Cocaine use in arrestees: a majority of the frequent cocaine users in the United States are contained in the criminal population. (Cover Story)

Estimating the number of cocaine users in the country is a worthwhile goal but an extremely difficult task. One must select a valid and representative sample of persons, locate them and measure their drug use.

But how can scientists access drug users when many of the worst cases are likely to be outside of the traditional work, school, or home environments? And even if one can locate the, how does the government survey person convince the respondent to report his or her illicit drug use voluntarily?

Satisfactory answers to all of these questions are not available. Nevertheless, policy makers need estimates of drug use and have routinely relied on several national surveys to obtain them. [1]

Many scientists have suspected that these surveys paint an incomplete picture of drug use in the country. However, with the introduction of a new measure of drug use in booked arrestees, it has become clear that some of our national estimates of drug use have been too low. This paper describe the new data system and discusses some of the implications for national drug policy.

Measures of Cocaine Use

Estimates of drug use trends in the United States are derived primarily from government-sponsored surveys of persons in households and seniors in high schools. The National Household Survey has been conducted every two or three years and is designed to estimate drug use in the U.S. household population aged 12 or older.

The High School Senior Survey consists of self-administered questionnaires completed each year by about 16,000 seniors enrolled in public and private high schools throughout the United States.

Because each of these surveys is conducted under rigorous, standardized conditions, one would expect that changes in drug use detected over time would reflect true changes in drug use in the populations measured. Nevertheless, the descriptions of the surveys have noted several limitations that may affect the estimated level of drug use and even the trends themselves.

First, each survey omits from its target sample persons who are likely to be among the most serious …