YOU ARE HERE: LAT HomeCollections

Administration's War on Drugs

November 13, 1986

While we agree that illegal use of drugs is a serious problem, your articles about large-scale drug testing in the workplace indicate that such testing has potential for great harm. Questions of invasion of privacy and other injury to civil liberties are substantial.

We want to amplify your discussion by noting that mass screening, however accurate on its face, may carry unacceptably high risks to the reputation and jobs of large numbers of innocent people. The phenomenon of so-called false positives , mentioned in your articles, is well known to professionals in survey work. The issue of quantitative determination of the percentage of positive tests that are "false" demands a place in the current public debate.

Readers may be surprised at how serious the problem is. Simple arithmetic is enough to see this. Suppose that 10% of the population uses drugs for which a test is to be administered (in line with the 9%-12% figure for drugs in the workplace mentioned in your articles). Suppose also that the test is 95% accurate in reporting the status of either a drug user or a drug-free person (in line with an estimate of accuracy of 95%-97% given in your article as a good figure for a test with confirmation ). With an "on average" analysis, among 1,000 employees, 100 are drug users and 95 will be reported as such by the test. Of the remaining 900 employees, 45 (5%) will be reported as drug users. Thus, of the 140 employees with positive reports, nearly 1 out of 3 will be erroneously labeled as drug users.

The problem is even worse than this for a more "law-abiding" population where only one person in a hundred is a drug user. With the same 95% accurate test, more than 5 out of 6 of those people with positive tests would not be drug users. Even if a better test, say 99% accurate, were used on the population with 1% drug users, one-half of those testing positive would be incorrectly labeled as drug users.

Given the high accuracy claimed by the testers, who would believe the one-third to five-sixths of those tested with positive results who honestly protest that they are innocent?

It is imperative to determine and discuss the accuracy of any proposed test, especially in a mass screening program where presumably dozens of laboratories would need to be monitored for their care in preserving the integrity of thousands of urine samples.

But as we have shown, even an apparently acceptable answer to this question (95% to 99%) may have unacceptable consequences for innocent people, all the more so in a law-abiding segment of the population.




Grabiner and Vitale are professors of mathematics at Pitzer College and the Claremont Graduate School.

Los Angeles Times Articles