Leveraging Health Plan Infrastructure, Patient Voices to Improve Food Insecurity Identification: Jonathan Wrathall, PhD

The American Journal of Managed Care^® (AJMC^®) recently spoke with Jonathan Wrathall, PhD, a senior advanced analytics consultant at Elevance Health, for an episode of Managed Care Cast. Wrathall authored 2 studies published in the May 2026 issue of AJMC, both focused on food insecurity: “Food Insecurity Identification Modeling for Medicare Enrollees Using Administrative Data” and “Making the Most of Limited Resources: Predicting Food Insecurity.”

In this excerpt from the interview, Wrathall explains why food insecurity is such a critical issue for health systems and payers to identify and address. He also highlights key findings from both studies and concludes by discussing the most important step stakeholders can take to improve the identification of patients experiencing food insecurity.

This transcript has been lightly edited for clarity.

AJMC: You had 2 food insecurity-related studies published in the May issue of AJMC. Before we get into those, why is food insecurity such an important issue for health systems and payers to identify and address in today’s care environment?

Wrathall: The role of health-related social needs, or HRSNs, is really well documented as a means of setting the stage for poor health conditions and health outcomes. So, medical intervention for clinical health tends to account for a relatively low degree of both impactable change and improving long-term health outcomes. Where HRSNs fit in, like food insecurity, is situated really at the intersection of social and health conditions that impact downstream health.

Food insecurity is really just not having enough food to eat to sustain daily life. In this case, food insecurity intervention can go a long way to improve health conditions before they start. It can mediate them as they progress and bolster health outcomes. An intervention, then, offers a natural starting point in addressing basic needs for at-risk individuals to help keep people healthy.

The hope with this work was to highlight what we can do to help keep people healthy, rather than waiting to intervene after a chronic condition or episodic medical care really occurs. Food insecurity, then, is really about addressing the whole person and the factors that influence their health.

AJMC: Beginning with your study titled “Food Insecurity Identification Modeling for Medicare Enrollees Using Administrative Data,” your model achieved strong predictive performance using administrative and social risk data. What do you think made prior documented social needs and dual Medicare-Medicaid enrollment such powerful predictors of food insecurity?

Wrathall: It's a little bit unfortunate reality that many HRSNs are comorbid, in that those suffering from food insecurity often face other clinical or social challenges together. One of the key questions facing managed care organizations (MCOs) in the social domain space is how much weight should be given to any one data point.

We show that we can take at face value the data that an MCO typically already has, which is quite helpful in providing an identification starting place. This points to the issues surrounding social determinant program investment. Because programs can be fairly costly to scale, we can simply allow HRSN attestation to speak for itself.

This model uses appropriate data sourcing in that one HRSN is likely to be related to others. In that vein, in the event of the attestation of one or more needs, there are others that are likely to follow. What we saw in the model, as you pointed out, was that HRSN clusters tended to contribute to a likely food insecurity attestation. That's what was somewhat expected, I guess, about the HRSN clusters, but as far as the dual enrollment finding, what we found was that it also had a fairly high predictive quality.

We reasoned that people who were eligible for a duals program likely faced a host of vulnerabilities from mobility or persistent mental health challenges, which generally compounded their food insecurity risk.

AJMC: Now turning to your second study, “Making the Most of Limited Resources: Predicting Food Insecurity,” insurance type emerged as the strongest predictor of food insecurity. Why do you think it carried so much predictive weight?

Wrathall: Similar to what we found in our other paper, in many instances, people self-select into their health plan. In this case, there are also several measured and nonmeasured factors rolled up into the insurance type, which can be indicative of sensitivities relating to food insecurity. These factors may be overt, like an indicator of poverty status, but also perhaps other issues, like benefit structures, such as Food is Medicine benefits, fresh food subsidies, or transportation vouchers, which are often more determined by their administering bodies.

In this case, it's likely that the signal produced by sensitive populations, like Medicaid, Medicare dual coverage, or some other designation, which, again, I should mention, corroborates our findings from our identification work, represents a considerable predictive ability simply by outweighing other signal factors. So, the signals from these vulnerabilities really become very strong.

AJMC: Based on the findings from both studies, what is the most important step payers and health systems can take right now to improve identification of patients experiencing food insecurity?

Wrathall: One takeaway from both studies is that the infrastructure of the health plan already exists to identify, target, and assess HRSNs and is relatively easy to do. For example, the Household Food Insecurity Assessment Scale, or another, the PREPARE assessment. These are screeners, which are just a few questions that can be used at program enrollment, for example, to gather information quickly and succinctly about food insecurity status. From there, the extrapolation can also be done by comparing similar groups, which is what this research is really about.

Another thing that health systems can do right now, which may go without saying, is to take seriously what a person says about their food security status. A consistent struggle with such programs comes up again and again, which has been the notion of data sourcing settings. The question follows: How do we know for sure that a patient has food insecurity if it wasn't done in a clinical setting, or if we asked them 12 months ago, or what have you?

In a lot of ways, they've already told us, and we just, in some ways, aren't listening. We can listen through their claims and through their assessments. We have to be ready for it to matter by considering what they tell us.