PHILADELPHIA - Computerized physician order entry (CPOE) systems normally raise a red flag whenever clinicians attempt to order both warfarin (an anticoagulant, also known as Coumadin) and trimethoprim-sulfamethoxazole (an antibiotic, also known as Bactrim or Septra) on the same patient. According to Brian Strom, MD MPH, chair of Department of Biostatistics and Epidemiology at the University of Pennsylvania School of Medicine, the two agents constitute "one of the very, very few drug interactions that have been shown to matter." Patients who take both medications are much more likely to develop complications of bleeding.
Like any software, explains Strom, CPOEs are not perfect; they have bugs in the system, and they can be awkward. Their goal is to help physicians avoid dangerous dosing mistakes and drug interactions. But, they can produce so many red flags that doctors get "alert fatigue" and essentially ignore the warnings.
Strom assumed a more assertive alternative would be more effective. So, he and his colleagues modified the CPOE so that it would generate a "nearly hard stop" if a doctor attempted to administer both these drugs to the same patient. (A hard stop blocks the order from going through; a soft stop simply issues an alert; a "nearly hard stop" blocks the order unless the clinician certifies by a deliberately inconvenient method that the pairing was required.)
Nearly 2,000 clinicians were randomized to receive either the new "nearly hard stop" intervention or the existing pharmacist intervention. Of some 8,826 prescriptions for either medication these clinicians ordered, 342 generated alerts; 57.2% of the alerts in the intervention group resulted in a "desired response" (that is, the clinician did not reorder the drug within 10 minutes of the alert) compared to 13.5% in the control group – a four-fold improvement. They describe their findings in this week’s Archives of Internal Medicine.
“The intervention worked," concludes Strom, but perhaps too well. After just six months the University's Institutional Review Board pulled the plug because four patients in the "nearly hard stop" group suffered the "unintended consequence" of delayed treatment – that is, they didn't receive either warfarin or trimethoprim-sulfamethoxazole when they needed it, although none were harmed as a result.
Currently, the Hospital of the University of Pennsylvania and Penn Presbyterian Medical Center have in place a CPOE system in which ordering both medications triggers a call from a pharmacist to determine if the prescription should be changed. Nevertheless, he says, some 300 patients a year receive both medications, which sometimes is warranted, for instance in HIV patients.
The irony, says Strom, is that the IRB initially rejected this study because it was so obvious the intervention would be effective, they felt it would be unethical to subject patients to the control intervention - the pharmacist intervention normally used.
"These results illustrate the importance of formal evaluation and monitoring for unintended consequences of interventions intended to improve prescribing," the authors conclude.
Put another way, says Strom, CPOEs cannot simply be installed and assumed to work; like all medical treatments, they need to be tested. "We need to evaluate these things as interventions because they can have side effects," he says.
The study was funded by the University of Pennsylvania Health System and the Agency for Healthcare Research and Quality.