Study: CPOE can increase risk of medication errors

Computerized physician order entry systems can increase the risk of medication errors, a study in the current issue of JAMA found. Researchers found that the CPOE system they studied created the potential for 22 types of medication errors.

Ross Koppel, a sociologist with the University of Pennsylvania School of Medicine, and colleagues in 2002 began a two-year qualitative and quantitative study of staff interaction with a CPOE system from TDS at a 705-bed urban teaching hospital. Healthcare IT vendor Eclipsys purchased TDS in January 1997. Researchers conducted five focus groups, 32 one-on-one interviews and surveyed 261 staff members, including nurses, physicians, pharmacists and IT staff, on their use of the CPOE system. Problems with the system included CPOE displays that blocked a coherent view of patients' medications, pharmacy inventory displays that were mistaken for dosage guidelines and rigid ordering formats that caused incorrect orders.

Although researchers only examined one system, Koppel said the lessons could be applied to any CPOE system. "We're not opposed to CPOE, but we're opposed to CPOE that's badly designed and not aggressively examined," he said.

Researchers said most studies of CPOE have been enthusiastic, but that the technology can create error risks in addition to reducing them. Oftentimes, failed CPOE implementations are blamed on physician resistance to the technology, said Robert Wears, MD, a professor in the department of emergency medicine and director of medical informatics the University of Florida, Jacksonville. Wears, who co-authored an accompanying editorial on CPOE systems, said this study proves there are several reasons why CPOE implementations fail and it's not just a matter of bad programming or physician resistance.

"In short, rather than framing the problem as 'not developing the systems right,' these failures demonstrate 'not developing the right systems' due to widespread but misleading theories about both technology and clinical work," Wears and his colleague wrote in the editorial.

Wears and Koppel said many CPOE systems don't take into account how physicians actually work in a hospital and are created for an idealized setting where the doctor works in a quiet environment and has time to carefully select orders. To prevent some of these mistakes, researchers made several recommendations that providers:

1. Focus on the organization of work, not on technology, to evaluate whether CPOE can improve patient care;

2. Examine the technology in use and some of the problems that are obscured when staff find workarounds of the system;

3. Aggressively fix technology when it is counterproductive;

4. Understand that there are multiple factors that contribute to errors and that these reasons may have several layers of complexity; and

5. Plan for continuous revisions and improvements of the system while recognizing that all changes can create new error risks.

CPOE experts say systems have value

While hospital CIOs should carefully evaluate CPOE implementations, David Bates, MD, said the study is flawed because it did not look at whether CPOE reduced the medication error rate and only asked staff about their perceptions of errors. Bates, medical director of clinical quality and analysis in information systems at Partners HealthCare System, has been involved with several studies of CPOE systems, including one that showed up to an 80 percent reduction in errors.

"A lot of the issues they [the researchers] identify as relating to the technology are not related to the technology at all," Bates said. "It's hard to blame the software." Some of the problems, such as system down time, are attributable to the hospital's underlying information system, he said.

Bates also pointed out the TDS system researchers studied has been replaced with an entirely different product.

Healthcare IT vendor Eclipsys said the study reflects some limitations of first-generation CPOE systems. "The report shows the value of studying new technology and identifying areas that require improvement," Eclipsys Chief Medical Information Officer Rick Mansour, MD, said in a written response. "Healthcare providers should be continuously aware of the ongoing innovations available to help marry information technology and clinical decision making in today's complex healthcare environment. They must also always be diligent in providing comprehensive training in the use of these systems for all of those individuals who will be using them to support high-quality patient care."

Koppel said it would be a mistake to conclude that the problems with the CPOE system that researchers examined are isolated. He said healthcare IT vendors should be more "humble" and candid about the difficulty of implementing CPOE. Vendors need to employ people who understand these technologies and can make them fit the reality of medical care, he said.

"We try to wrap, like a bad pretzel, the hospital and its people around the software," he said. "Software should make the practice of medicine better, not worse."

Koppel said he hopes to conduct future studies on the impact of newer CPOE systems.