Estimating Drug Use Prevalence Among Arrestees Using ADAM Data: An Application of a Logistic Regression Synthetic Estimation Procedure

• NCJ Number: 198829
• Author(s): Mary-Lynn Brecht; M. Douglas Anglin; Tzu-Hui Lu
• Date Published: January 2003
• Length: 98 pages
• Publication Series: NIJ Research Report
• Grant Number(s): 2000-IJ-CX-0017
• Agencies: NIJ-Sponsored
• Publication Type: Report (Study/Research)

Annotation

This study applied a logistic regression synthetic estimation approach to estimate the prevalence of illicit drug-using arrestees, using available ADAM (Arrestee Drug Abuse Monitoring) data for calendar year 2000 as a calibration sample for projecting to the national level.

Abstract

Prevalence estimates were made for any illicit drug use (any of 10 drugs tested by urinalysis) and specifically for cocaine. Drug use was estimated by gender, age group, and offense category. Estimation was performed for State and county data (California and its largest and smallest counties, Los Angeles and Alpine) to determine any illicit drug use by arrestees. The logistic regression synthetic estimation approach used in this study involved the application of the prevalence rates in a calibration sample (ADAM) to estimate the equivalent rates in a target population (national, State, or county) where the prevalence rates were unknown. This approach has the benefits of being low cost, relatively simple to understand and implement, and its use of existing available data. This method yielded an estimated number of U.S. arrestees with recent illicit drug use at 6.4 million, or approximately 65 percent of the arrestee population. Estimates were higher for males (4.8 million) than for females (1.5 million), which is approximately in the same ratio as arrests for males and females. For cocaine, the overall U.S. estimate of arrestees using it was 3.8 million (2.7 million males and 1.0 million females). The estimate for California drug-using arrestees was 780,000, approximately 61 percent of the arrestee population. An evaluation of the methodology affirmed the acceptability of most results regarding reasonability, replicability, and reliability; however, the method did not perform as well as anticipated in the estimation of opiate and methamphetamine prevalence; and results for some small subgroups were less reliable than desired. An assessment of potential estimate bias suggested that estimates were most likely conservative. The study recommends the continued application and refinement of this ADAM-based method for estimating the prevalence of drug use by arrestees. 8 tables, 6 figures, 52 references, and appended supplementary information on methodology and data