Measuring the impact of programs for mothers and infants on prenatal care and low birth weight: the value of refined analyses.

During the past two decades, intervention strategies designed to improve the health status of mothers and infants have been widely disseminated, yet relatively few have been evaluated for effectiveness. Moreover, most reported investigations have involved straightforward comparisons of aggregate data, employing various degrees of methodological control. In this study, vital statistics data were used to assess the effects of the North Carolina Maternity and Infant Care (MIC) Project on use of prenatal care and low birth weight. A weighted least squares procedure was used to control for selected maternal characteristics and identify significant interactions. Analyses of total population data indicated only minor MIC effects. However, more careful scrutiny of subpopulation data suggests that MIC impacts differed across categories of maternal risk status, with the greatest influence observed among mothers and infants at greatest risk. These findings raise several questions regarding the genesis of differential effects and suggest areas of special concern in conducting, interpreting, and using evaluations of programs for mothers and infants.