Health IT Should Accept Criticism

By M.L. Baker  |  Posted 2005-03-15

Health IT Should Accept Criticism

Software vendors are loudly criticizing a recent study showing that hospital computer systems can help physicians make mistakes while ordering medicines.

Advocates of CPOE (computerized physician order entry) complain that the study points out flaws in an old computer system at a single hospital and does not compare the errors that the system facilitates to those that it prevents.

The cheerleaders have some valid points, but mostly they need to be more willing to accept criticism than to dish it out.

Critics dismiss the study by saying that the flaws identified by the study were already recognized and have been fixed in a next-generation product.

Click here to read more about the study, which listed more than 20 ways that CPOE made medical errors more likely to happen.

But the lead investigator, Robert Koppel, could not recall a single improvement to the CPOE that occurred during his two-year study.

Even if all the glitches were eliminated, the study still has a point to make.

The main lesson is not that 22 sources of error were identified, but that vendors and administrators need to be more willing to find and fix sources of error.

"[The vendors] have to realize that its the software that has to be malleable. They have to be responsive to whats going on on the hospital floor," said Koppel.

There was also a lot of whining that the study did not compare the errors that CPOE facilitates to the ones it prevents.

Nor does the study consider other potential benefits of CPOE, like fewer unnecessary procedures or speedier medication delivery.

But Koppel, a sociologist at the University of Pennsylvania School of Medicine, is himself an advocate of CPOE.

He just thinks systems are in want of major improvements. The point is not whether or not to use CPOE, but how to use CPOE most effectively.

The necessary improvements actually seem relatively straightforward: like not forcing physicians to work through up to 20 screens to place an order, redesigning selection screens so that its not so easy to mix up patients and medications with similar names, and increasing integration so that a doctor can see all of a patients medications.

To improve CPOE systems, vendors need to know whats not working.

To better use CPOE, hospital managers need to know what can go wrong. Koppels study, as well as those of other researchers, meets this need.

These efforts should be welcomed, not censured.

Next comes the complaint that Koppels study was conducted at just one hospital, which has now updated its computer system.

But the problems that Koppel found have been reported across many facilities. (Indeed, another complaint of vendors was that the study found no surprises.)

Next Page: The source of the errors.

The source of the


For example, the United States Pharmacopeia has conducted annual Web-based surveys of medical errors since 1999. (The surveys are better known by the name of the error reporting program, Medmarx.)

In 2003, the latest year for which figures are available, 84 percent of the 570 participating facilities reported computer entry as a cause of error.

Computer entry was ranked as the fourth leading cause of error, and it has increased in rank every year since 2000.

Computer entry was reported as a cause for 13 percent of errors. [The top three causes of errors were performance deficit (38 percent), not following protocols (18 percent), and inaccurate or missing transcripts (13 percent); survey participants could select more than one cause of error.]

One response to such information is that physicians need more training and more willingness to be trained.

This is true, but the survey indicates that staff training can only go so far.

Participants who listed additional sources of error along with computer error most frequently picked distractions and performance deficit, meaning that the professional was able to use a computer competently, but simply made a mistake in a particular instance.

Read more here about a report that shows that IT can play a key role in creating a more integrated health care system.

Overall, evidence indicates that CPOE does prevent errors. In fact, the Medmarx survey found that inpatient facilities with CPOE reported one harmful medication error per 100,000 doses, while those without reported two such errors per 100,000 doses.

But these results do not leave room for complacency.

Instead of self-congratulation, vendors need vigilance.

In the words of the survey, they must "ensure that errors are truly averted and not perpetuated within the system."

M. L. Baker writes about biotechnology and health IT for She can be reached at

Rocket Fuel