Abstract
Introduction:
This study aimed to characterize progression from screening for food insecurity risk, to on-site food pantry referral, to food pantry utilization in pediatric primary care.
Methods:
This retrospective study included 14,280 patients ages 0–21 years with ≥1 pediatric primary care visit from March 2018-February 2020. Analyses were conducted in 2020–2022 using multivariable regression to examine patient-level demographic, clinical, and socioeconomic characteristics and systems-related factors associated with progression from screening positive for food insecurity risk to food pantry referral to completing ≥1 food pantry visit.
Results:
Of patients screened for food insecurity risk, 31.9% screened positive; 18.5% of food-insecure patients received an on-site food pantry referral. Among patients referred, 28.9% visited the food pantry. In multivariable models, higher odds of referral were found for patients living near the clinic [aOR 1.28 (95% CI, 1.03–1.59)], with each additional health-related social need reported [1.23 (1.16–1.29)], and when the index clinic encounter occurred during food pantry open hours [1.62 (1.30–2.02)]. Higher odds of food pantry visitation were found for patients with a preferred language of Haitian Creole [aOR 2.16 (95% CI, 1.37–3.39)], Hispanic race/ethnicity [3.67 (1.14–11.78)], when the index encounter occurred during food pantry open hours [1.96 (1.25–3.07)], with a clinician letter referral [6.74 (3.94–11.54)], or with a referral due to a screening-identified food emergency [2.27 (1.30–3.96)].
Conclusions:
There was substantial attrition along the pathway from screening positive for food insecurity risk to food pantry referral and utilization as well as patient-level characteristics and systems-related factors associated with successful referrals and utilization.
INTRODUCTION
The United States Department of Agriculture (USDA) describes households with food insecurity (FI) as those that are “at times, unable to acquire adequate food for one or more household members [due to] insufficient money and other resources.”1 FI is associated with poor childhood physical health2–9 and adverse developmental and behavioral outcomes.7–14 The American Academy of Pediatrics recommends incorporating screening and referral for health-related social needs, including FI, into routine pediatric practice.15 Optimizing these processes is critical, given that screening may cause unintentional harm in the absence of effective resource connection.16
In 2020, the USDA estimated 5.6 million US households with children (14.8%) were food insecure.1 FI’s episodic nature suggests that food assistance programs, even if intermittent, may be beneficial. Prior studies have identified patient- and clinician-level barriers to connecting families with food resources from healthcare settings17–23 and have documented low rates of resource connection among families with FI.24–28 In addition to federal food assistance programs (e.g., WIC, SNAP) and community food resources, on-site, healthcare-based food pantries and food bank partnerships represent feasible approaches to address FI.22,29 However, the most effective means of facilitating resource connection in these settings are not well understood, partially due to lack of healthcare system capacity to track referrals and resource uptake.25,30,31
This study aimed to leverage a health system’s unique ability to track FI screening and on-site food pantry referrals and visits to characterize progression along the pathway from FI screening to resource connection, and to identify patient-level characteristics and systems-related factors associated with each step. Characterization of modifiable failure points in the pathway may inform efforts to optimize screening, referral, and resource connection processes,25 thereby improving multiple domains of healthcare quality, including effectiveness, efficiency, and equity.32
METHODS
This retrospective cohort study included all patients ages 0–21 years with ≥1 pediatric primary care visit at Boston Medical Center (BMC) from March 2018-February 2020, who were eligible to be screened for FI at any outpatient appointment (primary care, urgent care, or subspecialty care). BMC is an academic medical center in Boston, Massachusetts and the largest safety-net health system in New England. BMC has an on-site food pantry, which is located in the same building as the pediatric clinics. Since 2016, pediatric primary care has conducted standardized screening in eight health-related social needs domains, including food, housing, education, employment, medication affordability, utilities, transportation, and caregiving responsibilities. Screening results, referrals, and food pantry visits have been captured in the electronic health record (EHR) since 2018.33 Pediatric patients are routinely screened during primary care visits, while screening in urgent or subspecialty visits is sporadic. Screening is administered on paper in English, Haitian Creole, Spanish, Portuguese, Vietnamese, or Arabic and entered into the EHR by medical assistants. All study data were extracted from EHR structured fields, and included all outpatient screening and referral instances. The Boston University Medical Campus Institutional Review Board assigned the study exempt status.
Measures
Three outcomes were assessed: screening positive for FI risk, food pantry referral based on a positive screening result, and completion of ≥1 food pantry visits following referral.
BMC’s social needs screening tool assesses for FI risk in three ways: the 2-item validated Hunger Vital Sign,34 a question regarding immediate food emergency, and an opportunity to indicate interest in food resources. Respondents were considered to have screened positive for FI risk (dichotomous outcome) if they responded “often true” or “sometimes true” to either of two Hunger Vital Sign34 statements (within the past 12 months, (1)The food you bought just didn’t last and you didn’t have money to get more or (2)You worried whether your food would run out before you got money to buy more), answered “yes” to the food emergency question (“Is this an emergency, do you need food for tonight?”), or marked “food” as a desired resource. For patients with >1 screening episode, the first positive screen during the study period was used for analyses.
A guide with information for community-based resources is automatically printed for patients who screen positive for FI risk (available in the same languages as the screener). Clinicians may also refer patients to BMC’s on-site food pantry regardless of screening status or result. These referrals can be generated in three ways: the clinician prints a referral letter for the patient to take to the pantry (“clinician letter”), the clinician submits an electronic referral (“electronic order”), or in the case of a screening-identified food emergency, the workstation automatically prints a referral (“emergency referral”). Electronic orders are visible in the EHR to food pantry staff, but do not generate a print-out or trigger a specific workflow. For patients with >1 food pantry referral during the study period, the first referral was used in analyses. A referral was considered to be based on a positive screen if it occurred ≤7 days after screening and before any subsequent clinic visits.
With any referral type, patients may visit the food pantry every two weeks and obtain three days of food for their household per visit. The pantry is open from 10am-4pm on weekdays. A food pantry referral was considered to result in a visit if the visit occurred before any subsequent referrals, as referrals do not expire.
A list of candidate predictors was developed for each outcome through conversations with key stakeholders (food pantry staff, patient navigators, clinicians, and leadership). Candidate predictors were categorized as patient-level or systems-related.
Patient-level demographic and clinical characteristics included patient age, gender (female/male), preferred language (English, Haitian Creole, Spanish, or another language), combined race/ethnicity as a proxy for experiences with racism35 (non-Hispanic Black, Hispanic, non-Hispanic white, another race/ethnicity, or unknown), home zip code bordering (or not bordering) BMC as a proxy for proximity, body mass index (BMI) category (overweight/obesity, lean BMI, or underweight), and number of primary care visits. Socioeconomic characteristics included zip code-based median household income (“neighborhood median household income”) from the 2019 American Community Survey,36 insurance type (Medicaid, commercial, or other), and number of additional health-related social needs identified from the eight-domain screener.
Systems-related factors for screening and food pantry referral encounters included clinician type for the encounter (resident physician, attending/advanced practice provider [APP], patient navigator, or other [e.g., nurse, social worker, registered dietician]), encounter type (primary care, urgent care, specialty care, or referral-only), whether the encounter occurred during food pantry open hours (yes/no), patient navigator involvement ≤7 days following the encounter (yes/no), and the three referral types: clinician letter, electronic order, and emergency referral (each yes/no, as multiple referral types can be placed per encounter).
Statistical Analysis
Descriptive statistics were used to characterize the overall study sample and bivariate analyses and multivariable logistic regression were conducted to assess patient-level characteristics and systems-related factors associated with the three outcomes. Non-Hispanic white was selected as the race/ethnicity reference as patients in this category had the lowest FI risk, therefore allowing for more straightforward interpretation of results.
Complete case analysis was used to account for missing data in multivariable models, with the exception of race/ethnicity (for which the “unknown” category was maintained due to the substantial proportion of patients in this group (17.9%)). In the food pantry referral model, patients who screened positive for FI risk were included, with the exception of those with referrals or food pantry visits that pre-dated the screening encounter and those whose first positive screen occurred ≤7 days before the study’s end (total excluded, n=393). Those referred by the “other” clinician type (n=26) were excluded due to small numbers. In the visitation model, all patients with food pantry referrals (regardless of screening status or result) were included, with the exception of those with any prior (pre-referral) food pantry visits and those referred ≤30 days before the study’s end (to ensure adequate time elapsed to assess the referral’s effectiveness; total excluded, n=106). Analyses were conducted with SAS, version 9.4.
RESULTS
A total of 14,280 patients ages 0–21 years with completed encounters in BMC pediatric primary care during the study period were identified (Table 1). Mean (SD) age was 8.1 (6.3) years. Preferred language was 67.6% English, 14.8% Haitian Creole, and 7.1% Spanish. Race/ethnicity was non-Hispanic Black for over half of the sample (53.2%); 17.1% were Hispanic and 6.0% non-Hispanic white. The majority (73.6%) had Medicaid insurance. Patient-level characteristics and systems-related factors by outcome are available in Appendix Tables 1-3. There was no evidence of substantial collinearity for any variables included in multivariable models.
Table 1.
Patient characteristics and food insecurity resource connection pathway (n= 14,280)
| Characteristics | N (%) or Mean (SD) |
|---|---|
| Patient-level (child) characteristics | |
| Age (years) | |
| <2 | 3583 (25.1%) |
| 2–5 | 2559 (17.9%) |
| 6–11 | 3654 (25.6%) |
| 12–17 | 3453 (24.2%) |
| 18–21 | 1031 (7.2%) |
| Age (years) | 8.1 (6.3) |
| Gender | |
| Female | 7207 (50.5%) |
| Male | 7073 (49.5%) |
| Preferred language | |
| English | 9659 (67.6%) |
| Haitian Creole | 2115 (14.8%) |
| Another language a | 1485 (10.4%) |
| Spanish | 1020 (7.1%) |
| Missing/unknown | 1 (<0.01%) |
| Race/ethnicity b | |
| Non-Hispanic Black | 7604 (53.2%) |
| Hispanic | 2440 (17.1%) |
| Non-Hispanic white | 854 (6.0%) |
| Another race/ethnicity | 831 (5.8%) |
| Unknown | 2551 (17.9%) |
| Zip code | |
| Not bordering BMC | 11347 (79.5%) |
| Bordering BMC | 2933 (20.5%) |
| BMI category c | |
| Lean BMI | 8355 (58.5%) |
| Overweight/obesity | 4763 (33.4%) |
| Underweight | 1119 (7.8%) |
| Missing/unknown | 43 (0.3%) |
| Neighborhood median household income d | $68,900 ($26,900) |
| Insurance type | |
| Medicaid | 10510 (73.6%) |
| Commercial | 3612 (25.3%) |
| Other | 34 (0.2%) |
| Missing/unknown | 124 (0.9%) |
| Number of primary care visits e | 4.4 (3.7) |
| Food insecurity resource connection pathway | |
| Screening for food insecurity risk during study period f | |
| Ever screened for food insecurity risk | 12442/14280 (87.1%) |
| Number of screens for food insecurity risk per patient (n=14280) | 2.3 (2.2) |
| Screening positive during study period | |
| Ever screened positive on Hunger Vital Sign | 3854/12442 (31.0%) |
| Ever screened positive for food emergency | 596/12442 (4.8%) |
| Ever requested food resources | 1241/12442 (10.0%) |
| Ever screened positive for food insecurity risk | 3964/12442 (31.9%) |
| Food pantry referrals during study period | |
| Referred to food pantry ≤7 days after first positive screen for food insecurity risk during study period g | 661/3571 (18.5%) |
| Ever referred to food pantry during study period (regardless of screening) h | 1623/14280 (11.4%) |
| Food pantry visits during study period | |
| Visited food pantry following first referral during the study period i | 439/1517 (28.9%) |
| Ever visited food pantry (regardless of referral timing) | 993/14280 (7.0%) |
| Number of food pantry visits during the study period among those who visited (n=993) | 4.9 (6.3) |
Tigrinya is the only other language with >1% prevalence (1.54%). “Another language” also includes: Albanian; American Sign Language; Amharic/Ethiopia; Arabic; Bengali/Hindi/Urdu; Bosnian/Croatian/Yulo; Cambodian; Chinese / Cantonese; Chinese/Mandarin; Dinka / Sudan; French; Fulani/Cameroon; German; Gujarati/India; Hebrew; Hindi/India; Indonesian; Italian; Japanese; Korean; Kurdish; Luganda/Uganda; Nepali; Oromo/Ethiopia; Other; Persian; Polish; Portuguese; Punjabi; Russian; Somali; Swahili/Kenya; Tamil/India; Turkish; Urdu/Pakistan; Vietnamese.
Race/ethnicity was categorized using electronic health record variables entered at an initial patient registration which indicate a patient’s race, ethnicity, and an indication (yes/no) of Hispanic origin. Individuals with an indication of Hispanic ethnicity, Hispanic/Latino race, or an ethnicity from a country within Latin America were considered to be ‘Hispanic.’ Of the remaining individuals, those with an indication of African American or Black race or ethnicity were considered to be ‘non-Hispanic Black.’ Of the remaining individuals, those with an indication of white race were considered to be ‘non-Hispanic white.’ Of the remaining individuals, those with an indication of American Indian/Native American, Asian, Asian Indian, Cape Verdean, Caribbean Islander, Middle Eastern, or Native Hawaiian/Pacific Islander, among others were considered to be ‘another race/ethnicity’ due to small sample sizes. Individuals for whom race/ethnicity data were not available or unknown, those who declined data collection, and those with an ethnicity of ‘American’ were considered to be of ‘unknown race/ethnicity.’
BMI=body mass index (kg/m2). BMI was unknown/missing for 43 patients. Categorization for patients < 20 years of age: BMI <5th percentile = Underweight; BMI 5th - 85th percentile= Lean BMI; BMI ≥85th = Overweight/Obesity. For patients ≥20 years of age: BMI <18.5= Underweight; BMI 18.5 to <25= Lean BMI; BMI ≥25=Overweight/Obesity.
Zip code-based median household income was evaluable for 14,239 patients.
Well-child or urgent care visits during study period
Screening for food insecurity risk includes the Hunger Vital Sign measure, a question regarding immediate food emergency, and the opportunity to indicate interest in connection to food resources.
Of the 3,964 patients who ever screened positive for food insecurity risk during the study period, 3,571 patients were evaluated for presence of a resulting food pantry referral. 393 patients were excluded due to having a positive screening result that occurred ≤7 days before the study’s end or having a food pantry referral or visit that pre-dated the screening encounter.
All patients that were referred to the food pantry were included, regardless of screening status or result and regardless of their exclusion in the food pantry referral analysis. Of the 1,623 patients who were referred to the food pantry, 661 were referred ≤7 days after the first positive screen for food insecurity risk and before any subsequent clinic visits, 104 were referred but did not have documented screening results for food insecurity risk during the study period, 306 were referred but screened negative for food insecurity risk during the study period, 222 were referred from the 393 that were excluded from the food pantry referral analysis, and 330 were referred to the food pantry >7 days after 1st positive screen (e.g. after a subsequent screen or encounter).
Of the 1,623 patients ever referred to food pantry during study period, 1,517 patients were evaluated for the presence of a resulting food pantry visit. 106 patients were excluded due to having a food pantry visit that pre-dated the referral episode captured during the study period or having a referral placed ≤30 days before the study’s end.
Of the 14,280 primary care patients identified, 12,442 (87.1%) were screened for FI risk (Table 1). Of those screened, 3,964 (31.9%) screened positive for FI risk at least once, including 31.0% who screened positive on the Hunger Vital Sign, 4.8% who endorsed food emergency, and 10.0% who desired food resources (Appendix Table 4). Of 3,571 patients with no prior food pantry visits who screened positive for FI risk, 661 (18.5%) had a food pantry referral ≤7 days after first positive screen. An additional 330 patients were referred >7 days after first positive screen (e.g., after a subsequent screen or encounter), and 410 patients were referred either without documented screening results or without screening positive (Appendix Figure 1). Of 1,517 total patients referred, 439 (28.9%) had a resulting food pantry visit.
A subset of screened patients with complete covariate data (n=12,315) were included in models assessing patient-level characteristics associated with a positive FI risk screen (Table 2). Patients with a preferred language of Haitian Creole or Spanish (vs. English) had higher odds of screening positive [adjusted odds ratios [aOR] 1.59 (95% CI, 1.42–1.78) and 1.80 (1.51–2.14), respectively]. Patients of non-Hispanic Black, Hispanic, or unknown race/ethnicity (vs. non-Hispanic white) also had higher odds of screening positive [aOR 2.22 (1.79–2.76), 2.03 (1.60–2.57), 1.50 (1.19–1.89), respectively]. Higher odds of screening positive were also noted in patients with Medicaid (vs. commercial) insurance [aOR 2.46 (2.22–2.73)] and with each additional primary care visit attended during the study period [aOR 1.06 (1.04–1.07)]. Older patient age [aOR 0.98 (0.97–0.98)], higher neighborhood median household income [aOR 0.95 (0.93–0.96)], and “another” language (vs. English) [aOR 0.71 (0.61–0.82)] were associated with lower odds of screening positive.
Table 2.
Factors associated with food insecurity risk, among patients screened for food insecurity risk (n= 12,315)
| Characteristics/Factors | Odds of Identification of Food Insecurity Risk |
|---|---|
| Patient-level (child) characteristics | Adjusted Odds Ratio (95% CI) |
| Age (years) | 0.98 (0.97, 0.98) |
| Gender | |
| Female | 0.98 (0.90, 1.06) |
| Male | 1.00 [Ref] |
| Preferred language | |
| Haitian Creole | 1.59 (1.42, 1.78) |
| Another language | 0.71 (0.61, 0.82) |
| Spanish | 1.80 (1.51, 2.14) |
| English | 1.00 [Ref] |
| Race/ethnicity | |
| Non-Hispanic Black | 2.22 (1.79, 2.76) |
| Hispanic | 2.03 (1.60, 2.57) |
| Another race/ethnicity | 1.25 (0.94, 1.65) |
| Unknown | 1.50 (1.19, 1.89) |
| Non-Hispanic white | 1.00 [Ref] |
| Neighborhood median household income (in ten thousands) a | 0.95 (0.93, 0.96) |
| Insurance type | |
| Medicaid | 2.46 (2.22, 2.73) |
| Other | 1.24 (0.46, 3.35) |
| Commercial | 1.00 [Ref] |
| Number of primary care visits | 1.06 (1.04, 1.07) |
Zip code-based median household income
Note: Boldface indicates odds ratio estimates are statistically different from the null according to the 95% confidence interval.
A subset of patients with complete covariate data and an eligible positive FI risk screening instance (n=3,445) were included in food pantry referral models (Table 3). For patient-level characteristics, higher odds of food pantry referral were found for patients with a home zip code bordering BMC (aOR 1.28 (95% CI, 1.03–1.59)) and with each additional health-related social need reported (aOR 1.23 (1.16–1.29)). Systems-related factors associated with food pantry referral included an index clinic encounter during food pantry open hours (aOR 1.62 (1.30–2.02)) and being seen by a resident physician (vs. attending/APP; aOR 1.48 (1.18–1.86)).
Table 3.
Factors associated with food pantry referral, among patients who screened positive for food insecurity risk (n= 3,445)
| Characteristics/Factors | Odds of Food Pantry Referral |
|---|---|
| Patient-level (child) characteristics | Adjusted Odds Ratio (95% CI) |
| Age (years) | 1.01 (0.99, 1.02) |
| Gender | |
| Female | 1.00 (0.84, 1.20) |
| Male | 1.00 [Ref] |
| Preferred language | |
| Haitian Creole | 0.91 (0.70, 1.17) |
| Another language | 1.14 (0.80, 1.62) |
| Spanish | 1.36 (0.95, 1.95) |
| English | 1.00 [Ref] |
| Race/ethnicity | |
| Hispanic | 1.26 (0.69, 2.30) |
| Non-Hispanic Black | 1.62 (0.93, 2.82) |
| Another race/ethnicity | 0.98 (0.46, 2.08) |
| Unknown | 0.97 (0.53, 1.79) |
| Non-Hispanic white | 1.00 [Ref] |
| Zip code | |
| Bordering BMC | 1.28 (1.03, 1.59) |
| Not bordering BMC | 1.00 [Ref] |
| BMI category a | |
| Overweight/Obese | 1.07 (0.88, 1.29) |
| Underweight | 0.88 (0.62, 1.26) |
| Lean BMI | 1.00 [Ref] |
| Neighborhood median household income (in ten thousands) b | 1.03 (0.99, 1.06) |
| Insurance type | |
| Medicaid | 1.06 (0.80, 1.40) |
| Commercial | 1.00 [Ref] |
| Number of health-related social needs (apart from food insecurity) | 1.23 (1.16, 1.29) |
| Systems-related factors | |
| Appointment type for screening encounter | |
| Specialty Care | 2.26 (0.93, 5.49) |
| Urgent Care | 0.80 (0.59, 1.08) |
| Primary Care | 1.00 [Ref] |
| Clinician type for screening encounter | |
| Resident | 1.48 (1.18, 1.86) |
| Attending Physician/APP c | 1.00 [Ref] |
| Patient navigator involved | |
| Yes | 1.22 (0.98, 1.52) |
| No | 1.00 [Ref] |
| Food pantry open during screening encounter d | |
| Yes | 1.62 (1.30, 2.02) |
| No | 1.00 [Ref] |
BMI= body mass index. If a BMI was not available from the encounter, the most proximal measurement within 6 months was used.
Zip code-based median household income
APP = Advanced Practice Provider
The food pantry is open from 10am-4pm on Monday-Friday. The food pantry was considered “closed” for weekend visits and for weekday visits scheduled to start at 3:30pm or later, to account for time to finish the visit and arrive at the food pantry.
Note: Boldface indicates odds ratio estimates are statistically different from the null according to the 95% confidence interval.
A subset of patients with an eligible food pantry referral and complete covariate data (n=1,026) were included in food pantry visit models (Table 4). Patient-level factors associated with higher odds of food pantry visitation included a preferred language of Haitian Creole or “another” language (vs. English; aOR 2.16 (95% CI, 1.37–3.39) and aOR 2.42 (1.31–4.46), respectively) and Hispanic race/ethnicity (vs. non-Hispanic white; aOR 3.67 (1.14–11.78)). Systems-related factors associated with food pantry visits included having an index clinic encounter during food pantry open hours (aOR 1.96 (1.25–3.07)), having a clinician letter referral (vs. no clinician letter; aOR 6.74 (3.94–11.54)), and having an emergency referral (vs. no emergency referral; aOR 2.27 (1.30–3.96)).
Table 4.
Factors associated with a food pantry visit, among patients with a food pantry referral (n= 1,026)
| Characteristics/Factors | Odds of Food Pantry Visit |
|---|---|
| Patient-level (child) characteristics | Adjusted Odds Ratio (95% CI) |
| Age (years) | 1.01 (0.98, 1.04) |
| Gender | |
| Female | 0.85 (0.62, 1.15) |
| Male | 1.00 [Ref] |
| Preferred language | |
| Haitian Creole | 2.16 (1.37, 3.39) |
| Another language | 2.42 (1.31, 4.46) |
| Spanish | 1.26 (0.71, 2.26) |
| English | 1.00 [Ref] |
| Race/ethnicity | |
| Non-Hispanic Black | 2.49 (0.81, 7.71) |
| Hispanic | 3.67 (1.14, 11.78) |
| Another race/ethnicity | 1.43 (0.34, 6.06) |
| Unknown | 0.98 (0.28, 3.40) |
| Non-Hispanic white | 1.00 [Ref] |
| Zip code | |
| Bordering BMC | 1.35 (0.93, 1.94) |
| Not bordering BMC | 1.00 [Ref] |
| BMI category a | |
| Overweight/Obese | 1.10 (0.78, 1.55) |
| Underweight | 0.80 (0.43, 1.49) |
| Lean BMI | 1.00 [Ref] |
| Neighborhood median household income (in ten thousands) b | 1.06 (1.00, 1.12) |
| Insurance type | |
| Medicaid | 0.87 (0.52, 1.46) |
| Commercial | 1.00 [Ref] |
| Number of health-related social needs | 0.94 (0.86, 1.02) |
| Systems-related factors | |
| Appointment type for referral encounter | |
| Specialty Care | 0.50 (0.14, 1.81) |
| Urgent Care | 1.27 (0.75, 2.17) |
| Referral-Only Encounter | 1.02 (0.43, 2.41) |
| Primary Care | 1.00 [Ref] |
| Clinician type for referral encounter | |
| Patient Navigator | 0.64 (0.16, 2.54) |
| Resident | 0.63 (0.39, 1.01) |
| Other c | 1.60 (0.52, 4.98) |
| Attending Physician/APP d | 1.00 [Ref] |
| Patient navigator involved | |
| Yes | 0.91 (0.62, 1.35) |
| No | 1.00 [Ref] |
| Food pantry open during referral encounter e | |
| Yes | 1.96 (1.25, 3.07) |
| No | 1.00 [Ref] |
| Clinician letter referral | |
| Yes | 6.74 (3.94, 11.54) |
| No | 1.00 [Ref] |
| Emergency referral | |
| Yes | 2.27 (1.30, 3.96) |
| No | 1.00 [Ref] |
| Electronic order referral | |
| Yes | 0.79 (0.48, 1.31) |
| No | 1.00 [Ref] |
BMI= body mass index. If a BMI was not available from the encounter, the most proximal measurement within 6 months was used.
Zip code-based median household income
Other includes: psychologist, social worker, health educator, registered nurse, registered dietitian, inpatient clinician.
APP = Advanced Practice Provider
The food pantry is open from 10am-4pm on Monday-Friday. The food pantry was considered “closed” for weekend visits and for weekday visits scheduled to start at 3:30pm or later, to account for time to finish the visit and arrive at the food pantry.
Note: Boldface indicates odds ratio estimates are statistically different from the null according to the 95% confidence interval.
DISCUSSION
In this retrospective cohort study of pediatric primary care patients, patient-level characteristics and systems-related factors associated with FI risk and food pantry referrals and utilization were identified. Similar to findings from other pediatric primary care settings,24,37 nearly a third of patients had FI risk. Likewise, in line with USDA survey data,1 higher odds of screening positive were identified for FI risk for patients of younger age, non-Hispanic Black or Hispanic race/ethnicity, and lower income (based on neighborhood median household income and Medicaid insurance), as well as non-English preferred languages. Substantial attrition was identified along the pathway, with only 18.5% of patients with screening-identified FI risk getting referred ≤7 days after screening and 28.9% of referrals resulting in a food pantry visit. Potentially modifiable systems-related factors associated with food pantry referrals and visits were identified, including referral method and food pantry open hours.
Few prior studies have examined factors associated with successful health-related social needs resource referral and connection pathways for pediatric patients. Fritz, et al. examined factors associated with community-based organization referral acceptance and food resource connection from multiple pediatric care locations.24 All families who screened positive for FI were offered referrals, though the most substantial attrition occurred at the referral acceptance step in the clinical setting.24 In contrast, all families who screened positive for FI in the current study were provided a resource guide, but clinicians had to initiate food pantry referrals (with the exception of food emergencies). As found in other non-pediatric studies related to food resource referrals, potential reasons for low referral and resource connection rates include both clinician- and patient-level barriers. Clinician-level barriers may include time constraints, competing demands, lack of awareness of screening results, insufficient process training, and lack of feedback on referral effectiveness.17–20 Additional patient-level factors may include declining a referral offer due to food pantry-related stigma, inaccessibility (i.e., transportation, hours of operation, or the capacity to carry groceries home), approachability (lack of information needed to access the resource), or the food resource not meeting their needs or preferences.18,20–23,38
Tracking progress in providing equitable, patient-centered care requires monitoring systems that can measure and provide feedback on multiple domains of healthcare quality in local contexts, including the effectiveness, efficiency, and equity of care.32 In the context of this study, EHR-based tools were leveraged to identify opportunities for process change in the FI screening, referral, and utilization pathway for an on-site food pantry. Several potentially modifiable systems-related factors were identified. First, having an index clinic encounter during food pantry open hours was associated with both successful referral and visitation, suggesting that effectiveness could be improved by expanding the hours of operation and by referring to community-based resources with evening or weekend hours. Second, increased odds of referral for patients who live closer to the clinic suggests that referrals may be offered or accepted based on proximity. To improve geographic equity, clinic staff must understand and address logistical barriers for patients, including transportation and food-carrying needs, while also ensuring that trackable processes for efficient referral to more convenient community-based resources are developed. Finally, related to efficiency and effectiveness, the findings suggest that current emergency and electronic order referral processes do not facilitate food pantry visitation as well as clinician letters. For example, the odds of food pantry visitation was <1 for electronic orders. Without a printed clinician letter or clear communication from the referring clinician, families may not know or recall that an electronic order was placed. Additionally, the lower odds of pantry visits based on emergency referrals relative to clinician letters may result from either lack of knowledge (on the part of either clinician or patient) that the referral was generated or a lapse in communication about the referral. Education to increase clinicians’ awareness of automatic processes may be helpful, as well as ensuring that operational constraints such as malfunctioning printers do not limit resource connection. While food pantry staff currently prefer printed referrals (clinician letter or emergency referral), as they maintain their primary records outside of the EHR, pantry visits may also increase with implementation of more robust EHR-based tools such that staff members or patient navigators could proactively monitor electronic orders and call patients to provide information about food pantry services, location, and hours.
Further qualitative research involving patients, families, and clinicians should examine factors associated with successful resource connection.25 For example, number of health-related social needs was associated with higher odds of food pantry referral, but lower odds (not significant) of visitation, suggesting there may be additional barriers to visiting the food pantry that the analysis did not elucidate. Likewise, the finding that patients of Hispanic race/ethnicity or those with certain non-English preferred languages had higher odds of completing a food pantry visit suggests other unmeasured cultural factors may impact food pantry visitation or that tailored guidance or support are being offered to these patient groups.
Strengths of this study include the large sample size as well as the unique ability to track food pantry referrals and confirm visits through the EHR, which is not possible with most healthcare-based resource referral processes. In addition, the study quantitatively examines referral and resource connection successes and failures, thereby identifying opportunities for improvement.25
Limitations
First, although the food pantry provides food to households, the EHR tracks referrals and visits at the individual level; this may result in an underestimation of process efficiency if members of the same household (e.g., siblings) screen positive for FI risk but only one receives a referral or completes a visit. Second, though all patients who screen positive for FI risk receive a community-based resource referral guide and some receive support from clinic-based patient navigators, the system was unable to track referrals and connection to resources other than the on-site food pantry (i.e., community food pantries, WIC, SNAP). As healthcare systems become increasingly involved in health-related social needs screening and referral, intentional investment and development of tracking systems for these processes is needed to facilitate identification of opportunities for resource connection optimization.25,31 Third, unlike clinical care cascades such as identification of and linkage to care for infectious diseases,39,40 the optimal progression in health-related social needs pathways is not as clear. Not all families with FI desire connection to resources41 and an on-site food pantry may not be an ideal resource for all families. In addition, although the prevalence of on-site food pantries has increased in recent years, there may be limited generalizability of these findings to other settings.
CONCLUSIONS
As clinic-based FI screening and referral becomes increasingly routine, health systems must develop effective, efficient, and equitable processes for tracking screening and referrals and connecting patients with desired resources. In this sample of pediatric primary care patients, the unique ability to analyze the pathway from FI screening to resource connection allowed for identification of several modifiable factors for process improvement. This study illuminates the need for more comprehensive solutions for tracking connection to community-based and federal food resources, in addition to on-site offerings. Optimization of health-related social needs referral and utilization processes is a crucial component of healthcare systems’ promotion of quality care, particularly given that screening may result in unintentional harm without effective resource connection.
Supplementary Material
Appendix Figure 1. Food Insecurity Resource Connection Pathway
Note:
a 393 patients were excluded from the food pantry referral analysis due to having a positive screening result that occurred ≤7 days before the study’s end or having a food pantry referral or visit that pre-dated the screening encounter.
b All patients that were referred to the food pantry were included, regardless of screening status or result and regardless of their exclusion in the food pantry referral analysis. Of the 1,623 patients who were referred to the food pantry, 661 were referred ≤7 days after the 1st positive screen for food insecurity risk and before any subsequent clinic visits, 104 were referred but did not have documented screening results for food insecurity risk during the study period, 306 were referred but screened negative for food insecurity risk during the study period, 222 were referred from the 393 that were excluded from the food pantry referral analysis, and 330 were referred to the food pantry >7 days after 1st positive screen (e.g. after a subsequent screen or encounter).
c 106 patients were excluded from the food pantry visitation analysis due to having a food pantry visit that pre-dated the referral episode captured during the study period or having a food pantry referral placed ≤30 days before the study’s end.
Appendix Table 1. Patients screened for food insecurity risk stratified by screening positive for food insecurity risk (n=12,442)
Appendix Table 2. Patients who screened positive for food insecurity risk stratified by food pantry referral (n=3,571)
Appendix Table 3. Patients with food pantry referral stratified by food pantry visit (n=1,517)
Appendix Table 4. Responses to food insecurity questions among patients who screened positive for food insecurity risk (N=3,964)
Acknowledgements:
The authors thank Latchman Hiralall from the BMC Preventive Food Pantry for his guidance and support of this study.
This work was supported by The Joel and Barbara Alpert Endowment for Children of the City and the BMC Pediatrics Center for the Urban Child and Healthy Family. Dr. Egan was funded under grant number T32HS022242 from the Agency for Healthcare Research and Quality (AHRQ), U.S. Department of Health and Human Services. Dr. Fiechtner was supported by grant number K23HD090222 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. Dr. Hsu was supported by grant number K01DA054328 from the National Institute on Drug Abuse. The content is solely the responsibility of the authors and does not necessarily represent the official views of the AHRQ, U.S. Department of Health and Human Services, the National Institutes of Health, or any other funders. The study sponsors did not have any role in study design, analysis, interpretation of data, writing the manuscript, or the decision to submit the manuscript for publication.
Footnotes
No financial disclosures have been reported by the authors of this paper.
The article contents have not previously been published elsewhere.
REFERENCES
- 1.Coleman-Jensen A, Rabbitt MP, Gregory CA, Singh A. Household Food Security in the United States in 2020; 2021. [Google Scholar]
- 2.Park K, Kersey M, Geppert J, Story M, Cutts D, Himes JH. Household food insecurity is a risk factor for iron-deficiency anaemia in a multi-ethnic, low-income sample of infants and toddlers. Public Health Nutr 2009;12(11):2120–2128. doi: 10.1017/S1368980009005540 [DOI] [PubMed] [Google Scholar]
- 3.Skalicky A, Meyers AF, Adams WG, Yang Z, Cook JT, Frank DA. Child food insecurity and iron deficiency anemia in low-income infants and toddlers in the United States. Matern Child Health J 2006;10(2):177–185. doi: 10.1007/s10995-005-0036-0 [DOI] [PubMed] [Google Scholar]
- 4.Cook JT, Frank DA, Berkowitz C, et al. Food insecurity is associated with adverse health outcomes among human infants and toddlers. J Nutr 2004;134(6):1432–1438. doi: 10.1093/jn/134.6.1432 [DOI] [PubMed] [Google Scholar]
- 5.Alaimo K, Olson CM, Frongillo J, Briefel RR. Food insufficiency, family income, and health in US preschool and school-aged children. Am J Public Health 2001;91(5):781–786. doi: 10.2105/AJPH.91.5.781 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.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;4:51–61. doi: 10.3945/an.112.003228 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Drennen CR, Coleman SM, Ettinger de Cuba S, et al. Food Insecurity, Health, and Development in Children Under Age Four Years. Pediatrics 2019;144(4):e20190824. doi: 10.1542/peds.2019-0824 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Thomas MM, Miller DP, Morrissey TW. Food Insecurity and Child Health. Pediatrics 2019;144(4):e20190397. doi: 10.1542/peds.2019-0397 [DOI] [PubMed] [Google Scholar]
- 9.Cook JT, Frank DA. Food security, poverty, and human development in the United States. Ann N Y Acad Sci 2008;1136(1):193–209. doi: 10.1196/annals.1425.001 [DOI] [PubMed] [Google Scholar]
- 10.Shankar P, Chung R, Frank DA. Association of Food Insecurity with Children’s Behavioral, Emotional, and Academic Outcomes: A Systematic Review. J Dev Behav Pediatr 2017;38(2):135–150. doi: 10.1097/DBP.0000000000000383 [DOI] [PubMed] [Google Scholar]
- 11.Grineski SE, Morales DX, Collins TW, Rubio R. Transitional dynamics of household food insecurity impact children’s developmental outcomes. J Dev Behav Pediatr 2018;39(9):715–725. doi: 10.1097/DBP.0000000000000598 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Whitaker RC, Phillips SM, Orzol SM. Food insecurity and the risks of depression and anxiety in mothers and behavior problems in their preschool-aged children. Pediatrics 2006;118(3):e859–68. doi: 10.1542/peds.2006-0239 [DOI] [PubMed] [Google Scholar]
- 13.Rose-Jacobs R, Black MM, Casey PH, et al. Household Food Insecurity: Associations With At-Risk Infant and Toddler Development. Pediatrics 2007;121(1):65–72. doi: 10.1542/peds.2006-3717 [DOI] [PubMed] [Google Scholar]
- 14.Alaimo K, Olson CM, Frongillo EA. Food insufficiency and American school-aged children’s cognitive, academic, and psychosocial development. Pediatrics 2001;108(1):44–53. [PubMed] [Google Scholar]
- 15.AAP COUNCIL ON COMMUNITY PEDIATRICS. Poverty and child health in the United States. Pediatrics 2016;137(4):e20160339. [DOI] [PubMed] [Google Scholar]
- 16.Garg A, Boynton-Jarrett R, Dworkin PH. Avoiding the unintended consequences of screening for social determinants of health. JAMA - J Am Med Assoc 2016;316(8):813–814. doi: 10.1001/jama.2016.9282 [DOI] [PubMed] [Google Scholar]
- 17.Greenthal E, Jia J, Poblacion A, James T. Patient experiences and provider perspectives on a hospital-based food pantry: a mixed methods evaluation study. Public Health Nutr 2019;(12):1–9. doi: 10.1017/S1368980019002040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Steeves-Reece AL, Totten AM, Broadwell KDB, Richardson DM, Nicolaidis C, Davis MM. Social Needs Resource Connections: A Systematic Review of Barriers, Facilitators, and Evaluation. Am J Prev Med 2022;62(5):e303–e315. doi: 10.1016/j.amepre.2021.12.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Walker D, Depuccio M, Clark A, Garner J, Coovert N, Hefner J. Examining a clinic-based referral program to address food insecurity: A multi-stakeholder qualitative study of implementation. Ann Fam Med 2022;20(Supplement 1):2868. [Google Scholar]
- 20.Depuccio MJ, Garner JA, Hefner JL, Coovert N, Clark A, Walker DM. Multi-stakeholder perspectives on the implementation of a clinic-based food referral program for patients with chronic conditions: a qualitative examination. Transl Behav Med 2022;12:927–934. doi: 10.1093/tbm/ibac027 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Marpadga S, Fernandez A, Leung J, Tang A, Seligman H, Murphy EJ. Challenges and Successes with Food Resource Referrals for Food-Insecure Patients with Diabetes. Perm J 2019;23(18–097). doi: 10.7812/TPP/18-097 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zack RM, Rodríguez Bronico J V., Babbin M, et al. Facilitators and Barriers to Patient Attendance at a Free Health Center Produce Market. Am J Prev Med 2022;63(3):S131–S143. doi: 10.1016/j.amepre.2022.03.034 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sandhu S, Lian T, Smeltz L, Drake C, Eisenson H, Bettger JP. Patient Barriers to Accessing Referred Resources for Unmet Social Needs. J Am Board Fam Med 2022;35(4):793–799. doi: 10.3122/jabfm.2022.04.210462 [DOI] [PubMed] [Google Scholar]
- 24.Fritz CQ, Thomas J, Brittan MS, Mazzio E, Pitkin J, Suh C. Referral and Resource Utilization Among Food Insecure Families Identified in a Pediatric Medical Setting. Acad Pediatr 2020;000. doi: 10.1016/j.acap.2020.11.019 [DOI] [PubMed] [Google Scholar]
- 25.Schweitzer A, Mohta NS. Pathways to Success in Meeting Health-Related Social Needs. NEJM Catal 2023;4(4). doi: 10.1056/cat.22.0352 [DOI] [Google Scholar]
- 26.Stenmark SH, Steiner JF, Marpadga S, Debor M, Underhill K, Seligman H. Lessons Learned from Implementation of the Food Insecurity Screening and Referral Program at Kaiser Permanente Colorado. Perm J 2018;22:18–093. doi: 10.7812/TPP/18-093 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Chagin K, Choate F, Cook K, Fuehrer S, Misak JE, Sehgal AR. A Framework for Evaluating Social Determinants of Health Screening and Referrals for Assistance. J Prim Care Community Heal 2021;12:1–8. doi: 10.1177/21501327211052204 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Schickedanz A, Sharp A, Hu YR, et al. Impact of Social Needs Navigation on Utilization Among High Utilizers in a Large Integrated Health System: a Quasi-experimental Study. J Gen Intern Med 2019;34(11):2382–2389. doi: 10.1007/s11606-019-05123-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Hickey E, Phan M, Beck AF, Burkhardt MC, Klein MD. A Mixed-Methods Evaluation of a Novel Food Pantry in a Pediatric Primary Care Center. Clin Pediatr (Phila) 2020;59(3):278–284. doi: 10.1177/0009922819900960 [DOI] [PubMed] [Google Scholar]
- 30.Freij M, Dullabh P, Lewis S, Smith SR, Hovey L, Dhopeshwarkar R. Incorporating social determinants of health in electronic health records: Qualitative study of current practices among top vendors. JMIR Med Informatics 2019;7(2):e13849. doi: 10.2196/13849 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Cartier Y, Fichtenberg C, Gottlieb LM. Implementing community resource referral technology: Facilitators and barriers described by early adopters. Health Aff 2020;39(4):662–669. doi: 10.1377/hlthaff.2019.01588 [DOI] [PubMed] [Google Scholar]
- 32.Institute of Medicine (IOM). Crossing the Quality Chasm: A New Health System for the 21st Century; 2001. [PubMed] [Google Scholar]
- 33.Buitron De La Vega P, Losi S, Sprague Martinez L, et al. Implementing an EHR-based Screening and Referral System to Address Social Determinants of Health in Primary Care. Med Care 2019;57(6):S133–S139. doi: 10.1097/MLR.0000000000001029 [DOI] [PubMed] [Google Scholar]
- 34.Hager ER, Quigg AM, Black MM, et al. Development and Validity of a 2-Item Screen to Identify Families at Risk for Food Insecurity. Pediatrics 2010;126(1):e26–e32. doi: 10.1542/peds.2009-3146 [DOI] [PubMed] [Google Scholar]
- 35.Lett E, Asabor E, Beltrán S, Cannon AM, Arah OA. Conceptualizing, Contextualizing, and Operationalizing Race in Quantitative Health Sciences Research. Ann Fam Med 2022;20(2):157–163. doi: 10.1370/afm.2792 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.United States Census Bureau. INCOME IN THE PAST 12 MONTHS (IN 2019 INFLATION-ADJUSTED DOLLARS). 2019: ACS 5-Year Estimates Subject Tables Accessed May 19, 2021. https://data.census.gov/cedsci/table?q=median income&y=2019&tid=ACSST5Y2019.S1901&hidePreview=true [Google Scholar]
- 37.Bottino CJ, Rhodes ET, Kreatsoulas C, Cox JE, Fleegler EW. Food Insecurity Screening in Pediatric Primary Care: Can Offering Referrals Help Identify Families in Need? Acad Pediatr 2016;17(5):497–503. doi: 10.1016/j.acap.2016.10.006 [DOI] [PubMed] [Google Scholar]
- 38.Caspi CE, Davey C, Bliss Barsness C, et al. Needs and Preferences Among Food Pantry Clients. Prev Chronic Dis 2021;18(200531). doi: 10.5888/pcd18.200531 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Campbell JI, Sandora TJ, Haberer JE. A scoping review of paediatric latent tuberculosis infection care cascades: Initial steps are lacking. BMJ Glob Heal 2021;6(5):e004836. doi: 10.1136/bmjgh-2020-004836 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Epstein R, Sabharwal V, Wachman EM, et al. Perinatal Transmission of Hepatitis C Virus: Defining the Cascade of Care. J Pediatr 2018;203:34–40. doi: 10.1016/j.jpeds.2018.07.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.De Marchis EH, Alderwick H, Gottlieb LM. Do patients want help addressing social risks? J Am Board Fam Med 2020;33(2):170–175. doi: 10.3122/jabfm.2020.02.190309 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Appendix Figure 1. Food Insecurity Resource Connection Pathway
Note:
a 393 patients were excluded from the food pantry referral analysis due to having a positive screening result that occurred ≤7 days before the study’s end or having a food pantry referral or visit that pre-dated the screening encounter.
b All patients that were referred to the food pantry were included, regardless of screening status or result and regardless of their exclusion in the food pantry referral analysis. Of the 1,623 patients who were referred to the food pantry, 661 were referred ≤7 days after the 1st positive screen for food insecurity risk and before any subsequent clinic visits, 104 were referred but did not have documented screening results for food insecurity risk during the study period, 306 were referred but screened negative for food insecurity risk during the study period, 222 were referred from the 393 that were excluded from the food pantry referral analysis, and 330 were referred to the food pantry >7 days after 1st positive screen (e.g. after a subsequent screen or encounter).
c 106 patients were excluded from the food pantry visitation analysis due to having a food pantry visit that pre-dated the referral episode captured during the study period or having a food pantry referral placed ≤30 days before the study’s end.
Appendix Table 1. Patients screened for food insecurity risk stratified by screening positive for food insecurity risk (n=12,442)
Appendix Table 2. Patients who screened positive for food insecurity risk stratified by food pantry referral (n=3,571)
Appendix Table 3. Patients with food pantry referral stratified by food pantry visit (n=1,517)
Appendix Table 4. Responses to food insecurity questions among patients who screened positive for food insecurity risk (N=3,964)