Accuracy of the Diagnosis of Physical Features of Fetal Alcohol Syndrome by Pediatricians After Specialized Training
OBJECTIVES. Accurate and early diagnosis of the fetal alcohol syndrome is important for secondary prevention, intervention, and treatment, yet many pediatricians lack expertise in recognition of the characteristic features of this disorder. After a structured training program for pediatricians, we examined the ability to accurately diagnose fetal alcohol syndrome.
METHODS. Two dysmorphologists conducted a 2-day training program in the diagnosis of the physical features of fetal alcohol syndrome for 4 pediatricians in Moscow. Dysmorphologists and pediatricians worked in teams to examine children, demonstrate techniques, and validate that pediatricians could identify physical features of this disorder under direct observation. Subsequently, pediatricians independently evaluated children in 41 boarding schools and orphanages. Those children diagnosed with fetal alcohol syndrome or deferred (possible fetal alcohol syndrome) by the pediatricians were then evaluated by the dysmorphologists. Accuracy of the diagnosis of fetal alcohol syndrome or deferred was assessed, as well as the interrater agreement for specific selected features of the disorder.
RESULTS. A total of 110 children were examined by both the pediatricians and the dysmorphologists. Of these, 79 were identified with fetal alcohol syndrome by the pediatricians; in 66 (83.5%) of these children, the diagnosis was confirmed by the dysmorphologists. Among 31 children who were classified as deferred by the pediatricians, 21 (67.7%) were confirmed with either fetal alcohol syndrome or deferred by the dysmorphologists. With respect to selected structural features characteristic of fetal alcohol syndrome, good interrater agreement was noted for height and head circumference ≤10th centile, whereas moderate-to-fair agreement was noted for smooth philtrum, long philtrum, presence of “hockey-stick” palmar crease, and palpebral fissure length ≤10th centile. Poor agreement was noted for thin upper lip.
CONCLUSIONS. After a relatively short training session, pediatricians were reasonably accurate in diagnosing fetal alcohol syndrome on the basis of physical features and in recognizing most of the selected specific features associated with the disorder.
Fetal alcohol syndrome (FAS) represents a specific pattern of malformations seen in the offspring of women who drink alcohol heavily during their pregnancies.1 The principal features of the disorder are alterations of growth and neurobehavioral development, as well as characteristic minor malformations, including short palpebral fissures, smooth philtrum, and thin vermilion border of the upper lip. These minor physical features are often difficult to recognize resulting in frequent underdiagnosis of the disorder. Accurate and early diagnosis of FAS is of vital importance in terms of secondary prevention, intervention, and treatment.2 However, pediatricians have been shown to lack expertise in recognition of the characteristic structural features of FAS and also to lack confidence in diagnosing it, particularly in the newborn period and in children whose features are at the more mild end of the spectrum of effects.3 For example, Stoler and Holmes4 compared the ability of pediatricians to recognize alcohol-related features in newborn infants with that of dysmorphologists who were blinded to the maternal alcohol-exposure status. Of 19 study infants who had documentation of maternal alcohol exposure, only 1 was noted to have any alcohol-related features by the pediatricians as opposed to 7 infants examined by the dysmorphologists. Two of the 19 infants were diagnosed with FAS by the dysmorphologists; neither of these children were diagnosed with a FAS by the pediatricians.4
In response to these difficulties, several guidelines for referral and diagnosis of FAS have been introduced,5–8 and a 4-digit code to facilitate classification of the full spectrum of fetal alcohol-exposed individuals has been developed.9 Beginning in 2003, an international Collaborative Initiative on the Fetal Alcohol Spectrum Disorders (CIFASD) has been funded in sites throughout the world by the National Institute of Alcohol Abuse and Alcoholism. This consortium was established to better understand the mechanisms, diagnostic approaches, prevention, treatment, and the full spectrum of effects in alcohol-exposed pregnancies. As part of that initiative, some sites have incorporated specialized training of pediatricians to increase the number of physicians who have expertise in the diagnosis of FAS. The purpose of this article is to evaluate the ability of general pediatricians with little or no experience in the diagnosis of FAS who have participated in one of these training sessions to correctly identify the characteristic structural features of FAS through a physical examination and to successfully diagnose or rule out this disorder on that basis.
Training of pediatricians was conducted as one component of a project in a CIFASD site located in Moscow among children in boarding schools and orphanages with special programs for the mentally handicapped. Trained pediatricians subsequently used active case ascertainment to identify affected children by evaluating all of the residents in the 41 participating institutions. Results of this cross-sectional study have been reported previously and indicated that the overall prevalence of FAS was 8.7% of children in these settings.10
The training protocol for pediatricians was based on the specific physical features described as sufficient for the diagnosis of FAS according to one schema, the revised Institute of Medicine criteria set forth by Hoyme et al8 Training of 4 local Russian pediatricians was conducted by 2 US dysmorphologists (Drs Jones and Robinson) over a 2-day period. As part of the training session, a 1-hour lecture describing the features of FAS was followed by a 1-hour session demonstrating the optimum techniques for measurement of the various craniofacial features characteristic of the disorder. In addition, the appropriate use of the Astley-Clarren “lip-philtrum guide,” a Likert scale photographic reference standard for classification of philtral smoothness and upper lip thinness, was demonstrated.9 Thereafter, 2 teams of 3 physicians (1 dysmorphologist and 2 Russian pediatricians training in the diagnosis of FAS) worked in parallel over the remainder of the 2 days. During that time, children diagnosed previously with FAS, as well as those who did not have that diagnosis, were examined simultaneously by the Russian pediatricians and the dysmorphologists in each of the 2 teams. For training purposes, results for individual children were compared after each examination.
A structured protocol and examination form were used to help pediatricians identify and systematically record dysmorphologic features that constitute FAS. Anthropometric measurements included weight, height, head circumference, palpebral fissure length, and philtral length. Palpebral fissure length and philtral length were measured with a rigid ruler marked in millimeters. The smoothness of the philtrum and the thinness of the upper lip were assessed separately using the Astley-Clarren lip-philtrum guide. Percentiles for weight, height, and head circumferences were determined on the basis of the US National Center for Health Statistics 2000 standard growth curves.11 Percentiles for palpebral fissure length12 and philtral length13 were determined using previously published charts. Additional features, such as a “hockey-stick” palmar crease, that are seen more frequently in children prenatally exposed to alcohol than in those who are unexposed, were also evaluated as part of the structured protocol.
As part of the larger CIFASD research protocol, a standard classification system was used for preliminary diagnosis of alcohol-affected children solely on the basis of structural features and growth deficiency. Under this scheme, children could receive 1 of 3 preliminary diagnoses: FAS, deferred, or no FAS. Consistent with the revised Institute of Medicine criteria, children who were growth deficient on weight and/or height (≤10th centile); had a head circumference ≤10th centile; and had ≥2 of the key characteristic facial features, including short palpebral fissures (≤10th centile), smooth philtrum (Astley-Clarren lip-philtrum guide score of 4 or 5), and thin vermilion border of the upper lip (Astley-Clarren lip-philtrum guide score of 4 or 5) were classified as FAS. Children who had evidence of growth deficiency but had an insufficient number of the characteristic facial features to qualify as FAS were classified as deferred. This category represented a temporary diagnosis pending more information on neurobehavioral performance, evidence of maternal alcohol exposure, and possible alternative diagnoses before a final diagnosis could be assigned. However, for purposes of this evaluation of the training protocol, preliminary diagnostic categories based strictly on the physical examination were used; thus, some children remained classified as deferred. Finally, children who had insufficient or no evidence of growth deficiency or any of the key facial features were classified as no FAS.
After completion of the training, local pediatricians independently examined children in a total of 41 institutions in Moscow over a several-month period of time. At the time of the examination, they were blinded to prenatal alcohol exposure histories. On the basis of their physical examinations, the pediatricians assigned each child 1 of the 3 preliminary diagnoses. At a later date, children initially diagnosed as FAS or deferred by the pediatricians received an identical standardized physical examination by 1 or both of the dysmorphologists who were blinded to the results of the pediatrician's evaluation, as well as prenatal alcohol exposure histories. Children were independently categorized by the dysmorphologists into 1 of the 3 diagnostic groups.
To evaluate the ability of these specially trained Russian pediatricians to correctly classify children as FAS or deferred on the basis of the physical examination alone, we compared the frequency distribution of these diagnostic categories of FAS between the 2 groups of examiners. To evaluate the ability of pediatricians to recognize specific alcohol-related features, we compared the interrater agreement between the pediatricians and the dysmorphologists on the following dichotomous variables: height ≤10th centile, head circumference ≤10th centile, palpebral fissure length ≤10th centile, philtral length ≥90th centile, smooth philtrum, thin upper lip, and presence or absence of a hockey-stick palmar crease.
Statistical analysis was performed in SAS 9.0 (SAS Institute, Inc, Cary, NC). Agreement or interrater reliability between pediatricians and dysmorphologists was estimated using Cohen's κ statistic.14 The κ coefficient is equal to 1 when the agreement is perfect, equal to 0 if the agreement is entirely attributable to chance, and <0 when the agreement is less than would be expected by chance alone. On the basis of Landis' and Koch's criteria describing the relative strength of agreement, κ values >0.80 should be interpreted as almost perfect, 0.61 to 0.80 as substantial, 0.41 to 0.60 as moderate, 0.21 to 0.40 as fair, 0.00 to 0.21 as slight, and <0 as poor.15 The direction of disagreement between the pediatricians and dysmorphologists was described using the dysmorphologist as the “gold standard.” For example, if a pediatrician indicated the presence of a certain feature that was not found by the dysmorphologist (false-positive), then identification of this feature by the pediatrician was considered an overestimate. Conversely, if a pediatrician missed a feature that was identified by the dysmorphologist (false-negative), identification of this feature was considered an underestimate.
For purposes of this study, the sample consisted of 110 children who were either diagnosed with FAS (n = 79) or deferred (n = 31) by the local pediatricians and who were later examined by the dysmorphologists. Because the dysmorphologists did not examine children who were classified as no FAS by the pediatricians, the sensitivity and specificity of the FAS diagnosis could not be estimated.
The age distribution of examined children ranged from 3 to 18 years of age. Among 79 children diagnosed with FAS by the pediatricians, 66 (83.5%) were confirmed to have FAS by the dysmorphologists, 10 (12.7%) were categorized as deferred, and 3 (3.8%) were determined to be false-positives, because they did not have FAS (Table 1). Among 31 children who did not have a definitive diagnosis of FAS but had some features of FAS and were classified as deferred by pediatricians, 13 (41.9%) were diagnosed with FAS by the dysmorphologists, 8 (25.8%) were classified as deferred, and 10 (32.3%) did not have FAS (Table 1).
Agreement between pediatricians and dysmorphologists in recognizing selected individual features associated with FAS is presented in Table 2. For height and head circumference ≤10th centile, interrater agreement was high (κ = 0.80 and 0.68, respectively). Moderate-to-fair agreement was observed for palpebral fissure length ≤10th centile (κ= 0.35), smooth philtrum (κ= 0.42), philtrum ≥90th centile (κ= 0.36), and hockey-stick palmar crease (κ= 0.41). Agreement in identification of thin upper lip was the poorest (κ = 0.14) despite the use of the Astley-Clarren lip-philtrum guide as a visual aid.
This study demonstrates that pediatricians with little or no experience in diagnosing FAS can be trained to accurately diagnose the extreme end of the spectrum of physical features of this disorder through intensive training over a 2-day period. In this study, in those children in which the pediatricians diagnosed FAS, the dysmorphologists confirmed that diagnosis in 83.5% of cases and deferred an additional 12.7% for additional information when the diagnosis could not be based solely on physical features. In those children for whom the pediatricians had concerns but did not feel confident in committing to a diagnosis of FAS, that is, assigned the deferred category, the dysmorphologists were in disagreement in ∼75% of cases. However, the majority (13 of 23 [56.5%]) of those children who were misclassified by the pediatricians as deferred were reclassified by the dysmorphologists as FAS.
After completion of training, the pediatricians were able to identify most individual features of FAS reasonably accurately. Agreement was particularly high for more objective features, such as measurements for height and head circumference, for which landmarks are easily identified. Agreement was less robust for features that required more subjectivity, such as palpebral fissure length, for which identification of the lateral canthus is often difficult. Similarly, correct identification of smooth philtrum and thin upper lip was inconsistent, because both required subjective judgment on the part of the examiner even with the assistance of the Astley-Clarren visual reference guide; this is especially surprising given that we used a categorical variable for these measures.
Agreement regarding the presence of another subjective feature, the hockey-stick crease, was modest. Furthermore, this was the one feature that pediatricians consistently overestimated compared with the dysmorphologists. It is possible that this was because of the pediatricians mistakenly classifying other abnormal palmar creases as a hockey-stick crease. This finding suggests that training of pediatricians to better differentiate between hockey-stick crease and other aberrant palmar creases might be improved by the use a photographic aid similar to the Astley-Clarren lip-philtrum guide.
Efforts have been made in the past to train nurses and school personnel in the identification of features characteristic of FAS. Although such screening has revealed some previously undiagnosed children, the overall efficiency of these programs has been low. For example, in a study conducted among kindergarten students in North Dakota, South Dakota, and Minnesota, a 32-item FAS screen was administered by school staff, teachers, social workers, and psychologists. Only 6 children, or 8.7% of the 69 who screened positively, were confirmed to have FAS by a dysmorphologist.16 In another study conducted among elementary school students in the state of Washington, school and public health nurses were trained in one 4-hour session to recognize the features of FAS and to refer children who screened positive to a special diagnostic clinic. Among those screened, 64 children were suspected of having FAS; however, only 7, or 10.9%, were confirmed to have the diagnosis by a dysmorphologist.17
There are several limitations of this study. First, only those children who the pediatrician diagnosed as FAS or deferred on the basis of physical features were evaluated by the dysmorphologists. Thus, the accuracy of a negative diagnosis made by the pediatricians could not be evaluated. Second, because examiners were looking specifically and only for FAS-related features in institutional settings where it was known that the prevalence of FAS was high, the potential exists for observer, as well as context, bias.18 With respect to observer bias, it is possible that examiners were more likely to identify alcohol-related features because they were carefully looking for them; however, this type of bias might be expected to occur equally in both pediatricians and dysmorphologists, thus negating the effect on interrater agreement. With respect to context bias, the pediatricians conducted examinations of children with 3 options for preliminary diagnosis; however, the dysmorphologists only examined children who they knew had been classified previously as either FAS or deferred by the pediatricians. On the basis of this narrowed context of possibilities, the dysmorphologists could have been more likely to identify features of FAS, which could have contributed to a higher degree of agreement between the 2 examiners. Third, this program did not involve training of pediatricians in the examination of newborn infants for FAS. Finally, it was not possible to measure the contribution of the training session itself in improving the pediatricians' diagnostic abilities, because no pretraining evaluation of their capabilities was performed. However, the pediatricians who participated in this study before initiating training described themselves as lacking any experience or expertise in recognizing the physical features of FAS.
Despite these limitations, the program herein described is the first published report of which we are aware in which pediatricians have been trained in a clinical setting to recognize the characteristic physical features of FAS. Analysis of the results in terms of accuracy of the pediatricians' diagnoses documents the effectiveness of such a program. Furthermore, it highlights the fact that often subtle differences in physical features, such as those characteristic of FAS, can best be appreciated by pediatricians through a careful demonstration of those features followed by observation and correction of technique.
Up to this time, pediatricians have functioned primarily as intermediaries in evaluating children prenatally exposed to alcohol who must then be definitively diagnosed by a panel of experts. Given the limited numbers of these experts, the situation is untenable for the number of children who should be evaluated for prenatal alcohol effects. This study suggests that after a short, interactive training program, pediatricians can contribute in a more definitive way to the accurate diagnosis of this disorder, thus enhancing potential access to early intervention and treatment for affected children.
This study was supported by the National Institutes of Health, National Institute of Alcohol Abuse and Alcoholism grants AA10437, AA14834, AA14815, and U24 AA14815 as part of the International Collaborative Initiative on Fetal Alcohol Spectrum Disorders.
- Accepted July 23, 2006.
- Address correspondence to Kenneth Lyons Jones, MD, Department of Pediatrics, University of California San Diego Medical Center, 200 West Arbor Dr, Mail Code 8446, San Diego, CA 92103. E-mail:
The authors have indicated they have no financial relationships relevant to this article to disclose.
The data in this article were presented at the Research Society on Alcohol meeting; June 2004; Vancouver, British Columbia, Canada.
- ↵Jones KL, Smith DW, Ulleland CN, Streissguth AP. Pattern of malformation in offspring of chronic alcoholic mothers. Lancet.1973;1(7815) :1267– 1271
- ↵Streissguth AP, Barr HM, Kogan J, Bookstein FL. Understanding the Occurrence of Secondary Disabilities in Clients With Fetal Alcohol Syndrome (FAS) and Fetal Alcohol Effects (FAE): Final Report of the Fetal Alcohol Syndrome Conference. Seattle, WA: Fetal Alcohol Drug Unit; 1996:4– 8
- ↵Stratton K, Howe C, Battaglia F. Fetal Alcohol Syndrome: Diagnosis, Epidemiology, Prevention and Treatment. Washington, DC: National Academy Press; 1996
- Centers for Disease Control and Prevention. Fetal Alcohol Syndrome: Guidelines for Referral and Diagnosis. Atlanta, GA: Centers for Disease Control and Prevention; 2004
- Chudley AE, Conry J, Cook JL, Loock C, Rosales T, LeBlanc N. Fetal alcohol spectrum disorder: Canadian guidelines for diagnosis. CMAJ.2005;172 (5 suppl) :S1– S21
- ↵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 :39– 47
- ↵Astley SJ, Clarren SK. Diagnosing the full spectrum of fetal alcohol-exposed individuals: introducing the 4-digit diagnostic code. Alcohol Alcohol.2000;35 :400– 410
- ↵Marintcheva G, Riley E, Konovalova V, et al. Fetal alcohol syndrome in different contingents of children and adolescents [in Russian]. Soc Clin Psychiatry.2003;3 :17– 22
- Copyright © 2006 by the American Academy of Pediatrics