Objective. This prospective, longitudinal project was designed to determine the effects of prenatal cocaine use on the neurodevelopmental outcomes of infants from a historically understudied rural public health population.
Methodology. We interviewed >2500 women prenatally, identified 154 cocaine users, and matched 154 controls on race, parity, socioeconomic status, and location of prenatal care (that related to level of pregnancy risk). Drug testing was required at enrollment and at delivery; detailed demographic, psychosocial, and drug histories were taken at each available trimester and follow-up visit. After birth, certified evaluators, blinded to maternal history of drug use, administered the Brazelton Neonatal Behavioral Assessment Scale (BNBAS) under controlled conditions. Evaluation time for preterm infants was adjusted for gestational age.
Results. After controlling for the effects of marijuana (users of other illicit drugs were excluded), alcohol, and tobacco use, the following results remained. There were significant drug group interactions on the BNBAS Qualifier Score of Alert Responsiveness that demonstrated lower scores among infants who were exposed to both cocaine and tobacco and among those exposed to both marijuana and tobacco. There were also significant correlations between the amount of cigarette, alcohol, marijuana, and cocaine exposure and several BNBAS scores. The reported amount of cocaine use in the third trimester was negatively related to scores of Orientation, Cost of Attention, and Alert Responsiveness (that was also related to amount of cocaine used over the entire pregnancy). When the effects of marijuana, alcohol, and tobacco were partialled out, the amount of cocaine use in the third trimester was negatively related to Regulation of State, a precursor of alertness, and the infant's ability to orient to the environment.
Conclusions. The observed decrement in state regulation, attention, and responsiveness among cocaine-exposed neonates raises concerns about later developmental abilities as well as the effect these infants may have on caregivers (who themselves may be compromised in their parenting abilities by their drug use). Follow-up of these infants will reveal if these disadvantages continue. These early results also emphasize the importance of considering amount and time of drug exposure as well as the interactive effects of drug exposure and other risk variables.
Concern was raised about the consequences of prenatal cocaine exposure when early studies1-3 reported neurobehavioral effects using the Brazelton Neonatal Behavioral Assessment Scale (BNBAS).4 These included poorer performance on Orientation, Motor Maturity, and State Regulation. Yet these studies were limited to samples of urban women who were heavy users of other drugs as well as cocaine. Also, most women were enrolled from programs providing prenatal care and drug treatment, making blinded evaluations more difficult.
Recent prospective, blinded studies have also examined the neurobehavioral effects of prenatal cocaine exposure using the BNBAS.5-11 Black et al5 reported that cocaine-exposed infants scored higher on the Depressive summary score and did more poorly than controls on cluster scores of Orientation, State Regulation, and Autonomic Regulation on the BNBAS. However, by 2 weeks, exposed neonates were poorer than controls only on Autonomic Regulation. Eisen et al7 and Mayes et al8 found cocaine-exposed neonates did significantly less well on the BNBAS Habituation Cluster. Coles et al6 reported no BNBAS differences until 2 to 4 weeks, most of the lower scores were in the clinically normal range, and the most negative responses were not found in the group with the most drug exposure. Neuspiel et al9also found no drug group differences at birth; at 11 to 30 days, the cocaine-exposed neonates did more poorly only on the Motor Cluster, a difference reduced when controlling for confounding variables. When Richardson and Day10 controlled for confounding variables, they demonstrated no differences on BNBAS between cocaine-exposed and control infants. Our group11 also found no differences in BNBAS examinations between cocaine-exposed infants and matched controls both at birth and at 1 month.
To date, the majority of research has examined the effect of cocaine exposure by comparing the outcome of exposed to nonexposed infants. Very few researchers have systematically studied the relationship of timing or amount of prenatal cocaine use on newborn behavior. Corwin et al12 found that newborns who had positive urine tests for cocaine (indicating recent use) were worse in measures of acoustical cry characteristics. Chasnoff et al3 compared women who used cocaine throughout pregnancy, only during the first trimester, and not at all. First trimester users, although similar to drug-free women in measures of intrauterine growth, had infants with more neurobehavioral problems, similar to those of full pregnancy users. It is difficult to draw conclusions about the effects of amount versus timing of use that are inherently confounded in these and most human studies.
This article describes the newborn behavioral assessment portion of the longitudinal study. Our primary null hypothesis was that prenatally cocaine-exposed neonates would not differ from matched controls on measures from the BNBAS. We also hypothesized that any infant behavioral effects observed would be related to reported amounts of cocaine used during each trimester of pregnancy.
A full description of the study design including power analysis, informed consent procedures, subject selection, and identification and documentation of drug use is included in Part I. Users of illicit drugs other than cocaine and marijuana were excluded; subjects were matched on race, SES, parity, and location of prenatal care, that related to pregnancy risk.
Neonatal Assessment Procedures
We administered the BNBAS4 which is commonly used in neonatal research to assess reflexes, motor maturity, autonomic regulation, the range and regulation of wake-sleep states, habituation to repeated stimuli, and the infant's alertness and orientation to visual and auditory stimuli. In the 1984 revised edition of the BNBAS, in addition to refined measurements of the 28 elicited and observed behaviors, nine supplementary scores were added to qualify the behavioral responses by evaluating the effort and physiologic cost of an infant's performance.
Before they began this research project, BNBAS examiners were certified by a clinician/researcher from the Child Development Unit at Harvard Medical School. The evaluation team was certified reliable on all BNBAS items, including qualifiers, and continued to practice together monthly to maintain intra-group reliability. Twice during the 2-year course of this portion of the project, a BNBAS trainer was on site to certify new examiners and provide reliability checks with national standards for those already certified.
Each evaluator was masked to drug group membership and was always accompanied by an interviewer to protect against disclosure of information about the infant or mother. A backup system for assessment was in place for the rare cases when a tester was unmasked.
The BNBAS assessment was attempted midway between feedings, before potentially tiring procedures. We tried to evaluate infants as close as logistically possible to the ideal 40 weeks postconceptional age. All infants evaluated were between 36 and 43 weeks postconceptional age (mean days from ideal or 40 weeks = 8). On average, control infants were examined 3.7 days after birth and cocaine-exposed infants were tested 5.2 days after birth. There were no significant drug-group differences between chronological or adjusted ages at testing. Assessments were performed in the Clinical Research Center under controlled conditions of light, sound and temperature. If infants were discharged before testing was appropriate, the mother and infant were brought back in the research van to the Clinical Research Center or clinic for assessment. BNBAS data were summarized using Lester's13 traditional seven cluster scoring method as well as the recent Excitable/Depressed scoring scheme.
Simple comparisons of continuous neurobehavioral outcomes between groups were made with the Wilcoxon rank sum test. Simple comparisons of categorical demographic variables were made using Fisher's exact test. Analyses adjusting for use of marijuana, alcohol, and tobacco were performed using a multiple regression analysis with two-way interaction terms for cocaine use with marijuana, alcohol, and tobacco. Spearman rank correlations and partial correlations were used to evaluate the effects of the amount of drug use during each trimester of pregnancy on the birth outcome measures.
Part I includes a description of sample characteristics and pregnancy outcome data. In the BNBAS, certain items can only be administered in a designated state of sleep or wakefulness. For example, if a infant awakens at the beginning of the first Habituation item, the test requires that those items not be completed, or if an infant can never be brought to a quiet, alert state, Orientation items cannot be presented. Thus, sample size for each item varies in this study. Of more interest is the finding that only 12% of the control infants tested failed to come to a quiet, alert state compared with 25% of the cocaine-exposed infants (P = .008).
In simple bivariate comparison analyses there were no significant differences between cocaine-exposed and control infants on any of the six BNBAS cluster scores or number of abnormal reflexes. Nor were there any differences on Excitable or Depressed scores. Cocaine-exposed infants did score significantly lower on six of the nine Qualifier measures as shown in Table 1. There were also differences in Qualifier scores related to other drug use in pregnancy. Tobacco users' infants scored lower (x̄ = 4.0; SD = 2.3) than nonusers' infants (x̄ = 4.7; SD = 2.1;P = .02) on Alert Responsiveness. Alcohol users had infants who were lower (x̄ = 4.6; SD = 1.2) on the score of Examiner Persistence than nondrinkers' infants (x̄ = 4.9; SD = 1.3; P = .02). Analyses by marijuana use revealed that infants of those who smoked marijuana scored lower (x̄ = 3.8; SD = 2.2) on Alert Responsiveness than nonusers' infants (x̄ = 4.5; SD = 2.2;P = .03). Infants of marijuana smokers also had worse scores (x̄ = 5.8; SD = 2.3) on General Irritability than nonmarijuana smokers' infants (x̄ = 6.4; SD = 2.1;P = .047) and scored more poorly (x̄ = 5.1; SD = 1.3) on Robustness and Endurance than infants whose mothers did not smoke marijuana (x̄ = 5.5; SD = 1.3;P = .04).
In regression analyses including cocaine, marijuana, alcohol, and tobacco use, only Alert Responsiveness remained significantly related to prenatal drug use (P = .03). There were two significant interactions, cocaine by tobacco and marijuana by tobacco; in both cases, Alert Responsiveness was less when infants were exposed to both drugs.
There were 288 possible correlations between maternal drug use (four drugs over three trimesters and total pregnancy) and BNBAS measures (seven Clusters, nine Qualifiers, two Summary Scores). Of these possible correlations, 215 were in the hypothesized inverse direction (higher amounts of reported drug use associated with poorer BNBAS scores) and 25 of these were statistically significant (14 or 15 significant correlations could be expected by chance). There were only 2 significant correlations in the unexpected direction. Table2 includes all the significant correlation coefficients between prenatal drug use and BNBAS measures. There appeared to be a pattern to the dose-effects of cocaine related to infant attention and responsiveness. The amount of money spent on cocaine in the third trimester was inversely related to the infant's score on the Orientation cluster of the BNBAS. Reports of the amount of cocaine use in the third trimester and over all of pregnancy were also significantly negatively related to Alert Responsiveness, and the amount of cocaine used during the third trimester was negatively related to Cost of Attention.
There were also several significant inverse relationships between the prenatal use of tobacco and Alert Responsiveness, General Irritability, Regulatory Capacity, and Reinforcement Value of the Infant's Behavior; of alcohol and Motor Maturity, Alert Responsiveness, Cost of Attention, and Reinforcement Value; and of marijuana and Alert Responsiveness. Average amounts of tobacco use in the first and second trimesters were also related to the Number of Abnormal Reflexes elicited on the BNBAS but not in the expected direction. As more cigarettes were smoked there were fewer Abnormal Reflexes observed.
A second set of analyses was performed to evaluate the effects of reported drug usage, using correlational techniques that partialed out the effects of each drug (cocaine, marijuana, alcohol, and tobacco) and their interaction terms. Table 3 gives the significant correlation coefficients found between amount of each drug used and BNBAS scores by each trimester of pregnancy. Holding constant the effect of other drugs used, tobacco was no longer significantly related to any BNBAS score. Alcohol use in the second trimester continued to be significantly negatively correlated with the Motor Maturity cluster but additionally was positively related to General Irritability. First trimester (but not third or total pregnancy) alcohol use continued to be related to the Cost of Attention Qualifer, but alcohol usage was no longer significantly related to the Qualifiers of Alert Responsiveness and Reinforcement Value of Infant's Behavior. Marijuana was no longer related to Alert Responsiveness, but usage in second trimester was significantly positively correlated to the Regulation of State cluster. Cocaine usage was no longer significantly related to the Alertness, Orientation, and Attention Qualifier scores as before, but was correlated with the Regulation of State cluster score. The number of significant partial correlations was no greater than expected by chance.
This report represents one of the largest prospective studies of BNBAS evaluations of cocaine-exposed neonates and matched controls and is among the first to present data on the more recently developed BNBAS Qualifier scores. We demonstrated few significant differences at birth on blinded evaluations of neonatal behavior using the traditional BNBAS cluster scores, confirming our earlier study11 as well as other recent, controlled studies in the literature6,9,10that found no Brazelton cluster differences related to cocaine exposure at birth. We differed from two similarly controlled studies7,8 that found differences only on the Habituation cluster and the study by Black et al5 that found differences on Orientation, State Regulation, and Autonomic Regulation, perhaps because there was less polydrug use or more prenatal care in our sample.14
However, we did find that significantly more cocaine-exposed infants failed to maintain a quiet alert state long enough to be administered Orientation items. Also, several of the BNBAS Qualifier scores were significantly different between cocaine users and matched controls. In regression analyses including cocaine, marijuana, alcohol, and tobacco, Alert Responsiveness continued to be related to prenatal cocaine exposure in interaction with tobacco. Infants tended to be less alert and responsive when exposed to both cocaine and tobacco or to both marijuana and tobacco.
The amount of prenatal drug exposure was also related to several BNBAS measures, again notably, in the areas of attention, alertness, and orientation. After partialing out the effects of usage of other drugs, the amount of cocaine use was no longer related to the BNBAS scores reflecting alertness, attention, and orientation. Cocaine usage was significantly correlated with the measure of State Regulation, in a sense, a precursor of an infant's ability to maintain alertness, attend to stimuli, and focus and follow, orienting successfully.
One must be cautious about the reliability of these behavioral findings due to the large number of correlational analyses. However, except in the most conservative analyses (where partialing out the variance of all drugs significantly reduced power), the number of significant correlations found was greater than would be expected by chance. Furthermore, the results do seem to fit into an overall pattern of effects. Twice as many cocaine-exposed infants as controls, one fourth of the sample, were unable to come to and maintain the quiet, alert state needed to be able to administer and score Orientation test items. In simple comparisons, multiple regression analyses, and in correlations and partial correlations with amount of usage, it was the cocaine-exposed infants who demonstrated the significant decrements in response and always in the areas of regulation of attention and responsiveness. Furthermore, it may be that differences related to cocaine were more often demonstrated by the qualifiers because they are more sensitive measures. Perhaps, experienced evaluators were able to help infants to perform well on most test items (reflected in the cluster scores), but at a cost to the exposed infant's physiologic and attentional systems (reflected by the qualifers).
We have reported neurobehavioral outcome at birth from a prospective, controlled study of prenatal cocaine exposure, determined by careful, extensive interviews and biologic samples. Using blinded developmental examinations, and adjusting for other prenatal drug exposure, we observed fewer alert periods and less Alert Responsiveness in infants exposed to cocaine and tobacco or to marijuana and tobacco as well as a dose effect of cocaine on Regulation of State. If reliable, these findings highlight possible early markers of later problems that exposed children may have in the area of attention and response to novel stimuli. Of more immediate concern is the effect that infants who have difficulty alerting and responding, may have on caregivers, who already may be compromised in parenting abilities by their drug use.
These results generally support those from other controlled studies and remind us that we have few data to warrant the alarm that grew out of earlier reports, often fueled by the media, of the possible devastating effects of prenatal cocaine-exposure. As we15 and others16,17 have warned, an even greater concern may be the potentially devastating effect of a self-fulfilling prophecy on a child who has been so negatively labeled. As researchers, we owe the public the truth, which would afford these children the benefit of hope and individualized care in lieu of negative labels of certain poor prognosis. However, the long-term future of children exposed prenatally to cocaine and any sequelae of their growth retardation and attentional problems, remains uncertain. For the children, their caregivers, health providers, and educators, we as researchers need to continue to monitor with care their progress relative to all factors that may place them at risk, as well as those that may enhance their resiliency.
Supported by National Institute on Drug Abuse grant DA05854 and Clinical Research Center grant RR00082.
- BNBAS =
- Brazelton Neonatal Behavioral Assessment Scale •
- SES =
- socioeconomic status •
- SD =
- standard deviation
- ↵Brazelton TB. Neonatal behavioral assessment scale. Clinics in Developmental Medicine (No. 88). 2nd ed. Philadelphia, PA: JB Lippincott Co; 1984
- Black M,
- Schuler M,
- Nair P
- Eisen LN,
- Field TM,
- Bandstra ES,
- et al.
- Mayes LC,
- Granger RH,
- Frank MA,
- Schottenfeld R,
- Bornstein MH
- Corwin MJ,
- Lester BJ,
- Sepkoski C,
- McLaughlin S,
- Kayne H,
- Golub HL
- ↵Lester BM, Freier K, LaGassse L. Prenatal cocaine exposure and child outcome: what do we really know? In: Lewis M, Bendersky M, eds. Mothers, Babies, and Cocaine: The Role of Toxins in Development. Hillsdale, NJ: Lawrence Erlbaum; 1995
- Eyler FD,
- Behnke M
- Zuckerman B,
- Frank DA
- Copyright © 1998 American Academy of Pediatrics