Prenatal alcohol exposure and family history of alcoholism in the etiology of adolescent alcohol problems. John S. Baer; Helen M. Barr; Fred L. Bookstein; Paul D. Sampson; Ann P. Streissguth.

Abstract: A study was conducted to examine adolescent drinking problems originating from prenatal alcohol exposure and family history of alcoholism. A group of 439 parents reported information about family history and problems of alcohol with their adolescent offspring. The 14-year old adolescents also reported information about the frequency and quantity of their alcohol consumption. The study found that prenatal alcohol exposure is predictive of adolescent alcohol use.

Author's Abstract: COPYRIGHT 1998 Alcohol Research Documentation, Inc.

Objective: To examine the relative importance of prenatal alcohol exposure and family history of alcoholism for the prediction of adolescent alcohol problems. Method: In 1974-75, a populationbased, longitudinal prospective study of alcohol and pregnancy began with self-report of alcohol use by pregnant women. In a 14-year followup, 439 parents provided information on the family history of alcohol problems for these adolescent offspring. The 14-year-old adolescents provided information on the frequency and quantity of their own alcohol consumption within the past month, on the consequences of their drinking over the past 3 years, and on their age at first intoxication. Additional covariates were assessed prenatally and at follow-up. Results: Prenatal alcohol exposure was more predictive of adolescent alcohol use and its negative consequences than was family history of alcohol problems. Prenatal exposure retained a significant predictive effect even after adjustment for family history and other prenatal and environmental covariates. By contrast, the nominally significant correlation of family history with adolescent drinking is weaker after adjustment for prenatal alcohol exposure and disappears entirely after adjustment for other relevant covariates. We observed no evidence for an interactive effect of fetal exposure and family history in predicting adolescent alcohol use. Conclusions: Fetal alcohol exposure is a risk factor for adolescent alcohol involvement and alcohol-related problems and may account for variance in prediction of problems otherwise attributed to family history of alcoholism. Studies of alcoholism etiology and family history need to include consideration of even modest levels of fetal alcohol exposure. (J. Stud. Alcohol 59: 533-543, 1998)

Full Text: COPYRIGHT 1998 Alcohol Research Documentation, Inc.

LONG-TERM NEGATIVE consequences of prenatal alcohol exposure have been demonstrated both in experimental studies of laboratory animals and in longitudinal studies of humans. Fetal alcohol effects include growth deficiency, attention problems, memory problems, and deficits in information processing (Streissguth et al., 1996). To date, however, fetal alcohol exposure has not generally been implicated in the etiology of alcoholism. Here we provide data from our 14-year longitudinal study pertaining to the relative roles of fetal alcohol exposure and family history of alcoholism in the prediction of adolescent alcohol problems.

The etiology and development of alcohol problems have received considerable attention for many years. Several reviews have concluded that children of alcoholics (COAs) are 3 to 4 times more likely to suffer from alcoholism than are children of nonalcoholics (nonCOAs). Evidence in support of this contention is based on a higher concordance for alcoholism among monozygotic twins than among dizygotic twins (e.g., McGue et al., 1992), and higher rates of alcohol problems among adopted-out COAs compared to nonCOAs (e.g. Cloninger et al., 1981). There is a large literature devoted to explaining multiple pathways (Babor et al., 1992) and multiple mechanisms of genetic risk of alcoholism, including factors such as sociopathy, impulsivity, neuropsychological performance and alcohol sensitivity (Schuckit and Smith, 1996; Sher, 1991; Zucker et al., 1995).

To our knowledge, however, no study of the heritability of alcohol problems has benefited from a study design that provides the data necessary to distinguish the effects of family history from the direct teratogenic effect of maternal prenatal alcohol use. The issue of prenatal alcohol exposure was noted by Cloninger et al. (1981) in their cross-fostering study. They stated that "strictly, the congenital predisposition of the adoptees includes not only genetic contributions from the biological parents but also the influence of maternal intrauterine environment" (p. 863; see also Bohman et al., 1981). Yet cross-fostering studies for alcoholism outcomes have not measured maternal drinking during pregnancy. Some do not acknowledge this potential confounding variable (e.g., Goodwin et al., 1973, 1977). Although the drinking of biological fathers contributes much to statistical prediction of problems in biological offspring in cross-fostering studies (e.g., Cloninger et al., 1981), such analyses do not account for assortative mating within parents. In fact, assortative mating has been documented with respect to alcohol use patterns (Hall et al., 1983). Ruling out maternal "alcoholic" behavior in cross-fostering analyses does not control for lower level drinking during pregnancy that can have deleterious effects on the adjustment of children (Carmichael Olson et al., 1997; Day, 1995; Goldschmidt et al., 1996).

Fetal alcohol exposure may also present a confound within twin studies. In twin studies, of course, maternal drinking is held constant and assortative mating is not an issue. Yet fetal alcohol exposure may interact with genetic susceptibility and thus account for differential rates of concordance of observed problems among twins. In one study, where the effects of teratogens were examined with respect to both monozygotic and dizygotic twins, higher concordance for fetal alcohol effects were observed among monozygotic twins (5/5) than were observed for dizygotic twins (7/11) (Streissguth and Dehaene, 1993).

Fetal alcohol exposure is of special concern because it is a possible explanation for some of the conduct problems and neuropsychological risk factors associated with family history of alcoholism. Pihl and Bruce, (1995), for example, conclude that at least a subgroup of COAs demonstrate deficits in verbal skills, abstract thinking, goal-directed planning, attention processes, visuospatial abilities, and learning and memory. Such differences are typically attributed to genetic processes. Yet findings vary considerably from study to study, and effect sizes tend to be small. Several authors have offered methodological critiques of this literature, raising concerns about operational definitions and sampling (Bates and Pandina, 1992). Fetal alcohol exposure has been addressed occasionally in neuropsychological studies of COAs, typically by ruling out subjects based on maternal alcoholism. In a recent meta-analysis, Pollock and Earleywine (1977) compared neuropsychological studies that did or did not rule out subjects based on maternal alcoholism and found no moderating effects. Yet these studies suffer from limitations similar to those found in genetic studies. Hesselbrock et al. (1991) noted the potential confound of fetal exposure for neuropsychological performance, and further suggested that ruling out maternal alcoholism does not control for assortative mating or low level doses of alcohol during pregnancy. Similar issues can be raised with respect to many psychophysiological and electrophysiological studies of COAs as well (see Hill et al., 1995, for a noteworthy exception).

Exposure to alcohol during pregnancy has been associated not only with fetal alcohol syndrome (FAS), characterized by growth deficiency, a pattern of facial dysmorphology and central nervous system (CNS) dysfunction, but also with more subtle cognitive and learning disabilities (Streissguth et al., 1996). That CNS dysfunction in humans can result from quite moderate doses of alcohol is not surprising, as experimental studies of prenatal alcohol effects using animal models demonstrate CNS effects at lower and more variable alcohol doses than those required to produce physical effects (Bonthius et al., 1996; Goodlett and West, 1992). In large longitudinal studies (Day and Richardson, 1991; Jacobson et al., 1993; Larroque et al., 1995), including our own (Sampson et al., 1989), prenatal alcohol exposure, even at quite moderate levels, has been associated with IQ decrements, attention and memory problems and neurobehavioral deficits during the first years of life (Streissguth et al., 1993), between 7 and 14 years of age (Streissguth et al., 1994) and at the age of 14 (Carmichael Olson et al., 1997). It is quite likely, therefore, that fetal alcohol exposure can lead to cognitive deficits that might otherwise be associated with family history of alcoholism and be implicated in risk models of alcohol involvement (Sher, 1991; Zucker, 1994). Also, although we are not aware of data directly pertaining to humans, several animal models have demonstrated changes in alcohol sensitivity in adults as a result of fetal alcohol exposure (Becker et al., 1995; Fulginiti et al., 1989; Taylor et al., 1981).

In this article we use data from a 14-year follow-up of single-bom children from the Seattle Longitudinal Study on Alcohol and Pregnancy. Interviews of 439 parents provided information on the adolescent's family history of alcohol problems and on the parent's perception of the adolescent's drinking. Independently, 14-year-old adolescents provided information on their own drinking habits and associated problems and their perception of their parent's drinking. In 1974-75, this population-based, longitudinal prospective study began with detailed prenatal interviews of the mothers regarding their pregnancy health habits including alcohol use (a measure of fetal exposure, or dose). We examine the relationships between fetal alcohol exposure and family history of alcohol problems in predicting drinking rate and problems in adolescence using a multivariate methodology. In our longitudinal design we are able to control for several aspects of the postnatal environment that might contribute to the prediction of adolescent alcohol problems.

Method

In 1974-75, 1,529 pregnant women who were receiving prenatal care at either of two hospitals that together represented the demographic characteristics of the Seattle area were interviewed during their fifth month of pregnancy regarding general demographic characteristics, and pregnancy health habits, including nutrition, smoking, alcohol and caffeine use, and use of prescription and nonprescription medications. Families were chosen for longitudinal follow-up using a strategy that included as many of the heaviest drinkers as possible while also reflecting a continuum of prenatal alcohol use. Cigarette use was stratified across alcohol as measured by an a priori alcohol scale. For greater detail regarding study design and methodology, strengths and weaknesses, see Streissguth et al. (1993).

As part of the 14-year follow-up conducted in 1989-90, 464 participating families were seen at the University of Washington Medical School. This represents 82% success of follow-up since birth with no differential loss of adolescents born to mothers using alcohol more heavily during pregnancy. Only 10 of the 14-year-olds were no longer living with either biological parent. Subjects described here are the 439 families with complete data from the three relevant data sets: adolescent alcohol use (from adolescent and parent in 1989-90), family history of alcohol problems (from parent report in 1989-90) and prenatal alcohol exposure history (from biological mother report during pregnancy in 197475). Mothers, who were primarily white (90%), averaged ([+ or -] SD) 26.5 [+ or -] 4.6 years of age, were married (89%) and had graduated from high school (90%) at the time of this pregnancy. At examination roughly 14 years later, the adolescents ranged in age from 13 years 11 months to 15 years 8 months (mean age--14 years 5 months); 53% male. For more details on the 14-year cohort characteristics and details of the 14-year testing battery, see Streissguth et al. (1994).

Prenatal alcohol exposure (fetal dose)

Prenatal alcohol was assessed during midpregnancy by self-report of the biological mother in a confidential quantityfrequency-variability interview (Mulford and Miller, 1960) conducted in her own home. Table 1 defines the alcohol scores and presents data on levels of drinking among drinkers only. Of these mothers 20% abstained from alcohol during both gestational periods assessed; 324 (74%) drank in the early part of pregnancy before they knew they were pregnant; 342 (78%) reported drinking during pregnancy. The mothers' drinking prior to pregnancy recognition averaged about 11 drinks per week, during midpregnancy about four drinks a week. Prior to pregnancy recognition 39% reported at least one heavy drinking episode (defined as at least five drinks), but during midpregnancy only 24% reported drinking five or more drinks on one occasion. The drinking patterns of mothers in this cohort range from total abstinence to very heavy drinking, but are mostly in the moderate or social range. Only three of these 439 mothers reported any lifetime history of problems due to alcohol such as marriage break-up, being hospitalized, losing a job or getting arrested due to drinking behavior.

TABLE 1. Alcohol scores and descriptive statistics (N = 439). Data pertain to drinkers only

Alcohol scores                                  Mean ([+ or -] SD)
 
Average ounces of absolute alcohol per day
  Prior to pregnancy recognition (AAP)          0.80 [+ or -] 1.78
  During midpregnancy (AAD)                     0.32 [+ or -] 0.60
5 drinks or more on any occasion
  Prior to pregnancy recognition                0.39 [+ or -] 0.49
   ([is greater than or equal to] 5DRINKSP)
  During midpregnancy                           0.24 [+ or -] 0.43
   ([is greater than or equal to] 5DRINKSD)
Average drinks per occasion
  Prior to pregnancy recognition (ADOCCP)       2.48 [+ or -] 1.45
  During midpregnancy (ADOCCD)                  2.19 [+ or -] 1.16
Max. drinks reported on any occasion
  Prior to pregnancy recognition (MAXP)         4.00 [+ or -] 2.56
  During midpregnancy (MAXD)                    3.61 [+ or -] 2.60
Monthly occasions of drinking
  Prior to pregnancy recognition (MOCCP)       16.76 [+ or -] 22.70
  During midpregnancy (MOCCD)                   8.00 [+ or -] 11.13
Quantity-Frequency-Variability Index
  Prior to pregnancy recognition (QFVP)         3.51 [+ or -] 1.03
  During midpregnancy (QFVD)                    3.02 [+ or -] 0.93
Ordered Exposure Code
  combines timing, dose, pattern (ORDEXC)       2.78 [+ or -] 1.22
 
Alcohol scores                                Min.   Med.    Max.
 
Average ounces of absolute alcohol per day
  Prior to pregnancy recognition (AAP)        0.01   0.39   25.76(a)
  During midpregnancy (AAD)                   0.01   0.16    8.55
5 drinks or more on any occasion
  Prior to pregnancy recognition                 0      0    1.00
   ([is greater than or equal to] 5DRINKSP)
  During midpregnancy                            0      0    1.00
   ([is greater than or equal to] 5DRINKSD)
Average drinks per occasion
  Prior to pregnancy recognition (ADOCCP)      1.5    2.2    13.0
  During midpregnancy (ADOCCD)                 1.5    1.7    13.0
Max. drinks reported on any occasion
  Prior to pregnancy recognition (MAXP)        1.5    3.5    13.0
  During midpregnancy (MAXD)                   1.5    3.5    13.0
Monthly occasions of drinking
  Prior to pregnancy recognition (MOCCP)       0.3    9.0   240.0
  During midpregnancy (MOCCD)                  0.3    4.8   105.3
Quantity-Frequency-Variability Index
  Prior to pregnancy recognition (QFVP)        2.0    3.0     5.0
  During midpregnancy (QFVD)                   2.0    3.0     5.0
Ordered Exposure Code
  combines timing, dose, pattern (ORDEXC)      1.0    3.0     4.0

Notes: AA (average oz of absolute alcohol per day) is a continuous variable; AA = 1.00 means an average of two drinks per day of wine, beer, spirits, or any combination. P (n = 324) refers to the month or so prior to pregnancy or pregnancy recognition; D (n = 342) refers to drinking during midpregnancy, assessed at the fifth month of pregnancy. [is greater than or equal to] 5DRINKS is a dichotomous variable representing whether or not five or more drinks were reported on at least one occasion. This score combines the VV = 5, VV = 8 and VV = 11 of Cahalan et al., 1969. ADOCC represents the average number of drinks reported per drinking occasion. MAX is the maximum number of drinks reported for any drinking occasion. MOCC is the number of occasions per month in which drinking is reported. QFV is a categorical score (quantity, frequency, variability) summarizing three dimensions of drinking from Cahalan et al., 1969. The order has been reversed for consistency with the other drinking scales, so that 5 corresponds to the heaviest drinking. ORDEXC is an a priori code (ordered exposure code) developed at the outset of this study to describe the presumed risk to the fetus of different drinking patterns, in order to enroll women in the follow-up study: 4 is highest presumed risk (n = 351).

(a) The next highest AAP score was 9.14 oz.

Adolescent alcohol use

As part of a 4-hour confidential evaluation, adolescents were given two short questionnaires pertaining to alcohol. The Lifestyle Choices Survey (LCS) includes five questions of adolescents' quantity and frequency of alcohol use, age of first feeling drunk, number of days during the past year that they drank enough to feel drunk and whether they had ever had treatment for alcohol problems (based on Grube and Morgan, 1986). A 13-item alcohol problems scale (Baer and Carney, 1993), a short version of the Rutgers Alcohol Problem Index (RAPI-A) (White and Labouvie, 1989), reflects how often during the past 3 years the adolescent experienced specific social and personal difficulties related to drinking and behaviors indicative of dependence symptoms.

Data on adolescent alcohol use and problems were obtained from adolescent self-report (for the adolescent' s "ever consuming alcohol," parental reports were used as confirmation). Figure 1 presents the six adolescent alcohol scores analyzed as dependent variables for this report. Alcohol was consumed at some time by 252 (ALCINDC, 57%) of the adolescents and during the past month by 111 (AMOCC, 25%); beer, 64 (15%); wine, 38 (9%); wine cooler, 57 (13%); and spirits, 39 (9%); numbers are not additive due to the use of multiple beverage types by some subjects. As estimated number of drinks in the past month and number of drinking occasions are highly correlated (r = .96) only the latter (AMOCC) are used for these analyses. Two adolescents report consuming alcohol at least every other day on average. The usual number of drinks consumed when drinking (not just in the past month) is one to two drinks, but 53/252 (21%) of these young adolescents who have ever had a drink usually have three or more drinks on an occasion (USDOCC), and 137 (31%) of 14-year-olds have at some time felt intoxicated from alcohol.

[Figure 1 ILLUSTRATION OMITTED]

The 13 RAPI-A questions yield a summary score for "negative" consequences due to drinking (RAPIASUM) which is calculated as the simple sum of the 0-4 point frequency codes from the 13 questions. Of the 439 adolescents, 112 (26%) have already experienced at least one consequence of drinking (44% of the 252 who have had a drink). The most frequently reported consequences include "neglected responsibilities" (15%) and "had a bad time" (14%). In addition, three of these consequences (tolerance, perceived problems with alcohol, and perceived dependence) were combined in a binary indicator (ALCDEPND) reflecting the endorsement of at least one criterion for alcohol dependence consistent with DSM-IV (American Psychiatric Association, 1994).

Several other measures of adolescent alcohol use were observed at frequencies too low to use for analyses. Only six of the 14-year-old adolescents had already undergone alcohol counseling/treatment programs; only seven parents said that they were concerned about their adolescent's alcohol use; and no parent requested intervention by our staff in response to adolescent drinking.

Family history of alcohol problems

Family history of alcohol problems was assessed with the Family Tree Questionnaire for Assessing Family History of Drinking/Drug Abuse Problems (FAMTREE/CONSEQ) (Mann et al., 1985). Almost all (93%) of the informants were biological mothers. We assessed history of drinking problems among first-degree and second-degree relatives of the adolescent. The determination of alcohol problems for each relative was based on nine questions that defined five dimensions (social and marital problems, employment, arrests, physical symptoms, and treatment) of alcohol problems. Any family member coded positive for two or more of these five dimensions was considered positive for family history of alcohol problems. Summary scores for family units have been computed in many different ways in the literature. We selected six different ways of using our ratings of first- and second-degree relatives to compute binary indicators of family history of alcohol problems (see Table 2, column a). In order not to confound the definition of family history with the measures of maternal alcohol consumption, the same six indicators were computed excluding consideration of the biological mother (Table 2, column b). As is evident in the table, the number of families identified as positive for history of alcohol problems is almost the same whether or not maternal alcohol problems are included in the definition. Analyses herein use only the six indicators that exclude consideration of the biological mother (Table 2, column b). Sixty biological fathers and 23 mothers are identified as having alcohol problems for a total of 73/439 (17%) families (for ten families, both the mother and father have alcohol problems). Of the 23 families with mothers identified as having alcohol problems, only three have no other first- or seconddegree relative with a problem.

TABLE 2. Indicators of family history of alcohol problems in first- and second-degree relatives

Variable                                            % Yes    % Yes
(each is code: 0: no, 1 = yes)                       (a)      (b)
 
1st/2nd-degree relative has alcohol problem
 (CPA12)                                              56      55
Someone in the family of origin has an alcohol
 problem (CPAfo)                                      40      38
Maternal 1st/2nd degree relative has alcohol
 problem (CPAmat12)                                   38      36
Paternal 1st/2nd degree relative has alcohol
 problem (CPApat12)                                   34      30
1st-degree relative has alcohol problem (CPA1)        18      15
Mother and/or father has an alcohol problem
 (CPApar)                                             17      14

Notes: A positive score on any of the indicators above results from one or more members of the target group having a positive score on the Consequences of Problem Alcohol use scale (CPA). Column (a) shows the percent of adolescents with a positive family history of problems with alcohol by the definitions as stated. Column (b) differs from column

(a) by the exclusion of mother's alcoholism from the calculation. For CPApat12, the restriction is to "males only" on the paternal side of the family. These are the scores used in subsequent analysis (e.g., Figure 3, Table 4).

Covariates

Covariates were assessed at various points in time. Prenatal covariates included basic demographic characteristics such as mother's age, education, SES and parity, and information on her pregnancy use of alcohol, cigarettes, drugs and medicines (concurrent information about prenatal fetal exposures). Stratification for smoking across alcohol categories in the selection of the follow-up cohort from the prenatal interview sample helped to minimize the correlation of alcohol with nicotine.

At the 14-year study, the parent interview asked about other major household changes believed to indicate stress (family changes in residence, foster care, divorces, deaths in the family) and whether the adolescent has lived his or her whole life with cigarette smokers. The adolescent questionnaires and interview contained information about contemporaneous alcohol use in the household (we coded the maximum frequency of alcohol use reported for any parent figure), the Rosenberg Self-Esteem Inventory (SEI) (Rosenberg, 1965) and summary scores from the Parenting Style Survey (PSS) (Sameroff et al., 1990).

[Figure 2 ILLUSTRATION OMITTED]

Statistical analysis

Simple and partial correlations were used to assess the strength of the relationships of prenatal alcohol exposure (13 measures) and family history (6 dichotomous measures) to adolescent alcohol use (6 measures).

We carried out one primary composite analysis of all six adolescent alcohol measures and two other analyses of the measures of consequences of adolescent drinking from the RAPI-A: RAPIASUM and ALCDEPND. Our primary analysis involves three composite latent variable (LV) scores representing the prenatal alcohol exposure/dose measures, the family history indicators and the adolescent alcohol use measures. These are computed by the Partial Least Squares (PLS) method explained in detail and demonstrated in a number of our previous publications (Bookstein et al., 1996; Sampson et al., 1989; Streissguth et al., 1993). In these articles we explain why the alcohol exposure LV score should be computed as a linear combination of nonlinearly transformed individual exposure measures (Table 1), with coefficients proportional to the correlations of the transformed exposure measures, with an adolescent alcohol use latent variable that is similarly defined in terms of its components. We refer to these coefficients as saliences. In this analysis we utilize the modestly nonlinear transformations of the 13 prenatal exposure measures computed in the comprehensive analyses of data from birth through 7 years reported in our earlier monograph (Streissguth et al., 1993, Figure 4.1).

The family history latent variable score is computed from its individual indicators (Table 2) in the same way with respect to an adolescent alcohol use latent variable. That is, the saliences (coefficients) of the family history LV are proportional to the correlations of the individual family history indicators with the adolescent alcohol LV. The adolescent alcohol LV is nearly the same (r = .99) regardless of whether the saliences defining it as a linear combination of the six measures of adolescent alcohol consumption and consequences are computed with respect to the family history LV or the prenatal alcohol LV. Subsequent regression analyses use the adolescent alcohol LV computed with respect to prenatal alcohol.

Regression analyses are used to assess the strength of the relationships of the prenatal alcohol LV and the family history LV with the adolescent alcohol LV and with the two consequence variables, RAPIASUM and ALCDEPND. The analyses describe the extent to which other predictors suggested in the literature and available within our database may alter the interpretation of the basic findings. The potential covariates were prenatal demographic characteristics and cigarette use (six variables), adolescent age and sex, parents' postnatal alcohol and cigarette use (two variables) other postnatal environmental descriptors (12 variables) and adolescent attributes of self-esteem and receipt of special programming in school in response to learning problems. We note that the postnatal environmental descriptors and adolescent self-esteem and learning difficulties may be directly or indirectly influenced by maternal prenatal alcohol consumption or family history. Thus, one must consider carefully the interpretation of conventional multiple regression analyses adjusting for these as covariates.

This study was designed to assess effects of prenatal alcohol exposure. There will be no attempt to launch a study of general predictors of adolescent drinking (a task undertaken by many earlier studies designed for that purpose). Many of these covariates are highly correlated with one another. They will be used here mainly to assess stability of findings regarding the relationship of prenatal alcohol exposure and family history to adolescent drinking and its consequences.

Results

Figure 2 presents the simple and partial correlations of four of the adolescent alcohol-related outcomes with fetal dose measures. For each prenatal dose/adolescent alcohol combination, the solid diamond indicates the simple (unadjusted) correlation, and the nearby dots are the partial correlations deriving from adjustment for each of the six family history indicators. (Relationships are of a strength somewhat greater than that which we have commonly observed across neurobehavioral adolescent outcome measurements [Streissguth et al., 1993].) Three of the four outcome measures in Figure 2 (USDOCC, RAPIASUM, AGEINTOX) as well as ALCINDC and ALCDEPND all correlate near 0.2 with alcohol measures, peaking at a correlation of 0.28 between USDOCC and maximum drinks per drinking occasion prior to recognition of pregnancy, MAXP. Overall average alcohol exposure (as represented by AAP and AAD) does not correlate with adolescent drinking as highly as do ADOCC, MAX and [is greater than or equal to] 5DRINKS scores of prenatal alcohol exposure. As shown in Figure 2, adjustment for any measure of family history of alcohol problems only slightly attenuates these relationships.

Figure 3 displays correlations of adolescent alcohol use outcomes with family history measures. Relationships generally are lower than those with prenatal exposure except in the case of first intoxication for which they reach approximately the same level, 0.19 (Figure 3). The partial correlations (dots below diamonds) display attenuation by adjusting for fetal exposure.

[Figure 3 ILLUSTRATION OMITTED]

Partial least squares (PLS) and regression analyses

The PLS analysis results in a linear combination of alcohol exposure scores that correlates 0.27 with a linear combination of the adolescent drinking variables (Table 3). The saliences for these two latent variables are derived from the 13 x 6 correlation matrix between alcohol exposure and adolescent alcohol, as explained in Streissguth et al. (1993). The elements of (four columns of) this correlation matrix are represented by the diamonds in Figure 2; 98% of the sum of squared correlations is "explained by" these two latent variables. (For further discussion and interpretation of this summary statistic, see Streissguth et al., 1993). The prenatal alcohol LV is a combination of the 13 individual measures weighted toward episodic drinking, with slightly greater weight on the prepregnancy recognition measures. The adolescent alcohol LV correlating with this prenatal alcohol LV gives the greatest weight to drinks per occasion (USDOCC), relatively little weight for the average monthly occasions measure (AMOCC), and intermediate weight to the other four scores.

TABLE 3. Summary of PLS analysis of 13 measures of prenatal
alcohol exposure measures and six measures of adolescent alcohol
use and consequences
 
Prenatal alcohol saliences         Adolescent alcohol saliences
 
ORDEXC                    0.25        ALCINDC        0.33
AAP                       0.15        AMOCC          0.28
AAD                       0.12        USDOCC         0.54
MOCCP                     0.16        AGEINTOX       0.40
MOCCD                     0.15        RAPIASUM       0.46
ADOCCP                    0.36        ALCDEPND       0.40
ADOCCD                    0.34
MAXP                      0.37
MAXD                      0.34
[is greater than or
  equal to] 5DRINKSP      0.30
[is greater than or
  equal to] 5DRINKSD      0.30
QFVP                      0.31
QFVD                      0.31

Prenatal alcohol-adolescent alcohol LV correlation = 0.27

Notes: Acronyms for alcohol scores are explained in Table 1. Acronyms for the adolescent alcohol use and consequences scores are explained in Figure 1.

The family history LV computed in the same way correlates 0.18 with a similar adolescent drinking latent variable (Table 4). The major difference between the definitions of the two adolescent drinking latent variables is in lower saliences for AMOCC and RAPIASUM in the analysis with respect to family history. However, as these adolescent LV scores correlate 0.99, we use only the adolescent alcohol LV score computed with respect to prenatal alcohol exposure in the regression analyses to be described next.

TABLE 4. Summary of PLS analysis of six indicators of family
history of alcohol problems and six measures of adolescent
alcohol use and consequences.
 
Family history saliences          Adolescent alcohol saliences
 
CPA 12            0.41               ALCINDC           0.46
CPAfo             0.33               AMOCC             0.12
CPAmat 12         0.21               USDOCC            0.47
CPApat 12         0.41               AGEINTOX          0.56
CPA1              0.52               RAPIASUM          0.28
CPApar            0.49               ALCDEPND          0.40
 
Family history-adolescent alcohol LV correlations = 0.18

Notes: Acronyms for family history scores are explained in Table 2. Acronyms for adolescent alcohol scores are explained in Figure 1.

A regression model combining the prenatal alcohol and family history LVs for prediction of the adolescent alcohol LV (Table 5, Model 1) shows that prenatal alcohol exposure retains a greater effect after adjustment for family history (t = 5.30, partial r = 0.25) than does family history after adjusting for alcohol exposure (t = 2.36, partial r = 0.11). Thus, prenatal exposure accounts for prediction otherwise attributed to family history. The reverse, that family history accounts for prenatal effects, is not evident. The family history and prenatal alcohol LV scores are correlated 0.23. There is no statistical evidence of an interaction effect expressed as the product of these two latent variable scores (t = 1.00).

TABLE 5. Three regression analyses of the adolescent alcohol
LV score on the prenatal alcohol LV, family history LV score
and covariates
 
                                Model 1             Model 2
 
Coefficients                Value    t value    Value    t value
 
Prenatal alcohol LV         0.16      5.30       0.13     4.29
Family history LV           0.11      2.36       0.04     0.36
Parity (log)                                    -0.36    -2.13
Prenatal nicotine (log)                         -0.21    -3.20
Age at the exam                                 -1.76    -4.13
PS S "control"
PSS "effectiveness"
Self-esteem
                          [R.sup.2] = .083      [R.sup.2]=. 149
 
                                        Model 3
 
Coefficients                     Value      t value
 
Prenatal alcohol LV               0.10       3.34
Family history LV                 0.03       0.55
Parity (log)                     -0.27      -1.70
Prenatal nicotine (log)          -0.20      -3.16
Age at the exam                  -1.50      -3.71
PS S "control"                   -0.08      -3.99
PSS "effectiveness"              -0.19      -4.25
Self-esteem                       0.66       3.37
 
                                  [R.sup.2] = .241

Our data generally support findings of other researchers who report that factors such as alcohol and cigarette use in the home, parenting style, religion, learning problems, poor relationship with parents and poor self-esteem are related to adolescent drinking (Baer et al., in press). We see significant correlations between these factors and adolescent alcohol outcomes in this sample. However, we reiterate that some of these "postnatal covariates" are themselves probably a result of prenatal exposure and are thus not appropriate to consider as covariates for the purpose of adjustment in an analysis of the significance of prenatal exposure effects. A second regression (Table 5, Model 2) shows that alcohol exposure retains its nominal significance after further adjustment for age of the adolescent at the time of the 14-year interview, parity and maternal smoking (prenatal nicotine), but family history does not. Adolescent gender did not enter the model, indicating no gender effects on adolescent drinking. When other postnatal environmental measures are considered (Table 5, Model 3), including measures of parenting style (PSS), current parental drinking, indicators of stressful household changes and an adolescent self-esteem score (SEI), the significant association of adolescent drinking with prenatal alcohol exposure, but not family history, remains. Table 5, Model 3, presents a summary of one of many nearly comparable regression models suggested by the exhaustive search of an "all subsets" regression procedure (Seber, 1977). We explicitly forced the family history LV score into the regression as it would not otherwise have been selected. The adolescent assessment of current parental drinking is likewise insignificant when forced into the regression. Even though it is correlated 0.26 with the alcohol LV score, adjustment for current parental drinking does not greatly affect the estimated effect of prenatal alcohol exposure.

Two measures of parenting style, "control" and "effectiveness," are the strongest correlates of adolescent drinking behavior, both marginally and in this multiple regression model. (Both lack of "control" and lack of "effectiveness" are associated with high levels of adolescent drinking and problems.) As noted previously, measures such as these and the similarly significant adolescent self-esteem score may themselves be reflections of the effects of prenatal alcohol exposure or family history of alcohol problems. In view of this, the stability and significance of the prenatal alcohol LV term across the regressions reported in Table 5 are perhaps even more convincing. That is, fetal alcohol exposure continues to contribute an apparently unique prediction (unaccounted for by other variables) to adolescent drinking behavior. We note also that, in this final analysis, maternal smoking (log nicotine) correlated almost as strongly with the outcome as does prenatal alcohol exposure.

Analysis of the RAPIASUM score for consequences of drinking results in a very similar regression model with a similarly significant alcohol LV effect (t = 2.99) and insignificant contribution from the family history LV. There are some differences in the most relevant postnatal covariates. (The "effectiveness" parenting style measure correlates highly with consequences while the "control" measure correlates mainly with the drinking rate measures, not consequences.) Logistic regression analysis of the binary indicator of alcohol dependency (ALCDEPND) points to similar conclusions (although with slightly lower estimated effects of the alcohol LV: t = 2.7 prior to adjusting for the postnatal factors, t = 2.14 after postnatal adjustment).

Discussion

The current study addresses a potential overlap and confound between alcohol-related factors that have previously been studied separately: fetal alcohol effects and family history of alcoholism. It specifically addresses whether fetal alcohol exposure is predictive of adolescent alcohol problems. Our data not only indicate that fetal alcohol exposure is related to adolescent alcohol use and related problems but also suggest that fetal alcohol exposure can account for effects that might otherwise be attributed to family history of alcohol problems. The predictive relationship of prenatal alcohol exposure for adolescent alcohol problems held even when several aspects of the postnatal environment were statistically adjusted for by linear regression. Consistent with previous research, family history of alcohol problems was associated with adolescent alcohol involvement. Yet this relationship was not as strong as the relationship with prenatal alcohol and was substantially lessened by adjustment for fetal alcohol exposure in statistical models. In our data there was no evidence that fetal alcohol exposure and family history of alcohol problems interacted in the prediction of adolescent alcohol use or related problems, nor did family history account for variance attributed to fetal exposure.

The multivariate PLS analyses that define the scores we have used for prenatal alcohol exposure, family history of alcohol problems and adolescent drinking and consequences are based on the premise that it is not possible to measure a single variable for each of these three domains that suffices for understanding the predictive relationships among them. We look instead for appropriately weighted combinations of the observable variables in each of these domains, with weights or saliences that reflect the importance of each variable in the cross-block predictions and that are derived from the simple cross-correlations illustrated in Figures 2 and 3. Consistent with previous analyses for prediction of other adolescent outcomes, the pattern of saliences on the prenatal alcohol measures (Table 3) points to the episodic drinking measures (particularly average drinks per occasion, ADOCC, and maximum drinks on an occasion, MAX) as most predictive of adolescent drinking and problems. These prenatal measures correlate substantially with all of the adolescent drinking and consequence measures, with the exception of the adolescent monthly occasions score, AMOCC (Figure 2; Table 3). We analyzed adolescent drinking rates, associated problems and a subset of dependence symptoms separately in order to evaluate possible specific prediction effects. Our data, however, suggested that fetal exposure is predictive of each of these adolescent outcomes at roughly equivalent salience.

The saliences of the family history indicators show that it does matter how family history is coded, as the paternal and first-degree relative indicators have substantially higher saliences for adolescent drinking and consequences than do the other family history indicators (Table 4). This result is consistent with other published studies that demonstrate specific statistical prediction of risk factors based on paternal and multigenerational alcoholism profiles (e.g., Harden and Pihl, 1995). Age at first intoxication is clearly the most salient adolescent outcome with respect to family history. However, we caution against overinterpretation of this univariate effect. The higher salience of AGEINTOX in Table 4 does not result in an appreciably different adolescent alcohol LV, as the scores derived from Tables 3 and 4 correlate r = .99. The singular association between age of first intoxication and paternal drinking could be explained by a variety of social mechanisms.

The implications of these results are manifold. First and foremost, for research, attention to fetal exposure to alcohol appears critical for the development of etiological risk models. As reviewed in the introduction, we know of no study that has controlled for maternal drinking during pregnancy in heritability studies of alcoholism. In fact, there is good reason to assume that assortative mating does occur with respect to alcohol consumption (Hall et al., 1983): Wives of alcoholic husbands drink more than women married to nonalcoholics. In family history studies, maternal alcoholism is sometimes assessed, typically in an attempt to study paternal vs maternal genetic influences. Yet fetal exposure poses a difficult assessment challenge for heritability studies. Recall that teratogenic effects are manifest statistically over a range of moderate ("social") rates of drinking, rates that would be missed by broadband assessment of maternal drinking as "problematic" or "alcoholic." Based on extant literature, and the analyses just presented, it appears that subtle effects of alcohol exposure during pregnancy should be accounted for in heritability studies.

The data also have implications for etiological models of alcohol involvement and alcohol-related problems in young adolescents. Our data, consistent with many studies of adolescents (see Baer et al., in press, for a review), suggest that 14-year-olds at higher psychosocial risk (as reflected in our data by covariates of parenting style and adolescent self-esteem) have greater involvement with alcohol and more negative consequences. Our regression analyses controlling for the psychosocial factors that we measured could account for family history effects but not for effects based on fetal exposure to alcohol. Thus, fetal alcohol exposure may have a specific or unique role in the prediction of adolescent drinking. While environmental factors clearly contribute to such risky behaviors of 14-year-olds, the implication of our research is that individual risk factors commonly attributed to genetic family history, including difficult temperaments, impulsivity, conduct disorder and executive functioning (cf. Zucker, 1994), could arise in part from exposure to alcohol during pregnancy.

The lack of a unique prediction based on family history was surprising. While our data suggest that fetal exposure is important for adolescent adjustment and can account for family history effects, our data do not suggest that family history of alcoholism contributes significantly in the prediction of adolescent outcomes beyond that which can be accounted for by fetal alcohol exposure and other covariates. Before concluding that there are no effects of family history, some limitations should be noted. It is possible that our dependent measures of drinking rates and problems are not sensitive to family history effects (although family history has been predictive of adolescent drinking using similar assessment methodology for a similar age group; Chassin et al., 1991). Further, our sample may not represent those at highest risk for familial alcohol problems. For a study of alcoholism etiology, our outcome measure, 14-year-olds' drinking, may be limited as well. We do not know much yet about the role of relatively early (age 14) adolescent drinking behaviors in predicting later, adult alcohol problems. In fact, some aspects of genetic risk, such as a reduced response to alcohol (Schuckit and Smith, 1996), may not manifest as risk factors until early or middle adulthood.

Yet some of the known linkages between adolescent and adult drinking problems tend to be associated with conduct problems and developing sociopathy (Zucker, 1994). It is quite possible that fetal exposure is a critical component in a developmental trajectory toward alcohol problems involving school failure, association with deviant peers, impulsive behavior and criminality (Carmichael Olson et al., 1997). Further, the role of fetal alcohol exposure in generating alcohol sensitivity, as has been shown in animal models, remains to be studied in humans.

Acknowledgments

We thank Heather Carmichael Olson, Ph.D., for directing the 14-year data collection and John Anzinger for technical assistance.

References

AMERICAN PSYCHIATRIC ASSOCIATION Diagnostic and Statistical Manual of Mental Disorders (DSM-IV), Washington, DC, 1994.

BABOR, T.F., HOFMANN, M., DELBOCA, F.K., HESSELBROCK, V., MEYER, R.E., DOLINSKY, Z.S. and ROUNSAVILLE, B. Types of alcoholics: I. Evidence for an empirically derived typology based on indicators of vulnerability and severity. Arch. Gen. Psychiat. 49: 599-608, 1992.

BAER, J.S. and CARNEY, M.M. Biases in the perceptions of the consequences of alcohol use among college students. J. Stud. Alcohol 54: 54-60, 1993.

BAER, J.S., MACLEAN, M.G. and MARLATT, G.A. Linking etiology and treatment for adolescent substance abuse: Towards a better match. In: JESSOR, R. (Ed.) New Perspectives on Adolescent Risk Behavior, New York: Cambridge Univ. Press, in press.

BATES, M.E. and PANDINA, R.J. Familial alcoholism and premorbid cognitive deficit: A failure to replicate subtype differences. J. Stud. Alcohol 53: 320-327, 1992.

BECKER, H.C., WEATHERSBY, R.T. and HALE, R.L. Prenatal ethanol exposure alters sensitivity to the effects of apomorphine given alone and in combination with ethanol on locomotor activity in adult male mouse offspring. Neurotoxicol. Teratol. 17: 57-64, 1995.

BOHMAN, g., SIGVARDSSON, S. and CLONINGER, C.R. Maternal inheritance of alcohol abuse: Cross-fostering analysis of adopted women. Arch. Gen. Psychiat. 38: 965-969, 1981.

BONTHIUS, D.J., BONTHIUS, N.E., NAPPER, R.M.A., ASTLEY, S.J., CLARREN, S.K. and WEST, J.R. Purkinje cell deficits in nonhuman primates following weekly exposure to ethanol during gestation. Teratology 53: 230-236, 1996.

BOOKSTEIN, F.L., SAMPSON, P.D., STREISSGUTH, A.P. and BARR, H.M. Exploiting redundant measurement of dose and developmental outcome: New methods from the behavioral teratology of alcohol. Devel. Psychol. 32: 404-415, 1996.

CAHALAN, D., CISIN, I.H. and CROSSLEY, H.M. American Drinking Practices: A National Study of Drinking Behavior and Attitudes. Rutgers Center of Alcohol Studies Monograph No. 6, New Brunswick, NJ, 1969.

CARMICHAEL OLSON, H., STREISSGUTH, A.P., SAMPSON, P.D., BARR, H.M., BOOKSTEIN, F.L. and THIEDE, K. Association of prenatal alcohol exposure with behavioral and learning problems in early adolescence. J. Amer. Acad. Child. Adolesc. Psychiat. 36:1187-1194, 1997.

CHASSIN, L., ROGOSCH, R. and BARRERA, M. Substance use and symptomatology among adolescent children of alcoholics. J. Abnorm. Psychol. 100: 449-463, 1991.

CLONINGER, C.R., BOHMAN, M. and SIGVARDSSON, S. Inheritance of alcohol abuse: Cross-fostering analysis of adopted men. Arch. Gen. Psychiat. 38: 861-868, 1981.

DAY, N.L. Research on the effects of prenatal alcohol exposure: A new direction. Amer. J. Publ. Hlth 85' 1614-1615, 1995.

DAY, N.L. and RICHARDSON, G.A. Prenatal alcohol exposure: A continuum of effects. Seminars Perinatol. 15: 271-279, 1991.

FULGINITI, S., ARTINIAN, J., CABRERA, R. and CONTRERAS, P. Response to an ethanol challenge dose on sleep time and blood alcohol level in Wistar rats prenatally exposed to ethanol during gestational day 8. Alcohol 6: 253-256, 1989.

GOLDSCHMIDT, L., RICHARDSON, G.A., STOFFER, D.S., GEVA, D. and DAY, N.L. Prenatal alcohol exposure and academic achievement at age six: A nonlinear fit. Alcsm Clin. Exp. Res. 20: 763-770, 1996.

GOODLETT, C.R. and WEST, J.R. Fetal alcohol effects: Rat model of alcohol exposure during the brain growth spurt. In' ZAGON, I.S. AND SLOTKIN, T.A. (Eds.) Maternal Substance Abuse and the Developing Nervous System, San Diego, CA: Academic Press, 1992, pp. 45-75.

GOODWIN, D.W., SCHULSINGER, F., HERMANSEN, L., GUZE, S.B. AND WINOKUR, G. Alcohol problems in adoptees raised apart from alcoholic biological parents. Arch. Gen. Psychiat. 28: 238-243, 1973.

GOODWIN, D.W., SCHULSINGER, F., KNOP, J., MEDNICK, S. and GUZE, S.B. Alcoholism and depression in adopted-out daughters of alcoholics. Arch. Gen. Psychiat. 34: 751-755, 1977.

GRUBE, J.W. and MORGAN, M. Smoking, drinking and other drug use among Dublin post-primary pupils, Dublin, Ireland: The Economic and Social Research Institute, 1986.

HALL, R.L., HESSELBROCK, V.M. and STABENAU, J.R. Familial distribution of alcohol use: I. Assortative mating in the parents of alcoholics. Behay. Genet. 13: 361-372, 1983.

HARDEN, e.w. and PIHL, R.O. Cognitive function, cardiovascular reactivity, and behavior in boys at high risk for alcoholism. J. Abnorm. Psychol. 104: 94-103, 1995.

HESSELBROCK, V., BAUER, L.O., HESSELBROCK, M.N. and GILLEN, R. Neuropsychological factors in individuals at high risk for alcoholism. Rec. Devel. Alcsm 9: 21-40, 1991.

HILL, S.Y., MUKA, D., STEINHAUER, S. and LOCKE, J. P300 amplitude decrements in children from families of alcoholic female probands. Biol. Psychiat. 38: 622-632, 1995.

JACOBSON, S.W., JACOBSON, J.L., SOKOL, R.J., MARTIER, S.S. and AGER, J.W. Prenatal alcohol exposure and infant information processing ability. Child Devel. 64: 1706-1721, 1993.

LARROQUE, B., KAMINSKI, M., DEHAENE, P., SUBTIL, D., DELFOSSE, M.J. AND QUERLEU, D. Moderate prenatal alcohol exposure and psychomotor development at preschool age. Amer. J. Publ. Hlth 85: 16541661, 1995.

McGuE, M., PICKENS, R.W. and SVIKIS, D.S. Sex and age effects on the inheritance of alcohol problems: A twin study. J. Abnorm. Psychol. 101: 3-17, 1992.

MANN, R.E. SOBELL, L.C., SOBELL, M.D. and PAVAN, D. Reliability of a family tree questionnaire for assessing family history of alcohol problems. Drug Alcohol Depend. 15: 61-67, 1985.

MULFORD, H.A. and MILLER, D.E. Drinking in Iowa: IV. Preoccupation with alcohol and definitions of alcohol, heavy drinking and trouble due to drinking. Q.J. Stud. Alcohol 21: 279-291, 1960.

PIHL, R.O. and BRUCE, K.R. Cognitive impairment in children of alcoholics. Alcohol Hlth Res. World 19: 142-147, 1995.

POLLOCK, V.E. and EARLEYWINE, M. Neuropsychological characteristics of individuals at risk for alcoholism: A recta-analysis. Univ. of Southem California, Los Angeles, unpublished manuscript, 1977.

ROSENBERG, M. Rosenberg Self-Esteem Scale, Princeton, NJ: Princeton Univ. Press, 1965.

SAMEROFF, A.J., THOMAS, S.L. and BARRETT, L.C. Anxiety and parental childrearing styles (abstract). Soc. Res. Adolesc. April, 1990.

SAMPSON, P.D., STREISSGUTH, A.P., BARR, H.M. and BOOKSTEIN, F.L. Neurobehavioral effects of prenatal alcohol: Part II. Partial least squares analysis. Neurotoxicol. Teratol. 11:477-491, 1989.

SCHUCKIT, M.A. and SMITH, T.L. An 8-year followup of 450 sons of alcoholic and control subjects. Arch Gen. Psychiat. 53; 202-210, 1996.

SEBER, G.A.F. Linear Regression Analysis, New York: John Wiley & Sons, Inc., 1977.

SHER, K.J. Children of Alcoholics: A Critical Appraisal of Theory and Research, Chicago: Univ. of Chicago Press, 1991.

STREISSGUTH, A.P., BOOKSTEIN, F.L. and BARR, H.M. A dose-response study of the enduring effects of prenatal alcohol exposure: Birth to 14-years. In: SPOHR, H.-L. and STEINHAUSEN, H.-C. (Eds.) Alcohol, Pregnancy and the Developing Child, New York: Cambridge Univ. Press, 1996, pp. 141-168.

STREISSGUTH, A.P., BOOKSTEIN, F.L., SAMPSON, P.D. and BARR, H.M. The Enduring Effects of Prenatal Alcohol Exposure on Child Development: Birth Through Seven Years, a Partial Least Squares Solution, Ann Arbor: Univ. of Michigan Press, 1993.

STREISSGUTH, A.P. and DEHAENE, P. Fetal alcohol syndrome in twins of alcoholic mothers: Concordance of diagnosis and IQ. Amer. J. Med. Genet. 47: 857-861, 1993.

STREISSGUTH, A.P., SAMPSON, P.D., OLSON, H.C., BOOKSTEIN, F.L., BARR, H.M., SCOTT, M., FELDMAN, J. and MIRSKY, A.F. Maternal drinking during pregnancy: Attention and short-term memory in 14-year-old offspring: A longitudinal prospective study. Alcsm Clin. Exp. Res. 18:202-218, 1994.

TAYLOR, A.N., BRANCH, B.J. and KOKKA, N. Neuroendocrine effects of fetal alcohol exposure. Progress Biochem. Pharmacol. 18' 99-110, 1981.

WHITE, H.R. and LABOUVIE, E.W. Towards the assessment of adolescent problem drinking. J. Stud. Alcohol, 50: 30-37, 1989.

ZUCKER, R.A. Pathways to alcohol problems and alcoholism: A developmental account of the evidence for multiple alcoholisms and for contextual contributions to risk. In: ZUCKER, R., BOYD, O. and HOWARD, J. (Eds.) The Development of Alcohol Problems: Exploring the Biopsychosocial Matrix of Risk. NIAAA Research Monograph No. 26, NIH Publication No. 94-3494, Rockville, MD: Department of Health and Human Services, 1994, pp. 255-289.

ZUCKER, R.A., FITZGERALD, H.E. and MOSES, H.D. Emergence of alcohol problems and the several alcoholisms: A developmental perspective on etiologic theory and life course trajectory. In: CICCHETTI, D. and COHEN, D.J. (Eds.) Developmental Psychopathology, Vol. 2, New York: John Wiley & Sons, Inc., 1995, pp. 677-711.

JOHN S. BAER, PH.D., HELEN M. BARR, M.A., M.S.,([dagger]) FRED L. BOOKSTEIN, PH.D.,([dagger]) PAUL D. SAMPSON, PH.D.,* and ANN P. STREISSGUTH, PH.D.([dagger])

Department of Psychology, University of Washington School of Arts and Sciences & Veterans Affairs Puget Sound Health Care System, Seattle, Washington

Received: October 7, 1997. Revision: December 1, 1997.

(*) This study was supported by National Institute on Alcohol Abuse and Alcoholism grant AA 01455-01-22 to Ann P. Streissguth.

([dagger]) Helen M. Barr and Ann P. Streissguth are with the Department of Psychiatry and Behavioral Sciences, University of Washington School of Medicine, Seattle, WA. Fred L. Bookstein is with the Institute of Gerontology, University of Michigan, Ann Arbor, MI. Paul D. Sampson is with the Department of Statistics, University of Washington School of Arts and Sciences, Seattle, WA.

Correspondence should be addressed to Ann P. Streissguth, Ph.D., Fetal Alcohol and Drug Unit, 180 Nickerson St., Suite 309, Seattle, WA 98109.