Cocaine in the hair of mother-infant pairs: quantitative analysis and correlations with urine measures and self-report.

Abstract

A sample of mothers who had used crack cocaine while pregnant was evaluated during the postpartum period before random assignment to drug treatment. Mean infant age was 74 days. Paired hair samples were acquired from 63 mothers and 63 infants. Maternal urine was screened for cocaine and benzoylecgonine; three different drug use self-report measures were also collected. Normalizing transformations of all hair and urine data preceded analyses. The initial correlation of mother and infant hair (r = .41, N = 62, P = .001) was strengthened (r = .62, N = 30, P < .0005) by removing from the dataset maternal hair independently judged to be damaged. Damage to hair is associated with certain types of "hair care" products. The damaged maternal hair bore no quantitative relationship to infant hair (r = -.04, N = 30, P = .41). Poor quality hair samples could have been partly anticipated because self-reported use of hair products had a significant contingent relationship to laboratory-judged damage to hair (Z kappa = 2.28, P = .01). Maternal urine benzoylecgonine correlated with maternal hair (r = .41, N = 60, P = .001) and, of course, with maternal urine cocaine (r = .63, N = 60, P < .0005). None of the three self-report measures (use in past 30 days, duration since first use, average regular use) significantly correlated with any of the hair or urine measures. Factor analysis of drug use variables identified three factors possibly representing long-term use, recent use, and some artifact of self-report. In summary, hair analysis may provide a quantitative index of exposure when the hair is not damaged. The amount of self-reported drug use could not be corroborated with analytic measures of hair or urine. Studies in which self-reported cocaine use is a scalar variable should be interpreted cautiously.