The University of New Mexico, Center on Alcoholism, Substance Abuse, and Addictions (CASAA), Albuquerque, NM 87106, USA.
Drug Alcohol Depend. 2011 Dec 1;119(1-2):18-27. doi: 10.1016/j.drugalcdep.2011.05.009. Epub 2011 Jun 11.
Previous research in South Africa revealed very high rates of fetal alcohol syndrome (FAS), of 46-89 per 1000 among young children. Maternal and child data from studies in this community summarize the multiple predictors of FAS and partial fetal alcohol syndrome (PFAS).
Sequential regression was employed to examine influences on child physical characteristics and dysmorphology from four categories of maternal traits: physical, demographic, childbearing, and drinking. Then, a structural equation model (SEM) was constructed to predict influences on child physical characteristics.
Individual sequential regressions revealed that maternal drinking measures were the most powerful predictors of a child's physical anomalies (R² = .30, p < .001), followed by maternal demographics (R² = .24, p < .001), maternal physical characteristics (R²=.15, p < .001), and childbearing variables (R² = .06, p < .001). The SEM utilized both individual variables and the four composite categories of maternal traits to predict a set of child physical characteristics, including a total dysmorphology score. As predicted, drinking behavior is a relatively strong predictor of child physical characteristics (β = 0.61, p < .001), even when all other maternal risk variables are included; higher levels of drinking predict child physical anomalies.
Overall, the SEM model explains 62% of the variance in child physical anomalies. As expected, drinking variables explain the most variance. But this highly controlled estimation of multiple effects also reveals a significant contribution played by maternal demographics and, to a lesser degree, maternal physical and childbearing variables.
南非之前的研究显示，幼儿中胎儿酒精综合征（FAS）的发病率非常高，达到每 1000 名儿童中有 46-89 名。该社区研究中的母婴数据总结了 FAS 和部分胎儿酒精谱系障碍（PFAS）的多种预测因素。
采用序列回归分析考察了母亲身体、人口统计学、生育和饮酒等四个方面的特征对儿童身体特征和畸形的影响。然后，构建了一个结构方程模型（SEM）来预测儿童身体特征的影响因素。
个体序列回归分析显示，母亲饮酒量是儿童身体异常的最有力预测因素（R² =.30，p <.001），其次是母亲人口统计学特征（R² =.24，p <.001）、母亲身体特征（R² =.15，p <.001）和生育变量（R² =.06，p <.001）。SEM 利用个体变量和母亲特征的四个综合类别来预测一系列儿童身体特征，包括总畸形评分。正如预测的那样，饮酒行为是儿童身体特征的一个相对较强的预测因素（β = 0.61，p <.001），即使在考虑所有其他母亲风险变量的情况下也是如此；饮酒水平越高，儿童身体异常的可能性越大。
总体而言，SEM 模型解释了儿童身体异常的 62%的方差。正如预期的那样，饮酒变量解释了最大的方差。但这种对多种影响的高度控制评估也揭示了母亲人口统计学特征和一定程度上的母亲身体和生育变量的显著贡献。
