Reducing warfarin medication interactions

From Clinfowiki
Jump to: navigation, search

Dr Feldstein et al published an interrupted time series controlled trial to determine the effectiveness of computerized decision support's warfarin-drug interaction alerts.


The study took place in a Health Maintenance Organization (HMO) where an ambulatory electronic medical record system was already implemented. The participants included primary care providers (239) from primary care clinics (15) and patients who were taking warfarin (9910). The targeted warfarin-drug interaction alerts included warfarin and the following interacting medications: acetaminophen, fluconazole, NSAIDs (non-steroidal anti-inflammatory drugs), metronidazole and sulfamethoxazole. The primary outcome measured to determine effectiveness was the “interacting prescription rate” which was defined as the “number of coprescriptions of warfarin-interacting medications per 10,000 warfarin users per month”.


All 15 clinics received the electronic warfarin-drug interaction alerts and, in addition, 7 clinics were randomized and received academic detailing. The preintervention baseline data was captured from January 2000 through November 2002. Development of the electronic warfarin-drug interaction alerts and completion of the academic detailing to the randomized clinics took place from December 2002 to March 2003. The postintervention period measured was from April 2003 to August 2004, during which time all of the electronic warfarin-drug interaction alerts remained active.

The researchers found that the electronic warfarin-drug interaction alerts significantly changed the interacting prescription rate trend from an increasing rate of 1.1 (preintervention) to a decreasing rate of 21.3 (postintervention) with the biggest reduction ocurring within the first month postintervention and the main contributor being from the effect of warfarin-acetaminophen alert. On the other hand, they found there was no statistical significance in the postintervention interacting prescription rates between the group that received electronic warfarin-drug interaction alerts plus academic detailing and the group that received electronic warfarin-drug interaction alerts only.


They went on to conclude that the electronic warfarin-drug interaction alerts were associated with an immediate and continuing reduction, with ~15% relative reduction in the overall rate at 12 months. They also noted that coprescribing of warfarin and these interacting medications is not necessarily contraindicated in all situations and could be a reason why rates did fall further.

The authors did provide some useful comments about the high incidence of coprescribing of warfarin with interacting medications in patients in the outpatient setting (more than one-third were prescribed 1 of the 5 study medications) and appropriately discussed the controversy over the validity of the warfarin-acetaminophen interaction. They also noted that the warfarin-drug interaction alerts comprised the majority of the electronic alerts presented to the clinicians at this organization and thereby possibly increasing clinician sensitivity to respond to the alerts in comparison to other EMR systems where many electronic interaction alerts are displayed and clinician sensitivity may be lower.


A fairly well designed interrupted time series study design which reconfirms the high incidence of the coprescribing of warfarin with interacting medications in patients in the outpatient setting. Unfortunately, the authors did not design the study to measure the effectiveness of the electronic warfarin-drug interaction alerts and academic detailing on patient outcomes. Because it is often clinically necessary for warfarin to be coprescribed with the study medications outlined, the appropriate intervention may be to adjust the warfarin dosing regimen or to increase INR monitoring, not necessarily avoid the interacting therapies all together. Because these details were not captured as part of the study design, pre and postintervention patient outcomes were not measured and a controversial warfarin-acetaminophen interaction accounted for a major portion of the relative rate reduction, the study results are not necessarily generalizable and are of limited value if attempting to assess actual reduction in patient harm from warfarin-drug interactions through the use of targeted electronic alerts.