It has been an article of faith in medical informatics that computerization of healthcare processes will reduce medical errors, but a study released in the current issue of the Journal of the American Medical Association adds to a counter-current of reports that computerization has its drawbacks, too.
The authors, in the opening paragraphs of their "Role of Computerized Physician Order Entry Systems in Faciliating Medical Errors," acknowledge the reigning orthodoxy, noting that by some estimates, 770,000 people are injured or killed by adverse drug events in hospitals each year and that "CPOE systems are widely viewed as crucial for reducing prescribing errors."
But using a 71-item questionnaire developed during an iterative process of observation as well as one-on-one interviews and focus groups at the Hospital of the University of Pennsylvania, Philadelphia, researchers queried 261 active users of an early CPOE system in service at the 750-bed teaching hospital between 1997 to 2004.
The survey "identified 22 previously unexplored medication-error sources that users report to be facilitated by CPOE."
In December, U.S. Pharmacopeia, while not focusing exclusively on CPOE, found in looking at its drug-error reporting data from 570 hospitals and healthcare systems that one in five medication errors involved some sort of computerization or automated dispensing system.
But neither study measured the net effect of CPOE on the volume of errors because they didn't take note of error rates before the electronic systems were introduced, said David Bates, M.D., medical director of clinical quality and analysis in information systems at Partners HealthCare System, Boston.
The Pennsylvania survey "does provide useful information about categorizing errors," Bates said. But, "The important point is that this study did not evaluate whether or not CPOE reduced the medication error rate."
The Pennsylvania researchers placed CPOE errors into two categories: information errors generated by fragmentation of data, principally caused by lack of integration of the CPOE with other hospital record-keeping systems, and human-machine interface flaws that occurred because the software didn't match the way physicians provided care.
One example of data fragmentation problems cited in the report was that the Pennsylvania CPOE system forced physicians to view up to 20 screens of information to see one patient's medication record. An example of a human-machine interface problem was more prosaic: "Common" crashes caused delays in ordering. A third of physicians surveyed indicated these delays occurred "a few times a week" while nearly 9% reported they happened "about once a day."
"I'm not saying we should not be doing CPOE," said study co-author Brian Strom, M.D., associate vice president of strategic integration at the Hospital of the University of Pennsylvania and associate vice dean of the University of Pennsylvania School of Medicine. "It certainly has shown to have some benefit. The point is, it's not a panacea. It's not a one-shot deal. You can't just put it in place and say you're OK. It needs to be continually evaluated, changed and reviewed."
Researchers, while blaming some of the problems on the system itself, also noted that between 2002 and 2004, the study period, the Eclipsys TDS computerized order entry system in use at the Pennsylvania hospital was the market leader. Thus, the report contends, "There was no reason to suspect that TDS is inferior to any other CPOE system." Strom also remarked that the Pennsylvania hospital has since upgraded its CPOE to the next generation of Eclipsys system, and so some of the problems encountered before may have been eliminated. Still, he concludes, "continued research is needed to find out what (CPOE) does well and what it doesn't do well."
Bates has been with Brigham and Women's Hospital in Boston since 1988 and was there when the Harvard Medical School-affiliated hospital launched its homegrown CPOE system in 1993. He and his colleagues subsequently published a landmark article in JAMA in 1998 on the impact CPOE systems had on improving patient safety at Brigham and Women's, using data gleaned both before and after the CPOE system was implemented.
"We looked at the errors that were present with paper ordering and at both the errors created and eliminated with CPOE and there was a more than an 80% reduction," Bates said.
Bates expressed concern the Pennsylvania study might cloud the picture at a time when hospital officials are anxiously deciding whether to invest huge sums in CPOE and other clinical IT systems.
He said that at Brigham and Women's, even after nearly 12 years of using CPOE, they regularly collect feedback from users and are constantly making improvements to the application. He agrees with Strom that hospital officials mustn't look at CPOE as a one-shot cure.
Today's CPOE systems are "much better than they were a few years ago, but still what you get off the shelf won't do all you'd like," Bates said. "We recognize that process is never going to stop (because) the evidence about what constitutes adequate care is always changing."