To err is human — or is
it? With thousands of patients dying every year from hospital
prescribing errors and other adverse drug events (ADEs),
computers seem to offer a deus ex machina solution
to one of medicine's most intractable problems. Computerized
physician order entry (CPOE) possibly eliminates a number
of obstacles to proper prescribing; like reducing transcription
errors, minimizing dosing inaccuracies and forestalling
dangerous drug-drug interactions. Various studies have
claimed that CPOE could potentially reduce ADEs by anywhere
from 18-81%. Yet research in the May 23 Archives of
Internal Medicine has found that ADEs continue apace
in even the most highly computerized hospitals.
"What I think is needed is a much
more comprehensive approach to ADE reporting, which
feeds data directly back into the system so we can learn
from mistakes," says Dr Jonathan Nebecker, a physician
at the VA hospital in Salt Lake City and the study's
lead author. The system should offer more decision-making
support, he adds, for example reminding doctors to provide
prophylaxis for likely drug reactions. Forgetting to
provide potassium with diuretics was one of the most
common errors in the study.
TECHNOLOGY
ON TRIAL
In this most recent study, 937 patients were randomly
selected from new admissions at the Veteran's Administration
Medical Center in Salt Lake City. The hospital has one
of the most advanced and comprehensive systems in the
world, featuring barcode-controlled medication delivery,
a complete electronic medical record, automated drug-drug
interaction checking and computerized allergy tracking.
During the study, 483 participants
suffered clinically significant ADEs (one that necessitated
a change to the patient's treatment plan) for a rate
of 70 ADEs per 1,000 patient-days or 52 ADEs per 100
admissions. Only about one in 12 ADEs involved a dosing
error, and 93% were adverse reactions. Additive drug-drug
reactions, in which one drug reinforces the effect of
another, accounted for 39% of the ADEs. Analgesics and
narcotics were the drug types most frequently involved,
followed by diuretics and cardiovascular-renal agents.
Six ADEs resulted in deaths, half of them from narcotics.
Of the 937 patients averaging a one-week stay in this
high-tech hospital, about 0.6% died from ADEs.
ERRORS
ABOUND
These figures point to a far higher rate of ADEs than
in most previous studies. The authors don't believe,
however, that their hospital is worse off. They attribute
the higher rate to the fact that pharmacists did the
ADE record-keeping and also to the accessibility of
data contained in the computer system.
Dr Nebecker doesn't go so far as
to suggest that computerization has increased or reduced
the number of ADEs, but says it has changed the distribution
of errors. Before, most mistakes were made in transcription
and administration. Now, ordering errors predominate,
prompting them to adjust their system accordingly. "We've
made some useful changes to our system since we began
this research," he says. It now includes standardized
'order sets' to make sure no vital elements of a treatment
program are omitted and takes better account of each
patient's renal status. "And we've reduced the number
of drug-drug interaction alerts to focus on the likely
and important ones. With fewer alerts, people are less
likely to ignore them," adds Dr Nebecker.
Arch Intern Med May 23, 2005;165:1111-6
|
