Computerized Drug Warning Alerts Remain Largely Ignored

A new study finds that 96% of computerized drug warning alerts generated in hospitals are overridden out of hand.

Researchers at the Johns Hopkins Bayview Medical Center (Baltimore, MD, USA) examined data from the 410-bed academic medical center to clarify the factors associated with accepting or overriding a warning from the hospital computerized provider order entry (CPOE) system, developed to generate warnings when there is the possibility of allergy, adverse events, interactions, and duplications, with the intent of increasing inpatient safety. To do so, they identified 6,646 patients - mean age 57, 44% male, with a mean length of hospital stay of four days. When differentiated, 45% were on medical wards, 29% in psychiatric departments, 28% on surgical wards, , and small numbers in other departments such as obstetrics and gynecology.

The researchers found that for the patients whose medication orders generated at least a single warning, the median range of medications per patient was four. The reasons for the warnings were duplications in 47%, potential interactions in 47%, allergies in 6%, and only 0.1% for potential adverse reactions. One-quarter of the warnings were for parenteral medications, and 13% were on the list of high-alert medications. While only 4% of the warnings for oral medications were accepted, warnings for parenteral medications were accepted twice as often, at 8%. The study was presented at the Society of Hospital Medicine (SHM) annual meeting, held during April 2012 in San Diego (CA, USA).

“The least common warnings were those that were most frequently accepted, with the average accepted warnings for drug interactions and allergy being 37% and 10%, respectively,” said lead author and study presenter Amy Knight, MD. “The problem with these computerized warning systems is that providers develop 'alert fatigue' and end up overriding many potentially important warnings.”

“Developers of these computerized provider systems need to revise their programs to more clearly differentiate between types of warnings and, in particular, to distinguish the warnings that are most likely to have severe consequences,” concluded Dr. Knight.

Related Links:
Johns Hopkins Bayview Medical Center