The same automated systems intended to prevent dangerous drug combinations may actually lead to potentially harmful treatment delays, according to a new study.
The same automated systems intended to prevent dangerous drug combinations may actually lead to potentially harmful treatment delays, according to research published in the Archives of Internal Medicine.
These findings provide a compelling example of why electronic systems need to be evaluated in the real world before they are put into widespread use. "We need to study the side effects of these interventions and make sure they do more good than harm,” said Brian L. Storm, of the University of Pennsylvania School of Medicine, in a Reuters Health article.
In a study conducted at two academic medical centers in Philadelphia, PA, Storm and colleagues evaluated the effectiveness of a computerized physician order entry (CPOE) prescribing alert intended to reduce concomitant orders for warfarin and trimethoprim-sulfamethoxazole. Nearly 2,000 clinicians were assigned to either an intervention group receiving a nearly hard stop alert—meaning the physician prescribing the drug combination could not continue the prescribing process without calling the pharmacy—or a control group receiving the standard practice (where a warning pops up on the computer screen that can be easily overridden).
Of the providers who were given the hard-stop alert, 57.2% did not proceed with prescribing the dangerous drug combination within 10 minutes of the alert, combined with 13.5% in the control group. However, the researchers note, the study was terminated prematurely because of “unintended consequences identified among patients in the intervention group”—a delay of treatment with trimethoprim-sulfamethoxazole in two patients and a delay of treatment with warfarin in another two patients.
The researchers concluded that although “an electronic hard stop alert as part of an inpatient CPOE system seemed to be extremely effective in changing prescribing,” the intervention triggered “clinically important treatment delay” in four patients who required immediate drug therapy. These results, they said “illustrate the importance of formal evaluation and monitoring for unintended consequences of programmatic interventions intended to improve prescribing habits.”
In an editorial accompanying the study, David W. Bates, MD, of Brigham and Women's Hospital in Boston, MA, noted that the delays in trimethoprim-sulfamethoxazole prescriptions were “especially concerning” given that "this is the single best drug for treating some serious specific infections." The risk to these patients from not receiving the best treatment as quickly as possible, he added, was greater than the risks of the potentially harmful warfarin-antibiotic combo.
The delays likely occurred because providers got distracted in the middle of the prescribing process and didn't follow through with a call to the pharmacist, said Strom in the Reuters Health article. Adjusting the system to make the process less complicated could potentially reduce the risk of these delays.
Currently, many hospitals use CPOE systems that include several alerts, not all of which are clinically necessary. "There are millions of different adverse drug interactions, very, very few of which have been proven to be of clinical importance," he said. This can lead to alert fatigue, he noted.
Hard alerts like the one in the current study "should be used extremely judiciously," Strom added, in situations where there is no alternative; for example if a doctor prescribes a pregnant women isotretinoin, which can cause birth defects.
CPOE systems are going to be a major focus of the $50 billion investment in health information technology planned by the US government, Bates said. "If the nation is to realize the desired benefits from the enormous investment in HIT, it will be critical to develop answers to many smaller questions, such as those posed in this study, and to share them widely.”