The Maternal Lifestyle Study is a large, multisite, longitudinal investigation of PCE being conducted at 4 geographically diverse, collaborating university centers (Wayne State University, University of Tennessee at Memphis, University of Miami, and Brown University). Each participating center had approval for the study from the institutional review board and a certificate of confidentiality from the National Institute on Drug Abuse. Informed consent was obtained from all participants.
Between May 1993 and May 1995, mothers at these centers were enrolled in the study within 24 hours after delivery.3,28,29
Initial screening included the mother’s labor and delivery chart, the newborn admission chart, and a meconium sample. A drug use questionnaire that addressed the mother’s use of nicotine, alcohol, marijuana, cocaine, opiates, and other illicit drugs was given by research staff who were trained and certified in the reliable administration of all of the study interviews. Exposure was determined on the basis of a mother’s admitting cocaine use during pregnancy and/or a positive meconium assay for cocaine metabolites including gas chromatography/mass spectrometry confirmation. Nonexposed children were those who were born to mothers who denied cocaine use, confirmed by negative meconium test results. As previously reported,30
participants for the longitudinal follow-up were recruited at a 1-month visit. The sample included a cohort of exposed infants (n
= 658) who were group matched within site with a group of nonexposed comparison children (n
= 730) by gestational age categories (<32, 33–36, and >36 weeks) and child gender, race, and ethnicity. At the 1-month visit, the biological mother was interviewed for a detailed inventory of her legal and illegal drug use during pregnancy using the Maternal Interview of Substance Use (MISU). Prenatal cocaine use was categorized as high, some, or none on the basis of standard criteria.31
“High” cocaine use referred to ≥3 times per week in the first trimester. Any other use was referred to as “some” cocaine use. MISU reports of the frequency and quantity of these substances per trimester were averaged to produce indices of the number of tobacco cigarettes (heavy use: ≥10 cigarettes per day), the amount of absolute alcohol (heavy use: ≥0.5 oz/day), and the number of marijuana joints (heavy use: ≥0.5 joints per day) consumed during the pregnancy.
Medical characteristics were collected at birth ().
Sample Characteristics According to Attrition
Parent/caregiver age, race, marital status, education level, and Medicaid insurance status were collected at 1 month.
Maternal/caregiver IQ was measured with the Peabody Picture Vocabulary Test at 30 months or 5 years.
Postnatal Substance Use
Postnatal substance use was measured at 4 months, 8 months, then yearly using the Caretaker Inventory of Substance Use, which quantifies frequency and amounts of cocaine, opiates, marijuana, tobacco, and alcohol with the same indices for heavy use as the MISU.
SES was measured with the Hollingshead Index of Social Position32,33
at 1 month, then yearly.
Quality of the home environment was measured with the Home Observation for Measurement of the Environment (HOME)34
at 10 months and 5.5 years.
Depression was measured with the Beck Depression Inventory, a proxy for caregiver depression assessed at 4 months, 30 months, and 5.5 years.
Domestic violence was assessed yearly between 4 and 7 years using a questionnaire (via Caretaker Inventory of Substance Use) on any experience of domestic violence by the caregiver, including physical or sexual abuse.
Child abuse was defined by removal of the child from the home as a result of suspicion of physical and/or sexual abuse or medical examination findings suggestive of physical or sexual abuse.
Changes in primary caregivers (child with biological mother or other caregiver) were recorded at 1 month, then yearly.
Child IQ was measured at 7 years using the Wechsler Intelligence Scale for Children III.
>Each child’s school performance data were gathered directly from school records at age 7 by research staff who were trained for interrater reliability and included the presence of the following for the current school year: (1) individualized education plan (IEP); (2) special education conditions (SEs) including mental retardation, learning disabilities, behavioral/emotional impairment, orthopedic impairment, attention-deficit/hyperactivity disorder, speech/language impairment, and physical or other health impairment; (3) special education classes (SECs) including math, language arts, social studies, science, and conduct (work effort, interpersonal skills); (4) support services (SSs) including occupational therapy, physical therapy, speech and language, counseling/behavior management, reading tutoring/assistance, math tutoring/assistance, and gifted and talented enrichment; and (5) speech and language services (SLSs), which include any service or designation from the previous categories that involves speech, language, or reading. All examiners were blind to prenatal and caregiver substance use.
One-way analysis of variance and x2 were used for continuous and categorical variables, respectively. Logistic regression was used to examine the effect of PCE on 7-year special education outcomes, including the presence of an IEP, SE, SEC, SS, and SLS, controlling for covariates. Covariates included a priori were gender, low birth weight (LBW) (<1500 g), and study site. Additional characteristics were examined in preliminary analyses as candidate covariates ( and ). Candidate covariates that were correlated with PCE and the outcome measures (P ≤ .10) were included in the logistic regression analysis. Measures that met this criterion included small for gestational age, defined as gender-specific weight < 10th percentile for gestational age; ethnicity (white versus nonwhite); low SES at 7 years (Hollingshead Index of Social Position level 5); any primary caregiver change (≥1); poor quality of the home environment as assessed by using the HOME total averaged at 10 months and 5.5 years and recoded into a dichotomous variable (lowest 20% cutoff); Medicaid insurance status; prenatal tobacco use; low caregiver IQ (<85); caregiver marital status at 1 month; maternal age at birth (in years); caregiver depression as indicated by average Beck Depression Inventory assessed at 4 months, 30 months, and 5.5 years and recoded into a dichotomous variable (score of >17, indicating moderate to severe depression); and any child abuse.
Sample Characteristics According to Cocaine Exposure
Regressions for each outcome variable used stepwise elimination of the covariates that made the least contribution to the models and had the least effect on other parameters. The eliminated covariates were then added back to the model in a stepwise manner to test for confounding effects. All covariates in the final models were required to have significant contribution (P < .10) to the outcomes except for the 3 selected a priori (gender, LBW, and study site).
Two sets of logistic regression analyses were performed: with and without low child IQ (score of <85 on the Wechsler Intelligence Scale for Children III) as a covariate. The IEP evaluation process involves a multi-disciplinary approach35
and often uses several diagnostic tools. Because IQ scores are a predictor of educational achievement and are often reviewed when assessing a student for the necessity of an IEP and other special education services, we thought that it was important to investigate the interactions of the hypothesized cocaine effects with low child IQ on special education.