03 · What You Need to Know
How to Define, Measure, and Interpret Intervention Adherence
What Is Intervention Adherence?
Intervention adherence concerns the extent to which participants follow or complete activities specified by an intervention protocol.
Its operational meaning depends on the intervention. In an educational study, adherence might involve attending instructional sessions, completing assigned learning activities, reviewing formative feedback, or using a learning platform according to a specified schedule.
In other disciplines, adherence may concern prescribed treatments, behavioral activities, or repeated procedures.
Researchers should identify the specific behaviors that constitute adherence and explain why they matter to the intervention's intended function.
For example, logging into an online learning platform may be a poor adherence indicator when the intervention requires students to review and act upon individualized feedback.
Why Might Adherence Be Important for Feasibility?
A feasibility study examines whether the proposed definitive research can be conducted and whether important procedures are workable.
When the intervention requires sustained participation, adherence can indicate whether its demands are realistic.
For example, an eight-week instructional intervention may require students to complete one structured reflection activity each week.
If many students complete only the first two activities, the research team needs to understand why before proceeding to a larger evaluation.
The problem may involve workload, unclear instructions, technical barriers, or a mismatch between the intervention and ordinary instructional practices.
Adherence evidence can help distinguish these possibilities and inform appropriate modifications.
Is Adherence the Same as Intervention Fidelity?
No, although the terminology sometimes overlaps across implementation frameworks.
Intervention fidelity generally concerns whether the intervention is delivered according to its intended design. Participant adherence concerns whether participants complete or engage with the prescribed activities.
Intervention fidelity
Whether implementers deliver the intended intervention components according to the protocol and relevant quality expectations.
Participant adherence
Whether participants undertake the activities or behaviors specified for them within the intervention.
A teacher may deliver all planned feedback sessions while students complete only some associated learning activities.
The intervention could therefore demonstrate high delivery fidelity but lower participant adherence.
Researchers should report these dimensions separately when both are relevant.
How Is Adherence Different From Intervention Exposure?
Exposure concerns the intervention content or activities participants encounter. Adherence concerns whether their participation meets specified expectations.
For example, a student may receive six feedback messages but open only three. Another student may open all six without undertaking the required revisions.
These observations represent different levels of exposure and engagement.
The appropriate adherence measure depends on the intervention's mechanism and intended activities.
Researchers should avoid assuming that access, receipt, or platform activity automatically demonstrates meaningful participation.
How Is Adherence Different From Participant Retention?
Retention concerns whether participants remain involved or available for follow-up. Adherence concerns whether they undertake the prescribed intervention activities.
A participant may remain enrolled and complete all outcome assessments while rarely engaging with the intervention.
Conversely, someone may complete every intervention activity but become unavailable for a later follow-up assessment.
These circumstances have different implications for feasibility and analysis.
Researchers planning longitudinal interventions should consider participant retention as a separate feasibility outcome when continued follow-up is essential.
When Should Adherence Be a Primary Feasibility Outcome?
Adherence may appropriately be primary when the central uncertainty concerns whether participants can complete the intervention as intended.
For example, a new digital learning intervention may require frequent student interaction, but researchers may have little evidence that the expected activity schedule is manageable.
In that situation, adherence findings could determine whether the intervention requires redesign before a definitive trial.
However, adherence need not be primary when recruitment, data completeness, or institutional implementation presents a greater uncertainty.
The outcome hierarchy should reflect the decisions the feasibility study is intended to inform.
What Types of Adherence Can Be Measured?
Adherence may involve attendance, completion, frequency, duration, timing, or the quality of required engagement.
| Adherence dimension |
Possible indicator |
Important limitation |
| Attendance |
Sessions attended relative to scheduled sessions |
Attendance does not establish active participation. |
| Activity completion |
Required activities completed |
Completion may not indicate quality of engagement. |
| Frequency |
Number of prescribed interactions undertaken |
Repeated actions may be superficial. |
| Duration |
Time spent undertaking relevant activities |
Longer duration is not necessarily better. |
| Timing |
Activities completed within intended periods |
Late participation may still be meaningful. |
| Behavioral implementation |
Evidence that prescribed behaviors were undertaken |
Self-report or digital traces may not capture actual behavior accurately. |
Not every intervention requires all these dimensions. Researchers should prioritize indicators that correspond to its essential components.
How Do You Calculate an Adherence Rate?
A simple activity-based adherence proportion can compare completed prescribed activities with the number expected.
This result describes completion relative to the specified schedule. It does not establish that the activities were completed thoughtfully or that the intervention produced the intended outcome.
When participants have different numbers of assigned activities, researchers should explain whether they calculate individual proportions, average those proportions, or aggregate completed and expected activities across participants.
These approaches may produce different summaries.
Should Adherence Be Reported at the Participant or Activity Level?
Both levels may be informative, but they answer different questions.
Participant-level adherence can show how many individuals meet a specified completion criterion.
Activity-level adherence can show the proportion of prescribed activities completed across the study.
For example, a high overall activity completion rate may conceal a subgroup of participants who rarely engage.
Conversely, a threshold-based participant measure may obscure substantial variation among individuals who fall just above or below the threshold.
Researchers should select and report summaries according to their feasibility objectives.
What Counts as Adequate Adherence?
There is no universal adherence percentage appropriate to every intervention.
A threshold should reflect the intervention's essential components, intended dose, theoretical rationale, and practical requirements.
For example, completing five of six sessions may be sufficient for one intervention but inadequate for another in which the final session contains an essential activity.
Researchers should avoid adopting a threshold such as 80% without explaining why it is meaningful for the particular intervention.
Where a minimum exposure is required, the justification should be established before interpreting the feasibility results.
Should All Intervention Components Be Weighted Equally?
Not necessarily. Some activities may be essential to the intervention's intended mechanism, while others provide reinforcement or supplementary practice.
For example, an AI-supported formative feedback intervention may require students to review feedback and revise their work. Merely opening the feedback message may not represent equivalent engagement.
Researchers may distinguish core from supplementary activities.
However, assigning different weights requires a defensible rationale.
Weights should not be selected retrospectively to make adherence appear more favorable.
How Should Adherence Be Measured in Digital Interventions?
Digital platforms may provide records of logins, activity completion, page access, or time spent using a system.
These records can support adherence assessment when the logged events correspond meaningfully to the intervention requirements.
However, digital traces are imperfect indicators of behavior.
A student may leave a learning resource open without engaging with it, or complete meaningful work outside the platform.
Researchers should examine what the system actually records and whether the indicator supports the intended interpretation.
Technical verification may be necessary to ensure that events are captured and linked correctly.
Can Self-Reported Adherence Be Used?
Yes, when self-report is suitable for the intervention and its limitations are acknowledged.
Participants may report completing activities that cannot be observed directly, such as independent reflection or practice outside a digital platform.
However, recall errors and social desirability may affect responses.
Where feasible, researchers may compare self-reports with other appropriate evidence, such as activity records or observations.
Agreement between sources can be informative, but disagreement should be investigated rather than automatically attributed to participant dishonesty.
Should Adherence Be Measured Repeatedly?
Repeated measurement may be useful when adherence can change over the intervention period.
For example, students may complete most activities during the first two weeks but participate less consistently during examinations.
An overall adherence percentage may conceal this decline.
Recording adherence by session or week can reveal whether particular activities or periods create difficulties.
However, the measurement process itself should not impose unnecessary burden.
The frequency of monitoring should reflect the intervention and the feasibility decisions being made.
What Factors May Explain Low Adherence?
Low adherence can arise from participant workload, unclear instructions, technical problems, competing responsibilities, perceived irrelevance, accessibility barriers, or intervention delivery failures.
For example, a teacher may schedule required activities at times when students have limited internet access.
In that situation, low participation may reflect an implementation problem rather than unwillingness.
Researchers should avoid interpreting adherence solely as an individual participant characteristic.
Qualitative interviews, observations, or implementation logs may help explain the observed patterns.
Should Adherence Be Evaluated Alongside Acceptability?
Often, yes. Adherence describes what participants do, while acceptability concerns how they experience or evaluate the intervention.
A participant may complete all activities despite finding them burdensome.
Another may consider the intervention valuable but be unable to participate because of scheduling or technical constraints.
These findings suggest different implications for redesign.
Where acceptability is uncertain, it may warrant a distinct feasibility assessment rather than being inferred from adherence alone.
How Does Adherence Relate to Intervention Fidelity?
Adherence should be interpreted alongside the extent to which the intervention was actually delivered.
For example, a participant cannot complete a prescribed activity that was never made available.
Researchers should therefore distinguish missed activities caused by non-delivery from activities offered but not completed.
Failure to make this distinction can produce misleading adherence estimates.
A feasibility study should document whether the intervention was delivered as planned and whether participants had a reasonable opportunity to engage.
Can Low Adherence Affect the Interpretation of Effectiveness?
Potentially. When participants receive substantially less of an intervention than intended, the relationship between assignment, exposure, and outcomes becomes more difficult to interpret.
However, researchers should not assume that excluding low-adherence participants produces an unbiased estimate of intervention effectiveness.
Adherence may be associated with motivation, baseline characteristics, or other factors related to outcomes.
In randomized trials, analyses based on adherence require careful consideration of the estimand and potential selection bias.
Feasibility adherence findings should inform intervention planning rather than encourage unplanned analyses designed to produce more favorable effects.
Should Adherence Determine the Definitive Sample Size?
Adherence may influence planning, but its relationship to sample size depends on the design and intended analysis.
For example, low adherence may reduce the contrast between randomized conditions and affect the precision of certain effect estimates.
However, simply recruiting additional participants may not resolve an intervention that is too burdensome or difficult to deliver.
Researchers should first determine whether adherence problems can be addressed appropriately.
Sample size decisions should reflect the definitive estimand, relevant assumptions, and suitable statistical expertise.
How Large Should an Adherence Feasibility Study Be?
There is no universal sample size.
The appropriate scope depends on the intervention's duration, complexity, variation across participants, and the precision needed for the intended progression decision.
A small study may identify obvious barriers but provide imprecise estimates of adherence proportions.
For interventions delivered across classes or institutions, variation between settings may also matter.
Researchers should justify the sample and observation period according to the adherence questions being investigated.
How Should Adherence Progression Criteria Be Established?
Progression criteria should reflect the level and pattern of adherence necessary for the definitive intervention to be evaluated meaningfully.
For example, researchers may establish a minimum completion level for essential activities, accompanied by an assessment of why activities are missed.
However, a single aggregate percentage may conceal serious problems with core components.
Criteria should be justified prospectively where possible and interpreted with attention to uncertainty.
A green-amber-red progression framework may support structured decisions, but its thresholds must reflect the particular intervention.
What If Adherence Is Lower Than Expected?
Researchers should investigate which activities were missed, when participation declined, and what barriers contributed to the pattern.
Possible modifications include clearer instructions, improved technical support, revised scheduling, or reduced unnecessary burden.
However, changes should preserve the intervention's essential components.
If substantial redesign is required, additional feasibility testing may be appropriate.
Researchers should not redefine adherence retrospectively simply to classify the intervention as feasible.
Does High Adherence Establish Intervention Effectiveness?
No. Participants may complete an intervention consistently without experiencing the intended benefit.
For example, students may complete every AI-supported feedback activity but show no meaningful improvement in academic performance.
High adherence supports a conclusion about observed participation under the tested conditions.
It does not establish that the intervention causes improved outcomes.
Effectiveness requires an appropriate research design and outcome evidence.
How Should Adherence Findings Be Reported?
The report should define adherence, specify expected activities, identify the observation period, and explain how completion was measured.
Researchers should report relevant numerators and denominators and distinguish participant-level from activity-level summaries.
Where appropriate, adherence patterns over time and reasons for noncompletion should be described.
Any deviations from the protocol or changes to adherence definitions should be documented.
Findings should be interpreted as feasibility evidence rather than definitive evidence of effectiveness.
Watch Out
Do not label participants as nonadherent without first establishing that the required activities were delivered, accessible, and clearly explained. Low adherence may reflect weaknesses in the intervention or its implementation rather than unwillingness to participate.
