Strata Academy

CONSORT Checklist Explained for RCT Reporting

25-item checklist, flow diagram, reporting vs risk of bias, and pairing with ROB 2

• CONSORT checks reporting transparency — ROB 2 checks whether the effect estimate is biased.
• The flow diagram reconciles randomised vs analysed numbers; discrepancies signal attrition or exclusion problems.
• High-yield items: randomisation, blinding, primary outcome pre-specification, harms, and trial registration.
• Use CONSORT extensions for cluster, non-inferiority, and pragmatic designs — the base checklist alone is not enough.
• In journal club, complete an abbreviated CONSORT table before assigning ROB 2 domain judgements.

1. What is CONSORT?

CONSORT (Consolidated Standards of Reporting Trials) is an evidence-based minimum set of recommendations for reporting randomised controlled trials. It helps readers assess whether a trial was conducted and reported with enough transparency to judge validity.

The current CONSORT 2010 statement, with its 25-item checklist and flow diagram, is maintained via the EQUATOR Network and endorsed by major medical journals including The BMJ and The Lancet. Most RCT manuscripts you encounter in journal club or during clinical attachments will cite CONSORT — or should.

CONSORT is a reporting guideline – not a risk-of-bias tool. Complete reporting makes ROB 2 easier; incomplete reporting forces 'unclear' judgements in bias domains. A beautifully written discussion section cannot compensate for missing randomisation detail in the methods.

For UK medical students, CONSORT often appears in OSCE-style critical appraisal stations, SSC projects, and intercalated dissertation marking. Examiners expect you to distinguish 'not reported' from 'not done' and to explain why that matters for interpreting the primary outcome.

The ICMJE recommends trial registration as a condition of publication. UK NIHR-funded trials must register — if a major trial lacks registration, note it as a serious reporting deficiency regardless of journal prestige.

Abstract reporting: CONSORT for Abstracts requires structured items including trial design, N, primary outcome, harms, and registration. Many student appraisals skip the abstract — start there for quick CONSORT screening.

Tip: Download the official CONSORT checklist PDF and keep it beside your ROB 2 worksheet — they answer different questions about the same trial.

2. CONSORT vs ROB 2

CONSORT asks: did authors report allocation, blinding, outcomes, and harms clearly? ROB 2 asks: given what was done, how biased is the effect estimate?

A trial can have good CONSORT reporting but high ROB 2 risk (e.g. high attrition with differential loss). Conversely, poor reporting obscures bias assessment even if conduct was sound — you may be forced to rate domains as 'some concerns' solely because methods were omitted.

Students frequently conflate the two in coursework. Writing 'well reported therefore low risk of bias' is a common marker comment. Reporting quality and internal validity are related but not identical.

• CONSORT → transparency and completeness of the write-up
• ROB 2 → validity of the causal inference
• Use both for RCT appraisal — always in that conceptual order
• Registry and protocol comparison supports both CONSORT Item 24 and ROB 2 Domain 5

3. CONSORT flow diagram

The flow diagram tracks participants from enrolment through allocation, follow-up, and analysis. Compare numbers randomised vs analysed – discrepancies signal attrition or exclusion bias.

Every CONSORT flow diagram should allow you to reconstruct the denominator for the primary analysis. If 200 patients were randomised but only 140 appear in the primary outcome table, you need to know why — and whether loss differed between arms.

Cluster trials, crossover designs, and factorial trials have CONSORT extension flow templates. Using the wrong template is itself a reporting deficiency worth noting in appraisal.

When appraising trials linked to NHS datasets or registry studies, flow diagrams may appear in supplementary files. Check there before concluding the diagram is absent.

• Assessed for eligibility (n)
• Randomised (n)
• Allocated to each group, received intervention, analysed
• Losses and exclusions with reasons at each stage
• Protocol deviations and whether analysis was ITT or per-protocol

Tip: If the flow diagram is missing, note it as a major CONSORT deficiency before judging primary results.

Note: Authors sometimes report 'efficacy population' analyses without showing how many were excluded from the ITT population — trace this in the flow diagram first.

4. High-yield checklist items

For student appraisal, these items most often drive interpretability. You do not need to memorise all 25 items for journal club, but you should know where to look when a result seems too good to be true.

Items 3–6 cover trial design, eligibility, settings, and interventions. Can you replicate the intervention from the methods alone? Vague descriptions ('standard care') make both CONSORT and clinical application fail.

Items 6–7 and 12 cover outcomes and sample size. The primary outcome must be defined a priori with time point and measurement method. Post-hoc switching is a CONSORT and ROB 2 Domain 5 issue.

Items 8–10 cover randomisation and blinding — the backbone of ROB 2 Domains 1, 2, and 4. 'Randomised' without sequence generation or concealment detail is incomplete reporting, not evidence of adequate randomisation.

Adaptive and platform trials have dedicated CONSORT extensions. If you encounter basket or umbrella trial designs in oncology journal club, download the matching extension before scoring the base checklist alone.

• Trial design description (parallel, factorial, crossover)
• Eligibility criteria and settings — who was excluded and why
• Interventions described so they can be replicated in another centre
• Primary outcome defined a priori with time point
• Sample size calculation and interim analyses declared
• Randomisation method and allocation concealment
• Blinding – who was blinded to what
• Statistical methods matched to design
• Harms and adverse events reported for each arm
• Trial registration number and protocol access

5. CONSORT extensions

Non-inferiority, equivalence, cluster, pragmatic, and patient-reported outcome trials have CONSORT extensions. Check EQUATOR for the matching extension when the design is specialised.

Cluster-randomised trials (common in public health and primary care research) require reporting of the unit of randomisation, intracluster correlation, and whether analysis accounted for clustering. The cluster CONSORT extension adds items the standard checklist omits.

Pragmatic trials deliberately allow co-intervention variation to mirror real NHS practice. The pragmatic trials extension helps you judge whether authors balanced external validity against internal validity — and whether they reported deviations clearly enough for ROB 2.

Non-inferiority and equivalence trials need pre-specified margins and justification. Without these, you cannot interpret whether 'no significant difference' supports non-inferiority or merely reflects imprecision.

• Cluster trials → cluster CONSORT extension
• Non-inferiority / equivalence → NI CONSORT extension
• Pragmatic trials → PRECIS-2 and pragmatic CONSORT guidance
• Patient-reported outcomes → PRO CONSORT extension
• N-of-1 trials → dedicated CONSORT extension

7. Trial registration and protocol access

CONSORT Items 23–24 require trial registration in a public registry (ISRCTN, ClinicalTrials.gov, EudraCT) and ideally access to the full protocol. Registration before first patient enrolment is best practice and reduces selective outcome reporting.

When the published primary outcome differs from the registry, flag CONSORT deficiency and ROB 2 Domain 5 (selection of reported result). Authors may explain changes in a protocol amendment — read the amendment before assuming misconduct.

UK trials often register on ISRCTN. EU trials may appear on CTIS. Cross-check registry entry date against first patient recruited — post-hoc registration is a reporting red flag.

Protocol access: many journals require submission of the full protocol as supplementary material. If only a summary protocol is available, note limited ability to assess pre-specified analyses.

For student appraisal, paste registry ID and primary outcome from registry vs paper into your worksheet — examiners reward this explicit comparison.

• Registry ID in abstract and methods (CONSORT Item 24)
• Compare registered vs published primary outcome
• Note registration date vs first enrolment
• Read statistical analysis plan if separate from protocol
• Check for post-hoc changes to sample size or endpoints

8. Worked example – landmark RCT

Apply CONSORT items to a published trial you can access in full text. The 4S trial below is a teaching staple for flow diagrams, pre-specified outcomes, and long-term follow-up reporting.

When reading any trial for journal club, open the CONSORT checklist alongside the paper. Tick items as present, absent, or unclear. Flag items that block ROB 2 assessment — those are your priority discussion points.

10. Journal club checklist (CONSORT)

First slide: participant flow diagram with randomised vs analysed highlighted. If absent, state this before any efficacy discussion.

Second slide: CONSORT gaps that block ROB 2 — randomisation detail, blinding, primary outcome registration.

Third slide: ROB 2 domain summary tied to CONSORT items. Show how reporting gaps became 'some concerns'.

Ask the room: 'Could we replicate this trial from the published report?' CONSORT-positive trials should answer yes for methods; ROB 2 answers whether we believe the result.

Time-box CONSORT scoring to ten minutes in journal club — full 25-item completion is for coursework appendices, not live presentation. Hit flow, randomisation, primary outcome, harms, and registration.

• Flow diagram on slide one
• CONSORT gaps linked to ROB 2 domains
• Registry vs published primary outcome
• Replication test for methods section
• Separate reporting from bias judgements verbally

11. Using CONSORT in journal club

Complete an abbreviated CONSORT table: item present / absent / unclear. Pair with ROB 2 domain judgements. Flag items that block bias assessment (e.g. no allocation concealment description).

Start with title and abstract — does the abstract report trial design, N randomised, primary outcome, and harms? CONSORT for Abstracts has its own checklist; many high-impact journals enforce it.

Compare ClinicalTrials.gov, ISRCTN, or EU Clinical Trials Register entries to the published paper. Outcome switching between registry and manuscript is a CONSORT Item 24 issue and ROB 2 Domain 5 concern.

For coursework, quote page numbers when marking items unclear. Examiners reward evidence-based appraisal, not generic statements that 'methods were well described'.

1. Skim abstract against CONSORT for Abstracts structure.
2. Locate or reconstruct the participant flow diagram.
3. Score high-yield items (randomisation, blinding, outcomes, harms, registration).
4. Note reporting gaps that force ROB 2 'unclear' or 'some concerns'.
5. Summarise: is the trial interpretable despite reporting limitations?

Tip: One slide with CONSORT gaps and one with ROB 2 domains is a strong journal club format.

12. Common CONSORT mistakes

Students and junior doctors repeat the same errors when first learning trial appraisal. Avoiding them makes your journal club contribution and coursework marks noticeably stronger.

• Treating CONSORT compliance as proof of low risk of bias.
• Ignoring the flow diagram and reading only the results table.
• Assuming 'double-blind' means all ROB 2 domains are low risk.
• Not checking trial registration when the primary outcome seems convenient.
• Applying the standard CONSORT checklist to a cluster or NI trial without the extension.
• Marking items 'present' when only one vague sentence mentions them.

13. StrataResearch and CONSORT

RCT manuscripts receive CONSORT-aligned reporting feedback alongside ROB 2 domain scoring – reporting gaps are separated from bias judgements.

Upload a trial PDF via quick analysis to see which CONSORT themes are addressed in the manuscript and which force uncertain ROB 2 ratings. The structured output helps you prepare journal club slides without re-reading the entire paper.

Pair StrataResearch output with your manual CONSORT checklist for coursework — supervisors often ask you to justify discrepancies between automated and human appraisal.

• Reporting gaps flagged separately from bias domains
• Flow and registration checks aligned to CONSORT items
• ROB 2 domain output on the same upload for side-by-side comparison

14. What to read next

CONSORT is one layer in full RCT appraisal. Pair it with ROB 2 for bias, our sample size and power guide for interpreting significance, and GRADE if you are synthesising multiple trials in a systematic review.

For non-randomised intervention studies, switch to ROBINS-I and STROBE — CONSORT does not apply. For systematic reviews of RCTs, use PRISMA and AMSTAR 2 on the review itself, then ROB 2 on each included trial.

Frequently asked questions

What is the CONSORT checklist?

CONSORT is a 25-item reporting guideline for randomised controlled trials. It ensures authors describe design, conduct, analysis, and results transparently — including a participant flow diagram. It does not score trial quality or risk of bias.

What is the difference between CONSORT and ROB 2?

CONSORT assesses whether the trial was reported completely enough to evaluate. ROB 2 assesses whether the effect estimate is at risk of bias. Good CONSORT reporting helps ROB 2 assessment but does not guarantee low bias.

Do I need the CONSORT flow diagram?

Yes — it is a core CONSORT item. The flow diagram shows how many participants were assessed, randomised, treated, and analysed, with reasons for exclusions. Without it, attrition and selective analysis are hard to judge.

When should I use a CONSORT extension?

Use an extension when the trial design differs from a standard parallel-group superiority RCT — for example cluster randomisation, non-inferiority, pragmatic designs, or N-of-1 trials. Check EQUATOR for the matching extension.

Can a trial fail CONSORT but still be valid?

Yes. Poor reporting limits your confidence and may force ROB 2 'unclear' judgements, but conduct may still have been rigorous. Conversely, perfect CONSORT reporting does not rule out high risk of bias from attrition, unblinding, or selective reporting.

Interactive walkthroughs and quizzes load when JavaScript is enabled — the checklist and tables above are fully readable without it.