Typical breath tests can only detect very recent alcohol use, but a new urine test called EtG promises to detect a biomarker for drinking up to 80 hours after the subject's last drink, the Wall Street Journal reported Aug. 12.

The EtG test has been adopted by the state of Pennsylvania, which uses it to verify that health professionals who are recovering from alcoholism have lived up to agreements not to drink. A lawyer for the state recently called the test, which screens for the alcohol metabolite ethyl glucuronide, the "gold standard" of alcohol testing.

About 10 percent of the 20,000 people being regularly screened with the EtG test because they are not supposed to drink at all are testing positive, industry experts say. The tests cost about $25 each.

However, EtG has one major flaw: it can detect ethyl glucuronide even in people who did not drink. The metabolite can show up when alcohol is inhaled, absorbed, or ingested through food, medicine, or even hand sanitizers. H. Westley Clark, director of the federal Center for Substance Abuse Treatment (CSAT), said the test "can't distinguish between beer and Purell."

"When you're looking at loss of job, loss of child, loss of privileges, you want to make sure," Clark said. CSAT is studying the EtG test and plans to issue a report.

Drug-testing firms say it is up to clients how they use the results. Kevin Knipe, manager of a Pennsylvania monitoring program for health professionals in recovery, contends that even accidentally absorbed or inhaled alcohol poses a threat of relapse. "They must abstain from alcohol in any form," he said.

Gregory Skipper, a recovering addict and physician who first promoted use of the EtG test in the U.S., said the "use of this screen has gotten ahead of the science." Even the testing industry has backed away from earlier claims that the test only produces positive results when subjects have been drinking. Still, in places like Pennsylvania, health professionals have lost their licenses as a result of EtG testing, even though the state isn't even trying to prove that testing subjects actually drank.