Joyce Mendelsohn, Post-Gazette UPMC Shadyside will switch to a new system that allows physicians to electronically enter orders for patients to undergo tests and receive drugs. Click photo for larger image.

Could a hospital computer system meant to save lives inadvertently lead to more deaths?

It's a question lurking in the background this month as UPMC Shadyside switches to a new system that allows physicians to electronically enter orders for patients to undergo tests and receive drugs.

The costly computer system being installed allows for "computerized physician order entry," an electronic way of directing patient care that's been promoted by business groups and patient safety experts as key to standardizing hospital care. The idea is that fewer patients will suffer the ill effects of medical errors, such as patients getting the wrong drug, due to miscommunication.

But the experience thus far in Pittsburgh hospitals has highlighted both the promise of the new technology and the unsettling prospect of unintended consequences. A study published in December found an unexpected mortality increase at Children's Hospital of Pittsburgh among critically ill patients following the implementation of a computer system in 2002.

Administrators at Children's say they strongly disagree with the study, and counter that the system -- as well as one installed at UPMC St. Margaret during 2004 -- has saved lives by helping make sure patients get the right medicine at the right time and in the right dose.

"We took it very seriously, and looked into it in great detail," said Dr. Eugene S. Wiener, medical director of the computer program at Children's. "I don't believe a single patient died because of this, and all of our data show that medication errors significantly decreased each year after implementing the system."

While supporters and critics of the new computer systems disagree on some things, they agree on why they are being adopted by more hospitals.

Starting with a landmark report in 1999 by the Institute of Medicine, public concern about the number of preventable deaths each year in hospitals due to medical errors has skyrocketed. Mistakes in prescribing drugs is one of the most common medical errors and computer systems are being promoted as a key prevention strategy. A majority of medical studies have found that the computer systems have a positive impact on such errors.

Still, a growing number of skeptics have questioned whether the systems simply introduce new types of problems. A 2005 article in the Journal of the American Medical Association, for example, reported that the computer systems actually facilitate 22 types of medication error risks related to mismatches between how computers and caregivers do their work.

The December article in the journal Pediatrics about the implementation at Children's Hospital added to the negative reports. The study found that the mortality rate for a subset of critically ill patients significantly increased from 2.8 percent before the computer system was implemented to 6.57 percent following implementation.

The study at Children's did not purport to show a cause-and-effect relationship between the computer system and the increased mortality. But it raised several examples of how the new computer system changed the way doctors cared for the patients, all of whom had been transported to the hospital for care.

After the system was installed, for example, orders could not be entered until after the patient had physically arrived at the hospital and been fully registered into the system. That led to potential delays in new therapies and diagnostic testing, the study authors said. And the process of entering medication orders that would help stabilize a patient's condition often took 1 to 2 minutes per order, compared with a few seconds needed to place an order before the computer system.

Moreover, nurses wound up spending less time at the bedside because of the computer system, the authors wrote, while doctors spent more time entering orders in the computer. And some medications that had been available in intensive care units were centralized in a pharmacy, the authors said.

As bad as it might sound, the study's authors -- two of whom declined interview requests -- urged caution when interpreting the results. "In a single institution, it is difficult to assess the causality of increased mortality when a new intervention is given, especially when the intervention affects the administration of every drug given to every patient," the authors wrote.

Dr. Wiener, the medical director of the computer project, said the study was correct on one point: Doctors quickly realized that orders needed to be ready to go once emergency patients being transported to the hospital arrived. But he denied that any patients died or were harmed as a result.

Some medications were centralized in the main hospital pharmacy, he said, but that change occurred about a year before the implementation of the computer system and was mandated by agencies that regulate hospitals

Children's officials did learn lessons along the way, Dr. Wiener said, and he acknowledged that every new process introduced in a hospital brings risks with it. But no patient died because of the computer change, he said, adding that he wouldn't have allowed for implementation had he thought a patient would die.

"I've talked to the doctors at Shadyside, and I've talked to them about this article to their satisfaction," Dr. Wiener said. "We've learned a lot, we've made improvements, but I still don't believe that a single patient died because of this."

On the contrary, the hospital has reported a number of positive results, including a sharp reduction in medication errors related to wrongly dosing a drug according to a child's height and weight, the elimination of errors stemming from penmanship and a 50 percent decline in delivery time for medications.

At both Children's and UPMC St. Margaret, doctors might well be spending more time entering orders at computers a result of the new system, said Dr. Daniel Martich, vice president of UPMC's electronic medical record project. But there are positive trade-offs since pharmacists and nurses spend less time interpreting physician handwriting, and the potential for miscommunication decreases, he said.

The implementation at UPMC Shadyside, which began June 4, is continuing in phases all month.