Development and validation of a postnatal risk score that identifies children with prenatal alcohol exposure

Fetal alcohol spectrum disorders prevention and clinical guidelines research - workshop report

It is estimated that up to 1 in 20 people in the United States may have a fetal alcohol spectrum disorder (FASD), or the array of physical, cognitive, emotional, and social disorders caused by exposure to alcohol during prenatal development (May et al., JAMA 319:474-82, 2018). While this condition is present in a broad range of individuals and families, it has not previously been examined in the military community, where cultural factors including an increased prevalence of alcohol misuse may pose a unique set of challenges (Health.mil, Alcohol misuse, 2024).The Uniformed Services University of the Health Sciences (USUHS), in conjunction with FASD United, hosted the second annual Workshop on Fetal Alcohol Spectrum Disorders Prevention and Clinical Guidelines Research on 20 September 2023 in Washington, DC. Organized as part of a four-year, federally-funded health services research initiative on FASD in the U.S. Department of Defense (DoD) Military Health System (MHS), the workshop provided a forum for exploring the initiative's focus and progress; examining current knowledge and practice in the research and clinical spheres; and identifying potential strategies to further improve prevention, screening, diagnosis, interventions, and family support. Building off of the 2022 workshop that covered the state of the science surrounding prenatal alcohol exposure and FASD, the 2023 focused primarily on FASD and efforts aimed at identification and management (Koehlmoos et al., BMC Proc 17 Suppl 12:19, 2023). One hundred and thirty attendees from academia, healthcare, federal agencies, and patient advocacy organizations gathered to share research findings; learn from lived experiences; and discuss initiatives to advance research, screening, and services for at-risk pregnant women as well as families and caregivers supporting individuals with FASD.

Machine learning approaches to the identification of children affected by prenatal alcohol exposure: A narrative review

Fetal alcohol spectrum disorders (FASDs) affect at least 0.8% of the population globally. The diagnosis of FASD is uniquely complex, with a heterogeneous physical and neurobehavioral presentation that requires multidisciplinary expertise for diagnosis. Many researchers have begun to incorporate machine learning approaches into FASD research to identify children who are affected by prenatal alcohol exposure, including those with FASD. This narrative review highlights these efforts. Following an introduction to machine learning, we summarize examples from the literature of neurobehavioral screening tools and physiologic markers of exposure. We discuss individual efforts, including models that classify FASD based on parent-reported neurocognitive or behavioral questionnaires, 3D facial imaging, brain imaging, DNA methylation patterns, microRNA profiles, cardiac orienting response, and dysmorphic facial features. We highlight model performance and discuss the limitations of these approaches. We conclude by considering the scalability of these approaches and how these machine learning models, largely developed from clinical samples or highly exposed birth cohorts, may perform in the general population.

Results of a screening tool for fetal alcohol spectrum disorders are associated with neuropsychological and behavioral measures

PURPOSE

This study assessed whether the outcome of a screening tool for fetal alcohol spectrum disorders (FASD), the FASD-Tree, was associated with neuropsychological and behavioral outcomes.

METHODS

Data for this study were collected as part of the fourth phase of the Collaborative Initiative on Fetal Alcohol Spectrum Disorders (CIFASD-4). Participants (N = 175, 5 to 16 years) with or without histories of prenatal alcohol exposure were recruited from San Diego and Minneapolis. Each participant was screened using the FASD-Tree and administered a neuropsychological test battery; parents or guardians completed behavioral questionnaires. The FASD-Tree incorporates physical and behavioral measures and provides an outcome regarding the presence of FASD (FASD-Positive or FASD-Negative). Logistic regression was used to test whether the FASD-Tree outcome was associated with general cognitive ability, executive function, academic achievement, and behavior. Associations were tested in two groups: the whole sample and only correctly classified participants.

RESULTS

Results of the FASD-Tree were associated with neuropsychological and behavioral measures. Participants classified as FASD-Positive were more likely than those classified as FASD-Negative to have a lower IQ score and exhibit poorer performance on measures of executive and academic functions. Behaviorally, participants classified as FASD-Positive were rated as having more behavior problems and adaptive difficulties. Similar relationships were found for all measures when including only participants correctly classified by the FASD-Tree screening tool.

CONCLUSION

Results from the FASD-Tree screening tool were associated with neuropsychological and behavioral measures. Participants classified as FASD-Positive were more likely to have impairment in all domains tested. The results support the effectiveness of the FASD-Tree as a screening tool for use in clinical settings, providing an efficient and accurate way to identify patients in need of additional evaluation.

Predicting Fetal Alcohol Spectrum Disorders Using Machine Learning Techniques: Multisite Retrospective Cohort Study

BACKGROUND

Fetal alcohol syndrome (FAS) is a lifelong developmental disability that occurs among individuals with prenatal alcohol exposure (PAE). With improved prediction models, FAS can be diagnosed or treated early, if not completely prevented.

OBJECTIVE

In this study, we sought to compare different machine learning algorithms and their FAS predictive performance among women who consumed alcohol during pregnancy. We also aimed to identify which variables (eg, timing of exposure to alcohol during pregnancy and type of alcohol consumed) were most influential in generating an accurate model.

METHODS

Data from the collaborative initiative on fetal alcohol spectrum disorders from 2007 to 2017 were used to gather information about 595 women who consumed alcohol during pregnancy at 5 hospital sites around the United States. To obtain information about PAE, questionnaires or in-person interviews, as well as reviews of medical, legal, or social service records were used to gather information about alcohol consumption. Four different machine learning algorithms (logistic regression, XGBoost, light gradient-boosting machine, and CatBoost) were trained to predict the prevalence of FAS at birth, and model performance was measured by analyzing the area under the receiver operating characteristics curve (AUROC). Of the total cases, 80% were randomly selected for training, while 20% remained as test data sets for predicting FAS. Feature importance was also analyzed using Shapley values for the best-performing algorithm.

RESULTS

Overall, there were 20 cases of FAS within a total population of 595 individuals with PAE. Most of the drinking occurred in the first trimester only (n=491) or throughout all 3 trimesters (n=95); however, there were also reports of drinking in the first and second trimesters only (n=8), and 1 case of drinking in the third trimester only (n=1). The CatBoost method delivered the best performance in terms of AUROC (0.92) and area under the precision-recall curve (AUPRC 0.51), followed by the logistic regression method (AUROC 0.90; AUPRC 0.59), the light gradient-boosting machine (AUROC 0.89; AUPRC 0.52), and XGBoost (AUROC 0.86; AURPC 0.45). Shapley values in the CatBoost model revealed that 12 variables were considered important in FAS prediction, with drinking throughout all 3 trimesters of pregnancy, maternal age, race, and type of alcoholic beverage consumed (eg, beer, wine, or liquor) scoring highly in overall feature importance. For most predictive measures, the best performance was obtained by the CatBoost algorithm, with an AUROC of 0.92, precision of 0.50, specificity of 0.29, F1 score of 0.29, and accuracy of 0.96.

CONCLUSIONS

Machine learning algorithms were able to identify FAS risk with a prediction performance higher than that of previous models among pregnant drinkers. For small training sets, which are common with FAS, boosting mechanisms like CatBoost may help alleviate certain problems associated with data imbalances and difficulties in optimization or generalization.

Identifying Prenatal Alcohol Exposure and Children Affected by It: A Review of Biomarkers and Screening Tools

PURPOSE

Early identification of prenatal alcohol exposure (PAE) and of those in need of services resulting from this exposure is an important public health concern. This study reviewed the existing literature on potential biomarkers and screening tools of PAE and its impact.

SEARCH METHODS

Electronic databases were searched for articles published between January 1, 1996, and November 30, 2021, using the following search terms: ("fetal alcohol" or "prenatal alcohol" or "FASD" or "alcohol-related neurodevelopmental disorder" or "ARND" or "ND-PAE") and ("screening" or "identification" or "biomarker"). Duplicate articles were electronically eliminated. Titles and abstracts were reviewed for appropriateness, and selected articles were retrieved for further analysis. Additional articles were added that were referenced in the reviewed articles or identified from expert knowledge. Information about the characteristics of the sample, the biomarker or screening tool, and the predictive validity outcome data were abstracted. A narrative analysis of the studies was then performed on the data.

SEARCH RESULTS

A total of 3,813 articles were initially identified, and 1,215 were removed as duplicates. Of the remaining articles, 182 were identified as being within the scope of the review based on title and abstract inspection, and 181 articles were successfully retrieved. Of these, additional articles were removed because they were preclinical (3), were descriptive only (13), included only self-report of PAE (42), included only mean group comparison (17), were additional duplicates (2), focused on cost analysis (9), missed predictive validity data (24), or for other reasons (23). The remaining articles (n = 48) were abstracted. An additional 13 manuscripts were identified from these articles, and two more from expert knowledge. A total of 63 articles contributed to the review.

DISCUSSION AND CONCLUSIONS

Biomarkers and screening tools of PAE and its impact fall short of ideal predictive validity characteristics. Higher specificity than sensitivity was found for many of the biomarkers and screening tools used to identify PAE and its impact, suggesting that current methods continue to under-identify the full range of individuals impacted by PAE. Exceptions to this were found in recent investigations using microRNAs related to growth and vascular development, proteomic changes associated with PAE, and combinations of markers estimating levels of various cytokines. Replications of these findings are needed across other samples to confirm the limited data available. Future research on biomarkers and screening tools should attend to feasibility and scalability of implementation. This article also recommends a systematic process of evaluation to improve early identification of individuals impacted by PAE so that harm reduction and habilitative care efforts can be implemented.

Current considerations for fetal alcohol spectrum disorders: identification to intervention

PURPOSE OF REVIEW

This review highlights recent findings regarding the prevalence, public health impact, clinical presentation, intervention access and conceptualization of fetal alcohol spectrum disorders (FASDs). Despite ongoing work in prevention and identification of this population, the rates of drinking during pregnancy have increased and significant gaps remain in diagnosis and intervention.

RECENT FINDINGS

Prenatal alcohol exposure is the most common preventable cause of developmental disability in the world. Research has focused on improving diagnostic clarity, utilizing technology and neuroimaging to facilitate identification, engaging broader stakeholders (including self-advocates) to inform understanding and needs, and increasing access to effective interventions. There is an emerging focus on developmental trajectories and experiences in young and middle adulthood. Public policy advocacy has also made great strides in recent years.

SUMMARY

Increases in public awareness, greater concordance of diagnostic schema, leveraged use of novel technology, and the development of targeted interventions within a holistic, strengths-based conceptualization are important considerations for this population.

Racial/Ethnic Disparity in Association Between Fetal Alcohol Syndrome and Alcohol Intake During Pregnancy: Multisite Retrospective Cohort Study

BACKGROUND

Alcohol consumption during pregnancy is associated with a range of adverse birth-related outcomes, including stillbirth, low birth weight, preterm birth, and fetal alcohol syndrome (FAS). With more than 10% of women consuming alcohol during pregnancy worldwide, it is increasingly important to understand how racial/ethnic variations affect FAS onset. However, whether race and ethnicity inform FAS risk assessment when daily ethanol intake is controlled for remains unknown.

OBJECTIVE

This study aimed to assess racial/ethnic disparities in FAS risk associated with alcohol consumption during pregnancy.

METHODS

We used data from a longitudinal cohort study (the Collaborative Initiative on Fetal Alcohol Spectrum Disorders) at 5 hospital sites around the United States of 595 women who consumed alcohol during pregnancy from 2007 to 2017. Questionnaires, in-person interviews, and reviews of medical, legal, and social service records were used to gather data on average alcoholic content (AAC) during pregnancy. Self-reports of maternal race (American Indian/Alaska Native [AI/AN], Asian, Native Hawaiian or other Pacific Islander, Black or African American, White, more than one race, and other) and ethnicity (Hispanic/Latino or not Hispanic/Latino), as well as FAS diagnoses based on standardized dysmorphological criteria, were used for analysis. Log-binomial regression was used to examine the risk of FAS associated with each 1-gram increase in ethanol consumption during pregnancy, stratified by race/ethnicity.

RESULTS

A total of 3.4% (20/595) of women who reported consuming alcohol during pregnancy gave birth to a baby with FAS. Women who gave birth to a baby with FAS had a mean AAC of 32.06 (SD 9.09) grams, which was higher than that of women who did not give birth to a baby with FAS (mean 12.07, SD 15.87 grams). AI/AN mothers with FAS babies had the highest AAC (mean 42.62, SD 8.35 grams), followed by White (mean 30.13, SD 4.88 grams) and Black mothers (mean 27.05, SD 12.78 grams). White (prevalence ratio [PR] 1.10, 95% CI 1.03-1.19), Black (PR 1.13, 95% CI 1.04-1.23), and AI/AN (PR 1.10, 95% CI 1.00-1.21) mothers had 10% to 13% increased odds of giving birth to a baby with FAS given the same exposure to alcohol during pregnancy. Regardless of race, a 1-gram increase in AAC resulted in a 4% increase (PR 1.04, 95% CI 1.02-1.07) in the chance of giving birth to a baby with ≥2 facial anomalies (ie, short palpebral fissures, thin vermilion border of the upper lip, and smooth philtrum) and a 4% increase (PR 1.04, 95% CI 1.01-1.07) in the chance of deficient brain growth.

CONCLUSIONS

The risk of delivering a baby with FAS was comparable among White, Black, and AI/AN women at similar levels of drinking during pregnancy. Regardless of race, a 1-gram increase in AAC resulted in increased odds of giving birth to a baby with facial anomalies or deficient brain growth.

Fetal alcohol spectrum disorders

Alcohol readily crosses the placenta and may disrupt fetal development. Harm from prenatal alcohol exposure (PAE) is determined by the dose, pattern, timing and duration of exposure, fetal and maternal genetics, maternal nutrition, concurrent substance use, and epigenetic responses. A safe dose of alcohol use during pregnancy has not been established. PAE can cause fetal alcohol spectrum disorders (FASD), which are characterized by neurodevelopmental impairment with or without facial dysmorphology, congenital anomalies and poor growth. FASD are a leading preventable cause of birth defects and developmental disability. The prevalence of FASD in 76 countries is >1% and is high in individuals living in out-of-home care or engaged in justice and mental health systems. The social and economic effects of FASD are profound, but the diagnosis is often missed or delayed and receives little public recognition. Future research should be informed by people living with FASD and be guided by cultural context, seek consensus on diagnostic criteria and evidence-based treatments, and describe the pathophysiology and lifelong effects of FASD. Imperatives include reducing stigma, equitable access to services, improved quality of life for people with FASD and FASD prevention in future generations.

Latent classes of neurodevelopmental profiles and needs in children and adolescents with prenatal alcohol exposure

BACKGROUND

Fetal alcohol spectrum disorder (FASD) resulting from prenatal alcohol exposure (PAE) is a common neurodevelopmental disorder, but substantial interindividual heterogeneity complicates timely and accurate assessment, diagnosis, and intervention. The current study aimed to identify classes of children and adolescents with PAE assessed for FASD according to their pattern of significant neurodevelopmental functioning across 10 domains using latent class analysis (LCA), and to characterize these subgroups across clinical features.

METHODS

Data from the Canadian National FASD Database, a large ongoing repository of anonymized clinical data received from diagnostic clinics across Canada, was analyzed using a retrospective cross-sectional cohort design. The sample included 1440 children and adolescents ages 6 to 17 years (M = 11.0, SD = 3.5, 41.7% female) with confirmed PAE assessed for FASD between 2016 and 2020.

RESULTS

Results revealed an optimal four-class solution. The Global needs group was characterized by high overall neurodevelopmental impairment considered severe in nature. The Regulation and Cognitive needs groups presented with moderate but substantively distinguishable patterns of significant neurodevelopmental impairment. The Attention needs group was characterized by relatively low probabilities of significant neurodevelopmental impairment. Both the Global and Regulation needs groups also presented with the highest probabilities of clinical needs, further signifying potential substantive differences in assessment and intervention needs across classes.

CONCLUSIONS

Four relatively distinct subgroups were present in a large heterogeneous sample of children and adolescents with PAE assessed for FASD in Canada. These findings may inform clinical services by guiding clinicians to identify distinct service pathways for these subgroups, potentially increasing access to a more personalized treatment approach and improving outcomes.

Validation of the FASD-Tree as a screening tool for fetal alcohol spectrum disorders

BACKGROUND

As many as 80% of individuals with fetal alcohol spectrum disorders (FASD) are misdiagnosed or not diagnosed. This study tests the accuracy and validity of a web-based screening tool (the FASD-Tree) for identifying children and adolescents with FASD.

METHODS

Children with histories of prenatal alcohol exposure (PAE) and controls (N = 302, including 224 with PAE and 78 controls) were examined for physical signs of fetal alcohol syndrome (FAS), and parents completed behavioral questionnaires. Data were entered into the FASD-Tree, a web-based decision tree application. The FASD-Tree provided two outcomes: a dichotomous indicator (yes/no) and a numeric risk score (0 to 5), which have been shown separately to identify children with PAE and neurobehavioral impairment and to correlate with neurobehavioral outcomes. Overall accuracy (ACC), sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated for the decision tree, risk score, and their combination. Misclassified cases were examined for systematic bias.

RESULTS

The FASD-Tree was successful in accurately identifying youth with histories of PAE and the subgroup of individuals with FASD, indicating its validity as an FASD screening tool. Overall accuracy rates for FASD-Tree components ranged from 75.0% to 84.1%, and both the decision tree outcome and risk score, and their combination, resulted in fair to good discrimination (area under the curve = 0.722 to 0.862) of youth with histories of PAE or FASD. While most participants were correctly classified, those who were misclassified differed in IQ and attention. Race, ethnicity, and sex did not affect the results.

CONCLUSION

The FASD-Tree is not a biomarker of PAE and does not provide definitive evidence of prenatal alcohol exposure. Rather it is an accurate and valid screening tool for FASD and can be used by clinicians who suspect that a patient has a history of PAE, even if the exposure is unknown.