This page separates reported trial information from statistical interpretation. The ClinicalTrials.gov record contains the registered design, endpoint definition, posted outcome-measure status, and serious adverse events by arm, but no formal statistical-analysis records with effect estimates, confidence intervals, or p-values.
1. Trial at a Glance
EMPACTA was a randomized, double-blind, parallel phase 3 trial evaluating the efficacy and safety of tocilizumab in hospitalized participants with COVID-19 pneumonia. The registered primary endpoint was the cumulative proportion of participants who died or required mechanical ventilation by Day 28.
| Feature | EMPACTA |
|---|---|
| Trial name | EMPACTA |
| NCT ID | NCT04372186 |
| Phase | Phase 3 |
| Therapeutic area | Infectious Disease |
| Condition | COVID-19 Pneumonia |
| Design | Randomized, double-blind, parallel |
| Primary purpose | Treatment |
| Enrollment | 377 |
| Interventions | Placebo; Tocilizumab |
| Results posted | Yes |
| Outcome measures posted | 7 |
| Lead sponsor | Genentech, Inc. |
| Sponsor type | Industry |
2. Clinical Question
The central question was whether tocilizumab could affect the occurrence of death or the requirement for mechanical ventilation by Day 28 in hospitalized participants with COVID-19 pneumonia compared with placebo.
Population
Hospitalized participants with COVID-19 pneumonia.
Intervention
Tocilizumab.
Comparator
Placebo.
Primary question
What was the cumulative proportion of participants who died or required mechanical ventilation by Day 28 in the randomized treatment comparison?
3. Trial Design
Participants were assigned to the trial arms using randomized allocation.
The registered design identifies the study as double-blind.
The two interventions were evaluated in parallel randomized groups.
The registry identifies treatment as the primary purpose of the study.
Tocilizumab
- Drug intervention: tocilizumab
- Randomized treatment arm
- Evaluated for efficacy and safety
Placebo
- Drug intervention: placebo
- Randomized comparator arm
- Evaluated for efficacy and safety
The ClinicalTrials.gov record does not provide the allocation ratio or the number randomized to each arm. Accordingly, the enrollment total of 377 should not be converted into arm-specific sample sizes on this page.
4. Trial Timeline
Trial start
The registered study start date was May 14, 2020.
Primary completion
The registered primary completion date was August 18, 2020.
Trial status
The registry status in the ClinicalTrials.gov record is COMPLETED.
5. Primary Endpoint
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Cumulative Proportion of Participants Who Died or Required Mechanical Ventilation by Day 28 | Cumulative proportion is measured as a percentage of participants meeting the endpoint. | Up to Day 28 | Binary |
This endpoint combines two clinically distinct events into a single binary outcome: death or requirement for mechanical ventilation by Day 28. A participant meeting either component satisfies the composite endpoint.
The registry defines the reported quantity as the cumulative proportion of participants meeting this binary endpoint.
6. Results
The ClinicalTrials.gov record indicates that results were posted and that 7 outcome measures were posted. However, the ClinicalTrials.gov recordset contains no statistical analyses for the registered primary endpoint. There are therefore no treatment-effect estimates, confidence intervals, or p-values in the ClinicalTrials.gov record to report.
What can be reported from the ClinicalTrials.gov record
| Measure | Available in the ClinicalTrials.gov record |
|---|---|
| Primary endpoint | Cumulative proportion of participants who died or required mechanical ventilation by Day 28 |
| Endpoint type | Binary |
| Results posted | Yes |
| Formal statistical analysis | None reported |
| Effect estimate | Not reported in the ClinicalTrials.gov record |
| Confidence interval | Not reported in the ClinicalTrials.gov record |
| P-value | Not reported in the ClinicalTrials.gov record |
7. Planned Analysis
For a binary primary endpoint such as the cumulative proportion of participants who died or required mechanical ventilation by Day 28, a treatment comparison would ordinarily begin by estimating the event proportion in each randomized group.
These are examples of standard effect measures for binary outcomes. The ClinicalTrials.gov record does not identify which of these measures, if any, was used for the EMPACTA primary analysis.
Why the endpoint can be analysed as binary
The endpoint asks whether a participant meets a prespecified condition by a fixed time point: Day 28. That structure differs from a time-to-event endpoint in which the exact time to death, progression, or another event is itself analysed.
A conventional analysis would compare the observed proportions between randomized treatment groups and quantify uncertainty around the estimated treatment contrast. Depending on the prespecified statistical analysis plan, possible approaches include a two-sample test for proportions, a chi-square procedure, or an exact method. The ClinicalTrials.gov record does not specify the formal test, adjustment strategy, or confidence-interval method used in EMPACTA.
8. Statistical Methodology
Randomization
Randomization is a central feature of EMPACTA's design. In a randomized parallel-group trial, assignment occurs before outcome evaluation and is intended to create treatment groups whose differences in outcomes can be interpreted in relation to the randomized intervention, subject to the trial's design and execution.
For a binary endpoint, randomization provides the foundation for comparing the proportion of participants experiencing the event under the two assigned strategies. It is important, however, to distinguish the design from the analysis: randomization does not itself specify whether the final comparison should be expressed as a risk difference, risk ratio, odds ratio, or another measure.
Double blinding
The registry identifies EMPACTA as double-blind. Blinding is particularly relevant when trial outcomes or clinical management could be affected by knowledge of treatment assignment. Even for an objectively defined endpoint, maintaining masking can reduce opportunities for differential treatment, assessment, or reporting behavior.
Randomization addresses allocation
Randomized assignment is intended to reduce systematic differences between treatment groups at the point of assignment.
Blinding addresses knowledge
Double blinding is intended to reduce the influence of treatment knowledge during trial conduct and outcome assessment.
Binary endpoint analysis
The primary endpoint is explicitly identified as binary in the ClinicalTrials.gov record. Binary analysis reduces each participant's Day 28 status to whether the registered composite event occurred. The resulting treatment-group proportions can then be compared using a prespecified statistical procedure.
Cumulative proportion
The registry defines the primary outcome as a cumulative proportion measured as a percentage of participants meeting the endpoint. This is an absolute measure at the specified Day 28 horizon. It differs conceptually from a hazard ratio, which summarizes a relative instantaneous event rate over time.
Confidence intervals
A confidence interval around a binary treatment-effect estimate quantifies sampling uncertainty under the statistical model or method used to construct the interval. It does not describe the probability that the true effect lies inside the interval in a literal repeated-trial sense.
P-values
A p-value, when reported for a prespecified hypothesis test, addresses compatibility of the observed data with the null hypothesis under the test's assumptions. It does not measure the size of the treatment effect, the clinical importance of the effect, or the probability that the treatment works.
9. How to Interpret a Binary Treatment Effect
If a risk difference were reported, it would describe the absolute difference in event proportions between the randomized groups. Unlike a relative measure, it directly expresses the difference in the percentage of participants meeting the endpoint.
A risk ratio would compare the event probability in the tocilizumab group with the event probability in the placebo group. A value below 1 would indicate a lower observed event probability in the tocilizumab group relative to placebo, while a value above 1 would indicate a higher observed event probability.
An odds ratio compares odds rather than probabilities. It should not automatically be described as a risk ratio, particularly when the outcome is not rare.
A p-value does not tell the reader how large the treatment difference is. The magnitude and uncertainty of the estimated effect require an effect measure and an appropriate confidence interval.
10. Statistical Methods Explained
Why was randomization important?
Randomization establishes the treatment comparison before outcomes are observed. Its purpose is not to guarantee identical groups in every characteristic, but to provide a principled basis for comparing outcomes between assigned interventions while reducing systematic allocation bias.
Why does double blinding matter?
Knowledge of treatment assignment can influence behavior, clinical management, assessment, or reporting. A double-blind design is intended to reduce these influences by limiting knowledge of assignment during the trial.
Why is the primary endpoint binary?
The registered endpoint asks whether a participant died or required mechanical ventilation by Day 28. Each participant can therefore be classified according to whether the composite outcome occurred within the specified time frame.
What does "cumulative proportion" mean?
It is the percentage of participants who meet the endpoint by the stated time horizon. It is an absolute event measure at Day 28 rather than a rate or hazard ratio.
How would a risk difference be interpreted?
A risk difference subtracts one treatment-group event proportion from the other. It directly expresses the absolute separation between the groups at the specified endpoint horizon.
How would a risk ratio be interpreted?
A risk ratio divides one group's event probability by the other's. A ratio of 1 would represent equal event probabilities; values below or above 1 indicate relative differences in the corresponding direction.
Why should the p-value not be used as an effect-size measure?
The p-value is tied to a hypothesis-testing framework and the amount of statistical information in the data. It does not quantify how clinically large the treatment difference is. Effect magnitude requires a measure such as a risk difference or risk ratio together with its uncertainty.
11. Safety Results
The ClinicalTrials.gov record reports serious adverse events by treatment arm. Because the denominators are provided as "affected/at risk," these values can be reported directly without reconstructing percentages.
| Safety measure | Tocilizumab | Placebo |
|---|---|---|
| Serious adverse events | 38/250 | 25/127 |
The reported serious-adverse-event figures describe affected participants relative to the corresponding at-risk populations reported in the ClinicalTrials.gov record. They should not be combined with the overall enrollment of 377 to create new arm-specific denominators.
Safety is a separate dimension of the randomized evidence. The presence of serious adverse events in both treatment groups does not, by itself, establish whether the intervention caused more, fewer, or the same number of events than placebo. That requires a prespecified comparative safety analysis using the appropriate denominators and definitions.
12. Analysis Population and Denominators
One of the most important statistical disciplines in a clinical-trial report is keeping denominators attached to the population from which they were calculated. EMPACTA's the ClinicalTrials.gov record provide an overall enrollment of 377, while the serious-adverse-event data provide separate "affected/at risk" counts for the two arms.
| Quantity | Reported value | Interpretation |
|---|---|---|
| Total enrollment | 377 | Overall enrolled trial population in the registry-reported profile |
| Tocilizumab serious AEs | 38/250 | 38 affected among 250 at risk |
| Placebo serious AEs | 25/127 | 25 affected among 127 at risk |
The ClinicalTrials.gov record does not define a separate efficacy analysis population, per-protocol population, modified intention-to-treat population, or as-treated population. Those labels should therefore not be assigned to EMPACTA based on inference alone.
13. Design Topics Supported by the Registry Data
Randomization
The allocation is explicitly identified as randomized. This supports an analysis framework based on comparison of outcomes according to assigned treatment.
Parallel groups
The design model is parallel, meaning the randomized treatment groups are evaluated concurrently rather than in a crossover sequence.
Double blinding
The masking field identifies the study as double-blind, making treatment-assignment knowledge an explicit design consideration.
Binary primary endpoint
The primary endpoint is a fixed-horizon binary outcome measured through Day 28.
Non-inferiority margin
The ClinicalTrials.gov record does not report a non-inferiority margin. The primary endpoint should therefore not be described using non-inferiority logic.
Crossover
The ClinicalTrials.gov record does not report a crossover design or crossover treatment. The registered design model is parallel.
Factorial design
The ClinicalTrials.gov record does not identify a factorial design. The registered design model is parallel with two interventions listed: placebo and tocilizumab.
Bayesian methods
The ClinicalTrials.gov record does not identify a Bayesian primary analysis. No Bayesian posterior probability, prior distribution, or Bayesian decision rule is reported in the provided statistical-analysis data.
Interim analysis
The ClinicalTrials.gov record does not identify an interim analysis, stopping boundary, or alpha-spending procedure. These methods should not be inferred from the fact that the trial was completed.
Missing data and imputation
The ClinicalTrials.gov record does not specify a missing-data strategy or imputation method for the primary endpoint. For a binary Day 28 endpoint, the handling of participants without an evaluable Day 28 status can materially affect the analysis population and must therefore be prespecified rather than assumed.
14. Limitations
- No formal statistical analyses reported: the statistical-analysis dataset contains no treatment-effect estimates, confidence intervals, or p-values for the primary endpoint.
- Posted results without analysis records: the registry profile states that results were posted and that 7 outcome measures were posted, but the ClinicalTrials.gov record does not provide formal comparative analyses.
- No arm-specific efficacy sample sizes reported: the overall enrollment is 377, but the ClinicalTrials.gov record does not provide the randomized enrollment in each efficacy arm.
- Composite endpoint: death and mechanical ventilation are clinically different events combined into one binary outcome. A composite can be statistically efficient, but interpretation requires understanding what contributes to the combined event.
- Fixed time horizon: the primary endpoint is defined through Day 28. It therefore answers a specific short-term outcome question rather than describing the entire subsequent clinical course.
- Unspecified statistical procedure: the ClinicalTrials.gov record does not identify the formal hypothesis test or effect measure used for the primary endpoint.
- Unspecified multiplicity strategy: the ClinicalTrials.gov record does not describe how the primary endpoint relates statistically to the other 6 posted outcome measures.
- Unspecified missing-data handling: the ClinicalTrials.gov record does not state how participants without a definitive Day 28 endpoint classification were handled.
- Safety denominators differ from total enrollment: serious adverse events are reported using arm-specific at-risk denominators of 250 and 127, so those figures should not be interpreted using 377 as the denominator.
15. Why This Trial Matters Statistically
EMPACTA is a useful teaching example because its design connects several fundamental clinical-trial concepts without requiring a time-to-event endpoint. The primary outcome is a binary composite endpoint at a fixed time point, evaluated in a randomized, double-blind, parallel phase 3 design.
| Concept | How it appears in EMPACTA |
|---|---|
| Randomization | Participants were assigned using randomized allocation. |
| Double blinding | The registry identifies the trial as double-blind. |
| Parallel design | The design model is registered as parallel. |
| Binary endpoint | The primary endpoint is classified as binary. |
| Composite endpoint | Death and mechanical ventilation are combined into the Day 28 endpoint. |
| Fixed time horizon | The registered time frame is up to Day 28. |
| Absolute event proportion | The registry defines the cumulative proportion as a percentage of participants meeting the endpoint. |
| Safety denominators | Serious adverse events are reported as affected/at risk separately for each arm. |
| Results interpretation | Outcome measures are posted, but no formal statistical analyses are reported in the ClinicalTrials.gov record. |
The trial therefore illustrates an important statistical principle: a clinical trial can have a clearly specified randomized design and endpoint while still requiring careful separation of what was registered, what was posted as an outcome, and what formal statistical evidence is actually available for interpretation.
16. Results Reporting: What Is Missing From the Statistical Record
The ClinicalTrials.gov record supports a precise description of the primary endpoint, but not a numerical treatment comparison. For this reason, an educationally responsible analysis should resist the temptation to reconstruct a result from outside sources.
| Question | Supported by the ClinicalTrials.gov record? |
|---|---|
| Was the trial randomized? | Yes |
| Was it double-blind? | Yes |
| Was it a parallel design? | Yes |
| Was the primary endpoint binary? | Yes |
| Was the primary endpoint posted? | Yes |
| Were 7 outcome measures posted? | Yes |
| Is a primary-endpoint effect estimate reported? | No |
| Is a confidence interval reported? | No |
| Is a p-value reported? | No |
| Is the formal primary-endpoint test specified? | No |
| Is an interim-analysis method reported? | No |
| Is a multiplicity strategy reported? | No |
| Is a missing-data strategy reported? | No |
17. How to Read the Day 28 Endpoint
The phrase "cumulative proportion of participants who died or required mechanical ventilation by Day 28" contains several important statistical details.
Cumulative
The measure counts participants who meet the endpoint during the specified follow-up period rather than considering only a single instantaneous time point.
Proportion
The outcome is expressed as the number meeting the endpoint relative to the relevant participant population.
Composite
Either death or mechanical ventilation is sufficient to meet the endpoint.
Day 28
The registered time frame is explicitly limited to follow-up up to Day 28.
This structure is different from a conventional survival endpoint. A time-to-event analysis retains information about when an event occurred and can account explicitly for censoring. The EMPACTA primary endpoint, as registered, instead summarizes whether the composite event occurred by the specified Day 28 horizon.
18. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The trial uses a randomized, double-blind, parallel phase 3 design with a binary Day 28 primary endpoint. The ClinicalTrials.gov record does not contain the formal comparative analysis needed to quantify the treatment effect.
Clinical interpretation
The registered clinical question concerns death or mechanical ventilation by Day 28 in hospitalized participants with COVID-19 pneumonia. The available data establish the question and design but do not supply a formal numerical treatment comparison in the ClinicalTrials.gov record.
19. Serious Adverse Events: Statistical Perspective
The safety data provide an example of why denominators and analysis populations matter. The registry-reported values are:
Serious adverse events
Tocilizumab · Placebo
The values are reported exactly as affected participants over participants at risk.
These counts permit a factual description of the reported serious-adverse-event burden in each registry-reported safety population. A comparative safety inference would require an appropriate statistical framework and additional information about the event definitions, follow-up, and prespecified safety analysis. Those details are not reported in the ClinicalTrials.gov recordset.
20. Data Discipline in Trial Analysis
EMPACTA also demonstrates why clinical-trial statistical pages should preserve the distinction between registered information and derived interpretation. Several tempting calculations would go beyond the ClinicalTrials.gov record.
- The total enrollment of 377 should not be split into treatment groups because the arm-specific efficacy sample sizes are not reported.
- The serious-adverse-event fractions 38/250 and 25/127 should not be converted into new denominators or combined with the enrollment total.
- No risk ratio, risk difference, odds ratio, confidence interval, or p-value should be constructed without the corresponding efficacy event counts and analysis specification.
- No superiority, non-inferiority, or equivalence conclusion should be assigned because the ClinicalTrials.gov record does not specify the formal hypothesis framework.
- No interim-analysis or multiplicity procedure should be inferred merely from the phase 3 designation.
This discipline is especially important when a registry indicates that results have been posted but the underlying statistical-analysis fields posted on ClinicalTrials.gov for an independent review are incomplete.
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Statistical Calculators
23. Sources
- ClinicalTrials.gov: NCT04372186 — EMPACTA.
- PubMed: PMID 33332779.
- PubMed: PMID 33201228.
Continue through Clinical Biostats
Build from trial design and endpoint interpretation into deeper statistical tutorials and practical calculation tools.
24. Record Summary
EMPACTA is a completed randomized, double-blind, parallel phase 3 trial with 377 enrolled participants evaluating tocilizumab versus placebo in hospitalized participants with COVID-19 pneumonia. Its registered primary endpoint is a binary cumulative proportion: participants who died or required mechanical ventilation by Day 28. The ClinicalTrials.gov record also report serious adverse events as 38/250 for tocilizumab and 25/127 for placebo.
The most important statistical limitation is the distinction between results being posted and a formal comparative statistical analysis being available in the ClinicalTrials.gov record. The registry profile states that results were posted and that 7 outcome measures were posted, but the ClinicalTrials.gov recordset contains no effect estimate, confidence interval, or p-value for the primary endpoint.