Food Security and Mental Health in the United States: Evidence from the Medical Expenditure Panel Survey

With over 40 million food insecure Americans, access to food is a significant policy challenge. Food insecurity is associated with many adverse health conditions, including poorer mental health outcomes. However, previous research generally does not address that poor mental health can both be a cause and a consequence of food insecurity.

We estimate the directional causal effect of food insecurity on mental health status and mental health treatment using bounding methods to partially identify the causal effects from food insecurity to mental health status and mental health treatment.

Data on food security, mental health status, mental health treatment, and individual and family socioeconomic characteristics for adults come from the nationally representative 2016 and 2017 Medical Expenditure Panel Survey. We use both the continuous score (0-10) of a 10-question module on food security as well as classifying adults as living in households that are food secure (0) or having marginal (1-2), low (3-5), or very low food security (6-10). Mental health status is measured using the Kessler-6 (K6) and the PHQ2 depression screening scales. A K6 score of 13 or greater indicates serious psychological distress while a score of 7 to 12 indicates moderate distress. A score of 3 or more on the PHQ-2 indicates probable depression. Mental health treatment is measured by ambulatory mental health visits, prescriptions for psychotropic medications, and total mental health expenditures. Standard parametric regression models are used as a baseline for partial identification models that bound the effects of food security on mental health. In our preferred specification, we impose the following assumptions: monotone treatment selection (MTS), monotone treatment response (MTR), and monotone instrumental variables (MIV) using household income as an instrument.

Those living in food insecure households are more likely to experience psychological distress and depression than those who in food secure households, but do not seek commensurately more mental health treatment. Non-parametric bounds suggest food insecurity increases the probability of moderate psychological distress by no more than 7.2 percentage points, serious psychological distress by no more than 3 percentage points, and probable depression by no more than 4.2 percentage points. The estimated effect sizes of food security on mental health treatment are much smaller, with treatment uptake increasing by no more than 2.4 percentage points.

Our parametric results are consistent with prior findings on the relationship between food security and mental health. We provide evidence for a causal effect of food insecurity which may account for about half the observed association of food security on mental health. A new and previously unreported result indicates that, despite poorer mental health, the food insecure do not show similar increases in mental health care. IMPLICATIONS FOR HEALTH CARE PROVISION AND USE / IMPLICATIONS FOR HEALTH POLICIES: Our results provide policy relevant bounds on the causal impact of food insecurity on mental health. These results raise concerns about the mental health treatment gap in the food insecure population. The relative lack of treatment may point towards deeper structural issues in access to mental health treatment.