Direct and Indirect Effects of Maternal Intelligence, Maternal Age, Income, and Home Environment on Intelligence of Preterm, Low-Birth-Weight Children

Low birth weight (LBW) infants are at elevated risk for adverse developmental outcomes, including impaired cognition. From a public health perspective, interventions designed primarily to prevent and/or secondarily to avert intellectual impairments could have substantial impact on social and fiscal costs associated with LBW. But LBW is a heterogenous condition correlated with demographic, biological, psychological, and social variables. In this study we examined direct and indirect effects of maternal intelligence, age, income, and home environment on 3year intelligence among premature LBW children. Hierarchical regression analyses were employed to determine the extent to which SES (measured by family income) and home environment mediate effects of maternal IQ and age on child IQ. Maternal IQ was found to have major direct and indirect consequences. Indirect effects were mediated by income and home environment. A small part of the influence of income is explained by home environment. Taken together with results of other studies, these findings suggest that standard family interventions to improve cognitive outcome for children of poor and intellectually compromised women are not likely to produce impressive results. Preterm, low-birth-weight infants (LBW <2500 g) are at significantly elevated risk for delayed cognitive development in comparison with full-term, normal-birth-weight infants (NBW) (Baumeister & Bacharach, 1996; Hack, Klein, & Taylor, 1995). Furthermore within the LBW range, a weight gradient describes this relationship in that risk for cognitive delay increases as weight decreases (Breslau, et aI., 1994; Escobar, Littenberg, & Petitti, 1991). A recent population-based study has shown that for both LBW and NBW prematurity (controlling for sex, birth order, maternal age, maternal race, and maternal education) adjusted odds ratios for mental retardation were 2.8 (Mervis, Decoufle, Murphy, & Yeargin-Allsopp, 1995). As might be expected, risk even was higher for very LBW children. The relationship among LBW, risk profile, and behavioral outcome has long been known to be multiply and inter-generationally determined, implicating numerous distal social and more proximal familial and biological factors (Birch & Gussow, 1970; Emanuel, Filaki, Albennan, & Evans, 1992; Drillen, 1964; Rutter & Madge, 1976). Over 20 years ago Hardy and Mellits (1977) listed more than 30 factors known to be associated with birthweight including behavior, medical, and economic variables. Obviously simple univariate and bivariate analyses are insufficient to capture the nature and complexity of variables that place a child at risk for LBW and that mediate the outcome (Willerman, 1979; Ramey & MacPhee, 1986). Epidemiological and clinical studies reveal LBW children are particularly vulnerable to socioeconomic, biological, and psychological risks that predispose all children to impaired cognitive development (Alberman, 1994). But simple extrapolation from circumstances that condition outcome for NBW children to those who are premature LBW is misleading because an important interaction involves birth weight and social circumstances. While both LBW and NBW children reared in high-risk social environments experience poorer developmental outcomes than those in more wholesome social milieus, LBW children are more greatly affected (McGauhey, Starfield, Alexander, & Ensminger, 1991; Siegel, 1982). Moreover, the social risk profile is different for LBW children (Starfield, 1992). Heightened relative risk for LBW premature children applies not only to specific health outcomes, but to broader performance indicators such as school achievement and behavior disorders. In view of these considerations, different and more creative intervention strategies are indicated for LBW children and their families. Maternal IQ is a major consideration in assessing developmental risks associated with LBW and what may be done to alleviate them. Using maternal education as a marker of intelligence, Wilson (1985) found among samples of monozygotic and dizygotic LBW twins, one born smaller than the other, that maternal intelligence was the strongest predictor of preschool IQ. Maternal intelligence accounted for variance otherwise due to socioeconomic factors associated with home environment. Heritage had a powerful effect that increased over time. LBW children making the greatest recovery from initial suppressing effects of prematurity are those from more advantaged families. Reared in supportive environments, most LBW infants will recover from early trauma, even those born at very LBW. This is a consideration that is too often overlooked because children are resilient, even in the face of multiple adversities. Resilience poses the questions as to which aspects of the familial-social-economic context contribute to development of LBW children, which children are most susceptible, and where is it possible to design tailored intervenetions. Some predisposing variables cannot be altered (e.g., low maternal intelligence), while some may be (e.g., events associated with low income). Hierarchical analyses of cognitive development among LBW children indicate causally prior effects of maternal intelligence are both direct and indirect. Some portion of the maternal IQ effect is mediated by experiences in the home environment, particularly those aspects that are cognitively enriching (Baumeister & Bacharach, 1996; Bradley, Caldwell, & Rock, 1990; Bradley, Whiteside, Caldwell, et aI., 1993). According to Bradley, Whiteside, and Caldwell, et al. (1993) the respective contributions of maternal IQ and home environment to child IQ may also be associated with maternal age. Although the correlation between parental and child IQ among LBW children appears to be partly dependent on the child's experiences within the home, there is reason to be cautious about overstating the case for improving the home environment. First, efforts to alter the developmental trajectories by home intervention have been less than impressive. Second, the correlation between child intelligence and home environment also has a genetic component in that more competent parents may provide more stimulating and enriching experiences (Plomin & Bergeman, 1991). We assume that a number of variables mediate effects of maternal IQ on cognitive development of LBW children. This assumption is difficult to test because conditions that predispose these children to developmental disabilities covary. Multicollinearity is to be expected in generating models for predicting intellectual performance among premature LBW children. Many predictor variables will have low tolerance. For this reason it is essential to employ hierarchical analyses in situations such where suspected risk and mediating variables are implicated in the outcome. For example, family income, maternal IQ, and maternal age are correlated, and all are related to developmental disabilities. There is also evidence that low maternal IQ and poverty exert independent detrimental effects on child IQ (Feldman & Walton-Allen, 1997). Descriptive epidemiology or demographic studies have identified many possible causes of disease processes such as prematurity or LBW. Descriptive studies are necessary, but it is analytical epidemiology that will reveal key causes. One approach to disentangling this complex web of associations is to employ hierarchical regression procedures incorporating variables known to mediate the connection between maternal IQ and child cognitive development. Given collinearity among predisposing variables, it is important to employ hierarchical models to identify direct and mediated effects and to estimate their relative contributions to child IQ. Several multivariate comprehensive frameworks have been proposed that link developmental outcomes to predisposing and systemic variables in an effort to describe paradigms for unraveling the relationships among variables at different levels, from proximal to more distal (e.g., Baumeister, Kupstas, & Woodley-Zanthos, 1993; Brofenbrenner, 1977; Garbino, 1991; Sameroff, 1983). Each conceptualizes compromised cognitive outcome in terms of multiple risks, occurring over time, and in a hierarchical manner. These models can thus identify points, either at individual or systemic levels, where interventions might be most efficacious and/or cost-effective. Inclusive conceptualizations delineate the various ways that cognitive development among preterm LBW infants is influenced by psychological, socioeconomic, and biological threats, including diminished maternal IQ. Although differing in detail and emphasis, these systems models share the common perspective of contextualism: the child is regarded as a biological entity in interaction and transaction with a network of family, community, service, social, economic, and political systems. Studies have consistently shown that, aside from maternal intelligence, certain predisposing variables influence cognitive development among LBW children. Intellectual development is related to maternal age (Liaw & Brooks-Gunn, 1993). Poverty, as measured by family income, also influences cognitive outcome (Duncan, Brooks-Gunn, & Klebanov, 1994; Thompson, Goldstein, Oehler, Gustafson, Catlett, & Brazy, 1994). Birthweight effects are exacerbated by absence of the father (Ricciute & Scarr, 1990). The relationship between resource variables, such as quality of the home environment, and cognitive development have been documented in other studies (Bradley & Casey, 1992; Lee & Barratt, 1993; Liaw & Brooks-Gunn, 1993). These variables are not independent of one another, and their relative and independent contributions to cognitive development among LBW children are uncertain. Aside from SES, maternal education is the most frequent variable examined in the literature regarding the connection between LBW and cognitive outcome. The strongest predictor of child IQ-maternal intelligence (Ramey & Ramey, 1992)-typically is not included in hierarchial multivariate analyses predicting cognitive development among premature LBW children. Although, as noted, a common procedure is to employ education as a proxy, in a previous study involving the LBW cohort included in the present analysis Baumeister and Bacharach (1996) found that the partial correlation for maternal education and child IQ, with maternal IQ controlled, was .09. Controlling for maternal education, the corresponding correlation for maternal IQ was .43. Influence of maternal IQ should be mediated by predisposing variables such as maternal age at parturition, the effects of which are, in turn, conditioned by income or SES. The impact of income or SES should be moderated by family influences such as the home environment, while resource variables (e.g., income) may have a direct effect on cognitive growth among at-risk LBW children. The purpose of this investigation is to test these hypotheses within a hierarchical regression model.