The Importance of Psychosocial Variables in Predicting Low Birth Weight

Objective: An analysis of a pre-existing data set of 606 inner city pregnant
women collected by the Fetal Alcohol Research Center of Wayne State University School of Medicine in Detroit,
Michigan, was conducted to determine if the inclusion of psychosocial variables would improve the prediction of
low birth weight. Method: Along with the usual medical and demographic data typically collected by obstetricians
interested in low birth weight, data collection had also included variables assessing maternal comfort, feelings of
hopefulness, intimacy of relationship with the baby’s father, and alcohol and drug use. The sample was divided
into cases for developing the mathematical models and test cases for comparing two different mathematical
approaches-discriminant function analysis (DFA) and a computer simulation modeling method derived from
chaos theory (dynamic systems modeling or DSM). Unlike DFA, DSM required a priori specification of the
relationships among the variables. Findings: Psychosocial variables were needed by both mathematical
approaches to achieve the best predictions of which women would deliver low birth weight infants. Alcohol and
drug use were also important. The DSM method correctly identified 74 of 78 women who did not deliver low
birth weight infants compared to 69 of 78 for DFA. It correctly identified 16 of the 22 low birth weight (LBW)
infants compared to 11 of 22 for DFA. Both models performed better than existing models in the literature which
do not consider psychosocial measures or drug and alcohol consumption. DSM performed better than DFA. Its
sensitivity was 80%; efficiency was 90%; and specificity, 92%. DSM’s a prior specification of variables included
both independent contributions of psychosocial factors to medical risk and modulating effects of psychosocial
factors on medical risks. Conclusions: Psychosocial variables and alcohol and drug use measures permitted
significant improvement in the ability to predict risk for low birth weight. Dynamic systems modeling was better
than DFA and should be further explored and developed to create an explanatory theory for low birth weight
which could be used to guide a clinical trial of psychosocial intervention its prevention. KEY WORDS: Low birth
weight, psychosocial variables, discriminant function analysis (DFA), dynamic systems modeling (DSM).