Duke 2008 | ASHE | Equity and Efficiency in Health and Healthcare

Asthma, one of the most common chronic conditions of childhood, profoundly affects quality of life for millions of U.S. children. Although asthma affects children of all race-ethnicities, black children are much more likely than other groups to be diagnosed with the condition. Further, among children with asthma, black children are much more likely than others to have poorly controlled asthma resulting in excess missed school days, activity limitations and avoidable emergency room (ED) visits and hospitalizations. Among children diagnosed with asthma from 2000-04, for example, black children (14.5%) were about three times as likely as white children (4.5%) to have an ED visit or hospitalization to treat the condition. This paper uses data from the 2000-05 Medical Expenditure Panel Survey (MEPS) to examine racial and ethnic differences in asthma treatment and outcomes for a nationally representative sample of school aged children (ages 5-18) children in the United States who have been diagnosed with asthma.

We begin by examining patterns of pharmaceutical treatment of asthma among black non-Hispanic, white non-Hispanic and Hispanic children diagnosed with the condition. Current asthma treatment guidelines emphasize the importance of using controllers (drugs that provide long-term control of underlying airway inflammation), as opposed to relievers (drugs that provide relief of acute symptoms). Short courses of oral steroids are recommended for the management of severe asthma attacks. We examine differences across racial-ethnic groups in controller, reliever and oral steroid use (any use, number of prescriptions purchased) among school aged children diagnosed with asthma.

If disparities in asthma treatment exist, they may affect observed disparities in health outcomes related to the condition. Accordingly, we examine disparities across racial ethnic groups in whether each child had any asthma attacks during the year, any missed school days or activity limitations due to asthma and any ED visits or hospital stays to treat the condition. Finally, we estimate models that directly examine the contribution of pharmaceutical treatment to disparities in outcomes. A central difficulty in examining this issue, however, is that differences in asthma severity may affect both treatment choices and outcomes. We control for severity in our models by including detailed information on health status and co-morbidities. In an alternative specification we use longitudinal models that include fixed effects to control for unobserved individual characteristics.

The functional form of our models varies by the type of dependent variable. We estimate logit models for binary variables and negative binomial models for count variables. For both types of models we estimate marginal effects using the method of recycled predictions and estimate standard errors using balanced repeated replication (BRR). In addition to controls for asthma severity, our models control for a wide range of individual characteristics (e.g., age, sex, insurance status, usual source of care), family level characteristics (e.g, family income, parental education and parental smoking), neighborhood characteristics (e.g. the percentage of buildings that are vacant, the median age of buildings) and population characteristics (e.g. the prevalence of poverty, the unemployment rate, population density), and geographic variables including census region and MSA.