"computers" and "doctors" in the same sentence, especially given
the heated debate over HMOs and individual health care, and people
tend to conjure Orwellian images of huge clinical bed labs manned
by faceless technicians diagnosing patients by the number.
Not David Schriger. To the contrary, the associate professor of
medicine in the Division of Emergency Medicine has devised a computer
program for UCLA's emergency room he believes will restore the personal
touch to medicine. By prompting physicians to ask the right questions,
order the proper tests and then prescribe the correct medications
and procedures, Schriger's program will unburden ER doctors from
some of the vast memorization otherwise required, thus giving them
more time to practice their bedside manners.
still want the Marcus Welby-model plus the technical knowledge,"
says Schriger, who came to UCLA as a fellow in 1987 and has been
working on the computer programs since 1989. "Maybe we can use the
technology to do some of the latter so that the doctor can concentrate
on the former. The technology is like a stethoscope or other tool;
it's a means, not an end."
software comes out of a larger movement in this country toward developing
clinical guidelines for everything from depression to bed- sores.
But it was also motivated by his own interest in determining and
modifying how doctors make decisions. The problem, he has found,
is encouraging doctors to implement standardized recommendations.
"Even if you get them to read the guidelines, they may not use them,"
Nowhere is this more apt to occur than in the chaotic environment
of the emergency room, where it's doubtful a doctor will stop to
read a standardized protocol for dealing with, say, cardiac arrest.
The computer, though, offers a carrot: If the doctor works on a
case through Schriger's software, he or she won't have to write
out prescriptions, test orders, discharge instructions or even medical-chart
reports. The computer will do it for them - and legibly. It will
also document their decision-making for later analysis.
So far, Schriger has developed programs to handle five frequent
complaints encountered in the emergency room. If a patient comes
in complaining of needle sticks, lower-back pain, seizures in known
epileptics, fever in healthy children under 3 or male urogenital
problems, then his or her chart is flagged as a computer case. The
physician can then choose to look up the complaint on the computer
and answer a series of diagnostic questions. The program suggests
which tests to run, marking some optional (in blue) and some required
(in red). The suggestions are strictly advisory, but if the physician
ignores a "red" test, he or she has to type out an explanation.
After the chosen
tests are conducted, the physician returns to the computer and enters
the results. The program then recommends treatment options or more
tests. Those suggestions are also color-coded, and again the physician
must explain any deviations from red recommendations.
more than 1,000 ER patients have been treated with computer assistance,
and Schriger is pleased with the results. In the case of needle
sticks (the first complaint for which he's fully analyzed the data),
96 percent of all cases were computer-aided, physician documentation
increased substantially and the quality of treatment improved. (A
Web site funded by the Centers for Disease Control will make the
needle-stick diagnostic available on-line in early 2000.) In addition,
charges for care have dropped 28 percent. But those remarkable savings
may not hold for all complaints.
is that we're going to see variance across the specific complaints,"
says Schriger. "The drop in costs, for example, may be true for
needle sticks more than for other complaints."
Schriger wants the computer to catalog protocols for 20 chief complaints,
covering 80 percent of all emergency room patients. And he expects
the human interface with the computer to become less and less obtrusive.
Theoretically, with advances like voice-recognition software, physicians
will one day be able to simply record a conversation with a patient
and have the computer parse it out.
There are still
drawbacks, though. "One of the challenges here is that it's based
on having one chief complaint," says Schriger. "How do you write
the program to cover more than one complaint? Also, there may be
a whole class of complaints for which this isn't very good."
As for those
who worry that the computer will put the doctors out of work, Schriger
isn't concerned. "Fundamentally, I'm a humanist. I do this computer
stuff as a way to free the doctor to focus more on the art of medicine,"
he says. "Anyway, if you got rid of the doctor, who would you complain
to?" - Michele Kort