CPOE study shows drop in hospital errors

Errors included lack of adequate decision support

In a study published in the February 15 issue of the American Journal of Health-System Pharmacy,1 Agency for Healthcare Research and Quality (AHRQ) researchers, using a national voluntary medication error-reporting database, found that facilities with Computerized Prescriber Order Entry (CPOE) systems in place had fewer hospital-based errors than facilities without a CPOE system.

The researchers found that the most common CPOE errors were dosing errors, and that using CPOE itself could lead to errors because of faulty computer interfaces, lack of interoperability, lack of adequate decision support, and human factors such as typing errors, distractions, inexperience or lack of knowledge.

The primary objective of the study, say the authors, "was to assess the potential benefits and problems associated with CPOE using a voluntary medication error-reporting system, Medmarx, sponsored by the United States Pharmacopeia [USP]."1

"At the time we undertook the study CPOE was a hot issue — and personally, I have put a lot of effort into voluntary error-reporting systems, which is a huge issue here and in patient safety in general," says Chunliu Zhan, MD, PhD, Senior Fellow at AHRQ and lead author of the study. "One of the key questions was, what can we do with voluntary medical error-reporting data; what useful information can we get out of that?"

The study used Medmarx data from 2003. Medmarx (www.medmarx.com), which was established in 1998, is an anonymous system in which about 600 hospitals nationwide currently participate.

In their conclusion, the authors wrote that a national voluntary medication error-reporting database "cannot be used to determine the effectiveness of a CPOE system in reducing medication errors because of the variability in the number of reports from different institutions."1 They added, however, that "It may provide valuable and useful information on the specific types of error related to CPOE systems."1

Pointing out characteristics

That is exactly what this research was able to show. For example, in describing the characteristics of medication errors related to CPOE, they found that 51.4% were dosing errors, while the next highest category, "unauthorized drug," was only 3.6%. In terms of cause of error, 57.9% were attributed to knowledge deficit, 43.4% to computer entry, 21.6% to abbreviations, and 20.9% to calculations (the report identified one or more causes.) Factors contributing to medication errors included distractions (78.3%); inexperienced staff (10%); workload increase (7.2%); and "computer system down" (2.8%).

Limitations seen

The fact that the system was voluntary made it difficult to explain some of the findings, however. For example, why there were fewer hospital-based errors in the systems that used CPOE but more outpatient errors? "It’s voluntary, so we have no idea what percentage of the errors that happened got reported," notes Zhan. "It could be that outpatient errors happen more, but that’s probably not reality; it’s more likely that’s what got picked up and reported."

This led Zhan to what he calls one of the study’s major findings. "By showing you the data, we kind of validate our conception that you just cannot use voluntary data to compare one type of provider with another type," he says.

Another significant finding, he adds, was how the data varied from provider to provider. "However," he notes, "voluntary data can provide you with rich information on what went wrong. In this case, can you say the error you reported was caused by CPOE and look at the patterns of what went wrong and why that kind of error happened."

In terms of simply evaluating or determining the effectiveness of CPOE, however, Zhan says there are better methods. "For example, you can do a pre/post study," he suggests. "Before you install the system, look at your error patterns, then after you install it, look again. That’s probably the most valid method for determining effectiveness."

CPOE, he insists, definitely reduces errors. "The extent to which it is effective depends on the system," he adds. "A lot of doctors are turned off because a system is too error-proof’ — it holds too much information for them to absorb. So, there has to be an optimal balance between efficiency, effectiveness, and sophistication. We don’t have to prove any longer that CPOE is effective; what we need to look at now is how we can improve CPOE systems."

For more information, contact:

Chunliu Zhan, MD, PhD, Center for Outcomes and Evidence, Agency for Healthcare Research and Quality, 540 Gaither Road, Rockville, MD 20850. Phone: (301) 427-1225. Fax: (301) 427-1640. E-mail: Chunliu.Zhan@.ahrq.hhs.gov.


  1. Zhan C, Hicks RW, Blanchette CM, Keyes MA and Cousins D.D. Potential benefits and problems with computerized prescriber order entry: Analysis of a voluntary medication error-reporting database. Am J Health Syst Pharm 2006; 63 (4): 353-358.