How to Make a CONSORT Diagram for an RCT (CONSORT 2025)

You finished your randomized controlled trial, and now you need the figure every journal requires before they'll even read your results: the CONSORT flow diagram. Built by hand in PowerPoint or Illustrator, it's a fiddly grid of boxes and arrows that breaks the moment your numbers change. This guide shows you how to make a CONSORT diagram for a randomized trial aligned with the CONSORT 2025 statement by describing it in plain language, so the AI lays out the four stages — enrollment, allocation, follow-up, and analysis — with the counts and exclusion reasons in the right place, and keeps them editable when your numbers move.

By the end of this guide you'll be able to:

• Make a CONSORT 2025 flow diagram for an RCT with the four standard stages — no manual box drawing.
• Show enrollment, allocation, follow-up, and analysis with the right counts in each box.
• Add exclusion reasons, discontinued-intervention counts, and primary-outcome losses as CONSORT 2025 requires (or adjust wording if your journal still asks for CONSORT 2010).
• Build parallel, multi-arm, or crossover versions and export a print-ready diagram for submission.

Everything below happens inside the SciDraw AI editor using the CONSORT flow diagram generator — you describe the trial and the numbers, the AI lays out the boxes, and you refine it with words.

What a CONSORT diagram must contain

The CONSORT (Consolidated Standards of Reporting Trials) flow diagram tracks every participant from screening to analysis. Under CONSORT 2025 (which supersedes CONSORT 2010), the standard two-group parallel diagram has four vertical stages, each a row of boxes:

1. Enrollment (or Enrolment) — number assessed for eligibility, those excluded (broken down into not meeting inclusion criteria, declined to participate, and other reasons), and the number randomized.
2. Allocation — how many were allocated to each arm, how many received the allocated intervention, and how many did not receive it (with reasons).
3. Follow-up — how many discontinued intervention (with reasons) and how many were lost to follow-up for primary outcome (with reasons) in each arm.
4. Analysis — how many were analysed for primary outcome in each arm, and how many were excluded from analysis (with reasons).

In CONSORT 2010, the follow-up box listed "Lost to follow-up" before "Discontinued intervention", and the analysis box used "Analysed" without specifying "for primary outcome". The numbers must still reconcile down each column: every participant is accounted for from top to bottom so reviewers can check for attrition bias.

What changed in the CONSORT 2025 flow diagram

Published in April 2025 by the SPIRIT–CONSORT Group (consort-spirit.org; Hopewell et al., BMJ 2025;389:e081123, DOI: 10.1136/bmj-2024-081123), CONSORT 2025 supersedes the CONSORT 2010 statement. For the participant flow diagram of a standard two-group parallel trial, the updates are targeted:

• The four stages are unchanged. The flow still moves through Enrollment, Allocation, Follow-up, and Analysis, with no fifth stage added and the same pre-randomization and allocation box structure.
• Follow-up items swapped order. Within each arm's follow-up box, Discontinued intervention (give reasons) now appears first, followed by Lost to follow-up for primary outcome (give reasons).
• Loss to follow-up is scoped to the primary outcome. The label now reads "Lost to follow-up for primary outcome" rather than a generic "Lost to follow-up".
• Analysis is scoped to the primary outcome. The analysis box label is now "Analysed for primary outcome" (alongside "Excluded from analysis (give reasons)"), clarifying which outcome the analyzed sample size refers to.
• Checklist numbering updated. The overall CONSORT checklist expanded from 25 to 30 numbered items, and the participant flow diagram items moved from 13a/13b in CONSORT 2010 to 22a/22b in CONSORT 2025.

Which version should you submit? CONSORT 2025 replaces 2010 as the default standard, and the official flow diagram template is available on consort-spirit.org. Because journals update their instructions for authors on different schedules — some already require CONSORT 2025, some still list CONSORT 2010, and some accept either version during the transition — check your target journal's author guidelines before submission. Also note that extensions for specific trial designs (such as cluster randomized trials and non-inferiority or equivalence trials) remain listed under the CONSORT 2010-based extensions directory until updated extensions are aligned.

Step 1: Describe the trial structure

Open the editor and tell the AI the trial type and the arms first. A parallel two-arm trial and a three-arm trial have different layouts, so set the shape before the numbers.

A filled-in example:

• "Create a CONSORT 2025 flow diagram for a parallel two-arm RCT comparing Drug A vs. Placebo. Four stages stacked vertically: Enrollment, Allocation, Follow-up, Analysis. After randomization the flow splits into two columns. Add a side box at enrollment for excluded participants with reasons, and in each arm show discontinued intervention, lost to follow-up for primary outcome, and analysed for primary outcome."

Step 2: Fill in the counts

Now add the numbers stage by stage using the CONSORT 2025 box labels (or tell the AI to use CONSORT 2010 labels if your target journal still requires them). The AI places them in the correct boxes and keeps the columns reconciling.

• "Enrollment: 540 assessed for eligibility, 120 excluded (80 did not meet inclusion criteria, 25 declined to participate, 15 other reasons), 420 randomized."
• "Allocation: 210 allocated to Drug A (200 received allocated intervention, 10 did not receive allocated intervention with reasons), 210 allocated to Placebo (205 received allocated intervention, 5 did not receive allocated intervention with reasons)."
• "Follow-up: Drug A discontinued intervention 7 (with reasons), lost to follow-up for primary outcome 8 (with reasons); Placebo discontinued intervention 5 (with reasons), lost to follow-up for primary outcome 7 (with reasons)."
• "Analysis: Drug A 195 analysed for primary outcome, 5 excluded from analysis (with reasons); Placebo 198 analysed for primary outcome, 7 excluded from analysis (with reasons)."

Listing specific reasons for exclusions, discontinued interventions, and losses to follow-up is what distinguishes a compliant CONSORT diagram from a generic flowchart.

Step 3: Add stage labels and side branches

CONSORT diagrams use a left-hand strip of stage labels and right-side branches for exclusions. Make both explicit.

• "Add the stage labels 'Enrollment', 'Allocation', 'Follow-up', and 'Analysis' down the left side."
• "Place exclusion boxes as side branches to the right, connected with a small horizontal arrow."
• "Draw the randomization step as a single box that splits into the two arms below it."
• "Use rectangular boxes with thin connecting arrows and consistent spacing."

The side-branch arrows visually peel participants away from the main flow, which is exactly how readers expect attrition to be shown. To tweak any single box's wording later, see how to edit text and labels in AI figures.

Step 4: Adapt to multi-arm, crossover, or extension designs

Not every trial is a standard two-group parallel RCT. The diagram structure adapts to your design:

• "Change this to a three-arm trial: Low dose, High dose, and Placebo, each as its own column from allocation down."
• "Make this a crossover trial showing Period 1, washout, and Period 2 with the sequence groups."
• "Add a multi-site note indicating participants were enrolled across 5 centers."

If you are reporting a specialized design such as a cluster randomized trial or a non-inferiority / equivalence trial, check the official CONSORT extensions — those design-specific extensions are still based on CONSORT 2010 until their 2025 updates are completed. Tell the AI the design and required cluster or per-protocol counts, and it rebalances the columns and arrows without you redrawing anything.

Common mistakes (and how to fix them)

• Numbers that don't reconcile. If 420 are randomized but the arms sum to 415, reviewers will catch it. Fix: "Check that allocated counts sum to the randomized total and fix the discrepancy."
• Exclusions without reasons. A box that says "120 excluded" with no breakdown fails CONSORT. Fix: "List the exclusion reasons with counts inside the enrollment side box."
• Missing a stage. Dropping the follow-up row hides attrition. Fix: "Add the Follow-up stage showing discontinued intervention and lost to follow-up for primary outcome for each arm."
• Crowded, misaligned boxes. Inconsistent spacing reads as careless. Fix: "Align the boxes into clean columns with even vertical spacing."
• No stage labels. Readers shouldn't have to infer the stages. Fix: "Add the four CONSORT stage labels down the left margin."
• Wrong arm count for the design. A three-arm trial drawn as two arms misrepresents the study. Fix: "Split allocation into three columns, one per arm."

Export and use

When the diagram is ready, export it for submission:

• "Export this as a 300 DPI PNG sized for a single journal column."
• "Give me an SVG so I can make final edits in Illustrator."
• "Export at high resolution with a white background for the manuscript."

The CONSORT diagram pairs naturally with two other reporting figures. If your work is a systematic review of trials, see how to make a PRISMA flow diagram for study selection — it's the review-level cousin of CONSORT. And for reporting your trial's survival outcomes, see how to make a Kaplan-Meier plot. Everything stays editable in the SciDraw AI editor.

Frequently asked questions

How do I make a CONSORT flow diagram without PowerPoint or Illustrator? Open the CONSORT flow diagram generator, describe your trial type and arms, then give the counts for each stage. The AI lays out the four-stage flow with reconciled boxes and side branches. You edit any number or label by typing — no manual box-and-arrow drawing.

What are the four stages of a CONSORT 2025 diagram? Enrollment (assessed for eligibility, excluded with reasons, randomized), Allocation (allocated, received allocated intervention, and did not receive it with reasons per arm), Follow-up (discontinued intervention and lost to follow-up for primary outcome with reasons per arm), and Analysis (analysed for primary outcome and excluded from analysis with reasons per arm). The counts must reconcile down each column.

What's the difference between a CONSORT and a PRISMA diagram? CONSORT tracks participants through a single randomized trial, while PRISMA tracks studies through a systematic review (identification, screening, eligibility, inclusion). They look similar but serve different reporting standards; use CONSORT for an RCT and PRISMA for a review.

How do I show exclusion reasons in a CONSORT diagram? Put them in a side box branching off the main flow at enrollment (e.g., did not meet inclusion criteria, declined to participate, other reasons), and list reasons inside each arm's allocation, follow-up, and analysis boxes for participants who did not receive the intervention, discontinued, were lost to follow-up for the primary outcome, or were excluded from analysis. The generator places these boxes and branches for you.

Can I make a CONSORT diagram for a three-arm or crossover trial? Yes. Tell the CONSORT flow diagram generator your design — parallel multi-arm, crossover, or multi-site — and it adds the right number of columns and adapts the allocation and follow-up rows accordingly.

Make your CONSORT diagram now

Stop nudging boxes in PowerPoint. Describe your trial, enter the counts once, and let the AI build a compliant, reconciled CONSORT 2025 flow diagram that stays editable when your numbers change. Open the CONSORT flow diagram generator or jump into the SciDraw AI editor to make yours now.