Social factors and infant mortality: Identifying high-risk groups and proximate causes

The infant mortality rate is regarded as one of the most revealing measures of how well a society is meeting the needs of its people (Newland, 1981:5). In comparisons of groups within a society, as well as comparisons among societies, it is considered an indicator of deprivation. Although the absolute risk is low for most groups in the United States, differences in risk nevertheless are socially meaningful and politically sensitive. Infant mortality is often cited as a cause for concern about adolescent pregnancy (Baldwin and Cain, 1980), illegitimacy (Berkov and Sklar, 1976), poverty and race (Food Research and Action Group, 1984), and federal spending on social welfare programs (Miller, 1985). The basis for such concerns is statistical data on group differences in infant mortality. It is well established that teenagers, blacks, unwed mothers, and mothers of low socioeconomic status are high-risk groups (Blum, 1984; Miller, 1985). Much of the evidence on group differences, however, consists of relatively simple, descriptive statistics, often just bivariate correlations. These descriptive statistics may be misleading in several respects. The purpose of this paper is to go beyond simple description and examine causal relations among social factors and infant mortality to provide a more solid basis for discussions of social problems. Research on social differentials in infant mortality is deficient in three respects. First, the risk factors are correlated, and descriptive statistics may include causally spurious components. The most frequently examined factors include maternal age (McCormick, Shapiro, and Starfield, 1984; Menken, 1980; Scott, Field, and Robertson, 1981), marital status (Berkov, 1981; Keller, 1978; Powell-Griner, 1981; Ventura, 1980), and race and socioeconomic status (Binkin et al., 1985; Bross and Shapiro, 1982; Eberstein and Parker, 1984; Gortmaker, 1979; McMahon, Kovar, and Feldman, 1972). Teenage mothers are disproportionately unwed, black, high school dropouts, and poor; it is controversial whether these factors account for teenagers' high risk of infant mortality, or whether age itself is a risk factor (Chilman, 1980; Horon, Strobino, and MacDonald, 1983). Similarly, marital status may not be an independent risk factor, as Keller (1978) found in the case of early neonatal mortality; babies born to unwed mothers may experience higher neonatal mortality because mothers are disproportionately young, black, and high school dropouts. In general, it is not known which social factor or combination of factors is causally responsible for the observed group differences. Because of limitation's in the data available from linked birth and infant death records, no more than three factors have been examined simultaneously. For example, Gee, Lee, and Forthofer (1976), Berkov (1981), and Powell-Griner (1981) examined race and marital status, with age controlled; Bross and Shapiro (1982) examined race, education, and age; and Keller and Fetterly (1978) examined education and marital status, with age controlled. Since results for any one variable differ depending on what other variables are controlled, they cannot be compared easily across studies. Many studies are also deficient in how social factors are specified in their models. Not only are the social factors correlated, they also interact in their associations with