Developmental outcome for preschoolers prenatally exposed to crack cocaine

MISSING IMAGE

Material Information

Title:
Developmental outcome for preschoolers prenatally exposed to crack cocaine behavioral and adaptive sequelae
Alternate title:
Developmental outcome for toddlers prenatally exposed to crack cocaine behavioral and adaptive sequelae
Behavioral and adaptive sequelae
Physical Description:
ix, 142 leaves : ; 29 cm.
Language:
English
Creator:
Holler, Karen Anne
Publication Date:

Subjects

Subjects / Keywords:
Research   ( mesh )
Prenatal Exposure Delayed Effects -- Child   ( mesh )
Prenatal Exposure Delayed Effects -- Infant   ( mesh )
Crack Cocaine -- adverse effects   ( mesh )
Child Development   ( mesh )
Child Behavior   ( mesh )
Child Behavior Disorders -- chemically induced   ( mesh )
Department of Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- College of Health Related Professions -- Department of Clinical and Health Psychology -- UF   ( mesh )
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 1996.
Bibliography:
Bibliography: leaves 131-141.
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Karen Anne Holler.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 002402910
oclc - 48663375
notis - AMA7840
System ID:
AA00011827:00001


This item is only available as the following downloads:


Full Text













DEVELOPMENTAL OUTCOME FOR PRESCHOOLERS PRENATALLY EXPOSED TO
CRACK COCAINE: BEHAVIORAL AND ADAPTIVE SEQUELAE






By

KAREN ANNE HOLLER



















A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1996











ACKNOWLEDGMENTS


I would like to thank my cochairs, Eileen B. Fennell,

Ph.D., ABPP, and Fonda D. Eyler, Ph.D., for their support and

guidance throughout this project. The breadth of their

combined expertise in the world of child psychology and

development is substantial. I would also like to thank my

committee members, Marylou Behnke, M.D., Michael Conlon,

Ph.D., Steven Boggs, Ph.D., and Sheila Eyberg, Ph.D., for

their advice and practicality in the management of the

research.

I would particularly like to thank Drs. Eyler, Behnke, and

Conlon for making this project possible by generously allowing

me to in participate on their research grant. My association

with their group was among the best of my graduate school

experiences. I would also like to thank Kathie Wobie, M.A.

and Lisa Maag for the all the help and friendship they have

offered before and during the course of this project. My

thanks are also extended to Annie, Kristen, Lorraine, Bruce,

Bridget, Jack and others who, at various times over the past

five years, have been the mainstay of the grant's success.









Last but not least, I would like to thank my parents for

their constant encouragement and support, and Tiffany,

Samantha, and Ben, who make the rest of my life better.

This research was supported in part by NIDA grant

R01DA05854 and GCRC grant M01RR00082.













TABLE OF CONTENTS

page

ACKNOWLEDGMENTS............. ...... .................... ii

LIST OF TABLES...................................... vi

ABSTRACT ................. .................. .......... viii

INTRODUCTION ......................................... 1

LITERATURE REVIEW................ .... .............. 8

Physiological and Behavioral Effects............. 8
Summary of Literature Review.................... 32

PURPOSE OF RESEARCH .................................. 35

MATERIALS AND METHODS.................................... 37

Subjects .......................................... 37
Parent Report Measures........................... 43
Behavioral Observation Measures................. 45
Procedures..................................... 47

ANALYSES .............................................. 49

Descriptive Analyses.................. ....... 49
Hypothesis 1..................................... 49
Hypothesis 2..................................... 51
Hypothesis 3..................................... 51
Hypothesis 4..................................... 51
Hypothesis 5.................................... 52

RESULTS.............................................. 53

Descriptive Analyses............................. 53
Initial Between Groups Analyses.................. 57
Initial Correlations............................ 61
Hypothesis 1..................................... 65








Hypothesis 2.................................... 70
Hypothesis 3..................................... 72
Hypothesis 4.................................... 74
Hypothesis 5..................................... 76

DISCUSSION........................................... 108

Background..................................... 108
Comparison to Norms............................. 110
Between Groups Comparisons...................... 116
Covariate Findings.............................. 122
Strengths and Weaknesses....................... 126
Suggestions for Further Study................... 129
Conclusion....................................... 130

REFERENCES......... .................... ........... 131

BIOGRAPHICAL SKETCH.................................. 142













LIST OF TABLES


Table page

1 Demographic Characteristics of the Total
Sample............................. ........... 79

2 Covariate Characteristics of the Total
Sample.......................................... 80

3 Reliability of Video Coding of Selected
DPICS-II Categories in the Child-Directed
Interaction................................... 81

4 Reliability of Video Coding of Selected
DPICS-II Categories in the Parent-
Directed Interaction ........................... 82

5 Between Group Differences Defined by
Maternal Depression............................ 83

6 Between Group Differences Defined by Home
Inventory..................................... 85

7 Between Group Differences Defined by
Foster Care Status............................ 87

8 Between Group Differences Defined by
Gender......................................... 89

9 Between Group Differences Defined by
Alcohol Use.................................... 91

10 Correlations between Covariates and
Behavioral and Adaptive Variables.............. 93

11 Correlations between Covariates................ 95

12 Correlations between Dependent Subscales........ 96








13 Means and Standard Deviations of Tested
Sample with Age-Based Norms Parent
Report.......................................... 98

14 Means and Standard Deviations of Tested
Sample with Normative Sample DPICS-II
Child Directed Interaction..................... 99

15 Means and Standard Deviations of Tested
Sample with Normative Sample DPICS-II
Parent Directed Interaction.................... 100

16 Dependent Variables Means and Standard
Deviations for Exposed and Nonexposed
Children for the Total Sample.................. 101

17 MANCOVA with Covariates Child Gender and
Maternal Depression and Dependent
Variables from the Connors..................... 104

18 MANOVA with Dependent Variables from the
ECBI............................................ 105

19 MANOVA with Dependent Variables from the
Vineland........................................ 106

20 MANOVA with Dependent Variables from the
DPICS-II Derived Variables.................. 107










Abstract of the Dissertation Presented to the Graduate
School of the University of Florida in Partial Fulfillment
of the Requirements for the Degree of Doctor of Philosophy

DEVELOPMENTAL OUTCOME FOR TODDLERS PRENATALLY EXPOSED TO
CRACK COCAINE: BEHAVIORAL AND ADAPTIVE SEQUELAE

By

Karen Anne Holler

August, 1996

Chairpersons: Eileen B. Fennell, Ph.D., ABPP
Fonda D. Eyler, Ph.D.
Major Department: Clinical and Health Psychology

Early studies indicated that infants exposed prenatally

to cocaine and cocaine derivatives may be at risk for

increased spontaneous abortions, fetal death, preterm labor,

abruptio placentae, decreased length, weight, and head

circumference, congenital malformations, and deviant

neurobehavior. However, very little long-term followup has

been done to assess the functioning of these children as they

mature. The present study compared two groups of three-year

old preschoolers who were a) exposed in utero to crack, or b)

not exposed in utero to crack. Information about child

behavior and adaptive functioning was collected by primary

caregiver report using the Vineland Adaptive Behavior Scales

Survey Form, the Connors Parent Rating Scale, and the Eyberg








Child Behavior Inventory. Child behavior was also assessed by

direct objective observation using the Dyadic Parent-Child

Interaction Coding System-II. Administered instruments

intended to determine whether a) scores for the total sample

conformed to norms provided by psychological measures, and b)

whether behavioral or adaptive deficits were present in the

crack-exposed preschoolers when compared to same-aged

nonexposed peers. Results showed significant differences

between the total study cohort which included exposed and

nonexposed children and normative groups on which many

standardized psychological measures are based. For example,

the total sample was more likely to display elevations in

conduct problems, learning problems, hyperactivity, and

impulsivity compared to normative groups. Second, at three

years of age, no differences were found between a prenatally

crack-exposed cohort on parent and objective measures of

behavior and adaptive functioning when compared with

nonexposed matched peers. Thus elevations on disordered

behavior compared to normative groups cannot be attributed to

exposure status. Implications of findings and possible

explanation for nonsignificant between groups findings are

discussed.













INTRODUCTION

Cocaine is one of the most popular illicit drugs used in

the United States. Approximately 30 million Americans have

tried cocaine at least once and as many as five million use

cocaine on a regular basis (Abelson & Miller, 1985). The two

common forms of illicit cocaine are cocaine hydrocholoride

(HC1) in a powder form, and 'crack,' a highly purified

alkaloidal base also known as 'freebase.'

Cocaine (methylbenzoylecgonine) is a central nervous

system (CNS) stimulant that works in part by increasing the

activation of the neurotransmitter dopamine in the mesolimbic

and/or mesocortical pathways. Increased activation results in

a stimulant induced reward or the 'euphoria' commonly

experienced by cocaine users (Goeders & Smith, 1983; Wise,

1984). Neurotransmitter reuptake is simultaneously reduced in

the cerebral cortex, hypothalamus, and cerebellum resulting in

the hyperaroused state also typical of cocaine intoxication

(Ryan, Ehrlich, & Finnegan, 1987). Cocaine taken intranasally

(or 'snorted') causes vasoconstriction of the nasal mucous

membranes which reduces its own absorption. Subsequently, the

plasma drug concentration rises relatively slowly for











insufflatted cocaine. Cocaine in powder form is also

frequently diluted or cut before sale further limiting plasma

drug concentrations in the body.

Crack on the other hand is almost 70% pure cocaine, made

by precipitating alkaloidal cocaine from an aqueous solution

of cocaine HC1. Unlike the powder form of cocaine, crack is

not destroyed by moderate heating and vaporizes at

temperatures over 98 degrees celcius (Medical Letter, 1986).

Crack, named for the popping sound it makes when heating, may

be smoked via freebasing with a base pipe or rolled with

tobacco in a cigarette. Crack may also be injected. Smoking

crack delivers large quantities of cocaine directly to the

vascular bed of the lungs. The plasma drug concentration in

this instance is substantially higher than for insufflatted

cocaine and rises almost instantaneously producing an effect

more intense than that experienced with intravenous injection

(Medical Letter, 1986).

Crack when smoked reaches the brain in approximately

eight seconds producing an intense, immediate euphoria

(Howard, Mofenson, & Caraccio, 1987). These euphoric effects

are short lived, lasting approximately 45 minutes followed by

a severe crash (dysphoria) during which the user becomes

depressed and agitated (Howard, 1989). This cycle is








3

reportedly shorter and more intense when smoking crack than

for either intranasally or intravenously administered cocaine.

Users may feel a greater compulsion to repeat the experience

when smoking crack both for its magnified euphoric effects as

well to avoid the subsequent crash (Howard et al., 1987).

Binges, characterized by frequent readministration of the

drug, reportedly may last 12 hours but may go on for several

days (Gawin & Kleber, 1985). In addition, the price of crack

continues to decrease, making it accessible to people of all

socioeconomic levels. A 'rock' may cost as little as five to

ten dollars (Howard, 1989). Combined, these factors

precipitate abuse.

While it is uncertain whether cocaine is physically

addictive, it clearly has powerful psychological effects. It

is estimated that five times as many people are addicted to

cocaine than to heroin (Grinspoon & Balkalar, 1980). Indeed,

treatment for cocaine use, and crack use in particular, is

escalating while age of first use appears to be dropping

(Adams, Gfroerer, Rouse, & Kosel, 1986). In reaction to

epidemic use, researchers are beginning to describe not only

the immediate effects of cocaine intoxication but also to

contemplate the long-term implications of using what had










initially been considered a 'safe' stimulant (Schnoll,

Karrigan, Kitchen, Daghestani, & Hansen, 1985; Siegel, 1985).

Associated with the use of cocaine are a number of

behavioral changes as well as substantial physical risk.

Behavioral effects that have been described in adult users

include impulsivity, disinhibition, repetitive actions,

anxiety, psychomotor activation and a loss of appetite which

can result in malnutrition (Howard et al., 1987; Levine,

Washington, Jefferson, Kieran, Moen, Feit, & Welch, 1987;

Medical Letter, 1986). Severe and toxic physiological effects

have been described in the literature. Hypertension,

tachycardia, ventricular arrhythmias, seizures, loss of

consciousness, stroke, and myocardial infarctions are

increasingly attributed to cocaine use. Even deaths due to

respiratory and cardiac arrest have been reported (Howard et

al., 1987; Levine et al., 1987; Medical Letter, 1986).

Further, there are indications that the inhibitory

receptors of dopamine neurons become supersensitive as an

adaptation to chronic activation of the reward pathways which

occurs with stimulant use (Gawin & Kelber, 1986). Thus, there

is some evidence that long-term cocaine use may lead to

permanent neurophysiological changes in the brain that impact

on mood states and the user's experience of pleasure. Such










findings are mirrored in clinical observations of protracted

anhedonia and anergia in some cocaine users (Gawin & Kleber,

1986). Generally, these symptoms resolve over time. However,

there are sporadic reports of high dose users with chronic

anhedonia, anergia and a craving for stimulants that does not

remit (Ellinwood, 1974; Schuster & Fischman, 1985).

Pregnant women are not immune to either the pleasurable

effects of crack or to the physiological consequences of use.

Over the last several years researchers have turned their

attention to the possible impact of in utero exposure to

cocaine and cocaine derivatives on the neonate. This interest

has accelerated due partially to shifting attitudes towards

drug use in general as well as justified alarm at the recent

explosion in crack use (Grinspoon & Balkalar, 1980). Still,

the toxic effects of cocaine use are only beginning to be

understood, and focus has been almost entirely on adult users.

Early reports concerning infant outcomes of in utero exposure

to cocaine has been limited, often anecdotal, contradictory

and predominantly focused on medical outcomes (Ryan et al.,

1987) Thus, while reports in the popular media of antisocial

or hyperactive toddlers created by in utero exposure to

cocaine make sensational copy, they are distinctly premature.











Research, medical or otherwise, on preschoolers exposed to

cocaine in utero is very limited.

The absence of literature notwithstanding, concern for a

generation of children exposed to cocaine is certainly

warranted. One survey suggested that 10% of women may use

cocaine at least once during pregnancy, and 50% of these use

other drugs in addition to cocaine (Howard et al., 1987).

Cocaine is highly water and lipid soluble and passes through

the placentae by simple diffusion. Cocaine may concentrate in

the fetus because fetal blood has a lower pH. In addition,

plasma cholinesterase, which is necessary to metabolize

cocaine, is less active in the fetus and in pregnant women

(Bingol, Fuchs, Diaz, Stone, & Gromisch, 1987). Conceivably,

even small doses of cocaine could have negative consequences

on a developmentally vulnerable embryo and fetus. Although

results in the literature are equivocal, cocaine has been

implicated in a number of negative outcomes including

spontaneous abortions, fetal death, preterm labor, precipitous

labor, abruptio placentae, fetal distress, fetal meconium

staining and other conditions which qualify the newborn for

high risk status (Chasnoff, Burns, Schnoll, & Burns, 1985; Oro

& Dixon, 1987; Ryan et al., 1987).








7

Exposed infants have also been shown to exhibit decreased

length, weight and head circumference, higher rates of

congenital malformations, deviant neurobehavior, rapid shifts

between irritability and lethargy, and be at increased risk

for Sudden Infant Death Syndrome (SIDS) although many of these

outcomes remain controversial (Bingol et al., 1987; Chasnoff,

1989; Chasnoff, Burns, & Burns, 1987; Oro & Dixon, 1987).

However, negative findings are not consistently found in well

controlled studies. Reports of investigations into the

behavioral and long-term developmental outcome of cocaine-

exposed infants remain unclear.













LITERATURE REVIEW


Physiological and Behavioral Effects

The physiological impact of in utero cocaine exposure has

been extensively modelled in animal investigations. The

underlying assumption in such studies is that cocaine may

affect development in two principle ways: 1) by disturbing

dopaminergic functioning in the developing fetus, and/or 2)

via hypoxic effects.

The dopamine system is implicated in the modulation of

reward systems, reinforcement, sensorimotor integration, and

environmental responding. Cocaine acts as an inhibitor of

dopamine reuptake presynaptically, serving to increase the

amount of amines in the synaptic cleft. There is evidence

that repeated cocaine administration decreases the amount of

dopamine synthesis presynaptically and increases the number of

dopamine receptors postsynaptically (Fung, Reed, & Lau, 1989).

The dopaminergic system is developing in the third trimester

of human fetal development and is functional and capable of

mediating behaviors in rat fetuses towards the end of their

gestational period. Moody, Robison, Spear, and Smotherman,











(1993) found that administration of cocaine to rat dams

resulted in increases in fetal activity which they

subsequently considered a reflection of altered CNS

development.

Subsequent investigations found that prenatal exposure to

cocaine in late gestation results in a decrease in the number

of spontaneously active midbrain dopaminergic cells in adult

rats, whereas rats exposed to cocaine only in adulthood showed

an increase in the number of spontaneously active dopaminergic

cells (Minabe, Ashby, Heyser, Spear, & Wang, 1992). The

decrease of midbrain dopamine activity in adult rats exposed

prenatally to cocaine has been used to explain early findings

of irritability and decreased interactive behaviors in human

infants. These reports suggest, not surprisingly, that

prenatal cocaine exposure may be toxic to the fetus or impact

negatively on the developing neuronal system in a manner that

does not necessarily parallel consequences of adult exposure.

Studying cell development in brain regions, Seidler and

Slotkin (1993) compared rat pups exposed in utero to cocaine

during late gestation to a nonexposed comparison group. They

found retarded maternal weight gain but no effect on pup body

or brain region weights. DNA content was also largely

unaffected. Although postnatal cell growth was reduced in the










forebrain, the magnitude of the reduction was small when

compared to the impact of other drugs such as the hypoxic or

ischemic effects of nicotine. Seidler and Slotkin (1992)

concluded that functional deficits attributable to fetal

cocaine exposure probably result from actions directed toward

specific cell or synaptic populations as opposed to global

effects on cell development.

In another investigation, Seidler and Slotkin (1992)

found somewhat conflicting results when studying the effect of

fetal cocaine exposure on rats given 30 mg/kg daily from

gestational days 2 to 20. As a toxic referent, a dose of

approximately 80 to 100 mg/kg of cocaine is generally fatal in

adult rats. The exposed group showed minor differences from

controls in body and brain region weights and in levels of

norepinephrine. However, investigators found marked

noradrenergic hyperactivity as assessed by noradrenergic

turnover, which they attributed to the effects of

glucocorticoids and hypoxia on noradrenergic cell

differentiation. These results suggest that hypoxia rather

than dopaminergic disruption may be the more critical result

of cocaine exposure in utero.

Animals prenatally exposed to cocaine generally show

differential physiologic responses to hypoxia compared to










controls. Weese-Mayer and Barkov (1993) studied the responses

of rabbit pups to hypoxia following exposure to 30 mg/kg of

cocaine from days seven to 15 gestation. They found the

cocaine-exposed pups had more significant oxyhemoglobin

desaturation and pulse deceleration during exposure to severe

hypoxia than controls, suggesting that cocaine exposure in

utero may reduce the normal defense mechanisms for metabolic

adjustment to low oxygen.

These results supported findings from an earlier study by

Weese-Mayer, Klemka-Walden, Barkov, and Gingras (1992) who

also found that, while baseline ventilation did not differ

significantly among study groups, cocaine-exposed pups had a

deficient ventilatory response during an hypoxic challenge.

Weese-Mayer et al. (1992) argued that these findings may

represent perturbed maturation of respiratory control. Other

researchers have found similar results which suggest that

infants exposed to cocaine have an impaired repertoire of

protective responses to hypoxia and hypercapnia during sleep

which may play a role in increased risk for SIDS (Ward,

Bautista, Woo, Chang, Schuetx, Wachsman, Sehgal, & Bean,

1992).

Similarly, Woods, Plessinger, and Clark (1987) found that

pregnant ewes exposed to cocaine produced increases in










maternal blood pressure and decreases in uterine blood flow

accompanied by fetal hypoxemia, hypertension, and tachycardia.

Cocaine administration to the fetus directly resulted in

relatively smaller increases in fetal heart rate and blood

pressure than those observed following maternal

administration, and no changes in fetal arterial blood gas

values. These results were interpreted to indicate that

maternal administration of cocaine alters fetal oxygenation by

reducing uterine blood flow and impairing oxygen transfer to

the fetus and that resulting fetal cardiovascular changes may

reflect fetal hypoxemia, increased fetal levels of cocaine, or

a combination of these events.

Birth outcome studies with animals demonstrate a range of

postnatal results following cocaine exposure in utero.

Henderson and McMillen (1990) studied postnatal development in

rats exposed to cocaine daily in utero. The cocaine-exposed

group weighed less, had more stillbirths, and more birth

defects when compared to a nonexposed group. In addition,

cocaine-exposed pups had delayed righting reflexes although no

delay was noted in opening their eyes, both indices of

developmental integrity. Their findings led Henderson and

McMillen (1990) to conclude that cocaine exposure in utero










impacted neonatal outcome and long term development in the

rats due to disruption of the dopamine system.

In a study designed to mimic moderate cocaine use by

humans during pregnancy, Fung et al. (1989) examined the

neurobehavioral responses and striatal dopaminergic system in

cocaine-exposed newborn rats. No change in length of

gestation, litter size, birth weight and length of pups, and

the ratio of male to female pups was noted. At 14 days old,

exposed and nonexposed pups showed similar locomotor

performance in righting reflex, position reflex and negative

geotaxic tests. Results demonstrated that prenatal exposure

to cocaine did not alter the development of motor coordination

suggesting no significant impact on the developing

dopaminergic and striatal systems with moderate exposure.

Spear, Kirstein, Bell, Yoottanasumpun, Greenbaum, O'Shea,

Hoffmann, and Spear (1989) also studied the early development

of cocaine-exposed rats. They found that when pups were

exposed a relatively large dose of cocaine daily (40 mg/kg)

late in gestation, there were no differences in maternal

weight gain, duration of pregnancy, or number of live

male/female pups per litter were found. Like Fung et al.

(1989), they found no differences in offspring body weights at

birth and weaning, physical maturation and reflex development.










However, cocaine-exposed pups were deficient in learning and

odor/milk association and showed enhanced locomotion. The

researchers concluded that prenatal cocaine exposure impacts

behavioral and cognitive function during the early postnatal

period even in the absence of overt physiological changes.

A review of studies of postnatal integrity of animals

exposed to moderate amounts of maternal cocaine administration

reveals some consistencies. Generally studies have found no

differences in dam weight gain, litter size or weight, or

early postnatal behavioral tests (Smith, Mattran, Kurkjian,

and Kurtz, 1989; Hutchings, Fico, & Dow-Edwards, 1989; Spear,

Kirstein, & Frambes, 1989). However, drug effects have been

noted on the level of locomotion, exploratory behavior, tail

flick, footshock sensitivity and some learning and/or

retention tasks. Arguments explaining significant results are

generally dependent on altered neurodevelopment during

critical prenatal periods, poor maternal nutrition, and

hypoxia.

While animal models for cocaine exposure in utero may be

useful, there are several difficulties in extrapolating the

effects found in animals to humans. In addition to the

obvious problem of equating human and rat, sheep, or rabbit

neurodevelopment, results of animal studies also tend to vary










by species or strain. Furthermore, they appear to be heavily

dose-dependent including some studies using extremely high

doses of cocaine, and demonstrate that impact appears

contingent upon the gestational period of drug administration.

Clearly fetal exposure to cocaine is not a benign perinatal

event. But the deficiencies noted in the animal literature

such as locomotor changes or alterations in learning and

"cognition" are frequently difficult to demonstrate in

parallel forms in human infants. In addition, it is unclear

whether such alterations in functioning found in the animal

literature are transient or long-term.

Preliminary attempts have been made to study human

infants in terms of the physiological impact of cocaine

exposure in utero. For example, Link, Weese-Mayer, and Byrd

(1991) performed magnetic resonance imagery (MRIs) on infants

exposed to cocaine prenatally to determine the presence of

hypoxic injuries. In all 21 infants studied, at a mean age of

3.6 years, they found myelination was appropriate compared to

age-matched norms; brain and brainstem anatomy were also

normal with no evidence of infarct or hemorrhage. Link et al.

(1991) concluded that negative findings could be due to the

small sample size, limited maternal use, or the possibility

that cocaine use without concomitant use of other illicit










substances may not be as significant as some studies have

suggested.

Doberczak, Shanzer, Senie, and Kandall (1988) likewise

examined 39 infants with in utero exposure to cocaine for

neurologic and electroencephalographic (EEG) abnormalities.

During the first week of life, 17 of the infants had abnormal

EEGs and abnormal behavior characterized by irritability. By

the second week of life, nine of the 17 EEGs remained

abnormal. By three to 12 months of age, however, findings had

normalized, suggesting effects attributed to cocaine may be

transient. There was no control group utilized in this study

and it is unclear whether examiners were blinded to drug

status. Despite these methodological flaws, it may be

significant that EEG findings could not be predicted by

neurologic dysfunction or perinatal variables, and ultimately,

findings normalized for all infants.

Preliminary study of human infants indicates that cocaine

exposure results in worse outcomes than nonexposed infants in

terms of birth weight, length and head circumference,

spontaneous abortion, fetal death, and sudden infant death

syndrome (SID) (Ryan et al., 1987). In an early study, Oro

and Dixon (1987) examined neonatal growth, behavior and

physiologic organization in 104 mother/infant pairs including










a cocaine and methamphetamine group, a narcotic group and a

drug free group. Findings in this study were plentiful and

dramatic. The researchers found significantly lower birth

weight, length, occipitofrontal head circumference (OFC), and

gestational ages in both the cocaine/methamphetamine and

narcotics groups compared to the illicit drug free group.

They also found significantly higher incidence of prematurity,

intrauterine growth retardation, fetal distress, and

complications during the neonatal period in both the

cocaine/methamphetamine and narcotics groups compared to the

illicit drug free group.

Significant increases in neurologic and physiologic

problems in the cocaine/methamphetamine group were also

reported including (in order of decreasing significance)

abnormal sleep patterns, tremors, poor feeding, hypertonia,

vomiting, sneezing, high pitched crying, frantic fist sucking,

tachypnea, loose stools, fever, yawning, hyperreflexia, and

excoriation. Physician and nurse descriptions of the behavior

of these exposed infants included disorganization, poor visual

processing of faces and objects, random sucking and long dull-

alert periods with eyes open. Decreased spontaneous activity

and fixed catatonic postures were seen in four of the

cocaine/methamphetamine group. Even accounting for various








18

maternal factors, cocaine/ methamphetamine and narcotic use

still made significant independent and negative contributions

to gestational age, birth weight, length and OFC (Oro & Dixon,

1987).

The results of this study and of many early studies like

it appeared to confirm the clinicians' worse fears regarding

cocaine exposure in utero. However, a critical evaluation of

the methodology mitigates the findings to a degree. For

example, in the above study, the combined effects of certain

drugs (in this case methamphetamine and cocaine) may have had

some unique or synergistic properties not found when the same

drugs are taken separately. Further, it is unclear whether

evaluators in this study were blinded to the drug group

affiliation of the babies being studied. Confounds and

researcher bias have been frequently apparent in this body of

literature on prenatal cocaine exposure, especially in earlier

reports.

In another study concerned with the teratogenic effects

of cocaine, Bingol et al. (1987) compared a polydrug, a

cocaine only, and a no illicit drug control group. The

cocaine group was comprised of 60% intranasal, 30% free base

inhalation, and 10% intravenous users. While no statistical

differences were reported in the spontaneous abortion rates










among the three groups, stillbirth rates were significantly

higher in the cocaine only group compared to the no drug

group. Also, there were significantly more congenital

malformations in the cocaine only group compared to both

contrast groups. Five of fifty infants in the cocaine only

group had major congenital malformations including

exencephaly, intraparietal encephalocele, and parietal bone

defects. Although complications occurred in women after all

three routes of cocaine administration, the sequelae were more

frequent in those who smoked crack regularly or injected it

intravenously. These findings are dramatic and suggest that

cocaine is a powerful teratogen to the developing fetus.

However, this study was conducted in a large inner city

hospital where the confound of polydrug use was common among

the cocaine-using group, and it is unclear whether examiners

were blinded as to the drug status of the infants.

Little, Snell, Klein, and Gilstrap (1989) also examined

perinatal outcome following maternal cocaine use during

pregnancy in 53 exposed infants compared to a 100 infant

control group. Cocaine use in their study was associated with

preterm labor; birth complications including meconium

staining, tachycardia, and lower birth weight; and an excess

of congenital cardiac anomalies. There were several








20

methodological shortcomings in this study which may impact the

reliability of the authors' conclusions. Mothers were self

reported drug users whose neonates were automatically referred

to a high risk nursery for observation of withdrawal symptoms

compromising blind evaluations. Furthermore, the cocaine-

using mothers were generally older, more likely to be black,

and use tobacco and other illicit drugs when compared to

controls. Controls were not matched to drug using mothers,

and it is unclear whether other risk variables were controlled

adequately in the analysis.

In a subsequent study, Little and Snell (1991) studied

the pattern of brain growth in cocaine-exposed newborn infants

where brain growth was defined as head circumference in

addition to other growth variables including birth weight,

length, and gestational age. Groups were cocaine-exposed

infants without alcohol exposure, alcohol but no cocaine

exposure, and neither cocaine nor alcohol exposure. Results

showed significant differences in head size between unexposed

and cocaine-exposed infants. Head circumference was reduced

proportionately more than birth weight in cocaine-exposed

infants. Cocaine and alcohol exposed groups were not

statistically different in terms of head size. Little and

Snell (1991) concluded that cocaine exposure results in a








21

pattern of growth retardation similar to alcohol exposure and

that this retardation may be asymmetrical with head size more

involved. However, mothers in the cocaine group tended to be

polydrug users, control groups were not matched, nor were

statistical controls used for possible confounding maternal

variables.

Chiriboga, Bateman, Brust, and Hauser (1992) reported

that, compared to a no illicit drug exposure control group (n

= 16), cocaine-exposed infants (n = 14) had significantly

lower birth weights, lengths and head circumferences.

Neurologic abnormalities were also associated with exposure

and included hypertonia, plantar extension, tremors and gaze

abnormalities. These findings led researchers to conclude

that prenatal cocaine exposure results in tone and movement

abnormalities in newborn infants. Strengths of this research

were the blinded status of the neurologic examiners and the

fact that toxicology screens were used in addition to

histories to determine cocaine use. However, this study also

was conducted in an inner city hospital serving primarily low

socioeconomic status (SES) women and information on

confounding polydrug use by mothers was considered either

somewhat unreliable or was unavailable to the researchers.








22

Alternately, Hadeed, and Siegel (1989), when studying

neonates of 56 mothers who used cocaine prenatally, found no

differences in the frequency of maternal preeclampsia or

caesarian section (C/S) rate, teratogenicity, narcotic

withdrawal symptoms, or illnesses compared to a no illicit

drug control group. Alternately, weight, length and head

circumference growth curves of infants born to cocaine-using

mothers shifted below the 25th percentile, although this

finding was not thought to be clinically significant. In

addition, cocaine use did seem to precipitate more spontaneous

abortions, abruptio placentae, and meconium stained amniotic

fluid. Advantages of this study included the use of a cocaine

only group and blinded evaluations. In addition, maternal

factors were controlled for differences in age, parity,

socioeconomic status (SES), ethnicity, and smoking, making

findings apparently more reliable.

One of the most prolific researchers in this area, Ira

Chasnoff, consistently reported a number of negative outcomes

for infants exposed to cocaine in utero in his early

investigations. In addition to the effects described by both

Oro and Dixon (1987) and Ryan et al. (1987), Chasnoff (1989)

reported a higher incidence of significant genitourinary tract

malformations. In one study of a pool of 70 infants with










cocaine-using mothers, Chasnoff (1989) found two infants with

prune belly syndrome, (a congenital nephrotic disorder) one

with female pseudohermaphroditism, two with hypospadias and

undescended testes, and three with hydroureter/hydronephrosis.

Further, two infants in this study suffered perinatal cerebral

infarctions that he attributed to maternal cocaine use in the

48 to 72 hours preceding delivery.

In a similar study, Chasnoff, Chisum, and Kaplan (1988)

found nine of 50 infants born to prenatal cocaine users had

some form of physical anomaly compared to only one of 30

infants born to polydrug noncocaine-using women. These

anomalies included two with ileal atresia and seven with

malformations of the genitourinary tract. While the

researchers attributed these effects to cocaine exposure, all

women in the study were known substance users specifically

enrolled for treatment possibly confounding outcome, and

polydrug use was common.

Researchers who have considered the outcome of cocaine-

exposed children in the context of other factors have found

more moderate results. For example, Bauchner, Zuckerman,

McClain, Frank, Fried, and Kayne (1988) assessed the risk of

SID among cocaine-exposed infants and found one of 175

prenatally cocaine-exposed infants died of SID and four of 821








24

unexposed children died of SID. These results suggested to

Bauchner et al. (1988) no increased risk of SID among infants

exposed in utero to cocaine. They observed that cocaine-using

mothers tended to use other drugs and alcohol, and tended to

have low birth weight babies. They further noted that SID

usually accounts for one to two deaths in a thousand; in poor

black groups, the rate may be as high as five to six per

thousand.

MacGregor, Keith, Chasnoff, Rosner, Chisum, Shaw and

Minogue (1987) again examined the perinatal outcome data of 70

women receiving care at a large urban drug treatment center.

In this study, the use of cocaine during pregnancy was

associated with younger gestational age at delivery, increase

in preterm labor and delivery, lower birth weight and delivery

of smaller for gestational age infants. However, no cocaine

related differences were found in the incidence of abruptio

placentae and congenital anomalies. Further, no differences

were noted on the basis of patterns of substance use. Again,

problems in this study included the possible biases of

unblinded investigators and confounding of polydrug use.

Schneider and Chasnoff (1992) also investigated the motor

development of cocaine/polydrug infants at four months of age.

Assessing 50 nonillicit drug exposed infants and 74 cocaine-










exposed infants, Schneider and Chasnoff (1992) found

significant differences between groups in muscle tone,

primitive reflexes and volitional movement with cocaine-

exposed infants performing more poorly. Results demonstrated

that motor differences were evident in cocaine-exposed infants

at a postnatal period beyond the first month of life. As in

other studies from the Chasnoff group, mother/infant

involvement in a drug treatment program was known by

investigators, and subjects in the cocaine group included

polydrug users.

In a two year followup, Chasnoff, Griffith, Freier and

Murray (1992) investigated the growth and developmental

outcome of cocaine-exposed infants. Comparing a cocaine-

exposed group (with marijuana and alcohol exposure), a

marijuana/alcohol exposed group, and a nonexposed control

group, they found that cocaine-exposed children no longer

lagged behind nonexposed children in terms of length and

weight although the cocaine group and marijuana/alcohol group

both lagged in head growth. No differences in performance on

the Bayley Scales of Infant Development (BSID) were noted

between the three groups. Again, these children were all

known to the researchers through a drug treatment program, and

polydrug use was a confound. Further, the study sample had a








26

high rate of attrition. The conclusion that head growth after

birth may be a biological marker for drug exposure may be

correct; however, this cannot be attributed solely to cocaine

exposure in utero, particularly since the marijuana/alcohol

but no cocaine group also lagged in head growth.

In the context of longer term development, Azuma and

Chasnoff (1993) subsequently assessed the three year outcome

of cocaine-exposed children using cognitive and behavioral

measures. They found that children prenatally exposed to

cocaine were no different from controls in terms of

intelligence measured on the Stanford Binet (SBIT). However,

statistical modelling using path analysis suggested that drug

exposure, home environment and level of perseverance

(attention) had direct effects on cognitive function, while

head growth and parent report of behavioral functioning did

not have a direct effect on test scores. The authors

considered this a "best case" outcome study because all women

were involved in a treatment program. However, as in the two

year followup of the same population, possible biases and

confounds existed particularly with regard to subject

attrition.

From the same research team, Griffith, Azuma, and

Chasnoff (1994) also evaluated the three-year behavioral and








27

developmental outcome of children prenatally exposed to

maternal substances of abuse. Their subjects included 93

children exposed prenatally to cocaine and other drugs, 24

children exposed to polydrugs without cocaine and 25 children

who were nonexposed. Drug exposed children had smaller head

circumferences than nonexposed children; the

polydrug/noncocaine group performed worse than nonexposed

children on tasks of abstraction and visual reasoning; the

cocaine/polydrug group performed worse on a task of verbal

reasoning; and a marijuana group performed more poorly on a

task of abstract/visual reasoning. Caregivers also rated drug

exposed children as more aggressive than nonexposed children.

The authors concluded that not all substance exposed

children suffer the same poor prognosis, and that

generalizations about outcome for drug-exposed children need

to be qualified pending more thorough investigations of the

roles of maternal and environmental factors such as SES

levels. Recognizing that cocaine may not account for all

negative findings and that bias on the part of researchers and

even parents could convolute findings was an important pivot

point, not only for this group of researchers but for other

investigators in the field as well.








28

Studies focusing on the behavioral outcome of cocaine-

exposed infants present a range of findings and methodology.

In a study of fetal behavioral state as a predictor of

neonatal outcome, Hume, O'Donnell, Stanger, Killam, and

Gingras (1989) performed fetal assessments which included

ultrasonographic examination, videotaping and scoring of a

behavioral protocol developed by the authors. Abnormal or

delayed state behavior was identified in 13 of 20 fetuses

exposed to cocaine. State organization was also suspect or

abnormal for 16 of 20 exposed newborns, and the disorganized

behavioral states in fetuses successfully predicted abnormal

newborn behaviors. The authors concluded that cocaine

disrupts (CNS) development. Again, problems in this early

study included lack of a control group, small sample size,

unblinded examiners, and the use of unnormed scales which,

because they were developed by the authors for this study, may

have led to biased observations.

Davis, Fennoy, Laraque, Kanem, Brown and Mitchell (1992),

who were also interested in behavioral outcome, studied 70

children with cocaine exposure in utero who had been targeted

by positive urine screen, maternal report, or by doctor/agency

notation. All children in the study had been referred to a

large inner city hospital for developmental evaluation. Davis










et al. (1992) found significant neurodevelopmental

abnormalities including language delay and a high frequency of

autism (11.4%) in the cocaine-exposed group. Since autistic

disorders have not been linked to alcohol or opiate exposure

alone, Davis et al. (1992) attributed their findings to

cocaine use. There are obvious deficiencies in this study

including the nonblinded status of clinical evaluators, the

skewed sampling of clinically referred children, the polydrug

use status of the mothers, and the conclusion that results

must be narrowly attributed to cocaine exposure in utero.

In a study of 51 cocaine-exposed infants and 60

nonexposed infants targeted by maternal report or positive

infant urine screening, Neuspiel, Hochberg, Greene, and

Campbell (1991) found no differences on the Brazelton Neonatal

Behavioral Assessment Scale (BNBAS) administered between one

and three days of age. A second examination at 11 to 30 days

of age showed a significant difference in motor functioning in

the cocaine-exposed group. However, this difference

disappeared when researchers controlled for confounding

variables such as perinatal and social factors. No

differences were detected in observed maternal or infant

behaviors indicating that prenatal cocaine use had no

significant impact on maternal/infant interactions.








30

In a study with variable outcomes, Eisen, Field,

Bandstra, Roberts, Morrow, Larson, and Steele (1991) studied

26 target and 26 control infants for the effects of maternal

cocaine use on the Brazelton Neonatal Behavior Assessment

Scale (BNBAS). Infants were excluded if their mothers used

opiates; and mothers were matched for maternal age, ethnicity,

gravida, previous abortions, and hepatitis. Cocaine-using

mothers were more likely to be tobacco and marijuana smokers

and alcohol drinkers.

Results showed no differences between controls and

cocaine-exposed infants in sex distribution, gestational age,

chronological age, birth weight, birth length or postnatal

complications. However, cocaine-exposed infants had smaller

head circumferences and more obstetric complications than the

nonexposed infants. In addition, cocaine-exposed infants

showed more stress behaviors defined as abnormal reflex

behavior and autonomic instability on the Neonatal Stress

Scale. However, regression analysis showed obstetric

complications and maternal alcohol use were the only

significant variables that contributed to the variance on the

Neonatal Stress Scale. Finally, the cocaine group

demonstrated impaired ability to habituate responses to

repeated stimuli; regression analysis showed cocaine exposure










to account for a significant amount of the variance in

habituation scores. Maternal polydrug use may confound

findings somewhat in this study.

In an attempt to improve upon the methodology in the

prenatal cocaine exposure literature, Woods, Eyler, Behnke,

and Conlon (1993) longitudinally assessed infant behavior

using the BNBAS at birth and again at one month. Mothers

enrolled in this study were from a rural, low income

population with little or no drug treatment. All examinations

were performed blindly. Woods et al. (1993) found cocaine-

exposed infants had lower birth weights and shorter gestations

than controls, a finding typical of many investigations.

However, there were no differences in neonatal performance on

the BNBAS at birth or at one month of age. These results

indicate that not all cocaine-exposed infants show

neurobehavioral deficits in the neonatal period.

Again, studies of infant neurobehavior report varying

results. For example, Mayes, Granger, Frank, Schottenefeld,

and Bornstein (1993) reported that exposed infants had poorer

orientation than nonexposed infants, while Richardson and Day

(1991) reported no differences on any Brazelton scale. In

general, studies documenting birth outcome using repeated

BNBAS assessments report a pattern of developmental recovery








32

over the first few weeks of life for the cocaine-exposed

infant (Black, Schuler, & Nair, 1993; Coles, Platzman, Smith,

James, & Felix, 1992).



Summary of Literature Review

A review of the literature concerning cocaine effects on

fetal development indicates several serious methodological

problems in human clinical studies, including probable

underreporting of cocaine use and the increased likelihood

that cocaine users also use more alcohol and other drugs.

There is also the likelihood of confounding of other maternal

risk factors that are more common in cocaine-using groups than

nonusers such as sexually transmitted diseases (including

HIV); previous spontaneous and elective abortions, and less

prenatal care; low birth weight infants; polydrug use; poor

health and nutrition; and poverty (Dow-Edwards, 1991). Other

methodological problems endemic to the literature include a

preponderance of unblinded evaluations. Further, most

research has concentrated on urban populations and

inappropriate control groups and analyses which frequently

ignore important environmental and maternal factors.

In terms of specific animal research findings, cocaine

generally has no effect on gestational length in the rat, a








33

few studies report an increase in stillbirth rate, and cocaine

appears to reduce litter size but only at the most toxic doses

which may not be relevant to human studies. In animals, odor-

associated learning appears to be impaired in some studies.

Altered neuronal circuits, including nigrostriatal pathway

(fine motor tuning) and mesolimbic dopaminergic system

(reinforcement), have been also demonstrated. Head size in

animals does not appear to be associated with prenatal cocaine

use.

While some reports in the human clinical literature

suggest that cocaine use during pregnancy is associated with

smaller babies, others do not find statistically significant

effects on fetal growth. Evidence of decreased length in

human fetal growth has been found, but it is unclear whether

these results can be attributed to cocaine or to other factors

such as maternal malnutrition and hypoxia. Further, no

specific constellation of fetal malformations is found with

prenatal cocaine use although case reports of craniofacial

defects, missing digits, and genitourinary malformations exist

(Dow-Edwards, 1991).

Transient neurological deficits have been noted in

neonates exposed to cocaine in utero which have included

tremors, rigidity, hyperactivity, abnormal EEG, seizures,











abnormal orientation, sensory and motor functions. Most of

these problems appear to resolve by six months of age. Some

studies have found exposed infants at risk for SID whereas

others have not found a significant association with cocaine.

Additionally, neonates exposed in utero to cocaine do not

appear to suffer from addiction to cocaine which is unlike

infants exposed to opiates in utero.

At the present time, reduced length and head

circumference appear to be the most consistent findings

reported for neonates exposed prenatally to cocaine, and

again, these deficiencies are usually resolved by one year of

age. There does not appear to be a clinical syndrome which

adequately captures the impact of cocaine exposure in utero on

the developing human. Individual variation in outcome appears

to be the most consistent finding. The methodological

difficulties of working with a drug-using population,

exaggerated media portrayals of 'crack babies,' and the need

to control for investigator bias in studies have added to the

confusion evident in the cocaine literature. Thus, while

cocaine exposure in utero may reasonably be expected to have

deleterious effects on some children, it is not possible at

this time to define a syndrome in any, even globally,

meaningful way.













PURPOSE OF RESEARCH

While the particular results of studies vary somewhat,

reports consistently portray cocaine as an agent strongly

implicated in a host of negative outcomes. Further, crack has

been singled out as the most potent form of cocaine both in

terms of abuse potential and morbid impact (Kaye, Elkind,

Goldberg, & Tytun, 1989). Clearly research on the

implications of in utero exposure to cocaine, and especially

crack, needs to be ongoing and continually refined from a

methodological, medical and psychosocial perspective.

At present, long-term followup on the effects of prenatal

exposure to cocaine and cocaine derivatives is extremely

limited. Because of the variability in outcome cited in the

neonatal literature, it remains unclear what long-term impact

in utero cocaine exposure may have on the developing child.

In the past, the media have portrayed cocaine-exposed infants

and toddlers as behaviorally disordered, impulsive,

nonempathetic, and autistic, generally without benefit of

sound empirical evidence. These prejudices have frequently

been supported by biased investigations making hasty

unwarranted projections as to future status while ignoring the








36

very significant presses of poverty, malnutrition and a

caregiving environment compromised by addiction. While the

potential toxicity of cocaine exposure should be in no way

minimized, it has been the recent onus of researchers in this

area to produce methodologically sound and conservatively

interpreted accounts of the impact of cocaine on the

developing infant and toddler. Only recently has sound

research on the infant and toddler been possible as

prospectively enrolled cohorts mature to preschool age and

older. Thus, there are a number of interesting questions to

be investigated. Whether the effects linked to cocaine

exposure in utero are stable over time, and how they translate

behaviorally and developmentally will require increasing

attention as these infants mature.

The present study is intended as an initial exploration

of some of these developmental issues within improved

methodological design. Addressed here is whether toddlers

exposed in utero to crack differ from same aged unexposed

peers in their level of behavioral and adaptive development.













MATERIALS AND METHODS


Subjects

Subjects were 30 children exposed to crack in utero and

30 nonexposed control children who were part of a large,

ongoing, prospective longitudinal research project conducted

at Shands Teaching Hospital through the Department of

Pediatrics, Division of Neonatology. Subjects were the

children of women who were originally recruited predominantly

from public health department (PHD) prenatal clinics in two

counties in an understudied rural area of north central

Florida. Recruitment began in the Fall of 1991. These women

represented a range of low to moderate reproductive risk and

all were scheduled to deliver at the referral hospital. There

was a wide range of drug use with most smoking crack cocaine

and few having access to drug treatment.

Criteria for inclusion in the original study were: 1)

women over 18 years of age to reduce the confounding influence

of perinatal risk known to be associated with pregnancy in the

young teenager and to eliminate problems with obtaining

informed consent from minors, 2) women with no major illnesses










diagnosed prior to pregnancy that are known to affect

pregnancy or developmental outcome or maternal interaction or

caregiving, such as diabetes, chronic hypertension, immune

complex disease, seizure disorders, mental illness or

retardation, etc., 3) women with no history of illegal drug

use other than marijuana and cocaine, 4) English speaking, and

5) women who did not use any of the following drugs without a

prescription or chronically with prescription: amphetamines,

benzodiazepines, barbiturates, opiates, methadone, and

methaqualone.

Target subjects were selected during a screening

interview from those who admitted any prenatal use of cocaine

or had a full toxicology screen positive for metabolites of

cocaine; maternal urine specimens were taken at two

unanticipated times, study enrollment and delivery. Any

patient who during the screening interview denied the use of

cocaine, the illicit substances and drugs listed above, and

had a negative cocaine toxicology screen were retained in the

subject pool as a potential match for a cocaine-using subject.

Matching was done within each county health department to

equate level of prenatal risk and included the following

subject conditions: 1) maternal race: black vs. nonblack; 2)

parity: primiparity vs. multiparity; and 3) socioeconomic










status: level of Hollingshead Index. These are factors which

have been shown to mediate caregiving and affect outcome of

high risk infants.

All subject mothers were interviewed once or twice

prenatally and/or at the time of birth, and their surviving

infants were evaluated at each followup time. Detailed drug-

use histories and psychosocial interviews were done by

helpful, nonjudgmental, experienced interviewers. Calendars

were used to help women identify timing of use, and probing

for drug use details was done about past but not present use

in an effort to be less threatening. Infants were assessed

one to three times in their first week of life and six more

times over three years. Three followup assessments were made

in the homes and three in the clinics. All infant assessments

were performed by trained, reliable examiners blinded to drug

exposure history. Subjects in the present study were seen at

three years of age plus or minus eight weeks, and measures

included parent report instruments and behavioral observation.

Several variables were chosen for use as covariates

during statistical analysis in order to strengthen results

including maternal alcohol use, maternal depression, child

gender, and home environment. Covariates were chosen a'priori

based on literature that indicated such variables correlate










significantly with infant/toddler development. Placement in

foster care was chosen post hoc to be included as a covariate

in the analysis due to the difference in frequency between the

two groups and due to the potential impact of foster placement

on child outcome at three years of age. Foster care was

defined as a child being placed in a home away from, or with

a relative other than, their biological mother. Thus, five

variables (alcohol use, maternal depression, child gender,

home environment, and foster placement) were included as

covariates in the initial analyses between groups.

In order to further strengthen the analyses, subjects

were excluded from the study based on several birth

characteristics which could significantly skew results. These

included any child born with a birth weight below 2,500 grams

and/or a congenital malformation. Children exposed to cocaine

in any form other than crack were also excluded from the

study. It was decided a'priori to exclude children from the

study who suffered any accident or illness (such as sickle

cell anemia) that might seriously affect their performance on

developmental measures independent of crack exposure in utero.

Altogether, 90 children from the longitudinal study were

evaluated between July 1, 1995 and January 30, 1996. Subjects

were assessed in the order that they came in for their three










year followup as part of the ongoing study. Thirteen children

(14.4%) were dropped from the study. Of the thirteen excluded

children, 9 were low birth weight (69.3%; 5 targets and 4

controls), 3 were exposed to forms of cocaine other than crack

(23.1%), and one had sickle cell anemia (7.7%). At the close

of enrollment for this study, the matches for 17 (18.9%) of

the original 90 had not been tested. Thus, the 17 unmatched

subjects had to be dropped from the study.

The remaining 30 pairs of matched children (66.7%)

constituted the study sample. Forty of the children (66.7%)

were from Alachua County, and 20 children (33.3%) were from

neighboring Marion County. Two of the mothers were

primiparous (3.3%) and 58 were multiparous (96.7%). Fifty of

the mothers were African American (83.3%) and ten were of

other ethnic origin (16.7%). Sixteen of the women earned a

Hollingshead Index of four (26.7%) while 44 of the women

earned a Hollingshead of five (73.3%). The Hollingshead Index

is a measure of socioeconomic status based on level of

education and employment. Scores range from one to five with

scores of four and five representing the lowest levels of

employment and education. Thirty of the children (50.0%) were

exposed to crack cocaine in utero and 30 were not exposed

(50.0%). Tables 1 and 2 outline basic demographic and








42

covariate characteristics for the sample. For the remainder

of this document, 'sample' will refer to the 60 children in

this study which was completed in the context of the larger

longitudinal study as previously described.

The children in the sample (n = 60) included 34 girls

(56.7%) and 26 boys (43.3%). Sixteen children were placed in

a home other than that of their biological mother (26.7%).

Average birth weight was 3326 grams (SD = 413 grams), and mean

age at time of testing was 36.8 months (SD = 1.9 months). The

nonexposed group of children was comprised of 18 girls (60.0%)

and 12 boys (40.0%). Only two children were in foster care

(6.7%). Average birth weight in nonexposed children was 3355

grams (SD = 401.469 grams), and mean age at time of testing

was 37.3 months (SD = 1.446 months). The exposed group was

comprised of 16 girls (53.3%) and 14 boys (46.7%). Fourteen

of the children were in foster placement (46.7%). Average

birth weight in exposed children was 3297 grams (SD = 431

grams), and mean age at time of testing was 36 months (SD =

2.3 months).

In the sample, maternal depression scores ranged from one

to 43 (M = 19.7, SD = 9.8) with 16 being the clinical cutoff

score. Home Environment total scores ranged from 18 to 52 (M

= 37.0, SD = 8.4) with a normative sample M = 37.54, SD =








43

10.41. Alcohol use, calculated as average number of 4-ounce

drinks per day over three trimesters, ranged from zero to 3.66

(H = .253, SD = .649). In the nonexposed group, average

maternal depression score was 20.9 (SD = 9.5), and average

home inventory score was 36.5 (SD = 8.1). Alcohol use in the

nonexposed groups averaged .2667 drinks per day (SD = .450).

In the exposed group, average maternal depression score was

18.4 (SD = 10.1), and average home inventory score was 37.6

(SD = 8.7). Mean drinks per day in the exposed group was

.9333 (SD = .640). For purposes of further analyses, alcohol

was coded as a dichotomous variable, either present or absent

in the prenatal period.



Parent Report Measures

Vineland Adaptive Behavior Scales

The Vineland (Sparrow, Balla, & Cicchetti, 1984) is a

nationally normed measure of adaptive behavior for children

birth to adulthood and assesses competency in four skill

areas: Socialization, Communication, Motor Skills, and Daily

Living. The Survey Form used in this study consists of 297

questions administered to the primary caregiver. Items to be

answered on a range from 'yes, usually' to 'don't know'

include such examples as "Uses sentences of four or more











words" or "Feeds self with spoon without spilling."

Administration time is approximately 15 minutes. Split-half

coefficients and test-retest reliabilities for the Adaptive

Behavior Composite are .94 and .95 respectively and factors

reportedly load in appropriate subdomains of each of the four

areas. Scores are percentile ranks, means, and standard

deviations.



Connors Parent Rating Scale

The Connors (1990) is a parent rating report of

hyperactive and attentional problems in children aged three to

17. It was normed on 383 children and consists of 48

questions which load into six factors including Conduct

Problem, Learning Problem, Psychosomatic, Hyperactivity Index,

Impulsive-Hyperactive, and Anxiety. Higher scores indicate

increased problems in the tested area. Administration time is

approximately 15 minutes. Factors loadings of the CPRS-48

range form .41 to .82 and appear stable over time. Scores are

percentile ranks, means and standard deviations.



Evbera Child Behavior Inventory (ECBII

The ECBI (Eyberg, 1992) is a parent report screening

measure for disruptive behaviors in children ages two through










16. It includes an Intensity Scale, which indicates how often

the behaviors presently occur, and a Problem Scale, which

identifies the specific behaviors that are currently problems

for the caregiver. There are a total of 36 items.

Clinically, a Problem Scale score of 11 or greater and an

Intensity Scale score of 127 or greater are used as cutoffs

for disordered conduct. The normative sample was 798 parents

of children drawn from six pediatric health care settings

sampled so as to approximate the demographic composition of

the Southeast. Split-half correlations for both scales were

.93. Administration time is approximately 5 minutes.



Behavioral Observation Measure

Dyadic Parent-Child Interaction Coding System-II (DPICS-II1

The DPICS-II (Eyberg, Bessmer, Newcomb, Edwards, &

Robinson, 1994) is a direct, objective observation measure to

discriminate normal from conduct-disordered child/parent

interactions. The system provides a measure of both child and

parent behaviors and was originally normed on 20 families

referred for treatment of a conduct-problem child and 22

families without such a conduct-problem child. Problem

behaviors include disobedience, aggression, destructiveness,

or hyperactivity. Reliability for DPICS-II by behavior ranged










from 38% to 99% (Bessmer, 1993). The DPICS-II includes three

situations that may be coded. For this study, we selected

five minutes of child directed play between caregiver and

child (CDI) and five minutes of parent-directed play (PDI).

Administration and scoring criteria for the standardized

protocol were not altered. The DPICS-II code categories

selected for observation were child behaviors that others have

suggested might differentiate between cocaine-exposed children

and nonexposed children. Exposed children may demonstrate

conduct disordered behaviors. Alternately, the exposed child

may be withdrawn in interactive situations. To capture a

potential range of child behaviors, both positive and

negative, included here were: 1) yell and whine, 2) laugh, 3)

destructive, 4) physical negative, and 5) physical positive.

These variables were measured by frequency over two, five

minute periods. Compliance data were drawn from several

variables as follows: 1) compliance or 2) noncompliance to a

3) direct or 4) indirect command, or 5) no opportunity for

compliance. Total frequency of commands were calculated for

each child. The five compliance variables were also combined

to derive a percentage representing the rate of compliance,

noncompliance, and no opportunity for compliance for each

child. These behaviors are defined in detail in the DPICS-II










manual (Eyberg et al., 1994). Toys used in the play situation

were standard and included: 1) large primary-color blocks, 2)

a farm, and 3) and a gender-neutral stuffed animal.



Procedures

Parents and children were seen during their scheduled

visits as part of the ongoing, longitudinal project. Informed

consent was obtained from parents at the intake of the

original project approximately three years previously.

Maternal demographic and perinatal medical information was

obtained from interviews conducted prenatally and/or at the

time of birth. Parent report measures for the present study

were incorporated into the existing project interview

scheduled as part of a home visit when the children were three

years old. The play observation for this study was scheduled

separately from the parent interview as part of the three-year

clinic visit. The observation period was situated in the test

period before procedures considered aversive (e.g., physical

examination) and after generally nonaversive testing (e.g.,

Bayley II). Administration of materials was standard for each

caregiver/child dyad.

Observation periods were videotaped in a room at either

the Public Health Department or delivery hospital. The video








48

camera was concealed behind a three sided screen to limit

distractions and to maximize play opportunity. The three toys

were placed in the center of the floor before each video

session began. Verbal instructions to the parent for the free

play or child-directed play, and for the parent-directed

portion of taping were standardized and followed the format

provided with the DPICS-II manual.

Tapes were scored by two advanced graduate students

trained to reliability on pilot tapes using the DPICS-II

(Eyberg et al., 1994) scoring criterion. Examiners were

blinded to cocaine status to reduce possible bias.









ANALYSES


Descriptive Analyses

Means and standard deviations were calculated for the

total sample for dependent variables including: 1) the

Connors' (Connors, 1990) subscales Conduct Problem, Learning

Problem, Psychosomatic, Impulsive-Hyperactive, Anxiety, and

Hyperactivity Index, 2) ECBI (Eyberg & Colvin, 1994) Problem

and Intensity scales, 3) the Vineland (Sparrow et al., 1984)

subscales Socialization, Communication, Daily Living, and

Motor Skills, and 4) DPICS-II (Eyberg et al., 1994) variables.

For the DPICS-II, ten behaviors were chosen for basic

frequency observation. For purposes of higher level analyses,

four summary variables were derived from DPICS-II compliance

data. Normality assumptions for dependent variables were

tested by graphing and computing skewness scores. Covariates

were included in the initial analyses between groups only for

derived summary scores.



Hypothesis 1

To test whether the performance of the entire sample is

depressed compared to age-based norms, scores on well normed

measures were compared to age-based norms using Welch's V for










unequal n and Studentized Maximum Modulus (SMM) tables for

familywise error rates to control for multiple comparisons at

the .05 significance level. Dependent variables were grouped

conceptually for comparison to norms as follows: 1) the six

subscales of the Connors (Connors, 1990) including Conduct

Problem, Learning Problem, Psychosomatic, Impulsive-

Hyperactive, Anxiety, and Hyperactivity Index, 2) the ECBI

(Eyberg & Colvin, 1994) Problem and Intensity scales, and 3)

the four subscales of the Vineland (Sparrow et al, 1984)

including Socialization, Communication, Daily Living, and

Motor Skills.

The ten DPICS-II basic observation frequency variables

were broken into two groups for analysis. The first group

included compliance information including frequency of direct

and indirect commands, compliance, noncompliance, and no

opportunity for compliance. The second group consisted of

child behaviors including frequency of laugh, yell-whine,

physical positive, physical negative, and destructive. Scores

were compared to normative data using Welch's V for unequal n

and Studentized Maximum Modulus (SMM) tables for familywise

error rates. Comparisons to normative groups were not

performed for DPICS-II derived summary scores as norms are not

available.











Hypothesis 2

It was hypothesized that the performance of the cocaine-

exposed group would be impaired compared to the nonexposed

group in terms of behavior as measured by parent report on the

six subscales of the Connors (Connors, 1990). After graphing

to inspect relationships between scales, MANCOVA with

Hotelling's T2 was used to analyze between group differences.



Hypothesis 3

It was hypothesized that the performance of the cocaine-

exposed group would be impaired compared to the nonexposed

group in terms of behavior as measured by parent report on the

two subscales of the ECBI (Eyberg & Colvin, 1994). After

graphing to inspect relationships between scales, MANCOVA with

Hotellings T2 was used to analyze between group differences.



Hypothesis 4

It was hypothesized that the performance of the cocaine-

exposed group would be depressed compared to the nonexposed

groups in terms of adaptive behaviors as measured by the four

subscales of the Vineland (Sparrow et al., 1984) by parent

report. After graphing to inspect relationships between








52

scales, MANCOVA with Hotellings T2 was used to analyze between

group differences.



Hypothesis 5

It was hypothesized that the performance of the cocaine-

exposed group would be depressed compared to the nonexposed

groups in terms of observed disordered behaviors as measured

on the DPICS-II (Eyberg et al., 1994). Because data for

individual behaviors are observed at very low frequencies, the

basic ten observed behaviors were compared for between group

differences using t-tests. The four derived summary scores

including Total Commands, Percent Compliance, Percent

Noncompliance and Percent No Opportunity for Compliance, were

treated like parent report variables. After graphing to

inspect relationships between scales, MANCOVA with Hotellings

T2 was used to analyze between group differences.









RESULTS


Descriptive Analysis

First, means and standard deviations were calculated for

the total sample for parent report dependent variables

including: 1) the Vineland (Sparrow et al., 1984) subscales

Communication, Daily Living, Motor Skills, and Socialization,

2) the Connors' (Connors, 1990) subscales Conduct Problem,

Learning Problem, Psychosomatic, Impulsive-Hyperactive,

Anxiety, and Hyperactivity Index, and 3) the ECBI (Eyberg &

Colvin, 1994) Intensity and Problem scales.



Parent Report Measures

Normality assumptions for parent report dependent

variables were then tested by graphing and computing skewness

scores. One subscale of the Connors' scale (Psychosomatic)

and one subscale of the Vineland (Daily Living) were found to

be skewed. These were corrected for further analyses using

log transformations. All other dependent variables on

standardized parent report measures were found to approximate

normality. Data were missing for two children on all Connors

subscales, and for one child on all Vineland subscales. Data

were missing for four children on the ECBI subscales. Missing











data were substituted with group means to allow analysis of

these variables to be performed with equal sample size.



Observational Measures

Second, means and standard deviations were calculated for

the total sample for the ten basic frequency observational

variables including: 1) DPICS-II (Eyberg et al., 1994) five

compliance scores including direct and indirect commands,

compliance, noncompliance and no opportunity to comply, and

child behaviors including yell and whine, laugh, destructive,

physical negative, physical positive, and 2) DPICS-II derived

variables as described previously including commands,

compliance, noncompliance, and no opportunity to comply.

It was determined a'priori' that the ten DPICS-II

observational variables would be consolidated to derive

various scores because the frequencies of target behaviors

tended to be low. Thus, for the present analysis we were not

concerned with testing normality assumptions for the ten

individual observational scales. However, normality

assumptions for consolidated observational dependent variables

including Total Commands, Percent Compliance, Percent

Noncompliance, and Percent No Opportunity to comply were

tested by graphing and computing skewness scores.











The Total Commands score was derived by adding together

total direct command comply, total direct command noncomply,

total direct command no opportunity to comply, total indirect

command comply, total indirect command noncomply, and total

indirect command no opportunity to comply. Percent Compliance

was the sum of total direct command comply and total indirect

command comply divided by the sum of total direct command

comply, total indirect command comply, total direct command

noncomply, total indirect command noncomply, total direct no

opportunity to comply, and total indirect no opportunity to

comply. Percent Noncompliance was the sum of total direct

command noncomply and total indirect command noncomply divided

by the sum of total direct command noncomply, total indirect

command noncomply, total direct command comply, total indirect

command comply, total direct command no opportunity to comply,

and total indirect command no opportunity to comply. Finally,

Percent No Opportunity to Comply was the sum of total direct

command no opportunity and total indirect command no

opportunity divided by total direct command no opportunity,

total indirect command no opportunity, total direct command

comply, total indirect command comply, total direct command

noncomply, and total indirect command noncomply.











Of the four derived DPICS-II observational scales, only

no opportunity to comply was found to be moderately skewed.

This scale was corrected for subsequent analyses using a log

transformation. The remaining three scales, command, comply,

and noncomply, were found to approximate normality. Data were

missing for six of the children from the nonexposed group.

Because data frequency was low on observational scales, it was

considered inappropriate to substitute missing data with group

means, and analysis of these variables was performed with

unequal sample size.



Reliability of Observational Measures

Interrater reliability calculations were performed for

two raters on DPICS-II behavioral variables for both child

directed interactions and parent directed interactions (Tables

3 and 4). In general, interrater reliability was very strong

using Pearson's Product Moment values ranged from r = .84 to

1.00 across the ten basic frequency observation variables in

two situations (parent and child directed play interactions).

Pearson's correlations can be inflated or deflated based on

the frequency of occurrence. Thus, it was decided to

determine interrater reliabilities using agreement of

occurrence, agreement of nonoccurrence, and finally the mean










of agree occur and agree nonoccur for low frequency data

measured in discrete time intervals (Page & Iwata, 1986).

Percent agreement for occurrence was based on summing the

agreement of occurrence across subjects and dividing by the

agreement of occurrence plus disagreements across subjects.

Percent agreement for nonoccurrence was based on summing the

agreement of nonoccurrence divided by the agreement of

nonoccurrence plus disagreements across subjects. In the

child directed interactions, agreement of a given behavior

ranged from 89.10% to 100.00%. Similarly, agreement of

occurrence of a given behavior in the parent directed

interaction ranged from 85.74% to 100.00%. In summary,

interrater reliabilities were strong across all behaviors.



Initial Between-Groups Analyses

Demographic Between Groups Differences

Basic demographic match criteria variables measured on a

nominal or ordinal scale were analyzed for significant

differences between the exposed and nonexposed groups using

Fisher's Chi-Square with the Yates Correction for small sample

sizes. Demographic variables included parity, race, county,

and Hollingshead Index. Due to matching upon entry to the

study and limited variability, it is not surprising that no








58

significant differences were found between groups on these

variables. Similarly, no significant differences on interval

scale data using t-tests were found on basic child

characteristics including birth weight and age at testing

(Table 1).



Covariate Between Groups Differences

Covariates measured on a nominal or ordinal scale were

also analyzed for significant differences between the exposed

and nonexposed groups using Fisher's Chi-Square with the Yates

Correction for small sample sizes. Covariates included child

gender, foster care, and alcohol use where alcohol use was

converted to an ordinal variable to control for outliers. No

differences were found between groups in terms of gender.

However, children exposed to crack were significantly more

likely to be in foster care than children in the nonexposed

group with X2(1) = 10.31, R = .0013. Similarly, children who

were exposed to crack in utero were more likely to be exposed

to alcohol than nonexposed children with X2(1) = 13.08, p =

.0003.

Remaining covariates measured on an interval scale were

analyzed for significant differences between exposed and

nonexposed groups using t-tests. These covariates included










maternal depression at the three year visit and HOME (Caldwell

& Bradley, 1984) inventory total score at the time of the

three year visit. No differences were found in terms of

maternal depression or HOME inventory total scores at three

years of age using exposure status as the independent variable

(Table 2).



Group Differences using Covariates as Independent Variables

Notably, significant differences were found on parent

report measures when the groups were defined by covariates.

For example, groups were defined by maternal depression based

on provided clinical cutoffs (Radloff, 1977), the low group

with scores of 16 or less (n = 21) and a the high group with

scores over 16 (n = 39) representing elevated depression. The

elevated maternal depression group demonstrated more conduct

problems t = -2.57, p < .013, learning problems t = -3.16, R

< .003, hyperactivity L = -2.55, p < .014, and increased

intensity of problem behaviors f = -2.42, p < .019 (Table 5).

Similarly, HOME inventory scores were divided based on a

provided normative mean of 37 (Caldwell & Bradley, 1984). The

first group (low) was comprised of scores of 37 or less (n =

29) and a second group (high) was comprised of scores of

greater than 37 (n = 31) representing more enriched











environments. Findings indicated that the group representing

children from less enriched environments had more conduct

problems f = 2.57, p < .013, learning problems t = 2.25, R <

.028, worse socialization t = -2.75, g < .008, and reduced

communication skills t = -2.07, p < .043 compared to children

from relatively more enriched environments (Table 6).

When groups were defined by foster care status, children

who were with their biological mothers (n = 44) tended to have

significantly more conduct problems t = 2.25, R < .028 than

children living away from their biological mothers (n = 16)

(Table 7). Further, when groups were defined by child gender,

boys (n = 26) were significantly more likely to display

learning problems t = -2.38, R < .021, and have poorer

communication skills than girls = 2.05, g < .045. Girls (n

= 34) were significantly more likely to feel anxious t = 2.17,

p < .034 than boys (Table 8). Interestingly, although alcohol

use was significantly different between exposed and nonexposed

groups, no significant differences were found on any parent

report measure using alcohol as the independent variable

(Table 9).

In terms of observational measures and covariates, no

between groups differences were found on the four derived

DPICS-II scores when groups were defined by either maternal











depression (exposed n = 36, nonexposed n = 18) or HOME

inventory (exposed n = 29, nonexposed = 25). Further, no

differences were found on derived DPICS-II observational

scales when groups were defined by foster care status (exposed

n = 15, nonexposed n = 39), gender (exposed n = 24, nonexposed

n = 30), or alcohol use (exposed n = 31, nonexposed n = 29)

(Table 9). Together, these results suggest that chosen

covariates may have complex relationships with outcome on

behavioral measures, and that their selection for inclusion as

covariates was warranted.



Initial Correlations

Several additional analyses were done to clarify the

nature of relationships between covariates and dependent

variables, between pairs of covariates, and between pairs of

dependent variables. It should be noted that correlations

with nominal, ordinal, and interval data were done with

Pearson's Product Moment rather than point biserial

correlations as the outcome is identical (Weinberg & Goldberg,

1979).








62

Covariates and Dependent Variables

The relationships between covariates and dependent

variables are presented in Table 10. Notably, elevated

maternal depression was significantly related to increased

conduct disorder r = .3358, p < .01, learning problems r =

.3977, p < .01, and hyperactivity r = .3534, p < .01 on the

Connors. Maternal depression was also related to the problem

scale of the ECBI r = .3387, p < .01, again suggesting that

problem behaviors are more strongly associated with higher

levels of maternal depression. Similarly, a more enriched

home environment as measured by HOME inventory was

significantly related to better communication i = .3488, p <

.01. Interestingly, no significant relationships were noted

between the ten basic and four derived DPICS-II variables and

five measured covariates. It should be noted that a more

conservative significance level of .01 was chosen in order to

partially control spurious relationships that may results from

multiple corrrelations.



Covariates

In general, relationships between pairs of covariates

were insignificant using a two-tailed analysis. However,

caregivers who were raising foster children reported










significantly less depression r = -.3363, p < .01 than mothers

raising their biological children (Table 11).



Dependent Variables

Not surprisingly, when assessing relationships between

pairs of dependent variables across the three parent report

instruments (Connors, ECBI, and Vineland), behavioral

subscales tended to vary together and adaptive functioning

subscales tended to vary together. The ECBI problem scale was

significantly related to the Connors conduct problem r =

.5569, p < .001, learning problem r = .5586, p < .001,

impulsivity E = .4915, p < .001, and hyperactivity scales r =

.6089, p < .001. The ECBI Intensity score was also

significantly related to these four Connors subscales with

increased intensity associated with conduct problems r =

.6635, p < .001, learning problems r = .5827, p < .001,

impulsivity r = .6014, p < .001, and hyperactivity r = .6331,

p < .001. Interestingly, the Connors subscale learning

problem was negatively correlated to socialization r = -.4672,

p < .001 and communication r = -.3768, p < .001. In other

words, higher levels of learning problems were associated with

relatively poor socialization and communication. Higher rates

of hyperactivity from the Connors were similarly related to








64

reduced socialization r = -.3343, R < .001. Neither ECBI

score was related to any measured domain of adaptive

functioning (Table 12).



Parent Report and Derived Observational Variables

Of particular interest were relationships between parent

report dependent measures and derived observational measures.

When examining derived DPICS-II scores, results indicate that

noncompliance measured objectively was related to caregiver

reports of higher learning problems r = .3578, p < .01, and

reduced daily living skills I = -.4229, p < .01. Similarly,

higher levels of caregiver perceived socialization were

positively related to compliance r = .4384, p < .01 (Table

12).

In terms of child behaviors, yell was positively related

to conduct problems and psychosomatic behavior with r = .4186,

p < .01 and r = .4399, p < .001 respectively. Finally, child

destructive behavior was significantly associated with conduct

problems with r = .3846, p < .01. It should be noted that a

more conservative significance level of .01 was chosen in

order to partially control spurious relationships that may

result from multiple correlations. Again, preliminary between











group and correlational findings suggest complex relationships

between measured scales (Table 12).



Hypothesis 1

It was hypothesized that the performance of both groups

would be depressed on all measures compared to age-based

norms. Analyses were performed with sample scores and

population norms. To test this, Welch's V for unequal sample

size was calculated comparing total sample scores on dependent

variables to standardized norms. Dependent variables were

grouped conceptually a'priori' and Studentized Maximum Modulus

(SMM) tables for familywise error rate were used to control

for multiple comparisons at the .05 significance level.

Tables 13, 14, and 15 summarize these results.



Connors Parent Rating Scale

The first group was comprised of the six subscales of the

Connors including: Conduct Problem, Learning Problem,

Psychosomatic, Impulsive, Anxiety, and Hyperactivity Indices.

Using the SMM familywise rate for six comparisons, Conduct

Problem was significantly different from the population norm

y = 5.03, R < .01 where the mean for the sample was higher

than the population mean. According to parent report,










learning problems were also more frequent in the sample than

the normative group with y = 2.98, R < .05. Children in the

sample tended to be more impulsive than the normative group

where y = 6.76, R < .01. Similarly, parent report indicated

increased hyperactivity in the total sample than in the

normative population where y = 5.30, p < .05 (Table 13).



Eybera Child Behavior Inventory

The second conceptual grouping was comprised of the two

ECBI scores including: Problem and Intensity. Using the SMM

familywise rate for two comparisons, the total sample was

significantly more likely to display a high intensity of

problem behaviors than the normative group where y = 2.637, p

< .05. Despite the high intensity of problem behaviors,

caregivers did not endorse items indicating that the behavior

posed a significant problem in the home (Table 13).



Vineland Adaptive Behavior Scales

The third group was comprised of the four subscales of

the Vineland related to adaptive functioning: Socialization,

Communication, Daily Living, and Motor Skills. Using the SMM

familywise rate for four comparisons, none of the sample










scores on any of these measured domains were significantly

different from provided norms (Table 13).



Observational Ratinas

The final two conceptual groupings were comprised of

basic objective observational data including: 1) direct

commands, indirect commands, compliance, noncompliance and no

opportunity to comply, and 2) laugh, yell, physical positive,

physical negative, and destructive behavior over the total ten

minute play period in two situations including both child-

directed and parent-directed play. These results are

summarized on Tables 14 and 15.

Using the SMM familywise rate for five comparisons,

significant differences were noted in sample scores in the

child-directed interaction when compared to normative data.

Specifically, the caregivers in the present sample tended to

have a higher rate of direct commands than either the clinic

referred or nonreferred normative group where y = 8.61, p <

.01 and v = 9.35, p < .01 respectively. The rate of indirect

commands was similarly elevated in the sample compared to

clinic referred and nonreferred normative groups where y =

4.01, p < .01 and y = 3.37, p < .01 respectively. Not

surprisingly given the higher rates of commands, the rate of








68

compliance was also significantly higher in the sample

compared to both the clinic referred and nonreferred reference

groups with y = 5.47, p <.01 and y = 6.61, p < .01

respectively. Noncompliance was likewise elevated in the

sample group compared to referred and nonreferred group with

M = 3.24, p < .01 and y = 3.93, p < .01 respectively.

Finally, no opportunity for compliance occurred at

significantly higher rates in the study sample compared to

normative groups including clinic referred where y = 7.96, p

< .01 and nonreferred where y = 8.12, E < .01.

No significant differences between the study sample, and

referred and nonreferred normative groups were noted on child

behaviors including yell, physical negative or destructive

using the SMM familywise rate for three comparisons. It was

not possible to test differences in the rate of laugh and

physical positive as normative data were not available for

these two variables. Results suggest that overall rates of

both indirect and direct commands were elevated in the present

study sample which in turn results in higher rates of

compliance, noncompliance, and no opportunity to comply (Table

14).

Using the SMM familywise rate for five comparisons,

significant differences were found between the study sample










and referred and nonreferred normative groups in the parent-

directed interactions (Table 15). Specifically, caregivers in

the sample group displayed significantly more direct commands

than either the referred or nonreferred normative groups with

Y = 4.91, p < .01 and y = 7.64, p < .01 respectively.

Interestingly, no differences were noted in the rate of

indirect commands in the parent-directed interaction. As in

the child-directed interactions, the rate of compliance was

significantly higher in the sample group during the parent-

directed interactions than in either the referred or

nonreferred normative groups with y = 6.13, R < .01 and v =

5.28, p < .01 respectively. Similarly, the study sample

tended to have a significantly higher rate of no opportunity

to comply than either the referred or nonreferred comparative

group with v = 3.76, p < .01 and y = 5.55, 2 < .01

respectively.

Alternately, no significant differences were noted in the

rate of noncompliance between the study sample and normative

groups. It was not possible to test between groups

differences for child behaviors for laugh, yell, physical

positive, physical negative or destructive due to either the

absence of normative data or variance in the normative group

(Table 15). Finally, it should be noted that the significance










for certain comparisons may have been diminished by the use of

familywise comparisons which tends to reduce power.



Hypothesis 2

It was hypothesized that behavioral difficulties would be

more pronounced in the exposed group. Descriptive data for

the exposed and nonexposed children are summarized on Table

16. To test this hypothesis, a multivariate analysis of

covariance was performed on the six dependent variables that

comprise the Connors: Conduct Problem, Learning Problem,

Psychosomatic, Impulsive-Hyperactive, Anxiety and

Hyperactivity. In the initial analysis, adjustment was made

for five covariates including alcohol, maternal depression,

foster care, HOME inventory, and child gender. The

independent variable was exposure status (exposed to crack

versus not exposed). SPSS' (Norusis, 1987) MANCOVA was used

for the analyses with the sequential adjustment for

nonorthogonality.

With the use of Hotelling's criterion, the dependent

variables were significantly related to the combined

covariates, with E(30,232) = 1.70, p < .05. To investigate

more specifically the power of the covariates to adjust the

dependent variables, multiple regressions were run with








71

covariates acting as multiple predictors. None of the

covariates provided adjustment to the Conduct Problem or

Anxiety subscales of the Connors. However, two of the five

covariates, child gender and maternal depression, provided

significant adjustment to Learning Problems with beta values

of .34 t(53) = 2.75, p < .05 and .41 t(53) = 3.27, i < .05

respectively. Maternal depression also provided adjustment to

Psychosomatic with a beta value of .30 significantly different

from zero where t(53) = 2.13, p < .05. Similarly maternal

depression provided significant adjustment to Impulsivity and

Hyperactivity with beta values at .36 t(53) = 2.60, p < .05,

and .36 t(53) = 2.67, p < .05 respectively. For none of the

dependent variables did alcohol, foster care, or HOME

inventory provide significant adjustment. These covariates

were dropped from further analyses.

The effects of exposure status on the dependent variables

after adjustment for significant covariates were investigated

in univariate and stepdown analysis. Dependent variables were

entered in the following order: Conduct Problems, Learning

Problems, Psychosomatic, Impulsivity, Anxiety, and

Hyperactivity Indices. In the stepdown analysis, each

dependent variable was analyzed in turn with higher priority

dependent variables treated as covariates and with the highest










priority dependent variables tested in a univariate ANOVA.

Conduct Problem was chosen as the highest priority dependent

variable since the literature suggests that this may be a

marker of prenatal cocaine exposure.

The results of this analysis are shown in Table 17. The

Hotelling's criterion for the combination of dependent

variables was not significant with E(6,51) = 1.26, p > .05.

Findings indicate that there are no differences between

exposed children and nonexposed children on this combination

of variables by parent report. Even in a univariate context

at the E < .05 level, there are no significant differences

between exposed and nonexposed children on these variables.

Specifically, no significant differences were found between

groups for conduct problem F((1,56) = 1.06, p > .05, learning

problem (1,55) = 1.22, R > .05, psychosomatic E(1,54) = .74,

p > .05, impulsivity (1,53) = .55, E > .05, anxiety (1,52)

= 3.25, p > .05, or hyperactivity (1,51) = .71, p > .05.



Hypothesis 3

As before, it was hypothesized that behavioral

difficulties would be more pronounced in the exposed group.

Descriptive data for the exposed and nonexposed children are

summarized on Table 16. To test this, a multivariate analysis











of covariance was performed on the two dependent variables

that comprise the ECBI: Problem and Intensity Scales. In the

initial analysis, adjustment was made for five covariates

including alcohol, maternal depression, foster care, HOME

inventory, and child gender. The independent variable was

exposure status (exposed to crack versus not exposed to crack

in utero). SPSS' MANCOVA was used for the analyses with the

sequential adjustment for nonorthogonality.

With the use of Hotelling's criterion, the dependent

variables were not significantly related to the combined

covariates with (10,102) = .921, p > .05. Thus, all five

covariates were dropped from further analyses. The effects of

exposure status on the dependent variables without inclusion

of covariates were investigated in univariate and stepdown

analysis using SPSS' MANOVA. Dependent variables were entered

in the following order: Problem Scale and Intensity Scale.

In the stepdown analysis, each dependent variable was analyzed

in turn with higher priority dependent variables treated as

covariates and with the highest priority dependent variables

tested in a univariate ANOVA. The Problem Scale was chosen as

the highest priority dependent variable as the literature

suggests that behavioral problems may result from prenatal

cocaine exposure.










The results of this analysis are shown in Table 18. The

Hotelling's criterion for the combination of dependent

variables was not significant with E(2,57) = 1.32, R > .05.

Findings indicate that there are no differences between

exposed and nonexposed children on this combination of

variables by parent report. Inspection of univariate and

stepdown results at the a < .05 level reveal no significant

differences between exposed and nonexposed children on these

variables. Specifically, groups were not significantly

different on either the ECBI problem scale (1,58) = 2.63, Q

> .05, or intensity scale (1,57) = .05, R > .05.



Hypothesis 4

It was also hypothesized that adaptive difficulties would

be more pronounced in the exposed group. Descriptive data for

the exposed and nonexposed children are summarized on Table

16. To test this, a multivariate analysis of covariance was

performed on the four dependent variables that comprise the

Vineland: Socialization, Communication, Daily Living, and

Motor Skills. In the initial analysis, adjustment was made

for five covariates including alcohol, maternal depression,

foster care, HOME inventory, and child gender. The

independent variable was exposure status (exposed to crack










versus not exposed to crack in utero). SPSS' MANCOVA was used

for the initial analyses with the sequential adjustment for

nonorthogonality.

With the use of Hotelling's criterion, the dependent

variables were not significantly related to the combined

covariates, with (20,194) = 1.55, R > .05. Thus, all five

covariates were dropped from further analyses. The effects of

exposure status on the dependent variables without inclusion

of covariates were investigated in univariate and stepdown

analysis using SPSS" MANOVA. Dependent variables were entered

in the following order: Socialization, Communication, Daily

Living, and Motor Skills. In the stepdown analysis, each

dependent variable was analyzed in turn with higher priority

dependent variables treated as covariates and with the highest

priority dependent variables tested in a univariate ANOVA.

Socialization was chosen as the highest priority dependent

variable as the literature suggests that social relatedness

may be disturbed by prenatal cocaine exposure.

The results of this analysis are shown in Table 19. The

Hotelling's criterion for the combination of dependent

variables was not significant with E(4,50) = 1.17, a > .05.

These findings suggest that there are no differences between

exposed and nonexposed children on this combination of











variables by parent report. Inspection of univariate and

stepdown results at the p < .05 level likewise reveal no

significant differences between exposed and nonexposed

children on these variables. Specifically, no significant

between groups differences were noted for socialization

E(1,58) = .40, p > .05, communication (1,57) = .51, p > .05,

daily living (1,56) = 1.12, R > .05, or motor skills E(1,55)

= 1.67, R > .05.



Hypothesis 5

Finally, it was hypothesized that disordered behaviors by

objective observation would be more pronounced in the exposed

group. Descriptive data for the exposed and nonexposed

children are summarized on Table 16 collapsed across parent-

and child-directed interactions. To test this, a multivariate

analysis of covariance was performed on the four dependent

variables derived from compliance data on the DPICS-II: Total

Commands (frequency data), and Percent Comply, Percent

Noncomply, and Percent No Opportunity to comply (percentages).

The ten basic DPICS-II frequency observation variables were

not tested as they were not normally distributed. In the

initial analysis, adjustment was made for five covariates

including alcohol, maternal depression, foster care, home










inventory, and child gender. The independent variable was

exposure status (exposed to crack versus not exposed to crack

in utero). SPSS" MANCOVA was used for the initial analyses

with the sequential adjustment for nonorthogonality.

With the use of Hotelling's criterion, the dependent

variables were not significantly related to the combined

covariates, with E(20,142) =.229, 1 > .05. Thus, all five

covariates were dropped from further analyses. The effects of

exposure status on the dependent variables without inclusion

of covariates were investigated in univariate and stepdown

analysis using SPSS' MANOVA. Dependent variables were entered

in the following order: comply, noncomply, no opportunity to

comply, and commands. In the stepdown analysis, each

dependent variable was analyzed in turn with higher priority

dependent variables treated as covariates and with the highest

priority dependent variables tested in a univariate ANOVA.

Comply was chosen as the highest priority dependent variable

as the literature suggests that child compliance may be

disturbed by prenatal cocaine exposure.

The results of this analysis are shown in Table 20. The

Hotelling's criterion for the combination of dependent

variables was not significant with E(4,42) = .876, p > .05.

These findings suggest that there are no differences between








78

exposed and nonexposed children on this combination of

variables by parent report. Inspection of univariate and

stepdown results at the R < .05 level likewise reveal no

significant differences between exposed and nonexposed

children on these variables. Specifically, significant

differences were not found between groups for commands E(1,45)

= .26, R > .05, comply (1,44) = .86, p > .05, noncomply

(1,43) = .11, R > .05, or no opportunity (1,42) = .02, R >

.05.










Table 1


Demographic Characteristics of the Total Sample


Non-
MATERNAL Exposed Exposed % df p-value


County"
Alachua 20 20 66.7
Marion 10 10 33.3 1 1.0000

Hollingshead"
Four 8 8 26.7
Five 22 22 73.3 1 1.0000

Race"
Black 25 25 83.3
Non-Black 5 5 16.7 1 1.0000

Parity'
Prima 1 1 3.3
Multi 29 29 96.7 1 1.0000


Non-
CHILDREN Exposed Exposed df p-value

Birthweight (gm)b 3355 3297 56 .600


Test Age (months)' 37.3 36 56 .056

* Using Fisher's Chi-Square with the Yates correction for small sample sizes
bUsing Student's t-test
* = < .05










Table 2


Covariate Characteristics of the Total Sample


NOMINAL/ Non-
ORDINAL Exposed Exposed % df p-value


Child Gender
Female 18 16 56.7
Male 12 14 43.3 1 .7945


Fostercare
Yes 2 14 26.7
No 28 16 73.3 1 .0013*


Alcohol
Yes 8 23 51.7
No 22 7 48.3 1 .0003*


Non-
INTERVAL Exposed Exposed t p-value


Maternal 18.4 20.9 .99 .326
Depression


HOME 37.6 36.5 -.52 .605
Inventory


* Using Fisher's Chi-Square with the Yates correction for small sample sizes
b Using Student's t-test
*= p < .05










Table 3
Reliability of Video Coding of Selected DPICS-II Categories in the Child-Directed
Interaction

%Agree %Agree
r Occur Nonoccur Disagree Mean



Direct Command .99 95.03 96.16 .26 95.59

Indirect Command .98 90.18 95.03 .29 92.60

Compliance .97 93.56 97.44 .12 95.50

Noncompliance .84 94.19 97.00 .09 95.59

No Opportunity for Compliance .99 89.32 95.14 .29 92.23

Laugh 1.00 100.00 99.68 .02 99.84

Yell 1.00 100.00 100.00 .00 100.00

Physical Positive .89 89.10 98.90 .04 94.00

Physical Negative 1.00 100.00 100.00 .00 100.00

Destructive .98 100.00 100.00 .00 100.00


Percent agreement for occurrence is based on summing the agreements across subjects
and dividing by the agreements plus disagreements across subjects. Similarly, percent
agreement for nonoccurrence is based on summing agreements across subjects and
dividing by the agreements plus disagreements across subjects.










Table 4

Reliability of Video Coding of Selected DPICS-II Categories in the Parent-Directed
Interaction

%Agree %Agree
I Occur Nonoccur Disagree Mean


Direct Command .99 90.63 96.16 .24 93.39

Indirect Command .95 85.74 96.35 .38 91.04

Compliance .97 91.88 96.68 .18 94.14

Noncompliance .98 90.20 98.40 .11 94.34

No Opportunity for Compliance .98 86.85 93.12 .35 89.98

Laugh .99 100.00 99.62 .02 99.81

Yell .98 100.00 100.00 .00 100.00

Physical Positive .99 96.15 100.00 .00 98.07

Physical Negative 1.00 100.00 100.00 .00 100.00

Destructive .97 91.66 100.00 .00 95.83

Percent agreement for occurrence is based on summing the agreements across subjects
and dividing by the agreements plus disagreements across subjects. Similarly, percent
agreement for nonoccurrence is based on summing agreements across subjects and
dividing by the agreements plus disagreements across subjects.










Table 5
Between Group Differences Defined by Maternal Depression

Low High
MATERNAL M M P
DEPRESSION (n = 21) (n = 39)


CONNORS

Conduct 53.4286 63.2564 .013*

Learning 48.3810 58.9744 .003*

Psychosomatic 49.2381 54.4103 .217

Impulsive 55.9524 58.9487 .193

Anxiety 49.9048 51.5128 .502

Hyperactivity 53.5714 62.0769 .014*

ECBI

Problem 7.3810 11.5385 .069

Intensity 107.5238 126.9231 .019*

VINELAND

Socialization 102.6190 96.8462 .108

Communication 102.6190 102.1026 .882

Daily Living 105.4286 104.1026 .775

Motor Skills 97.0952 96.3333 .856

(Continued)










Table 5 Continued


Low High
MATERNAL M M B
DEPRESSION (n = 18) (n = 36)


DPICS-II (Derived Scores)

Total Commands' 46.0256 46.1333 .989

Percent Comply" .3460 .3514 .908

Percent Noncomplyb .0996 .0896 .696

Percent No Opportunityb .5544 .5591 .915

*= p < .05
Frequency data
percentages










Table 6
Between Group Differences Defined by HOME Inventory


Low High
HOME M M 1
INVENTORY (n = 29) (n = 31)


CONNORS

Conduct 64.6552 55.2903 .013*

Learning 59.1379 51.6452 .028*

Psychosomatic 55.2759 50.0968 .195

Impulsive 58.6552 57.1935 .508

Anxiety 52.6897 49.3226 .138

Hyperactivity 62.0690 56.3226 .085

ECBI

Problem 11.7586 8.5161 .140

Intensity 127.0000 113.7097 .095

VINELAND

Socialization 94.2414 103.1935 .008*

Communication 98.8621 105.4839 .043*

Daily Living 104.0000 105.0968 .804

Motor Skills 92.8621 100.0968 .067


(Continued)










Table 6 Continued


Low High
HOME M Mp
INVENTORY (n = 25) (n = 29)


DPICS-II (Derived Scores)

Total Commands' 45.8000 46.2759 .945

Percent Complyb .3316 .3651 .445

Percent Noncomplyb .0947 .0914 .892

Percent No Opportunityb .5738 .5435 .458

*= p < .05
* frequency data
b percentages










Table 7
Between Group Differences Defined by Foster Care Status


No Yes
M M a
FOSTER CARE (n = 44) (n = 16)


CONNORS

Conduct 62.3182 52.9375 .028*

Learning 55.6818 54.1250 .692

Psychosomatic 54.1364 48.3750 .202

Impulsive 58.3864 56.5625 .464

Anxiety 51.6818 48.9375 .287

Hyperactivity 59.8409 57.0625 .465

ECBI

Problem 11.1136 7.2500 .119

Intensity 123.3636 111.2500 .180

VINELAND

Socialization 99.7727 96.3750 .385

Communication 102.8182 100.8125 .593

Daily Living 105.9545 100.7500 .295

Motor Skills 95.6136 99.3125 .413


(Continued)










Table 7 Continued


No Yes
M M p
FOSTER CARE (n = 39) (n = 29)


DPICS-II (Derived Scores)

Total Commands' 46.0256 46.1333 .989

Percent Complyb .3592 .3247 .481

Percent Noncomplyb .0836 .1170 .212

Percent No Opportunityb .5572 .5583 .981

*=p < .05
* frequency data
percentages










Table 8
Between Group Differences Defined by Gender


Girls Boys
M M P
GENDER (n = 34) (n = 26)


CONNORS

Conduct 61.0882 58.1538 .451

Learning 51.8235 59.7692 .021*

Psychosomatic 54.0588 50.6923 .406

Impulsive 57.0294 59.0385 .366

Anxiety 53.0294 48.2308 .034*

Hyperactivity 56.7941 62.1154 .114

ECBI

Problem 10.2059 9.9231 .899

Intensity 121.3824 118.5000 .732

VINELAND

Socialization 100.3235 96.9615 .335

Communication 105.1471 98.5385 .045*

Daily Living 105.7647 103.0000 .535

Motor Skills 95.4118 98.1538 .496

(Continued)










Table 8 Continued


Low High
M M P
GENDER (n = 30) (n = 24)


DPICS-II (Derived Scores)

Total Commands' 44.1333 48.4583 .532

Percent Complyb .3746 .3183 .198

Percent Noncomplyb .0801 .1089 .233

Percent No Opportunityb .5453 .5728 .500

* =p < .05
' frequency data
b percentages










Table 9
Between Group Differences Defined by Alcohol Use


No Use Use
M M P
ALCOHOL (n = 29) (n = 31)


CONNORS

Conduct 63.0000 56.8387 .107

Learning 57.1724 53.4839 .287

Psychosomatic 56.5172 48.9355 .056

Impulsive 59.0000 56.8710 .334

Anxiety 50.7931 51.0968 .895

Hyperactivity 61.8966 56.4839 .105

ECBI

Problem 11.7931 8.4839 .132

Intensity 127.4438 113.2903 .075

VINELAND

Socialization 99.1724 98.5806 .865

Communication 100.4483 104.0000 .284

Daily Living 102.6897 106.3226 .410

Motor Skills 96.0690 97.0968 .797


(Continued)




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EAI0MWZTV_IJBSEC INGEST_TIME 2012-09-24T12:40:21Z PACKAGE AA00011827_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES