How Much Participant Attrition Should You Expect When Planning a Study?

How Many Participants Should You Expect to Lose Before Your Study Ends?

You calculate that your study needs 200 participants with complete outcome data. But recruitment is only the beginning. Some participants may withdraw, miss follow-up assessments, become unreachable, or provide incomplete measurements.

Should you recruit 220 participants to allow for 10% attrition? Would 250 be safer? Perhaps 300 would provide enough protection against unexpected dropout.

The difficulty is that attrition depends on the population, research procedures, follow-up period, and circumstances in which participation occurs. A short, one-session survey and a two-year longitudinal study face very different risks.

Researchers therefore need an attrition estimate that is defensible rather than merely convenient. An inaccurate assumption can leave a study underpowered, create unnecessary recruitment costs, or conceal serious threats to the interpretation of its findings.

Estimate Attrition From Comparable Evidence, Not a Universal Percentage

Estimating Attrition and Translating It Into a Defensible Recruitment Plan

What Counts as Participant Attrition?

Participant attrition generally refers to the loss of participants from continued study participation or follow-up after enrollment. The precise operational definition should be established in the research protocol.

Attrition can occur when participants formally withdraw, become unreachable, stop attending scheduled assessments, or otherwise cease providing required information.

However, several related events should be distinguished.

These categories can overlap, but they are not equivalent.

A participant who stops using an intervention may still complete every follow-up assessment. Conversely, a participant who remains formally enrolled may miss the primary outcome measurement.

For planning purposes, the most consequential quantity is often the proportion of participants expected to provide information sufficient for the primary analysis.

Researchers should avoid counting every form of nonadherence as attrition unless the study's design and analysis justify that classification.

Why Is There No Standard Attrition Rate?

Researchers sometimes use allowances such as 10%, 15%, or 20% because these figures are familiar from previous protocols or sample size examples.

Such percentages may be reasonable in particular circumstances, but they are not universal methodological standards.

Attrition varies with the population, study duration, research setting, participant burden, perceived relevance, contact arrangements, and many other conditions.

A study requiring one follow-up questionnaire after two weeks may have different retention prospects from a study requiring repeated clinical visits over several years.

Even studies with similar follow-up periods can differ substantially if one involves invasive procedures, frequent travel, or participants experiencing unstable living circumstances.

Consequently, researchers should not transfer an attrition percentage from an unrelated study simply because its sample size calculation used that figure.

Use Comparable Studies as the Starting Point

Published research can provide useful information about retention under conditions resembling the proposed study.

When reviewing comparable studies, examine more than the reported attrition percentage.

• Population: Were participants similar in age, circumstances, eligibility, and relevant characteristics?
• Setting: Did recruitment and follow-up occur in comparable institutions or communities?
• Follow-up duration: Were participants observed over similar periods?
• Assessment burden: Did the studies require comparable numbers and types of measurements?
• Retention procedures: Were reminders, flexible appointments, reimbursements, or other support provided?
• Outcome definition: Did the reported attrition concern complete withdrawal, missing primary outcomes, or another measure?

A published retention rate is useful only when its denominator and endpoint are understood.

For randomized trials, CONSORT reporting guidance emphasizes accounting for participant flow, including numbers randomized, receiving interventions, and analyzed. For observational studies, STROBE similarly emphasizes reporting participant numbers across study stages and reasons for nonparticipation where relevant.

These reporting principles help researchers assess whether published attrition figures are comparable to their own planning needs.

Pilot and Feasibility Studies Can Inform Attrition Assumptions

A pilot or feasibility study may provide direct evidence about whether participants can be recruited, retained, and assessed under the proposed procedures.

For example, a small feasibility study may reveal that participants frequently miss evening assessments because of work commitments, or that online follow-up is difficult for individuals with limited connectivity.

Such observations can guide revisions to scheduling, communication, and measurement procedures.

However, pilot attrition estimates can be imprecise, especially when based on small samples.

If two of twenty pilot participants fail to provide a final assessment, the observed missing-outcome proportion is 10%. This does not establish that the definitive study will experience exactly 10% attrition.

The estimate is subject to sampling uncertainty and may change when recruitment expands to different settings or populations.

Researchers should therefore consider plausible ranges rather than treating a small pilot estimate as a precise prediction.

Attrition Should Be Defined for the Relevant Follow-Up Point

Longitudinal studies may have several follow-up assessments, each with a different proportion of available observations.

For example, a study might obtain measurements from 95% of enrolled participants at one month, 88% at six months, and 76% at twelve months.

These hypothetical figures describe different follow-up endpoints. They should not be combined into a single attrition rate without clarifying the intended definition.

If the primary outcome is measured at twelve months, the expected availability of that measurement is particularly relevant to planning.

Nevertheless, earlier measurements may still contribute to longitudinal analyses, depending on the statistical model and missing-data assumptions.

Researchers should also distinguish intermittent missingness from monotone dropout. A participant who misses one assessment may return for a later one.

The appropriate planning assumption depends on the structure of the data and the intended analysis.

How Do You Calculate the Observed Attrition Rate?

For a simple study in which attrition is defined as failure to provide the required final outcome, the observed proportion can be calculated from the number initially enrolled and the number without that outcome.

The result means that 12% of the defined cohort did not provide the specified outcome. It does not establish why those observations are missing or whether the remaining 88% are representative of the original cohort.

For studies involving staggered entry, competing events, varying follow-up opportunities, or time-to-event outcomes, a simple proportion may be inadequate. The appropriate retention or censoring measure should reflect the study design.

How Do You Inflate a Recruitment Target for Expected Attrition?

If a sample size calculation indicates that a specified number of participants must provide complete data for a planned analysis, researchers may estimate the number to recruit by accounting for expected outcome availability.

This calculation assumes that the expected retention proportion applies to the relevant enrolled cohort and that the original sample size requirement is appropriate for the planned analysis.

It also assumes a simplified relationship between the number recruited and the number contributing complete data.

The formula does not guarantee adequate statistical power if missingness changes the composition of the analyzed sample, if the analysis uses partial observations, or if other design features affect precision.

Why Adding the Attrition Percentage Directly Is Incorrect

A common mistake is to increase the required sample size by the expected attrition percentage.

Suppose 200 complete observations are needed and 20% attrition is expected.

Adding 20% gives 240 participants. But if 20% of 240 participants are lost, only 192 remain.

The correct simplified inflation calculation divides by the expected retention proportion, producing a recruitment target of 250.

The difference becomes more consequential as the expected attrition proportion increases.

Researchers should round recruitment targets upward to whole participants and account separately for screening failures, ineligibility, or refusal before enrollment when those occur before the attrition denominator begins.

Attrition Planning Should Reflect the Statistical Analysis

Not every analysis requires complete measurements from every participant.

Longitudinal mixed-effects models, multiple imputation, and other approaches may use available information under particular assumptions. Their efficiency and validity depend on the missing-data mechanism, model specification, and observed data structure.

For example, a repeated-measures analysis may retain information from participants who miss one follow-up but complete others.

Nevertheless, having an analysis method that accommodates incomplete data does not make attrition harmless.

Missing data can reduce precision, limit the ability to assess long-term outcomes, and introduce bias when assumptions about missingness are violated.

For complex designs, simulation-based power analysis may be more appropriate than a simple sample size inflation formula.

Such simulations can incorporate expected measurement patterns, correlations, allocation ratios, and plausible dropout mechanisms.

Attrition May Differ Between Study Groups

In comparative studies, participants assigned to different interventions or exposed to different conditions may have different probabilities of remaining in follow-up.

For example, a demanding intervention may produce more discontinuation than a less burdensome comparison condition.

Assuming identical attrition across groups may therefore be unrealistic.

Researchers should examine comparable evidence and consider group-specific assumptions where justified.

Differential attrition can affect both statistical precision and the validity of comparisons, particularly when missing outcomes are associated with prognosis or treatment response.

Recruiting more participants cannot automatically resolve these concerns.

Use Sensitivity Scenarios Instead of a Single Optimistic Estimate

Because attrition is uncertain, planning should consider several plausible scenarios.

Suppose the required complete-data sample is 200 participants.

These figures are mathematical planning scenarios, not recommended attrition rates.

The purpose is to examine how sensitive recruitment requirements are to assumptions about retention.

If a modest increase in expected attrition makes recruitment impractical, the study may need changes to its design, follow-up procedures, recruitment period, or research objectives.

Recruitment Inflation Cannot Correct Attrition Bias

Increasing the initial sample can help preserve the number of observed outcomes, but it does not necessarily preserve the original sample's composition.

Suppose participants with poorer outcomes are more likely to withdraw. Even if enough participants remain to meet the numerical sample size target, the observed outcomes may systematically differ from those of the original cohort.

The problem is especially consequential when missingness depends on unobserved outcomes or other unmeasured factors.

Researchers should therefore distinguish loss of statistical precision from bias caused by selective missingness.

Retention planning, missing-data analysis, and transparent reporting are complementary responsibilities.

Attrition Assumptions Should Be Revisited During Feasibility Assessment

An attrition estimate should inform more than the initial sample size calculation.

It affects recruitment costs, staffing, participant contact arrangements, follow-up schedules, and the time needed to complete the study.

If the projected recruitment target exceeds the accessible population or available resources, the study may not be feasible under its current design.

Researchers should evaluate whether reducing unnecessary burden, improving follow-up arrangements, or modifying measurement schedules could make retention more realistic.

These considerations connect to planning retention strategies within the research protocol and assessing when attrition threatens longitudinal study feasibility.

Planning Recruitment for a Twelve-Month Educational Intervention Study

The team then evaluates alternative assumptions.

The researchers determine whether recruiting 320 participants would be feasible if attrition proved higher than expected.

They also consider whether participants in the intervention group might face greater burden and whether follow-up arrangements should differ in accessibility while preserving appropriate comparability across groups.

The final protocol documents the evidence used to select the planning assumption, the recruitment target, the retention procedures, and the missing-data analysis strategy.

The calculation does not guarantee that 240 complete outcomes will be obtained. It provides a transparent planning basis whose assumptions can be examined by reviewers and revised when justified.

Common Mistakes When Estimating Participant Attrition

Choose an Attrition Assumption You Can Defend

Start with the number of participants or observations required by the primary analysis. Define what constitutes usable outcome data and identify the follow-up point at which those data are needed.

Then estimate attrition using the strongest available evidence, recognizing uncertainty rather than concealing it behind a single percentage.

Researchers should also identify factors that may increase participant dropout in the proposed population.

Attrition planning is strongest when recruitment assumptions, retention procedures, outcome measurement, and statistical analysis are developed together rather than treated as separate administrative tasks.

Before Finalizing an Attrition Allowance

Frequently Asked Questions About Expected Participant Attrition

Plan for Attrition Using Evidence and Explicit Assumptions

How to Cite This Guide

This guide is intended to be read, shared, and used in research, teaching, and academic work. If you draw on its ideas, explanations, or other content, please acknowledge the source by citing the guide. Doing so gives appropriate credit and helps your readers locate the original resource.