Computerized prescriptions only as good as the system and user

Does typing a prescription into a computer mean automatically reducing the risk of medication errors, compared with the old-fashioned written script?

One study's authors say the answer is no -- but their data show that electronic prescribing provides the opportunity for better quality of care as long as the systems have safety checks and features that are used properly.

"It's being worked on, but it's not fixed yet by any means," said study co-author Karen Nanji, MD, MPH, an anesthesiologist at Massachusetts General Hospital in Boston.

The research looked at the types and causes of errors associated with 3,850 computer-generated prescriptions sent to a commercial outpatient pharmacy chain in three states over four weeks in 2008. Researchers found about one in 10 included at least one error, a third of which had the potential for harm. The results were published online June 29 in the Journal of the American Medical Informatics Assn.

The rate and the kinds of errors were similar to those from paper prescriptions. Errors found in those prescriptions included unclear information about doses, quantities and drug names; conflicting information such as length of treatment, frequency and quantity; and clinical errors such as incorrect duration, strength, instructions and overdoses. The most common error was omitted information such as dose, frequency and quantity.

The authors, in a July 8 statement issued with e-prescribing network Surescripts, took pains to note that what they studied was not electronic prescribing. The statement was issued in response to media reports that described the study as covering e-prescribing. Surescripts did not participate in the study.

"The more common use of faxed or printed computerized prescriptions in 2008 [the year the data was gathered] has led many to equate their use with e-prescribing," the statement read. "By today's standards, e-prescribing refers to prescriptions that are generated using computers in practitioners' offices and sent electronically in coded form to pharmacies. Therefore, the results of the study should not be generalized to e-prescribing as there are clear and significant differences between these prescription transmission methods, and the terms and technologies are not interchangeable."

However, Dr. Nanji said the study is an opportunity to look at current systems to ensure they have the capability to detect the sort of errors spotlighted in the study. Since the time of the study, more physicians have used e-prescribing systems because of Medicare incentives that will soon turn to penalties for noncompliance.

Many of those areas of improvement are starting to be addressed by vendors and increasingly stringent meaningful use rules.

Ken Whittemore Jr., senior vice president of professional and regulatory affairs for Surescripts, said meaningful use has brought certain safeguards and checks to the e-prescribing process.

Meaningful use may not explicitly require some of the safety features to be used, or even be present -- at least not yet. But most of the systems certified to meet meaningful use, both electronic medical records and their e-prescribing components, have safety features that address the common errors the researchers found. For example, most modern e-prescribing systems have mandatory data fields. The systems will refuse to send a prescription if a mandatory data field is left blank.

The study found that forced use of functions designed to prevent omitted information could have eliminated 71.7% of the total errors and 63.2% of potential adverse drug events found in the prescriptions analyzed for the study. The addition of drug decision support tools, such as maximum dose checking, can also help reduce errors, the study's authors wrote. Errors such as wrong dose and frequency made up 7.5% of errors and 13.5% of potential adverse drug events.

Dr. Nanji said functionality is not enough. The systems have to be used properly. Even though specific functions are present in modern-day e-prescribing systems, prescribers don't have to use them. This has led to creating workarounds to avoid things like "alert fatigue," she said.

Meaningful use criteria require systems to have drug interaction checks, both drug-to-drug checks as well as allergy checks. Though the checks only have to be implemented in the systems to meet stage 1 requirements, proposed stage 2 criteria state that the systems must be employed and have the ability to be refined by physicians. But just meeting the criteria does not guarantee improved safety.

The authors said physicians need to take vendor selection more seriously and choose systems that are designed, or can be tweaked, to meet individual needs. "For example, while the [meaningful use] criteria require drug-drug interaction checking, there is no check of whether or not the key interactions are in place or whether the warnings when delivered adhere to human factors design principles," the authors said in the report.

The researchers said training is important. A vendor's training capabilities should be part of the selection. Physicians and their staff need to learn how to use the system to improve safety. Constant monitoring of errors is also needed to address issues by upgrading or changing systems. Changes and upgrades to systems might be needed as meaningful use requirements change.

Vendors who are "unwilling to commit to resolving problems as they arise through system revisions" should be eliminated, the authors wrote.