To improve Part D adherence and raise PDC, identify the members who are about to fall below 80 percent proportion of days covered before they actually do, then take a real action on each one: confirm the prescription is active, get the next refill out of the pharmacy, chase the prescriber for a lapsed script, and remove the specific barrier keeping the member from filling. The plays are well known. The hard part is running thousands of them on time, every week, which is exactly the work an execution layer does and a dashboard does not.

This post covers the operational plays that actually move PDC. If you need the underlying definitions first, the Part D adherence measures explainer covers what each measure is, and the PDC math guide walks through the calculation step by step.

What PDC is, and why 80 percent is the line

Proportion of days covered (PDC) is the share of days in the measurement year that a member had medication on hand for a given drug class. A member who is supplied for 292 of 365 days has a PDC of 80 percent. The Pharmacy Quality Alliance (PQA) set 80 percent as the adherence threshold for the Part D measures, the level at which a chronic medication is likely to deliver its clinical benefit (pqa.org). The threshold is binary at the member level: a member at 79 percent counts exactly the same as one at 20 percent, which is to say, not at all. That cliff is why the days-to-fall-below math below matters so much. Full detail is in the PDC math guide and the PDC glossary entry.

The three triple-weighted adherence measures

Part D has three medication adherence measures, and they are the highest-leverage measures on the Star Ratings scorecard:

  • Diabetes medications (non-insulin antidiabetics), measured as PDC-DR.
  • Hypertension medications (renin-angiotensin system antagonists, or RAS antagonists), measured as PDC-RASA.
  • Cholesterol medications (statins), measured as PDC-STA.

Each is the percentage of eligible members who hit a PDC of 80 percent or higher for that class. In a normal year all three carry a weight of 3 in the CMS Star Ratings, the highest weight any outcome measure gets, which is why they move a contract's overall rating more than nearly anything else (cms.gov). One scheduling note: for measurement year 2026 CMS temporarily reduced these measures to single weight, and they are expected to return to triple weight in 2027. That timeline does not change the operational work, and the contracts that build the outreach muscle now are the ones that hold their rating when the weight goes back up. The 2027 Star Ratings changes post covers the schedule, and the triple-weighted glossary entry defines the term.

The operational plays that actually move PDC

Five plays do the work. None of them are a report. All of them end in a completed action on the phone.

1. Find members about to fall below 80 percent, early

The single most important play is timing. For every member, you can compute how many more covered days they can afford to miss before 80 percent becomes mathematically impossible for the year. A member who is at 85 percent in June with a chronic statin has a cushion; a member at 81 percent in October with one refill left is one missed fill away from falling off the measure permanently. Working the worklist by days-to-fall-below, not by alphabetical order or by who called in, is what protects the rate. Late in the year, some members can no longer reach 80 percent no matter what you do, and the right move there is to stop spending outreach on a lost measure-year and protect next year instead.

2. Proactive refill outreach before the gap opens

Most adherence loss is not a decision to stop. It is a refill that slipped: the member ran out, did not get to the pharmacy, or the auto-refill lapsed. Reaching the member a few days before the supply runs out, confirming the refill is ready, and getting it picked up or delivered closes the gap before it ever counts against PDC. This is cheaper and far more effective than chasing a member who is already 30 days dark.

3. Prescriber follow-up for lapsed scripts

Some gaps are not the member's fault at all. The prescription expired, the refills ran out, or a prior authorization stalled. No amount of member outreach fixes a script that no longer exists. Closing these means going back to the prescriber, getting the renewal or the PA resolved, and confirming the new script reached the pharmacy. This is the step most adherence programs skip, because it requires actually calling the practice.

4. 90-day fills and medication synchronization

Two structural moves quietly raise PDC for whole panels. Converting eligible members from 30-day to 90-day fills removes two-thirds of the refill events where adherence leaks. Medication synchronization aligns a member's chronic medications to a single monthly pickup, so one trip covers everything instead of three. Both reduce the number of moments where a member can fall off, which is the surest way to lift a population's PDC.

5. Address the real reason, in the member's language

When a member is non-adherent by choice, the reason is almost always concrete: the copay is too high, they have no ride to the pharmacy, or a side effect scared them off. A generic reminder call does nothing for any of these. Lowering the cost (a generic switch, a 90-day mail-order price, a copay program), arranging delivery, or routing a side-effect concern back to the prescriber are the actions that change behavior. And the call has to happen in the language the member actually speaks, or it does not land at all.

What an AI execution layer does that a dashboard does not

Most adherence tools are analytics: they score PDC, rank members by risk, and hand a coordinator a list. The list was never the problem. A team of six managing 100,000 members cannot make thousands of pharmacy, prescriber, and member calls a week by hand, so the list goes stale and the members at the bottom of it fall off the measure.

An execution layer does the calls. Pelica's voice AI agents call the pharmacy to confirm a fill is ready, call the prescriber's office for a lapsed script or a stalled prior authorization, and call the member in their own language to clear a cost or transport barrier. The agents follow up across days until the gap is actually closed, log every call, and escalate to a live human only when judgment is genuinely required. The coordinator stops dialing and starts handling the handful of cases that truly need a person.

Working pharmacies and prescribers autonomously this way, Pelica reached 96 to 98 percent medication adherence at its flagship physician-led IPA in New York. That is the difference between knowing who is about to fall below 80 percent and doing something about every one of them.

96 to 98%
Medication adherence reached at a physician-led IPA in New York, working pharmacies and prescribers with voice AI
80%
PDC threshold per member, per class, set by the Pharmacy Quality Alliance
3x weight
The normal Star Ratings weight on each of the three adherence measures

How to choose adherence tooling

If the goal is to raise PDC and not just watch it, the tool has to take the action, not just surface it. The questions that separate an execution layer from a dashboard are:

  • Does it compute, per member, the days-to-fall-below 80 percent, and prioritize outreach by that, not by a static risk score?
  • Does it actually place the calls, to the pharmacy, the prescriber, and the member, or does it hand a coordinator a list to dial?
  • Can it reach members in their own language and clear concrete barriers (cost, transport, side effects), not just send a reminder?
  • Does it follow up across days until the gap is closed, and escalate to a human only when needed?
  • Does it close the prescriber loop on lapsed scripts and stalled prior authorizations, not only the member loop?

If the honest answer is "no, but it shows me a report," the tool is naming the problem, not solving it. A fuller version of this evaluation lives in the Part D adherence software comparison, and the Pharmacy and Quality solution page shows how this runs alongside the rest of a VBC operation.

Sources and further reading