Predicting fetal alcohol spectrum disorders in preschool-aged children from early life factors.

BACKGROUND: Early life factors, including parental sociodemographic characteristics, pregnancy exposures, and physical and neurodevelopmental features measured in infancy are associated with fetal alcohol spectrum disorders (FASD). The objective of this study was to evaluate the performance of a classifier model for diagnosing FASD in preschool-aged children from pregnancy and infancy-related characteristics. METHODS: We analyzed a prospective pregnancy cohort in Western Ukraine enrolled between 2008 and 2014. Maternal and paternal sociodemographic factors, maternal prenatal alcohol use and smoking behaviors, reproductive characteristics, birth outcomes, infant alcohol-related dysmorphic and physical features, and infant neurodevelopmental outcomes were used to predict FASD. Data were split into separate training (80%: n = 245) and test (20%: n = 58; 11 FASD, 47 no FASD) datasets. Training data were balanced using data augmentation through a synthetic minority oversampling technique. Four classifier models (random forest, extreme gradient boosting [XGBoost], logistic regression [full model] and backward stepwise logistic regression) were evaluated for accuracy, sensitivity, and specificity in the hold-out sample. RESULTS: Of 306 children evaluated for FASD, 61 had a diagnosis. Random forest models had the highest sensitivity (0.54), with accuracy of 0.86 (95% CI: 0.74, 0.94) in hold-out data. Boosted gradient models performed similarly, however, sensitivity was less than 50%. The full logistic regression model performed poorly (sensitivity = 0.18 and accuracy = 0.65), while stepwise logistic regression performed similarly to the boosted gradient model but with lower specificity. In a hold-out sample, the best performing algorithm correctly classified six of 11 children with FASD, and 44 of 47 children without FASD. CONCLUSIONS: As early identification and treatment optimize outcomes of children with FASD, classifier models from early life characteristics show promise in predicting FASD. Models may be improved through the inclusion of physiologic markers of prenatal alcohol exposure and should be tested in different samples.

Alcohol-related dysmorphic features as predictors of neurodevelopmental delay in infants and preschool-aged children: Results from a birth cohort in Ukraine.

BACKGROUND: Cardinal and non-cardinal dysmorphic features are associated with prenatal alcohol exposure (PAE); however, their association with neurodevelopment is less clear. The objective of this study was to determine whether alcohol-related dysmorphic features predict neurodevelopmental delay in infants and toddlers. METHODS: We analyzed a prospective pregnancy cohort in western Ukraine enrolled between 2008 and 2014. A dysmorphology examination comprising body size and three cardinal and 14 non-cardinal dysmorphic features was performed at approximately 6 to 12 months of age. PAE was self-reported and operationalized as absolute ounces of alcohol per day around the time of conception. Neurodevelopment was assessed at 6 to 12 months with the Bayley Scales of Infant Development-II (BSID-II), and at 3.5 to 4.5 years of age with the Differential Ability Scales-II, the Child Behavior Checklist, and multiple measures that were used to create an executive functioning factor score. We performed logistic regression to predict children's neurodevelopment from dysmorphic features, growth measures, sex, and PAE. RESULTS: From an analytic sample of 582 unique children, 566 had BSID-II scores in infancy, and 289 completed the preschool battery. Models with all cardinal and non-cardinal dysmorphic features, growth measures, sex, and PAE performed better than models with subsets of those inputs. In general, models had poor performance classifying delays in infancy (area under the curve (AUC) <0.7) and acceptable performance on preschool-aged outcomes (AUC ~0.75). When the sample was limited to children with moderate-to-high PAE, predictive ability improved on preschool-aged outcomes (AUC 0.76 to 0.89). Sensitivity was relatively low for all models (12% to 63%), although other metrics of performance were higher. CONCLUSION: Predictive analysis based on dysmorphic features and measures of growth performed modestly in this sample. As these features are more reliably measured than neurodevelopment at an earlier age, the inclusion of dysmorphic features and measures of growth in predictive models should be further explored and validated in different settings and populations.

Patterns of Prenatal Alcohol Exposure and Alcohol-Related Dysmorphic Features.

BACKGROUND: In animal models, it is possible to induce different alcohol-related dysmorphic abnormalities based on the timing of prenatal alcohol exposure (PAE). Our objective was to assess whether patterns of PAE differentially predict alcohol-related dysmorphic features in 415 infants. METHODS: We analyzed a prospective pregnancy cohort in western Ukraine enrolled between 2008 and 2014. Five distinct trajectories were previously identified to summarize PAE: (i) minimal/no PAE (n = 253), (ii) low/moderate PAE with reduction early in gestation (n = 78), (iii) low/moderate sustained PAE (n = 20), (iv) moderate/high PAE with reduction early in gestation (n = 45), and (v) high sustained PAE (n = 19). A dysmorphology examination of body size, 3 cardinal, and 15 noncardinal dysmorphic features was performed at approximately 6 to 12 months of age. A modified dysmorphology score was created based on previously published weights. Univariate comparisons were made between each dysmorphic feature and trajectory group. Features that differed by trajectory group were assessed in multivariable analyses. Models were adjusted for maternal age, prenatal vitamin use, socioeconomic status, smoking, and child's age at dysmorphology examination, with censoring weights for losses to follow-up. RESULTS: The 3 highest trajectories predicted total dysmorphology score, with larger effects in sustained exposure groups. Cardinal features: The 3 highest trajectories were each associated with a 2- to 3-fold increased risk of having 2 + cardinal facial features. When assessed individually, there were no consistent associations between the individual trajectories and each cardinal feature. Noncardinal features: The 3 highest trajectories were associated with increased risk of hypotelorism. Only the highest trajectory was associated with heart murmur. The highest trajectory predicted <10th centile for sex and age on height, weight, and head circumference; and moderate/high with reduction trajectory also predicted height. CONCLUSIONS: While we did not observe differential results based on specific trajectories of exposure, findings support the wide range of dysmorphic features associated with PAE, particularly at high and sustained levels.