Blog Post

How MetroHealth reduced alert firing by almost 60%—and how you can, too

January 31, 2018

    Clinicians and hospital leaders across the country recognize that low-value, interruptive alerts are a problem. Unfortunately, the conventional wisdom for solving this problem is to remove alerts one at a time—an approach that is both resource- and time-intensive. 

    Leading organizations are taking a different approach: Turning off entire categories of low-value alerts. Keep reading to see how MetroHealth, an academic medical center located in Cleveland, Ohio, removed whole categories of alerts to reduce the alert firing rate on drug orders by almost 60%.

    MetroHealth pre-emptively minimized alert burden by removing alerts before system go-live

    In 2013, MetroHealth was preparing to roll out their new Medi-Span drug database system and wanted to ensure it wouldn't overburden clinicians with alerts. So, hospital leaders formed a team to identify and pre-emptively cut unnecessary alerts before going live.

    The team of three physician informaticists and one clinical pharmacist ran the new drug database system silently—i.e., it wasn't firing alerts to clinicians—for three months to identify categories of drug alerts that could be safely removed. They found that the system fired an alert on 12% of drug orders when using default settings (104,098 alerts fired on 834,911 drug orders).

    Data analysis reveals categories of alerts that can be safely removed

    The team's goal was to reduce the number of times alerts were fired. Rather than removing alerts individually, they looked for entire categories of alerts they could turn off. When the team reviewed three months of data, they realized they could remove alerts related to incomplete information and minimum dose.

    Next, the team recognized that some of the default thresholds for drug overdose alerts were too low, causing clinicians to receive unnecessary overdose alerts. After reviewing current literature and internal practice on drug dose and frequency thresholds, they found that single and daily drug dose thresholds for all drugs could be increased by 25%, and dose frequency thresholds could be increased by two doses per day.

    The team found that making these categorical changes to the drug database system would reduce alert firing by almost 60%.

    Drug-specific changes offer further opportunity to reduce alert firings

    In addition to the categorical changes to drug alerts, the MetroHealth team identified further opportunities to reduce unnecessary alert firings by focusing on specific drugs. They leveraged pharmacy expertise to safely alter the maximum dose threshold on 20 of the most frequently alerting drugs. These drug-specific changes to sensitivity and specificity reduced alert firings by an additional 40%.

    Post-study analysis shows modified alert system is safe and effective

    To ensure the safety of the categorical cuts and drug-specific changes, the team studied the modified alert system. The analysis showed that even though the modified alert system fired fewer alerts, it was as effective at identifying critical drug dosing errors.

    Realizing MetroHealth's success at your organization

    The changes MetroHealth made to drug dosing alerts occurred before alerts were turned on out of the box—but the same principles apply to scaling back high-volume alerts that have already gone live. Categorical cuts, coupled with drug specific tweaks, allow high-value alerts to stand out to physicians and are one key strategy to reduce alert fatigue and recapture clinician attention.

     

    Get our executive briefing on reducing care variation

    Download our research to learn 10 Insights on reducing care variation from pioneer health systems.

    Download Now

    Have a Question?

    x

    Ask our experts a question on any topic in health care by visiting our member portal, AskAdvisory.

    X
    Cookies help us improve your website experience. By using our website, you agree to our use of cookies.