Problem Addressed

• A common, preventable problem: Each year at least 1.5 million preventable ADEs occur in the United States,¹ mostly in outpatient settings (where the prevalence of such errors has been found to be as high as 27 percent.²)
• Causing great harm: Each year, more than 770,000 injuries or deaths occur in hospitals due to ADEs,³ with many additional injuries and deaths in outpatient settings.
• Failure to heed BBWs: Physicians continue to prescribe medications in violation of BBWs, due in part to lack of awareness that the proposed prescription conflicts with guidance contained in the warnings. A study of 51 ambulatory practices in the Boston area found that roughly 7 percent of prescriptions for drugs with a BBW violated the warning.⁴
• Unrealized potential of alerts electronic ordering systems: Multiple studies suggest that revamped electronic ordering systems that make use of alerts can reduce prescribing errors in hospitals.^5,6,7 Yet relatively little experience exists with such systems in outpatient settings.^8,9

Description of the Innovative Activity

• Broad universe of alerts: Based on a knowledge base created by expert panels of physicians and pharmacists, the alerts covered situations judged to be clinically relevant to the ambulatory care setting, including all BBWs related to drug–pregnancy interactions, most related to drug–drug interactions and the need for laboratory test monitoring, and some related to drug–disease interactions.
• Tiered alerts based on potential consequences: The system reviewed each patient’s active medication list, problem list, and laboratory results, using applied logic rules to identify potential contraindications. Different types of alerts appeared based on the potential for harm, with more important warnings listed first. The BBW warnings appeared alongside other warnings already embedded within the CDS system, with no specific notation distinguishing BBW-based alerts from others. More details on the various levels of alerts appear below:
  
  • Intrusive popups for life-threatening, other serious situations: If the prescribed drug could lead to a fatal interaction or serious injury (deemed “level 1” and “level 2” alerts, respectively), the system used popup boxes that displayed the alert clearly and required the doctor to take a specific action before proceeding.
    
    • Level 1 alerts: For potentially fatal situations, the system did not let the doctor continue without taking specific action. For example, a physician ordering isotretinoin received a popup warning about the potential for death or serious injury for patients who are pregnant or of childbearing potential. The physician could not proceed without entering the date of the last pregnancy test and attesting to the fact that the patient was not of childbearing potential, or was not pregnant and had been made aware of the risk of using this drug. The alert also included a reminder to document a negative pregnancy test within 24 hours of initiation of therapy and every month during therapy. In essence, these level 1 alerts represented a “hard stop” that prevented the doctor from violating the BBW.
    • Level 2 alerts: These alerts covered situations with the potential for serious injury. Physicians could continue with the order only if they documented a legitimate reason for doing so. For example, a physician ordering the drug ritonavir to a patient already taking a certain dosage of cisapride received an alert warning of potentially fatal cardiotoxicity, along with instructions to discontinue one of the drugs. The doctor could proceed with the order only if he/she documented a reason for the override, such as adjusting the dosage of one of the drugs to a safe level, agreeing to monitor as recommended, noting that the patient already tolerated this combination of drugs or that no reasonable alternatives existed, or providing another reason.
  • Informational, unobtrusive warnings for less serious situations: For situations with little or no potential for immediate harm, the system embedded an informational alert requiring no specific action within the patient record, as outlined below:
    
    • Level 3 alerts: These alerts reminded the physician to collect certain information in conjunction with prescribing the drug, usually laboratory test data. For example, a doctor initiating clozapine received a reminder to collect baseline information on white blood cell count.
    • Level 4 alerts: The alerts reminded physicians to monitor certain laboratory test results on an ongoing basis after initiating the drug. For example, the system reminded the doctor to run a serum creatinine test for patients on metformin for more than a year.

References/Related Articles

Contact the Innovator

Innovator Disclosures

Results

• No impact on overall likelihood of BBW violation: Post-implementation, 5.1 percent of the 56,869 orders violated a BBW, slightly higher than the 4.8-percent rate in 28,359 orders before implementation.⁴
• Positive impact in specific areas: The alerts reduced BBW violations related to potential drug–drug and drug–pregnancy interactions (particularly high-priority alerts in these areas) and tended to be more effective with primary care clinicians and in clinics outside the hospital.⁴
  • Fewer potential drug–drug and drug–pregnancy interactions: The proportion of orders violating a BBW related to drug–drug interactions fell from 6.1 to 1.8 percent after implementation. Similarly, the proportion violating a BBW related to drug–pregnancy interactions fell from 5.1 to 3.6 percent. Both declines met the test of statistical significance. In general, reductions tended to be concentrated in level 1 and 2 alerts. (As noted, the system prevented physicians from violating level 1 alerts.)
  • Positive impact in primary care and nonhospital settings: Primary care physicians prescribed fewer drugs in violation of BBWs after implementation, while specialists prescribed more. Similarly, physicians working in clinics outside the hospital had fewer violations after implementation, while those in hospital-based clinics had more.