Impact of Food Delivery and Health Coaching on Outcomes and Costs of Care: A Payer’s Perspective
Summary
Blue Cross and Blue Shield of North Carolina (Blue Cross NC) launched a 6-month food delivery and health coaching pilot among low-income members with type 2 diabetes as a proof-of-concept of the feasibility and potential efficacy of a payer-delivered food insecurity intervention. Participants received $60 in groceries delivered to their home twice per month. Health advisors provided weekly support to identify, refine, and advance toward achieving one or more health goals. Baseline and 3- and 6-month surveys assessed self-reported food security, body mass index (BMI), hemoglobin A1c levels, physical and mental health, and member satisfaction. Medical expenses were extracted from claims data for the 6 months before and 6 months after pilot enrollment. Blue Cross NC analyzed results for the 555 members who completed the program (Completes) and the 327 members who partially completed it (Partials). Participants were satisfied with food delivery frequency (81%) and amount of food in each box (82%). Partial participation was associated with decreases in food insecurity (25% to 18%), BMI (35 to 34 kg/m2) and percent obese (76% to 71%), and increases in the percent at or above U.S. average physical (57% to 74%) and mental (68% to 94%) health scores. Complete participation was associated with decreased food insecurity (38% to 20%), BMI (35 to 33 kg/m2) and percent obese (72% to 61%), and increases in the percent scoring at or above U.S. average physical (51% to 65%) and mental (70% to 80%) health measures. Complete participation was associated with a reduction of $139 per member per month (PMPM) in total medical costs and an increase of $8 PMPM in pharmacy costs, suggesting greater medication adherence. Partial participation was associated with a decrease of $10 PMPM in pharmacy costs and increases in all other cost types. Extrapolating the estimated reduction of $139 in PMPM with 6-month participation to all eligible members, the authors estimate the potential for medical expense reductions as $8.5 to $13.1 million annually. The results of this proof-of-concept are a first step in understanding the role payers could play in addressing unmet social needs that have profound consequences for health.
Introduction
In North Carolina, 12% of households are food insecure, which is slightly higher than the national average of 10.5%.1,2 Estimates for some groups, such as households with children, suggest that rates are twice as high.3 Recognizing the fundamental importance of food security, Blue Cross and Blue Shield of North Carolina (Blue Cross NC) developed a pilot program in partnership with a digital health coaching vendor (Pack Health, a Quest Diagnostics Company) to implement and evaluate a food delivery and health coaching (FDHC) program with fully insured commercial group and individual Affordable Care Act (ACA) members. The aim of the evaluation was to: (1) assess the impact of FDHC on food security, medical expenses, health outcomes, and member experience; and (2) affect and sustain, through health coaching support, individual health behavior change.
Food security is a predominant unmet social need resulting from poverty, unemployment, and low wages, exacerbated by a lack of education, inadequate public supports, and environmental- and community-based factors. Food insecurity negatively influences physical and mental health, access to care, health care utilization, and health outcomes.4 Although the rate of U.S. and North Carolina food-insecure households has improved since 2018, progress is slow, and any long-lasting impacts of the Covid-19 pandemic may not yet be fully realized.1,2
Individuals who are food insecure are at increased risk for chronic diseases and other negative health outcomes.5-7 Approximately 12% of people diagnosed with type 2 diabetes report being food insecure, with the prevalence among low-income people diagnosed with type 2 diabetes possibly more than twice as high.8,9 Food insecurity is particularly problematic for individuals with type 2 diabetes, a condition in which diet plays a significant role.9-11
In addition to this greater prevalence of chronic illness, the use of health care services and magnitude of medical expenditures differ between those with and without food insecurity. Adults in food-insecure households are more likely to forego needed medications.12,13 Those experiencing food insecurity have a greater number of ED visits, inpatient hospitalizations, and, when hospitalized, number of inpatient days.14 In addition, individuals with food insecurity are more likely to use mental health services and also more likely to have no usual source of care.15 Consequently, annual medical costs may be as much as 25%–30% higher for individuals experiencing food insecurity.16-19
Common approaches to addressing food insecurity include food referral programs and providing food vouchers.20 However, those who experience food insecurity may also have additional socioeconomic barriers to healthy and sufficient food access.21-23 A promising approach to mitigating these barriers is direct FDHC. Evidence for the effectiveness of these programs is primarily being collected in the context of large health care systems. A fresh fruit and vegetable delivery and nutrition education intervention implemented in the Children’s National Hospital, Washington DC, USA, was associated with a decrease in high food insecurity, from 32% to 7%.24 Through its Fresh Food Farmacy program, Geisinger, Danville, PA, USA, is providing food sufficient to make 10 meals per week to its patients with diabetes.25 Results from the pilot program suggest $16,000 to $24,000 per year in medical savings.26 The Community Servings: Food as Medicine for Diabetes clinical trial delivered 10 medically tailored meals (MTMs) per week for 12 weeks to patients with type 2 diabetes.27 The estimated program impact was decreased food insecurity, less hypoglycemia, and fewer days during which mental health interfered with quality of life. Lastly, Kaiser Permanente is conducting several clinical trials to evaluate the effectiveness of MTMs with and without nutrition counseling.28-30 Published results from one of Kaiser Permanente’s MTM clinical trials found no significant effects; however, the authors note several limitations, including a lack of power.31
Blue Cross NC implemented a proof-of-concept pilot with members with documented type 2 diabetes to determine the feasibility and effectiveness of a payer-implemented food-insecurity intervention and to establish if a business case exists for scaling up such a program. To our knowledge, a food-delivery program aimed at reducing food insecurity and health care costs has never been implemented by a payer that is not part of an integrated delivery system such as Kaiser or Geisinger. Most commonly, efforts to address food insecurity through food delivery have been driven by government agencies, community groups, or health care systems and have focused on Medicare or Medicaid populations.20,32 Information about the effectiveness of these programs is limited: programs for which evaluation results are available have small sample sizes, and few have included the impact on medical costs.
Blue Cross NC partnered with Pack Health to implement a 6-month FDHC program to active Blue Cross NC members identified as being at high risk for food insecurity.
Methods
This work is part of ongoing Blue Cross NC operations and program evaluation and not classified as research.
Intervention
Blue Cross NC partnered with Pack Health to implement a 6-month FDHC program to active Blue Cross NC members identified as being at high risk for food insecurity. Pack Health adapted their 3-month coaching program by including the delivery of healthy food boxes and expanding the program to 6 months. With the addition of food boxes to health coaching, the intervention addresses both the lifestyle support and access to healthy food that are necessary for type 2 diabetes management. Blue Cross NC launched the program in December 2020, with the first invitations sent in January 2021. The program consisted of two core components — health coaching and food box delivery.
Health Coaching
The coaching component involved at least weekly communication between each program participant and an individually assigned health advisor (HA). HAs were national board-certified health and wellness coaches employed by Pack Health. The HAs engaged participants to understand individual specific food needs, situation, and cultural preferences and to identify one or more health goals (e.g., changing eating habits, increasing exercise). During weekly contacts, HAs supported participants in refining goals and developing healthy choices aimed at advancing toward achieving the identified goals. HAs and participants communicated using a variety of modes, including telephone, email, and text messaging.
Food Box Delivery
Twice per month, participants received $60 worth of groceries delivered directly to their home. The amount of food provided by the program was based on average Food and Nutrition Service/Supplemental Nutrition Assistance Program benefits as well as amounts reported in the literature describing other food programs. Based on the needs of — and conversation with — each member, HAs chose from three to four food boxes that were designed on the basis of U.S. Department of Agriculture Healthy Plate guidelines. Two vendors prepared and delivered food boxes to participants during the pilot. Due to logistical issues, the initial vendor was replaced at the midpoint of the pilot. The initial vendor included fresh produce in its food boxes, whereas the second vendor did not offer fresh produce. A sample food box included beans, salmon, rice, crackers, carrots, applesauce, pasta, sauce, crushed tomatoes, and milk. Each delivery was accompanied by easy-to-make recipes.
Eligibility
Potential participants were identified from Blue Cross NC membership data. Eligible individuals must: (1) have an active Blue Cross NC ACA or employer group policy; (2) have at least one claim indicating type 2 diabetes since 2019; (3) be at least 18 years of age; and (4) be positive on a proxy measure of food insecurity. Food insecurity for ACA members was determined by their cost-sharing reduction subsidy level. We used 93% cost-sharing reduction, approximately 100%–150% of the federal poverty limit (FPL), to indicate food insecurity.33 For employer group members, we used 100%–150% FPL to indicate food insecurity. FPL was determined by using household size and estimated household income fields from the third-party data.
Recruitment and Enrollment
Using the above inclusion criteria, Blue Cross NC created an eligibility file (Outreach) and messaging to introduce the program. Eligible members were sent an email or postcard with an invitation to participate. Postcards were sent to members who had chosen direct mail as their preferred means of communication with Blue Cross NC or who did not have a valid email address. Both email and direct mail invitations provided a link to a Pack Health/Blue Cross NC web page as well as a telephone number to contact Pack Health to enroll. Up to two reminder messages — 2 and 4 weeks after the initial message — were sent to eligible members not responding to the invitation. To encourage enrollment, Blue Cross NC followed up with eligible members who had visited the web page but did not enroll. Blue Cross NC case managers and social workers were also asked to identify members with type 2 diabetes, screen them for food insecurity, and recruit them to participate (Referrals). If interested, members were directed to the Pack Health/Blue Cross NC enrollment web page or Pack Health telephone number to enroll. Enrollment via Pack Health was cross-checked with referrals from case managers and social workers to avoid enrolling any member more than once. The 6-month enrollment period began January 2021.
The purpose of the proof-of-concept pilot was to determine the feasibility and scalability of a payer-implemented FDHC intervention. Thus, all eligible members were invited to join until the budgeted number of participants was reached (N = 1,250). Because we reached our enrollment target during the fourth month of enrollment, the number of permitted participants was increased to 2,200. Enrollment closed on June 30, 2021.
Eligible individuals must: (1) have an active Blue Cross NC ACA or employer group policy; (2) have at least one claim indicating type 2 diabetes since 2019; (3) be at least 18 years of age; and (4) be positive on a proxy measure of food insecurity.
At completion of the program, eligibility file and referral members were categorized into two participation groups: (1) participants who received 12 food boxes and 6 months of coaching (Completes); and (2) participants who received 6–11 food boxes and 3–5 months of coaching but did not participate for the entire period (Partials). To provide context, we summarized descriptive characteristics and medical claims costs for eligible members who did not participate in the FDHC program (nonparticipants).
Implementation Outcomes
Several outcomes were tracked assessing the feasibility of program implementation, including enrollment, engagement (i.e., at least 3 months of participation or 6 food boxes), touch points (i.e., number of HA-initiated contacts), and number of food boxes delivered. After the first two food box deliveries and again at 6 months, participant satisfaction with the program was assessed. Using a 5-point scale (1 = extremely dissatisfied to 5 = extremely satisfied), participants rated their satisfaction with the frequency of delivery, recipes, and quality, amount, and variety of food in each delivery. Participants also rated the overall food delivery component using the same 5-point scale. At baseline and 3 and 6 months, participants rated their satisfaction with Blue Cross NC as their insurer using a 5-point scale (1 = very dissatisfied to 5 = very satisfied). Lastly, participants were asked to rate their intention to renew their Blue Cross NC after participating in the FDHC program using a 5-point scale (1 = extremely likely to 5 = extremely unlikely).
Patient-Reported Outcomes
A set of patient-reported outcomes measures was assessed via the baseline, 3-month, and 6-month surveys.
Patient-Reported Outcomes Measurement Information System Global-10
The Patient-Reported Outcomes Measurement Information System (PROMIS) is a 10-item validated measure assessing general domains of health and functioning.34 Global physical health and mental health items were scored. Scores range from 4 to 20, with higher scores indicating better health.
Household Food Security Survey Module: 6-Item Short Form
The U.S. Household Food Security Survey Module 6-item short form is a validated short form of the U.S. Household Food Security Survey Module.35 Items with responses of “Yes,” “Often,” “Sometimes,” “Almost every month,” or “Some months” are scored as “1.” All other responses are scored as “0.” Scores for all 10 items are summed, and food security status is assigned as: high or marginal (0–1), low (2–4), and very low (5–6). Participants were categorized as food insecure if their score was low or very low.
Hemoglobin A1c
Participants were asked to report their hemoglobin A1c (HbA1c) level, which is the amount of blood sugar attached to hemoglobin, from their most recent provider visit. Testing is recommended at least two times per year for individuals with type 2 diabetes. We assumed participants would have at least two tests — one near enrollment (baseline) and a second closer to completion of the 6-month program. The American Association of Clinical Endocrinology recommends most individuals with diabetes aim for HbA1c levels less than 6.5%.36 We therefore assessed the impact of participation on maintaining or gaining control of diabetes as HbA1c levels below 6.5%.
Medical Costs
We examined medical costs to further inform our understanding of the efficacy of a payer-implemented FDHC program. Learnings from these results will assist in establishing hypotheses and directions for future work. Medical costs were extracted from Blue Cross NC claims data for the 6-month period before and the 6-month period after the participant’s index date. The index date was defined as the date of enrollment for Completes and Partials. For contacted nonparticipants, the index date was calculated as the first email/direct mail date plus 18, the median number of days from first contact to enrollment for participants. For nonparticipants who were not contacted (i.e., email bounced back or they opted out), the index date was calculated as the eligibility file creation date plus 36, the median number of days from the date added to the file to enrollment for participants. Medical costs assessed include the total cost of care (TCOC), pharmacy, inpatient emergency (IP ER), inpatient nonemergency (IP Non-ER), ED, outpatient nonemergent (OP Non-ER), and nonfacility professional (PROF) claims.
All eligible members were invited to join until the budgeted number of participants was reached.
Evaluation Methods
We assessed the feasibility of implementation, efficacy, and potential effectiveness of the FDHC program on food insecurity, relevant health outcomes, and medical costs of Blue Cross NC members at high risk of food insecurity. Enrollment progress was tracked, and baseline, 3-month, and 6-month follow-up data were collected by Pack Health (Table 1). At baseline and follow-up, HAs administered surveys collecting self-reported health outcomes and quality of life measures. Feedback from participants assessing implementation was collected after the first two food boxes were delivered, at 6 weeks, and again at 12 weeks.
Table 1
| Pilot Phase | Baseline | Weekly | Biweekly | 1 Month | 3 Months | 6 Months |
|---|---|---|---|---|---|---|
| Intervention implementation | ||||||
| Coaching | X | |||||
| Food box delivery | X | |||||
| Data collection | ||||||
| Participant feedback | X | X | X | |||
| Patient-reported outcomes | X | X | X | |||
| Medical costs | X |
Implementation and Data Collection Interval
Source: Blue Cross and Blue Shield North Carolina
Cost and utilization data were extracted from Blue Cross NC medical and pharmacy claims data covering the 6-month period before and the 6-month period after enrollment for all members identified in the eligibility file. Member demographic information was extracted by using third-party commercially available data and Blue Cross NC member information.
Proof-of-Concept
Program Implementation
Email/direct mail invitations were sent to 20,537 eligible members inviting them to enroll in the program (Table 2). Of all eligible members contacted, 1,439 (7%) enrolled in the program (Outreach). A total of 208 members referred by a case manager or social worker (Referrals) enrolled in the FDHC program. The number of members whom case managers and social workers attempted to recruit was not tracked, and the rate of enrollment among referrals is unknown.
Table 2
| Disposition | Total | Self-Referral (Eligibility File) | Referral (Case Managers or Social Workers) | |||
|---|---|---|---|---|---|---|
| N | Percent | n | Percent | n | Percent | |
| Total | 20,537 | 100.0 | ||||
| Contacted | 16,002 | 77.9 | ||||
| Enrolled | 1,647 | 100.0 | 1,439 | 7.0 | 208 | |
| Not engaged | 411 | 25.0 | 368 | 22.6 | 43 | 20.7 |
| Engaged | 1,275 | 77.4 | 1,114 | 77.7 | 161 | 77.4 |
| Completes | 648 | 50.8 | 565 | 50.7 | 83 | 51.6 |
| Partials | 392 | 30.7 | 342 | 30.7 | 50 | 31.1 |
| Dropouts | 235 | 18.4 | 207 | 18.6 | 28 | 17.4 |
| Controls | 19,513 | 19,423 | 94.6 | 90 | 43.3 | |
Disposition of Outreach to and Recruitment of Eligible Members
Source: Blue Cross and Blue Shield North Carolina
Once enrolled, similar percentages of Referrals and Outreach members engaged with an HA for at least one telephone call (77% vs. 78%) and completed the program (i.e., 12 food boxes over 6 months) (52% vs. 51%). The percentage of members who received 6–11 food boxes was similar among Referrals and Outreach members (31%).
A total of 235 (18%) members dropped out of the program before 3 months. The dropout rate was lower among Referrals (17%) than among Outreach members (19%). The most frequent reason for dropping a participant from the program was an inability to make contact (49%), followed by a loss of interest (20%) and a lack of time (9%). Other reasons included dissatisfaction with the program (4%), poor health (2%), and misunderstanding the program’s intent (2%).
For the remainder of this evaluation, we combined participants according to type of participation, as no further differences were observed between Outreach and Referrals. The average length of participation was approximately 26 weeks for Completes and 19 weeks for Partials. Completes and Partials engaged in an average of one telephone call per week and two emails per month. Completes exchanged seven text messages per week, whereas Partials exchanged six text messages per week.
During implementation, it was necessary to switch grocery package vendors due to participant dissatisfaction and confusion with vendor delivery (e.g., multiple boxes sent rather than one box). Participants were slightly less satisfied with the food boxes after the switch (81% before vs. 79% after). Overall, however, 81% of FDHC participants reporting being satisfied with the frequency of food delivery, and 82% were satisfied with the amount of food in each box. Participants were less satisfied with the recipes provided (70%) and the quality (66%) and variety (64%) of the food in each delivery.
Characteristics and Health Status
Differences in demographic and health characteristics were observed between eligible members who participated in the FDHC program and those who did not participate (Table 3). Compared with nonparticipants, a higher percentage of participants (Completes and Partials) were female, were African American, and had completed more than a high school education. Compared with nonparticipants, program participants had longer active plan tenures of approximately 2 months. Partials had the highest average household income and Completes had the lowest, a difference of approximately $4,000. The total number of household members did not differ across program participation; however, Partials lived in households with more children per household on average.
Table 3
| Characteristic | Completes | Partials | Nonparticipants | P** |
|---|---|---|---|---|
| Total no. | 555 | 327 | 14,061 | |
| Sex, % | <.001 | |||
| Female | 76.2 | 72.7 | 54.2 | |
| Male | 23.8 | 27.3 | 45.8 | |
| Age, mean (SD), y | 54.5 (8.3) | 53.0 (9.8) | 53.8 (9.8) | .273 |
| Plan tenure since enrollment, mean (SD), d | 314 (50) | 305 (52) | 314 (49) | .002 |
| Race/ethnicity, % | <.001 | |||
| European | 52.6 | 50.5 | 53.3 | |
| African American | 40.6 | 40.6 | 28.8 | |
| Hispanic | 4.3 | 5.6 | 9.1 | |
| Other | 2.5 | 3.3 | 8.6 | |
| Education, % | .071 | |||
| Less than high school | 12.1 | 11.5 | 14.8 | |
| High school | 39.2 | 37.2 | 38.5 | |
| More than high school | 48.8 | 51.3 | 46.6 | |
| Household income, mean (SD), $ | 55,982 (43,937) | 60,661 (44,723) | 57,053 (47,147) | .030 |
| Household members, mean (SD) | 3.2 (2.1) | 3.4 (2.2) | 3.3 (2.1) | .570 |
| Adults | 2.7 (1.6) | 2.7 (1.6) | 2.8 (1.7) | .069 |
| Children | 0.5 (1.0) | 0.7 (1.2) | 0.5 (0.9) | <.001 |
| Single-member households, % | 26.1 | 27.3 | 24.5 | .297 |
| Health conditions, mean (SD) | 3.8 (1.8) | 3.7 (1.7) | 3.3 (1.6) | <.001 |
| Health conditions, % | ||||
| Hypertension | 77.8 | 78.8 | 73.5 | .013 |
| Hyperlipidemia | 74.9 | 74.3 | 71.8 | .203 |
| Obesity | 61.9 | 66.2 | 50.8 | <.001 |
| Osteoarthritis | 37.7 | 35.0 | 26.3 | <.001 |
| Cataract | 17.4 | 17.7 | 13.7 | .008 |
| Asthma | 15.9 | 13.5 | 9.0 | <.001 |
| Coronary heart disease | 13.4 | 13.2 | 13.2 | .992 |
| Cancer | 10.5 | 10.3 | 8.8 | .261 |
| Chronic kidney disease | 10.2 | 9.3 | 8.6 | .404 |
| Chronic obstructive pulmonary disease | 7.5 | 7.4 | 6.0 | .239 |
| Walking assistance | 7.1 | 6.8 | 4.5 | .004 |
| Congestive heart failure | 6.1 | 8.4 | 4.7 | .004 |
| Arthritis | 4.2 | 1.3 | 2.0 | .002 |
| Stroke | 5.0 | 3.5 | 3.5 | .179 |
| Hearing loss | 3.1 | 1.6 | 2.6 | .436 |
| Care provider dependency | 2.3 | 2.3 | 1.6 | .373 |
| Supplemental oxygen | 1.7 | 1.3 | 0.8 | .049 |
| Smoker | 0.8 | 0.6 | 1.6 | .127 |
Baseline Characteristics of Participants by Participation Group*
SD = standard deviation. *Excludes outliers (more than $125,000 in costs in the 6-month postperiod). **Analysis of variance tests were conducted on continuous variables; χ2 with Fisher exact tests, when appropriate, were conducted on categorical variables. Bolded P-values indicate a statistically significant difference at P < 0.05. Source: Blue Cross and Blue Shield North Carolina
We assessed the impact of participation on maintaining or gaining control of diabetes as HbA1c levels below 6.5%.
With respect to health, Completes and Partials have more documented health challenges compared with nonparticipants. Although the difference is less than one condition on average, this does suggest the potential for the FDHC program to target members at increased risk for negative health outcomes. Hypertension, obesity, osteoarthritis, cataract, asthma, and congestive heart failure were documented less frequently in the year before the start of the program (P < .05) for nonparticipants. The need of walking assistance and supplemental oxygen was also documented less frequently for nonparticipants.
Outcomes
We only present self-reported health outcomes for program participants, as the evaluation did not attempt to survey nonparticipating members. Response rates to the baseline survey were similar for Completes (83%) and Partials (82%), whereas the 3-month survey response rate was higher for Completes (89% vs. 56%). For the 6-month follow-up survey, the response rate for Completes was 41% compared with 4% for Partials. The relative number of Partials responding to the 3- and 6-month surveys was low; therefore, we do not discuss the 6-month results of this group, nor do we attempt to draw meaningful conclusions regarding differences across participation groups.
Food Insecurity
At baseline, 42% of Completes and 37% of Partials reported either low or very low food security. By the 3-month follow-up, the percentage of Completes and Partials reporting food insecurity declined to 22% and 17%, respectively. By the 6-month follow-up, reports of food insecurity among Completes declined further to 19%.
Body Mass Index
From baseline to the 3-month survey, a slight increase in self-reported body mass index (BMI) was observed among Completes (34 to 35 kg/m2), and, for Partials, a slight decrease was observed (35 to 34 kg/m2). We did observe a modest decrease in the percent obese for Partials but not for Completes at the 3-month follow-up. By 6 months, decreases in both self-reported BMI (−3%) and the percent obese (−10%) were observed among Completes.
Hemoglobin A1c
At baseline, 37% of Completes and 34% of Partials reported baseline HbA1c levels less than 6.5%. The percentage of Partials reporting HbA1c levels less than 6.5% at 3 months increased to 40%, whereas the percentage of Completes reporting controlled diabetes at 3 months increased to 44%. At 6 months, the percentage of Completes with controlled diabetes declined to 42%; however, the 6-month rate was higher than baseline and represents an overall increase of 13%. Almost 86% of Completes and 82% of Partials with baseline HbA1c levels less than 6.5% reported maintaining control of their diabetes at 3 months. At the 6-month follow-up, 87% of Completes with baseline control of their diabetes reported HbA1c levels less than 6.5%.
Mental and Physical Health (PROMIS Global-10)
Average baseline Mental Health scores (MHS) were similar between Completes (mean [standard deviation (SD)], 44 [8]) and Partials (mean [SD], 44 [7]). A slightly higher percentage of Completes scored equal to or above the U.S. adult average of 50 (68% vs. 67%). By the 3-month follow-up, the percentage above average MHS increased among both Completes and Partials (to 81% and 92%, respectively). At 6 months, the average MHS and percent at or above average declined slightly for Completes but were still higher than baseline.
Average baseline Physical Health scores (PHS) were slightly higher among Partials (mean [SD], 43 [7]) compared with Completes (mean [SD], 41 [7]). Likewise, the percentage of Partials scoring at or above the U.S. adult average of 50 was higher than that of Completes (57% and 51%, respectively). By the 3-month follow-up, we observed increases in the average PHS among Partials and Completes of 10% and 8%. Likewise, the percent scoring at or above 50 increased among Completes (69%) and among Partials (72%). Similar to MHS, the average PHS and percent above the average for Completes declined at 6 months but remained higher than baseline.
Medical costs were extracted from Blue Cross NC claims data for the 6-month period before and 6-month period after the participant’s index date.
Member Experience
Continuous Membership
On average, Completes (mean [SD], 314 [50] days) and Partials (mean [SD], 305 [52] days) had active Blue Cross NC plans for similar tenures after enrolling in the FDHC program plan.
Member Satisfaction
On a scale of 1–5, satisfaction with their Blue Cross NC plan was relatively high and similar at baseline across program participants: Completes (mean [SD], 4 [1]) and Partials (mean [SD], 4 [1]) (Table 4). Satisfaction within both groups increased slightly between baseline and the 3-month follow-up. Self-reported satisfaction with Blue Cross NC among Completes at the 6-month follow-up did not change from the average satisfaction reported at the 3-month follow-up. Among Completes responding to the 6-month survey, 79% reported that they were “very likely” to renew their health plan with Blue Cross NC after participating in the FDHC program.
Table 4
| Completes | Partials | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Baseline | 3 Months | 6 Months | Baseline | 3 Months | 6 Months | |||||||
| n* | Value | n* | Value | n* | Value | n* | Value | n* | Value | n* | Value | |
| Food insecurity score, mean (SD) | 364 | 3 (3) | 337 | 1 (2) | 131 | 1 (2) | 227 | 3 (3) | 90 | 1 (2) | 4 | 0.3 (0.5) |
| Food insecure** | 42% | 22% | 19% | 37% | 17% | 0.0% | ||||||
| BMI, mean (SD) | 432 | 34 (7) | 390 | 35 (8) | 228 | 33 (7) | 259 | 35 (7) | 183 | 34 (6) | 13 | 33 (5) |
| Obese (BMI ≥30 kg/m2) | 69% | 70% | 63% | 73% | 73% | 69% | ||||||
| HbA1c level, mean (SD) | 331 | 7 (2) | 236 | 7 (2) | 84 | 7 (1) | 178 | 7 (2) | 73 | 7 (2) | 4 | 7 (1) |
| HbA1c <6.5% | 37% | 44% | 42% | 38% | 40% | 25% | ||||||
| PROMIS MHS, mean (SD)# | 458 | 44 (8) | 344 | 47 (8) | 131 | 47 (8) | 266 | 44 (7) | 96 | 48 (7) | 5 | 46 (10) |
| MHS ≥50 | 68% | 81% | 79% | 67% | 92% | 8% | ||||||
| PROMIS PHS, mean (SD)§ | 461 | 42 (7) | 343 | 45 (8) | 133 | 45 (8) | 268 | 42 (7) | 97 | 47 (8) | 5 | 47 (14) |
| PHS ≥50 | 501% | 69% | 63% | 55% | 72% | 8% | ||||||
| BCNC satisfaction, mean (SD) | 349 | 4 (1) | 334 | 5 (1) | 75 | 5 (1) | 213 | 4 (1) | 83 | 4 (1) | 5 | 4 (2) |
Self-Reported Health and Satisfaction Outcomes by Program Participation and Follow-up Time Point
SD = standard deviation, BMI = body mass index, HbA1c = hemoglobin A1c, PROMIS = Patient-Reported Outcomes Measurement Information System, MHS = Mental Health score, PHS = Physical Health score, BCNC = Blue Cross and Blue Shield of North Carolina. *Responses to the specific item. **Responses included from participants completing baseline and 3- and/or 6-month surveys. #Food-insecure score of ≥3 (low/very low food security). §PROMIS Global-10 PHS and MHS of 50 are considered average for a U.S. adult. Source: Blue Cross and Blue Shield of North Carolina
Health Care Costs
A disproportionate number of program participants had zero costs for some categories of medical costs. This affects our ability to reliably assess the impact of participation on these costs. Therefore, we dropped the following outcomes from subsequent discussion: IP ER (n = 944 with $0), IP Non-ER (n = 604 with $0), OP Non-ER (n = 3,597 with $0), and Nonemergent ER (n = 1,216 with $0).
Average preperiod costs across all categories were highest among Completes and lowest among nonparticipants. Except for pharmacy costs, average per-member per-month (PMPM) costs across types were lower in the postperiod compared with the preperiod for Completes (Table 5). Average costs among Completes were lower in the postperiod for TCOC excluding pharmacy (−$216), PROF (−$18), and OP Non-ER (−$64) claims. Average pharmacy costs among Completes increased by $55 PMPM. Conversely, average costs across all types increased among Partials in the postperiod. Average costs among Partials were higher in the postperiod for TCOC, excluding pharmacy ($56), PROF ($21), OP Non-ER ($49), and pharmacy ($75).
Table 5
| TCOC | TCOC (excluding Rx) | Rx | PROF | OP Non-ER | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Pre- | Post- | Pre- | Post- | Pre- | Post- | Pre- | Post- | Pre- | Post- | |
| Program participation | |||||||||||
| Completes | 545 | 2,280 | 2,135 | 1,495 | 1,279 | 785 | 855* | 520 | 502 | 444 | 380 |
| Partials | 322 | 2,084 | 2,214 | 1,345 | 1,401 | 738 | 813 | 444 | 465 | 443 | 492 |
| Nonparticipants | 13,777 | 1,571 | 1,563 | 993 | 953 | 577 | 609* | 345 | 328* | 316 | 312 |
Average Per-Member Per-Month Costs Pre- and Postenrollment by Program Participation, Food Security, and Cost Type
TCOC = total cost of care, Rx = pharmacy, PROF = nonfacility professional, OP Non-ER = outpatient nonemergency. *Pre-enrollment to postenrollment difference is significant, P < .05, based on paired-sample t tests. Source: Blue Cross and Blue Shield of North Carolina
Further investigation of the differential changes in costs between participants who completed the program and those who partially completed was not feasible given the limitations of the evaluation design. Testing the impact of different levels and duration of the program (e.g., one food box rather than two per month or 12 months rather than 6 months) may inform our understanding of the impact of FDHC on members’ costs. We did, however, walk through the exercise of estimating the impact of the program accounting for lack of randomization and the panel nature of the data using difference-in-difference and fixed-effects estimation. Results of this exercise suggest a directionally positive impact of participation for the full 6 months (Appendix).
Discussion
The expansion of food-aid programs during Covid-19 may have prevented more U.S. households from falling into food insecurity. Although the rate of food insecurity nationally remained unchanged in 2020 (10.5%) from 2019, the rate in North Carolina declined from 13.5% in 2019 to 12.0% in 2020.1,2 However, emergency aid programs have expired, and food prices have been rising. To address food insecurity among its members, Blue Cross NC developed and implemented a proof-of-concept FDHC program, the first such attempt by a payer.20
At baseline, 42% of Completes and 37% of Partials reported either low or very low food security. By the 3-month follow-up, the percentage of Completes and Partials reporting food insecurity declined to 22% and 17%, respectively.
Completion of the 6-month program was associated with directionally meaningful impacts on food security, self-reported health, and medical expenses. We observed decreases in self-reported food insecurity and obesity rates, and participation in the program may have helped approximately one-fifth of participants with uncontrolled diabetes gain control of their disease by the 3-month follow-up. Importantly, 87% of participants reported maintaining control of their diabetes and 19% reported gaining control over the 6-month evaluation period. Participants also reported improvements in overall physical and mental health after 3 months of participation. Goals set by participants included control of diabetes without medication, increasing exercise, and improving eating habits. HAs reported that several participating members shared that they were monitoring blood sugar levels more often, exercising more frequently, and improving their food choices. These observations suggest that the FHDC program may have a positive impact on members’ HbA1c levels and exercise and eating habits. The current work was limited by both HbA1c and BMI being self-reported. We are evaluating more reliable and systematic means of collecting these data, including accessing medical records information or direct testing.
Because we are not a research institution, we do not typically do repeated, extensive surveying of our members. Through this proof-of-concept pilot, we identified several improvement opportunities for future model tests. The survey data collected were self-reported and, as such, may be subject to measurement error. Not all desired outcome measures were available via electronic health and claims databases; however, for those that are available, we are evaluating methods to collect information directly from existing data. Furthermore, we had no way to collect these outcome measures among nonparticipants (beyond comparison of expenditures using claims data), and, therefore, this study lacked a full control group. Without outcomes among nonparticipants, we are unable to ascribe the change in these measures to participation in the program. The directionally positive changes observed, however, should be viewed as encouraging.
Enrollment in the program was low, although higher than the average enrollment for Blue Cross NC programs, and responses to the 3- and 6-month surveys were also low. Reasons for dropping out were documented, but it is also important to understand members’ reasons for electing not to participate in the first place. In other pilot work, Blue Cross NC is examining the messaging sent to eligible members and testing different approaches, including text messaging and reminders, as well as documenting reasons for lack of interest. Blue Cross NC is also investigating approaches, such as focus groups, for understanding members’ preferences so that programs are designed to better meet member needs.
In the 6 months’ postenrollment, we observed decreases in total costs of care among nonparticipants and increases in total costs among Partials. We hypothesize that the costs among Partials, similar to Completes with respect to health conditions, may have been serving as a more accurate counterfactual to the experience of the Completes. The current evaluation included twelve $60 food boxes and 6 months of coaching. Evaluating the impact of fewer boxes or less intense coaching on health outcomes and medical costs will provide additional evidence for designing the program for the market and continuing to build the business case.
Our efforts to reach members most in need may also be limited by the method used to identify eligible members. Our proxy for food insecurity (i.e., FPL) may not be sensitive enough given that less than one-half of enrolled members reported experiencing food insecurity. Self-reported food insecurity is not systematically captured in claims data and, except for the referral participants, needed to be determined through evaluation-related data collection. Blue Cross NC, along with many other health entities, is working to integrate screening for food insecurity and other nonclinical drivers of health to inform the types of products to offer members and to have the capability of assessing the impact on health outcomes and costs.
At baseline, 37% of Completes and 34% of Partials reported baseline HbA1c levels less than 6.5%. The percentage of Partials reporting HbA1c levels less than 6.5% at 3 months increased to 40%, whereas the percentage of Completes reporting controlled diabetes at 3 months increased to 44%.
Lastly, this evaluation was limited to assessing impacts over a 6-month period. To continue evaluating the business case, it is important to assess whether improvements in health and reductions in costs are sustained beyond the initial evaluation period and whether, with increased participation, these results are generalizable to members in other lines of business. Analyses are planned to evaluate the long-term impact of the FDHC program.
The design of the pilot was informed by the literature supporting the effects of providing food for those affected by food insecurity as well as the impact of coaching on health outcomes. Because all program participants received health coaching, the design makes it impossible to disentangle the effects of the food delivery component from those of the health coaching. Interaction with the HA was regular, with program participants and HAs communicating at least once per week by telephone and six to seven times per week via text message. The potential impact of the health coaching component was evident in the feedback shared through the HAs:
“Member was struggling to manage her diabetes but since starting the program 2 weeks ago, she started checking her blood sugar levels more regularly, taking her medication as prescribe[d], choosing healthy options, and waking at 6 a.m. to walk with her husband. She is so incredibly grateful for our help and credits the weekly check-in and text messages for her success.”
“This member thin[k]s very highly of the program! She said it really helps her to have plenty of food in the house for her and her daughter, and she appreciates the accountability working toward her health goals.”
Blue Cross NC continues to develop new programs and plans for implementation and evaluation of these programs using the lessons learned. The evaluation plan for the FDHC proof-of-concept did not include a qualitative component that would have provided more detailed insight into members’ experiences, particularly into the reasons for electing not to participate. Developing partnerships across a large, nonresearch organization is essential to successfully implementing programs designed to address nonclinical drivers of health. There is a substantial literature documenting the association of food insecurity and health outcomes. Likewise, the impact of interventions such as MTMs, produce prescriptions, and food delivery on food insecurity have been documented. Within a food-as-medicine framework, these interventions have been delivered through health care providers, and “treatment” is funded through the health care system, the government, or philanthropy.6 Such efforts require more sustainable financing models, however. Payers are in a unique position to integrate food-as-medicine interventions into more sustainable financial models. As a payer, the business case needs to be made for offering products and services that address the food and nutritional needs of its members. The results of this evaluation are a first step in understanding the role that payers could play in addressing unmet social needs that have profound consequences for health.
Related Content: Watch our related March 9 event, “Social Needs and SDOH: Impact on Patients & Communities.”
Notes
We thank the Pack Health data team, especially Stephen Burton and Vanessia Tran; the HAs for the support and encouragement given to our members; the Blue Cross NC members for their participation and valuable input; and Dr. Tunde Sotunde for his support and thought leadership.
John R. Lumpkin, Lori H. Taylor, Aiko Hattori, and Jenefer M. Jedele have nothing to disclose.
Appendix
References
1.
Feeding America. Food Insecurity Among Overall (All Ages) Population in North Carolina. 2020. Accessed September 28, 2022. https://map.feedingamerica.org/county/2020/overall/north-carolina.
2.
Coleman-Jensen A, Rabbitt MP, Gregory CA, Singh A. Household Food Security in the United States in 2020. Economic Research Report No. ERR-298. September 2021. Accessed September 20, 2022. https://www.ers.usda.gov/publications/pub-details/?pubid=102075.
3.
Berner M. The State of Food Insecurity in North Carolina During the Pandemic. Carolina Across 100, University of North Carolina. December 21, 2021. Accessed September 20, 2022. https://carolinaacross100.unc.edu/the-state-of-food-insecurity-in-north-carolina-during-the-pandemic/.
4.
Serchen J, Atiq O, Hilden D. Health and Public Policy Committee of the American College of Physicians. Strengthening food and nutrition security to promote public health in the United States: a position paper from the American College of Physicians. Ann Intern Med 2022;175:1170-1
5.
Laraia BA. Food insecurity and chronic disease. Adv Nutr 2013;4:203-12
6.
Aspen Institute. Advancing Health Through Food Security: A Multi-Sector Approach to Address the Disease Burden and Costs of U.S. Food Insecurity on Our Health System. February 2016. Accessed September 20, 2022. https://www.aspeninstitute.org/wp-content/uploads/2021/07/Advancing-Health-Through-Food-Security.pdf.
7.
Gregory CA, Coleman-Jensen A. Food Insecurity, Chronic Disease, and Health Among Working Age Adults. Economic Research Report No. ERR-235. July 2017. Accessed September 27, 2022. https://www.ers.usda.gov/publications/pub-details/?pubid=84466.
8.
Walker RJ, Garacci E, Ozieh M, Egede LE. Food insecurity and glycemic control in individuals with diagnosed and undiagnosed diabetes in the United States. Prim Care Diabetes 2021;15:813-8
9.
Seligman HK, Jacobs EA, López A, Tschann J, Fernandez A. Food insecurity and glycemic control among low-income patients with type 2 diabetes. Diabetes Care 2012;35:233-8
10.
Kirby JB, Bernard D, Liang L. The prevalence of food insecurity is highest among Americans for whom diet is most critical to health. Diabetes Care 2021;44:e131-2
11.
Seligman HK, Laraia BA, Kushel MB. Food insecurity is associated with chronic disease among low-income NHANES participants [published correction appears in J Nutr 2011;141:542]. J Nutr 2010;140:304-10
12.
Berkowitz SA, Seligman HK, Choudhry NK. Treat or eat: food insecurity, cost-related medication underuse, and unmet needs. Am J Med 2014;127:303-10.e3
13.
Herman D, Afulani P, Coleman-Jensen A, Harrison GG. Food insecurity and cost-related medication underuse among nonelderly adults in a nationally representative sample. Am J Public Health 2015;105:e48-59
14.
Berkowitz SA, Seligman HK, Meigs JB, Basu S. Food insecurity, healthcare utilization, and high cost: a longitudinal cohort study. Am J Manag Care 2018;24:399-404
15.
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;121:2177-86.e3
16.
Dean EB, French MT, Mortensen K. Food insecurity, health care utilization, and health care expenditures. Health Serv Res 2020;55(Suppl 2):883-93
17.
Berkowitz SA, Basu S, Gundersen C, Seligman HK. State-level and county-level estimates of health care costs associated with food insecurity. Prev Chronic Dis 2019;16:E90
18.
Berkowitz SA, Basu S, Meigs JB, Seligman HK. Food insecurity and health care expenditures in the United States, 2011-2013. Health Serv Res 2018;53:1600-20 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6426124/pdf/nihms-1000556.pdf
19.
Johnson KT, Palakshappa D, Basu S, Seligman H, Berkowitz SA. Examining the bidirectional relationship between food insecurity and healthcare spending. Health Serv Res 2021;56:864-73
20.
De Marchis EH, Torres JM, Benesch T, et al. Interventions addressing food insecurity in health care settings: a systematic review. Ann Fam Med 2019;17:436-47
21.
Arena O, Salerno C. Four Ways to Address Food Insecurity Through Transportation Improvements. Urban Institute: Urban Wire. January 28, 2020. Accessed September 27, 2022. https://www.urban.org/urban-wire/four-ways-address-food-insecurity-through-transportation-improvements.
22.
Shieh JA, Leddy AM, Whittle HJ, et al. Perceived neighborhood-level drivers of food insecurity among aging women in the United States: a qualitative study. J Acad Nutr Diet 2021;121:844-53
23.
Banks AR, Bell BA, Ngendahimana D, Embaye M, Freedman DA, Chisolm DJ. Identification of factors related to food insecurity and the implications for social determinants of health screenings. BMC Public Health 2021;21:1410
24.
Fischer L, Bodrick N, Mackey ER, et al. Feasibility of a home-delivery produce prescription program to address food insecurity and diet quality in adults and children. Nutrients 2022;14:2006
25.
Bulger JB. Fresh Food Farmacy: A Randomized Controlled Trial. ClinicalTrials.gov. Bethesda, MD: National Library of Medicine. NLM Identifier: NCT03718832. October 25, 2018. Accessed September 27, 2022. https://clinicaltrials.gov/ct2/show/NCT03718832.
26.
Hess A, Passaretti M, Coolbaugh S. Fresh Food Farmacy. Am J Health Promot 2019;33:830-2
27.
Berkowitz SA, Delahanty LM, Terranova J, et al. Medically tailored meal delivery for diabetes patients with food insecurity: a randomized cross-over trial. J Gen Intern Med 2019;34:396-404
28.
Kaiser Permanente. Food as Medicine Study. ClinicalTrials.gov. Bethesda, MD: National Library of Medicine. NLM identifier: NCT05407376. June 7, 2022. Accessed September 27, 2022. https://clinicaltrials.gov/ct2/show/NCT05407376.
29.
Kaiser Permanente. Food Prescriptions to Promote Affordable Diets (FoodRx). ClinicalTrials.gov. Bethesda, MD: National Library of Medicine. NLM identifier: NCT05309395. April 4, 2022. Accessed September 27, 2022. https://clinicaltrials.gov/ct2/show/NCT05309395.
30.
Kaiser Permanente. Evaluation of Medically Tailored Meals in Adults with Medical Conditions at High Readmission Risk (NOURISH). ClinicalTrials.gov. Bethesda, MD: National Library of Medicine. NLM identifier: NCT05166525. December 22, 2021. Accessed September 27, 2022. https://clinicaltrials.gov/ct2/show/NCT05166525.
31.
Go AS, Tan TC, Horiuchi KM, et al.; KP NOURISH Study Investigators. Effect of medically tailored meals on clinical outcomes in recently hospitalized high-risk adults. Med Care 2022;60:750-8
32.
Committee on National Statistics; Division of Behavioral and Social Sciences and Education; Food and Nutrition Board; National Research Council; Institute of Medicine. Research opportunities concerning the causes and consequences of child food insecurity and hunger: a workshop summary. 6, Community responses to food insecurity and hunger. Washington, DC: National Academies Press (US). December 23, 2013. Accessed September 20, 2022. https://www.ncbi.nlm.nih.gov/books/NBK201387/.
33.
Kaiser Family Foundation. Explaining Health Care Reform: Questions about Health Insurance Subsidies. October 27, 2022. Accessed January 23, 2023. https://www.kff.org/health-reform/issue-brief/explaining-health-care-reform-questions-about-health-insurance-subsidies/.
34.
Reeve BB, Hays RD, Bjorner JB, et al.; PROMIS Cooperative Group. Psychometric evaluation and calibration of health-related quality of life item banks: plans for the Patient-Reported Outcomes Measurement Information System (PROMIS). Med Care 2007;45(Suppl 1):S22-31
35.
United States Department of Agriculture. Survey Tools. October 17, 2022. Accessed September 27, 2022. https://www.ers.usda.gov/topics/food-nutrition-assistance/food-security-in-the-u-s/survey-tools/.
36.
American Association of Clinical Endocrinology. What You Need to Know About Diabetes. October 2021. Accessed January 23, 2023. https://www.aace.com/disease-and-conditions/diabetes/what-you-need-know-about-diabetes.
Information & Authors
Information
Published In
NEJM Catalyst Innovations in Care Delivery
Copyright
Copyright © 2023 Massachusetts Medical Society.
History
Published online: March 15, 2023
Published in issue: March 15, 2023
Topics
Authors
Metrics & Citations
Metrics
Altmetrics
Citations
Export citation
Select the format you want to export the citation of this publication.
