BETTER CARE-HF: Embedded EHR Alert Boosts MRA Prescribing in HFrEF

Alerts were more helpful than messages and worked best in older patients and those seen by cardiologists or HF subspecialists.

BETTER CARE-HF: Embedded EHR Alert Boosts MRA Prescribing in HFrEF

NEW ORLEANS, LA—Tailored alerts embedded in electronic health records (EHRs) directing physicians to consider mineralocorticoid receptor antagonists (MRAs) in their ambulatory patients with heart failure with reduced ejection fraction (HFrEF) more than doubled the prescribing rates, according to results from the BETTER CARE-HF trial.

Up to two-thirds of patients who may be eligible for MRA therapy are not prescribed it. Closing the treatment gap could save over 20,000 lives per year in the United States alone, said Amrita Mukhopadhyay, MD (NYU Grossman School of Medicine, New York, NY), who presented the findings here at the American College of Cardiology/World Congress of Cardiology (ACC/WCC) 2023 meeting.

EHR tools can be a rapid, low-cost and high-impact method to increase prescription of lifesaving therapies across large populations,” she said.

But choosing which tools work best without interfering with physician workflow and patient engagement can be challenging. For BETTER CARE-HF, which was simultaneously published in the Journal of the American College of Cardiology, two specific tools embedded in EHRs were tested: alerts and messages. While alerts pop up during patient visits, messages are usually seen by the physician between patient visits, and may include information about multiple patients as well as take up more of the clinician’s time.

Commenting in a press conference following Mukhopadhyay’s presentation, Lee R. Goldberg, MD, MPH (Hospital of the University of Pennsylvania, Philadelphia), said the study will be gratifying for physicians who often feel like they enter information into the EHR, but see no results after spending their time and the patients’ time doing so.

It’s “absolutely critical,” said Goldberg, to do studies like this that shed light on how pop-up interventions impact physician workflow and whether they actually translate into improved patient outcomes.

Alerts Modifiable, Part of Larger Puzzle

The study was conducted over 6 months in 2022 across a single healthcare system that used the Epic EHR system (Verona, WI). There were 60 cardiologists in each of the three arms who enrolled 2,211 heart failure patients (median age 73 years; 30% female; 11% Hispanic). Median EF was 35%, median potassium level was 4.3 mmol/L, median estimated glomerular filtration rate was 67.0 mL/min/1.73m2, and median systolic blood pressure was 122 mm Hg.

Most patients (80%) were already on background beta-blocker therapy and 74% were on ACE inhibitors, ARBs, or angiotensin receptor-neprilysin inhibitors (ARNIs).

The primary outcome of newly prescribed MRA therapy at 6 months was seen in 29.6% of patients treated by cardiologists randomized to the alerts as compared with 15.6% of patients seen by those randomized to messages and in 11.7% of those randomized to usual care. That translated into a RR of 2.53 (95% CI 1.77-3.62) for the alert group versus usual care and an RR of 1.67 (95% CI 1.21-2.29) for the alert group versus the message group. The number needed to treat to elicit one new MRA prescription was 5.6 in the alert group and 25.6 in the message group.

For the secondary endpoint of new prescription of beta-blockers, ACE inhibitors, ARBs, or ARNIs, there were no differences between the alert, message, or usual care groups. In prespecified subgroup analyses, patients over age 65, and those being seen by general cardiologists or HF subspecialists saw the greatest effect of the alert versus usual care. Other patient-level factors, such as female versus male sex, Black versus white race, Hispanic versus non-Hispanic ethnicity, and public versus private insurance, showed no significant interactions by intervention.

Physicians randomized to the alert saw it at the top of the patient chart, among other alerts. It included real-time data on HF therapies the patient was or wasn’t taking, as well as any contraindications. The alert linked to a preselected order set that included orders for MRA prescription, other HF therapy prescription, laboratory tests, and advanced HF referral. Those who got the messages, on the other hand, received them monthly through the EHR and had to click a link to view real-time patient data.

As Mukhopadhyay explained, when the alert pops up, the physician has the option to accept it or give one of the following reasons for not accepting it: planning to address at a future visit; allergy/adverse reaction; pregnancy; hyperkalemia/hypotension/renal dysfunction; EF score not accurate; or other.

To TCTMD, Mukhopadhyay said the alerts are modifiable such that if a physician has a reason to not prescribe at a specific visit or point in time, such as if the patient is pregnant or has hyperkalemia, the alert can be set to pop up at a later time.

Another observation made during the trial, she said, was that many of the cardiologists appeared to be using the alerts to order lab work, presumably because they were considering putting the patient on MRA but wanted to check their potassium levels first.

Overcoming Inertia

Speaking with TCTMD, Goldberg said one thing that will be important over time is to measure whether the intervention “decays.”

“Initially there's a novelty factor where you may have a bigger impact, and then as months go by, you may see that the impact becomes less and less in terms of motivating clinical decision changes,” he said.

While ways to combat it are multifactorial and often specific to the intervention and the patient population, Goldberg said changing the wording might be effective, or even combining the intervention to encompass another guideline-directed medical therapy (GDMT) that the patient might be missing, such as a sodium-glucose cotransporter 2 (SGLT2) inhibitor.

According to Mukhopadhyay, the alerts were developed over time and included interviews with physicians using it to understand how the alerts affected their workflow. She said they are still gathering information on what users like and dislike about it as well as other data that can only be gathered through longer-term use.

Goldberg cautioned that the innovations that these embedded alert systems offer must be tempered with the reality that they are only a piece of a much larger puzzle.

“It’s so much more complex, . . .  and fixing or addressing this part is a good start, but it may not be enough to fully solve problems, real-world problems, that we know our patients are having and that can interfere with treatment,” Goldberg told TCTMD. “Let’s say we solve the issue of clinicians ordering the correct meds and doing the right thing, and then there's a cohort of patients still not getting the meds they need. Maybe it's a cost barrier, maybe it's a transportation barrier in getting to the pharmacy. We can’t overcome all of that but we can take these . . . incremental steps.”

“It's crucial that we deploy strategies to overcome inertia in use and dosing of GDMT,” commented Bhavadharini Ramu, MD (Medical University of South Carolina, Charleston), the discussant for Mukhopadhyay’s presentation. “This is a really cost-effective and scalable intervention to improve heart failure care.”

  • Mukhopadhyay reports no relevant conflicts of interest.