Sometimes, A ‘Bad Day’ Is Fatal New Tests May Help Determine If People In High-Risk Professions Are A Danger
Good people can have a bad day. And when those people are pilots or railroad engineers, a bad day can spread to hundreds of others.
For years, scientists have wondered whether it would be possible to develop a simple, unobtrusive “bad-day” test for workers in critical industries. It would have to be fast and reliable, and it would have to detect a potential for impaired performance that might result from various factors, such as alcohol and drug use, fatigue, illness and extreme stress.
It is possible now to measure for alcohol or drugs, but the tests hardly qualify as “simple” or “unobtrusive.” Besides, many accidents involve neither alcohol nor drugs, including two of the most recent.
Those two disasters involved veterans with clean records and good reputations.
Capt. Nicholas Tafuri, with an unblemished record of 26 years with American Airlines, was in charge of Flight 965 on Dec. 20 when it slammed into a mountainside near Cali, Colombia, and killed 160 people. Engineer Richard Orr, with a similarly unblemished 22-year record, was running a Maryland commuter train Feb. 16 when it slammed into Amtrak’s Capitol Limited and killed 11 people.
The National Transportation Safety Board has not yet determined the cause of either accident, but initial indications are that Tafuri and Orr were distracted and perhaps under stress.
Will there be a simple method to determine in advance whether professionals such as Tafuri and Orr are experiencing problems that could affect their work? And will there ever be a way to monitor them minute by minute to determine whether their performance is deteriorating during their duty periods?
Scientists are optimistic that such technology can be developed, but there is widespread disagreement about whether several devices now hitting the market can fill the bill.
This burgeoning branch of behavioral science is called “fitness for duty” or “readiness to perform” testing. Much of it is in the experimental stage, but several companies are marketing devices. They generally break down into two groups:
Computer-based skill tests that measure simple motor skills, reaction time, memory, spatial perception or problem solving.
Eye-reaction tests that measure the rate of eye movement and the reaction of the pupil to light.
Devices that use either approach can keep a computerized record of each worker’s performance and compare each new test to an individual average. None of them can determine exactly what has caused an abnormal or substandard result.
“I think all of these are interesting approaches,” said Mark Rosekind, head of the fatigue countermeasures program at NASA’s Ames Research Center in California. “The problem is, there’s no data to say one of these is better than the other.”
Other scientists disagree.
Martin Moore-Ede of Circadian Technologies Inc. in Cambridge, Mass., a prominent researcher in the field of biological clocks, has adopted the eye-reaction approach, using PMI Inc.’s 30-second eye test. The test measures involuntary eye reactions to light plus the eye’s saccadic velocity - the speed with which it follows a moving target.
“We’ve taken a look at many of them,” Moore-Ede said. “This was the one that struck me as superior in a number of ways.”
He said the eye functions measured by the device are affected by numerous physiological and psychological factors and can’t be consciously controlled by the person being tested. In contrast, he said, skill-based programs can be fooled, an assertion that is disputed by their manufacturers.
PMI, of Rockville, Md., is one of several companies marketing a commercial version of such eye tests. Its product has undergone extensive validation studies, including research at an Alcoa Aluminum plant in Texas, a Federal Railroad Administration fatigue study, and a fatigue study of soldiers in combat conducted at Walter Reed Army Medical Center.
Other companies, such as Performance Factors Inc. of Alameda, Calif., have focused on skill testing. Performance Factors markets Factor 1000, an eye-hand coordination test that can operate on any IBM-compatible computer. Using a control panel with a knob, a subject attempts to hold steady a pointer on the screen that appears to move randomly.
Eventually the movement of the pointer exceeds the subject’s ability to control it. Once each worker establishes an individual base line over several days, the test can detect when psychomotor performance capabilities are impaired.
Civil liberties groups support fitness-for-duty tests because they do not involve invasion of individual privacy. The American Civil Liberties Union says those tests get to the heart of the issue: whether a person can safely perform a task, not whether the individual has violated a law.
Some scientists say more work needs to be done before the testing devices come into widespread use.
Kirby Gilliland and Robert E. Schlegel, of the University of Oklahoma, in a 1993 report for the Federal Aviation Administration, said “numerous issues” need to be addressed, including more research to determine what factors best predict readiness to perform, as well as the possible hidden costs of such testing.
But the key problem with fitnessfor-duty tests is that, so far, they measure only a worker’s condition at the beginning of a shift. There is no technology to continuously rate a person’s fitness during work periods.
“The technology, as far as it goes, appears to be valid,” said William Keppen, a vice president with the Brotherhood of Locomotive Engineers who has spent more than a decade working on fatigue issues. “But the concern is, it really only measures an individual’s state when he comes to work.”
Preliminary data from an engineer fatigue study, sponsored by the engineers union and the Federal Railroad Administration, indicate that current schemes to keep engineers alert don’t always work. Engineers put through a week of “severe” work schedules at the Illinois Institute of Technology’s locomotive simulator in Chicago were often able to reset the locomotive “alerter” even when they were asleep.
But Keppen said the ongoing simulator tests have produced an unexpected bonus. Researchers have been able to determine that engineers’ alertness was beginning to flag long before they experienced sleep or “near-sleep events.”
This offers hope, he said, that systems could be developed to monitor an engineer for dangerously low levels of alertness, perhaps signaling a dispatcher or stopping the train.