Health Care Quality/Patient Safety

Computerized physician order entry needs further refinement to substantially reduce medication errors in primary care

When doctors use a computer to enter prescriptions for medications for hospital patients, so-called computerized physician order entry (CPOE), it prevents a substantial number of medication errors. However, the development of a more sophisticated CPOE system is needed to achieve this in outpatient primary care, concludes a study supported in part by the Agency for Healthcare Research and Quality (K08 HS11644).

When physicians at five adult primary care practices used a common CPOE system, they overrode 91 percent of drug allergy alerts and 89 percent of high-severity drug interaction alerts. Yet there were no adverse drug events (ADEs, injuries due to medication) in cases where the physicians observed the alert, and there were three ADEs among patients with alert overrides, a nonsignificant difference.

Two independent physician reviewers agreed with prescribers' decisions in 98 percent of the override cases, judging 36.5 percent of the 189 alerts invalid. In most cases, the reviewers found that physician overrides were because the patient was no longer taking the medication cited in the alert, the drug interaction was not clinically significant, the patient tolerated the drug(s), or the benefits of treatment outweighed the disadvantages. Physicians were nearly 18 times more likely to override alerts for renewals compared with new prescriptions.

These results suggest that CPOE designers need to identify and eliminate inappropriate alerts that physicians don't find credible, and they should change the threshold for generating alerts on renewals of medications that patients currently tolerate in combination, suggests Saul N. Weingart M.D., Ph.D., of Beth Israel Deaconess Medical Center. Dr. Weingart and colleagues calculated the override rate among 3,481 consecutive alerts generated by the CPOE system used by the five primary care practices. They selected a random sample of 67 alerts in which physicians did not prescribe an alerted medication and 122 alerts that resulted in a written prescription in order to identify factors associated with physicians' decisions to override a medication alert and whether the override resulted in an ADE.

See "Physicians' decisions to override computerized drug alerts in primary care," by Dr. Weingart, Maria Toth, M.D., Ph.D., Daniel Z. Sands, M.D., M.P.H., and others, in the November 24, 2003, Archives of Internal Medicine 163, pp. 2625-2631.

Return to Contents
Proceed to Next Article