Changes in Prenatal Care Timing and Low Birth Weight by Race and Socioeconomic Status: Implications for the Medicaid Expansions for Pregnant Women - Statistical Data Included

Health Services Research, June, 2001 by Lisa Dubay, Ted Joyce, Robert Kaestner, Genevieve M. Kenney

Objective. To conduct the first national study that assesses whether the Medicaid expansions for pregnant women, legislated by Congress over a decade ago, met the policy objectives of improved access to care and birth outcomes for poor and near-poor women.

Data Sources/Study Setting. Data on 8.1 million births using the 1980, 1986, and 1993 National Natality Files. We use births from all areas of the United States except California, Texas, Washington, and upstate New York.

Methods. We conduct a before and after analysis that compares obstetrical outcomes by race and socioeconomic status for the periods 1980-86 and 1986-93. We examine whether women of low socioeconomic status showed greater improvements in outcomes during the 1986-93 period compared to the 1980-86 period. We analyze two obstetrical outcomes: the rate of late initiation of prenatal care and the rate of low birth weight.

Data Collection. Natality data were aggregated to race, socioeconomic status, age, and parity groups.

Results. During the 1986-93 period, rates of late initiation of prenatal care decreased by 6.0 to 7.8 percentage points beyond changes estimated for the 1980-86 period for both white and African American women of low socioeconomic status. For some white women of low socioeconomic status, the rate of low birth weight was reduced by 0.26 to 0.37 percentage points between 1986 and 1993 relative to the earlier period. Other white women of low socioeconomic status and all African American women of low socioeconomic status showed no relative improvement in the rate of low birth weight during the 1986-93 period.

Conclusions. The expansions in Medicaid lead to significant improvements in prenatal care utilization among women of low socioeconomic status. The emerging lesson from the Medicaid expansions, however, is that increased access to primary care is not adequate if the goal is to narrow the gap in newborn health between poor and nonpoor populations.

Key Words. Medicaid, pregnant women, access, outcomes

Significant differences in the incidence of adverse birth outcomes by race and socioeconomic status have long characterized infant health in the United States (Kleinman and Kessel 1987; Schoendorf et al. 1992). These differences have been attributed, in part, to inadequate access to prenatal care services (IOM 1985). Spurred by these disparities and by the United States' low international ranking in infant mortality, Congress extended Medicaid coverage to poor and near-poor pregnant women and infants through a series of legislative reforms between 1986 and 1990. Despite subsequent increases in Medicaid participation rates and declines in the numbers of deliveries to uninsured women, evaluations of the expansions have reported inconsistent evidence regarding whether prenatal care use increased and have found virtually no improvement in birth outcomes (Piper, Ray, and Griffin 1990; Haas et al. 1993; Piper, Mitchel, and Ray 1994a; Curie and Gruber 1996; Ray, Mitchel, and Piper 1997).

Published evaluations of the effects of Medicaid expansions, however, have limitations. In the only national study published to date, researchers tested whether state low birth weight and infant mortality rates were correlated with the proportion of women eligible for Medicaid in that state but did not control for potential confounding from time-varying factors that affect birth outcomes of poor and nonpoor women differently (Currie and Gruber 1996). All other evaluations have been conducted in individual states in an effort to exploit linkages between birth certificates and Medicaid administrative files or discharge abstracts that are not available at the national level. Some of these more narrowly focused evaluations may have had limited power to detect statistically significant effects because they examined small incremental changes in Medicaid or included relatively few women affected by the expansions (Piper, Ray, and Griffin 1990; Haas et al. 1993; Piper, Mitchel, and Ray 1994a). Other evaluations did n ot address temporal confounding (Ray, Mitchel, and Piper 1997; Long and Marquis 1998).

A national evaluation of the Medicaid expansions for pregnant women and infants based on well-defined treatment and control groups has been difficult to accomplish for two reasons: first, there are no nationally representative data with information on family income, insurance status, and birth outcomes necessary for defining treatment and control groups; second, even assuming the requisite data existed, an appropriate comparison group would be difficult to define. The Medicaid expansions were not limited to increases in income eligibility thresholds. Improvements in the eligibility determination process, increased fees, and reimbursement for enhanced prenatal care services affected women already covered by Medicaid as well as women eligible but not covered. Thus, one potential control group--women on Medicaid prior to the expansions--was likely affected by the reforms.

To evaluate whether the Medicaid expansions achieved the policy objective of increased access to care and improved birth outcomes among poor and near-poor women, we conducted a before and after analysis using national natality files to compare rates of delayed initiation of prenatal care and rates of low birth weight by race and socioeconomic status for the periods 1980-86 and 1986-93. If the extensive changes in Medicaid that took place between 1986 and 1993 were effective, we would expect to find improvements in obstetrical outcomes among women of low socioeconomic status between 1986 and 1993 that exceeded changes observed in this group between 1980 and 1986, a period of no major programmatic changes in Medicaid. As an alternative comparison, we contrasted changes in obstetrical outcomes between 1986 and 1993 for women of low and high socioeconomic status, since the latter were unaffected by changes in the Medicaid program.