Symposium
Chairs: Fonda Davis Eyler and Marylou Behnke
Discussant: Ronald Seifer
The teratogenic model provided the conceptual framework for most of the early research onprenatal cocaine exposure, wherein the issues of timing, frequency, and amount of exposure havebeen explored. In general, front-end design issues have been emphasized including attempting toenroll subjects based on accurate identification of drug exposure. Investigators have struggled tofind the most reliable techniques of identifying exposure patterns, including use of variousbiologic specimens and/or types of self-report measures. Methods of analyzing the variouspatterns of exposure have also been problematic. Because identification of illegal drug use isfraught with uncertainties, results of studies have been accepted with skepticism.The transactional model of development includes not only perinatal risk factors affecting thenewborn infant, but variables in the care-giving environment that change and are changed byinteraction with the developing child over time. Many researchers have attempted to use thisconceptual model, but few have integrated it into prospective, longitudinal research in this verydifficult area of prenatal drug exposure. We will argue that both the teratogenic and transactional models are helpful to research in thisarea and that both models require good study design, including first the selection of arepresentative sample, and unbiased enrollment and maintenance of subjects. In addition, it isimportant to minimize the sheer number of potentially confounding variables in any one study,by a priori exclusion and selection criteria, matching and/or statistical controls.However, part of the art in this area of research is to retain a sample that is representative of mostusers, not only to more widely generalize findings, but to maintain those study variables that ourliterature and developmental theories would suggest are critical elements of developmentalprogress for any child. That is, we need not to make nuisance variables of classic predictors ofdevelopmental outcome. Therefore, good design, including elements of the teratogenic modelare necessary but not sufficient to answer the questions of relative effects of prenatal cocaineexposure on developmental outcome. >From our research-based theories of development, we should be identifying key variables toinclude in our conceptual model and study design, not to eliminate, but to study, in the context ofprenatal drug exposure. Furthermore, our ultimate answers will come from transactional effectsacross the course of development. Some of these key study variables might include care-giverattitudes, behaviors, knowledge and skills; changing and multiple care-givers/out-of-homeplacements; family functioning; home environment; neglect, abuse; and experience with violencein home or community. Each presenter will provide examples from their prospective studies to illustrate challenges theyhave had and solutions they have developed in attempting to meld the best (or most helpfulaspects) of the teratogenic and transactional models. Data they have collected as part of theirlongitudinal interactive models, analyses, and interpretations of findings will be used to illustratethese issues. The discussant will be asked to comment on the state of the art with respect to themodels, study design and findings to date and make suggestions for future research.
Details of individual items:
paper
The theoretical framework for many prenatal cocaine exposure studies has been rooted in thebasic concepts of teratology/developmental toxicology described by Vorhees and others whichinclude the effect of amount of drug reaching the fetus and the timing of the drug exposurerelative to the developmental stage of the fetus. While these principles are important inunraveling the effects of prenatal cocaine exposure, at this time, it appears that cocaine is only amild teratogen at best. This realization has led researchers to investigate more fully not onlycocaine's teratogenic potential but its complex interplay with postnatal risk factors. Thisthinking has evolved into the recognition of the importance of the transactional model ofdevelopment originally proposed by Sameroff and Chandler. This model recognizes not only theinfluence of static prenatal and postnatal factors but also more importantly the change that occursbased upon interaction of these factors with the developing child over time. Integration of thesetwo models requires the identification of which variables in the prenatal and postnatalenvironment are important to pursue, development of appropriate measurement tools, the abilityto quantify outcomes, and then a distillation of the data and analysis of effects transactionallyover time. In our prospective, longitudinal study of prenatal cocaine exposure we have been challenged withthe integration of these two models. This presentation will focus on one major issue developedfrom each line of reasoning and how we have attempted to resolve the problems surroundingeach: (1) the effect on developmental outcome of amount, timing, and pattern of drug use duringpregnancy and (2) the influence of out-of-home placement of the child on developmentaloutcome at 6 months. We will include such information as the measurement tools we have usedfor these outcomes, how the data were quantified and reduced for analysis, and the identificationof important confounding variables. Simple analyses of the relationship between these variablesand 6-month Bayley scores have shown our cocaine-exposed infants have lower PDI but notMDI scores. A more complex determination of amount of exposure showed that infants ofmothers with heavier prenatal use had lower MDI and PDI scores. Additionally, out-of-homeplacement, greater prenatal and neonatal risk, and abnormal performance on the Brazelton at 1month (lower motor and orientation cluster scores) were related to significantly lower MDI andPDI scores. And poorer scores on the HOME evaluation at 1 month were related to lower MDIbut not PDI scores. However, these variables are not independent of each other and thus must beanalyzed using a multivariate, transactional model to fully understand their complex interactionsand effects on development. For example, our data have indicated an effect of exposure on thehome environment that on closer inspection appeared to be related to the biologic mother/out ofhome placement variable. Transactional analyses will be presented using the model shown in thefollowing figure.
paper
The long-term effects of cocaine exposure during pregnancy and the effects of cocaine use on thechild's environment are major public health concerns. However, very little information existsregarding these effects. In our program of research, we use the teratologic model as ourconceptual framework. The basic principles of this model can be summarized as follows(Vorhees, 1989): (1) Outcomes are expressed cumulatively as central nervous system (CNS)effects, growth deficits, and malformations as the dose increases. At low doses, effects will befound on the CNS in the absence of morphologic changes. (2) Effects are stage-specific. Theeffect of a teratogen will be related to the stage of fetal development at the time of exposure. (3) Some effects on the developing CNS will be manifested only at a later stage of development,often referred to as latent or 'sleeper' effects. (4) The observed outcomes will reflect the effectof both the teratogenic exposure and the environment in which the organism is raised.Embedded in this model are two definitions for the action of a teratogen, a direct effects modeland a vulnerability model. By contrast to the direct effects model, the vulnerability modelhypothesizes that the child is made vulnerable by the prenatal exposure, and the interactionbetween the vulnerability and the environment predicts the outcome. The environmentmoderates the phenotypic expression of the teratogen. A third model must also be considered.This is the possibility that prenatal cocaine exposure is not a teratogen, but rather theenvironment in which the child is raised mediates the relationship between exposure andoutcome. In this model, the environment is the important factor in determining outcome.Currently, there are insufficient data in the literature to judge which model is most applicable tothe long-term effects of prenatal cocaine exposure.We will present data from our longitudinal study of prenatal cocaine exposure. We conducteddetailed interviews during pregnancy using standardized instruments and data collection points,measured the quantity, frequency, and pattern of cocaine, alcohol, marijuana, tobacco, and otherillicit drug use for each trimester and the postpartum period, and assessed other risk factors, suchas sociodemographic factors, life events, social support, psychological characteristics, andenvironmental factors. Women were interviewed at the end of their 1st, 2nd, and 3rd trimesters,and at 1, 3, and 7 years postpartum. Offspring were evaluated at delivery, 1, 3, and 7 years byexaminers who were blind to exposure status. We will present several examples from our data of both direct and indirect effects of prenatalcocaine exposure. At each phase of our follow-up, we have found direct effects of prenatalcocaine exposure on the development of the CNS and on growth. Cocaine use also affects theenvironment in which the child is raised. For example, current cocaine users provided lessstimulating and organized home environments than non-users. Thus, the children are exposed tothe direct effects of cocaine prenatally and to the indirect effects of cocaine in their postnatalenvironment. We will discuss the importance of carefully considering which model bestdescribes the data.
paper
The study of human development is daunting, due in part to the complexity of dynamicallychanging interactive effects over time. In our longitudinal study of the effects of prenatal cocaineand other drug exposure on developmental outcome we have derived a model based on boththeoretical and empirical considerations. It suggests that difficulties with early emotionalregulation and coping, as well as environmental factors, mediate the effect of prenatal cocaineexposure on outcomes. We expect that prenatal cocaine and other drug exposures have a directeffect on later outcomes as a result of permanent structural and functional changes in the brain.We also expect indirect effects due to the impact of drug exposure on the reactivity andregulatory abilities of the child, as well as on environmental risk including the caregiver-childrelationship. These in turn should affect later outcomes. We are studying multiple measures over time of arousal, emotional regulation, and coping withstress due to our belief that these are functions specifically affected by cocaine exposure. We alsohave collected multiple measures of the child's social environment. These include distalvariables, such as the amount of stress and social support in the mother's life, and proximalvariables, including direct measures of maternal-child interaction, and the caregivingenvironment.The following study of impulse control when the subjects were 24 months of age illustrates thecomplex interplay between organism and environmental factors. The children were observed at atable upon which the examiner placed a cookie. They were told not to take the cookie and weregiven a toy to distract them. Mothers remained in the room and were instructed to watch theirchildren but not to interfere. Despite these instructions mothers varied in how much theyinteracted with their children. The latency to pick up the cookie was recorded. Hierarchicalmultiple regressions were used to examine the relative impact on the toddler's impulse control ofa set of predictors including prenatal cocaine exposure, amount of maternal distraction, neonatalmedical complications, environmental risk, and amounts of prenatal alcohol and cigaretteexposure. Prenatal cocaine exposure showed a significant relation to the latency to pick up thecookie. Exposed infants had shorter latencies. The amount of maternal distraction also affectedlatency to take the cookie. The greater the distraction the longer the latency to take the cookie.Of particular interest is that cocaine was related to both maternal and infant behavior resultingin opposite effects on the infant's impulse control. Mothers who used cocaine tended to distracttheir infants more, which had the effect of increasing the latency to take the cookie. However,cocaine exposure was directly related to decreased latency. Of note is that the failure to examineboth variables would have misrepresented the total impact of cocaine on impulse control. Thesedata are a reminder that the determinants of behavior are numerous and there can be suppressoreffects due to complex relations among the variables.
paper
The Maternal Lifestyle Study (MLS) is a large, multicenter, prospective, longitudinal study oftwo well defined cohorts of cocaine/opiate exposed and non-cocaine, non-opiate exposed infants,their mothers, their families, and their environment. The major strengths of this study are itslarge sample size, the multi-site collaboration providing access to varied sociodemographics, thevalidated definition of exposure and the training/certification of the research team in theadministration of the interviews, the exams and the developmental assessments at each site. Inspite of these major advantages over previous single site, small sample size studies, many of thecomplex confounders persist, but, due to the very large sample size, controlling for multiplefactors was possible with adequate cell size for detecting reliable associations. The use ofdichotomous as well as continuous exposure analytic techniques, both cross sectionally andlongitudinally over the first several years of life provided a characterization of this populationthat had not been previously possible. The overall prevalence of medical conditions correlatedwith drug use/abuse, such as sexually transmitted diseases, utilization of prenatal care, and othersuch factors was remarkably low in this population, but continued to exhibit the expecteddifferences between the exposed/non-exposed cohorts. However, the previously reportedteratogenic potential of these drugs was not confirmed.Of greatest interest was the impact of sociodemographic/environmental issues which were moreclosely correlated with both mother and infant outcomes, than was the actual exposure to drugs. Infant outcome reflected maternal issues, again showing low prevalence, but statisticallydifferent complications in the two cohorts in the expected direction, but not confirming a directeffort of cocaine/opiates on either acute medical outcomes or developmental accomplishments. With over 8000 mother/infant dyads available for acute outcome analyses and approximately1400 dyads participating in a highly successful follow-up effort, it was possible to control formany of the typical confounding variables that plague drug abuse research.Although this study is continuing its follow-up into the school age years whensociodemographics have traditionally become more prominent influences in outcome, this studyhas been able to demonstrate that these issues are more influential in determining even acuteoutcome than the toxic/teratogenic potential of the drug exposure itself. However, it cannot beunderstated that although specific complications are of overall low prevalence, exposure to drugsin-utero is indeed associated with a multifold increase in many major adverse health outcomes,some having chronic or even life threatening potential (ie. Hepatitis, Herpes, AIDS exposures),and cannot be under-emphasized. Yet it is variables like environment, maternal education,exposure to violence and crime, placement in foster care and for adoption, that may mostsignificantly impact the long-term multidimensional outcome of these infants and determine theirability to function effectively in society.