Computerized physician order-entry (CPOE) systems, designed in part to reduce prescribing errors, may in certain situations actually increase the risk of medication mistakes, according to a new study. Oncologists, however, with their well-defined treatment protocols, might get better use of these systems compared with other specialists, the authors said.
Prescribing errors are the largest identified source of preventable hospital medical error, and the computerized systems are widely viewed as key to helping to reduce prescribing errors and potentially cut costs.
The researchers, led by Ross Koppel, PhD, Professor of Sociology at the Center for Clinical Epidemiology and Biostatistics at the University of Pennsylvania School of Medicine, noted that although studies indicate that the computerized systems reduce medication errors up to 81%, few studies have focused on the existence or types of medication errors facilitated by computerized physician order entry.
“The main issue is the integration of the CPOE system into the hospital,” Dr. Koppel said in an interview. “This requires constant checking and feedback and an understanding of how the system is used in situ.”
This computer technology appears to be a double-edged sword. “As CPOE systems are implemented, clinicians and hospitals must attend to the errors they cause, in addition to the errors they prevent,” he and his coauthors wrote in the study, which was published in the March 9 issue of the Journal of the American Medical Association (2005;293:1197–1203).
One problem specific to oncology was that the lifetime limits on cancer drugs were not integrated into the computerized physician order entry system. Pharmacists had them on a separate listing, and had to call the house staff if a patient had maxed out on his or her lifetime limit.
“Without studies of the advantages and disadvantages of CPOE systems, researchers are looking at only one edge of the sword. This limitation is especially noteworthy because many problems we identified are easily corrected.”
The study assessed CPOE-related factors that enhance the risk of prescription errors, using a qualitative and quantitative study of house staff interaction with a computerized system at a teaching hospital.
The team interviewed 261 house staff (88% of CPOE users); conducted five focus groups and 32 intensive one-on-one interviews with house staff, information technology (IT) leaders, pharmacy leaders, attending physicians, and nurses; shadowed house staff and nurses; and observed them using CPOE.
The results showed that the CPOE system studied facilitated 22 types of medication error risks.
* Fragmented CPOE displays that prevent a coherent view of patients' medications.
* Pharmacy inventory displays mistaken for dosage guidelines.
* Ignored antibiotic renewal notices placed on paper charts rather than in the CPOE system.
* Separation of functions that facilitate double dosing and incompatible orders.
* Inflexible ordering formats generating wrong orders.
Three-quarters of the house staff reported observing each of these error risks, indicating that they occur weekly or more often. “Many of the problems had been identified for years and were not addressed,” Dr. Koppel said.
One problem specific to oncology was that the lifetime limits on cancer drugs were not integrated into the CPOE system. Pharmacists had them on a separate listing, and had to call the house staff if a patient had maxed out on his lifetime limit.
The particular way that CPOE products are developed and implemented makes all the difference in whether quality is improved, noted Michael Cohen, RPh, MS, ScD, President of the Institute for Safe Medication Practices.
“I see a lot of computer systems across the country. Honestly, vendors haven't done the greatest job. The systems are complex and hard to learn. A physician on different hospital staffs can go from one system to another with no standardization.”
However, he noted that the technology in the Penn computerized system was 20 years old: “It's not modern technology. The system was not designed to prevent some of the things the researchers saw, whereas today's systems are.”
Dr. Cohen added, “It worries me that oncologists might read this and say, ‘I told you so, we're not ready yet.’”
Oncology is more likely to have a successful CPOE system because of a long history of building and interpreting practice protocols, noted the coauthor of an accompanying editorial, Robert L. Wears, MD, MS, of the University of Florida.
“There is a body of expertise in oncology on how to use treatment protocols,” he said. “At least a portion of this could be built into a CPOE system with some success.”
Complicated oncology treatment protocols, first developed in the real world to get the bugs out of them, “might be able to be implemented into a computer system,” Dr. Wears said. “The system designers would have to watch when, how, and why oncologists deviate from the protocols, and program that into the system.”
In general, computerized systems are much safer than hand-written orders, Dr. Cohen said.
“Modern systems flash messages to note potential drug interactions or allergies, whether a physician is prescribing too high a dose, or whether a patient has missed a course of therapy.”
What's more, the Penn researchers did not measure medication errors before and after the system was installed, so the system may have prevented prescription errors, he said.
“Some of the 22 medication error risks were related to lack of training or understanding, and were not the fault of the computer system.”
For example, the house staff looked at the drug inventory and took the lowest and highest dosages to mean appropriate dosing levels. “That's tablet strength, not dose range,” said Dr. Cohen, who also noted that the older system was not designed to pick up dosing duplication.
Dr. Wears said, “This is a design flaw, not a training flaw. If the system misleads experienced, expert users, then training is not the solution.”
‘Not Based on How Clinicians Work’
He agrees that newer systems are probably better, but even new systems have a fundamental problem. “CPOE systems that look at the epidemic of medication errors make a mistake—they are not based on how clinicians work,” he said.
“The complexity of care and interruptive nature of care needs to be addressed.”
Dr. Koppel added: “Software is malleable. You're supposed to fix it when it allows an easy mistake.”
In the editorial, Dr. Wears and his coauthor, Marc Berg, MA, MD, PhD, of Erasmus University in the Netherlands, said that the results of the study are disappointing but should not be surprising.
“There is a long-standing, rich, and abundant literature on the problems associated with the introduction of computer technology into complex work in other domains, as well as occasional notes in health care,” they said noting that about 75% of all large IT projects in health care fail.
Dr. Cohen commented that a number of physician practice types of prescriber systems have come onto the market without proper financial backing or with poor designs, and they have failed.
Can Be Well Designed
“But this is not the same as a computer system in a hospital,” he said, adding that he could think of only one computerized hospital system that has failed—at Cedars-Sinai Hospital in Los Angeles, which was due to design flaws that stemmed from not consulting with front-line practitioners.
“Systems can be well designed,” he said. “When vendors install a system, they need to tailor it to the specific hospital and make sure that the staff spends time learning it and living with it.”
Dr. Wears said it's relatively easier to build a computer system to support a two-person pediatric practice than it is for a multispecialty medical center.
In a medical center, there are many different work processes that are always changing, he said.
He said he believes that current CPOE systems are experiments, not solutions, although he did note that Brigham & Women's Hospital in Boston and Later Day Saints Hospital in Salt Lake City have large, established, long-standing medical informatics support systems.
“These institutions are well-resourced and engage in development work in computerized systems,” Dr. Wears said.
“Most hospitals in the US are not well-resourced for development projects, which these systems really are.”
Only 5% to 8% of all hospitals have computerized physician order entry systems, which are designed to take orders in, not to provide clinical, interaction guidelines, Dr. Koppel said.
“I'm pro CPOE. I want better integrated CPOE, not a knee-jerk defense of technology, which is often poorly integrated into hospital information flow.”
The focus of new systems should not be with the software but with the organization of work, he added. “We need systems that meet doctors' needs to do better medicine.”