03 · What You Need to Know
What a Run-In Period Does and How It Changes Trial Design
What Is a Run-In Period?
A run-in period is a planned interval before randomization during which potential participants undergo specified procedures or receive a defined treatment condition. Their experience may determine whether they proceed to randomized allocation.
Depending on the protocol, participants may receive placebo, an active treatment, a standardized background regimen, or instructions designed to assess their ability to follow study procedures.
The defining feature is not simply that time passes before randomization. It is that the interval serves a specified methodological purpose, often with criteria that determine who can enter the randomized phase.
A run-in can occur after initial screening and consent procedures but before treatment assignment. Its duration and activities should be defined prospectively.
How Is a Run-In Different From Screening, Baseline Assessment, and Washout?
These procedures can occur before randomization, but they serve different purposes.
| Procedure |
Primary purpose |
Typical implication |
| Screening |
Determine whether candidates satisfy eligibility criteria |
Identifies eligible participants |
| Baseline assessment |
Measure characteristics before the relevant intervention comparison |
Establishes pretreatment information |
| Run-in |
Assess prespecified conditions such as adherence, tolerance, or stability |
May determine eligibility for randomization |
| Washout |
Allow residual effects of a previous treatment to diminish |
May reduce carryover or treatment interference |
These procedures may overlap. A run-in can include a washout, and a washout can occur without an adherence-based run-in. Researchers should distinguish the actual objectives rather than using the terms interchangeably.
The more specific question of when a trial needs a washout period concerns residual treatment effects rather than the broader purposes of run-in selection.
When Is an Adherence Run-In Appropriate?
An adherence run-in evaluates whether potential participants can follow a specified regimen or complete required procedures before randomization.
For example, participants might be asked to take a study preparation according to a schedule, attend required visits, or complete repeated assessments.
Researchers may consider this approach when nonadherence would make the intended efficacy question difficult to answer, particularly in a tightly controlled explanatory trial.
However, adherence during a short run-in does not guarantee adherence during longer follow-up. Participants may behave differently when they know they are being assessed for eligibility.
More importantly, excluding individuals with poor initial adherence creates a selected randomized population. The trial may estimate an effect among participants able and willing to comply with the preliminary regimen rather than among all patients or participants who would ordinarily be offered the intervention.
If the research objective concerns effectiveness under routine conditions, this selection may undermine the trial's practical relevance.
When Is a Placebo Run-In Used?
During a placebo run-in, potential participants receive an inactive preparation before randomization. The procedure has historically been used to assess adherence, examine nonspecific responses, or identify participants who meet specified entry criteria.
For example, investigators may monitor whether participants follow a medication schedule during a placebo period.
Some designs exclude participants whose symptoms improve substantially during placebo run-in. Such exclusions have been proposed as a way to improve the ability to detect treatment differences.
Nevertheless, selecting participants based on placebo response can alter the trial population and may not reliably eliminate placebo-related responses during the randomized phase. The resulting treatment effect may not apply to people excluded by the procedure.
Researchers should therefore avoid assuming that removing placebo responders necessarily produces a more valid or clinically useful trial.
When Is an Active-Treatment Run-In Useful?
An active-treatment run-in exposes potential participants to a specified active intervention before randomization.
It may be used to identify people who tolerate treatment, demonstrate a defined response, or achieve stabilization under a common regimen.
One example is a randomized withdrawal design. Participants initially receive an active treatment, and those satisfying prespecified response or tolerance criteria are subsequently randomized to continue treatment or switch to an alternative condition.
Such a design can be useful when the scientific question concerns the benefit of continuing treatment among initial responders.
However, the resulting comparison does not directly estimate the treatment effect among everyone who might start the intervention. It estimates an effect in a population selected through the active-treatment experience.
That distinction should be reflected in the eligibility criteria, estimand, and interpretation of results.
When Is a Stabilization Run-In Justified?
Some trials require participants to reach a reasonably stable condition before randomization.
For example, a study may need participants to be receiving a stable background treatment or to have completed a standardized adjustment period before evaluating an additional intervention.
Stabilization can help ensure that important changes immediately preceding randomization do not complicate the intended treatment comparison.
However, stability should be defined through clinically or scientifically meaningful criteria. Requiring stability without a clear rationale may exclude participants whose fluctuating conditions are common in routine practice.
Researchers should also consider whether the run-in itself changes outcomes or creates a new baseline that differs from participants' original condition.
How Can Run-In Selection Affect Generalizability?
Randomization can support a credible comparison among participants who reach the randomized phase. It does not eliminate the consequences of selecting those participants through a run-in.
Suppose 300 individuals initially consent to participate, but 90 are excluded because they do not meet adherence or tolerance criteria during run-in. The randomized sample then contains 210 selected participants.
Even if random assignment is properly conducted among those 210 participants, the treatment effect may not apply to the 90 excluded individuals.
This is not necessarily a flaw when the target population is explicitly defined as people meeting the run-in criteria. It becomes problematic when researchers generalize the findings to a broader population without acknowledging the selection.
The key distinction is between internal validity for the selected randomized population and applicability to the population in which the intervention might ultimately be used.
Can Run-In Selection Create Bias?
Run-in selection does not automatically invalidate randomization. If eligibility is determined before allocation and participants are subsequently randomized properly, treatment assignment can remain independent of baseline characteristics within the selected population.
Nevertheless, selection can affect the estimand and external validity. It may also complicate interpretation if run-in procedures produce lingering treatment effects or if the selected population differs systematically from the intended population.
For example, selecting participants who respond favorably to active treatment before randomization may produce a population unusually likely to benefit from continuing that treatment.
Similarly, excluding individuals who experience early adverse effects can make later tolerability estimates less representative of everyone who initiates therapy.
The appropriate interpretation depends on whether the trial seeks to evaluate treatment initiation, treatment continuation, or effectiveness among a specifically selected population.
Should Run-In Participants Be Included in the Final Analysis?
The primary randomized comparison generally concerns participants who reach randomization and are assigned to trial conditions.
Individuals excluded before randomization are not part of that randomized comparison, although their screening and run-in outcomes may be important for assessing recruitment, tolerability, feasibility, and applicability.
Researchers should report how many individuals entered the run-in, how many completed it, how many were excluded, and the reasons for exclusion.
When relevant, run-in outcomes may be summarized separately from randomized treatment outcomes.
For randomized withdrawal trials, interpretation should clearly distinguish events occurring during initial active-treatment exposure from outcomes observed after randomization.
How Long Should a Run-In Period Last?
There is no universal run-in duration. The interval should be long enough to assess the specific condition it is intended to establish, but not longer than necessary.
An adherence assessment may require enough observations to distinguish isolated missed doses from persistent difficulty following a regimen. A stabilization period may depend on the time required for a relevant clinical or biological process to reach the intended condition.
The duration should be justified using the intervention's characteristics, prior evidence, participant burden, and ethical considerations.
An unnecessarily long run-in can increase attrition, delay access to interventions, and create additional selection before randomization.
When Is a Run-In Unnecessary or Inappropriate?
A run-in may add little value when the intervention can be initiated immediately, eligibility is already clear, and there is no important uncertainty requiring pre-randomization observation.
It may also be inappropriate when excluding participants based on early adherence or tolerance would conflict with the study's objective of evaluating effectiveness under ordinary practice.
In pragmatic trials, where researchers often seek evidence relevant to real-world implementation, extensive run-in selection can produce a population that is less representative of routine service users.
Other approaches may be preferable, including improved participant instructions, appropriate monitoring, prespecified adherence analyses, or pilot testing of procedures.
The decision should follow the trial's scientific purpose rather than a general preference for highly compliant participants.
