Predictors of cocaine use in pregnancy.

Several hundred thousand infants are exposed to cocaine in utero every year in the United States. In an attempt to provide services for women abusing drugs, various programs have been proposed for prenatal drug testing. This study was conducted to assess the sensitivity, specificity, and predictive values of screening programs based on a variety of clinical and historical factors. Between October 1988 and March 1989, 1111 parturients who delivered at an inner-city hospital had urine samples and data collected. The urine, linked by code number to the data, which contained no identifiers, was tested anonymously for illicit drugs. We then assessed the associations of a variety of historical and clinical variables with positive toxicologies. An acknowledged history of drug use or a combination of no prenatal care and cigarette use provided the most valid prediction of cocaine use, with a 59.8% sensitivity, 97.5% specificity, 70% positive predictive value, and 94.9% negative predictive value for positive urine toxicologic examinations. The addition of clinical data such as low birth weight did not improve predictive values significantly. Although the use of easily available historical data allowed correct identification of 60% of parturients who used cocaine, no combination of factors could detect all users. If it is considered beneficial to identify all users at a given point, then urine toxicology screening would be more appropriate than sole reliance on patient history.