Reducing Co-payments for P2Y12 Inhibitors Upped Ticagrelor Use but Didn’t Budge MACE: ARTEMIS

ORLANDO, FL—Giving patients vouchers to cover co-payments for a P2Y12 inhibitor post-MI leads to more patients taking the pricier but more potent ticagrelor instead of clopidogrel, new data from the ARTEMIS trial show. Not only were physicians in the study more likely to prescribe the more effective antiplatelet, but also patients were more apt to continue taking their pills over 1 year of follow-up.

Better adherence did not, however, help reduce the MACE risk at 1 year, Tracy Y. Wang, MD (Duke University Medical Center, Durham, NC), reported at the American College of Cardiology (ACC) 2018 Scientific Session here. This was “unexpected,” Wang told TCTMD. “It was a learning experience for us.”

Around 30% to 60% of patients in the United States stop taking P2Y12 inhibitors before the 1-year mark, Wang told the media during an ACC press briefing. “Affordability is often cited as a reason” not only for early cessation, she said, but also for the fact that clopidogrel is more widely prescribed than ticagrelor (Brilinta; AstraZeneca) or prasugrel (Effient; Eli Lilly). The hope was that “by reducing and equalizing the out-of-pocket costs for generic and brand-name antiplatelet agents that we would have treatment decisions that are driven more by evidence than by what we think the patient can afford,” Wang explained.

Taking the cost piece out of this allows us to do the best thing for our patients. Tracy Y. Wang

Notably, “the biggest impact was on us as clinicians,” she observed, adding, “We all want to do the right thing, but we’re constrained by this cost concern. That tells us that taking the cost piece out of this allows us to do the best thing for our patients.”

Taking Money Out of Decision-Making

Using a cluster-randomized design, Wang and colleagues enrolled 11,001 MI patients at 301 US hospitals. All had health insurance (64% private, 42% Medicare, and 9% Medicaid), though 17% said they had previously avoided filling a prescription due to concerns about drug cost. Within the study, the choice of P2Y12 inhibitor, whether clopidogrel or ticagrelor, was left to the patient’s clinical team. Hospitals were randomly assigned to continue usual care or to give patients vouchers that would cover the co-payment for their prescribed drug for 1 year.

Clinicians practicing at hospitals in the intervention arm were less likely than those in the usual-care arm to prescribe clopidogrel (36.0% vs 54.7%) and prasugrel (4.4% vs 12.9%) and more likely to prescribe ticagrelor (59.6% vs 32.4%; P < 0.001).

For patients, continued use of the prescribed antiplatelet drug at 1 year (without a gap in use of ≥ 30 days) was more common in the voucher group, which had significantly less “nonpersistence” compared with the usual-care group (adjusted OR 0.84; 95% CI 0.72-0.98). On the other hand, the rate of MACE, defined as MI, stroke, or all-cause death, did not differ at 1 year for patients at hospitals assigned to provide vouchers or usual care (10.17% vs 10.63%; P = NS).

As-treated analyses hinted, though, that vouchers could be beneficial. The gap in nonpersistence favored the voucher group more strongly (adjusted OR 0.65; 95% CI 0.55-0.78), and the voucher arm had a lower crude MACE rate (7.49% vs 10.63%; P = 0.0001). However, the difference in MACE was no longer significant after adjustment.

Wang called this finding “encouraging,” despite being underpowered, noting that MACE curves began to diverge based on voucher use.

In another wrinkle, patients who were given vouchers but never used them had the highest rates of nonpersistence and MACE within the overall ARTEMIS population. Of note, the study was designed to allow participants to use the vouchers with any insurance payer and at the pharmacy of their choosing. On the other hand, Wang noted, this placed the onus on patients, requiring them to carry in their vouchers when filling prescriptions.

Getting That Horse to Drink

Speaking to journalists, Craig J. Beavers, PharmD (University of Kentucky Healthcare, Lexington), said, “ARTEMIS emphasizes the complexity of medication adherence.” He called for a “multipronged approach” that goes beyond vouchers and for better understanding of what barriers discouraged their use here. “The real trick is, even if you take a horse to the water, how do you get them to drink it?” Beavers asked.

As to why ARTEMIS did not affect clinical outcomes, Wang said one explanation is that the study looked at only one drug class. “We know that we discharge our patients on a number of really important evidence-based therapies,” so a single drug alone may carry relatively little weight, she observed.

Also, it’s unclear why 28% of patients given vouchers did not actually use them, Wang added. “Would this intervention have been more effective if instead of saying, ‘Here’s a voucher. Go use it,’ we implemented it on the back end? When you go to the pharmacy to fill this really important medication, we’re just going to waive your co-pay costs . . . so that 100% of patients get it as opposed to 72% of patients.”

Another tactic, Beaver suggested, would be to target the patients at highest risk of being nonadherent and of having events.

Laura Mauri, MD (Brigham and Women’s Hospital, Boston, MA), commenting after Wang’s presentation in the main arena, said ARTEMIS represents a great example of an attempt to “tackle the cost barrier.” Its cluster design, which involved randomizing at the hospital level rather than on a per-patient basis, likely sped up the trial but meant that some differences in baseline factors required adjustment, she pointed out.

“We selected the cluster-randomized design because it allowed us to look not only at the clinical outcomes [for patients] but also be able to look at physician prescribing behavior, which we would not have been able to do in a traditional patient-randomized study,” Wang responded.

In a trial where patients learned after randomization that they didn’t qualify for free medications, the dropout rate might spike in the usual-care arm, she said. Further, Wang added, because centers participating in ARTEMIS were encouraged to consecutively enroll patients, “the imbalances were actually fairly modest between the two groups,” with slightly lower proportions of women and those with private insurance in the voucher arm. There also was a nonsignificant trend toward fewer nonwhite subjects within that group.

Panelist Glenn N. Levine, MD (Baylor College of Medicine, Houston, TX), praised the ARTEMIS investigators for having the “noblest goals” among the day’s late-breakers. The trial did not seek to “get patients to use drug X over drug Y or device Z, but rather to actually take their medicines,” he commented. “We can write all the guidelines we want, but if the patients don’t take their medicines, then it’s really an academic exercise.”