To determine the prevalence and characteristics of fetal alcohol spectrum disorders (FASD) among first grade students (6- to 7-year-olds) in a representative Midwestern US community.
From a consented sample of 70.5% of all first graders enrolled in public and private schools, an oversample of small children (≤25th percentile on height, weight, and head circumference) and randomly selected control candidates were examined for physical growth, development, dysmorphology, cognition, and behavior. The children’s mothers were interviewed for maternal risk.
Total dysmorphology scores differentiate significantly fetal alcohol syndrome (FAS) and partial FAS (PFAS) from one another and from unexposed controls. Alcohol-related neurodevelopmental disorder (ARND) is not as clearly differentiated from controls. Children who had FASD performed, on average, significantly worse on 7 cognitive and behavioral tests and measures. The most predictive maternal risk variables in this community are late recognition of pregnancy, quantity of alcoholic drinks consumed 3 months before pregnancy, and quantity of drinking reported for the index child’s father. From the final multidisciplinary case findings, 3 techniques were used to estimate prevalence. FAS in this community likely ranges from 6 to 9 per 1000 children (midpoint, 7.5), PFAS from 11 to 17 per 1000 children (midpoint, 14), and the total rate of FASD is estimated at 24 to 48 per 1000 children, or 2.4% to 4.8% (midpoint, 3.6%).
Children who have FASD are more prevalent among first graders in this Midwestern city than predicted by previous, popular estimates.
Comments
Re:Non-specific FASD diagnostic criteria lead to over-diagnosis
Fetal alcohol spectrum disorders (FASD) are best diagnosed by a medical team headed by a geneticist/dysmorphologist with multidisciplinary input. Malformation syndromes with features similar to FASD must first be ruled out before a FASD diagnosis is assigned. As with any medical condition, no alcohol-related diagnosis is made solely on explicit criteria without the sound clinical judgment of an experienced clinician.
Astley has stated that the presence of all three "sentinel" facial features of FAS is 99.8% specific for FASD. Further, she has stated that "without a specific facial phenotype, a valid diagnosis of fetal alcohol syndrome cannot be rendered."(1) Instead of debating the number of requisite facial features needed for a FASD diagnosis, these findings should be considered in terms of their embryopathogenesis, that is as forme fruste signs of holoprosencephaly. Animal and human models confirm that these features can be genetically determined in addition to being the result of teratogenic exposures.(2,3,4,5) Even with documented teratogenic exposure, the facial phenotype may vary depending on timing.(4) This has been documented in at least two previous human studies,(2,3) and will be borne out in unpublished genetic testing data from our recent study. Thus, FASD cannot be diagnosed on the facial dysmorphology alone.
The case of "Johnny" in Dr. Davies' letter over-simplifies our diagnostic process. We gather data and apply diagnostic criteria only after other conditions have been ruled out. Not only are dysmorphic features considered, but also testing results, functional abilities of the child, evidence of prenatal alcohol exposure and other maternal risk factors.
All families receive useful feedback. Families of children for whom there is not an alcohol-related concern receive a report of the testing results detailing their child's performance on an extensive battery of tests. This provides useful information about strengths/deficits and is often, at the request of the families, included in the child's school records.
For children who have a FASD or another condition needing further attention, meetings conducted by a dysmorphologist and psychologist/diagnostician are held with each family. Feedback is provided about the preliminary diagnostic decision of the physician and the child's testing results. None of the results are shared until the families' concerns are discussed openly and a final diagnosis assigned. Therefore, the hypothetical "Johnny" outlined in the letter does not exist.
References:
1. Astley SJ. Comparison of the 4-digit diagnostic code and the Hoyme diagnostic guidelines for fetal alcohol spectrum disorders. Pediatrics 2006;118;1532.
2. Douzgou S, Breen C, Crow YJ, et al. Diagnosing fetal alcohol syndrome: new insights from newer genetic technologies. Arch Dis Child 2012 Sep;97(9):812-7.
3. Abdelmalik N, van Haelst M, Mancini G, et al. Diagnostic outcomes of 27 children referred by pediatricians to a genetics clinic in the Netherlands with suspicion of fetal alcohol spectrum disorders. Am J Med Genet A 2013 Feb;161A(2):254-60.
4. Lipinski RJ, Hammond P, O'Leary-Moore SK, et al. Ethanol-induced face-brain dysmorphology patterns are correlative and exposure-stage dependent. PLoS One. 2012;7(8):e43067.
5. Leibson T, Neuman G, Chudley AE, Koren G.J. The differential diagnosis of fetal alcohol spectrum disorder. Popul Ther Clin Pharmacol. 2014;21(1):e1-e30.
Conflict of Interest:
None declared
Re:FAS/PFAS Prevalence Over Estimated
Using active case ascertainment methods to provide medical geneticists/dysmorphologists the opportunity to perform in-person examinations of children in representative schools provides access to more cases, and therefore uncovers high rates of fetal alcohol syndrome (FAS) and also other fetal alcohol spectrum disorders (FASD). About the same time that we reported such findings in South African studies, Dr. Astley's own colleagues from the University of Washington reported that only one of six of the FAS cases (17%) they found in elementary schools had been previously diagnosed.(1) Conversely, prevalence rates generated in clinic- based, surveillance, or referral studies will always be under estimates.
There are several facial phenotypes of FAS, let alone other FASD. Sensitivity and specificity of a diagnosis of complex syndromes like FAS and partial fetal alcohol syndrome (PFAS) must be carefully balanced in any diagnosis and diagnostic system. The final diagnosis must be made by expert dysmorphologists using sound empirical evidence and medical judgment on a multitude of alcohol-linked physical traits and from the best evidence possible from: a targeted battery of cognitive and behavioral tests and sophisticated, multivariate maternal risk factor interviews. Using only three specific facial features to rule in or rule out FAS is neither as sensitive nor as predictive as it could be. The three cardinal facial features of FAS may represent what we now know is only one of the facial phenotypes of FAS.(2) As presented previously, our research in multiple populations indicates that using only the three cardinal features to rule in or rule out a case of FAS is insufficient, for these three features alone are 72% sensitive, 92% specific, have a positive predictive value for only 53% of the cases, and accuracy of 90%. Ruling out a diagnosis of FAS if all three features are not present results in many false negatives. As with other anomalies, individual phenotypic variation dictates that critical evidence must include multiple additional traits that may indicate other influences on child outcomes: specific patterns of alcohol use; maternal health variables and childbearing history; postnatal environment; and genetic and epigenetic factors.(3)
Finally, alcohol has been meticulously linked to all child traits utilized in our studies of FASD in several populations. Studies in South African populations, where mothers have freely provided detailed information on specific prenatal drinking practices, have linked each of the dysmorphic features we employ to alcohol by: quantity, frequency, duration, and timing during gestation.(4) Sophisticated statistical models that control for many other prenatal influences confirm this strong link to the diagnostic variables employed.(5) Furthermore recent publications of imaging and 3-D photographic studies that were carried out by other scholars using children diagnosed by members of our team have confirmed the link to prenatal alcohol exposure.
The range of estimates for FASD published in our paper from the Midwestern City are accurate. The prevalence lies somewhere therein: between 6 and 9 per 1,000 children for FAS and 2.4 to 4.8% for FASD.
References:
1. Clarren SK, Randels SP, Sanderson M, et al. Screening for fetal alcohol syndrome in primary schools: a feasibility study. Teratology. 2001;63:3-10.
2. Lipinski RJ, Hammond P, O'Leary-Moore SK, et al. Ethanol-induced face-brain dysmorphology patterns are correlative and exposure-stage dependent. PLoS One. 2012;7(8):43067.
3. Sulik KK. Fetal alcohol spectrum disorders: pathogenesis and mechanisms. Handbook of Clinical Neurology. 2014;125:463-475.
4. May PA, Blankenship J, Marais AS, et al. Maternal alcohol consumption producing fetal alcohol spectrum disorders (FASD): quantity, frequency, and timing of drinking. Drug Alcohol Depend. 2013;133(2):502- 5012.
5. May PA, Tabachnick BG, Gossage JP et al. Maternal risk factors predicting child physical characteristics dysmorphology in fetal alcohol syndrome and partial fetal alcohol syndrome. Drug Alcohol Depeden. 2011;119(1-2):18-27.
Conflict of Interest:
None declared
Non-specific FASD diagnostic criteria lead to over-diagnosis
We need to talk about Johnny.
He's a representative first-grader in a representative Midwestern city. His mother did not drink alcohol during pregnancy. Johnny's doing great in school and has an IQ of 120, with no evidence of developmental or behavioral problems.
Johnny's palpebral fissure lengths are at the 10th percentile, and he has a somewhat thinner upper lip (rank 4), but a deep philtrum (rank 1). This is not the face of FAS, by any reasonable definition.
In fact, my colleague Dr. Astley empirically confirmed that the FAS face as defined by Hoyme et al. is NOT specific to fetal alcohol exposure. In fact, 25% percent of a control group with above-average intellectual functioning and no prenatal alcohol exposure met the overly relaxed Hoyme FAS facial criteria. [Astley SJ. Comparison of the 4- Digit Diagnostic Code and the Hoyme diagnostic guidelines for fetal alcohol spectrum disorders. Pediatrics. 2006;118(4):1532-1545 doi:10.1542/peds.2006-0577] Just like Johnny.
Johnny's weight (or height) is at the 10th percentile. His pediatrician has reassured his family that his growth is normal, since he's growing steadily and comes from smaller parents.
According to these authors, Johnny has Partial FAS Without Confirmed Maternal Alcohol Exposure.
What if Johnny's head circumference is also tracking at the 10th percentile? This is also in the normal range, according to his pediatrician. After all, Johnny has no evidence of neurobehavioral impairments.
Now Johnny has FAS Without Confirmed Maternal Alcohol Exposure.
Johnny is a hypothetical, but hardly an edge case. Subjects like him are likely to have contributed to the high estimated prevalence of PFAS in this study. Note that 39% of the PFAS diagnoses in this study were "without confirmed exposure." The estimates of FAS are similarly suspect (2/3 of the FAS diagnoses in this study lacked confirmed exposure).
Overestimating FASD prevalence rates is one problem, that requires more critical peer review. Using non-specific diagnostic criteria in clinical practice is an even greater concern.
Would you feel comfortable sharing Johnny's "diagnosis" with his mother? If not, why have diagnostic criteria that require clinical judgment to prevent misdiagnosis in a child without medical or developmental issues, and no history of prenatal alcohol exposure?
It would be arguably worse if Johnny did have developmental problems of some sort.
When the facial criteria are not specific to FAS, and the growth and neurobehavioral features are not uniquely caused by prenatal alcohol exposure, is it valid to effectively blame the mother for Johnny's impairments with a diagnosis of Fetal Alcohol Syndrome Without Confirmed Maternal Alcohol Exposure?
The authors have stated that they "validated" these proposed revisions to the IOM criteria. It is impossible to know, as they have not published sufficient measures of reliability, validity, or accuracy. Both common pediatric sense and the empiric evaluation in this journal [Astley 2006] show that these criteria are far too relaxed to be used in FASD diagnosis and research.
Conflict of Interest:
None declared
FAS/PFAS Prevalence Over Estimated
The authors report FAS prevalence in this population is 6-9/1,000 children (3-fold higher than the FAS prevalence estimated by CDC/IOM (0.2- 3.0/1,000)). The authors also report both dysmorphology and maternal data link the teratogenic agent, alcohol, to the cases. The study methodology does not support these conclusions.
The Hoyme(1) diagnostic guidelines used in this study, unlike all other current FASD diagnostic guidelines (CDC, 4-Digit, Canadian, and Australian), relax the diagnostic criteria for the FAS facial phenotype from 3 features (PFL <=2%tile, smooth philtrum and thin upper lip (Rank 4 or 5 on UW Lip-Philtrum Guide) to any two of these three, with the PFL relaxed to <=10%tile.
The relaxation of the FAS facial criteria results in a facial phenotype that is no longer specific to (caused only by) prenatal alcohol exposure. The 3 facial features defined by the 4-Digit Code have a specificity of > 95%(2). The 2 facial features defined by Hoyme(1) have a specificity of only 71.5% (reported by Hoyme at the 5th International FASD Conference,2013). What happens when the specificity is this low? In a 2006 study(3), 25% of a group of high-functioning children (mean IQ 120) with confirmed absence of prenatal alcohol exposure met the Hoyme(1) criteria for the FAS facial phenotype.
When the specificity of the FAS facial phenotype falls below 95%, two problems arise. First, the diagnostic label FAS is rendered medically invalid. If one labels the patient's outcome FAS, one is declaring the patient has a syndrome caused by their mother's consumption of alcohol during pregnancy. But if the face, growth, and CNS abnormalities are not specific to (caused only by) prenatal alcohol exposure, one has no medical or scientific evidence to support this declaration of causation in an individual patient. Second, a diagnosis of FAS can no longer be made in the absence of a confirmed prenatal alcohol exposure. Note, this is why the Hoyme(1) ARND diagnosis cannot be made when alcohol exposure is unknown.
Since the FAS (and PFAS) facial phenotype in the Hoyme(1) guidelines is not specific to prenatal alcohol exposure, a diagnosis of FAS (or PFAS) cannot be rendered when prenatal alcohol exposure is unknown. Only 33% (n=4) of the 12 FAS diagnoses in this study had confirmed alcohol exposure. Thus the prevalence of FAS in this study is at most 1.9- 2.7/1,000 (4/2,033 to 4/1,433); comparable to the CDC/IOM FAS estimates (0.2-3.0/1,000).
The dysmorphology and maternal data do not link the teratogenic agent, alcohol, to the cases in this study. The facial dysmorphology are not specific to prenatal alcohol exposure. Half the FAS/PFAS cases had no confirmed alcohol exposure. No significant link between the "dysmorphology -score" and any alcohol measure was reported. Only 1 of 7 maternal alcohol measures differentiated FASD from Controls in Table 4, but this too fails to establish a causal link between a case's alcohol exposure and their outcomes. None of the children's outcomes reported in this study are specific to (caused only by) prenatal alcohol exposure. The mothers may well have been forthcoming and truthful.
1. Hoyme HE, May PA, Kalberg WO, et al. A practical clinical approach to diagnosis of fetal alcohol spectrum disorders: clarification of the 1996 institute of medicine criteria. Pediatrics. 2005;115(1):39-47.
2. Astley SJ. Validation of the fetal alcohol spectrum disorder (FASD) 4- Digit Diagnostic Code. J Popul Ther Clin Pharmacol Vol 20(3):e416-467; November 15, 2013.
3. Astley SJ. Comparison of the 4-Digit Diagnostic Code and the Hoyme Diagnostic Guidelines for Fetal Alcohol Spectrum Disorders. Pediatrics 2006;118(4):1532-1545.
Conflict of Interest:
None declared