The effectiveness of pre-employment drug testing for identifying problem workers may have been greatly exaggerated, according to one of the largest analyses of the effects of workplace drug use.

In a 2 1/2-year study of 2,537 postal workers in Boston published today in the Journal of the American Medical Association, researchers found that employees who tested positive for marijuana had 55 percent more industrial accidents, 85 percent more injuries and 78 percent more absenteeism than those who did not test positive. Cocaine users had a 145 percent increase in absenteeism and an 85 percent increase in injuries.

But the frequency of such problems was substantially lower than predictions commonly quoted in drug-testing literature, where drug users have been estimated to have 1,500 percent more absenteeism and sustain 400 percent more injuries. The authors of the study also stated that at least some of the increased problems seen among drug users had nothing to do with drugs, but were due to alcohol use, for which they were unable to test.

Although drug users had a higher-than-average frequency of workplace problems, the incidence of problems even among users was small. Of workers who tested positive for drugs -- 12 percent of those hired during the study period -- the majority had no accidents or injuries.

The findings are part of a small but growing body of scientific literature casting doubt on claims made during the mid-1980s about the effectiveness of pre-employment screening for illicit drugs. In an accompanying editorial, Eric Wish, director of the Center for Drug Abuse Research at the University of Maryland, argued that the low rates of workplace impairment and inability of drug screens to distinguish between casual drug users and habitual users suggest that pre-employment drug screening has "limited potential."

"We will literally test 100,000 employees to find roughly 1,000 marijuana users,but you can go into a booking facility of any large city in the United States and you can find 100 cocaine users by just testing 200 arrestees," Wish said in an interview.

"When you put it all together, it looks to me that if we really want to counteract the drug abuse problem we should be looking elsewhere than the employee population," he added.

The drug use study, conducted by a group of researchers at the University of Iowa, was begun at the Boston area's main post office four years ago to test estimates of many drug-testing advocates and corporations about the economic damage done by drug users in the workplace.

At the time the study was started, the Postal Service did not prohibit hiring employees who tested positive for drugs. The researchers received permission to screen job applicants for illicit drug use. They then followed all of those hired for an average of a year to see whether employees testing positive were any more likely to have accidents, suffer injuries, take sick leave or lose their jobs than those who tested negative.

The researchers stress that their results are not definitive, chiefly because they were unable to separate the effects of alcohol use, which is strongly correlated with workplace problems and use of other drugs, from cocaine and marijuana use.

"It is conceivable that the results we found in our study weren't caused by marijuana and cocaine at all but that we were simply using {them} as a way of identifying alcohol abusers," said University of Iowa researcher Craig Zwerling, who led the study.

Zwerling said he thought that drugs and alcohol probably played an equal role in causing workplace problems.