Disposition Tables: CONSORT-Style Reporting

Introduction

In a clinical trial, the disposition table answers one of the most basic — and most important — questions in a Clinical Study Report: What happened to the patients who entered the study?

A well-designed disposition table traces patients from randomization or treatment assignment through treatment exposure, treatment discontinuation, study completion, and the analysis populations used for the major efficacy and safety summaries.

Although disposition tables are often treated as simple counting exercises, they are actually an important part of the reporting architecture of a clinical trial. The numbers must reconcile across the study, treatment arms, analysis datasets, and related tables. A reader should be able to move from the disposition table to the CONSORT flow diagram, subject-level listings, exposure summaries, and analysis populations without encountering unexplained differences.

What Is a Disposition Table?

A disposition table summarizes the status of study participants at important stages of a clinical trial. Depending on the study design, it may include:

• Patients randomized
• Patients treated
• Patients who completed treatment
• Patients who discontinued treatment
• Reasons for treatment discontinuation
• Patients who completed the study
• Patients who discontinued the study
• Reasons for study discontinuation
• Full Analysis Set or other efficacy-population membership
• Safety Analysis Set membership
• Other prespecified analysis populations

The exact structure should follow the protocol, Statistical Analysis Plan (SAP), table shells, and reporting requirements for the study. There is no single disposition-table layout that is correct for every trial.

Disposition Table vs. CONSORT Flow Diagram

Disposition tables and CONSORT-style flow diagrams communicate related information, but they serve different purposes.

The strongest reporting package often uses both: the CONSORT-style diagram provides the rapid visual overview, while the disposition table provides the numerical detail.

What Does “CONSORT-Style” Mean?

A CONSORT-style presentation emphasizes the movement of participants through the trial. For a randomized trial, the reader should be able to distinguish at least:

• Patients assessed for eligibility
• Patients excluded before randomization
• Patients randomized
• Patients allocated to each treatment
• Patients who received the assigned treatment
• Patients who did not receive the assigned treatment
• Patients who discontinued treatment and why
• Patients included in the relevant analyses

The presentation should make the denominators and treatment-arm assignments unambiguous. A disposition table should never force the reader to guess whether a count refers to randomized patients, treated patients, or an analysis population.

An Actual CONSORT-Style Disposition Example

The following example uses simulated data from a two-arm randomized clinical trial. The numbers are illustrative and are designed to demonstrate reconciliation.

The Fundamental Reconciliation Rule

The first rule of disposition reporting is simple:

For the example above:

and:

The same principle applies throughout the table. If a treatment-arm total does not reconcile, the discrepancy must be investigated before the table is released.

Typical Disposition Table Structure

A common CSR disposition table can be organized into several logical sections.

A Detailed Disposition Table

The following simulated table illustrates a detailed treatment-disposition presentation. Percentages are calculated using the treatment-arm denominator shown in the section.

Do the Discontinuation Reasons Need to Sum?

Yes — when the table is intended to present mutually exclusive primary reasons for discontinuation, the reason categories should reconcile to the corresponding discontinuation total.

For Treatment A in the example:

For Treatment B:

However, the exact rules depend on how the source data were collected. A patient can have multiple adverse events leading to discontinuation, but a disposition table that asks for a primary reason should generally use one mutually exclusive reason per patient.

Primary Reason vs. Multiple Reasons

This distinction is easy to overlook.

Treatment Discontinuation vs. Study Discontinuation

Treatment discontinuation and study discontinuation are not necessarily the same event.

A patient may stop study treatment because of toxicity but remain in follow-up for survival or other study assessments. Conversely, a patient may discontinue the study entirely and therefore have no subsequent study assessments.

The disposition table should preserve this distinction rather than treating every treatment discontinuation as a study withdrawal.

Study Completion

A study-completion section may be particularly important when the protocol includes long-term follow-up after treatment.

For example, a patient could:

• Complete treatment but later discontinue the study;
• Discontinue treatment but complete study follow-up;
• Remain on treatment and complete the study; or
• Discontinue both treatment and study participation.

These pathways should be defined in the SAP and implemented consistently in the analysis datasets.

Analysis Populations

Disposition reporting often leads directly into the analysis populations used in the CSR. Common examples include:

These are examples rather than universal definitions. The exact population definitions must come from the protocol and SAP.

An Analysis-Population Reconciliation Example

The differences between these populations should be explainable from the prespecified definitions and patient-level data. A programmer should be able to identify every patient who is included or excluded from each population.

Denominators Matter

One of the most common reporting errors is presenting percentages without making their denominators clear.

Suppose 98 of 100 Treatment A patients received at least one dose:

If 8 of the 26 patients who discontinued treatment did so because of an adverse event, however, the percentage could be reported as:

or, if the table is explicitly designed around the randomized denominator:

Both calculations can be mathematically valid, but they answer different questions. The table must make the denominator convention obvious.

Recommended Denominator Conventions

The key requirement is consistency with the table shell and SAP.

Disposition by Treatment Arm

For randomized trials, treatment-arm presentation is essential. Readers should not have to reconstruct the arm assignment from a total column.

A useful layout is:

• Experimental treatment
• Control treatment
• Total randomized population

If there are multiple doses or treatment combinations, each arm should be represented explicitly and the column headers should remain unambiguous.

What Happens in a Single-Arm Trial?

The CONSORT-style concept still applies, but treatment-arm columns are replaced by the single treatment population.

The same reconciliation principles apply.

Screening Failures

Screening disposition is particularly important in trials with a substantial screening population.

A useful screening summary may distinguish:

• Patients screened
• Patients eligible
• Patients not eligible
• Patients who withdrew before randomization
• Other reasons for non-randomization

Screening failures should not be silently incorporated into randomized or treated populations.

Screened vs. Enrolled vs. Randomized

These terms should not be used interchangeably.

Disposition Tables and Subject-Level Traceability

Every aggregate number in the disposition table should be traceable to subject-level records.

For example, if the table reports 26 Treatment A patients who discontinued treatment, the programmer should be able to produce the exact list of those 26 patients and their coded discontinuation reasons.

CDISC-Oriented Data Considerations

For a CDISC-oriented reporting workflow, disposition information commonly draws on subject-level information from domains such as Demographics and Disposition, with study-specific derivations and analysis datasets supporting the final table.

The exact variables and derivation logic depend on the study's implementation. Typical concepts include:

These examples are illustrative. The final table should be driven by the study's validated data model and prespecified derivations rather than by variable names copied from another study.

R Implementation

A simple disposition table can be produced in R after creating a subject-level dataset with one record per patient and the required disposition flags.

library(dplyr)

disposition <- subject_level %>%
  group_by(TRT01P) %>%
  summarise(
    Randomized = n(),
    Treated = sum(SAFETYFL == "Y", na.rm = TRUE),
    Completed_Treatment = sum(TRT_COMPLETED == "Y", na.rm = TRUE),
    Discontinued_Treatment = sum(
      TRT_DISCONTINUED == "Y", na.rm = TRUE
    ),
    Completed_Study = sum(STUDY_COMPLETED == "Y", na.rm = TRUE),
    Discontinued_Study = sum(
      STUDY_DISCONTINUED == "Y", na.rm = TRUE
    ),
    .groups = "drop"
  )

The aggregation step should occur only after the subject-level derivations have been validated.

Calculating Percentages in R

disposition <- disposition %>%
  mutate(
    Treated_Pct = 100 * Treated / Randomized,
    Completed_Treatment_Pct =
      100 * Completed_Treatment / Randomized,
    Discontinued_Treatment_Pct =
      100 * Discontinued_Treatment / Randomized
  )

If a table requires percentages among patients who discontinued treatment, use that population as the denominator instead.

reason_summary <- subject_level %>%
  filter(TRT_DISCONTINUED == "Y") %>%
  group_by(TRT01P, TRT_DISC_REASON) %>%
  summarise(N = n(), .groups = "drop") %>%
  group_by(TRT01P) %>%
  mutate(Pct = 100 * N / sum(N))

Why Subject-Level Derivation Should Come First

A common programming mistake is to begin with an aggregate table and attempt to back-calculate the underlying patient status. This makes it difficult to identify overlaps, missing patients, or inconsistent definitions.

A safer architecture is:

Handling Patients Who Never Receive Treatment

Patients who are randomized but never receive study treatment require particular attention. They remain part of the randomized population, but they may not belong to the Safety Analysis Set if the SAP defines that set based on actual treatment exposure.

The disposition table should make the distinction visible rather than silently dropping these patients.

Handling Patients Who Receive Only One Dose

If the Safety Analysis Set is defined as patients who received at least one dose, then a patient who receives a single dose may belong to the safety population even if the patient subsequently discontinues immediately.

This is one reason why treatment exposure and treatment completion should be reported separately.

Handling Patients With Multiple Disposition Events

A patient can have multiple disposition events during a study. For example, a patient may discontinue treatment because of an adverse event and later discontinue study participation because of withdrawal of consent.

The table must therefore distinguish the event being summarized. A treatment discontinuation table should not automatically use the final study-disposition reason.

Early Discontinuation

Early treatment discontinuation is not necessarily a data-quality problem. It is an important part of the clinical-trial experience and should be represented accurately.

Common reasons include:

• Adverse events
• Progressive disease
• Withdrawal of consent
• Protocol deviation
• Investigator decision
• Lack of efficacy
• Patient decision
• Other administrative or study-related reasons

Disposition and Exposure

Disposition tells the reader whether patients continued or discontinued treatment. Exposure summaries answer a different question: how much treatment did the patients actually receive?

These sections should reconcile logically without being treated as interchangeable summaries.

CONSORT-Style Reporting and Missing Patients

Missing follow-up information should not automatically be interpreted as study completion. If a patient has no final assessment, the disposition status should be derived from the actual disposition information rather than inferred from missing data.

Disposition Tables in the CSR

A CSR disposition table should generally provide enough information for a reviewer to understand:

• How many patients entered each treatment arm;
• How many received treatment;
• How many completed or discontinued treatment;
• Why patients discontinued treatment;
• How many completed or discontinued the study;
• How the major analysis populations relate to the randomized and treated sets.

The exact table numbering and placement will depend on the study's CSR shell.

Suggested CSR Table Footnote

The wording should be adapted to the actual SAP and table shell.

Common Disposition Reporting Errors

1. Using the wrong denominator. A percentage can look reasonable while being calculated against the wrong population.
2. Mixing randomized and treated populations. A patient can be randomized without receiving treatment.
3. Confusing treatment and study discontinuation. Stopping treatment does not necessarily mean leaving the study.
4. Allowing reason categories to overlap unintentionally. If primary reasons are required, each patient should contribute once.
5. Dropping patients who do not receive treatment. These patients remain relevant to randomized disposition.
6. Inferring completion from missing data. Completion should be based on defined disposition criteria.
7. Manually entering aggregate counts. Counts should be generated from validated subject-level data whenever possible.
8. Failing to reconcile the CONSORT diagram and disposition table. The two displays should describe the same underlying patient flow.
9. Changing categories late in reporting. Disposition categories should follow the prespecified coding and reporting conventions.
10. Failing to document the analysis population. Readers need to know exactly who is represented by each section.

Disposition Table Quality Control

Disposition tables should undergo both programming QC and clinical/statistical review.

Manual Validation Example

Suppose the source data contain the following Treatment A patients:

The resulting counts must therefore be:

• Randomized = 4
• Received treatment = 3
• Did not receive treatment = 1
• Completed treatment = 1
• Discontinued treatment = 2
• Adverse event discontinuations = 1
• Progressive disease discontinuations = 1

This small example illustrates the same reconciliation logic used for hundreds or thousands of patients.

Reconciling the CONSORT Diagram and the Table

The diagram and table should be treated as two views of the same underlying patient-level accounting.

Disposition and the Statistical Analysis Plan

The SAP should determine the definitions used for the major disposition and analysis-population summaries. The statistical programmer should not independently invent definitions during table production.

Before programming, confirm:

• The population denominator for each section;
• The treatment assignment used for each summary;
• The definition of treatment exposure;
• The definition of treatment completion;
• The definition of study completion;
• The hierarchy for discontinuation reasons;
• The rules for analysis-set membership;
• The rounding convention for percentages.

Disposition Tables and Statistical Programming Specifications

Disposition Tables and Auditability

A high-quality disposition table should be reproducible. If the same validated subject-level data and specifications are used again, the same counts should be generated.

This is especially important during late CSR updates. A small change in one patient's disposition status can affect multiple tables, figures, and analysis populations.

Waterfall + Disposition + CONSORT

Different clinical-trial displays answer different questions.

Together, these displays provide a much more complete picture of the clinical-trial population than any one table or figure can provide alone.

Practical Disposition Checklist

Common Reporting Language

Key Principles

The most important principles of disposition reporting are straightforward:

• Define the population before counting it.
• Keep randomized, treated, and analysis populations distinct.
• Make treatment and study discontinuation separate concepts.
• Use mutually exclusive reasons when the table requires reconciliation.
• Make denominators explicit.
• Trace every aggregate number to subject-level data.
• Reconcile the table with the CONSORT-style flow diagram.
• Follow the protocol, SAP, and approved table shell.

References

Schulz, K.F., Altman, D.G., Moher, D., for the CONSORT Group (2010). CONSORT 2010 Statement: updated guidelines for reporting parallel group randomised trials. BMJ, 340, c332.
Moher, D., Hopewell, S., Schulz, K.F., et al. (2010). CONSORT 2010 Explanation and Elaboration: updated guidelines for reporting parallel group randomised trials. BMJ, 340, c869.
International Council for Harmonisation (ICH) (1996). ICH E3: Structure and Content of Clinical Study Reports.
International Council for Harmonisation (ICH). ICH E9: Statistical Principles for Clinical Trials.