Medication Adherence

How medication adherence changes over time

Medication adherence describes how closely a person's medication-taking behaviour corresponds with an agreed treatment plan. This page explains the stages of adherence, the reasons it may change, how it can be measured and how adherence affects clinical outcomes, healthcare costs and economic evaluation.

Adherence is not a fixed personal characteristic. It can vary between medicines, doses and periods of treatment as symptoms, beliefs, costs, routines, adverse effects and access change.

The three stages of adherence

Medication adherence can be understood as initiation, implementation and persistence. Separating these stages helps identify where a treatment pathway breaks down and which intervention might help.

• Initiation occurs when a patient takes the first dose of a prescribed medicine.
• Implementation describes how the patient's actual dosing corresponds with the prescribed regimen after initiation.
• Persistence describes the time from initiation until the patient stops treatment.
• Discontinuation marks the end of persistence when the next dose is no longer taken.

A person may never initiate treatment, may initiate but take doses inconsistently, or may implement the regimen well before discontinuing earlier than intended. A single adherence percentage can conceal these different patterns.

Adherence should reflect an agreed plan

Adherence is best interpreted in relation to a treatment plan agreed between the patient and healthcare professional. The plan should specify the medicine, dose, frequency, route and intended duration and should be revised when treatment changes.

The relevant question is not simply whether a patient followed an instruction. It is whether the medication-taking behaviour corresponds with a plan that the patient understands, can carry out and continues to consider acceptable.

Shared decision making can improve the fit between the plan and the patient's goals, but agreement does not remove practical barriers or guarantee continued use.

Intentional and unintentional non-adherence

Non-adherence may be intentional, unintentional or a combination of both. The distinction matters because different causes require different responses.

Intentional non-adherence may reflect:

• Concern about adverse effects.
• Doubt about necessity or benefit.
• Preference to avoid long-term treatment.
• Previous negative experiences.
• Competing priorities or treatment burden.
• Beliefs about illness or medicines.
• A decision to adjust treatment according to symptoms.

Unintentional non-adherence may reflect:

• Forgetting.
• Complex dosing schedules.
• Difficulty opening packaging or administering the medicine.
• Cognitive, visual or physical impairment.
• Prescription, dispensing or supply failure.
• Cost or lack of insurance coverage.
• Travel, work or caring responsibilities.
• Language, literacy or communication barriers.

Labelling all missed doses as forgetfulness can lead to reminder-based interventions that do not address concerns, affordability or access.

The factors that shape adherence

Medication-taking behaviour develops through the interaction of the patient, condition, treatment, healthcare system and social environment. A complete assessment should avoid placing all responsibility on the patient.

Relevant factors include:

• Symptom severity and whether the condition feels immediate or asymptomatic.
• Expected benefit and time before benefit occurs.
• Adverse effects and perceived treatment burden.
• Number of medicines and dosing complexity.
• Cost, coverage and pharmacy access.
• Quality of communication and trust.
• Continuity of care and refill systems.
• Mental health, cognition and health literacy.
• Family, caregiver and social support.
• Housing, transport and daily routines.
• Cultural beliefs and previous experiences of healthcare.

The same factor may affect initiation, implementation and persistence differently. Cost may prevent initiation, for example, while adverse effects may lead to later discontinuation.

Measuring medication adherence

No single method perfectly determines whether a medicine was taken as intended. Each method measures a different part of the medication-use pathway and introduces different assumptions.

Methods include:

• Patient self-report.
• Structured adherence questionnaires.
• Pill counts.
• Pharmacy dispensing or claims records.
• Electronic medication-monitoring devices.
• Prescribing and administration records.
• Drug or metabolite concentrations.
• Biomarkers influenced by treatment.
• Directly observed therapy.

Dispensing data show that medicine was supplied, not that it was swallowed or administered. Drug concentrations can confirm recent exposure but may not describe usual behaviour and can be influenced by metabolism, timing and dose.

Measurement should match the decision. A method suitable for population monitoring may be insufficient for determining whether an individual patient took a particular dose.

Calculating proportion of days covered

Proportion of days covered uses dispensing records to estimate the share of an observation period for which medicine was available. Overlapping supplies are generally carried forward, and covered days are capped so the measure does not exceed 100%.

$$ PDC = \frac{Number\ of\ days\ covered}{Number\ of\ days\ in\ the\ observation\ period} $$

For a 180-day period with 153 covered days:

$$ PDC = \frac{153}{180} = 0.85 = 85% $$

The observation window, eligible medicines, treatment switching, inpatient days and death should be handled explicitly. Different rules can produce different values from the same dispensing history.

Calculating medication possession ratio

Medication possession ratio commonly divides the total days supplied by the number of days in the observation period. Unlike proportion of days covered, it may exceed 100% when refills overlap or are obtained early.

$$ MPR = \frac{Total\ days\ supplied}{Number\ of\ days\ in\ the\ observation\ period} $$

If 195 days of medicine are supplied during a 180-day period:

$$ MPR = \frac{195}{180} = 1.083 = 108.3% $$

A value above 100% does not show that the patient took more than prescribed. It shows that the calculated supply exceeded the observation period under the selected rules.

Proportion of days covered and medication possession ratio should not be treated as interchangeable. The chosen measure and calculation rules should be reported.

Measuring persistence

Persistence measures how long treatment continues before a gap or discontinuation. The analysis requires a permissible gap that defines when a patient is no longer considered persistent.

Time to discontinuation can be measured from initiation to the end of supplied medication plus the allowable gap. The result depends on the gap definition, stockpiling rules, treatment switching and whether temporary clinical interruptions are permitted.

A 30-day and a 90-day allowable gap can produce materially different persistence estimates. The chosen threshold should reflect the medicine, dosing and clinical context rather than a universal convention.

Adherence thresholds

Studies often classify patients as adherent when a measure reaches a threshold such as 80%. A convenient threshold does not necessarily represent the level required for clinical benefit.

The relationship between medication use and outcome may be:

• Approximately linear.
• Characterised by a clinically meaningful threshold.
• Different across medicines or diseases.
• Dependent on the timing rather than only the number of missed doses.
• Influenced by treatment interruptions or dose changes.

Continuous results should be retained when possible, and any classification threshold should have a clinical rationale. Dichotomising adherence can hide differences between people just above and below the cut-off.

Timing and patterns of missed doses

Two patients can have the same overall adherence percentage but very different medication-taking patterns. Consecutive missed doses, weekend omissions or long gaps may have different clinical effects from occasional isolated doses.

Pattern-sensitive assessment may examine:

• Dose timing.
• Length and frequency of gaps.
• Clustering of missed doses.
• Adherence before or after clinical visits.
• Changes after adverse events or treatment adjustments.
• Differences between weekdays and weekends.

Aggregate measures such as proportion of days covered cannot represent every pattern. More detailed monitoring is justified only when timing materially affects safety or effectiveness.

Measuring adherence to several medicines

Patients with several long-term medicines can be adherent to one treatment and not another. A single summary measure requires rules for which medicines are included and whether simultaneous availability is required.

Possible approaches include:

• Reporting adherence separately for each medicine.
• Averaging medicine-specific adherence measures.
• Measuring the proportion of days on which all required medicines are available.
• Measuring adherence to a therapeutic class rather than a specific product.
• Defining clinically essential medicines separately from optional supportive treatment.

The selected method should reflect the clinical question. Averaging can conceal complete non-adherence to one essential medicine when adherence to others is high.

Adherence in clinical trials

Clinical trials often provide more support and monitoring than routine care, so adherence may be higher than after implementation. Trial adherence should be reported alongside the estimated treatment effect.

Important issues include:

• How adherence was measured.
• Whether measurement differed between groups.
• Treatment switching and rescue therapy.
• Reasons for missed doses and discontinuation.
• Whether outcomes were collected after treatment stopped.
• Whether analysis estimated assignment, receipt or sustained use.

Excluding non-adherent participants can break the randomised comparison because adherence is influenced by prognosis, treatment response and adverse effects. Per-protocol analyses require careful causal interpretation.

Adherence in routine care

Real-world adherence may differ from trial adherence because monitoring, cost, access, patient characteristics and clinical support differ. Routine data can describe dispensing and persistence across large populations but may not capture patient reasoning or actual ingestion.

Observed associations between adherence and outcomes may be confounded. People who adhere more consistently may also differ in health behaviours, access, social support and engagement with care.

The healthy-adherer effect describes this potential bias. Statistical adjustment can reduce measured confounding but cannot guarantee that adherence itself caused the observed outcome difference.

Clinical consequences of non-adherence

The consequences depend on the medicine, condition, amount and timing of missed treatment. Non-adherence may reduce benefit, create withdrawal or rebound effects, increase resistance or make treatment response difficult to interpret.

Potential consequences include:

• Poor symptom control.
• Disease progression.
• Avoidable complications.
• Emergency care or hospitalisation.
• Treatment escalation based on an apparent lack of response.
• Unnecessary diagnostic testing.
• Drug resistance for selected anti-infective treatments.
• Adverse effects after irregular use or restarting.

Non-adherence does not always cause harm. A patient may stop a medicine appropriately because of adverse effects or after clinical advice, so measurement should distinguish agreed treatment changes from unexplained gaps.

Economic consequences of adherence

Greater adherence usually increases medication acquisition costs because more treatment is supplied and taken. It may reduce other healthcare costs when effective treatment prevents complications, but it can also increase adverse-event costs.

Economic consequences include:

• Medication and dispensing expenditure.
• Monitoring and adherence-support costs.
• Changes in hospital and outpatient use.
• Changes in productivity or caregiver time.
• Health gains from improved treatment effect.
• Harms from increased exposure.
• Waste from medicine supplied but not taken.

Reduced medical spending should not be assumed to exceed increased medicine cost. The net effect depends on treatment effectiveness, baseline risk, adherence change, adverse events and the time horizon.

Representing adherence in economic models

Economic models can incorporate adherence through treatment exposure, effectiveness, cost, persistence and switching. The structure should reflect the evidence rather than apply an arbitrary proportional reduction to every outcome.

Modelling questions include:

1. Define the adherence construct represented by the model.
2. Estimate adherence under each strategy using evidence relevant to routine practice.
3. Link adherence to treatment effectiveness using a defensible relationship.
4. Apply medicine costs according to dispensing or consumption, as appropriate.
5. Represent discontinuation and switching over time.
6. Include intervention and monitoring costs.
7. Test uncertainty in adherence levels, persistence and effect relationships.

The relationship between adherence and effectiveness may be nonlinear and affected by timing. Assuming that 80% adherence always produces 80% of the treatment effect can be seriously misleading.

Interventions to support adherence

Adherence interventions should address the identified barrier rather than apply the same solution to every patient. Multi-component approaches may be needed when several barriers interact.

Interventions include:

• Shared decision making and treatment review.
• Simplified dosing or reduced treatment burden.
• Education and clear written information.
• Reminder systems and prompts.
• Pill organisers or adherence packaging.
• Pharmacy refill support and synchronisation.
• Financial assistance or reduced cost sharing.
• Management of adverse effects.
• Motivational interviewing or problem solving.
• Family or caregiver support with patient agreement.
• Digital monitoring and feedback.
• Outreach after missed refills or appointments.

Reminders can help unintentional forgetting but are unlikely to resolve unaffordable treatment or concerns about harm. The intervention should preserve patient autonomy and avoid punitive surveillance.

Evaluating an adherence intervention

An evaluation should measure whether the intervention changes medication-taking behaviour and whether that change improves outcomes. An intervention can increase refill collection without changing medicine use or health.

Relevant outcomes include:

• Initiation.
• Dose implementation.
• Persistence.
• Clinical outcomes.
• Adverse events.
• Health-related quality of life.
• Patient burden and acceptability.
• Healthcare use and costs.
• Equity of reach and benefit.

The comparator should include any adherence support already provided in usual care. Follow-up should be long enough to show whether change persists after intensive support ends.

Equity and structural barriers

Medication adherence is affected by coverage, pharmacy location, housing, employment, language, disability and trust. Policies that treat non-adherence as individual failure can worsen inequality.

Equity assessment should examine:

• Out-of-pocket cost and insurance restrictions.
• Geographic and transport access.
• Pharmacy and prescriber continuity.
• Language and accessible communication.
• Digital requirements.
• Ability to store or administer treatment safely.
• Work and caregiving constraints.
• Differential monitoring or penalties.

Removing structural barriers may produce more sustainable improvement than education or reminders alone.

A simplified example

Suppose a patient is observed for 365 days and has medicine available for 292 days. The proportion of days covered is:

$$ PDC = \frac{292}{365} = 0.80 = 80% $$

The patient has a 45-day gap late in the year and otherwise takes medicine regularly. A second patient also has 80% coverage but misses one or two doses each week throughout the year.

The annual proportion is the same, but the patterns and clinical consequences may differ. Interpretation requires the medication, allowable gaps, outcome relationship and reasons for missed treatment.

Distinguishing adherence from related concepts

Medication adherence overlaps with several terms, but each describes a different part of treatment use or decision making. Clear terminology helps match the measure and intervention to the problem.

• Persistence is the duration from initiation to discontinuation.
• Implementation is how actual dosing corresponds with the regimen while treatment continues.
• Initiation is taking the first dose.
• Compliance is an older term that can imply passive obedience rather than an agreed plan.
• Concordance refers to the process of reaching agreement between patient and clinician.
• Medication possession indicates available supply rather than confirmed ingestion.
• Treatment fidelity describes whether an intervention is delivered as intended and may apply to providers as well as patients.

These concepts should not be collapsed into one percentage without stating what is being measured.

Common misunderstandings

Medication adherence is not simply a matter of remembering and should not be interpreted as evidence of a patient's character or commitment. It is a behaviour shaped by treatment, healthcare and social conditions.

Common misunderstandings include:

• Dispensing records do not prove that medicine was taken.
• A single threshold such as 80% is not clinically valid for every medicine.
• High medication possession ratio can reflect early refills rather than excessive use.
• Proportion of days covered and medication possession ratio are not interchangeable.
• Adherence and persistence are not the same concept.
• Non-adherence is not always intentional.
• More adherence is not always beneficial when treatment is ineffective or harmful.
• Association between adherence and better outcomes does not automatically prove causation.
• Reminder systems do not correct affordability or access barriers.

Interpreting adherence evidence

Adherence results should be interpreted using the specific medicine, prescribed regimen, stage of treatment, measurement method and observation period. The analysis should distinguish supply, ingestion, timing and continuation rather than use adherence as an undefined label.

High-quality evidence links a clearly measured adherence pattern to outcomes and explains the barriers producing that pattern. This supports interventions that improve appropriate medicine use without blaming patients or assuming that greater exposure is always better.