Computerized order entry systems may miss medication errors

Simulations of computerized physician order entry found a large potential for the systems to miss alerting doctors to medication errors, according to a study by the Leapfrog Group, an employer-backed organization that rates hospitals on patient safety.

In simulations at 214 U.S. hospitals, testers reported that more than half of the entered drug orders that contained non-fatal errors did not trigger warnings. Thirty-three percent containing potentially fatal errors also were not detected.

After the initial simulation, 104 hospitals adjusted their systems and protocols and took the test a second time. Nearly all showed sizable improvements in receiving warnings.

"We consider that very important information and very promising information, because that tells us that it is possible to improve your performance as a hospital," said Leah Binder, CEO of the Leapfrog Group, which has long promoted CPOE as a safer alternative to written and verbal communication of physician orders.

Binder said most hospitals install CPOE systems because they want to provide the best patient safety, but they rely too much on vendors to develop an adoption strategy. After adoption, most believe that the systems are functioning in the ways they are intended and are surprised when tests shows otherwise.

"There's no other way for them to know," Binder said.

More physicians are expected to start using CPOE in coming months. Currently fewer than 50% of hospitals have a CPOE in place, and fewer than 5% are demanding that physicians use them, according to the Healthcare Information and Management Systems Society. That number is expected to grow significantly in coming years as hospitals and physicians try to meet meaningful use standards, which initially will require that more than 30% of patients with at least one medication order have at least one of those ordered through CPOE.

Binder said the Leapfrog Group believes the government should establish testing criteria, and hospitals should be required to report their performance on those tests, to qualify for incentives available through Medicare. It wrote a letter to CMS during the public comment period on meaningful use, urging the agency to include the testing in the final meaningful use definition.

Leapfrog, which conducts an annual hospital survey, added an outlet in 2008 for participating hospitals to test their CPOE systems. Hospitals with CPOE that do not take the test cannot achieve the "fully meets" or "good progress" requirements for the survey.

For the Leapfrog CPOE test, hospitals were given a scenario of 10 patients and 50 medication orders. Pediatric hospitals were given 10 test patients and 51 medication orders. Each simulated order would result in an adverse drug event in one of eight categories, some potentially fatal.

Many hospital CPOE systems are set up to detect certain types of medication errors, but not others, which resulted in them not meeting Leapfrog's criteria for having a fully implemented system. Hospitals also struggle with finding a balance between having an appropriate number of useful alerts without reaching the point of "alert fatigue" caused when too many alerts are prompted, many of which are not needed or helpful.

The University of Maryland Medical Center was one hospital that participated in Leapfrog's study. Agnes Ann Feemster, PharmD, assistant director of clinical pharmacy services and investigational drug services at UMMC, said an initial test revealed gaps in its CPOE system.

Over the next year, changes were made to the system to add more of the medication combinations that were used in Leapfrog's test, as well as customized administrative tasks performed on the system. The next year, UMMC received Leapfrog's stamp of approval for having a fully implemented CPOE system.

A separate study, published in the May 2009 Journal of the American College of Surgeons, looked at the impact of a CPOE system at the Mayo Clinic in Phoenix. It found that the system helped improve efficiencies but did not reduce medication errors. Researchers concluded that "refinements will be required" for the system to have an impact on patient safety. The authors also found that there was no consistent standard for identifying and tabulating errors.

Physicians' role

The Leapfrog Group requires that physicians participate in any simulated testing it conducts of a hospital CPOE system.

CEO Binder said doctors should play a lead role in making sure testing is done at their hospitals, because doctors' reputations are on the line when something goes wrong.

UMMC's Feemster said physicians played an important role when the medical center made changes to its CPOE. The medical center developed a work group after its initial test in 2008 that included physicians offering input regarding the usefulness of each alert and the physicians' likelihood of acting on it.

After UMMC updated its system, it found that about 25% of orders produced an alert. The most common alert was one identifying a therapeutic duplication (14%). Medical students and emergency physicians were most likely to modify or discontinue an order when an alert was given.