Outcomes data have been long heralded as a possible solution to high drug prices. By better understanding a drug’s benefit to patients, the prevailing thinking goes, the healthcare system can better understand its value.

In practice, this data are meant to inform contracts that tie rebates and discounts to drug performance in patients. And yet outcomes data are expensive to produce and difficult to interpret. To address that issue, pharmacy benefit managers are increasingly measuring a patient’s outcome in the context of adherence, according to ZS Associates managing principal Pratap Khedkar.

“Express Scripts has said, ‘I don’t have any way to measure outcomes, [because] even when payers have the data it is difficult to measure. I will use the patient’s behavior as a proxy for outcomes,’” Khedkar explained. He added that some PBMs have adopted a new stance along the lines of  “if you didn’t take your pills, of course you didn’t get better. But if you took your pills, I will credit you with a full outcome.”

See also: How pharma marketers are using behavioral science

Express Scripts spokesperson Jennifer Luddy confirmed that three of the company’s SafeGuardRx programs offer plan sponsors reimbursement when a patient stops taking their drug. She noted, however, that they stop short of a direct adherence guarantee. The company offers these programs for inflammatory conditions and oncology therapeutic areas right now, with plans to expand into multiple sclerosis next year.

New technology, such as artificial intelligence and predictive analytics, has helped PBMs better understand why patients aren’t taking their medication and personalize interventions that can prevent them from abandoning their treatment regimen in the first place.

Mark Bini, VP of new solutions at Express Scripts, said that the company uses these tools to identify the patients most likely to stop taking their medication. “We target patients in a proactive way by looking at the data and tailoring an intervention based on the root cause. As adherence increases, our downstream medical costs will decrease. Better adherence leads to fewer adverse outcomes,” he explained.

See also: Biogen, Merck use predictive modeling to better negotiate value-based contracts

In the past five years, there has been a significant interest among payers and PBMs in using predictive analytics to address adherence woes, said Clifford Jones, CEO of AllazoHealth.

“For some conditions, the impact is very clear on adherence correlating with outcomes. It can be used as a good proxy: ‘We’ve improved adherence rates by this much, that should translate into this range of dollars,’” Jones explained.

What’s behind the increased usage of new analytics tools in adherence? “If you wait until someone already becomes nonadherent, it’s much harder to change that behavior,” Jones suggested. “Being able to predict [nonadherence] ahead of time allows you to target patients with appropriate interventions.”

See also: Power to the payers: PBMs leave behind a trail of tiers

Khedkar said that pharma has long tried to boost adherence rates, but that these tools help them “get a lot of patient-level data to figure out when they’re going to drop off their regimen.”

To validate their offering, AllazoHealth and BlueCross BlueShield of North Carolina conducted a randomized controlled study to see if the company’s engine, which uses predictive analytics and AI, can boost adherence rates. Their research found that patients receiving interventions based on the company’s technology were 6% more likely to be adherent to their medications compared to the control group.

“That’s a pretty big jump,” Jones said, noting that the patients selected in the study had already refilled their medicine twice and had an overall adherence rate of close to 80%.  Patients in the study were either taking a statin, oral diabetes drug, or blood-pressure pill.

It’s less clear how effective certain interventions may be in getting patients back on a regimen. An August study published in JAMA Internal Medicine investigated whether wireless pill bottles, financial incentives, and social support could improve adherence rates among patients that had recently suffered a heart attack. Researchers found that this comprehensive approach did not significantly improve patients’ ability to consistently follow a four-drug prescription comprised of a statin, aspirin, beta-blocker, and antiplatelet agent.