OUP user menu

DIAGNOSING THE FULL SPECTRUM OF FETAL ALCOHOL-EXPOSED INDIVIDUALS: INTRODUCING THE 4-DIGIT DIAGNOSTIC CODE

Susan J. Astley, Sterling K. Clarren
DOI: http://dx.doi.org/10.1093/alcalc/35.4.400 400-410 First published online: 1 July 2000

Abstract

The medical/research records of 1014 patients diagnosed at the Washington State Fetal Alcohol Syndrome (FAS) Diagnostic and Prevention Network (DPN) of clinics were used to develop a new, comprehensive, reproducible method for diagnosing the full spectrum of outcomes among patients with prenatal alcohol exposure. This new diagnostic method, called the 4-Digit Diagnostic Code, was compared to the standard gestalt method of diagnosis on the first 454 patients who had received a gestalt diagnosis of FAS, atypical FAS (AFAS) or possible fetal alcohol effect (PFAE) prior to the development of the 4-Digit Diagnostic Code. The outcomes of the patients were more accurately and comprehensively documented by the 4-Digit Diagnostic Code, because of its use of quantitative, objective measurement scales and specific case definitions. The four digits in the code reflect the magnitude of expression of the four key diagnostic features of FAS in the following order: (1) growth deficiency; (2) the FAS facial phenotype; (3) central nervous system damage/dysfunction; (4) gestational alcohol exposure. The magnitude of expression of each feature is ranked independently on a four-point Likert scale with 1 reflecting complete absence of the FAS feature and 4 reflecting a strong ‘classic’ presence of the FAS feature. The 4-Digit Diagnostic Code is being used effectively for diagnosis, screening, and surveillance efforts in all Washington State FAS DPN clinics.

INTRODUCTION

The fetal alcohol syndrome (FAS) is a permanent birth defect syndrome caused by maternal consumption of alcohol during pregnancy. The definition of the FAS has changed little since the 1970s when the condition was first described and refined (Jones and Smith, 1973; Clarren and Smith, 1978; Rosett, 1980; Sokol and Clarren, 1989; Stratton et al., 1996). The syndrome has been broadly characterized by pre- and/or postnatal growth deficiency, a characteristic set of minor facial anomalies, central nervous system (CNS) dysfunction and prenatal alcohol exposure. The presentation of each individual feature of the syndrome may be variably expressed with age.

For trained clinicians, dysmorphologists, or clinical geneticists, there is likely to be full agreement on a diagnosis of FAS only when the anomalies in growth, face, and brain are all very extreme and the alcohol exposure is conclusive and substantial. But the features are not dichotomous, that is either normal or clearly abnormal. Rather, the features, and indeed the history of alcohol exposure, all range along separate continua from normal to clearly abnormal and distinctive. In the absence of accurate, precise, and unbiased methods for measuring and recording the severity of exposure and outcome in individual patients, diagnoses will continue to vary widely from clinic to clinic (Chavez et al., 1988; Aase, 1994; Stratton et al., 1996). From a clinical perspective, diagnostic misclassification leads to inappropriate patient care, increased risk for secondary disabilities (Streissguth and Kanton, 1997) and missed opportunities for primary prevention. From a public health perspective, diagnostic misclassification leads to inaccurate estimates of incidence and prevalence (Stratton et al., 1996). Inaccurate estimates thwart efforts to allocate sufficient social, educational, and healthcare services to this high-risk population and preclude accurate assessment of primary prevention intervention efforts. From a clinical research perspective, diagnostic misclassification reduces the power to identify clinically meaningful contrasts between groups. Non-standardized diagnostic methods prevent valid comparisons between studies.

The primary limitations in the current practice of diagnosing individuals with prenatal alcohol exposure include the following.

  1. While there are diagnostic guidelines that physicians and medical researchers are encouraged to follow, the guidelines are not sufficiently specific to assure diagnostic accuracy or precision. While the diagnostic guidelines published by Sokol and Clarren (1989), which were a minor modification of the definition by the Fetal Alcohol Study Group of the Research Society on Alcoholism (Rosett, 1980), which, in turn, were derived from the work of Clarren and Smith (1978), do provide guidance, they are not sufficiently specific to assure diagnostic accuracy and precision. They reflect a more gestalt approach to diagnosis. The guidelines for CNS dysfunction do not address how many areas of deficit must be present, how severe the deficits must be or what level of documentation must exist to substantiate the presence of the deficit (i.e. parental history, psychometric testing or structural imaging). The guidelines for the facial phenotype are equally non-specific. Thus, how many facial features must be present, how severe must the features be and what scale of measurement should be used to judge their severity? One needs only to read the clinical literature or review medical records, birth certificates, birth defect registries or ICD-9 codes to see how variably these criteria are interpreted, applied and reported (Cordero et al., 1994; CDC, 1995a, CDC, 1995b; Ernhart et al., 1995; Stratton et al., 1996). Although the most recent guidelines published by the Institute of Medicine (Stratton et al., 1996) have not been out long enough to judge their impact on diagnostic accuracy and precision, the Institute guidelines present with the same limitations as previous guidelines.

  2. There is a lack of objective, quantitative scales to measure and report the magnitude of expression of key diagnostic features. For example, although a thin upper lip and smooth philtrum are key diagnostic features (Jones and Smith, 1973; Clarren and Smith, 1978; Astley and Clarren, 1996; Stratton et al., 1996), quantitative measurement scales have never been used to measure thinness or smoothness and guidelines have never been established for how thin or smooth the features must be. Objective quantitative scales would not only improve accuracy and precision, they would also greatly increase the statistical power to detect clinically important exposure– outcome patterns (Polit and Hungler, 1995) by increasing the level of measurement from the current nominal scales (e.g. upper lip thin/not thin) to ordinal scales (e.g. five-point Likert pictorial scale for lip thinness) or continuous scales (e.g. upper lip circularity: perimeter2/area). Ordinal and continuous scales reflect better the true continuum of outcome and exposure in FAS. Objective, quantitative scales also establish a common descriptive language for communicating outcomes in medical records and in the medical literature (Polit and Hungler, 1995).

  3. The term fetal alcohol effects (FAE) is broadly used and poorly defined. The term ‘suspected fetal alcohol effects’ was first introduced into the medical literature in 1978 and was defined as ‘less complete partial expressions’ of FAS in individuals with prenatal alcohol exposure (Clarren and Smith, 1978). Based on this definition, an individual whose mother drank a few glasses of wine intermittently throughout pregnancy and presented with attention deficit hyperactivity disorder would meet the criteria for FAE. So would an individual whose mother drank a fifth of vodka (757 ml) daily throughout pregnancy and presented with microcephaly, severe mental retardation, growth deficiency, and no facial anomalies. The broad use of this term and the reluctance to abandon it point to the clear need to develop diagnostic terms for individuals with prenatal alcohol exposure who present with physical anomalies and/or cognitive/behavioural disabilities, but do not meet the criteria for FAS.

  4. Clinical terms like FAE, alcohol-related birth defects (ARBD) and alcohol-related neurodevelopmental disorder (ARND) (Stratton et al., 1996) inappropriately imply a causal link between exposure and outcome in a given individual. With the likely exception of the full facial phenotype, no other physical anomalies or cognitive/behavioural disabilities observed in an individual with prenatal alcohol exposure are necessarily specific to (caused only by) their prenatal alcohol exposure (Stratton et al., 1996). There have already been formal appeals by noted dysmorphologists in the field to discontinue the use of the term FAE (Sokol and Clarren, 1989; Aase et al., 1995). The diagnostic terms ARBD and ARND introduce the same limitations as FAE, namely, implying that alcohol exposure caused the birth defect or neurodevelopmental disorder in an individual patient.

  5. The terms FAS and FAE fail to convey the diversity of disability present in these individuals. No two individuals with FAS present with precisely the same constellation of anomalies and disabilities. Growth, facial phenotype, CNS dysfunction, and alcohol exposure all vary along separate continua. The term FAS only conveys that the condition is permanent and was caused by prenatal alcohol exposure. The term does not convey what the individual's disabilities are. A nomenclature that not only conveys the diversity of outcomes among individuals with prenatal exposure, but also separates outcome from exposure would benefit both the patient and their medical/social/educational care network.

Each of these limitations has been largely overcome with the development of a comprehensive manual for the diagnosis of FAS, Diagnostic Guide for Fetal Alcohol Syndrome and Related Conditions (Astley and Clarren, 1999), introducing a new quantitative approach to diagnosis, the 4-Digit Diagnostic Code. The diagnostic method was developed through the combined expertise of the University of Washington FAS Diagnostic and Prevention Network (FAS DPN) multidisciplinary, clinical team and the comprehensive records of 1014 FAS DPN patients (birth to 51 years of age) with reported prenatal alcohol exposure.

The creation of the 4-Digit Diagnostic Code was developed to assure accurate and precise diagnosis of individuals with prenatal alcohol exposure across all seven FAS Clinics in the Washington State FAS DPN (Clarren and Astley, 1997). The FAS DPN (expanded from the CDC-sponsored FAS Clinic at the University of Washington) was mandated by the 1995 Washington State Legislature in response to the high, statewide demand for diagnostic services. The guide was developed to meet the needs of a broad range of professionals in an equally broad range of settings. A core team that includes a physician, a psychologist, a language pathologist and an occupational therapist staffs each clinic, and also a family advocate. Each core team has community based links to alcohol treatment centres, genetics clinics, schools, and social and legal service agencies. The six clinics have been established in the following settings: a children's hospital neurodevelopmental clinic, two public health clinics, an alcohol treatment clinic, a private psychological services clinic paired with an academic institution, and a comprehensive children's medical/ social services centre. These six clinics are led by the FAS DPN core centre at the Center for Human Development and Disability at the University of Washington.

The need to standardize the criteria for FAS was recognized early on by the Fetal Alcohol Study Group of the Research Society on Alcoholism, resulting in published guidelines by Rosett (1980) followed by several efforts to hone, clarify and express concern about the guidelines (Sokol and Clarren, 1989; Aase et al., 1995; Stratton et al., 1996). In the absence of specific case definitions, the FAS DPN has responded to both a mandate by its State legislature and recommendations by the Institute of Medicine (Stratton et al., 1996) to establish a diagnostic method which could be administered accurately and reproducibly.

Below are a brief description of the 4-Digit Diagnostic Code and a comparison of the gestalt (Sokol and Clarren, 1989) and 4-Digit Diagnostic Code outcomes for 454 patients seen in the FAS DPN who received both diagnostic approaches. A more detailed description of the 4-Digit Diagnostic Code can be found in a 111-page manual (Astley and Clarren, 1999), distributed by the University of Washington in Seattle. The guide includes a standardized FAS Diagnostic Evaluation Form accompanied by instructions, case definitions, normal anthropometric charts, pictorial Likert scales for ranking lip thinness and philtrum smoothness, and a New Patient Information Form which is completed by the patient's family to document the patient's exposure and developmental history. The guide is accompanied by an instructional CD-ROM ‘Fetal Alcohol Syndrome Tutor Medical Training Software’ (Astley et al., 1999). We have used the 4-Digit Diagnostic Code to diagnose over 1000 patients and have found the system to be very helpful in clinical and research areas. We describe it here and present preliminary assessments of its accuracy, precision and power, so that others can consider and evaluate its use.

METHODS

The 4-Digit Diagnostic Code was developed through the expertise of the multidisciplinary FAS DPN clinical staff and use of the medical research records of 1014 patients diagnosed in the FAS DPN. The purpose was not to redefine, but rather, more specifically case define the key diagnostic components of FAS as presented across several published FAS diagnostic guidelines (Clarren and Smith, 1978; Rosett, 1980; Sokol and Clarren, 1989; Stratton et al., 1996). The first working draft of the method completed in 1997 (Astley and Clarren, 1997), was pilot tested on all patients seen in the FAS DPN from 1997–1999 (n ~ 400) and was refined to its current form. Prior to the development of the 4-Digit Diagnostic Code, all 598 patients seen in the University of Washington FAS DPN Clinic (1993–1997) were diagnosed using the gestalt (Sokol and Clarren, 1989) method. In 1997, the FAS DPN clinics stopped using the gestalt method and started using the 4-Digit Diagnostic Code method. The charts of all patients who had previously been diagnosed with the gestalt method were retrofitted to the 4-Digit Diagnostic Code system for research purposes. The gestalt and 4-Digit Diagnostic Code outcomes are compared among the 454 patients who had received gestalt diagnoses of FAS, atypical FAS (AFAS), or possible fetal alcohol effect (PFAE). The University of Washington Human Subjects Review Board approved the use of these data.

Preliminary assessments of precision, accuracy, and power are presented to assist the reader in their evaluation of this new diagnostic method. The diagnostic evaluation forms of 20 patients were randomly selected from the 736 patients who received a 4-Digit Diagnostic Code 1 to 4 years ago at the University of Washington FAS DPN Clinic by S.K.C. and S.J.A. The standardized diagnostic forms document the clinical and psychometric data that were available for the patients at the time of their diagnoses. The 4-Digit Diagnostic Codes were deleted from the forms by the research assistant and rederived by S.K.C. and S.J.A. independently. Inter-rater reliability between S.K.C. and S.J.A. was assessed by comparing their rederived 4-Digit Diagnostic Codes. Intra-rater reliability was assessed by comparing the rederived codes to the original 4-Digit Diagnostic Codes. An additional assessment of inter-rater reliability was conducted on all 16 patients who had received 4-Digit Diagnostic Codes from one of the six FAS DPN clinics without consultation with the FAS DPN Core team at the University of Washington. The level of agreement between the 4-Digit Diagnostic Codes derived by the Network and University of Washington clinical teams was assessed. The κ-statistic was computed to test intra- and inter-rater agreement. Accuracy (the degree to which a measurement represents the true value of the parameter that is being measured) was assessed by comparing the 4-Digit Diagnostic Code outcomes to the gestalt diagnostic outcomes of the 454 patients who were diagnosed by both methods. Each of the diagnostic outcomes was also compared to the published diagnostic guidelines (Sokol and Clarren, 1989) available when the gestalt diagnoses were made. Power, the probability of detecting an effect in a study sample if an effect of a specified size or greater truly exists in the population, was computed using SamplePower (SPSS Inc., 1997).

RESULTS

The 4-digit diagnostic code

The four digits of the diagnostic code reflect the magnitude of expression of four key diagnostic features of FAS in the following order: (1) growth deficiency; (2) the FAS facial phenotype; (3) brain damage/dysfunction; (4) gestational alcohol exposure (Fig. 1). The 4-Digit Diagnostic Code is generated by first recording key clinical data on the standardized FAS Diagnostic Evaluation Form and following specific case definitions to generate each digit.

Fig. 1.

4-Digit Diagnostic Code grid and nomenclature.

This grid is used to record the 4-Digit Diagnostic Code following the guidelines presented in the text. A code of 3 or 4 in the Growth or Face column is referred to as a ‘sentinel physical finding’. A code of 2 in the Brain column is a ‘neurobehavioural disorder’; a code of 3 or 4 is ‘static encephalopathy’. A code of 3 or 4 in the alcohol column is ‘alcohol exposed’. The code 3244 would receive the name ‘sentinel physical findings, static encephalopathy, alcohol exposed’. A subset of 4-Digit Diagnostic Codes that fall within the category ‘sentinel physical findings, static encephalopathy,

alcohol exposed’ are referred to as FAS when the four digits are sufficiently high to meet the criteria for FAS (see Table 1).

The magnitude of expression of each feature is ranked independently on a four-point Likert scale with 1 reflecting complete absence of the FAS feature and 4 reflecting a strong ‘classic’ presence of the FAS feature. Each Likert rank is specifically case defined. The 4-Digit Diagnostic Code can be used to diagnose individuals of all ages.

Clinical nomenclature

There are 256 possible 4-Digit Diagnostic Codes ranging from 1111 to 4444. Each 4-Digit Diagnostic Code falls into one of 22 unique clinical diagnostic categories (labelled A to V) (Table 1). The 22 diagnostic categories are named to reflect the Likert ranking of each digit in the 4-Digit Diagnostic Code. The names are constructed sequentially from four terms: ‘sentinel physical findings’, ‘neurobehavioural disorder’, ‘static encephalopathy’, and ‘alcohol exposure status’ as presented in Fig. 1 and Table 1. Note that the names for diagnostic categories E–J, K–P, and Q–V only differ by alcohol exposure, thus there are essentially only nine unique diagnostic outcome categories ranging from ‘no cognitive/ behavioural or sentinel physical findings detected’ to ‘FAS’. This is in contrast to the five diagnostic outcome categories (FAS, partial FAS, ARBD, ARND, and FAE) currently in use across the various gestalt guidelines (Clarren and Smith, 1978; Sokol and Clarren, 1989; Stratton et al., 1996).

View this table:
Table 1.

Diagnostic categories

CategoryDiagnostic category name and 4-Digit Diagnostic Codes within each category
The 4-Digit Diagnostic Code reflects the magnitude of expression of four key diagnostic components of FAS in the order growth, facial phenotype, CNS damage/dysfunction and alcohol exposure. Each component is measured on a four-point Likert scale, thus producing 256 possible combinations of 4-Digit Diagnostic Codes. These codes are collapsed into 22 diagnostic categories as presented above.
AFetal alcohol syndrome (alcohol exposed)
34334433343444343443444334444444
BFetal alcohol syndrome (alcohol exposure unknown)
3432443234424442
CAtypical fetal alcohol syndrome (alcohol exposed)
14431434243433344334244314442444334443444343
DFetal alcohol syndrome phenocopy (no alcohol exposure)
34314341444134414431
ESentinel physical findings/static encephalopathy (alcohol exposed)
13331433234431433243413342334333133423332433
3144324441344234134323343133323333334143
42431344234331343234334341444244
FStatic encephalopathy (alcohol exposed)
11331144124321342233224411341233124421432234
11431234213321442243
GSentinel physical findings/neurobehavioural disorder (alcohol exposed)
13232323312333234123432313242324312433244124
43241423242332233423422344231424242432243424
42244424
HNeurobehavioural disorder (alcohol exposed)
11232123112421241223222312242224
ISentinel physical findings (alcohol exposed)
13132313311333134113431313142314311433144114
43141413241332133413421344131414241432143414
42144414
JNo cognitive/behavioural or sentinel physical findings detected (alcohol exposed)
11132113111421141213221312142214
KSentinel physical findings/static encephalopathy (alcohol exposure unknown)
13322332313233324232134223423142334242421432
243232324132433214422442324241424342
LStatic encephalopathy (alcohol exposure unknown)
11321232213222321142124221422242
MSentinel physical findings/neurobehavioural disorder (alcohol exposure unknown)
13222322312233224122432214222422322234224222
4422
NNeurobehavioural disorder (alcohol exposure unknown)
1122122221222222
OSentinel physical findings (alcohol exposure unknown)
13122312311233124112431214122412321234124212
4412
PNo cognitive/behavioural or sentinel physical findings detected (alcohol exposure unknown)
1112211212122212
QSentinel physical findings/static encephalopathy (no alcohol exposure)
13312341323141411341243132414231143124413331
42411441313133414331233131414131
RStatic encephalopathy (no alcohol exposure)
11312131114121411231223112412241
SSentinel physical findings/neurobehavioural disorder (no alcohol exposure)
13213121412114213221422123213321432124213421
4421
TNeurobehavioural disorder (no alcohol exposure)
1121212122211221
USentinel physical findings (no alcohol exposure)
13113111411114113211421123113311431124113411
4411
VNo cognitive/behavioural or sentinel physical findings detected (no alcohol exposure)
1111211112112211

The first two diagnostic categories (A and B) meet the criteria for a clinical diagnosis of FAS and are named as such (Table 1). The term Atypical FAS (category C) is introduced for use with a relatively small group of patients who present with static encephalopathy, most, but not all of the sentinel physical findings of FAS, and who were alcohol exposed. The term FAS Phenocopy (category D) applies to patients who present with all of the features of FAS, but have a confirmed absence of gestational ethanol exposure. We have not yet observed such a patient. The remaining 19 categories (E–V) do not meet the minimum criteria for FAS and are subsequently named to reflect the Likert ranking of each digit in the 4-Digit Diagnostic Code. For example, a code of 4342 is the diagnostic category sentinel physical findings/static encephalopathy (alcohol exposure unknown). Many of these patients might have previously been referred to variably as having PFAE, ARBD, or ARND (Stratton et al., 1996). This new nomenclature replaces all of these terms. These terms are not used here, because the inclusion of ‘alcohol-related’ or ‘alcohol effect’ in the diagnostic name inappropriately implies a confirmed causal link between exposure and outcome in an individual. Diagnostic categories E–I would have previously been referred to as ‘fetal alcohol effects’, ‘alcohol-related birth defects’ or ‘alcohol-related neurobehavioural disorder’. Categories J–V are new categories that describe a large number of patient groups who have never been adequately classified or described in the past. The Likert ranks for the four digits of the code are case defined for consistent application. The case definitions are presented briefly below and are more fully presented in the diagnostic guide for FAS (Astley and Clarren, 1999).

Case defining the growth component of the 4-Digit Diagnostic Code

Age- and gender-adjusted height- and weight-centiles are ranked by circling A, B or C in the ABC score table (Table 2A). The Height–Weight ABC score recorded in Table 2A is converted to a 4-Digit Diagnostic Code rank using Table 2B. The 4-Digit Diagnostic Code rank is transferred to the 4-Digit Diagnostic Code grid (Fig. 1). Detailed instructions are provided in the diagnostic guide for FAS (Astley and Clarren, 1999) for ranking growth when growth measures are available at more than one time point.

View this table:
Table 2.

ABC score and case definitions for growth deficiency

Circle the ABC Score for
ABC rankCentile rangeHeightWeight
(A)C≤3rdC C
B>3rd and ≤10th B B
A>10thAA
4-Digit Diagnostic Code rankGrowth deficiency categoryHeight–weight ABC score combinations
(A) The first step in deriving the Likert rank for growth is to derive the ABC score for growth. If apatient's height centile was 8% and weight centile was 2%, an ABC score of BC would beassigned.
(B) The final step in deriving the Likert rank for growth is to convert the ABC score for growth into a4-Digit Diagnostic Code rank. A score of BC translates into a 4-Digit Diagnostic Code rank of 3.This rank would serve as the first digit in the 4-Digit Diagnostic Code (see Fig. 1).
(B)4SevereCCC
3 Moderate CB,BC
2MildCA, BB, AC
1NoneBA, AB, AA

Case defining the facial phenotype component of the 4-Digit Diagnostic Code

Three key diagnostic features characterize the FAS facial phenotype: small palpebral fissures, a smooth philtrum, and thin upper lip (Clarren and Smith, 1978; Astley and Clarren, 1996). Palpebral fissure length (PFL) z-scores are computed with adjustment for age and when possible, race (Hall et al., 1989). The thinness of the vermilion border of the upper lip and the smoothness of the philtrum are coded independently on 5-point pictorial Likert scales using Fig. 2. Lips must be gently closed with no smile. The magnitude of palpebral fissure length deficiency, philtrum smoothness, and upper lip thinness are ranked by circling A, B, or C in each column in the ABC score table (Table 3A). The ABC score is con-verted to the 4-Digit Diagnostic Code rank for face using Table 3B.

View this table:
Table 3.

ABC score and case definitions for facial phenotype

Circle the ABC scores for
Five-point Likert scale for philtrum and lipZ-Score for largest PFLPalpebral fissurePhiltrumUpper lip
(A)4 or 5≤ –2 SD C CC
3> –2 SD and ≤ –1 SDBB B
1 or 2> –1 SDA A A
4-Digit Diagnostic Code rankLevel of expression of FAS facial phenotypePalpebral fissure, philtrum, lip ABC score combinations
(A) The first step in deriving the Likert rank for facial phenotype is to derive the ABC score for facialphenotype. If a patient's palpebral fissure lengths (PFL) were > 2 SD below the norm andtheir philtrum and upper lip received Likert scores of 2 and 3 respectively (Fig. 2), the facial phenotype would receive an ABC score of CAB.
(B) The final step in deriving the Likert rank for facial phenotype is to convert the ABC score forfacial phenotype to a 4-Digit Diagnostic Code rank. A CAB score translates into a 4-Digit DiagnosticCode rank of2. This rank would serve as the second digit in the 4-Digit Diagnostic Code (Fig. 1).
(B)4SevereCCC
3ModerateCCB, CBC, BCC
2 Mild CCA, CAC, CBB, CBA,CAB, CAA
BCB, BCA, BBC, BAC
ACC, ACB, ACA, ABC, AAC
1AbsentBBB, BBA, BAB, BAA
ABB, ABA, AAB, AAA
Fig. 2.

Lip-philtrum guide.

Pictorial examples of the five-point Likert scale, upper lip circularity scale, and the ABC scale used to rank upper lip thinness and philtrum smoothness. Circularity (perimeter2/area) is a continuous measure of upper lip thinness that can be used to facilitate the ranking of the upper lip. Circularity ranges from 12.8 for a circle to infinity as the circle is squashed into a line (or becomes thinner). Circularity is measured by outlining the vermilion border of the upper lip using image analysis software such as SigmaScan Pro (1996) (Astley and Clarren, 1996; Astley et al., 1999). It is important that the individual's lips are gently

closed with no smile.

Facial phenotype can be assessed directly or from digitized facial photographs (Astley and Clarren, 1996; Astley et al., 1999). It has been our experience that palpebral fissure length and upper lip thinness can be more accurately measured from digitized photographs using image analysis software (SigmaScan, 1996). A standardized, digital, facial photograph is taken of the patient with an internal measure of scale (2-cm sticker) placed on the forehead. The image is displayed on a computer monitor and PFL is measured by clicking the mouse on the endocanthion and exocanthion landmarks and comparing the distance between the landmarks relative to the size of the internal measure of scale. Upper lip thinness is measured by tracing the outline of the vermilion border with the mouse and computing circularity (perimeter2/area). The thinner the lip the smaller the circularity. Circularity is used to guide the five-point ranking of upper lip thinness as demonstrated in Fig. 2. All patients seen in the FAS DPN clinic have their digital facial photographs analysed during their diagnostic evaluation. The process takes approximately 10 minutes and is described in detail in the FAS Tutor™ CD-ROM (Astley et al., 1999).

Case-defining the brain damage/dysfunction component of the 4-Digit Diagnostic Code

Brain damage/dysfunction is the most significant disability for individuals damaged by prenatal alcohol exposure. Ethanol alters the developing brain in a variety of ways, from structural to gross anomalies of grey and/or white matter and/or to subtle alterations in neurochemical levels (West, 1986). Accurately quantifying and qualifying brain damage/dysfunction is important for both diagnosis and treatment planning. Brain damage can be defined in a large number of ways that are each associated with a broad spectrum of disability. The four-point Brain Damage Likert Scale (Table 4) allows the clinician to separate patients with clear evidence of brain damage (Likert rank 4) from patients with no evidence of brain damage (Likert rank 1). The 4-Digit Diagnostic Code rank for brain does not rank the severity of structural, neurological or functional problems faced by the patient. Rather, it ranks the strength of evidence supporting the presence of an organic cause for cerebral/cerebellar dysfunction.

View this table:
Table 4.

Case definitions for brain damage

4-Digit Diagnostic Code rankBrain damage scaleConfirmatory findings
A patient presenting with a head circumference below the second centile would receive a 4-Digit Diagnostic Code rank of 4. This rank would serve as the third digit in the 4-Digit Diagnostic Code (Fig. 1). OFC, occipitofrontal circumference; FSIQ, full-scale intelligence quotient.
4 Definite Microcephaly, OFC ≤ –2 SD and/or
Referred to as static encephalopathyAbnormalities on brain images diagnostic of prenatal alteration and/or
Evidence of persistent neurological findings likely to be of prenatal origin and/or
FSIQ score ≤ 60
3 Probable Substantial deficiencies or discrepancies across multiple areas of brain performance, such as cognition, achievement, adaptation, neurological ‘soft’ signs, and language.
Referred to as static encephalopathyThree or more areas should be found aberrant
2 Possible Historical information/personal observations strongly suggest the possibility of brain damage, but data up to now do not permit a rank 3 or 4 classification
Referred to as neurobehavioural disorder
1 Absent No problems likely to reflect brain damage are presented

A rank of 4 is reserved for patients who present with ‘medical’ evidence of structural or neurological brain damage. Examples include any one of the following: microcephaly, structural alterations on brain-imaging studies, hard neurological findings, such as a primary seizure disorder or cerebral palsy, or a full-scale intelligence quotient (FSIQ) that is clearly below the normal distribution (<60).

A rank of 3 is reserved for patients who present with ‘psychometric’ evidence of brain damage. Clearly, there are patients who have organic brain damage at a level not detectable by the current technology that allows us to derive a rank 4. In the absence of advanced technology, we feel it is important to identify patients who present with cognitive/ behavioural dysfunction as measured on standardized psychometric tests. At this time, we case define rank 3 to mean that a patient has had an age-appropriate battery of tests in the areas of intelligence, adaptation, academic achievement, language, and neuropsychology. The pattern of abnormality on the test battery, when taken as a whole, must be clinically interpreted by the assessing team to strongly support abnormal brain function. Patients who do not meet the criteria for a rank 4, yet have psychometric test outcomes that document abnormal brain function (greater than 2 SD below the mean) across three or more areas listed above, receive a rank 3. Although there are no scientific data to support that a criterion of three or more failures is more reflective of brain damage than a criterion of one or two failures, our experience with over 1000 patients has demonstrated that the criteria we have selected have good face validity (e.g. the team is more likely to clinically interpret the battery as a whole as strongly supporting abnormal brain function when there are three or more failures). We anticipate that further clinical research coupled with rapidly advancing technology is likely to provide more objective scientific data from which to judge the validity of these criteria. It is important to note that it is possible for a patient to meet the criteria for both a rank 3 and rank 4, since these are not mutually exclusive categories. If this occurs, the higher rank (rank 4) is inserted into the 4-Digit Diagnostic Code, because, for diagnostic purposes, it reflects the strongest clinical evidence of brain damage. The psychometric outcomes, whether normal or abnormal, facilitate the development of the treatment plan for all patients.

Likert rank 2 is given to two subgroups of patients. All patients in rank 2 should have histories of behavioural and/or cognitive problems that strongly suggest underlying brain dysfunction. One group of patients has not yet had the types of testing that would move them into rank 3 or 4, if positive. The reason for this lack of testing is usually because the patients are too young to be reliably or conclusively tested (i.e. less than 6 years of age). The other group of patients includes those who have had testing that did not reveal compelling evidence for rank 3 or 4 classification, and yet, in the clinician's judgement, a strong possibility of brain damage cannot be entirely dismissed. Alternative testing and/or follow-up testing should usually be considered. If adequately sensitive and appropriate testing has been carried out without clear evidence of brain dysfunction, it is unlikely a rank 2 classification would be given.

Patients are classified as rank 1 when no structural, neurological or cognitive/behavioural problems measured by clinical/ psychometric assessment or caregiver interview are discerned.

Case defining the gestational alcohol exposure component of the 4-Digit Diagnostic Code

Alcohol exposure is ranked according to the quantity, timing, frequency, and certainty of exposure during pregnancy (Table 5). The case definitions address the facts that exposure information is often unavailable and/or inaccurate and a clear consensus is not available concerning the amount of alcohol that can actually be toxic to each individual foetus (Stratton et al., 1996). The case definitions differentiate four clinically meaningful exposure groups (4: confirmed high exposure; 3: confirmed exposure, but level is low or unknown; 2: unknown exposure; 1: confirmed absence of exposure).

View this table:
Table 5.

Case definitions for prenatal alcohol exposure

4-Digit Diagnostic Code rankPrenatal alcohol exposure categoryDescription
The case definitions used to derive the 4-Digit Diagnostic Code rank for alcohol exposure. If a birth mother reported drinking a fifth of alcohol (757 ml) several times a week throughout pregnancy, alcohol exposure would receive a 4-Digit Diagnostic Code rank of 4. This rank would serve as the fourth digit in the 4-Digit Diagnostic Code (Fig. 1).
4 High risk Alcohol use during pregnancy CONFIRMED and
Exposure pattern is consistent with the medical literature placing the foetus at ‘high risk’ (generally
high peak blood-alcohol concentrations delivered at least weekly in early pregnancy)
3Some riskAlcohol use during pregnancy CONFIRMED and
Drinking occurred in gestation in frequencies and volumes less than in rank 4 or exact amounts unknown
2Unknown riskGestational exposure is simply not known or information is of questionable reliability
1No riskAlcohol use during pregnancy is CONFIRMED to be completely ABSENT

High exposure is defined generally to be a blood-alcohol concentration greater than 100 mg/dl (a level that typically can be reached by a 55-kg woman consuming six to eight beers) weekly, early in pregnancy. In the absence of a clear consensus on the amount of alcohol that can actually be toxic to the foetus, this general definition should only serve as a guide, not a threshold. One example of a ‘rank 4’ exposure is birth mother reported drinking to intoxication weekly throughout pregnancy. Two examples of ‘rank 3’ exposures include: (a) birth mother was observed to be drinking during pregnancy, but the amount is unknown; (b) birth mother reported drinking 1 glass of wine once a week, but stopped drinking as soon as she learned she was pregnant. A few examples of when alcohol exposure is ultimately unknown and thus coded as a ‘rank 2’ include: (a) the child is adopted and the records are closed; (b) birth father reports birth mother drank while pregnant, but birth mother reports she did not drink; (c) birth mother started drinking at the age of 13 years, was never known to have a prolonged period of sobriety, thus the family assumed she drank during pregnancy.

Other prenatal and postnatal exposures/experiences

A comprehensive diagnostic process must take into consideration the risks associated with prenatal and postnatal exposures and experiences other than prenatal alcohol exposure. Most of the features associated with FAS are not specific to prenatal alcohol exposure. A variety of prenatal (poor prenatal care, prenatal complications, familial genetics, and exposure to other potentially teratogenic agents, etc.) and/or postnatal (physical/sexual abuse, disrupted placement histories, head injuries, chronic substance abuse by the patient, etc.) events could explain all or some of the symptoms presented by the patient. The 4-Digit Diagnostic Code method requires the clinician to record pertinent prenatal and postnatal exposures and events on the standardized FAS Diagnostic Evaluation Form, rank their severity using case defined four-point Likert scales and report them in the standardized medical summary template provided in the diagnostic guide for FAS (Astley and Clarren, 1999).

Comparison of the gestalt and 4-Digit Diagnostic Code Methods

The gestalt (Sokol and Clarren, 1989) and 4-Digit Diagnostic Code outcomes for 454 patients who initially received gestalt diagnostic evaluations at the University of Washington FAS DPN clinic are compared in Tables 6 and 7. Table 6 presents a cross-tabulation of the gestalt and 4-Digit Diagnostic Code outcomes. Table 7 illustrates the variable magnitude of expression of the key diagnostic features of FAS (growth, face, brain, and alcohol) for the three gestalt diagnostic outcomes (FAS, PFAS, and PFAE) and for the equivalent 4-Digit Diagnostic Code categories (categories A and B are equivalent to the gestalt FAS; category C is equivalent to the gestalt PFAS; and categories E–I would be equivalent to the gestalt category of PFAE). The study population was 57.7% male, ranged in age from birth to 51 years old with a mean ± SD of 10.1 ± 7.0 years and had the following racial distribution: Caucasian (57.5%), African American (9.0%), Native American/Alaskan (14.1%), other (19.4%). Race, age, and gender were equally distributed across the 4-Digit Diagnostic Code and gestalt diagnostic categories.

View this table:
Table 6.

Cross-tabulation of gestalt and 4-Digit Diagnostic Code outcomes

Gestalt diagnostic category
4-Digit Diagnostic Code categoryFAS (n = 69)AFAS (n = 41)PFAE (n = 344)Total (n = 454)
Cross-tabulation of the gestalt and 4-Digit Diagnostic Code outcomes was made for 454 patients diagnosed by both methods in the Washington State FAS DPN clinics.
AE, alcohol exposure during gestation confirmed; AE unknown, alcohol exposure during gestation unknown; FAS, fetal alcohol syndrome; AFAS, atypical fetal alcohol syndrome; PFAE, possible fetal alcohol effects.
A FAS (AE)82010
B FAS (AE unknown)1001
C Atypical FAS (AE)122216
E Sentinel physical findings/static encephalopathy (AE)10101737
F Static encephalopathy (AE)886973
G Sentinel physical findings/neurobehavioural disorder (AE)15111541
H Neurobehavioural disorder (AE)117179197
I Sentinel physical findings (AE)0167
J No cognitive/behavioural or sentinel physical findings (AE)011819
K Sentinel physical findings/static encephalopathy (AE unknown)1023
L Static encephalopathy (AE unknown)0156
M Sentinel physical findings/neurobehavioural disorder (AE unknown)2057
N Neurobehavioural disorder (AE unknown)102526
P No cognitive/behavioural or sentinel physical findings (AE unknown)0011
View this table:
Table 7.

Comparison of gestalt and 4-Digit Diagnostic Code outcomes

Diagnostic outcomes
FASAFASPFAEAll other
4-Digit Diagnostic Code ranks for key diagnostic featuresGestalt (n = 69)4-Digit1 (n = 11)Gestalt (n = 41)4-Digit2 (n = 16)Gestalt (n = 344)4-Digit3 (n = 365)4-Digit4(n = 62)
Magnitude of expression of key FAS diagnostic features was compared between the gestalt (Sokol and Clarren, 1989) and 4-Digit Diagnostic Code (Astley and Clarren, 1999) outcomes of 454 patients diagnosed by both methods in the Washington State FAS DPN clinics.
FAS, fetal alcohol syndrome; AFAS, atypical fetal alcohol syndrome; PFAE, possible fetal alcohol effects. 14-Digit diagnostic code categories A and B; 24-Digit Diagnostic Code category C; 34-Digit Diagnostic Code categories E–I; 4all other 4-Digit Diagnostic Code categories D, J–V; 5defined in Tables 2A and 2B; 6defined in Tables 3A and 3B; 7defined in Table 4; 8defined in Table 5.
Growth deficiency5 (n)
1 None37028829529557
2 Mild1001224312
3 Moderate835211172
4 Significant14874142262
FAS facial phenotype6 (n)
1 Absent0010716012
2 Mild27021024725343
3 Moderate15011620364
4 Severe27118106173
Brain damage7 (n)
1 Unlikely002025720
2 Possible29018022423833
3 Probable1329744534
4 Definite27912951675
Prenatal alcohol exposure8 (n)
1 No risk0000000
2 Unknown risk511038043
3 Some risk2651721601888
4 High risk385231414617711

Of the 69 patients who received a gestalt diagnosis of FAS, only nine met the 4-Digit Diagnostic Code criteria for FAS (categories A and B) (Table 6). In the absence of specific case definitions, quantitative measurement scales and only three diagnostic choices (FAS, PFAS, or PFAE), the gestalt method for diagnosing FAS produced a very heterogeneous population, more heterogeneous than would be supported by the gestalt guidelines (Sokol and Clarren, 1989). For example, 37 of the 69 patients had no evidence of growth deficiency, 27 had only one of the three facial features, 29 had no psychometric or structural evidence of brain damage, and five had unknown exposure to alcohol (Table 7). Of the 344 patients who received a gestalt diagnosis of PFAE, the outcomes of these patients are also remarkably variable. These patients fall into 13 different 4-Digit Diagnostic Code categories (Table 6) and present with every combination of diagnostic features (Table 7). The term PFAE clearly fails to convey the diversity of outcomes within this group. Some patients received a diagnosis of PFAE based solely on alcohol exposure (n = 18), whereas other patients received a diagnosis of PFAE based on outcomes that fell just short of the full syndrome (n = 2). Research studies that treat this diverse group of patients as one ‘homogeneous’ group are at great risk of failing to identify clinically meaningful outcomes.

Precision: inter- and intra-rater reliability

The 4-Digit Diagnostic Codes of 20 randomly selected patient files were rederived by S.K.C. and S.J.A. independently while masked to the original 4-Digit Diagnostic Code that had been derived 1 to 4 years ago by the University of Washington clinical team. The codes rederived by S.K.C. and S.J.A. matched the original 4-Digit Diagnostic Codes across all four digits for all 20 subjects (inter- and intra-rater reliability was 100%, κ = 1.0, P = 0.000). The 4-Digit Diagnostic Codes for the 20 randomly selected patients spanned the entire spectrum of normal to FAS (1124 to 1444). Inter-rater reliability between the six FAS DPN regional clinics and the University of Washington FAS DPN core clinic resulted in an exact match across all four digits on 15 of 16 (94%) patients (κ = 0.93, P = 0.000) and an exact match on diagnostic category on all 16 (100%) of the patients (κ = 1.0, P = 0.000). The one 4-Digit Diagnostic Code that did not match was coded by the regional FAS DPN clinic as 1223 and the University FAS DPN clinic as 1123. The mismatch in the facial score was due to the network physician not pulling the epicanthal fold back before measuring the PFL, resulting in an underestimate of the length.

POWER

To demonstrate the statistical power of the 4-Digit Diagnostic Code over the gestalt method of diagnosis, the hypothesis that the FSIQ decreases with increasing magnitude of expression of the FAS facial phenotype was tested among 216 patients who had been diagnosed by both the gestalt and 4-Digit Diagnostic Code systems. Of the 216 patients, 31 received a gestalt diagnosis of FAS. The difference in the mean FSIQ between the patients with and without the gestalt FAS facial phenotype (82.3 and 85.0 respectively) was not statistically significant (t = –1.56, P = 0.13). In contrast, when the same 216 patients were classified by their four-point Likert rank, reflecting the magnitude of expression of the FAS facial phenotype, a statistically significant, inverse, linear association was revealed. The mean FSIQ among the patients with Likert facial ranks of 4, 3, 2, and 1 were 78.5, 83.8, 84.8, and 87.7 respectively (F = 4.1, P = 0.04). The power of the t-test to detect a contrast in facial phenotype between the two gestalt groups was only 23%, whereas the power of the ANOVA to detect the linear trend was 85%. By convention, the minimum power of a clinical research study is set at 80% (Hulley and Cummings, 1988). Thus, a clinically important linear association between face and brain, that was detected by the 4-Digit Diagnostic Code, failed to be detected by the gestalt method of diagnosis.

DISCUSSION

The 4-Digit Diagnostic Code method has been used in all seven FAS DPN clinics in Washington State for over 3 years, demonstrating that it can be taught to a broad array of social and healthcare professionals in an equally broad array of clinical settings. Some of our FAS DPN colleagues were hesitant to make diagnoses in patients with prenatal alcohol exposure prior to using this system, precisely because the old nomenclature was too simplistic and did not offer consistent or helpful diagnostic outcomes. After 3 years of field testing this method, both prospectively and retrospectively on over 1000 patients, it continues to uniformly reflect clinical judgement (a measure of face validity) and provide tremendous power to identify clinically meaningful patterns of outcome.

The 4-Digit Diagnostic Code presents with many strengths. It offers an intuitively logical digital approach to reporting outcomes and exposure that reflects the true diversity and continuum of disability associated with prenatal alcohol exposure. Preliminary assessments of precision, accuracy and power appear to be greatly increased over the ‘gestalt’ method of diagnosis. This can be attributed, in large part, to the use of objective, ordinal and continuous measurement scales, specific, comprehensive case definitions (Polit and Hungler, 1995), and the use of a multidisciplinary clinical team approach. This study as well as others (Abel, 1990; Hannigan et al., 1992) have demonstrated that the current gestalt approach to diagnosis can often lead to diagnoses of FAS made solely on exposure, made in the absence of CNS dysfunction or made when only a single facial anomaly is present. The 4-Digit Diagnostic Code prevents this from occurring. Outcomes and exposures are reported independently so as not to imply that an individual's disabilities and/or anomalies are confirmed to be caused by their prenatal alcohol exposure. The 4-Digit Diagnostic Code serves as a standardized, descriptive language that will allow clinicians and researchers to communicate clearly and objectively the exposures and outcomes of their patients. Although the FAS DPN has gone one step further and clinically categorized and labelled the codes, use of the 4-Digit Diagnostic Code is independent of this step, much like measuring and reporting birth weight in grams and centiles is independent of defining the cut-off for ‘low birth weight’. Failure to reach consensus on the categorization and labelling of the codes need not prevent the use of the 4-Digit Diagnostic Code. The 4-Digit Diagnostic Code is fully comprehensive. It can be used to diagnose individuals of all ages and races who present across the full spectrum of exposure and outcomes. This is achieved by directing the clinician to age-, gender-, and race-adjusted anthropometric and psychometric measures when available and appropriate. The availability and reliability of outcome and exposure information varies across patients. The derivation of the 4-Digit Diagnostic Code addresses this reality by encoding both the presence and absence of outcome and exposure information. This method can be taught to a wide array of healthcare and social service providers, thus greatly expanding the availability of diagnostic services. Multidisciplinary clinical teams from six States in the USA and three Canadian Provinces have been trained to use the 4-Digit Diagnostic Code to date.

Although the 4-Digit Diagnostic Code was developed for prospective use in a fully staffed, multidisciplinary clinic, it can also be used in active and passive screening and surveillance efforts. Surveillance generally uses methods distinguished by their practicality, uniformity, and frequently their rapidity, rather than by complete accuracy (Last, 1988). A key feature of the 4-Digit Diagnostic Code is that it not only documents exposure and outcome, but also documents how much data were available (or not available) to support the diagnostic outcome. The standardized FAS Diagnostic Evaluation Form served as an efficient tool for conducting the retrospective chart review on the 736 patients whose gestalt diagnoses were upgraded to 4-Digit Diagnostic Codes. The method provides an efficient and reproducible tool for conducting retrospective chart reviews, a process that is the very essence of passive surveillance. The computerized facial analysis component of the 4-Digit Diagnostic Code also serves as an efficient and highly effective photographic method for screening for FAS (Astley et al., 1999). This method is currently being used to screen all children entering foster care in one county in Washington State.

Meaningful progress in the areas of screening, diagnosis, intervention, surveillance, and primary prevention all hinge on development of an accurate, precise, valid, and efficient method for identification of individuals damaged by prenatal alcohol exposure. The 4-Digit Diagnostic Code was developed to achieve that goal in Washington State.

Acknowledgments

We wish to acknowledge our clinical colleagues Diane Bailey, Sharon Beck, Heather Carmichael Olson, Sandra G. B. Clarren, Truman Coggins, Tracy Jirikowic, Robin LaDue, Christina Talbot, and the clinical teams at the FAS DPN clinics who have helped us to sharpen this diagnostic system. We also wish to thank Kathleen Tharp, Julie Gelo, Kathryn Briggs-Jones, Heather Grigg, Heather Wicklein Sanchez, and Cliff Astley who readily offered their assistance when needed. Finally, a very special thanks is extended to all of our patients and their families who have contributed a wealth of knowledge and information to the development of this methodology. The development of this 4-Digit Diagnostic Code was supported in part by the following agencies and contributors: Centers for Disease Control and Prevention; Center for Human Development and Disability at the University of Washington, Seattle, Washington; Division of Alcohol and Substance Abuse, Washington State Department of Social and Health Services; March of Dimes Birth Defects Foundation, and the John B. Chavez FAS Memorial Fund.

Footnotes

  • * Author to whom correspondence should be addressed at: Children's Hospital and Medical Center, 4800 Sand Point Way NE, CH-47, Seattle, WA 98105, USA.

REFERENCES

View Abstract