01 · The Question
Does a Smaller Follow-Up Sample Become a Different Study?
A study originally enrolled 1,000 participants. Two years later, another paper reports results for only 680 of them. The title is different, the sample size has changed, and perhaps the author list has changed too. Should you treat the follow-up paper as another study?
Usually not if those 680 people came from the original participant pool. Longitudinal studies commonly lose participants over time, and follow-up analyses may also apply additional eligibility or data-availability requirements. A smaller analytic sample therefore does not by itself establish a new independent population.
The important task is to reconstruct how the follow-up sample relates to the original cohort and then determine how the later results fit your review question.
03 · What You Need to Know
Why Follow-Up Samples Often Differ From the Original Sample
A Follow-Up Paper Can Be Another Report of the Same Study
A study can generate reports at several stages. PRISMA 2020 distinguishes the underlying study from the reports that provide information about it, while Cochrane recommends linking multiple reports so that the study rather than the individual publication remains the unit of interest.
A later publication containing only surviving or reassessed participants can therefore remain part of the same underlying study. Cochrane specifically identifies differences in follow-up length and the selection of subgroups for secondary goals as characteristics that can explain why sample sizes differ between reports.
Before interpreting the smaller sample, establish whether the papers originate from the same underlying investigation .
Start With the Original Participant Flow
Do not compare only the final sample sizes. Reconstruct the path from original enrollment to the follow-up analysis.
Original sample How many participants entered the parent study?
Eligible for follow-up Were all original participants eligible to be reassessed?
Reached or reassessed How many participants actually completed the later assessment?
Included in analysis Were further participants excluded because of missing outcomes, covariates, protocol requirements, or another analytical decision?
This sequence helps distinguish ordinary longitudinal attrition from a deliberately selected subsample.
Attrition Is Not the Only Reason the Sample May Shrink
A follow-up paper can contain fewer participants for several reasons. Some participants may withdraw, become unreachable, decline further participation, or die. Others may lack the outcome required for the particular analysis. A companion paper might also focus on a subgroup or on participants from selected sites.
These mechanisms are not methodologically interchangeable. Loss to follow-up concerns who remained observable over time, whereas a deliberately restricted analysis concerns who the investigators chose to include for a particular question.
Why the Sample Is Smaller
What Happened
What to Examine
Loss to follow-up
Some original participants were not observed at the later time point
Attrition rates, reasons for loss, and differences between retained and lost participants
Missing outcome data
Participants remained in the study but lacked the measurement needed for this analysis
Missing-data pattern and analytical handling
Subgroup analysis
The paper intentionally analyzes only participants with particular characteristics
Subgroup definition and whether it was prespecified
Site restriction
Only participants from selected study locations are included
How and why sites were selected
Additional eligibility requirement
The follow-up analysis applies criteria not required for the parent sample
Whether the restriction changes the population represented by the result
Measurement subsample
Only participants receiving an additional test or assessment are analyzed
How participants entered that subsample
The Follow-Up Sample Is Usually Nested Within the Original Population
If 700 of 1,000 original participants remain at follow-up, those 700 are ordinarily part of the original population rather than 700 newly recruited independent observations.
This creates a nested relationship. The baseline and follow-up samples are not identical, but they overlap because the later sample originates from the earlier one. Understanding this relationship is part of determining whether papers use the same participants .
Different Sample Sizes Do Not Prove Independence
This point deserves emphasis because sample size is an appealing shortcut. A baseline paper with 1,000 participants and a follow-up paper with 630 participants can look like two distinct datasets in an extraction spreadsheet.
The numbers tell you only that the analytic populations differ in size. They do not tell you whether the people are different. Study identifiers, recruitment information, explicit descriptions of the parent cohort, baseline characteristics, and citations to earlier reports are much more informative.
Later Follow-Up Can Provide Genuinely New Information
Saying that the follow-up is not an independent study does not mean it contributes nothing new. Measuring the same participants later can answer an important question about persistence, delayed effects, recurrence, long-term harms, or other temporal outcomes.
New independent participants
Additional people who were not members of the original participant population.
New follow-up information
Additional observations collected later from some or all of the original participants.
A longitudinal follow-up can add substantial information of the second kind without providing the first.
Check Whether Your Review Distinguishes Follow-Up Windows
The relevance of a later paper depends partly on your review's outcome framework. A review may distinguish immediate, short-term, medium-term, and long-term outcomes, or it may define another set of time windows appropriate to the field.
If both the original and follow-up results fall into separate prespecified windows, both may be useful. They should nevertheless remain connected to the same study.
If several time points fall within one outcome window, you may need a prespecified rule for selecting among them rather than entering correlated observations as though they came from separate studies.
Attrition Can Change What the Follow-Up Result Represents
A smaller follow-up sample is not merely an accounting issue. If participants who remain differ systematically from those lost, the later result may represent a selected subset of the original population.
For example, participants with more severe symptoms might be less likely to return, or participants benefiting from an intervention might be more willing to remain involved. The direction and importance of such differences depend on the study and cannot simply be assumed.
When assessing a follow-up result, examine the extent and reasons for missing outcome data and how investigators handled them analytically. For randomized trials, the Cochrane risk-of-bias framework treats bias due to missing outcome data as a distinct domain requiring consideration.
A Subsample Follow-Up May Answer a Narrower Question
Sometimes the later paper intentionally studies only a subset of the parent sample. Perhaps imaging was available at only two centers, or only participants above a particular age received an additional assessment.
The result may still be useful, but it now applies to a narrower analytic population. Do not casually transfer the subgroup result to the entire original cohort unless the analysis supports that interpretation.
Do Not Add the Original and Follow-Up Sample Sizes Together
If a study enrolls 1,000 participants and 700 are reassessed later, the study has not thereby generated 1,700 unique participants. Reporting participant totals by simply summing samples across linked publications can substantially exaggerate the size of the evidence base.
This is a straightforward form of an overlapping study population . Publication-level sample sizes should therefore not be summed unless you have established that the populations are mutually exclusive.
Watch Out
A smaller follow-up sample can affect both independence and validity. The participants are usually connected to the original study, while attrition or selective inclusion may also change which part of the original population the later result represents. These are separate issues and both deserve attention.
04 · A Practical Example
Tracing 1,000 Original Participants Into a Smaller Follow-Up Sample
Hypothetical Example
A Three-Year Follow-Up of an Educational Intervention
Suppose a randomized educational intervention originally enrolled 1,000 students. The primary paper reports end-of-year outcomes for 940 students. Three years later, a follow-up paper reports an outcome for 620 students.
Original enrollment 1,000 students were randomized.
Primary analysis 940 students provided the outcome used in the original publication.
Follow-up contact 710 of the original students completed the three-year assessment.
Follow-up analysis 620 of those students had complete information for the outcome and covariates required by the reported analysis.
Interpretation The 620-person analysis is a later, more restricted observation of the original study population, not a new independent study of 620 newly recruited students.
If your review includes three-year outcomes, the follow-up paper may be exactly the source you need. You would record the 620-person analytic sample for that result while preserving its relationship to the original 1,000-person trial.
You would also examine why 380 original participants are absent from the reported analysis. That question concerns the interpretation and potential bias of the follow-up result, not merely its bibliographic relationship to the earlier paper.
06 · What This Means for You
Record the Parent Sample and Every Step Leading to the Follow-Up Analysis
Treat the follow-up publication as a linked report and extract enough information to reconstruct its analytic population. At minimum, record the original enrollment, number eligible for follow-up, number assessed, number analyzed, reasons for exclusions or missingness, outcome, and follow-up time.
A simple decision framework
If the follow-up participants come from the original sample
Keep the publication linked to the parent study rather than treating it as an independent study.
If the follow-up reports a prespecified later outcome window
Use the result in the appropriate follow-up synthesis while preserving the study relationship.
If only a subgroup of original participants is analyzed
Record the subgroup definition and interpret the result for that analytic population rather than automatically generalizing it to the full cohort.
If substantial participants are missing at follow-up
Examine the reasons, patterns, and analytical handling of missing outcome data when assessing the result.
If the later paper adds genuinely new participants
Determine whether the design represents an extension of the original cohort, a new recruitment wave, or a distinct study before deciding how the populations should be classified.
When the follow-up paper supplies information unavailable elsewhere, use the companion report without treating its retained participants as newly independent .
07 · A Quick Checklist
What to Check in a Partial-Sample Follow-Up Paper
Before using a follow-up result, check:
Whether the later publication explicitly identifies the original study, trial, cohort, registry, or participant pool.
How many participants originally entered the study and how many were eligible for the later assessment.
How many original participants were successfully followed and how many entered the reported analysis.
Why participants were lost, excluded, or missing from the later analysis.
Whether the later paper uses the full retained cohort or an intentionally selected subgroup.
Whether retained and lost participants differ in characteristics relevant to the outcome.
Which outcome and follow-up time the later report represents within your review framework.
Whether the same participants already contribute another estimate to the synthesis in which you plan to use the follow-up result.
Whether participant totals in your review refer to unique people rather than sums of overlapping samples across reports.
11 · Cite this Guide
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.
Recommended (Field Guide)
APA
MLA
Chicago
Copy Citation
