Medication Errors At 'Wired' Hospital
Study: High-tech system also saw bad drug experiencesMay 25, 2005 | www.cbsmarketwatch.com
For many doctors, problems with starting and keeping patients on the right drugs crop up consistently, even in the best of circumstances.
But even a hospital that uses technology designed to make the process easier and more accurate isn't immune to medication errors, according to a study in the May 23 issue of the Archives of Internal Medicine.
Hospitals in the Veterans Administration Healthcare System have been among the early adopters of platforms such as computerized physician order entry to reduce transcription errors, bar code-controlled medication delivery, electronic medical records, automated drug interaction checking and computerized allergy tracking and alerting.
But a quarter of inpatient stays at a Salt Lake City VA hospital had at least one adverse drug event, meaning the patient suffered injuries resulting from the use of a drug, according to the study. Many patients had more than one problem related to a drug.
Such injuries account for as many as 41% of all hospital admissions and more than $2 billion in inpatient costs every year, the study said.
The vast majority 91% of those adverse events were seen as moderate, requiring doctors to monitor, discontinue or adjust the dose of the problematic drug. But 9% of such drug issues resulted in serious harm.
The computer systems were doing their job, but the study suggests decision support for selection, dosing and monitoring was lacking, said Dr. Jonathan Nebeker, co-author of the report and associate director of the Salt Lake Informatics Decision Enhancement and Surveillance Center, part of the Department of Veterans Affairs.
"All were working as designed, but there was minimal support for preventing adverse drug events associated with problems in the ordering and monitoring stages," he said.
Indeed, of the adverse drug events associated with errors, 61% were connected with ordering and 25% had to do with monitoring the drugs. Administration accounted for 13%, while just 1% of negative drug events happened in the dispensing stage. None occurred during transcription.
"The most dangerous type of errors originate in the ordering process and in the monitoring process," Nebeker said.
"In the outpatient setting, there are fairly good prompts for monitoring certain types of patient conditions, but in inpatient settings we still have work to do on that."
Nebeker and his colleagues reviewed daily electronic medical records from inpatients admitted during a 20-week period in 2000, leading to what he calls "the only study that broadly considers all types of inpatient adverse drug events in a fully computerized setting."
The authors considered an adverse drug event "clinically significant" when it required a change in the patient's treatment plan.
The drug classes most commonly found to have adverse drug events were narcotic pain medications, diuretics and cardiovascular or kidney agents, the study said. Those classes were either used independently or in combination with other drugs.
The VA system is popular with the Bush administration, which has made a priority of improving information technology at U.S. hospitals. Internationally, hospitals in Finland, Germany and some African countries use the VA's computerized protocol.
While some of the errors had to do with failing to counterbalance an expected effect of a drug, not all adverse drug events are related to medication errors, Nebeker said. Some patients have severe reactions that are difficult to predict.
Recognizing and treating adverse drug events may elude doctors who have to juggle a host of other factors that influence a patient's care, he said. About three out of four adverse drug events at the Salt Lake City hospital were identified as such a rate much better than the 25% recognition another study found a decade ago, Nebeker said.
The VA system has been upgraded since the time the study was conducted, offering more guidance on what doses doctors should prescribe and more sophisticated drug interaction checking routines, he said.
"There's a wider variety and smarter implementation of those alerts than was available in 2000," Nebeker said, estimating the system now sends a quarter of the alerts it sent before. "We used to just ignore all of these alerts because there were so many we didn't trust them."
"This technology now helps us identify both derangements in patients that are not yet frank adverse drug events, sort of emerging adverse events, to prevent those from happening, and to identify full-blown adverse drug events so we can intervene more effectively."
Another area of focus is making patients' history of drug experiences more visible, Nebeker said.