While it is well known the health of American children tends to decline with family income, the evidence so far has mostly relied on survey data that limited the ability to look at many different diagnoses at different levels of income. Now, using novel linkages between large-scale administrative data on diagnoses and mortality, researchers have been able to zoom in on the details of the relationship between income and health among low-income children.

Maria Polyakova, an assistant professor of health policy, Victoria Udalova, a Senior Economist at the U.S. Census Bureau, and Vinayak Bhatia, a graduate student at Carnegie Mellon University and former Stanford Health Policy researcher, have published their original investigation in the Journal of the American Medical Association (JAMA.) Taking advantage of the new data linkages enabled by the EHealth initiative of the Census Bureau, led by Udalova, the authors were able to assemble a large sample size of 795,000 children to look at detailed patterns of clinically recorded diagnoses and family income.

They focused on children ages 5 to 17 who lived in families with annual income under 200% of the U.S. poverty level — or about $44,000 for a family of three — and who were insured under Medicaid or CHIP programs in years 2011 and 2012. They used diagnoses records in Medicaid claims data to relationship between income and mental health disorders — including ADHD, anxiety, autism spectrum disorders and depressive disorders — as well as infections, injuries, asthma, anemia, and drug, alcohol and tobacco disorders. The authors made the data on the prevalence of each condition by income ranges publicly available

They also used administrative records of the Social Security Administration to examine the relationship between income and 10-year mortality. Thirty-six percent of children in their data have experienced an infection within a two-year period of measurement, and 13% had a chronic mental health diagnosis. 0.6% of all children in their sample died within a 10-year period of follow-up, at the average age of 19.8 years.

The study was unique in its ability to examine the association of income with many different conditions during childhood, measured in the same data for the same large sample of children.

In addition to bringing in new data into the discussion, the study had two goals. First, to examine the shape of the relationship between income and health at lower-income levels. Work that has studied the mortality-income gradients among adults has found particularly steep gradients at lower-income levels, so the authors wondered if this also reflected a lifelong accumulation of steeper gradients in health among children in the lowest-income families.

“We’re not comparing children in very poor to very rich families, but instead looking at children all of whom were living in relatively low-resource families,” Polyakova said. “Some could have argued that all these kids have similar access to medical care through Medicaid and maybe that should substantially reduce any differences in their health outcomes — but that is not the case.”

However, the study found no association between 10-year mortality and income among children ages 5 to 9, raising the question of whether access to health insurance may be playing a role in reducing this gradient.

In an accompanying editorial, two independent researchers note: "The headline finding is that an additional $20,000 in annual income was associated with a 15% reduction in the probability of mental health disorders for children aged 5 to 9 years."

The second big issue the study was able to address with new precision is the so-called “equalization hypothesis.” The question of how health-income gradients among children evolve as children age has been debated in the literature, as some researchers have found that such gradients decline as children enter the school system and socialize with peers. Other studies have found that such gradients increase. The findings by Polyakova and co-authors strongly support the hypotheses  proposed by Princeton’s Janet Currie and co-authors, who have argued that the association between income and health grows stronger, rather than diminishes, as children age and that it is driven more by frequent exposure to health shocks among lowest-income children than their ability to recover from shocks. This is consistent with the higher rates of infections and injuries at lower incomes that Polyakova and co-authors found in their data.

The results for mental health diagnoses and substance use disorders were particularly striking. The paper found pronounced gradients in many types of mental health diagnoses, such as ADHD, anxiety, depression, and PTSD. Moreover, given the large sample size, the authors were able to document the early emergence of the income gradient in the prevalence of drug use disorder among 10- to 17-year-olds, more than 1% of whom had this diagnosis in the lowest income families and somewhat less than 1% in the highest income families in the sample.

The authors cautioned that while their results are helpful in understanding the facts, evidence-based policymaking would require being able to understand the causal mechanisms behind these patterns, which this paper was not able to do.