Abstract
Introduction
The Centers for Medicare and Medicaid Services has made screening for health-related social needs (HRSNs) mandatory since 2024. Patients screening positive for HRSN food insecurity (FI) are connected with Supplemental Nutrition Services using 1115 Demonstration waivers. With a shift in funding, Massachusetts adopted a more stringent FI-level screening protocol (very low food security [VLFS]). To date, there is no screening tool that identifies VLFS alone; thus, we developed a sensitive, specific, and valid screen to determine risk for VLFS.
Methods
Sensitivity, specificity, and positive- and negative-predictive values were used to test combinations of questions compared with the USDA Household Food Security Survey Module using the 2022 Current Population Survey.
Results
The best VLFS screen was, “In the last 12 months, did you or other adults in your household ever cut the size of your meals or skip meals because there wasn't enough money for food?” (yes/no). The screen was highly sensitive, specific, and valid for detecting VLFS risk in US households with and without children (sensitivity, 92.2% and 97.2%; specificity, 96.2% and 97.4%, respectively), and with diverse populations.
Conclusion
The VLFS screen is recommended for use by states adopting a more stringent FI-level screening protocol.
Keywords: food insecurity, screening tool, health-related social needs, nutrition, hunger
Introduction
Social conditions are drivers of health, which, in turn, affect medical costs and quality.1,2 Food insecurity (FI) is a health-related social need (HRSN) that negatively impacts cognitive function and physical and behavioral health of adults and children.3 Food insecurity aggravates almost any co-occurring adverse health condition, disproportionately burdening families with low socioeconomic status and families of color. Additionally, poor health care access and low quality of care can adversely impact FI.4
Food insecurity is associated with high avoidable costs for the health care system,5 calculated in 2016 to be more than $160 billion dollars per year. Accounting for inflation, this figure is $215 billion in 2025 dollars. Given FI's strong relationship to both worse health and higher costs and its remediable nature, the Centers for Medicare and Medicaid Services (CMS) has approved state HRSN Section 1115 Demonstration waivers (“1115 waivers”), a mechanism to test new approaches in Medicaid that promote the objectives of Medicaid.6 In 2022, CMS approved Massachusetts Medicaid's (MassHealth) request to extend its 1115 waiver through 2027. As of December 4, 2025, Massachusetts and 11 other states have CMS-approved 1115 waivers to test nutrition provision through partner organizations (HRSN providers).7
In 2025, MassHealth issued guidance requiring accountable care organizations (ACOs) and HRSN providers to screen members for FI and refer qualifying members (criteria: an eligible health needs-based criterion and positive FI screen) to HRSN Nutrition Services. To assess FI, MassHealth requires that “the member meets either the USDA [US Department of Agriculture] definition of low or very low food security.”8 Nevertheless, MassHealth instructed ACOs that eligibility would be granted only to members experiencing very low food security (VLFS), the most severe level of FI. However, to date, there are no screening tools that measure VLFS. Thus, MassHealth recommended using any tool that can appropriately distinguish VLFS. Scales such as the 6-item Short Form of the Household Food Security Survey Module (6-item HFSSM)9 and the 8-item Abbreviated Child and Adult Food Security Scale (ACAFSS)10 can identify VLFS, but at a cost to clinical efficiency given the multiple questions required. Thus, this study developed a 1-item sensitive, specific, and valid VLFS screening tool, based on items from the established HFSSM, to be used in states that adopt more stringent screening levels. Based on this study, MassHealth updated its current policies to replace the 6-item HFSSM with the 1-item screener, effective January 1, 2026.
Methods
The December supplement of the Current Population Survey11 (CPS, 2022) annually collects nationally representative data on FI using the 18-item or 10-item HFSSM for households with or without children, respectively. In 2022, approximately 54 000 households were interviewed across the nation. Of those, the CPS 2022 file comprised 125 489 individuals/records. The analytic sample for this study was restricted to records with complete responses to the food security survey, resulting in a sample of 74 971 records. Given that the food security responses are replicated for all members of the household, this sample was further restricted to records from household reference persons to avoid duplication of responses, resulting in a sample of 31 948 records. Finally, to examine the validity of a screen, 2 samples were created: (1) sample restricted to households with children, resulting in a final sample of 8306 records representing 37 235 419 total households with children nationwide, and (2) sample restricted to households without children, resulting in a final sample of 31 884 records representing 132 729 812 922 total households without children nationwide. The HFSSM was used to calculate 4 psychometric properties (sensitivity, specificity, positive-predictive value [PPV], and negative-predictive value [NPV]) to determine a screening tool capable of identifying members living in households at risk for VLFS. Sensitivity identifies the screen's ability to correctly identify VLFS (minimizing false negatives), whereas specificity describes the screen's ability to correctly identify those not with VLFS (minimizing false positives). Positive-predictive value provides the percentage of those identified by a VLFS positive screen that actually have VLFS; NPV provides the percentage of those identified by a negative VLFS screen that actually do not have VLFS. The PPV and NPV are contingent on the prevalence of the VLFS condition; thus, the higher the prevalence, the higher the PPV.
Contingency tables were used to test psychometric properties of 1 item or 2 items compared with the HFSSM 18-item and HFSSM 10-item for households with and without children, respectively. Following standard methods for coding responses determined for households with and without children,12 the HFSSM food security status was dichotomized: not very low food security (not-VLFS: food security and low food security combined) or VLFS. The best screening tool was selected using the following criteria: (1) 1-item or 2-item candidates, to decrease burden; (2) sensitivity greater than 90%, specificity greater than 90%, PPV greater than 55%, and NPV greater than 95% to provide a highly effective tool; (3) 2-item candidates with the same level of response (both questions: often true, sometimes true, never true or yes, no), to provide response consistency and decrease data collection burden; and (4) candidate items to be the same for households with and without children, to decrease implementation burden. In case of multiple candidates with similar psychometric characteristics, the decisive element was based on alleviating burden for those implementing the tool; thus, the 1-item candidate was selected as the VLFS screen vs 2-item candidates. The best candidate screen was also tested within households with diverse characteristics: race (White, Black, American Indian/Alaskan Native, Asian, or Hawaiian/Pacific Islander), ethnicity (Hispanic), immigration (born outside of the United States), disability, and household income (<185% federal poverty level). Data were analyzed using Stata/IC 16.1. To correctly reflect stratification and clustering effects of the complex sampling design, analyses were performed using the command svy to represent the US population.
Results
For households both with and without children, the recommended 1-item VLFS screen asks: “In the last 12 months, did you or other adults in your household ever cut the size of your meals or skip meals because there wasn't enough money for food?” (response: yes/no). In households with and without children, sensitivity was 92.2% and 97.2% and specificity was 96.2% and 97.4%, respectively. Validation included both 12-month and 30-day recall periods. High sensitivity and specificity demonstrate the 1-item VLFS screen's capacity to minimize false negatives and false positives, respectively. Given that PPV and NPV are contingent on VLFS prevalence, results are also shown for simulated prevalences (Table 1). Subgroup validation analysis shows the 1-item VLFS screen's consistent ability to identify VLFS accurately across varied strata of the population (Table 2). For full psychometric analyses of combinations of questions in households with and without children as well as final candidate questions, please refer to Tables S1–S5.
Table 1.
Psychometric analyses of the 1-item VLFS screening tool relative to the 18- or 10-item US Household Food Security Survey Module in households with and without children, respectively, using simulated VLFS prevalences with a 12-month and 30-day recall periods: Current Population Survey, 2022.
| Measure | Prevalence, % | Sensitivity, % | Specificity, % | PPV,a % | NPV,a % |
|---|---|---|---|---|---|
| 12-Month recall period | |||||
| Households with children | |||||
| National | 5.3 | 92.2 | 96.2 | 57.3 | 99.5 |
| Simulation | 6 | 92.2 | 96.2 | 60.5 | 99.5 |
| 7 | 92.2 | 96.2 | 64.3 | 99.4 | |
| 8 | 92.2 | 96.2 | 67.6 | 99.3 | |
| 9 | 92.2 | 96.2 | 70.3 | 99.2 | |
| 10 | 92.2 | 96.2 | 72.7 | 99.1 | |
| 15 | 92.2 | 96.2 | 80.9 | 98.6 | |
| 20 | 92.2 | 96.2 | 85.7 | 98.0 | |
| 25 | 92.2 | 96.2 | 88.9 | 97.4 | |
| 30 | 92.2 | 96.2 | 91.1 | 96.7 | |
| 35 | 92.2 | 96.2 | 92.8 | 95.8 | |
| Households without children | |||||
| National | 5.2 | 97.2 | 97.4 | 67.3 | 99.8 |
| Simulation | 6 | 97.2 | 97.4 | 70.5 | 99.8 |
| 7 | 97.2 | 97.4 | 73.8 | 99.8 | |
| 8 | 97.2 | 97.4 | 76.5 | 99.8 | |
| 9 | 97.2 | 97.4 | 78.7 | 99.7 | |
| 10 | 97.2 | 97.4 | 80.6 | 99.7 | |
| 15 | 97.2 | 97.4 | 86.9 | 99.5 | |
| 20 | 97.2 | 97.4 | 90.3 | 99.3 | |
| 25 | 97.2 | 97.4 | 92.6 | 99.1 | |
| 30 | 97.2 | 97.4 | 94.1 | 98.8 | |
| 35 | 97.2 | 97.4 | 95.3 | 98.5 | |
| 30-Day recall period | |||||
| Households with children | |||||
| National | 2.8 | 93.4 | 97.1 | 48.0 | 99.8 |
| Simulation | 5 | 93.4 | 97.1 | 62.8 | 99.6 |
| 10 | 93.4 | 97.1 | 78.1 | 99.2 | |
| 15 | 93.4 | 97.1 | 85.0 | 98.8 | |
| 20 | 93.4 | 97.1 | 88.9 | 98.3 | |
| 25 | 93.4 | 97.1 | 91.4 | 97.8 | |
| 30 | 93.4 | 97.1 | 93.2 | 97.2 | |
| 35 | 93.4 | 97.1 | 94.5 | 96.5 | |
| Households without children | |||||
| National | 2.8 | 97.9 | 97.8 | 56.3 | 99.9 |
| Simulation | 5 | 97.9 | 97.8 | 70.2 | 99.9 |
| 10 | 97.9 | 97.8 | 83.3 | 99.8 | |
| 15 | 97.9 | 97.8 | 88.8 | 99.6 | |
| 20 | 97.9 | 97.8 | 91.8 | 99.5 | |
| 25 | 97.9 | 97.8 | 93.7 | 99.3 | |
| 30 | 97.9 | 97.8 | 95.0 | 99.1 | |
| 35 | 97.9 | 97.8 | 96.0 | 98.9 | |
aPPV and NPV are contingent on prevalence.
Abbreviations: NPV, negative-predictive value; PPV, positive-predictive value; VLFS, very low food security.
Table 2.
Psychometric analyses of the 1-item VLFS screening tool relative to the 18- or 10-item US Household Food Security Survey Module in households with and without children, respectively, among several strata of the population with a 12-month recall period: Current Population Survey, 2022.
| Measure | National prevalence by group, % | Sensitivity, % | Specificity, % | PPV,a % | NPV,a % |
|---|---|---|---|---|---|
| Households with children | |||||
| Overall | 5.3 | 92.2 | 96.2 | 57.3 | 99.5 |
| Respondent is White | 4.3 | 92.4 | 96.3 | 53.2 | 99.6 |
| Respondent is Black | 10.0 | 88.8 | 93.4 | 60.0 | 98.7 |
| Respondent is American Indian/Alaskan Native, Asian, Hawaiian/Pacific Islander | 3.3 | 96.0 | 98.1 | 63.6 | 99.9 |
| Respondent is Hispanic | 7.3 | 86.8 | 94.8 | 56.6 | 98.9 |
| Respondent is an immigrant | 5.1 | 80.2 | 96.2 | 53.3 | 98.9 |
| Respondent has a disability | 11.6 | 100.0 | 96,7 | 79.7 | 100.0 |
| Household income <185% federal poverty level | 11.4 | 91.8 | 92.4 | 60.8 | 98.9 |
| Households without children | |||||
| Overall | 5.2 | 97.2 | 97.4 | 67.3 | 99.8 |
| Respondent is White | 4.5 | 97.3 | 97.6 | 66.1 | 99.9 |
| Respondent is Black | 9.1 | 96.9 | 95.5 | 68.3 | 99.7 |
| Respondent is American Indian/Alaskan Native, Asian, Hawaiian/Pacific Islander | 3.1 | 94.4 | 97.9 | 58.9 | 99.8 |
| Respondent is Hispanic | 6.9 | 94.9 | 96.0 | 63.8 | 99.6 |
| Respondent is an immigrant | 5.0 | 93.9 | 97.1 | 62.8 | 99.7 |
| Respondent has a disability | 17.2 | 100.0 | 94.3 | 78.4 | 100.0 |
| Household income <185% federal poverty level | 12.5 | 97.1 | 94.0 | 69.8 | 99.6 |
aPPV and NPV are contingent on prevalence.
Abbreviations: NPV, negative-predictive value; PPV, positive-predictive value; VLFS, very low food security.
Discussion
Nationally representative data indicate that a 1-item screening tool can accurately identify members experiencing risk for VLFS. The 1-item VLFS screening tool was validated for use by states adopting a more stringent FI screening level to identify and refer members to HRSN Supplemental Nutrition Services. The screen's strong psychometric properties with consistent performance across diverse populations reliably identify members at risk of VLFS.
Section 1115 waivers have been the catalyst for innovation in screening and intervening for HRSN in multiple states.7 Several studies have evaluated the outcomes of nutrition supports and found reductions in health care spending, emergency department utilization, frequency of hospitalizations, and prevalence of FI for service participants.13-15 Standardized screening tools and protocols, along with a supportive organizational culture, can help health care providers identify and address FI.16 Thus, the 1-item VLFS screen may help improve clinical efficiency by (1) decreasing screening time significantly, (2) reducing documentation burden, (3) streamlining clinical workflows, and (4) supporting faster connection of individuals to services.
Research demonstrates that households experiencing even marginal food security—included in national counts as food secure—are at greater risk of poor health outcomes than food-secure peers.3 Therefore, while the 1-item VLFS screen is a strong tool, the VLFS threshold excludes individuals living in households classified with less severe levels of FI (ie, marginal food security or low food security) but also experiencing lack of adequate food for a healthy life. More inclusive eligibility criteria and nutrition interventions would better support the health of populations at risk of FI, but where resources or other circumstances dictate the need for more stringent criteria, the 1-item VLFS screen provides reliable results.
Given that many ACOs currently screen for FI using the Hunger Vital Sign (HVS), a 2-item screening tool that includes low and very low food security, we strongly recommend a 2-stage screening approach that preserves the validated psychometric properties of both instruments, if understanding of severity is desired. The HVS should continue to serve as the initial screener for FI risk. For patients who screen positive for FI risk, clinicians should then administer the 1-item VLFS screening tool to determine eligibility for specific interventions.
Despite the validated nature of the 1-item VLFS screener, use of this screener moves away from the original intent of HRSN nutrition services to intervene upstream to prevent negative health outcomes associated with FI. Instead, a choice to limit services to those with VLFS becomes a downstream-only intervention, essentially waiting for harm to occur before action. Therefore, standards of practice should appropriately focus on screening for FI risk using the HVS, rather than screening only for VLFS. While states with Section 1115 waivers may restrict certain services to patients experiencing VLFS (such as MassHealth's HRSN services), identifying and addressing low food security remains both clinically relevant and worthwhile for preventing progression to higher levels of FI severity.
It is worth noting that we explicitly do not recommend combining the 1-item VLFS screening tool with only 1 HVS item to create a hybrid 2-item instrument. This guidance is grounded in the premise that both screening tools were independently validated against the USDA HFSSM, which conceptualizes FI along a continuum from food security through marginal food security, to low food security, and VLFS. The HVS was validated to identify households at risk for low and/or VLFS, while the 1-item VLFS screening tool specifically identifies those experiencing VLFS. Creating hybrid combinations of these questions would invalidate the established psychometric properties of both instruments.17
Conclusion
The 1-item screen is highly sensitive, specific, and valid for detecting VLFS risk in US households with and without children, and with diverse populations. The VLFS screen is recommended for use by states adopting a more stringent FI level screening protocol to refer patients to nutrition services.
Supplementary Material
Contributor Information
Ana Poblacion, Department of Pediatrics, Boston Medical Center, Boston, MA 02119, United States; Department of Pediatrics, Boston University Chobanian and Avedisian School of Medicine, Boston, MA 02119, United States.
Richard Sheward, Department of Pediatrics, Boston Medical Center, Boston, MA 02119, United States; Department of Pediatrics, Boston University Chobanian and Avedisian School of Medicine, Boston, MA 02119, United States.
Stephanie Ettinger de Cuba, Department of Pediatrics, Boston University Chobanian and Avedisian School of Medicine, Boston, MA 02119, United States; Department of Health Law, Policy and Management, Boston University School of Public Health, Boston, MA 02118, United States.
Supplementary material
Supplementary material is available at Health Affairs Scholar online.
Notes
- 1. Potharaju K, Gottlieb LM, Pantell M, Jones DH. Do new quality measures for social risk screening ‘measure up'? Curr Opin Pediatr. 2025;37(1):7–12. 10.1097/MOP.0000000000001414 [DOI] [PubMed] [Google Scholar]
- 2. Murray GF, Rodriguez HP, Lewis VA. Upstream with a small paddle: how ACOs are working against the current to meet patients’ social needs. Health Aff (Millwood). 2020;39(2):199–206. 10.1377/hlthaff.2019.01266 [DOI] [PubMed] [Google Scholar]
- 3. Cook JT, Black M, Chilton M, et al. Are food insecurity's health impacts underestimated in the U.S. population? Marginal food security also predicts adverse health outcomes in young U.S. children and mothers. Adv Nutr. 2013;4(1):51–61. 10.3945/an.112.003228 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Jia J, Fung V, Meigs JB, Thorndike AN. Food insecurity, dietary quality, and health care utilization in lower-income adults: a cross-sectional study. J Acad Nutr Diet. 2021;121(11):2177–2186.e3. 10.1016/j.jand.2021.06.001 [DOI] [PubMed] [Google Scholar]
- 5. Cook JT, Poblacion A. Estimating the Health-related costs of food insecurity and hunger. Bread for the World Institute. Published 2015. Accessed August 30, 2025. http://hungerreport.org/costofhunger/
- 6. Hinton E, Diana A. Medicaid Section 1115 waivers: the basics. Kaiser Family Foundation. Published January 24, 2025. Accessed September 29, 2025. https://www.kff.org/medicaid/medicaid-section-1115-waivers-the-basics/
- 7. Kaiser Family Foundation . Waivers with SDOH & other DSR changes. In: Medicaid waiver tracker: approved and pending Section 1115 waivers by state. KFF website. Published December 4, 2025. Accessed December 11, 2025. https://www.kff.org/medicaid/medicaid-waiver-tracker-approved-and-pending-section-1115-waivers-by-state/#Table 3
- 8. Massachusetts Department of Health and Human Services . Attachment P: Protocol for Assessment of Beneficiary Eligibility and Needs, Infrastructure Planning, and Provider Qualifications for HRSN Services. MassHealth Medicaid and CHIP Section 1115 Demonstration. Published December 13, 2024. Accessed September 13, 2025. https://www.mass.gov/doc/attachment-p-health-related-social-needs-hrsn-approved-121324/download
- 9. Blumberg SJ, Bialostosky K, Briefel RR, Hamilton WL. The effectiveness of a short form of the Household Food Security Scale. Am J Public Health. 1999;89(8):1231–1234. 10.2105/AJPH.89.8.1231 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Poblacion A, Ettinger de Cuba S, Frank DA, et al. Development and validation of an abbreviated child and adult food security scale for use in clinical and research settings in the United States. J Acad Nutr Diet. 2023;123(10):S89–S102.e4. 10.1016/j.jand.2023.02.004 [DOI] [PubMed] [Google Scholar]
- 11. US Census Bureau . Current population survey. Published 2022. Accessed December 20, 2024. https://www.census.gov/data/datasets/time-series/demo/cps/cps-basic.2022.html#list-tab-1979780401
- 12. Bickel G, Nord M, Price C, Hamilton W, Cook J. Guide to measuring household food security, revised 2000. US Department of Agriculture, Food and Nutrition Service, Alexandria VA. Published March 2000. Accessed August 10, 2025. https://nhis.ipums.org/nhis/resources/FSGuide.pdf
- 13. Berkowitz SA, Archibald J, Yu Z, et al. Medicaid spending and health-related social needs in the North Carolina Healthy Opportunities Pilots program. JAMA. 2024;331(14):1197–1206. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Hager K, Sabatino M, Williams J, et al. Medicaid nutrition supports associated with reductions in hospitalizations and ED visits in Massachusetts, 2020-23. Health Aff (Millwood). 2024;43(4):413–421. [DOI] [PubMed] [Google Scholar]
- 15. Project Bread . Evaluation of Project Bread's flexible services program: study findings. Project Bread website. Published 2024. Accessed December 11, 2024. https://projectbread.org/research/evaluation-project-breads-flexible-services-program
- 16. Yaun JA, Rogers LW, Marshall A, McCullers JA, Madubuonwu S. Whole child well-child visits: implementing ACEs and SDOH screenings in primary care. Clin Pediatr (Phila). 2022;61(8):542–550. 10.1177/00099228221093279 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Makelarski JA, Abramsohn E, Benjamin JH, Du S, Lindau ST. Diagnostic accuracy of two food insecurity screeners recommended for use in health care settings. Am J Public Health. 2017;107(11):1812–1817. 10.2105/AJPH.2017.304033 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Citations
- US Census Bureau . Current population survey. Published 2022. Accessed December 20, 2024. https://www.census.gov/data/datasets/time-series/demo/cps/cps-basic.2022.html#list-tab-1979780401