CONSORT Flow Charts for Clinical Trial Reporting

ClinicoPath

2025-07-13

Introduction to CONSORT Flow Charts

The CONSORT (Consolidated Standards of Reporting Trials) statement provides evidence-based recommendations for reporting randomized controlled trials (RCTs). A key component is the flow diagram that tracks participants through each stage of the trial, from initial assessment through final analysis.

Why CONSORT Flow Charts Matter

CONSORT flow diagrams are essential for:

• Transparency: Showing exactly how many participants entered and completed the study
• Quality Assessment: Allowing readers to evaluate potential bias from dropout patterns
• Reproducibility: Providing clear documentation of study conduct
• Journal Requirements: Most medical journals require CONSORT-compliant reporting
• Regulatory Compliance: FDA and EMA expect CONSORT-style reporting

The ClinicoPath CONSORT Function

The consort() function in ClinicoPath generates publication-ready CONSORT flow diagrams automatically. Simply input your trial numbers, and it creates both visual flowcharts and summary statistics.

Basic CONSORT Flow Chart

Let’s start with a simple two-arm trial example:

# Basic cardiovascular trial
basic_consort <- consort(
  data = data.frame(),  # consort doesn't analyze patient data
  initialN = 1000,
  notEligibleN = 300,
  notEligibleText = "Age >80 years (n=150), Prior MI <6 months (n=100), Other (n=50)",
  randomizedN = 700,
  arm1Label = "ACE Inhibitor",
  arm1N = 350,
  arm1ReceivedN = 340,
  arm1LostN = 45,
  arm1AnalyzedN = 295,
  arm2Label = "Placebo",
  arm2N = 350,
  arm2ReceivedN = 345,
  arm2LostN = 40,
  arm2AnalyzedN = 305,
  excludedText = "Lost to follow-up, protocol violations"
)

The function produces: 1. Summary Table: Key numbers with percentages 2. Flow Diagram: Visual CONSORT chart 3. Text Summary: Detailed narrative description

Real-World Examples Using Sample Data

ClinicoPath includes the consort_examples_data dataset with realistic trial scenarios:

# Load the example data
data(consort_examples_data)
str(consort_examples_data)

# View the different trial types
consort_examples_data$trial_name

Example 1: Large Cardiovascular Outcomes Trial

# Extract cardiovascular trial data
cv_trial <- consort_examples_data[1, ]

# Create CONSORT flowchart
cv_consort <- consort(
  data = data.frame(),
  initialN = cv_trial$initial_assessed,
  notEligibleN = cv_trial$not_eligible,
  notEligibleText = cv_trial$not_eligible_reasons,
  randomizedN = cv_trial$randomized,
  arm1Label = cv_trial$arm1_label,
  arm1N = cv_trial$arm1_allocated,
  arm1ReceivedN = cv_trial$arm1_received,
  arm1LostN = cv_trial$arm1_lost_followup,
  arm1AnalyzedN = cv_trial$arm1_analyzed,
  arm2Label = cv_trial$arm2_label,
  arm2N = cv_trial$arm2_allocated,
  arm2ReceivedN = cv_trial$arm2_received,
  arm2LostN = cv_trial$arm2_lost_followup,
  arm2AnalyzedN = cv_trial$arm2_analyzed,
  excludedText = cv_trial$exclusion_reasons
)

This represents a typical large-scale cardiovascular outcomes trial with: - 5,000 patients initially screened - 30% exclusion rate (typical for cardiovascular studies) - Balanced randomization - ~10% dropout rate (acceptable for long-term studies)

Example 2: Cancer Immunotherapy Trial

# Extract cancer trial data
cancer_trial <- consort_examples_data[2, ]

cancer_consort <- consort(
  data = data.frame(),
  initialN = cancer_trial$initial_assessed,
  notEligibleN = cancer_trial$not_eligible,
  notEligibleText = cancer_trial$not_eligible_reasons,
  randomizedN = cancer_trial$randomized,
  arm1Label = cancer_trial$arm1_label,
  arm1N = cancer_trial$arm1_allocated,
  arm1ReceivedN = cancer_trial$arm1_received,
  arm1LostN = cancer_trial$arm1_lost_followup,
  arm1AnalyzedN = cancer_trial$arm1_analyzed,
  arm2Label = cancer_trial$arm2_label,
  arm2N = cancer_trial$arm2_allocated,
  arm2ReceivedN = cancer_trial$arm2_received,
  arm2LostN = cancer_trial$arm2_lost_followup,
  arm2AnalyzedN = cancer_trial$arm2_analyzed,
  excludedText = cancer_trial$exclusion_reasons
)

Cancer trials often have: - Strict inclusion criteria (29% exclusion here) - Higher dropout rates due to disease progression - Time-to-event endpoints requiring careful follow-up

Example 3: Small Pilot Study

# Extract pilot study data
pilot_trial <- consort_examples_data[5, ]

pilot_consort <- consort(
  data = data.frame(),
  initialN = pilot_trial$initial_assessed,
  notEligibleN = pilot_trial$not_eligible,
  notEligibleText = pilot_trial$not_eligible_reasons,
  randomizedN = pilot_trial$randomized,
  arm1Label = pilot_trial$arm1_label,
  arm1N = pilot_trial$arm1_allocated,
  arm1ReceivedN = pilot_trial$arm1_received,
  arm1LostN = pilot_trial$arm1_lost_followup,
  arm1AnalyzedN = pilot_trial$arm1_analyzed,
  arm2Label = pilot_trial$arm2_label,
  arm2N = pilot_trial$arm2_allocated,
  arm2ReceivedN = pilot_trial$arm2_received,
  arm2LostN = pilot_trial$arm2_lost_followup,
  arm2AnalyzedN = pilot_trial$arm2_analyzed,
  excludedText = pilot_trial$exclusion_reasons
)

Pilot studies typically have: - Smaller sample sizes (150 assessed, 100 randomized) - Focus on feasibility and safety - Higher proportional exclusions due to strict criteria

Advanced Usage Scenarios

Unequal Randomization (2:1 Ratio)

Some trials use unequal randomization to get more safety data on the experimental treatment:

unequal_consort <- consort(
  data = data.frame(),
  initialN = 900,
  notEligibleN = 300,
  notEligibleText = "Failed eligibility criteria",
  randomizedN = 600,
  arm1Label = "Experimental Drug (2:1)",
  arm1N = 400,  # 2/3 of randomized patients
  arm1ReceivedN = 390,
  arm1LostN = 60,
  arm1AnalyzedN = 330,
  arm2Label = "Standard Care (2:1)",
  arm2N = 200,  # 1/3 of randomized patients
  arm2ReceivedN = 195,
  arm2LostN = 25,
  arm2AnalyzedN = 170,
  excludedText = "Lost to follow-up, adverse events"
)

High Dropout Scenario

Some studies face high dropout rates that need careful documentation:

high_dropout_consort <- consort(
  data = data.frame(),
  initialN = 500,
  notEligibleN = 100,
  notEligibleText = "High-risk population with multiple comorbidities",
  randomizedN = 400,
  arm1Label = "Intensive Intervention",
  arm1N = 200,
  arm1ReceivedN = 185,
  arm1LostN = 75,  # 40% dropout
  arm1AnalyzedN = 110,
  arm2Label = "Standard Care",
  arm2N = 200,
  arm2ReceivedN = 195,
  arm2LostN = 45,  # 23% dropout
  arm2AnalyzedN = 150,
  excludedText = "High burden intervention, patient withdrawal, adverse events"
)

CONSORT Guidelines and Best Practices

Essential Elements

Every CONSORT flow diagram should include:

1. Enrollment:
   • Number assessed for eligibility
   • Number excluded (with reasons)
   • Number randomized
2. Allocation:
   • Number allocated to each intervention
   • Number receiving intended treatment
3. Follow-up:
   • Number lost to follow-up (with reasons)
   • Number discontinuing intervention
4. Analysis:
   • Number analyzed for primary outcome
   • Exclusions from analysis

Common Patterns by Study Type

Phase I Dose Escalation

# Small numbers, safety focus
phase1_consort <- consort(
  data = data.frame(),
  initialN = 60,
  notEligibleN = 15,
  notEligibleText = "Strict safety criteria",
  randomizedN = 45,
  arm1Label = "Dose Level 1",
  arm1N = 15,
  arm1ReceivedN = 15,
  arm1LostN = 1,
  arm1AnalyzedN = 14,
  arm2Label = "Dose Level 2", 
  arm2N = 15,
  arm2ReceivedN = 14,
  arm2LostN = 2,
  arm2AnalyzedN = 12
)

Phase II Efficacy Studies

# Moderate size, efficacy focus
phase2_consort <- consort(
  data = data.frame(),
  initialN = 300,
  notEligibleN = 80,
  randomizedN = 220,
  arm1Label = "Experimental",
  arm1N = 110,
  arm1AnalyzedN = 95,
  arm2Label = "Control",
  arm2N = 110,
  arm2AnalyzedN = 98
)

Phase III Confirmatory Trials

# Large scale, definitive evidence
phase3_consort <- consort(
  data = data.frame(),
  initialN = 3000,
  notEligibleN = 1000,
  randomizedN = 2000,
  arm1Label = "New Treatment",
  arm1N = 1000,
  arm1AnalyzedN = 850,
  arm2Label = "Standard Care",
  arm2N = 1000,
  arm2AnalyzedN = 875
)

Integration with jamovi

The CONSORT function integrates seamlessly with jamovi:

1. Open jamovi and load your module
2. Navigate to Exploration → ClinicoPath Reports → CONSORT Flowchart
3. Enter your trial numbers in the interface:
   • Enrollment numbers
   • Randomization details
   • Arm-specific data
   • Exclusion reasons
4. Review outputs:
   • Summary statistics table
   • Visual flow diagram
   • Text summary

Tips for jamovi Users

• Start with estimates: Begin with approximate numbers and refine
• Use descriptive labels: Clear arm names improve readability
• Include detailed reasons: Specific exclusion text enhances transparency
• Export for publication: Use the generated diagrams directly in manuscripts

Statistical Considerations

Sample Size and Power

CONSORT diagrams help readers assess whether the study had adequate power:

# Calculate effective sample sizes from our examples
cv_effective_n <- cv_trial$arm1_analyzed + cv_trial$arm2_analyzed
cancer_effective_n <- cancer_trial$arm1_analyzed + cancer_trial$arm2_analyzed

cat("Cardiovascular trial analyzed N:", cv_effective_n, "\n")
cat("Cancer trial analyzed N:", cancer_effective_n, "\n")

# Dropout rates
cv_dropout_rate <- (cv_trial$arm1_lost_followup + cv_trial$arm2_lost_followup) / cv_trial$randomized * 100
cancer_dropout_rate <- (cancer_trial$arm1_lost_followup + cancer_trial$arm2_lost_followup) / cancer_trial$randomized * 100

cat("CV trial dropout rate:", round(cv_dropout_rate, 1), "%\n")
cat("Cancer trial dropout rate:", round(cancer_dropout_rate, 1), "%\n")

Bias Assessment

CONSORT diagrams help identify potential sources of bias:

• Imbalanced dropout: Unequal loss between arms
• High overall dropout: >20% may compromise validity
• Differential reasons: Different dropout reasons between arms
• Post-randomization exclusions: Should be minimal and balanced

Reporting Standards

Different journals may have specific requirements:

• High-impact journals: Often require detailed CONSORT compliance
• Specialty journals: May have field-specific modifications
• Regulatory submissions: FDA/EMA have specific formatting expectations

Troubleshooting Common Issues

Inconsistent Numbers

Ensure logical flow:

# Check: randomized = arm1 + arm2
randomized_check <- arm1N + arm2N == randomizedN

# Check: analyzed <= received <= allocated
arm1_logic <- arm1AnalyzedN <= arm1ReceivedN & arm1ReceivedN <= arm1N
arm2_logic <- arm2AnalyzedN <= arm2ReceivedN & arm2ReceivedN <= arm2N

Missing Data Handling

Be explicit about: - Patients with missing primary endpoint data - Protocol violations - Consent withdrawals - Lost to follow-up vs. study discontinuation

Complex Study Designs

For crossover, cluster randomized, or adaptive designs: - Modify the standard CONSORT template - Add additional flowchart elements as needed - Consult CONSORT extensions for specific designs

Conclusion

The ClinicoPath CONSORT function provides:

✅ Easy generation of publication-ready flow diagrams
✅ Automatic calculations of percentages and summaries
✅ Professional formatting meeting journal standards
✅ Comprehensive examples for different trial types
✅ Integration with jamovi for interactive use

Next Steps

1. Practice with the provided examples
2. Adapt the parameters to your own trial data
3. Export diagrams for your publications
4. Combine with other ClinicoPath functions for complete trial reporting

References

• Schulz KF, Altman DG, Moher D, for the CONSORT Group. CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMJ 2010;340:c332.
• CONSORT website: http://www.consort-statement.org/
• Clinical trial reporting guidelines: https://www.equator-network.org/

---

For more information about ClinicoPath, visit our documentation at ClinicoPath Guide