Validation of the ND-PAE Diagnosis in Children with Heavy Prenatal Alcohol Exposure

This study evaluated criteria for Neurobehavioral Disorder Associated with Prenatal Alcohol Exposure (ND-PAE). Kable et al. (2022) assessed the validity of this diagnosis in a sample with low exposure to alcohol. The current study expanded this assessment to a sample with a wider age range and heavier alcohol exposure. Data were collected from participants (5-17y) with prenatal alcohol exposure (PAE) and typically developing controls at six Collaborative Initiative on Fetal Alcohol Spectrum Disorders sites using neuropsychological assessment and caregiver reports. Impairment was tested at 1SD, 1.5SD, and 2SD below the normative average and a modification of the adaptive functioning requirement was tested. Testing impairment at 1SD resulted in the highest endorsement rates in both groups. Our findings replicated the study by Kable et al. and show that current criteria captured a high rate of those with PAE and that requiring fewer adaptive functioning criteria resulted in higher sensitivity to PAE.


INTRODUCTION
Prenatal alcohol exposure (PAE) may lead to fetal alcohol spectrum disorders (Hoyme et al., 2016) which include fetal alcohol syndrome (FAS), partial FAS (PFAS), alcohol-related neurodevelopmental disorder (ARND), and alcohol-related birth defects (ARBD).As a group, these disorders are associated with neurobehavioral de cits, growth de cits, facial dysmorphology, and/or other medical symptoms (Hoyme et al., 2016).Recent prevalence studies have suggested that FASD affects 1.1 to 5% of children in the U.S. (May et al., 2018), and it is one of the leading causes of birth defects and neurobehavioral de cits in children (Williams et al., 2015), even more prevalent than autism spectrum disorder (Maenner et al., 2020).Despite these prevalence rates, diagnosis is particularly di cult due to confounding factors, such as incomplete or inaccurate self-report, lack of biomarkers, and comorbidities with other diagnoses (i.e., ADHD; Wozniak et al., 2019), and delays in diagnosis can lead to an increased risk of adverse outcomes in childhood (Streissguth et al., 2004).Diagnosis of FAS and PFAS is made, in part, based on facial dysmorphology (Del Campo & Jones, 2017) which can increase certainty of exposure and outcome.However, the majority of individuals with PAE do not manifest physical markers of their exposure (Wozniak et al., 2019), limiting the diagnostic utility of these features.Further, neurobehavioral challenges are present in individuals with PAE even in the absence of the characteristic physical features, and studies have shown that alcohol-exposed individuals with and without physical features have de cits in neuropsychological domains such as IQ, motor functioning, and learning (Mattson & Riley, 1998;Mattson et al., 1997).Neurobehavioral di culties associated with FASD encompass cognition, behavior, and mental health.Cognitively, people with FASD experience de cits in general intelligence, executive functioning, attention, learning, memory, motor skills, and language (Mattson et al., 2019).FASD is associated with an elevated risk of mental health disorders, such as anxiety and depression (Fryer et al., 2007;Hellemans et al., 2010) as well as externalizing disorders such as attention-de cit/hyperactivity disorder, conduct disorder, and oppositional de ant disorder (Fryer et al., 2007).Some studies have shown that fewer dysmorphology features is associated with more behavioral problems (Fagerlund et al., 2011); however, other studies have shown that these symptoms occur at the same rate with and without physical features (Mattson & Riley, 2000).Therefore, creating a diagnosis that speci cally re ects neurodevelopmental and mental health symptoms of PAE, rather than physical features, will increase early diagnosis and intervention, leading to fewer adverse outcomes.
In the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), an FASD diagnosis termed neurodevelopmental disorder associated with prenatal alcohol exposure (ND-PAE) was added under "Conditions for Further Study" to better detail the neurodevelopmental and mental health symptoms associated with PAE that are not captured with current FASD criteria (American Psychiatric Association, 2013, 2022; Kable et al., 2016).In contrast to other diagnoses described under the FASD diagnostic umbrella which are medical diagnoses, ND-PAE is a mental health diagnosis and is not reliant on a dysmorphology evaluation (Hoyme et al., 2016;Kable et al., 2016).In contrast to the ARND diagnosis, the ND-PAE diagnosis, which is based on de cits in neurocognition, self-regulation, and adaptive functioning, is independent of facial dysmorphology and can be made in individuals with or without FAS or PFAS (Hoyme et al., 2016).Like ARND, a diagnosis of ND-PAE requires con rmed PAE (Hoyme et al., 2016;Kable et al., 2016).The ND-PAE diagnosis also increases the number of individuals who have the expertise needed to diagnose and treat an individual with PAE.ND-PAE criteria are categorized into three domains: neurocognitive, self-regulation, and adaptive functioning, and symptoms in all domains must rst appear in childhood (American Psychiatric Association, 2013, 2022).For neurocognitive impairment (NI), impairment must be present in one of the following categories: global intellectual functioning (which is represented by an IQ or parallel assessment of 70 or below), executive functioning, learning, memory, and visual spatial reasoning.For self-regulation (SR), impairment must be seen in mood or behavioral regulation, attention, and/or impulse control.Finally, in the adaptive functioning (AF) domain, de cits must be seen in either communication (i.e., acquiring and understanding language) or social communication with one of the following: impairment in daily living or motor impairment.However, with the exception of global intellectual functioning (an IQ of 70 or below or a score of 70 or below on a similar comprehensive development assessment), the threshold for impairment in these domains is not clearly de ned (Kable & Coles, 2018).
Previous studies have assessed the validity of the diagnosis and have found that the AF domain may be too restrictive (Kable & Coles, 2018).Speci cally, requiring two out of four criteria for the AF domain (AF 2/4) may not be necessary, and requiring that one of the two rst criteria for adaptive functioning (communication and social communication) is not supported by the literature and therefore may lead to low sensitivity, or fewer true positives (Sanders et al., 2020).Sanders et al. (2020) also evaluated construct and factorial validity of the ND-PAE diagnosis and found through principal component analysis (PCA) that symptoms only loaded on one factor, which included neurocognitive and adaptive functioning criteria.Their assessment of convergent and divergent validity showed that convergence, or the correlation among symptoms, was low among SR symptoms, speci cally the correlation between mood regulation and attention symptoms (Sanders et al., 2020).Due to the unspeci ed impairment threshold in the DSM-5, previous studies have assessed symptom severity at 1.0, 1.5, and 2.0 standard deviations (SDs) from the mean (Kable & Coles, 2018;Sanders et al., 2020).Internal validity analyses have shown that the symptoms have a fair internal consistency at both 1.0 and 1.5 SDs (Sanders et al., 2020).However, these studies have small sample sizes (n < 58).
A recent paper by Kable et al. (2022) evaluated the ND-PAE diagnosis in a community sample of over 1000 1st-grade children (5-7y) as a part of the Collaboration on FASD Prevalence (CoFASP) study.Participants were categorized as "at-risk" if they were exposed to more than minimal (> 13 drinks per month or 2 or more drinks on one occasion) amounts of alcohol prenatally or if they exhibited physical de cits such as facial dysmorphology or growth delays.This at-risk group was compared to a no risk group, categorized as not having any PAE and no related physical de cits (i.e., growth de cits and facial dysmorphology).They assessed criteria at both 1.0 and 1.5 SDs and assessed two different AF criteria for impairment: only one AF criterion needed (AF 1) or the original AF 2/4 criteria.Memory impairment for the NI domain was not assessed because it was not measured in all of the participants of this sample.
Results showed higher endorsement of the ND-PAE diagnosis using 1.0 SD than 1.5 SD as the indicator of impairment and signi cant differences in ND-PAE endorsement between the at-risk and control groups at 1.0 SD.In the AF domain, having one necessary criterion (AF 1) led to a signi cant difference between the at-risk and no risk groups in AF endorsement at both 1.0 and 1.5 SDs, in contrast to the AF 2/4 criteria, which only had a signi cant difference between groups at the 1.0 SD cutoff.PCA was also performed to determine categories for the symptoms, which resulted in two factors: one with self-regulation, adaptive social skills, and adaptive independent living skills, and the other with IQ, visual-spatial functioning, adaptive communication, and adaptive motor skills.The authors used receiver operating characteristic (ROC) curves to test if each item and domain contributed to discrimination in ND-PAE endorsement (sensitivity and speci city).ROC curves produce area under the curve (AUC) values with values under 0.70 considered poor and values over 0.90 considered excellent.The results indicated that having an IQ score below 70 (i.e., 2.0 SD) did not signi cantly contribute to the diagnosis with either AF model in either of the groups (AUC < 0.60) and that the NI domain had poor prediction of the diagnosis with either AF model at 1.0 SD (AUC < 0.70).Therefore, they assessed IQ at 1.0 and 1.5 SDs and found that global intellectual impairment better predicted ND-PAE categorization at both 1.0 and 1.5 SD levels.Overall, the endorsement for the diagnosis was low among the at-risk group, which may be because these participants were recruited from the community instead of a healthcare facility and therefore may not be experiencing symptoms at the same rate or intensity as more heavily exposed individuals.Gender also seemed to play a role, as males had higher endorsement of symptoms than females in their sample.
Additional validation for this diagnosis is needed in a larger, more heavily exposed sample.The purpose of the current study is to evaluate the ND-PAE criteria in a sample with heavier exposure and a wider age range.

Participants
Data were collected from participants with and without PAE at six sites (San Diego, Los Angeles, Northern Plains, Atlanta, New Mexico, and Minnesota) as part of an ongoing study for the Collaborative Initiative on Fetal Alcohol Spectrum Disorders (CIFASD), phases 2 through 4. Participants were categorized into two groups: an alcohol-exposed group (AE; n = 486) and a control group (CON; n = 679) with minimal or no PAE.General methodology for the CIFASD clinical projects has been described previously (see Jones et al., 2006;Mattson et al., 2010) Histories of PAE were collected from parent questionnaires as well as from medical, legal, and social service records (Mattson et al., 2010) at each site.Participants were included in the AE group if they had con rmed reports of heavy (de ned as an average of 14 or more standard alcoholic drinks per week or 4 or more drinks per occasion during pregnancy) or moderate (de ned as an average of 1-13 drinks per week or 2-3 drinks per occasion during pregnancy) alcohol exposure.Participants were also included in the AE group if exposure was reasonably suspected based on documentation indicating signi cant problems related to alcohol use during pregnancy.Participants were included in the CON group if they had a con rmed lack of PAE or minimal exposure (de ned as an average of < 1 drink per week and never more than 2 drinks per occasion during pregnancy; see Control group from Mattson et al., 2010).While recruitment varied slightly from phase to phase, all CIFASD phases included a heterogeneous control group that did not exclude participants based on other diagnosis (e.g., ADHD), clinical concerns or conditions, or receipt of special educational or social services (see Goh et al., 2016;Mattson et al., 2010;Mattson et al., 2023).Participants were excluded from analysis if they were missing data required for grouping (e.g., exposure information) or if they were entirely missing data for one or more of the domains.

Neuropsychological Assessment
As part of the larger CIFASD study, children received a standardized neuropsychological battery administered by a trained examiner who was blind to exposure and diagnosis (see Mattson et al., 2010).Parents or caregivers completed standardized questionnaires about medical and social history and child behavior.
The test battery varied slightly depending on the phase of the CIFASD study, however all phases included measures of the domains required for the ND-PAE diagnosis: neurocognition, self-regulation, and adaptive function.Thus, not all participants had all measures, but all had at least one measure per domain, depending on criteria necessary for ND-PAE diagnosis (one criterion required for AF 1 versus two criteria required for AF 2/4).Criteria were assessed using neuropsychological testing measures and caregiver reports that best captured the symptoms of ND-PAE (see Doyle & Mattson, 2015).Individual measures were mapped to each domain required for the ND-PAE diagnosis, as detailed in Table 1.
Neurocognitive Domain.Neurocognitive impairment (NI) was captured using measures of global intellectual functioning, executive functioning, learning, visuospatial reasoning, and memory.For impairment in global intellectual functioning, the General Conceptual Ability (GCA) score from the Differential Ability Scales, Second Edition (DAS-II; Elliot, 2007) and the Full Scale IQ (FSIQ) score from the Wechsler Intelligence Scale for Children, Fourth and Fifth Edition (WISC-IV; WISC-V; Wechsler, 2003Wechsler, , 2014) ) were used.The original ND-PAE criteria specify that global intellectual functioning, captured by methods such as IQ, should be a standard score of 70 or below to be considered impaired (2.0 SD levels below the normative average).Following Kable et al. (2022), IQ was also evaluated at 1.0 and 1.5 SD levels below the normative average.For the remaining NI criteria, impairment was assessed using measures from the Behavior Rating Inventory of Executive Function (BRIEF; Gioia et al., 2000), the Cambridge Neuropsychological Test Automated Battery (Cambridge Automated Neuropsychological Test Assessment Battery (CANTAB), 1999), the Child Behavior Checklist (CBCL; Achenbach, 1991) Adaptive Functioning Domain.Adaptive functioning (AF) was captured using measures of communication, social skills, daily living skills, and motor functioning.Impairment in this domain was assessed using subtests from the BASC-3, CBCL, and VABS-II/III.

Statistical Analysis
Each test was administered and scored according to standardized procedures provided by the test publisher.Prior to analysis, data were converted to z scores to re ect standard deviations to put all data on the same scale (with the exception of the DBD, which only outputs a "1" for impairment or "0" for no impairment).Higher (positive) z scores indicated stronger scores and lower (negative) z-scores indicated weaker scores.For example, IQ scores were converted from standard scores (M = 100; SD = 15) to z-scores (M = 0; SD = 1).All tests were corrected for age and/or sex, depending on the measure based on publisher-provided normative data.
Participants were each given scores for endorsement of the NI, SR, and AF domains at each SD level (i.e., 1.0, 1.5, and 2.0), as well as for the overall diagnosis with two AF criteria required (AF 2/4) or just one (AF 1).If the participant scored in the impairment range for any of the measures used for each criterion, they were coded as impaired for that criterion; if they did not score as impaired in any of the measures for a criterion, they were coded as "0".Following the proposed ND-PAE diagnostic criteria, if a participant received a "1" in any of the criteria for the NI or SR, they were scored as impaired for that domain.For the AF 2/4 domain, participants had to be scored as "1" in the communication or social communication (AF criteria 1 and 2) measures and one of the other two measures.For the AF 1 domain, participants were required to be scored as "1" in only one of the following AF criteria: communication, social communication, or impairment in daily living skills.Participants with impairment in the NI domain, SR domain, and one of the AF domains (AF 2/4 or AF 1) were categorized as ND-PAE.It is important to note that although we examined whether each individual met criteria for ND-PAE, in clinical settings the diagnosis of ND-PAE requires knowledge of PAE.
ROC curves were computed for each criterion and each domain at 1.0, 1.5, and 2.0 SD levels using SPSS v. 28.0.1.1.ROC analysis was also performed for ND-PAE risk at all three SD levels.Internal consistency among the symptoms was calculated using Cronbach's alpha.Principal component analysis (PCA) was used to assess how factors load onto each domain at the different impairment levels (1.0, 1.5, and 2.0 SDs).

Demographics
Participant demographics are listed in 1 IQ is from test data, not based on parent report.

ND-PAE Endorsement
See Table 3 for ND-PAE symptom and domain endorsement rates for each SD cutoff level.Using the original ND-PAE criteria with two of four AF criteria required (AF 2/4), 69.2% of participants from the AE group were categorized as ND-PAE (i.e., met criteria for ND-PAE excluding con rmed PAE) at the 1.0 SD level compared to 39.6% at the 1.5 SD level and 20.4% at the 2.0 SD level.When revising the AF criteria to only require one criterion (AF 1), 81.7% of the AE group were categorized as ND-PAE at the 1.0 SD level, compared to 52.5% at the 1.5 SD level and 32.4% at the 2.0 SD level.In the CON group, 22.6% of participants were identi ed as ND-PAE using AF 2/4 criteria at the 1.0 SD level, compared to 8.9% at the 1.5 SD level and 2.8% of participants at the 2.0 SD level.Using AF 1 criteria, 35.4% of the CON group were identi ed as ND-PAE at the 1.0 SD level, 15.2% at the 1.5 SD level, and 6.4% at the 2.0 SD level.There were signi cant differences between the AE and CON groups at each SD level in both the AF

Sensitivity, Speci city, and Accuracy Analyses
Speci city, sensitivity, and accuracy were calculated for both models at each SD level (see Table 4).The original ND-PAE criteria at a lowered impairment threshold produced the most accurate results (sensitivity: 69.1%, speci city: 77.5%, accuracy: 74.0%).However, the most sensitive outcome occurred using a lowered impairment threshold of 1.0 SD with the revised adaptive symptoms requirement (sensitivity: 81.7%), and the most speci c outcome occurred using the original adaptive symptoms requirement and higher impairment threshold of 2.0 (speci city: 98.5%).
Although our analyses con rmed that the AF 2/4 model at the 1.0 SD level produced the most accurate categorization results, we tested whether this model was improved by using the original IQ criteria (2.0 SD levels below the mean) increased speci city results while keeping all other criteria at the 1.0 SD level.
Results of this analysis were comparable to the AF 2/4 model with the impairment level for all criteria at the 1.0 SD level (sensitivity: 68.9%; speci city: 77.5%; accuracy: 73.9%).

Internal Consistency
Internal consistency among the ND-PAE criteria was assessed using the entire sample and then within each group (AE and CON) at each SD level.For the entire sample analysis, only 413 individuals were included in the analysis because they were not missing data for any of the ND-PAE criteria.Using the entire sample, Cronbach's alpha was 0.833 at the 1.0 SD level, 0.831 at the 1.5 SD level, and 0.812 at the 2.0 SD level.In the AE group (n = 178), Cronbach's alpha was 0.701 at the 1.0 SD level, 0.753 at the 1.5 SD level, and 0.766 at the 2.0 SD level.In the CON group (n = 233), Cronbach's alpha was 0.772 at the 1.0 SD level, 0.764 at the 1.5 SD level, and 0.686 at the 2.0 SD level.This suggests that there is moderate to strong intercorrelation among the ND-PAE criteria at each SD level with the exception of the CON group at the 2.0 SD level (Akoglu, 2018).

Principal Component Analysis
Principal Component Analysis (PCA) was performed using the 12 criteria at the 1.0 and 1.5 SD levels.Due to low accuracy levels from the sensitivity, speci city, and accuracy analyses, PCA was not performed for the 2.0 SD level as it was determined to not be a good cutoff for this diagnosis.A Promax rotation was used to allow for correlation among the symptoms.Factors were retained using Kaiser normalization and con rmed with parallel analysis (Patil et al., 2017).Only 413 participants were included in this analysis because they had data for all criteria.
At the 1.0 SD level, items that loaded on factor 1 included impairment in executive function skills (NI domain), impairment of mood and behavioral regulation symptoms (SR domain), impairment in attention skills (SR domain), and impairment in impulse control (SR domain); items that loaded on factor 2 included impairment in global intellectual functioning (NI domain), impairment in adaptive communication (AF domain), social impairment (AF domain), impairment in independent living skills (AF domain), and adaptive motor functioning impairment (AF domain); factor 3 included learning impairments (NI domain), impairments in memory functioning (NI domain), and impairment in memory functioning (NI domain).These results were also consistent at the 1.5 SD level with the exception of impairment in global intellectual functioning (NI domain), which shifted to factor 3.

Receiver Operating Curve (ROC) Analysis
ROC analysis was used to assess the predictive validity of both domain (see AUC values in Table 5) and individual criteria endorsement (see AUC values in  (Nahm, 2022).The impairment in adaptive motor skills symptoms was not included in this analysis for the AF 1 model because it was not used as a criterion for impairment.In the overall sample, each of the ND-PAE symptoms signi cantly contributed to the diagnosis in both models at all three SD levels.In the CON sample, all ND-PAE symptoms were signi cant predictors of diagnostic endorsement in both models at the 1.0 SD, but at the 1.5 SD and 2.0 SD levels, visual-spatial impairment was no longer a signi cant predictor for both models, and at the 2.0 SD level, global intellectual functioning also was no longer a signi cant predictor in the AF 2/4 model.In the AE group at the 1.0 SD level, all symptoms were signi cant predictors of being categorized as at-risk for ND-PAE with the exception of impairment in global intellectual functioning in the AF 2/4 model and impairment in memory and impairment in visual-spatial skills in both models.Visual-spatial impairment was also the only insigni cant symptom at the 1.5 SD level in both models.At the 2.0 SD level in the AE group, all ND-PAE symptoms in the AF 2/4 model except for motor skills were signi cant predictors of diagnosis.
At the 1.0 SD level, the NI domain had poor (< 0.70) predictive validity for both models and for the overall sample, CON group, and AE group.However, the predictive validity becomes fair (≥ 0.70) for the overall sample and CON group at the 1.5 SD level, and it becomes fair to good (≥ 0.70 and < 0.90) for all three groups at the 2.0 SD level.For the SR domain, predictive validity was fair (≥ 0.70) at the 1.0 SD level for the overall sample and CON group for the AF 1 model only.At the 1.5 and 2.0 SD levels, the SR domain had fair to good predictive validity (≥ 0.70 and < 0.90) for only the overall sample and CON group for both models.For the AF domain, predictive validity was fair to excellent (≥ 0.70 and > 0.90) for both models and at all three SD levels.
AUC values for individual criteria in the NI domain varied from poor (< 0.70) to fair (≥ 0.70) for both models among the different samples and SD levels.
Impairment in executive functioning was the only fair (≥ 0.70) predictor at all three SD levels but only in the overall sample and CON group and only within the AF 1 model.In addition, at the 2.0 SD level, impairment in executive functioning is also a fair predictor in the AE group in the AF 1 model and in the overall sample in the AF 2/4 model.At the 2.0 SD level, impairment in memory was also a fair predictor in the CON group only in the AF 2/4 model.
In the SR domain, both impairment in attention problems and impulse control had fair to good (≥ 0.70 and < 0.90) predictive validity at all three SD levels for the oversample and CON group for both models, with the exception of the AF 2/4 model in the overall sample at the 2.0 SD level.Impairment in attention problems and impulsivity also had fair predictive validity in the AE group at the 1.0 SD level but only in the AF 1 model.Impairment in mood or behavioral regulation was a fair predictor (≥ 0.70) of ND-PAE risk for the overall sample and CON group for the AF 1 model at the 1.0, 1.5, and 2.0 SD levels, and at the 1.5 and 2.0 SD levels, it was also a fair predictor (≥ 0.70) for the overall sample and CON group in the AF 2/4 model.
In the AF domain, AUC values ranged from fair to excellent (≥ 0.70 and > 0.90) in all three samples, all three SD levels, and both models for the following criteria: impairment in adaptive communication, impairment in adaptive social skills, and impairment in adaptive living skills.The last AF criterion, impairment in motor skills, only had fair to good (≥ 0.70 and < 0.90) in the CON group in the AF 2/4 model at the 1.5 and 2.0 SD models.
Our ndings replicated the study by Kable et al. (2022) in two important ways: (1) current criteria captured a high rate of those with PAE and (2) requiring fewer adaptive functioning criteria was more sensitive to PAE.This higher rate of endorsement was re ected in both the alcohol-exposed group and the control group, again showing that an increase in capturing those at risk for ND-PAE will lead to an increase in false categorization of those not at risk (i.e., a sensitivity-speci city trade-off).
criteria were modi ed (AF 1).Increasing sensitivity may be helpful in this population as it can increase earlier diagnosis and, in consequence, earlier treatment.Though the modi ed global intellectual functioning criterion tested (considering impairment as 1.0 or 1.5 SD levels below the normative average) did not exceed chance levels of prediction, it improved predictive validity, and therefore a lower impairment threshold for global intellectual functioning should be considered.Also, certain criteria, such as visual-spatial impairment and impairment in adaptive motor skills, did not exceed chance levels for prediction for ND-PAE risk, and therefore these criteria should be modi ed or excluded from the diagnosis.
By including this diagnosis in future versions of the DSM, children may receive earlier FASD diagnoses and earlier access to care.In addition, this diagnosis can expand the number of individuals with the expertise needed to diagnose and treat conditions associated with PAE.Therefore, adapting the criteria of this diagnosis to increase utility in diagnosing alcohol-exposed individuals may improve care in the FASD population.

Declarations Author Contribution
Data were collected under the supervision of J.K., J.W., C.C., P. M., W.O., E.S., K.J., E.R., and S.M. Data were managed and coordinated by M.T. Analyses were done by C.V., M.T., J.K., and S.M.The manuscript was written by C.V. and S.M.All authors reviewed the manuscript.

Table 1
Measures used for each criterion.
CBCL School Competence Scale Score; CVLT-C List A Total Trials 1-5 Total Correct Scaled Score, List A Trial 5 Free Recall Total Correct Z Score, Learning Slope Z Score, Percent Recall Consistency Z Score; TRF Academic Performance Scale Score and Learning Scale Score; WIAT-III Math Problem Solving and Word Reading Scores Memory Impairment CANTAB Working Memory Composite score, the Spatial Working Memory Score, and Delayed Matching Score; CVLT-C Short Delay and Long Delay Free Recall Scores; NEPSY-II Narrative Memory Free Recall Score, Memory for Designs Total Scaled Score, Memory for Names Total Scaled Score; NIH Toolbox Picture Sequence Memory Subtest and List Sorting Working Memory Subtest Scores; WISC-IV/V Working Memory Composite Score BASC-3 Externalizing Problems and Internalizing Problems Scores; BRIEF Emotional Control Score; CBCL Anxious/Depressed and Withdrawn/Depressed Scores; VABS-II/III Externalizing and Internalizing Scores Attention De cit BASC-3 Hyperactivity Score; CBCL Attention Problems Score; Conners 3 Inattention, Attention De cit/Hyperactivity Disorder Predominantly Inattentive Type and the Attention De cit/Hyperactivity Disorder Predominantly Hyperactive-Impulsive Type Scores; DBD ADHD Combined Diagnosis Score; NIH Toolbox Flanker Subtest Score Impairment in Impulse Control BASC-3 Aggression and Conduct Problems Scores; CBCL Rule Breaking Score; Conners 3 Hyperactivity/Impulsivity, Conduct Disorder, and Oppositional De ant Disorder Scores; DBD OD Diagnosis and CD Diagnosis Scores

Table 2 .
Participants ranged from ve to 17 years old.Participants were, on average, 10 years old with a slight majority of male participants (n = 486, mean age = 10.9y,female = 47%) and controls (n = 679, mean age = 10.9y,female = 43.6%).Average IQ was in the Low Average range for the AE group (mean IQ = 87.2,SD = 14.86) and in the Average range for the CON group (mean IQ = 94.4,SD = 16.98).Both groups included participants with ADHD (AE: 77.7%; CON: 24.5%).

Table 4
Sensitivity, speci city, and accuracy analyses.

Table 6 )
in three different samples (i.e., overall sample, AE group only, CON group only) and using the original AF criteria requirements (AF 2/4 model) compared to using the modi ed AF requirements (AF 1 model).AUC values range from 0 to 1.0 with a chance level of 0.5.Values less than 0.70 indicate poor prediction, values between 0.70 and .80indicate fair prediction, values between 0.80 and 0.90 indicate good prediction, and values higher than 0.90 indicate excellent prediction

Table 5
AUC values for each domain for each SD level.Signi cant values are indicated by bold font.

Table 6
Kable et al. (2022) criterion for each SD level.Signi cant values are indicated by bold font.This study aimed to validate the ND-PAE diagnosis currently located under the "Conditions for Further Study" section of the DSM-5.Using data from six CIFASD sites, individuals with moderate to heavy PAE were compared to individuals with minimal or no PAE and with and without other neurodevelopmental disorders on neuropsychological assessment measures that were tied to each ND-PAE criterion.Though the sample was smaller than the sample studied inKable et al. (2022), participants involved had experienced heavier exposure to alcohol prenatally and ranged in age and our sample was still large.Participants were of equal age and evenly divided in sex.Participants in the control group had higher IQ scores, and more participants in the AE group had IQ scores less than 85.ND-PAE endorsement was evaluated at 1.0, 1.5, and 2.0 SD levels, and a revision of the ND-PAE diagnosis (AF 1) was tested.