Introduction: Epidemiology studies often rely on maternal self-reports for drug use information, however, the degree of drug use under-reporting among pregnant women is largely unknown. The purpose of this study is to assess the accuracy of self-reports for methadone, buprenorphine, opioids (prescription opioids and heroin), marijuana, benzodiazepines, amphetamines/methamphetamines, and cocaine/crack-cocaine in a population of pregnant women. Methods: Analysis was based on 102 pregnant women enrolled in the 'Biomarkers in Pregnancy Study' (BIPS) cohort at the University of New Mexico. Women attending the UNM Milagro clinic, designated to pregnant women with the current or past history of substance abuse, were enrolled during one of the first prenatal care visits and followed up to term. Self-reported information about drug use was compared with the results of the urine drug screens conducted during the third trimester. Simple kappa and prevalence-and-bias-adjusted kappa coefficients were calculated as measures of agreement. Sensitivity and specificity of self-reports for each drug class were also estimated using urine toxicology screening as the gold standard. In addition, logistic regression was conducted to evaluate the effect of number of toxicology screens on agreement. Results: The mean maternal age of the sample was 26.4 ± 4.9 years and included a large proportion of ethnic minority (78% Hispanics/Latina) and socially disadvantaged (51% < less than high school education and 95% Medicaid-insured) pregnant women. On average, these patients had 4.8 ± 3.0 urine drug screens in the third trimester. For methadone-maintenance therapy, there was a perfect agreement between self-reports and urine screens (k and PABAK =1.0, 100% sensitivity and specificity). Simple kappa coefficients for other classes of drugs revealed varied levels of agreement, however, PABAK coefficients indicated moderate to almost perfect agreement for other classes of drugs. Sensitivity of self-reports was low for all classes of drugs, with marijuana and opioids more acceptable than other classes of drugs. The specificity of self-report was high for classes of drugs. Logistic regression revealed no association between number of toxicology screens and agreement. Discussion: These results indicated that sensitivity of self-reports for all classes of drugs was low with opioids and marijuana more acceptable than other drugs.
First Committee Member (Chair)
Second Committee Member
Level of Degree
College of Pharmacy
Pregnant women--Drug use--New Mexico, Drug abuse surveys--New Mexico--Evaluation.
Garg, Mahek. "Validity of self-reported drug use during pregnancy." (2013). http://digitalrepository.unm.edu/phrm_etds/13