Diabetes has long been one of the most expensive medical conditions. In 2013 spending on care for people with the disease in the United States topped $100 billion. But it is also one of the most amenable to simple, low-cost behavioral interventions. At Geisinger, we set out to improve the health of diabetic adults by providing them with free, nutritious food and a comprehensive suite of medical, dietetic, social, and environmental services. This program, our Fresh Food “Farmacy,” has had clinical impacts superior to those provided by medications that cost billions of dollars to develop, and has done so at dramatically lower cost. Finding effective, less expensive treatments for diabetes is critical because of its enormous social and financial costs and its growing prevalence: One in 10 people currently has diabetes, and it is estimated that by 2050 the figure will rise to one in three.
People who are food insecure, meaning they cannot reliably get nutritious food, are more likely to have diabetes and to be obese and in poorer health. Food insecurity is widespread, and is particularly serious in many of the areas that Geisinger serves. While 12.7% of the U.S. population and 18% of children are food insecure, in many of the counties we serve those numbers are even worse: 14% of their overall populations and 23% of children. One in eight of these food-insecure people has diabetes. People with severely limited incomes are often faced with difficult choices that lead them to buy inexpensive, nutrition-poor food. And hunger itself may increase the likelihood that a person will have unhealthy behaviors. Together, these behaviors can contribute to the development of diabetes and exacerbate it.
Building and Running the “Farmacy”
Since its inception in 2016, this initiative has been a purposeful and iterative learning endeavor. We started the program in a county that has particularly high rates of diabetes, food insecurity, poverty, and unemployment. First, we queried our electronic health record’s (EHR) database for adult patients in selected zip codes who had a diagnosis of type 2 diabetes and hemoglobin A1c (HbA1c) levels ≤8 (indicating that their disease was not controlled). Most people without diabetes have HbA1c levels below 6, while those in well-controlled diabetic patients are usually under 7. This is not just academic: For diabetic patients, every one-point drop in this measure corresponds to a more than 20% decrease in chance of death and serious complications from the disease, such as blindness and kidney failure.
We then screened these people for food insecurity with a simple tool linked to our EHR, asking them to respond to two questions: (1) “Within the past 12 months I/we worried whether our food would run out before we got money to buy more,” and (2) “Within the past 12 months the food I/we bought just didn’t last, and we didn’t have money to get more.” Anyone who agreed with one or both of these statements was considered food insecure.
Today, patients who meet these criteria and express interest in our program are referred to an enrollment class where they meet their care team and receive a “prescription” for healthy, diabetes-appropriate food. This multidisciplinary team comprises a program coordinator, nurse, primary care physician, registered dietitian, pharmacist, health coach, community health assistant, and, importantly, nonclinical administrative-support personnel.
We built a food pantry at one of our clinical centers so that patients can pick up food and receive care at one location. Because the consequences of hunger affect entire families, the pantry provides patients and their families with the food, menus, and recipes needed to prepare two healthy, fresh meals five days per week. However, providing nutritious meals is not nearly enough; achieving sustainable health and lifestyle improvements requires educating patients about healthy eating habits, goal setting, exercise, mindfulness, and diabetes management.
Thus, in addition to receiving standard diabetes medical care, our program patients also participate in 15 hours of group classes on diabetes self-management. They receive direct medication-management assistance from a pharmacist, follow-up with a registered dietitian, health coaching, and ongoing case management. This care is provided through a medical home model so that participants receive reliable, patient-centered, multidisciplinary collaborative care.
Financing is managed through a collaboration between Geisinger Health and our health insurance company, Geisinger Health Plan (GHP). We currently fund the program through grants (40%), in-kind reciprocal contributions with GHP (30%), and private donations (30%). Geisinger provides rent-free space for the food pantry and clinical areas, though we anticipate that, as we expand, it’s likely that the program will pay rent. Employee expenses form the bulk of the costs; food accounts for only a small portion of the total. Program costs (before clinical and behavioral gains) initially amounted to $3,000 to $5,000 per year per patient across the initiative, with that cost decreasing to about $2,200 as we have progressed and learned.
Because Geisinger has a health insurance company, it is easy to access to payer-side claims data. This gives us a deep view into the true financial costs of care and disease. Strikingly, when we look back several years at our program patients before the initiative started (a particularly complex group of patients that includes some outliers), their care cost GHP $8,000 to $12,000 average per person per month. We have seen these payer-side costs drop by two-thirds on average across the program.
This initiative touches nearly all segments of the provider and administrative environments and many external resources. In our experience, programs of this type need executive-level endorsement in addition to administrators, staff, and population health personnel dedicating their time to participate. As mentioned, administrative staff are essential to success, as they free program clinicians to perform at their highest skill levels.
What We’ve Learned
Over the first nine months, we scaled from our original six patients to 50 patients and their families. Now, over a year in, more than 80 patients and their families are actively enrolled, and we feed approximately 250 people 10 meals each week.
The clinical results have been striking. Usually, when medically challenging diabetic patients have a second or even third diabetes medication added to their treatment, HbA1c levels drop between 0.5 and 1.2 points. With 12 months of healthy food and lifestyle changes, we have seen HbA1c levels drop more than two points, from an average of 9.6 before the program to 7.5. Our initiative has had a greater impact on diabetes control (albeit in a small population) than expensive medications that have significant side effects. We have also seen significant improvements in patients’ cholesterol, blood sugars, and triglycerides — improvements that can lower the chances of heart disease and other vascular complications. Among the program’s other benefits, we’re doing a better job than before at making sure patients are on appropriate medications and getting regular diabetic foot exams (to prevent them from one day needing amputations).
Other organizations contemplating similar programs can learn from our experience. Three points stand out:
First, we noticed early on that when patients switched from their existing low-nutrition diets to eating the foods we provided, some experienced potentially dangerous drops in blood sugar levels. Essentially, their bodies, having become accustomed to certain blood-sugar-lowering medications, had come to expect unhealthy high-sugar and high-fat foods. When those were replaced with healthier foods, the medications depressed blood sugar levels too much. Care teams should anticipate this risk.
Second, patient engagement is crucial to success. Though we’ve offered diabetes-education classes from the start, we did not initially require attendance. Yet patients who reliably attend the classes have significantly greater improvements in blood sugar and HbA1c. For that reason, we now require class attendance. We initially suspected that free, healthy food itself would lead to significant engagement and positive response. However, many of these patients have transportation, housing, and family-care challenges that make full engagement hard. How can a person who takes care of a disabled child or elderly parent full-time come to pick up food and attend classes? Challenges like these led us to hire a community health assistant to help patients navigate these real-world socioeconomic problems that are barriers to exceptional care.
And finally, as human resource costs account for most of the initiative’s expenses, it’s critical that providers and staff practice at the highest level their skill allows. For example, our initial instinct was to have registered dietitians carry much of the clinical work and leadership burden. But we have found that support staff, coaches, and nutritionists can successfully perform many of the functions that do not require a dietician’s specialized training. Carefully allocating work and onboarding more volunteers have helped us to remedy some initial missteps in capacity-cost management.
The Next Steps
Integrated systems like Geisinger that have their own health insurance company can directly financially benefit from programs like this that save payers money. However, most provider systems are still heavily reliant on fee-for-service care, and are not sharing in payer-side efficiencies. As we rapidly expand our program to additional locations and thousands of participants, a major focus is on demonstrating the programs’ financial impact so that payers and large health care purchasers (such as employers) will support it.
We are also looking to partner with a national retail supermarket chain to help us scale the initiative. To grow further, we will need to significantly improve our data-capture processes. To this end, we are beta-testing a novel iOS app that integrates with Apple’s HealthKit; this app enables patient engagement, communication, and real-time data exchange with a provider-facing clinical dashboard on iPads.
Ultimately, we hope to extend this program to reach food-insecure children with diabetes as well as diabetics for whom food insecurity is not an issue. If programs like ours can be scaled to a national level, they could improve the health of millions of people with diabetes, profoundly reduce costs, and even help decrease the prevalence of diabetes.
We are grateful to the Central Pennsylvania Food Bank for their partnership and assistance.