Medication adherence is measured through a combination of methods, and no single approach captures the full picture. The tools range from simple patient questionnaires to lab tests that detect drug metabolites in urine or blood. Each method trades off between accuracy, cost, and practicality, which is why researchers and clinicians often combine two or more approaches.
Before diving into specific methods, two terms are worth distinguishing. Adherence (sometimes called compliance) refers to whether a patient takes medication at the right dose, time, and frequency. Persistence refers to how long a patient continues treatment before stopping. A person might be perfectly adherent while taking their medication but discontinue it after three months, making them non-persistent. Most measurement tools focus on adherence, though pharmacy data can capture both.
Direct vs. Indirect Methods
The World Health Organization classifies adherence measures into two broad categories: subjective (relying on someone’s judgment) and objective (relying on data or testing). Another common framework splits them into direct and indirect. Direct methods confirm that a drug was actually ingested. These include watching a patient swallow their pills, or testing blood or urine for the drug or its byproducts. Indirect methods infer adherence from other signals: refill records, pill counts, electronic bottle openings, or patient self-reports.
Direct methods are more accurate but harder to scale. Indirect methods are easier to implement across large populations but introduce uncertainty, since none of them prove a pill was actually swallowed.
Self-Report Questionnaires
Self-report is the cheapest and most widely used approach. Patients answer a short set of questions about their medication-taking habits, and the responses generate an adherence score. Two validated tools dominate the field.
The Morisky Medication Adherence Scale (MMAS) has eight items covering forgetfulness, medication-taking behavior, and problems like side effects. In validation studies, it has shown sensitivity as high as 0.93 when compared against clinical outcomes like blood pressure control, though specificity is lower (around 0.53 in some samples), meaning it catches most non-adherent patients but also flags some adherent ones as non-adherent. Internal consistency varies across populations.
The Brief Medication Questionnaire takes a different angle, using nine items to evaluate regimen complexity and medication-taking problems. In validation against electronic monitoring data, it achieved both high sensitivity (0.80) and perfect specificity (1.00) in one study of 43 patients, though those numbers come from a small sample.
The core limitation of all self-report tools is social desirability bias. Patients tend to overstate how well they take their medications, especially when a clinician is asking the questions. Still, these scales are free or low-cost, take minutes to administer, and can flag patients who need closer follow-up.
Pharmacy Refill Records
For large health systems and insurers, pharmacy data offers a way to measure adherence across thousands of patients without asking anyone a single question. Two metrics are standard: the Medication Possession Ratio (MPR) and the Proportion of Days Covered (PDC).
MPR adds up the total days of medication supplied across all refills, then divides by the number of days in the measurement period. The problem is that overlapping refills (picking up a new bottle before the old one runs out) get counted twice, so MPR can exceed 100%, artificially inflating adherence scores.
PDC fixes this by counting each calendar day only once. If you had medication available on 292 out of 365 days, your PDC is 80%. Because it can’t exceed 100% and accounts for the timing of refills, PDC is the preferred measure for quality reporting and research.
One practical consideration: the data source matters. Insurance claims data, collected for billing purposes, may miss prescriptions paid out of pocket and often arrive with a one-to-three month administrative lag. Pharmacy fill data from within a health system’s own network captures those cash-pay prescriptions and is available in near real time. In a direct comparison, PDC calculated from pharmacy fill data ran about 2.6% higher than PDC from insurance claims, with sensitivity of 0.89 and specificity of 0.80 when using claims as the reference. The two sources correlate moderately well but are not interchangeable.
Electronic Monitoring Devices
Electronic monitoring, most commonly through Medication Event Monitoring System (MEMS) caps, has been called the “gold standard” of adherence measurement. These caps fit onto a standard pill bottle and record the date and time of every opening. The resulting data shows not just whether doses were taken, but whether they were taken on schedule.
In practice, the gold standard label is generous. MEMS caps are an indirect measure: they record bottle openings, not pill swallowing. A patient might open the bottle and discard the pill, remove multiple doses at once for later use, or open the bottle out of curiosity. Studies of HIV medication adherence found inconsistencies including multiple doses removed in a single opening, missed doses despite openings, and openings with no dose taken. Caps also occasionally malfunction, recording no data for extended periods, which researchers only discover at study visits.
There’s an irony embedded in the technology: the reliability of MEMS caps depends on the patient using them correctly, which itself requires a form of adherence. Patients who struggle with their medication regimen are the same patients most likely to use the caps inconsistently. MEMS devices also prevent some adherence-promoting strategies, like using weekly pill organizers, since the medication must stay in the monitored bottle.
Smart Pill Bottles and Digital Tools
Newer smart pill bottles build on the electronic monitoring concept but add intervention features. A smart bottle paired with a smartphone app can beep and blink at dose time, track remaining pills automatically, send alarm reminders for missed doses, and notify a designated caregiver when a dose is skipped. If a patient misses a dose, the caregiver receives an automated phone call one hour after the scheduled time.
In a study of breast cancer survivors taking oral antiestrogen therapy, participants using smart pill bottles achieved 97.3% adherence over 28 days, compared to 88.3% in the control group. That gap was statistically significant. The results outperformed educational interventions alone and were comparable to weekly text message reminders. These tools blur the line between measurement and intervention, simultaneously tracking and improving adherence.
Directly Observed Therapy
The most certain way to confirm someone took their medication is to watch them do it. Directly observed therapy (DOT) means a healthcare worker or designated person watches the patient swallow each dose. It is the standard of care for tuberculosis treatment, where incomplete courses drive drug resistance. The CDC recommends DOT for all patients with TB disease, including children and adolescents.
DOT can happen in person or remotely through a smartphone or computer, a variation called electronic DOT (eDOT). It should be scheduled at a time and place convenient for the patient. During each encounter, the observer also asks about side effects or problems with the medication. The obvious limitation is resource intensity: someone must be present for every single dose, which is feasible for a disease like TB with defined treatment courses but impractical for lifelong chronic conditions.
Biochemical Testing
Blood or urine tests can detect whether a drug or its metabolites are present in the body, providing direct proof of ingestion. This approach is used in research settings and increasingly in clinical practice for conditions like hypertension, diabetes, and high cholesterol.
A study of type 2 diabetes patients screened urine for oral diabetes medications, blood pressure drugs, and statins. The method could detect most drug classes, including common blood pressure medications (diuretics, beta-blockers, calcium channel blockers, ACE inhibitors, and angiotensin receptor blockers) and two statins (atorvastatin and rosuvastatin). However, not all drugs leave detectable urinary metabolites. Some diabetes medications like tolbutamide and glibenclamide, and blood pressure drugs like methyldopa and clonidine, produce no measurable urine traces.
Biochemical testing captures a snapshot, not a pattern. A positive result confirms recent ingestion but says nothing about whether the patient took their pills last Tuesday or will take them tomorrow. Patients who skip doses routinely but take one before a clinic visit, sometimes called “white coat adherence,” will test positive despite poor overall adherence.
The 80% Threshold and Why It’s Complicated
You’ll frequently see 80% cited as the cutoff between adherent and non-adherent. This number traces back to a 1980 study of blood pressure medications, where a regression analysis showed that diastolic blood pressure only dropped reliably when patients took at least 80% of their pills. Researchers adopted it widely, and it became a default across nearly every disease area.
The problem is that 80% is not a universal number. A systematic review of adherence thresholds found that the clinically meaningful cutoff varies dramatically depending on the condition and the medication. For congestive heart failure, the threshold linked to reduced hospitalizations was as low as 58%. For schizophrenia, it was 76%. For diabetes, it was 85%. For statins targeting cholesterol reduction, it was 90%. In one diabetes study, the threshold shifted between 46% and 92% depending on patient characteristics like insulin use and prior hospitalizations.
Treating 80% as a one-size-fits-all benchmark is, as one research group put it, “one remaining myth in 40 years of adherence science.” The takeaway for anyone measuring adherence: the threshold that matters depends on what outcome you’re trying to achieve and which medication you’re measuring.
Combining Methods for Better Accuracy
Every method has a blind spot. Self-report overestimates adherence. Pill counts can’t tell you if pills were discarded. Electronic monitors can’t confirm ingestion. Pharmacy records can’t account for medications obtained outside the tracked system. Biochemical tests capture only a moment in time.
The most reliable approach combines at least two methods that compensate for each other’s weaknesses. A common pairing in research is pharmacy refill data (which captures long-term patterns) plus a validated self-report scale (which captures reasons for non-adherence, like side effects or forgetfulness). In clinical settings, refill data flagging potential non-adherence can trigger a conversation using a structured questionnaire, giving providers both the what and the why. For high-stakes situations like clinical trials or TB treatment, direct observation or biochemical verification adds a layer of certainty that indirect methods cannot provide.