Pharmacogenomics pharmacists at Sanford Health in Sioux Falls, South Dakota, were conducting an annual review of medication alerts in their clinical decision-support system when they spotted an area for improvement: Many of the more than 1,000 alerts for tricyclic antidepressants (TCAs) didn't apply to the patients due to dosing regimens.
The alerts were set to fire when clinicians ordered TCAs for patients with a genetic variation that impacts the metabolism of the medicine. The system used guidelines from the Clinical Pharmacogenetics Implementation Consortium (CPIC) based on higher doses typically used for the treatment of depression.
Yet these drugs are also used at lower doses for indications such as chronic pain, migraine, or insomnia, so the alerts lacking dose criteria wouldn’t necessarily apply, said Amanda Massmann, lead clinical pharmacogenomics pharmacist with Sanford Imagenetics at Sanford Health, the largest rural health system in the country. She said too many alerts can cause treatment delays or create alert fatigue.
Clinical Matters
The ASHP News Center is publishing a series of stories on new research helping shape pharmacy practice.
So Massmann and colleagues adjusted the seven existing alerts related to TCAs to display only if an ordered dose exceeded 25 mg, in line with CPIC guidelines and published literature on drug-drug interactions. The result? The number of alerts decreased by 75%.
A description of the work was published in AJHP.
Clinical decision support and alerts are important tools for prescribers who “may be very unfamiliar with how to use genetic information as part of their ordering behavior or patient assessment,” said study co-author Sarah Mills, a clinical pharmacist practitioner in pharmacogenomics at the Department of Veterans Affairs in Durham, North Carolina. Mills was a postgraduate year 2 resident at Sanford Health at the time the study was conducted.
But the authors said they wanted to make sure the alerts are meaningful.
“We don’t want to be causing additional alerts for a 3-mg dose when they’re using it for sleep,” said Massmann. “Studies show upwards of 95% of alerts are overridden, so if you’re going to take the time to implement them, take the time to analyze them as well, to make sure that they are functioning as you intend for your users.”
Clinical decision support for TCAs originally was implemented at Sanford Health in 2014 as interruptive, or pop-up style, alerts for inpatient and ambulatory care electronic medical records, based on the 2013 CPIC guideline. But Massmann and colleagues observed in an annual review that alerts related to TCAs represented 6% of all alerts (1,037 of 16,278 alerts).
To measure the impact of their intervention, they compared reports of all patients for whom a TCA alert was generated from June to August 2022 (before the alerts were amended) and from February to April 2023 (after they were amended). In the pre-amendment group, 127 alerts displayed in three months, and 63% of the alerts were for doses of 25 mg or less. After the alert refinement, just 32 alerts were generated, all for clinically relevant doses of greater than 25 mg. Alert volume decreased the most in areas other than behavioral health.
Massmann said clinicians in the health system’s gastroenterology group, who don’t routinely prescribe doses exceeding 25 mg, were appreciative of the changes and said they now don’t have as many interruptions.
Sanford Health staff have changed some alerts from pop-ups to an inline style that displays in yellow but doesn’t require action by users. Alerts also can be adjusted by dosing criteria, or specifically for inpatient or outpatient populations, said study co-author Natasha Petry, a pharmacogenetics clinical pharmacist with Sanford Imagenetics.
“Alerts don’t have to be a one-size-fits-all approach,” said Petry.
