Review ArticleSystematic Review

Interventions Addressing Food Insecurity in Health Care Settings: A Systematic Review

Emilia H. De Marchis, Jacqueline M. Torres, Tara Benesch, Caroline Fichtenberg, Isabel Elaine Allen, Evans M. Whitaker and Laura M. Gottlieb
The Annals of Family Medicine September 2019, 17 (5) 436-447; DOI: https://doi.org/10.1370/afm.2412

Article Figures & Data

Figures

  • Figure 1
    Figure 1

    Study selection flow diagram.

    FI = food insecurity.

  • Figure 2
    Figure 2

    Forest plots for individual and pooled SMDs by study outcomes using random effects models.

    CalWORKS = Californial work opportunities and responsibilities to kids program; SMD = standard mean difference; WIC = women, infants, and children supplemental nutrition assistance program.

    aChange in receipt of WIC.

    bChange in receipt of CalWORKS.

    cChange in receipt of food stamps.

    dChange in vegetable consumption.

    eChange in fruit consumption.

    f95% CI and variance not calculable as mean gain for control group was zero. Note: Weights are from random effects analysis.

Tables

  • Table 1

    Types of Food Insecurity Interventions and Quality Scores for Included Studies (N = 23)

    StudyScreened for FI? Y/N (Screening Tool)aType of Intervention Quality (GRADE)
    Referral Food
    Education & PassiveNavigation & ActiveFood VouchersFood
    Beck,31 2014Y (2-item Hunger VS)✔✔Low
    Berkowitz,23 2018N✔Moderate
    Bryce,32 2017N✔✔Low
    Cavanagh,25 2017N✔✔Moderate
    Cohen,17 2017Y (1-item screener)✔✔Low
    Fleegler,35 2007Y (TOA: 6-item USDA FSS)✔Very low
    Fox,29 2016Y (2-item Hunger VS)✔✔Very low
    Freedman,33 2013Y✔✔Very low
    Freedman,26 2014Y (1-item screener)✔✔Low
    Gany,38 2015Y (18-item USDA FSS)✔✔✔Very low
    Garg,16 2007Y (WE CARE: 1-item screener)✔Moderate
    Garg,22 2015Y (WE CARE: Baseline 18-item USDA FSS; F/U 1-item screener)✔Moderate
    Hassan,37 2015Y (TOA: age specific USDA FSS)✔✔Low
    Knowles,34 2018Y (2-item Hunger VS)✔✔Very low
    Martel,40 2018Y (2-item Hunger VS)✔Very low
    Morales,24 2016Y✔✔Moderate
    Nguyen,27 2016N✔Very low
    Patel,30 2018N✔Low
    Saxe-Custack,36 2018N✔✔✔Very low
    Sege,21 2015Y (SEEK: 2-item screener)✔✔Moderate
    Smith,39 2017Y (6-item USDA FSS)✔✔✔ bVery low
    Watt,18 2015N✔✔Very low
    Weintraub,28 2010N✔✔Low

    • FI = food insecurity; F/U = follow up; GRADE = Grading Recommedations Assessment Development and Evaluation; N = no; SEEK = Safe Environment for Every Kid49; TOA = The Online Advocate (now known as HelpSteps)48; 2-item Hunger VS = 2-item Hunger Vital Sign; USDA FSS = United States Department of Agriculture-Food Security Survey; WE CARE = Well Child Care, Evaluation, Community Resources, Advocacy, Referral, Education16; Y = yes.

    • ↵a Type of food insecurity screening tool used, if noted in manuscript.

    • ↵b Only a subset of participants, those with diabetes mellitus, were eligible for food.

  • Table 2

    Process Outcomes of Interventions to Address Food Insecurity in Health Care Settings (N = 17)

    StudyDesignPopulationSampleInterventionProcess OutcomesStatistics
    Intervention type: referrals
    Garg,16 2007RCTCaregivers of pediatric patients aged 2 months to 10 years at well-child visits98 intervention, 95 controlIntervention caregivers screened with 10-item questionnaire for social needs in waiting room before well-child visitsReferral to food resource (pantry, foods stamps, WIC)1.42 (0.28-2.56), 0.34a
    Garg,22 2015Cluster RCTAdult caregivers of pediatric patients aged ≤6 months at well-child visits in 8 urban community health centers336 mothers (168 per study arm)Intervention familes screened with WE CARE tool for referral to social resourcesEnrollment in community resources
    Referral to food resources    		Food assistance program: 0.14 (–0.30 to 0.58), 0.05a
    Food pantry: 0.40 (–0.38 to 1.17), 0.16a
    0.67 (0.25-1.09), 0.05a
    Fleegler,35 2007Cross-sectionalFamilies of children aged 0-6 years who attended well-child visits at 2 urban pediatric clinics205 parents (68 with FI)Families screened with computer-based questionnaire for referrals to resourcesReferral to food resources Frequency of contacting referral agency35% (24/68) of FI patients referred 67% (16/24) contacted food resource; 94% (15/16) deemed referral helpful
    Fox,29 2016Pre-/post-intervention, pilotNew patients at a pediatric weight management clinic116 patientsIntervention to partner clinic with Second Harvest Heartland food bank with SNAP enrollment outreachEnrollment in SNAP34% (40/116) eligible for referral; 75% (30/40) accepted; 20% (3/15) completed enrollmentb
    Hassan,37 2015Prospective observationalPatients aged 15-25 years at an urban adolescent/young adult clinic401 youthWeb-based screening and referral toolFrequency of contacting any referral agency (not food specific)40% (104/259)
    Knowles,34 2018Mixed methodsCaregivers of pediatric patients aged <5 years eligible for benefits103 familiesIntegrated clinic-based referral interventionEnrollment in SNAP42% (43/103) eligible completed 85 applications; 32% (27/85) approved; 8% (7/85) denied; 60% (51/85) unknown 63% (12/19) enrolled
    Martel,40 2018Retrospective observationalPatients of urban county hospital/emergency department1,519 patientsClinic parntership with Second Harvest Heartland food bankFrequency of contacting referral agency Enrollment in SNAP74% (1,129/1,519) successfully contacted; 63% (954/1519) accepted; 92% (878/954) connected with >1 food resource 76% (338/446) of SNAP eligible completed applications
    Morales,24 2016Retrospective observational cohort with propensity score matchingPregnant patients with food insecurity at obstetrical clinic145 adult female patientsIntegrated screening and referral to Food for Families; program for referral to food resourcesEnrollment in benefits67% (97/145) enrolled
    Nguyen,27 2016Retrospective observational, pre-/post-intervention, pilotSelf-identified Hispanic patients aged ≥60 years with DM, at FQHC18/28 participants followed up at 3 monthsReferrals from clinic integrated Health Connector ProgramFrequency of contacting referral agency33% (6/18) requested food referral; 22% (4/18) contacted food resources
    Patel,30 2018Pre-/post-intervention, pilotAdult patients with DM at endocrinology clinic with access to telephone and documented financial difficulties104 patientsFinancial burden resource toolIncrease in use of farmers markets, groceries that accept food assistance0.12 (–0.16 to 0.40), 0.02a
    Sege,21 2015RCTFamilies with newborns aged <10 weeks at pediatric primary care clinic167 intervention, 163 controlIntervention group was paired with a trained family specialist who provided support (including home visits) and direct assistance accessing resourcesFood resource use0.18 (–0.08 to 0.43), 0.02a
    Weintraub,28 2010Prospective cohortPediatric patients at Peninsula family advocacy program109 participants of family advocacy program, 102 enrolled, 54 completed follow-upIntegrated clinic- and hospital-based legal servicesIncrease in use of food supportWIC: 0.73 (0.18-1.28), 0.08a; CalWORKS: 0.65 (0.11-1.20), 0.08)a; Food stamps: 0.73 (0.18-0.28), 0.08a
    Intervention type: referrals & food
    Beck,31 2014ObservationalFamilies with infants aged <1 year with FI that stretched formula or infants with failure-to-thrive at large, urban, academic pediatric primary care clinic1,042 familiesSupplemental formula and educational materials for as-needed referrals were provided directly (eg, to social workers, MLP, or food pantries)Use of social resources (social work and MLP)0.11 (0.05-0.16), <0.01a
    Cohen,17 2017Quasi-experimental; pre-/post-interventionSNAP-enrolled adult primary care patients177 patientsBrief clinic-based intervention associated with increase in uptake of SNAP incentive programDouble-up food bucks useUnadjusted OR 9.2 (95% CI, 6.1-13.8); Adjusted OR 19.2 (95% CI, 0.3-35.5)
    Freedman,26 2014Pre-/post-interventionAdult patients of FQHCs with farmers markets336 patients enrolled in Shop N Save (financial incentive for farmers market)Intervention to increase use of clinic-based farmers market and government food resourcesFarmers market revenue Use of government food assistanceIncreased from $14,285.60 to $15,719.73 (P <.001) Use of all forms food assistance: 0.51 (0.44-0.59), <0.01a; Senior farmers market nutrition program: 0.76 (0.65-0.86), <0.01a; SNAP: 0.64 (0.48-0.81), 0.01a
    Gany,38 2015Nested cohort, observationalHospital-based food pantries at 5 cancer clinics351 adult patientsUse of hospital-based food pantry after enrollment in programRepeat use of food pantryMedian return visits = 2; mean = 3.25 (SD = 3.07)
    Smith,39 2017Cross-sectionalStudent-run free clinic463 adult patientsIntegrated FI screening and intervention at free clinicUse of onsite food boxes, off-site food pantry, and SNAP enrollment43% (201/463) receiving monthly boxes of food; 14% (66/463) using off-site food pantry; 14% (64/463) enrolled in SNAP

    • CalWORKS = Calif. work opportunities and responsibilities to kids program; DM = diabetes mellitus; FI = food insecurity; FQHC = Federally Qualified Health Center; MLP = medical-legal partnership; OR = odds ratio; RCT = randomized controlled trial; SD = standard deviation; SMD = standardized mean difference; SNAP = supplemental nutrition assistance program; WE CARE = Well Child Care, Evaluation, Community Resources, Advocacy, Referral, Education; WIC = women, infants, and children supplemental nutrition assistance program.

    • ↵a Statistical results for standard mean differences are shown in format with SMD, (95% CI), varience.

    • ↵b Follow-up available for only 15 participants.

  • Table 3

    Non-Process Outcomes of Interventions to Address Food Insecurity in Health Care Settings (n = 11)

    StudyDesignPopulationSampleIntervention or Experimental ConditionOutcomesEffect Size: SMD, (95% CI), variancea
    Intervention type: referrals
    Hassan,37 2015Prospective observationalPatients aged 15-25 years at an urban adolescent/young adult clinic401 youthWeb-based screening and referral toolFood security: Complete resolution of food as priority problem58% (7/13)
    Nguyen,27 2016Retrospective observational, pre-/post-intervention, pilotSelf-identified Hispanic patients aged ≥60 years with DM at FQHC18/28 participants followed up at 3 monthsReferrals from clinic integrated Health Connector ProgramSelf-efficacy: Change in mean scores on the Stanford Diabetes Self-efficacy Scale
    Diabetes self-efficacy    		Diet/healthy eating plan: –0.14, (–0.79 to 0.51), 0.11
    Physical activity: –0.07, (–0.73 to 0.58), 0.11
    Diabetes self-efficacy: 0.30, (–0.35 to 0.96), 0.11
    General self-efficacy: 0.13, (–0.52 to 0.79), 0.11
    Morales,24 2016Retrospective observational cohort with propensity score matchingPregnant patients with FI at obstetrical clinic145 adult female patients enrolled; 145 matched not referredIntegrated screening and referral to Food for Families; program for referral to food resourcesHealth: Blood glucose Health: SBP Health: DBP0.10, (–0.13, to 0.33), 0.01 0.33, (0.09-0.56), 0.01 0.27 (0.04-0.51), 0.01
    Intervention type: referrals & food/food vouchers
    Beck,31 2014ObservationalFamilies with infants aged <1 year with FI that stretched formula or infants with failure-to-thrive at large, urban, academic primary care clinic1,042 families with infantsSupplemental formula and educational materials for as-needed referrals were provided directly (eg, to social workers, MLP, or food pantries)Utilization: Completed preventative care
    Utilization: ED visits    		Completed lead test and ASQ: 0.09, (0.04-0.15), <0.01
    Received full set of well-infant visits by 14 months: 0.11, (0.05-0.16), <0.01
    0.11, (0.05-0.16), <0.01
    Bryce,32 2017Pre-/post-interventionAdult, non-pregnant patients with type 2 DM and HbA1c >6.5 in last 3 months referred by medical provider65 patientsVoucher for fruits and vegetables, and health education/coaching at health center-based farmers marketHealth: Weight change Health: SBP change Health: DBP change Health: Drop in HbA1c–0.08, (–0.30 to 0.13), 0.01 –0.04, (–0.26 to 0.17), 0.01 0.15, (–0.06 to 0.37), 0.01 0.39, (0.17-0.60), 0.01
    Cavanagh,25 2017Retrospective matched cohort; pre-/post-interventionAdult low-income patients with obesity, hypertension, and/or type 2 DM54 intervention, 54 matched controlsVoucher (prescription coupon) for weekly mobile produce marketHealth: BMI change–0.11, (–0.18 to –0.05), <0.01
    Cohen,17 2017Quasi-experimental, pre-/post-interventionSNAP-enrolled adult primary care patients177 patientsBrief clinic-based intervention associated with increase in use of SNAP incentive programHealth behavior: Increased fruits/vegetable consumptionb0.49, (0.25-0.73), 0.01
    Freedman,33 2013Pre-/post-intervention, pilotAdult patients of FQHCs with farmers markets with DM41 patientsCommunity-based participatory research approach for onsite farmers market; financial incentive program to purchase food at marketHealth behavior: Increased fruits/vegetable consumptionc0.41, (–0.02 to 0.85), 0.05 at 2-3 months 0.15, (–0.28 to 0.58), 0.05 at 5 months
    Saxe-Custak,36 2018QualitativeAdult caregivers of pediatric patients at an urban pediatric clinic32 caregiversProvided vouchers for farmers market or bag of food when market closed; cooking/nutrition classesAcceptability
    Health behavior: Increased fruits/vegetable consumption
    Food security    		Appreciated convenience of clinic within farmers market building
    Preferred prescription vouchers over food bags Reported increased
    Improved food security and access to healthy foods
    Watt,18 2015Quasi-experimental prospectiveAdult Hispanic pregnant women at low-income Texas primary care clinic32 intervention, 29 controlPrenatal care-based nutrition education, food resources education, and farmers market vouchersHealth behavior: Increased fruits/vegetable consumptiond
    Health: Depression (mean gain PHQ2 score)
    Health: Excess maternal weight gain Health: Breastfeeding at age 6 months
    Health: Pass ASQ screening    		Fruits: d = 0.47e,f
    Vegetables: –0.71, (-1.19 to -0.22), 0.06 d-0.34,(–0.91 to 0.22), 0.08f
    –0.19, (–0.80 to 0.41), 0.09
    0.64, (–0.06 to 1.34), 0.13
    0.71, (–0.05 to 1.48), 0.15
    Intervention type: food only
    Berkowitz,23 2018Matched cohortAdult patients with dual Medicaid/Medicare eligibility; members of Common-wealth Care AllianceMedically tailored meals program: 133 intervention, 1,002 matched controls. Nontailored food program: 624 intervention, 1,318 matched controlsProvided food: impact of medically tailored meal delivery and Meals on WheelsUtilization: ED visits, inpatient admissions, use of ET
    Cost: Medical spending    		Medically tailored: ED visits: –0.26, (–0.4 to –0.10), 0.01; Inpatient admissions: –0.09, (–0.27 to 0.09), 0.01; Use of ET: –0.15, (–0.34 to 0.03), 0.01
    Non-medically tailored: ED visits: –0.15, (–0.25 to –0.06), <0.01; Inpatient admissions: –0.03, (–0.13 to 0.06), <0.01; Use of ET: –0.07, (–0.17 to 0.02), <0.02
    Medically tailored: lower medical spending; net savings $220 per participant
    Nontailored: lower medical spending: Net savings $10 per participant

    • ASQ = Ages and Stages Questionnaire; BMI = body mass index; DBP = diastolic blood pressure; DM = diabetes mellitus; ED = emergency department; ET = emergency transportation; FI = food insecurity; FQHC = Federally Qualified Health Center; HbA1c = glycated hemoglobin; MLP = medical-legal partnership; PHQ2 = Patient Health Questionnaire-2; SBP = systolic blood pressure; SMD = standard mean differences; SNAP = supplemental nutrituion assistance program.

    • ↵a Effect sizes are presented as standardized mean differences (d) unless sufficient alternatives were provided in the reviewed manuscripts (eg, Odds Ratios [ORs]). Effect sizes were not calculated when a plausible control/comparison group was not available to compare with the intervention group and/or if insufficient details were provided in the manuscript and we did not receive responses to requests for further information from study authors.

    • ↵b Increase in fruit/vegetable consumption (servings/day) at 5-month follow-up (n = 138).

    • ↵c Servings/day.

    • ↵d Reported as change from less than 3 servings to 3 or more servings per day; raw data unavailable to adjust results to report as servings per day, as would need to adjust standard deviation.

    • e 95% CI and varience not calculable as mean gain for control group = 0.

    • ↵F Author provided additional data points to enable effect size calculation.

Additional Files

  • Supplemental Appendixes

    Supplemental appendixes

    Files in this Data Supplement:

  • The Article in Brief

    Interventions Addressing Food Insecurity in Health Care Settings: A Systematic Review

    Emilia De Marchis , and colleagues

    Background Many health care settings are exploring ways to address social determinants of health, including the reduction of patient food insecurity. Health systems have begun experimenting with providing interventions to address food insecurity as a strategy to improve health; however, the impacts of such interventions on food security, patient health and health behaviors, and health care utilization and cost are unclear.

    What This Study Found A systematic review of articles covering food insecurity interventions in health care settings from 2000-2018 found that interventions that included either referrals or direct provision of food or vouchers both suffered from poor follow-up, a general lack of comparison groups, and limited statistical power and generalizability. Of the twenty-three studies included in analysis, 74% were rated low or very low quality. In studies of referral-based interventions, there were moderate increases in process outcome--like food program referrals (Standardized Mean Differences: 0.67 95% CI 0.36 to 0.98; 1.42 95% CI 0.76 to 2.08) and resource use (pooled SMD: 0.54 95% CI 0.31 to 0.78). In studies of direct food or voucher interventions, outcomes were mixed, with changes in fruit and vegetable intake averaging to no impact when pooled (-0.03 95% CI -0.66 to 0.61). Studies were limited in their evaluation of health or utilization outcomes, with generally small but positive effects.

    Implications

    • This review revealed limitations in growing literature on health care-based food insecurity interventions. Low quality and lack of studies examining health or health care outcomes limit inferences about their impacts. More rigorous evaluation of food insecurity interventions is needed that includes health and utilization outcomes to better understand roles for the health care sector in addressing food insecurity.
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools
Get Permissions

Jump to section

Subjects

Keywords