By MedClinRes.org Editorial Team
A headline number — “1,200 patients enrolled” or “320 participants completed the study” — tells you almost nothing on its own about how a clinical trial actually ran. The number that makes it into an abstract or a press release is the end of a chain of decisions: who was screened, who was excluded and why, who was randomized, who dropped out along the way, and who was ultimately included in the final analysis. Each link in that chain can be read before treating a single final sample size as the whole story.
That is the specific job of a CONSORT participant-flow diagram. It does not replace the protocol, the published methods section, or the statistical analysis plan, but it gives readers a structured map of where participants entered and exited a trial.
What the CONSORT Flow Diagram Is
CONSORT stands for CONsolidated Standards Of Reporting Trials. The CONSORT Statement is a reporting guideline built around a 30-item checklist and an accompanying flow diagram, intended to help trial authors report their work completely and transparently so readers have what they need to critically appraise, interpret, and use the research. The checklist covers how a trial was designed, analyzed, and interpreted. The flow diagram has a narrower, more mechanical role: it maps how every participant moved through the trial, from initial assessment to the final analysis.
In short, the checklist tells a reader what was done and why. The flow diagram tells a reader what happened to each person who walked through the door.
Why a Single Final Number Undersells the Study
As a hypothetical illustration: imagine two trials that both report “300 participants analyzed.” In the first, 310 people were screened, 305 were found eligible, 300 were randomized, and all 300 completed follow-up with no losses. In the second, 900 people were screened, 450 were excluded for not meeting eligibility criteria, 450 were randomized, 120 withdrew or were lost to follow-up, and 330 were eventually available for analysis — of whom only 300 had complete data for the primary outcome.
Those are very different studies. The analyzed sample size is identical in both, but what it represents — how selective enrollment was, how much attrition occurred, and how close the analyzed group is to the originally randomized group — is not. A reader who sees only “n = 300” cannot tell these two scenarios apart. A reader who sees the flow diagram can.
Reading the Diagram in Order: Screening, Allocation, Follow-Up, Analysis
A CONSORT-style flow diagram is organized as a sequence of stages, each with its own count. Reading them in order, rather than skipping straight to the bottom box, is what turns the diagram from decoration into information.
- Screening/enrollment: How many people were assessed for eligibility, and how many were excluded before randomization, ideally broken out by reason (did not meet inclusion criteria, declined to participate, or other stated reasons).
- Allocation/randomization: How many eligible participants were actually randomized, and how they were split across study arms.
- Follow-up: How many participants in each arm completed the intervention period versus discontinued, with reasons where reported (adverse events, withdrawal of consent, lost contact, and so on).
- Analysis: How many participants in each arm were included in the final statistical analysis, and whether that count matches the number originally randomized or is smaller because of later exclusions.
Each count answers a different question. The screening numbers show how selective the trial's entry criteria were in practice. The allocation numbers confirm how the study population was actually split. The follow-up numbers show where attrition occurred and in which arm. The analysis numbers show whether the people analyzed are close to the people originally randomized, which matters for judging how representative the reported result is of the full randomized group.
An Annotated Key: What Each Count Actually Answers
Used as a working reference rather than a one-time glance, here is what each stage of the diagram is there to tell you:
- Assessed for eligibility (n): How wide was the initial pool before any exclusions?
- Excluded before randomization, with reasons: Was the final study population narrowed mainly by clinical criteria, by participant choice, or by something else? This shapes how broadly the results should be read.
- Randomized (n), by arm: Did allocation produce roughly balanced groups, and does the randomized total match what the methods section describes as the planned sample?
- Received intervention as allocated vs. did not: Was there meaningful non-adherence to assigned treatment before follow-up even began?
- Lost to follow-up / discontinued, with reasons, by arm: Did one arm lose participants faster than the other, and for what stated reasons? Uneven attrition between arms is a detail worth noting, not a verdict on the trial.
- Analyzed (n), by arm: Does the analyzed count match the randomized count? If not, that gap is the first thing worth checking in the statistical analysis plan, which should explain whether the trial used an intention-to-treat population, a per-protocol population, or another defined analysis set.
Read this way, the diagram works less like a static figure and more like an index: each box points to a question the full study report is positioned to answer in more detail.
What the Diagram Can Reveal — and What It Cannot
The flow diagram is a structural account of participant movement, not an assessment of trial quality or conduct. A few boundaries matter here:
- The diagram can show how many participants were lost at each stage and, if reported, the stated reason category. It cannot by itself explain why attrition happened beyond what is listed, and dropout is not evidence of misconduct. Participants leave trials for ordinary reasons — relocation, time burden, resolved symptoms, unrelated life events — and attrition counts alone say nothing about how a trial was conducted.
- The diagram shows which analysis population was used in terms of participant counts, but the definition of that population — intention-to-treat, modified intention-to-treat, per-protocol — and the reasoning behind it belong to the statistical analysis plan and methods section, not the diagram itself.
- The diagram does not report outcome results. It shows who was measured, not what was found. Effect sizes and confidence intervals are reported elsewhere in the paper.
- The diagram reflects what the trial team chose to report. Questions about the actual eligibility criteria, what counted as a protocol deviation, or how missing data were handled sit one level deeper, in the protocol itself, and have to be checked there directly.
In short, the diagram is a map, not an explanation. It tells a reader where to look next — in the protocol, the methods section, or the statistical analysis plan — rather than answering every question on its own.
Why This Matters for NIH-Funded Trials
Participant flow isn't only a reporting courtesy; for many trials it intersects with a federal requirement. NIH-funded clinical trials are expected to register and submit results information to ClinicalTrials.gov, under the NIH Policy on Dissemination of NIH-Funded Clinical Trial Information, which implements the reporting requirements of Section 801 of the Food and Drug Administration Amendments Act (FDAAA 801), as carried out through 42 CFR Part 11. That policy exists specifically so that a trial's registration and results, including how participants moved through the study, are findable and checkable rather than left to a single published summary number.
Reading the Diagram Alongside the Rest of the Report
None of this replaces reading a study's methods directly. A flow diagram works best paired with the sections of a trial report that explain the decisions behind the numbers:
- Eligibility criteria and the reasoning behind specific exclusions, detailed in the protocol and methods.
- What counted as a successful or failed outcome, and how confident the trial's effect estimate is, covered in outcomes and confidence-interval reporting.
- How a crossover design accounts for carryover effects between treatment periods, where relevant to the trial design.
- How study status terminology on a registry differs from posted results, since a trial can be marked “completed” well before results are posted, and status alone doesn't indicate what was found.
For readers following up on these specific points: how washout periods work in crossover trial designs, and why a trial's status and its posted results are two different fields on a registry.
The same habit of checking the actual analyzed population before taking a headline claim at face value applies to product-specific evidence reviews as well: see an evidence review of a specific supplement's claims and an evaluation of health-label claims against available evidence.
The Takeaway
A final sample size is a single data point. A participant-flow diagram is a record of the decisions and losses that produced that data point. Reading screening, allocation, follow-up, and analysis counts in sequence, and knowing which question each stage answers and which it doesn't, is what separates taking a trial's headline number at face value from actually understanding what it represents.