Extensive evidence from animal model and epidemiological studies has linked prenatal alcohol exposure (PAE) to a broad range of cognitive and behavioral deficits, growth impairment, and physical anomalies. However, to date there has been no systematic attempt to use sophisticated meta-analytic techniques to integrate data across studies to improve identification of affected individuals, and virtually no information is available regarding the levels of exposure associated with an increased risk of clinically meaningful adverse effects.
The aim of this study is integrate extensive data collected from five large prospective longitudinal cohorts to better define the nature of the adverse effects associated with PAE and to derive more reliable and robust estimates of effect size and critical dose. Although these five studies were conducted independently, there was substantial convergence in the approaches used to measure exposure and developmental outcomes. Three complementary approaches?Bayesian hierarchical meta-analysis, structural equation modeling, and growth curve modeling?will be used to evaluate effect size in four domains: cognitive function, externalizing behavior problems, growth, and physical anomalies. These analyses will also consider the degree to which (a) larger effect sizes are associated with greater average daily dose vs. dose/occasion; (b) larger effect sizes are associated with drinking earlier vs. later in pregnancy; (c) effect sizes within these domains are stable or increase or attenuate across development; (d) effect sizes differ for different aspects of cognition (e.g., IQ, learning and memory, executive function) and behavioral function (e.g., aggression, social problems); and (e) effect sizes are moderated by maternal age at delivery, history of alcohol use disorders, and/or body mass index. In addition, we will (a) examine the shape of the dose-response curves to identify nonlinearities and inflection points that may suggest threshold effects; (b) use benchmark dose analysis to determine critical doses of PAE at which there is an increased likelihood of clinically significant adverse effects; and (c) evaluate the sensitivity and specificity of critical doses suggested by these analyses for predicting a range of developmental outcomes. This project will be conducted by a team of leading fetal alcohol researchers in collaboration with an internationally respected biostatistician. The proposal is responsive to NIAAA PAR-14- 338, Secondary Analyses of Existing Alcohol Epidemiology Data, which highlights the application ?of new analytic techniques and statistical methods for alcohol research? to ?currently available data sets? to ?help develop accurate measurement of?risk relationship and outcomes of alcohol consumption [and] to establish reliable and plausible thresholds.? Data from the proposed study will be critically important to the further development and refinement of the new tentative diagnostic criteria proposed in the Diagnostic and Statistical Manual of Mental Disorders, 5th ed., recognizing FASD for the first time as a ?condition in need of further study.? !
Although diagnosis of fetal alcohol spectrum disorders (FASD) requires a confirmed history of ?risky? maternal alcohol consumption during pregnancy, there is virtually no empirically-derived information in the scientific literature regarding levels of prenatal alcohol exposure that are reliably associated with clinically-meaningful adverse effects. The sophisticated meta-analyses planned for the proposed study will integrate data from five major U.S. longitudinal cohorts to provide new, more reliable estimates of effect size and dose that can be used to further refine diagnosis of FASD by (a) identifying the most sensitive endpoints seen in children with these disorders and (b) determining the levels and patterns of exposure most reliably associated with adverse effects. These data will provide information that will be critical for the further refinement of the new provisional diagnosis for FASD (designated as a ?condition in need of further study? in the latest edition of the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders) and will lay the ground work for development of new interventions more specifically targeted at the deficits that characterize these disorders. The work will also generate new statistical methodology and insights that can be applied in other epidemiological settings.
