.56, 3.82) for Hispanic PWH. Food insecurity prevalence was higher for PWH with heterosexual/other but not those with injection drug use (IDU) as a risk factor, compared with men who have sex with men (MSM). PWH with a VL ≥200 copies/mL had a PR of 1.42 (0.98, 2.05) compared with PWH with a VL <200 copies/mL. Having no health insurance, having Medicaid, and having Medicare were all associated with a food insecurity prevalence more than twice as high as for PWH with private insurance. Having a lower income was also associated with having food insecurity, with a PR of 2.09 (1.11, 3.92) for PWH with an income ≤100% FPL compared with ≥201 FPL.

Characteristics of PWH with food insecurity

Among the 174 PWH with food insecurity, similar to the overall sample, 49% had hypercholesterolemia, 48% had hypertension, and 27% had diabetes mellitus. A majority were obese (34%) or overweight (33%), and 26% had normal weight. Sixteen (9%) food insecure PWH had a VL ≥200 copies/mL.

Of the 174 PWH with food insecurity, 126 were assessed for food services and barriers. Of these, 42% were enrolled in SNAP at the time of assessment, and 21% used a food pantry service; 44% were accessing neither SNAP nor a food pantry. Ninety-six percent had stable housing. When assessing barriers to accessing food resources, 60% reported lack of education/knowledge of available government assistance programs, how to complete applications, or location of pantries/shelters, 18% lack of transportation, and 3% lack of time.

DISCUSSION

In this study based at a large tertiary care center in North Carolina from 2021 to 2022, among 479 PWH targeted for food insecurity screening, we found that 36% were positive for food insecurity. Being food insecure was associated with being a cisgender woman or transgender person, being of minority race/ethnicity, and having public or no health insurance. Among those with food insecurity, close to half were not using a food pantry or enrolled in SNAP, and the most commonly identified barrier to food security was lack of knowledge on how to access the necessary resources.

The food insecurity prevalence in our study, 36%, is lower than what has been reported by studies in vulnerable groups of PWH, such as homeless or unstably housed PWH, PWH using social services, or PWH with hepatitis C virus co-infection, which have generally found estimates of at least 50% (8–13). The food insecurity prevalence in our study, however, is higher than several other clinic-based estimates, including a prevalence of 24% among PWH in the Veterans Health Administration and 26% in an HIV clinic in Miami (2, 4). A study in a New York City HIV clinic found a prevalence similar to our study (3). Differences in food insecurity ascertainment might have contributed to variation in prevalence estimates. For example, the UNC ID clinic intervention from which our data came used a 2-item subset of the USDA HFSS, while some studies used the 6-item form of the HFSS, and other studies used the USAID Household Food Insecurity Access Scale (HFIAS). In addition, the food insecurity screening program in the UNC ID clinic prioritized PWH receiving RWHAP benefits and those referred for nutrition services, which might have contributed to a higher prevalence. Our study also took place during the COVID-19 pandemic, which could also have worsened food security problems for PWH in our sample (41–43), though we did not have pre-COVID data for comparison in these analyses.

In this study, food insecurity prevalence was higher among cisgender women and transgender PWH, PWH who were Black or Hispanic, and those with public or no health insurance. These findings confirm in a southeastern, non-urban HIV cohort what has been previously reported in other settings. A systematic review found that food insecurity was generally higher for women versus men with HIV, including in high-income countries like the US (44). Studies have also shown that PWH of minority race/ethnicity have a high burden of food insecurity. In a sample of vulnerable Hispanic PWH in Boston, food insecurity prevalence was 83% (45). In a substudy of the Medical Monitoring Project, 24% of heterosexual Black men with HIV reported food insecurity (46). A study among veterans with HIV found that those who were food insecure were more likely to be Black or Hispanic and less likely to be white (47). In the general population, food insecurity is also more prevalent among women and people of minority race/ethnicity (48, 49). The disparately high food insecurity burden in people of minority race/ethnicity may be explained in part by higher rates of poverty (50). Persons with Medicaid or no health insurance in the general population also have a higher food insecurity prevalence compared with privately insured persons (51), though gaining Medicaid coverage may facilitate access to food services and reduce food insecurity (52). Prior studies have found food insecurity in PWH to be associated with lower CD4 counts and unsuppressed VLs (6, 53). PWH with VL ≥200 copies/mL had a higher food insecurity prevalence in our study, although the PR estimate had a wide confidence interval. Food insecurity has been linked to ART non-adherence, while food assistance programs help improve ART adherence (54–56). Addressing food insecurity could therefore play an important role in increasing viral suppression rates among PWH. We did not find any strong evidence of an association between current CD4 count and food insecurity. PWH in our sample had high CD4 counts overall compared with earlier studies, which may explain our findings. CD4 count could be more strongly associated with food insecurity among PWH who are less well engaged in care. Together, our findings highlight the importance of ensuring vulnerable populations are screened for food insecurity and provided adequate resources.

Cardiometabolic conditions including obesity, hypercholesterolemia, hypertension, and diabetes mellitus were not associated with food insecurity in this study, in contrast to prior research. In the general population, persons with food insecurity have been shown to be at greater risk of incident obesity, hypertension, and diabetes mellitus (16), though there is conflicting evidence on the link between food insecurity and dyslipidemia (57, 58). It is possible that, because the UNC ID clinic food insecurity screenings targeted RWHAP-eligible PWH and those referred for nutritional services, even food secure PWH in our study experienced economic conditions leading to poor diets and other drivers of cardiometabolic risk factors. It is also possible that food insecurity prevalence could be associated with the level of disease control (i.e., blood pressure, lipid, and glycemic levels), which our study did not capture. Future studies should investigate this potential association in PWH.

Despite the lack of association with food insecurity in this study, the high prevalence of diet-sensitive cardiometabolic risk factors has important implications for the health of food insecure PWH. Food insecurity may compromise disease management because access to nutritious food is limited (59, 60). In the general population, food insecurity has been shown to lead to worse diabetes control (17, 61). One study among veterans with HIV found that food insecurity was associated with worse control of hypertension and diabetes (47). Food insecurity could thus contribute to the incidence of cardiovascular, renal, and other diseases through poor risk factor control (14, 16). Given the high burden of both food insecurity and cardiometabolic conditions among PWH, there is a need for more research to help understand the role of food insecurity in the incidence and control of these cardiometabolic conditions.

Early intervention to address food insecurity is essential to prevent the incidence of both precursor conditions and more severe morbidity. For example, meal delivery interventions have been shown to improve diabetes control and reduce health care utilization, including in PWH (62, 63). A clinical trial providing PWH with medically tailored meals, groceries, and nutrition education reported a reduction in hospitalizations, depressive symptoms, ART non-adherence, unprotected sex, and fatty food consumption (56). In addition to improving access to nutritious food, interventions may consider targeting other structural factors that are associated with food insecurity, such as HIV-related stigma (64). Finally, while food insecurity screening may be beneficial for all PWH, some interventions may need to target subgroups at highest risk (e.g. Black, Hispanic, and transgender PWH in our study) to ensure patients most affected by food insecurity receive adequate services.

Many food insecure PWH in our study were obese or overweight by BMI classification. This paradox has previously been reported, and may be explained by the consumption of calorie-rich foods of poor nutritious value (60). Notably, PWH have been shown to experience poorer diets compared with the general population (65). BMI is also known to imperfectly predict adiposity, with variations by sex and race/ethnicity, and some patients in our study with BMIs classified as obese might not have been classified as obese by body fat or other measures (66).

Our findings provide important information for clinic-based efforts to reduce food insecurity among PWH, which is challenging and often requires addressing structural barriers. In this study, a common barrier to food security (60% of food insecure PWH) was lack of knowledge of available resources and how to access them. Providing information about local food pantries and other resources and support to enroll in benefits such as SNAP could have a substantial impact on patients’ access to food. We also found that one-fifth of PWH with food insecurity reported lack of transportation as a barrier. Transportation problems have previously been highlighted as a barrier to HIV care in PWH in the South (67), and may be a particularly important barrier to food security for PWH who do not reside in dense, urban settings. While we did not examine rural residence in our study, up to 35% of PWH entering care in the UCHCC reside in rural areas (68). In general population studies, people in rural areas have lower access to healthy food, as do people in low-income neighborhoods (69, 70), which may particularly impact RWHAP-eligible PWH. In addition, rural areas may have fewer resources available to support people with food insecurity. A study of 12 US states found that rural census tracts were less likely to have food pantries than urban tracts (71). Transportation assistance, and interventions such as grocery or meal delivery programs (62, 63, 72), may be needed to reduce food insecurity among PWH, particularly in the south.

Strengths and limitations

The strengths of this study include data on food insecurity and food resources and barriers on almost 480 PWH, with a gender and racial/ethnic diverse sample of PWH. We also had granular clinical data including medical diagnoses, medication prescriptions, and laboratory measures, which enabled us to examine HIV clinical outcomes such as CD4 count and viral load, as well as medical comorbidities defined using rigorous, validated algorithms. In addition, our data were recent, providing evidence on food insecurity in PWH after the onset of the COVID-19 pandemic.

One limitation of our study is that our findings partly reflected a targeted at-risk population and may not be representative of our entire clinic population. Similarly, we examined PWH at a single site, with a high virologic suppression rate (93%), and our findings may not be generalizable to PWH in other clinical settings or who are not engaged in care. While 6% of PWH in our studies were of Hispanic ethnicity, we did not have sufficient sample sizes to categorize these patients by race, nor did we have data on primary language spoken, country of birth, or immigration status, and future studies should further investigate food insecurity among Hispanic PWH across these factors. Another limitation is that we did not have data on the severity of food insecurity. Studies have shown that severe food insecurity and food insecurity with hunger are common among PWH and have a worse impact than less severe food insecurity (5, 30, 73, 74). Future studies should further investigate severe food insecurity, particularly in the south and among vulnerable populations, as well as evaluate longitudinal outcomes of PWH screened for food insecurity, including changes in accessing resources, becoming food secure, and the clinical impact on HIV and cardiometabolic health outcomes.

CONCLUSIONS

Over a third of PWH in a southeastern HIV clinic who were screened reported food insecurity, and food insecurity was more common among gender and racial/ethnic minorities and PWH without private insurance. Among food insecure PWH, close to half were not accessing any food services, most commonly due to a lack of knowledge. Food insecure PWH also had a high prevalence of diet-sensitive cardiometabolic conditions. Screening and addressing food insecurity in routine HIV care may help to promote adequate control of HIV and comorbidities and prevent serious morbidity and mortality.

Availability of data and material:

Due to the sensitivity nature of data, including HIV seropositivity information, data sets are not available for public sharing.