Adherence conversations usually begin after a missed refill. That’s already late. By the time a refill is overdue, a share of patients were never in the count to begin with, because they never started the medication at all.

Primary Nonadherence Is the Blind Spot

In 2010, Fischer and colleagues published an analysis of 195,930 electronic prescriptions in the Journal of General Internal Medicine and found that 28% were never filled. Not filled late. Never picked up, with meaningful variation by drug class.

The mechanism is boring, which is why it persists. An e-prescription leaves the office and, from the prescriber’s side of the screen, the encounter looks finished. Nothing in the chart contradicts it. The patient nods in the room, goes home, thinks about the copay or reads the leaflet, and the prescription waits in a queue until the pharmacy purges it. No alert fires. At the next visit, the assumption in the room is that the drug has been on board for four months.

Eighty Percent Adherent Means Missing One Day in Five

The standard measure is proportion of days covered, and the threshold for counting a patient as adherent is a PDC of 80%. CMS uses that cutoff in the Part D Star Ratings adherence measures for diabetes medications, RAS antagonists, and statins.

Read the threshold carefully. A patient sitting at exactly 80% missed 73 days last year. They pass. They will appear as adherent in every quality report anyone runs. For a statin, that is probably survivable. For an anticoagulant it is a different conversation entirely, and PDC won’t distinguish between the two, because the measure treats every day as interchangeable and doesn’t care whether the 73 missed days were scattered or consecutive.

Fourteen straight days off apixaban and fourteen isolated missed days across a year both score the same. Only one of them lands someone in the hospital.

How PDC Gets Calculated, and Who Falls Out of It

The arithmetic is simple enough to do on paper. Count the days in the measurement period, count the days the patient had the drug on hand based on fill dates and days supply, and divide.

Item	Value
Days in the measurement period	365
Days with medication on hand	292
PDC	292 ÷ 365 = 80.0%
Days with nothing on hand	73

Now the part that matters for everything above. The adherence measures generally require at least two fills in the period before a patient enters the denominator at all. A patient who filled once and never returned, or who never filled, mostly isn’t in your adherence rate. They fell out before the measurement started.

So the measure used to police adherence is structurally blind to the 28%. Your PDC can improve while your actual population does worse, and nothing in the report will tell you.

Why People Stop

The reasons are mostly mundane. Cost, side effects nobody warned them about, a schedule that doesn’t fit the shape of their day, and a quiet suspicion that the drug isn’t doing anything.

Asymptomatic conditions are the hardest. Hypertension doesn’t hurt, so the medication has no visible payoff while the side effect is perfectly visible. From the patient’s chair, stopping produces no consequence they can detect for years. That is a rational decision built on incomplete information, and arguing with it as though it were carelessness gets you nowhere.

Then there is the question itself. “Are you taking your medications?” reliably produces a yes. “How many days in the last week did you miss it?” produces a number, and often an honest one, because the phrasing has already conceded that missing days is normal. The second question isn’t harder to ask. It just has to be the one on the intake form.

Where the Structure Has to Live

Somebody has to close the loop between the prescription and the bottle, and it can’t be whoever has a free afternoon. In practice that means checking fill data for new starts, a call around day ten, and a real reconciliation at least monthly for anyone on five or more medications. None of that is clinically difficult. All of it is operationally fragile, because it depends on someone noticing an absence.

Medicare funds a version of it. Chronic care management pays for monthly clinical staff time for patients with two or more chronic conditions, and medication reconciliation is squarely the kind of work those minutes are meant to cover. Practices that get results from it have a named person with those calls sitting on their calendar, working from a virtual care platform that flags who is overdue instead of depending on anyone to remember.

We spend a lot of time on this with practices, and the split is consistent. The programs that move adherence numbers have a person and a list. The ones that don’t have a policy document.

What Actually Changes the Number

• Verify the first fill, not the first refill. A new prescription with no fill after ten days is the highest-yield phone call in the whole workflow.
• Ask about cost before the patient has to raise it. People will abandon a medication rather than say they can’t afford it.
• Name the likely side effect out loud at the start, along with how long it usually lasts. An expected symptom is tolerable; a surprising one ends the prescription.
• Track PDC, and then look at the shape of the gaps for anything where consecutive missed days are dangerous.
• Reconcile against what the patient has in the house, not against the med list. Those two documents disagree more often than anyone is comfortable admitting.

The prescription is a recommendation until something gets swallowed. Every measurement we use, PDC included, is a proxy for that one event, and the weakest link sits in the first two weeks, where almost nobody is looking.

This article discusses medication adherence measurement and Medicare care management billing rules as of 2026. It is general professional information and not clinical or legal advice. Verify current requirements against CMS guidance and your own state pharmacy regulations.