Inequality in life and death: What drives racial trends in U.S. child death rates?

Published

Journal Article (Chapter)

This chapter examines the trends in and determinants of child death rates in the United States over the period 1980 to 1998. The annual death rate (number of deaths per 100,000 population) of children age zero to nineteen declined by 39.6 percent over this period, from 117.6 deaths per 100,000 in 1980 to 71.0 in 1998. Several explanations have been offered for this marked decline in the child death rate. For example, David Cutler and Ellen Meara (2000) focus on the role that innovations in medical technology played in the declines in infant mortality. Sherry Glied (2001) draws attention to the dramatic declines in child death rates due to unintentional injuries, or accidents, and the importance of information about public safety and better-educated parents. We extend the previous literature on childhood death rates in three ways. First, we look at how trends differ across causes of death, ages of children, and race. As we document in this chapter, the overall decline in childhood death rates masks important differences along these dimensions. We show that while the overall death rates between black and nonblack children narrowed over the final two decades of the twentieth century, the gap was still sizable in 1998. For example, as of 1998 black children were still twice as likely to die from unintentional injuries as white children. Furthermore, this pattern of narrowing gaps did not hold across all causes of death. Among young children, the black-white gap in death rates due to auto accidents actually increased, and for almost all ages the racial gap in deaths due to intentional injuries, including those involving firearms, increased over this period. Second, our chapter examines a variety of different determinants of these trends that have been identified in the literature and assesses the robustness of their estimated effects to the inclusion of various controls, including state fixed effects and state time trends. We examine the influence of three broad sets of factors: income and inequality as measured by state-level median household income, as well as the differences between the ninetieth and fiftieth percentiles and between the fiftieth and tenth percentiles of the income distribution; other indicators of socioeconomic status (SES), such as maternal education, maternal employment, and the incidence of single-headed households; and indicators of access to medical and trauma care. However, there are many general improvements in factors affecting health and safety that are difficult to measure. Our third innovation is to ask whether child mortality rates are primarily affected by innovations in the factors just mentioned, or whether these trends can be explained by improvements that also drive death rates among adults. For example, safer cars might be expected to benefit both adults and children, while the use of car seats would primarily affect young children. To address this question, we construct regression-adjusted death rates for twenty-four- To forty-four-year-old men, where the regression adjustments net out the effects of our measures of income, socioeconomic status, and medical access. Presumably, these "residual" measures of adult male death rates capture changes in factors that are not readily measured and that affect both adults and children within a state, year, and race group, including changes in medical technology, product safety and regulation, and health care practices. Our models generally do a good job of explaining the level of death rates and the gap in death rates between blacks and whites. Income, inequality, and measures of socioeconomic status such as maternal education are often important predictors of death rates, and gaps in black and nonblack incomes can explain gaps in death rates for some ages and causes of death. However, it is more difficult to explain trends in death rates over time or to explain the narrowing or broadening of the gap between black and nonblack death rates. For example, the gap between black and nonblack median incomes increased slightly over the period, while the gap in death rates narrowed. Moreover, our measures of medical access typically have little explanatory power, though this may be because they are relatively crude. It is possible that year effects, state effects, state time trends, and the male residuals are capturing a good deal of the improvements in medical access and medical technology that took place over the period. In contrast, our estimates indicate that the "male residuals" can explain much of the change in the gaps between black and nonblack death rates. This finding suggests that the gap between black and nonblack overall death rates narrowed because of factors that affected adults as well as children rather than solely because of factors affecting children. In the next section, we discuss the previous literature concerning childhood death rates and their potential causes. The third section describes the data sources we use and the variables we have constructed. In the following sections, we lay out the trends in childhood death rates by cause, age, and race; we also describe the trends in our explanatory variables. Next, we present the results from our regression analysis and discuss our findings. We conclude with a short summary of our findings and their implications. Copyright © 2004 by Russell Sage Foundation.

Duke Authors

Cited Authors

  • Currie, J; Joseph Hotz, V

Published Date

  • December 1, 2004

Start / End Page

  • 569 - 632

Citation Source

  • Scopus