daily alcohol intake prior to pregnancy was 0.8 ounces, although levels of consumption varied from less than 0.1 ounce per day to more than 25 ounces (50 drinks) per day. The investigators reported dose-related alcohol effects on physical development and growth in infants examined at birth (Graham et al. 1988). Followup examinations demonstrated that 80 percent of the children who were identified at birth as having FAE were similarly classified at 4 years of age.

The Ottawa study addressed substance use during pregnancy and included women who smoked tobacco and marijuana in addition to drinking alcohol in moderation during pregnancy (Fried et al. 1980). This group of volunteers consisted predominantly of white middle-class women. The sample size was quite small. In general, the results of this study did not demonstrate an appreciable effect of moderate or "heavy social" drinking (the latter defined as more than 0.85 ounces of absolute alcohol per day) on growth parameters at 12 or 24 months of age (Fried and Watkinson 1988). Measures of dysmorphology were not assessed.

Investigators in the Atlanta study examined pregnancy outcome in a population consisting primarily of low-income African-American women. The sample was divided into three groups: women who consumed alcohol throughout their pregnancy (continued-to-drink group); women who stopped drinking following intervention sometime in the second trimester; and a control group of women who completely abstained from alcohol during their pregnancies. This design enabled the researchers to compare outcomes in children exposed to alcohol throughout the pregnancy with outcomes in both those who were exposed only during the first two trimesters and those who were not prenatally exposed to alcohol at all. The findings indicated that children of mothers in the continued-to-drink group (average daily consumption of 1.8 ounces absolute alcohol) had retarded growth and alcohol-related birth anomalies (Coles et al. 1991; Smith, Coles et al. 1986). Alcohol-related effects were not as severe in the children whose mothers stopped drinking as in children whose mothers drank throughout pregnancy.

Various factors that predicted continued drinking in women suggest that the continuedto-drink group may have experienced more chronic and severe alcohol-related problems than the group that discontinued drinking during pregnancy. Thus, the differences in period of

exposure between the two groups may not be the only factor that negatively influenced pregnancy outcome in the continued-to-drink group.

The Buffalo study, originally designed to test methods of screening for alcohol-related problems among obstetric and gynecologic patients, included assessments of pregnancy outcome at birth (Russell and Skinner 1988) and at 6 years of age (Russell et al. 1991). Maternal self-reports of drinking prior to pregnancy recognition predicted spontaneous abortion and decreased Apgar scores at birth. Indications of problem drinking in mothers predicted retarded growth (particularly small head circumference), low birth weight, and decreased Apgar scores at birth (Russell and Skinner 1988). The 6-year followup study revealed that heavier alcohol intake prior to pregnancy recognition was associated with growth deficits, increased dysmorphology (in particular, facial features that are associated with FAS), and a higher prevalence of possible FAE.

In the Cleveland study, women were asked about their drinking during pregnancy prospectively while they were pregnant and retrospectively in a followup interview 5 years postpartum. Amounts reported during the followup interview were often larger than those reported in the antenatal interviews, thus suggesting that use during pregnancy is underreported. Average alcohol use per day reported during pregnancy was 0.07 ounces. Ernhart et al. (1985, 1989) found a positive relationship between the amount of prenatal alcohol exposure and a tally of neonatal anomalies, including the craniofacial anomalies associated with FAS (Ernhart et al. 1989). This research group has extensively investigated threshold levels, in particular levels related to dysmorphology. Although it has been difficult to firmly establish a critical threshold level for alcoholinduced injuries, results from this study indicate that heavy drinking (more than an average of 1.5 ounces of absolute alcohol per day, or about three drinks per day) prior to recognition of pregnancy clearly places the fetus at increased risk for birth defects, including the craniofacial anomalies that, as a group, characterize FAS.

Women in the Pittsburgh study were interviewed during their fourth and seventh months of pregnancy and at delivery. Those selected for the study reflected the entire spectrum of drinking during pregnancy. The results demonstrated an association between prenatal alcohol exposure and increased incidence of minor physical anomalies and retarded growth (Day et al. 1989). Drinking during the first and second months of

pregnancy was associated with increased risk of low birth weight in offspring. The investigators noted that the average amount of alcohol consumed daily by a mother during pregnancy predicted growth deficits in offspring at the 18month followup (Day, Goldschmidt et al. 1991). Alcohol-related physical anomalies and growth retardation were found to persist at age 3 in this cohort (Day, Robles et al. 1991).

Neurobehavioral Effects

The Atlanta study, with a sample of children who were exposed prenatally to rather high levels of alcohol (e.g., mothers in the continuedto-drink group consumed an average of 1.8 ounces per day), reported neurobehavioral deficits in infants examined 3 days after birth (Coles et al. 1985). As measured on the Brazelton Neonatal Behavioral Assessment Scale (BNBAS), infants exposed prenatally to alcohol were more active physically, were more likely to show signs of autonomic or central nervous system instability (e.g., tremors or asymmetric reflexes), and had less mature motor behavior compared with unexposed infants. Several alcohol-related neurobehavioral effects were less severe in infants whose mothers stopped drinking during the second trimester than in those whose mothers drank throughout their pregnancies. Alcoholrelated neurobehavioral effects were still present 14 and 30 days after birth (Coles et al. 1987).

In the 6-year followup, children of mothers in the continued-to-drink group had impaired intellectual functioning, including deficits in shortterm memory and overall mental processing, compared with children of the two other groups. Both groups of alcohol-exposed children displayed significant deficits in math and reading skills (Coles et al. 1991).

The Seattle study reported neurobehavioral deficits at birth, as measured by the BNBAS, at significantly lower levels of prenatal alcohol exposure (an average of 0.8 ounces of absolute alcohol consumed per day) than reported in the Atlanta study (Streissguth et al. 1983). Infants exposed prenatally to alcohol had a diminished ability to habituate to aversive stimuli (Streissguth et al. 1983). Effects of maternal drinking on cognitive and neurobehavioral measures persisted throughout development (Barr et al. 1990; Streissguth et al. 1989). At 7 years of age, children of heavier drinking mothers had cognitive deficits (as measured by reduced IQ scores) along with problems in short-term memory,

arithmetic, and spatial organization (Sampson et al. 1989; Streissguth et al. 1989, 1990). Streissguth et al. (1990) also reported that of those mothers who reported at least one occasion of drinking five or more drinks, 24 percent had children who presented some type of learning problem. In the 6-year followup of the Buffalo study, investigators found that children born to mothers with more than one indicator of problem drinking had an increased likelihood of cognitive deficits, defined by lower scores on verbal IQ and receptive language function tests (Russell et al. 1991).

Neurobehavioral deficits, as measured at birth by the BNBAS, were not found in the Cleveland (Ernhart et al. 1985) or Pittsburgh (Richardson et al. 1989) studies. However, the Pittsburgh investigators reported relationships between alcohol exposure during pregnancy and abnormalities of neurophysiological status. Prenatal alcohol exposure affected infants' sleep state cycling and body movement during sleep (Scher et al. 1988). The Cleveland investigators found no significant relationship between moderate prenatal alcohol exposure and language indices in a 3-year followup (Greene et al. 1990).

Attention Measures

Investigators in the Seattle study reported that children of heavier drinking mothers exhibited impulsivity and attention decrements. These children were more easily distracted than children exposed prenatally to lower levels of alcohol, and they performed worse on vigilance tests at 4 and 7 years of age (Streissguth et al. 1984; Streissguth, Barr et al. 1986). In the Atlanta study, although hyperactivity and impulsivity were not evident, children exposed to alcohol throughout pregnancy showed deficits in laboratory tests of sustained attention. In addition, teachers often reported that these children had attention and behavior problems (Brown et al. 1991).

Although some consistent results emerge from these studies, research is only beginning to gauge the impact of low-level maternal drinking on children's physical and cognitive development. In particular, among the problems that prospective studies have had to address is that posed by the need to rely on self-reports of maternal drinking levels. Because women may deny or forget to report drinking episodes during pregnancy, underestimation of prenatal alcohol exposure is a potential problem (Day and Robles 1989; Ernhardt et al. 1988; Morrow

Tlucak et al. 1989; Russell 1988). Underestimation may result in heavy drinkers being classified with moderate drinkers, thereby producing a situation in which deficits that are truly related to heavy drinking are mistakenly attributed to moderate consumption (Russell 1991). In addition, underreporting can hinder efforts to determine precise threshold or dose-response measures for various alcohol-induced effects in the developing fetus.

The long-term consequences of prenatal alcohol exposure demonstrated in several of the studies justify continued attention and rigorous examination of individuals exposed to alcohol in utero. The importance of such findings gains particular significance in light of the large number of women of childbearing age who drink.

animal research has shown that different profiles of alcohol-related birth defects are related to differences in the periods of alcohol exposure (i.e., critical periods) during fetal development.

Critical Periods, Threshold
Doses, and Drinking
Patterns

Unfortunately, such questions as how much alco-
hol is too much and when the fetus is at greatest
risk have proven difficult to address in human
studies. These limitations in our knowledge can
produce much anxiety among women who con-
sume alcohol prior to recognition of pregnancy.
In a recent study, results from three separate in-
terview instruments used to assess drinking dur-
ing pregnancy (a measure that investigators
presume correlates with drinking prior to recog-
nition) were found to support an inferred con-
servative threshold of 1.5 ounces of absolute
alcohol per day (an average of about three
drinks per day during the period prior to knowl-
edge of pregnancy) for FAS-related neonatal
physical anomalies (Ernhart et al. 1989). How-
ever, for a number of these mothers, this average
number of drinks resulted from weekend drink-
ing. Therefore, many of the fetuses whose moth-
ers averaged three or more drinks per day may
have been exposed to much heavier doses on in-
dividual occasions. Research on laboratory ani-
mals suggests that binge-like exposure to alcohol

may be more harmful than exposure to the same or larger amounts of alcohol spread out more evenly over time (Bonthius and West 1990).

The use of animal models in studies probing outcomes of alcohol use in pregnancy allows for controlled analysis of threshold doses, critical periods of drinking, and drinking patterns. Indeed, animal research has shown that different profiles of alcohol-related birth defects are related to differences in the periods of alcohol exposure (i.e., critical periods) during fetal development (Randall 1987).

It is likely that threshold doses will be difficult to define fully. Because an enormous range of defects can result from prenatal alcohol exposure, it is not unreasonable to assume that different aspects of development will be found to be sensitive to different levels of exposure. As expressed by Clarren et al. (1987, p. 345), "It is probable that there is no single dose-response relationship for ethanol teratogenesis, but rather that each abnormal outcome in brain structure or function, morphology, and growth has its own dose-response and gestational timing parameters."

Prevention of Alcohol-
Related Birth Defects

Alcohol-related neonatal injuries are completely preventable with changes in the drinking behavior of pregnant women. To date, researchers have not established a clear threshold of daily drinking for alcohol-related impairments. For this reason, women have been advised that abstaining from alcohol during pregnancy is the only known way to prevent alcohol-related injuries in offspring (Surgeon General's advisory 1981).

Findings from the 1985 National Health Interview Promotion and Disease Prevention survey underscore the need to better inform the public about the risks of drinking during pregnancy (Fox et al. 1987). Of the more than 18,000 men and women interviewed in this survey, only 55 percent had heard of FAS. Among those who had heard of the syndrome, less than 25 percent identified FAS as a pattern of birth defects. A higher percentage of light and moderate drinkers than of heavy drinkers were aware of the syndrome.

Studies suggest that various community education programs, including those that target health care providers, pregnant women, women of childbearing years, and the wider community,

are effective in modifying drinking behavior, particularly among social drinkers. In 1974 to 1975, also in 1980 to 1981, Streissguth et al. (1984) interviewed women about drinking and smoking habits. FAS had just been identified at the time of the first interviews, and consequently there were no published reports or papers explaining the effects of maternal drinking on offspring. In contrast, by 1981, many reports addressing the dangers of prenatal drinking had been published. In addition, 1981 marked the issuance of the Surgeon General's warning about the harm that daily drinking by a mother could cause to a fetus. Streissguth et al. noted a marked decline in alcohol use by pregnant women involved in the second interviews compared with participants in the first interviews. The most significant decline in drinking was observed in highly educated and older pregnant women. Smith, Lancaster, and Falek (1986) observed similar results over a 5-year period with a program undertaken in a low-income, predominantly African-American population.

The Indian Health Service conducted a comprehensive macrolevel FAS prevention program for Native Americans and Alaska Natives (May and Hymbaugh 1989). This prevention study was designed to provide native communities in the United States with the information and training they needed to initiate their own prevention measures. Trainers were instructed in all levels of prevention. The results of the study indicated that local trainers successfully disseminated FAS information to various community groups, including pregnant women and school children. In addition, the groups retained over time the knowledge furnished by trainers.

Women at greatest risk may require more intensive prevention approaches than those offered by community education programs. To reach these women, prevention programs should incorporate intervention strategies (Smith and Coles 1991). Identifying women whose drinking places them at heightened risk for having alcohol-affected children is key to prevention efforts. If women at increased risk can be identified, intervention may be possible in the health care setting. Indeed, studies have shown that therapeutic interventions in the prenatal clinic setting can be instrumental in promoting abstinence, even in high-risk drinkers (Larsson 1983; Rosett and Weiner 1984).

However, various factors can complicate efforts to identify high-risk women. For example, when asked about their drinking, pregnant

women may underreport their consumption levels (Emhardt et al. 1988; Morrow-Tlucak et al. 1989) or may not consider their alcohol intake prior to recognition of pregnancy (i.e., drinking during the first trimester) (Day and Robles 1989). Indirect measures of drinking during pregnancy can be useful. Russell and Skinner (1988) and Russell et al. (1991) suggested that indices of drinking prior to pregnancy might serve as useful predictors of pregnant women at high risk because these indices were found to be predictive of adverse pregnancy outcome. Smith et al. (1987) reported that the best predictors of the high-risk behavior of continued drinking in pregnancy were length of drinking history, reported tolerance for alcohol, a history of alcohol-related illness, and a preference for drinking with other family members.

Studies suggest that various community education programs, including those that target health care providers, pregnant women, women of childbearing years, and the wider community, are effective in modifying drinking behavior, particularly among social drinkers.

Self-administered questionnaires about drinking habits are effective tools for collecting data about maternal alcohol consumption (Russell and Bigler 1979) and can facilitate the identification of women whose drinking places them at risk for adverse pregnancy outcomes. Questionnaires have been devised to be brief and nonthreatening (Russell et al. 1991; Sokol et al. 1989). The T-ACE and the TWEAK are two instruments that have been developed to screen for alcohol use among pregnant women. The T-ACE questionnaire, developed by Sokol et al. (1989), incorporates the C, A, and E items of the CAGE (see Chapter 13, Screening and Brief Intervention) and replaces the G item with a question that addresses alcohol tolerance (T). Sokol et al. considered a woman to be alcohol-tolerant if she requires more than two drinks to feel the effects of alcohol. The acronym TWEAK is derived from the questions pertaining to alcohol Tolerance, others Worried about drinking, Eye-opener or morning drinking, Amnesia or blackouts, and the need to (K) cut down. Preliminary results indicate that these tools may have better predictive

value for identifying risky drinking behavior in pregnant women than other screening instruments, such as the CAGE and Michigan Alcoholism Screening Test (see chapter 13) (Russell et al. 1991). The questions on tolerance in the T-ACE and the TWEAK are key because many persons being screened are less likely to suspect that tolerance is an indication of heavy drinking. Thus, questions on tolerance of alcohol are less likely to trigger denial about drinking levels. These and related screening tools should prove to be useful aids for identifying high-risk drinking among pregnant women.

Characteristics and

Benefits of Animal Models

There is a remarkable similarity and correspondence of findings in animal models with findings in the clinical setting (Driscoll et al. 1990). This correspondence applies to both physical anomalies and behavioral teratogenic effects (table 2). A mouse model is ideal, because mice are sensitive to the full spectrum of prenatal alcohol effects, ranging from dysmorphology to cognitivebehavioral deficits in the absence of physical abnormalities, and therefore may best model the clinical condition (Becker and Randall 1989; Becker et al. 1989). Moreover, most of the deleterious effects of prenatal alcohol exposure

observed in animals have been identified after levels of maternal alcohol exposure (blood alcohol levels) that approximate typical levels of exposure reported in pregnant alcohol-consuming women (Driscoll et al. 1990).

Another advantage of using animal models is that the combined effects of gestational exposure to multiple drugs may be examined under more controlled circumstances. Although both licit and illicit drug use or abuse may occur in women who drink during pregnancy, other drug use is typically adjusted for statistically in human studies in an attempt to isolate an association between a particular pregnancy outcome measure and prenatal alcohol exposure. A limited number of experimental studies have found that drugs such as nicotine (Leichter 1989), marijuana (Abel and Dintcheff 1986), and cocaine (Church et al. 1991) magnify the negative consequences associated with maternal alcohol consumption during pregnancy in animals. Given the prevalence of multiple drug use in society, additional attention to this subject is warranted.

Sensorimotor Deficits

Various sensorimotor deficits identified in children exposed prenatally to alcohol have been demonstrated and examined in greater detail in animal studies. These deficits include visual, auditory, vestibular, and motor coordination