(e.g., poor nutrition, parental stress and psychiatric illness, sexually transmitted diseases) that may also impact on development (Frank et al., 1993; Finnegan, 1994). Further, the majority of women who use heroin, marijuana, or crack cocaine also use varying amounts of alcohol and/or nicotine and may use one or more illicit drugs in combination. Thus, rarely is it possible to speak, for example, of a "pure" crack cocaine effect.
Additionally, longitudinal studies of the developmental outcome of prenatal drug exposure in human infants face four methodological issues that cut across the specific agent of abuse and exposure (reviews by Jacobson and Jacobson, 1995; Neuspiel, 1995; Olson et al., 1995). First, there are difficulties in ascertaining the amount, frequency, and duration of drug abuse during pregnancy due to inaccuracies of maternal selfreport and limitations of current biological markers of exposure (Coles, 1992; Kidwell, 1992). Second, the high rate of attrition is a problem in studies of drug-abusing populations (Mayes and Cicchetti, 1995). Third, there are difficulties in choosing the appropriate comparison group (e.g., determining whether the comparison group is drug free or free only of the primary drug of interest, choosing appropriate demographic comparison cohorts). Fourth, determining the appropriate time (developmentally) and length of time to assess infants is another crucial issue. Traditional models of behavioral teratology presume effects that are present at least early in infancy but may or may not persist through childhood. Less frequently discussed are drug-related effects that are not apparent until later in development, when central nervous system processing of information or social skills are required, or during periods of major central nervous system reorganization (e.g, between age 4 and 5 years or during puberty) (see Weiss, 1995).
Animal models provide some basis for comparison because the amount of exposure and environmental conditions may be controlled. Animal models have been particularly useful for studies of the effects of prenatal exposures and for modeling drug-related effects on brain development at the structural, cellular, and functional levels. Neurobehavioral data from animal models should be viewed carefully, however, when extrapolating results from animal models to the complex developmental capacities found in higher primates and humans (e.g., language, complex problem-solving tasks, and neuropsychological functions such as certain domains of memory) (see Stanton and Spear, 1990).
Nicotine acts as a vasoconstrictor, reducing placental blood flow and