Participant Flow Diagram¶
Reporting diagram — instantiates Attrition and Dropout Monitoring
Draws the study as a cascade of boxes — assigned, retained, measured, analyzed — so every unit lost between stages is visible on one page.
A report that jumps straight to "300 patients were analyzed" has already hidden its most important fact: how many started and where the rest went. Participant Flow Diagram is the static, top-to-bottom picture that refuses that shortcut. It draws each denominator state as a box — screened, eligible, randomized, retained, measured, analyzed — and each loss as a labelled arrow between boxes, so the reader can trace a single unbroken chain from the population the study began with to the sample the conclusion rests on. Its one job is visibility of the cascade: it is a picture that makes disappearance impossible to overlook. It is not a live monitor and it is not an analysis — it neither watches loss accrue in real time nor asks whether the loss biased anything; it simply insists that every unit be accounted for on the page.
Example¶
A dermatology team is reporting a trial of a new injectable biologic for plaque psoriasis. The Participant Flow Diagram is the first figure. It opens with 620 patients screened, narrows to 400 who met eligibility, then splits into two parallel columns at randomization: 200 to the biologic, 200 to placebo. Below each column the losses are drawn stage by stage with counts and reasons — drug arm: 14 discontinued for injection-site reactions, 6 lost to follow-up; placebo arm: 3 discontinued for perceived inefficacy, 5 lost to follow-up — down to the numbers actually analyzed at week 24. Laid out this way, the figure surfaces something a summary table would bury: the drug arm lost noticeably more people, and mostly for a treatment-specific reason. The diagram does not say what that does to the efficacy estimate. It just makes sure nobody can read the result without seeing it.
How it works¶
The distinguishing discipline is reconciliation, not detection:
- Boxes are denominator states, kept strictly separate. Enrolled is not randomized is not analyzed; the diagram never lets one silently stand in for another.
- Arrows carry counts, annotated with loss reasons where known. Each drop between boxes shows how many were lost and, at each node, why.
- Arms run as parallel columns so per-group loss is read off the same figure rather than pooled.
- Every unit must reconcile — units into a box equal units retained plus units lost. A figure that does not balance is flagging a bookkeeping error, not a design choice.
Tuning parameters¶
- Stage granularity — how many denominator boxes to draw; more stages expose where loss concentrates but crowd the figure.
- Arm splitting — whether to draw arms in parallel or pool them; pooling is cleaner but can hide differential loss.
- Reason-annotation depth — bare counts versus counts-with-reasons at each node; richer labels aid diagnosis but lengthen the figure.
- Axis choice — organizing by study stage versus by calendar time; the time axis shows when loss clustered, the stage axis shows where.
- Reconciliation strictness — whether unexplained residuals are permitted or must resolve to zero before the figure ships.
When it helps, and when it misleads¶
Its strength is that it drags the missing denominator into plain sight and makes silent completer bias hard to commit: a reader can see at a glance that the analyzed sample is smaller and differently composed than the assigned one. The CONSORT flow diagram[n1] became a reporting standard for exactly this reason.
It misleads when it is drawn only at the end, so it documents loss rather than catching it; when arms are pooled so an acceptable total conceals concentrated loss in one group; or when a tidy, well-balanced figure lends false confidence to counts that were themselves wrong upstream. A diagram is only as honest as the register feeding it. The guarding discipline is to reconcile every box, split by arm, and treat the diagram as a reporting obligation rather than the whole of monitoring.
How it implements the components¶
denominator_transition_frame— the boxes are the denominator states held apart; the figure's whole structure is this frame made visible.attrition_event_register— the between-box counts are individual loss events from the register summed into stage totals, which is why the figure must reconcile against it.
It does not watch loss rates diverge across arms and time as a running signal (differential_attrition_monitor, its nearest twin Attrition Dashboard) — the diagram is a static end-of-study snapshot — nor test whether the conclusion survives the missing (missingness_assumption_record, Missing-Data Sensitivity Analysis).
Related¶
- Instantiates: Attrition and Dropout Monitoring — the flow diagram is the archetype's reporting face, making denominator loss legible to any reader.
- Sibling mechanisms: Attrition Dashboard · Withdrawal Reason Survey or Interview · Completer Balance Table · Missing-Data Sensitivity Analysis · Retention Outreach Protocol · Data Monitoring Review
Editorial Notes¶
Form Classification¶
Form family: Representation, Specification & Plan
Rationale: Participant Flow Diagram operates as a static representation, map, specification, schema, or prospective plan that externalizes information because it draws the study as a cascade of boxes — assigned, retained, measured, analyzed — so every unit lost between stages is visible on one page.
Independent corroboration: The frozen evidence defines Participant Flow Diagram as 'Draws the study as a cascade of boxes — assigned, retained, measured, analyzed — so every unit lost between stages is visible on one page', so its operative form is Representation, Specification & Plan.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Medicine & Healthcare
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Participant Flow Diagram is most directly rooted in medicine and healthcare's clinical protocols, safety systems, evidence practices, and accountable care delivery. The lineage fits its defining practice: Draws the study as a cascade of boxes — assigned, retained, measured, analyzed — so every unit lost between stages is visible on one page.
Related originating lineages:
- Statistics & Experimental Design — Participant Flow Diagram also draws materially on experimental design and statistics' methods for comparison, uncertainty, sampling, sensitivity, and inferential validation, which shaped this mechanism rather than merely adopting it as an application.
Review resolution: Both independent reviews agree on primary origin medicine_healthcare; reconciliation resolves domain_reach_disagreement. Formative alternate lineages retained: statistics_experimental_design. The broader reach of later applications is kept separate as domain_reach=multi_domain; origin_mode=cross_disciplinary_synthesis records how the formative lineages relate. Confidence is conservatively reconciled to high, and encyclopedia_synthesis=false preserves the reviewers' boundary judgment.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
The flow diagram shows that loss happened and where in the cascade — but never why it biases or whether the reasons cluster. Those questions belong to the Completer Balance Table and Missing-Data Sensitivity Analysis. Keeping the diagram to pure bookkeeping is what makes it trustworthy: it is the one artifact whose only claim is that the numbers add up.
[n1] The CONSORT (Consolidated Standards of Reporting Trials) flow diagram is the widely adopted template for reporting participant movement through a randomized trial — enrollment, allocation, follow-up, and analysis — with losses and exclusions counted at each stage. It is cited here as the canonical instance of this mechanism, not as a specific study. ↩