CDIX program makes alerts more meaningful

Alerts limited to those deemed critical

Electronic drug interaction alerts can be useful in preventing harmful drug-drug interactions, but too many clinically insignificant alerts can lead to "alert fatigue" and clinically significant alerts may be overridden.

Failing to detect significant drug-drug interactions can lead to adverse health outcomes for patients and increased health care costs. Some analysts say drug-drug interactions are responsible for up to 2.8% of hospitalizations each year, with an estimated annual cost of $1.3 billion.

To minimize the risk of bypassing significant alerts, the Institute for Safe Medication Practices recommends using a tiered system that requires staff to make a text entry to describe their response to more significant alerts. It also recommends a regularly updated list of significant alerts that require direct prescriber notification.

Pharmacists at the University of Colorado School of Pharmacy and Kaiser Permanente Colorado developed an electronic critical drug interaction alert program (CDIX) and evaluated it when coupled with active intervention to prevent dispensing of critically interacting drug combinations. The researchers hypothesized that, compared with a baseline period, the rate of co-dispensing of drugs that critically interact would be lower in the health care system after introducing CDIX.1

Elizabeth Chester, PharmD, a clinical assistant professor at the University of Colorado Pharmacy School and quality and clinical pharmacy manager in the Kaiser Permanente Colorado Pharmacy Department, tells Drug Formulary Review CDIX was developed by a multidisciplinary team.

"Before its implementation," she says. "outpatient pharmacists relied upon a passive drug interaction alert system to warn them about drug interactions of varying severity. An electronic alert notified the pharmacist of a drug-drug interaction, but the alert could be easily bypassed or ignored. CDIX was developed to intercept critical drug interactions after a prescription is ordered but before the medication is dispensed. An electronic medical record and the electronic pharmacy information management system were used to identify when newly prescribed medications interact with any previously prescribed medications."

Pharmacist must respond to alerts

With CDIX, when a potential critical drug interaction is detected, an alert is triggered to notify the pharmacist, the prescription label is not printed, and the pharmacist must consult with the prescriber. The planners developed a decision-support guide to help pharmacists interpret and resolve critical alerts. The guide details alternative therapies to discuss with the provider when critical drug interactions occur.

For the study, the primary outcome measure used to assess the impact of the CDIX intervention was the proportion of subjects co-dispensed two critically interacting drugs. To avoid including interacting drug combinations taken chronically, dispensed prescriptions were included only if the subject was not exposed to the medication in the previous 180 days for non-antibiotics and 75 days for antibiotics.

"During the study period," Chester says, "623 instances of co-dispensing interacting drug pairs were observed in 555 subjects. Of those, 367 instances of co-dispensing were observed in 348 subjects during the 20-month pre-CDIX period, and 256 instances of co-dispensing were observed in 248 subjects during the 37-month post-CDIX period.

"This study showed that implementation of a CDIX that requires electronic documentation and provider consultation decreases co-dispensing of critically interacting drugs. A decrease in the co-dispensing rate of critically interacting drugs of approximately one-third was observed immediately after the CDIX implementation and persisted throughout the remainder of the study."

She points out that knowledge of drug interactions does not guarantee timely recognition or appropriate intervention. Managing drug interactions is highly dependent upon an effective process for adequately informing providers. Also, as alert fatigue has been identified as a problem, selective identification of which drug interactions should be labeled "critical" is also important.

Panel identified critical interactions

Those components were built into the CDIX. Electronic alerts were limited to drug interactions deemed critical by a panel of pharmacists and physicians. Pharmacists were required to consult with the prescriber before dispensing the prescription, and the consultation had to be documented in the information management system before the alert could be overridden. Pharmacists also provided details to the prescriber on the drug-drug interaction and recommended alternative therapy.

Pharmacist review is essential, Chester says, to prevent providers from having to deal with an excess of unnecessary alerts.

She tells Drug Formulary Review that the vast majority of pharmacist-to-prescriber contacts to discuss a critical drug interaction alert were well received. "Some of our physician colleagues specifically commented that they preferred our pharmacist-managed intervention to a purely automated intervention alert at the point of prescribing," she says. "One of the reasons cited for this preference was that there are too many false alerts with the point of prescribing system leading to alert fatigue. Physicians often took the pharmacists' critical drug interaction alerts more seriously as their clinical relevance had already been vetted by another trained clinician and likely warranted intervention or enhanced monitoring."

She points out that implementing the CDIX in the Kaiser Permanente Colorado integrated health care system may have made communication between pharmacists and prescribers easier, but says implementation of a similar system should be possible in most health care systems.

Try to combine CDIX with existing program

Chester says interested institutions should consult with local pharmacy IT resources to determine if their existing drug interaction program could be modified to support a critical drug interaction program or if with additional programming a critical drug interaction program could be layered on top of their existing program. (Kaiser layered its CDIX on top of their pharmacy vendor-supported drug interaction screening program.)

Existing committees such as pharmacy and therapeutics, medication use, or medication safety committees may be appropriate venues to initiate discussions regarding whether an institution might benefit from developing a CDIX, she says.

Asked for her view of the critical success factors that made the CDIX work, Chester highlighted these elements:

  • Engage a multidisciplinary group of key stakeholders, including physician leaders, pharmacy leaders, frontline physicians and pharmacists, researchers, and information technology specialists, early in the development process for CDIX;
  • Maintain open communication and consultations with key clinicians such as physician department chiefs and clinical pharmacy specialists to help evaluate appropriateness of requests for additions to the program;
  • Attempt to limit included drug interactions to those that should never be used in combination and for which there is almost always a therapeutic alternative available;
  • Suggest scripting to support pharmacists in having conversations with physicians and patients about critical drug interaction alerts;
  • Educate and train pharmacists on new processes related to the critical drug interaction program and educate and train physicians to increase awareness of the program and the calls they could be receiving; and
  • Review annually all critical drug interaction monographs to ensure information is current and that drugs continue to meet criteria to be included in the program.

She says it is possible the definition of drug interactions will vary between institutions and/or practice settings. For example, she says, given her plan's ambulatory setting, there are drug interactions they are not likely to encounter, particularly with intravenous medications, and therefore would not be likely to incorporate into their local critical drug interaction program.

"Regardless of setting," Chester says, "there is also the potential for local stakeholders to influence the types of interactions that are elevated to the level of a critical drug interaction. Sometimes the available literature to support a particular drug interaction is limited or flawed. We engage physician department chiefs, clinical pharmacy specialists, and other clinical leaders to help us determine the clinical significance and relevance of these drug interactions for our setting."

Although this study demonstrated improvements in dispensing of critical drug-drug interactions, Chester concludes, the impact on adverse outcomes was not evaluated. As that was not the study's primary aim, sample size was not sufficient to evaluate clinical outcomes. But because the co-dispensing of critically interacting drugs may result in serious adverse events or hospitalization, decreases in co-dispensing rates should translate into improved health care for patients.

[Editor's note: Contact Dr. Chester by e-mail at]


  1. Humphries TL, Carroll N, Chester E, et al. Evaluation of an electronic critical drug interaction program coupled with active pharmacist intervention. Ann Pharmacother 2007;41:1979-1985.