This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical results on this page are restricted to the ClinicalTrials.gov record.
1. Trial at a Glance
REMDACTA was a completed, randomized, double-blind, parallel-group phase 3 trial evaluating remdesivir plus tocilizumab versus remdesivir plus placebo in hospitalized participants with severe COVID-19 pneumonia. The primary endpoint was a time-to-event outcome measuring time to hospital discharge or "Ready for Discharge" up to Day 28.
| Feature | REMDACTA |
|---|---|
| Trial name | REMDACTA |
| Brief title | A Study to Evaluate the Efficacy and Safety of Remdesivir Plus Tocilizumab Compared With Remdesivir Plus Placebo in Hospitalized Participants With Severe COVID-19 Pneumonia |
| Phase | Phase 3 |
| Status | Completed |
| Condition | COVID-19 Pneumonia |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Enrollment | 649 |
| Lead sponsor | Hoffmann-La Roche |
| Start | 2020-06-16 |
| Primary completion | 2021-02-01 |
| ClinicalTrials.gov | NCT04409262 |
2. Clinical Question
The central statistical question was whether adding tocilizumab to remdesivir changed the time to hospital discharge or "Ready for Discharge" compared with remdesivir plus placebo in hospitalized participants with severe COVID-19 pneumonia.
Population
Hospitalized participants with severe COVID-19 pneumonia.
Intervention
Remdesivir plus tocilizumab.
Comparator
Remdesivir plus placebo.
Primary question
Does remdesivir plus tocilizumab improve time to hospital discharge or "Ready for Discharge" up to Day 28 compared with remdesivir plus placebo?
3. Trial Design
Remdesivir + Placebo
- Remdesivir
- Placebo
- Comparator arm for the registered statistical analyses
- Serious adverse events: 76/213 affected participants
Remdesivir + Tocilizumab
- Remdesivir
- Tocilizumab
- Intervention arm for the registered statistical analyses
- Serious adverse events: 141/429 affected participants
4. Trial Timeline
Trial start
The REMDACTA trial began on June 16, 2020.
Primary completion
The registry lists February 1, 2021 as the primary completion date.
Results posted
ClinicalTrials.gov results include 19 posted outcome measures and 21 statistical analyses in the ClinicalTrials.gov record.
5. Primary Endpoint
| Endpoint | Registry definition and time frame | Type |
|---|---|---|
| Time to Hospital Discharge or "Ready for Discharge" up to Day 28 | Defined as days from randomization to hospital discharge or "Ready for Discharge" not followed by ordinal scale category >1, hospital readmission or death. Hospital discharge or "Ready for Discharge" is defined as an ordinal score of 1 on the 7-point ordinal scale. Participants who die are censored at Day 28. | Time-to-event; up to Day 28 |
This endpoint is not simply a proportion discharged by Day 28. It uses the timing of the qualifying discharge or "Ready for Discharge" event, with the registry's specified handling of subsequent clinical status, readmission, death, and Day 28 censoring.
6. Statistical Methodology
Log-rank test for the primary time-to-event endpoint
The primary endpoint was analyzed using a log-rank test. This is a standard method for comparing time-to-event distributions between randomized groups while accounting for the fact that participants can have different follow-up times.
The registry identifies the hypothesis type as superiority and the confidence interval as two-sided.
Hazard ratio
The registry reports the treatment effect for the primary endpoint as a hazard ratio. A hazard ratio compares the estimated instantaneous event rates between the two groups within the time-to-event framework.
The HR is a relative measure of event hazard. It is not a difference in the number of days to discharge, and it does not directly state the percentage of participants discharged by a particular day.
Logistic regression for clinical-status endpoints
The Day 7, Day 14, Day 21, Day 28, and Day 60 clinical-status endpoints were analyzed using logistic regression and reported as odds ratios. The registry identifies these outcomes as binary and reports the outcome unit as percentage of participants.
Cochran-Mantel-Haenszel testing for categorical outcomes
Several binary outcomes were analyzed using the Cochran-Mantel-Haenszel test, with the treatment effect reported as a weighted percentage difference. These included initiation of mechanical ventilation, being alive and free of respiratory failure among participants requiring mechanical ventilation at baseline, mortality, discharge or "Ready for Discharge," and initiation of mechanical ventilation or death.
Linear regression for duration of mechanical ventilation
Duration of mechanical ventilation among participants requiring mechanical ventilation at baseline was analyzed using linear regression. The reported effect measure was the mean difference in final values.
| Method | Trial application | Effect measure |
|---|---|---|
| Log-rank test | Time-to-event endpoints | Hazard ratio |
| Logistic regression | Binary clinical-status endpoints | Odds ratio |
| Cochran-Mantel-Haenszel test | Binary categorical outcomes | Weighted percentage difference |
| Linear regression | Duration of mechanical ventilation | Mean difference |
7. Primary Result: Time to Hospital Discharge or "Ready for Discharge"
The registry reports a formal statistical analysis of the primary endpoint in the modified intention-to-treat (mITT) population. The analysis used a log-rank test, with the treatment effect reported as a hazard ratio.
Primary endpoint hazard ratio
95% CI: 0.78–1.19 · P = 0.7414
Comparison: RDV+Placebo vs RDV+Tocilizumab · Superiority hypothesis
| Primary endpoint | Analysis population | Method | Effect | 95% CI | P-value |
|---|---|---|---|---|---|
| Time to Hospital Discharge or "Ready for Discharge" up to Day 28 | mITT | Log-rank test | HR 0.965 | 0.78–1.19 | 0.7414 |
The reported HR of 0.965 means that, under the time-to-event comparison as reported, the estimated instantaneous rate of the qualifying discharge event was about 96.5% as large in the RDV+Placebo group as in the RDV+Tocilizumab group. Equivalently, the point estimate corresponds to an estimated 3.5% lower event hazard in the first-listed group.
This does not mean that participants receiving remdesivir plus placebo spent 3.5% fewer days in hospital, nor does it mean that 3.5% more or fewer participants were discharged. A hazard ratio describes a relative event-rate comparison over the time-to-event analysis.
The 95% confidence interval of 0.78–1.19 indicates substantial uncertainty around the point estimate. It spans values below and above 1, so the data represented by this confidence interval are compatible with either a lower or higher estimated event hazard for the first-listed group relative to the second.
The P-value of 0.7414 is a measure used for testing the specified statistical hypothesis; it is not a measure of the size or clinical importance of the treatment effect. A P-value does not tell us that the treatment effect is "74.14% likely" to be zero or that one treatment has a particular probability of being better.
Because this is a time-to-event endpoint, interpretation also depends on the event definition, censoring rules, and assumptions underlying the survival-analysis framework. The registry specifically states that participants who die are censored at Day 28 for this endpoint. The ClinicalTrials.gov record does not provide a separate assessment of the proportional-hazards assumption.
8. Secondary Results: Time-to-Event Endpoints
The registry reports several additional time-to-event outcomes. Each comparison below uses the same treatment-group ordering: RDV+Placebo versus RDV+TCZ. All confidence intervals are two-sided 95% intervals.
| Endpoint | Time frame | Method | Effect | 95% CI | P-value |
|---|---|---|---|---|---|
| Time to Mechanical Ventilation or Death | Up to Day 28 | Log-rank | HR 0.980 | 0.72–1.34 | 0.8993 |
| Time to Death | Up to Day 28 | Log-rank | HR 0.948 | 0.65–1.39 | 0.7867 |
| Time to Death | Up to Day 60 | Log-rank | HR 0.882 | 0.63–1.23 | 0.4602 |
| Time to Improvement of at Least 2 Categories Relative to Baseline on a 7-category Ordinal Scale of Clinical Status | Up to Day 28 | Log-rank | HR 0.982 | 0.80–1.21 | 0.8664 |
| Time to Recovery | Up to Day 28 | Log-rank | HR 0.957 | 0.78–1.18 | 0.6778 |
These estimates should be read as separate secondary analyses rather than as alternative versions of the primary endpoint. For example, "Time to Death up to Day 28" and "Time to Death up to Day 60" use different follow-up horizons and therefore answer different questions about the timing of death.
Time to Mechanical Ventilation or Death
The registry reports an HR of 0.980 with a 95% CI of 0.72–1.34 and P = 0.8993. The confidence interval crosses 1, and the estimate is close to 1. The result therefore does not provide a precise indication of a difference in the event hazard between the two randomized groups.
Time to Death
For death through Day 28, the reported HR was 0.948 (95% CI 0.65–1.39; P = 0.7867). For death through Day 60, the reported HR was 0.882 (95% CI 0.63–1.23; P = 0.4602). The wider confidence intervals reflect uncertainty in these event-time comparisons; the ClinicalTrials.gov record does not provide event counts or Kaplan-Meier estimates from which additional absolute measures could be calculated.
Time to clinical improvement and recovery
The HR for time to improvement of at least 2 categories on the 7-category ordinal scale was 0.982 (95% CI 0.80–1.21; P = 0.8664). The HR for time to recovery was 0.957 (95% CI 0.78–1.18; P = 0.6778).
9. Secondary Results: Clinical Status by Day
Clinical status assessed by the investigator using the 7-category ordinal scale was analyzed as a binary endpoint at several fixed time points. Logistic regression was used, with odds ratios reported for RDV+Placebo versus RDV+TCZ.
| Clinical-status endpoint | Time frame | Effect | 95% CI | P-value |
|---|---|---|---|---|
| Clinical Status as Assessed by the Investigator Using a 7-category Ordinal Scale | Day 7 | OR 1.01 | 0.75–1.35 | 0.9569 |
| Clinical Status as Assessed by the Investigator Using a 7-category Ordinal Scale | Day 14 | OR 1.05 | 0.77–1.44 | 0.7648 |
| Clinical Status as Assessed by the Investigator Using a 7-category Ordinal Scale | Day 21 | OR 1.09 | 0.78–1.52 | 0.6331 |
| Clinical Status as Assessed by the Investigator Using a 7-category Ordinal Scale | Day 28 | OR 0.99 | 0.70–1.40 | 0.9622 |
| Clinical Status as Assessed by the Investigator Using a 7-category Ordinal Scale | Day 60 | OR 1.14 | 0.79–1.64 | 0.4850 |
How to interpret the odds ratios
An odds ratio of 1 corresponds to equal odds between the groups for the binary outcome being analyzed. Values above or below 1 indicate relative differences in odds according to the registry's treatment-group ordering. An odds ratio should not be interpreted as a risk ratio or as a percentage-point difference in the proportion of participants experiencing the outcome.
For example, the Day 14 OR of 1.05 is an estimate of the odds ratio for the binary clinical-status outcome, not a statement that 5% more participants had the favorable status. The 95% CI of 0.77–1.44 also shows that the point estimate has considerable statistical uncertainty.
10. Secondary Results: Mechanical Ventilation and Respiratory Failure
Initiation of Mechanical Ventilation
This endpoint was restricted to participants who did not require mechanical ventilation at baseline. The registry used the Cochran-Mantel-Haenszel test and reported weighted percentage differences.
| Endpoint | Time point | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Proportion requiring initiation of mechanical ventilation post-baseline among participants not requiring mechanical ventilation at baseline | Day 28 | Weighted % difference | -2.2 | -10.2 to 5.9 | 0.5915 |
| Proportion requiring initiation of mechanical ventilation post-baseline among participants not requiring mechanical ventilation at baseline | Day 60 | Weighted % difference | -2.5 | -10.5 to 5.6 | 0.5494 |
The negative estimates indicate that the weighted percentage difference, as reported in the registry's RDV+Placebo versus RDV+TCZ comparison, was negative at both time points. The confidence intervals extend from negative to positive values, so the data do not establish a precise directional difference from these estimates alone.
Alive and Free of Respiratory Failure
This analysis was restricted to participants requiring mechanical ventilation at baseline. The outcome was evaluated at Days 28 and 60 using the Cochran-Mantel-Haenszel test.
| Endpoint | Time point | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Proportion alive and free of respiratory failure among participants requiring mechanical ventilation at baseline | Day 28 | Weighted % difference | -14.1 | -37.4 to 9.2 | 0.2590 |
| Proportion alive and free of respiratory failure among participants requiring mechanical ventilation at baseline | Day 60 | Weighted % difference | -14.6 | -37.0 to 7.8 | 0.2290 |
These estimates are particularly important to interpret in the context of their restricted analysis populations. They do not describe the entire enrolled population; they apply specifically to participants who required mechanical ventilation at baseline.
11. Secondary Result: Duration of Mechanical Ventilation
Among participants requiring mechanical ventilation at baseline, duration of mechanical ventilation up to Day 28 was analyzed using linear regression. The effect measure was the mean difference in final values.
Mean difference in duration of mechanical ventilation
95% CI: -2.16 to 8.38 · P = 0.2434
Analysis population: participants requiring mechanical ventilation at baseline
The point estimate is a mean difference rather than a hazard ratio or odds ratio. Its interpretation therefore concerns the difference in the analyzed final-value measure between the two groups. The confidence interval ranges from a negative value to a positive value, indicating uncertainty about the direction and magnitude of the difference.
This result illustrates why the effect measure must be read together with the analysis method. A value of 3.1125 is not a 3.1125-fold increase and does not mean a 311.25% difference. It is the registry's reported mean difference in final values from the linear-regression analysis.
The 95% CI of -2.16 to 8.38 contains zero. The P-value of 0.2434 is a hypothesis-test result and should not be treated as a measure of the clinical importance of the observed mean difference.
12. Secondary Results: Mortality
Mortality was analyzed at Days 14, 28, and 60 using the Cochran-Mantel-Haenszel test. The registry reports weighted percentage differences.
| Mortality endpoint | Time point | Effect measure | Estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Difference in mortality | Day 14 | Weighted % difference | 0.6 | -4.4 to 5.6 | 0.8222 |
| Difference in mortality | Day 28 | Weighted % difference | -1.3 | -7.8 to 5.2 | 0.6944 |
| Difference in mortality | Day 60 | Weighted % difference | -3.0 | -10.1 to 4.0 | 0.3919 |
The mortality estimates vary by assessment day, which is expected because each time point represents a different binary outcome definition.
13. Secondary Results: Discharge and Composite Clinical Outcomes
| Endpoint | Time frame | Method | Effect | 95% CI | P-value |
|---|---|---|---|---|---|
| Proportion discharged or "Ready for Discharge" | Up to Day 28 | Cochran-Mantel-Haenszel | Weighted % difference -0.9 | -8.7 to 6.8 | 0.7692 |
| Proportion requiring initiation of mechanical ventilation post-baseline or dying | Up to Day 28 | Cochran-Mantel-Haenszel | Weighted % difference -0.3 | -7.8 to 7.2 | 0.9334 |
These fixed-horizon categorical endpoints complement the primary time-to-event analysis but should not be substituted for it. A time-to-event endpoint uses information about when the event occurs, whereas a fixed-horizon binary endpoint classifies participants according to whether the specified outcome has occurred by a particular assessment point.
14. Statistical Methods Explained
Why was a log-rank test used for the primary endpoint?
The primary endpoint measures time from randomization until a qualifying discharge or "Ready for Discharge" event, rather than simply recording whether a participant achieved the event. Time-to-event data can include participants who have not experienced the event by the end of observation or who are censored according to the protocol-defined rules. The log-rank test is designed to compare the event-time distributions of two groups under this framework.
What does the primary HR of 0.965 mean?
The HR of 0.965 is the estimated relative event hazard for RDV+Placebo compared with RDV+TCZ. A value below 1 means the point estimate places the event hazard lower in the first-listed group. It does not mean that participants were discharged 3.5% faster or that the average hospital stay was 3.5% shorter.
Why is the confidence interval more informative than the point estimate alone?
A point estimate such as 0.965 is only one estimate from the observed data. The 95% CI of 0.78–1.19 displays the statistical uncertainty around that estimate under the analysis framework. Because the interval spans 1, it includes both lower and higher relative event hazards for the first-listed group.
Why are odds ratios used for the clinical-status endpoints?
The registry treats the Day 7, Day 14, Day 21, Day 28, and Day 60 clinical-status outcomes as binary endpoints and specifies logistic regression. Logistic regression naturally expresses a treatment comparison as an odds ratio. The OR should not be read as a percentage-point difference or automatically as a risk ratio.
Why are some outcomes analyzed with the Cochran-Mantel-Haenszel test?
The Cochran-Mantel-Haenszel framework is commonly used for categorical comparisons when a weighted comparison across strata is desired. In REMDACTA, the registry reports this method for several binary outcomes and expresses the treatment effect as a weighted percentage difference.
Why does the analysis population change for some endpoints?
Some endpoints only make sense for participants who meet a particular baseline condition. For example, initiation of mechanical ventilation after baseline is analyzed among participants who did not require mechanical ventilation at baseline, while being alive and free of respiratory failure is analyzed among participants who did require mechanical ventilation at baseline. Restricting the analysis population in this way aligns the outcome with the clinical question being measured.
15. Confidence Intervals and P-values
Across the reported REMDACTA analyses, confidence intervals and P-values serve different statistical purposes. The confidence interval describes uncertainty around the estimated effect, while the P-value evaluates the data relative to the specified null hypothesis under the test's assumptions.
Confidence interval
The 95% confidence interval gives a range of effect estimates compatible with the statistical model and sampling framework used for the analysis. It should not be interpreted as saying that there is a 95% probability that the true effect lies inside this particular interval.
P-value
The P-value measures how unusual the observed data would be under the specified null hypothesis and test assumptions. It is not a measure of effect size, clinical importance, or the probability that the null hypothesis is true.
The distinction is especially clear in the primary result: an HR of 0.965 is close to 1, while the 95% CI of 0.78–1.19 is relatively broad. The P-value of 0.7414 should be considered alongside the estimated effect and its precision rather than used as a standalone description of the treatment difference.
16. Analysis Populations and Missing Data
The primary endpoint analysis is identified as being conducted in the mITT population. Several secondary endpoints have explicitly restricted analysis populations based on baseline mechanical-ventilation status.
| Endpoint category | Analysis population information reported |
|---|---|
| Primary endpoint | mITT population |
| Day 7 clinical status | Two RDV+TCZ participants with no post-baseline clinical-status data to Day 7 were excluded. |
| Day 14 clinical status | Two RDV+TCZ participants with no post-baseline clinical-status data to Day 14 were excluded. |
| Day 21 clinical status | Two RDV+TCZ participants with no post-baseline clinical-status data to Day 21 were excluded. |
| Day 28 clinical status | Two RDV+TCZ participants with no post-baseline clinical-status data to Day 28 were excluded. |
| Day 60 clinical status | Two RDV+TCZ participants with no post-baseline clinical-status data to Day 60 were excluded. |
| Initiation of mechanical ventilation | Only participants not requiring mechanical ventilation at baseline were included. |
| Alive and free of respiratory failure | Only participants requiring mechanical ventilation at baseline were included. |
| Duration of mechanical ventilation | Only participants requiring mechanical ventilation at baseline were included. |
17. Multiplicity and Statistical Inference
The ClinicalTrials.gov record identifies the hypothesis type for the reported analyses as superiority. They also report one registered primary endpoint and numerous secondary analyses. However, the ClinicalTrials.gov record does not provide an alpha-allocation strategy, multiplicity-adjustment procedure, hierarchical testing sequence, or interim-analysis plan.
This distinction is important because the trial has a large number of reported outcome analyses. Looking at many endpoints creates more opportunities for apparently notable results to arise by chance if each is interpreted independently. A statistically rigorous review therefore considers not only individual estimates and P-values, but also the prespecified role of each endpoint and the trial's multiplicity strategy when that information is available.
18. Blinding and Randomization
REMDACTA used randomized allocation and double masking in a parallel-group phase 3 design. Randomization is the foundation for a causal treatment comparison because, in expectation, it balances measured and unmeasured prognostic factors across treatment assignments.
Randomization
The allocation was randomized, creating the basis for comparing outcomes according to assigned treatment rather than according to participant characteristics selected after enrollment.
Double masking
Double masking is a design feature intended to reduce the potential influence of treatment knowledge on participant care, outcome assessment, and other trial processes.
The ClinicalTrials.gov record does not identify the specific individuals or groups who were masked, so this page does not expand the registry's "double" masking designation into a more detailed masking structure.
19. Safety Results
The ClinicalTrials.gov record reports serious adverse events by treatment arm. These figures are presented as affected participants divided by participants at risk.
| Arm | Serious adverse events | Reported affected / at risk |
|---|---|---|
| Remdesivir + Placebo | Serious adverse events | 76/213 |
| Remdesivir + Tocilizumab | Serious adverse events | 141/429 |
The denominators reported for the serious-adverse-event summary are 213 and 429, respectively. They should not be replaced with the overall enrollment figure of 649 or assumed to represent the number randomized to each arm, because the ClinicalTrials.gov record specifically identify their own affected/at-risk denominators.
A safety count such as 76/213 describes the number of participants affected among the stated at-risk population. It is a different type of information from a hazard ratio, odds ratio, or weighted percentage difference. Safety should therefore be analyzed according to the relevant exposure and risk-set definitions rather than combined mechanically with efficacy effect measures.
20. What the Primary Hazard Ratio Does — and Does Not — Mean
The primary HR of 0.965 is a relative time-to-event measure comparing RDV+Placebo with RDV+TCZ. Under the reported analysis, it represents the estimated instantaneous rate of the qualifying discharge event in the first-listed group relative to the second-listed group.
It does not mean that hospital stay was 3.5% shorter, that 3.5% more participants were discharged, or that an individual participant had a 3.5% change in their personal probability of discharge.
The 95% CI of 0.78–1.19 is important because it is substantially wider than the point estimate alone. The interval includes values below and above 1, so the observed data do not identify a precise direction or magnitude of the relative event-hazard difference.
The P-value of 0.7414 should not be described as the probability that the treatments are equivalent or as the probability that there is no effect. It is a test statistic probability calculated under the specified null hypothesis and assumptions of the statistical procedure.
21. Reading the Secondary Effect Measures
| Effect measure | Example from REMDACTA | What it describes | What it does not directly describe |
|---|---|---|---|
| Hazard ratio | Primary HR 0.965 | Relative event hazard over the time-to-event analysis | Absolute difference in days or probability of discharge |
| Odds ratio | Day 14 clinical status OR 1.05 | Relative odds for a binary outcome | Automatically the relative risk or percentage-point difference |
| Weighted percentage difference | Day 28 mortality difference -1.3 | Difference on the percentage scale under the reported weighted analysis | A hazard ratio or odds ratio |
| Mean difference | Mechanical ventilation duration difference 3.1125 | Difference in the analyzed mean final values | A relative percentage change unless separately defined |
Using the correct effect measure is essential. A common statistical error is to treat all estimates as though they were interchangeable "treatment effects." They are not. Each effect measure has a specific scale and interpretation determined by the outcome and statistical model.
22. Limitations
- Registry-level reporting: this analysis is limited to the ClinicalTrials.gov record. The ClinicalTrials.gov recordset does not contain a full statistical analysis plan, baseline table, subgroup forest plot, or detailed event-count tables.
- No reconstructed absolute time-to-event estimates: the registry supplies hazard ratios and confidence intervals but not the underlying Kaplan-Meier event data needed to reconstruct curves or additional absolute survival estimates.
- Primary endpoint definition: the registry definition includes specific discharge, "Ready for Discharge," ordinal-scale, readmission, death, and censoring rules. These rules make the endpoint more specific than a simple hospital-discharge proportion.
- Different analysis populations: some secondary outcomes are restricted to participants with or without mechanical ventilation at baseline. Results from those populations should not be generalized automatically to all enrolled participants.
- Missing clinical-status observations: two RDV+TCZ participants were excluded from each specified Day 7, Day 14, Day 21, Day 28, and Day 60 clinical-status analysis because they lacked post-baseline clinical-status data at the relevant assessment.
- Multiplicity information: the ClinicalTrials.gov record does not specify an adjustment strategy for the large number of secondary analyses, so individual secondary P-values should not automatically be interpreted as multiplicity-controlled confirmatory evidence.
- Proportional-hazards assumption: the ClinicalTrials.gov record does not report an assessment of proportional hazards. A hazard ratio is a model-based summary and may be less descriptive if the relative hazard changes substantially over time.
- Safety denominators: the serious-adverse-event summary uses arm-specific at-risk denominators of 213 and 429. These are not interchangeable with the overall enrollment count.
- Unreported methods: the ClinicalTrials.gov record does not specify a general missing-data imputation strategy, stratification factors, Bayesian methods, crossover, or interim-analysis procedure. Those features are therefore not inferred here.
23. Why This Trial Matters Statistically
REMDACTA is a useful teaching case because its registry results bring several core clinical-trial methods together in one study. The primary endpoint uses time-to-event analysis, while the secondary endpoints demonstrate how different outcome structures lead to different statistical models and effect measures.
| Concept | How it appears in REMDACTA |
|---|---|
| Randomization | Randomized parallel-group phase 3 design |
| Blinding | Double masking |
| Time-to-event analysis | Primary discharge endpoint and several secondary outcomes |
| Log-rank test | Primary and other time-to-event comparisons |
| Hazard ratio | Primary and secondary time-to-event effect measure |
| Confidence interval | Reported for every registry-reported formal treatment-effect estimate |
| Logistic regression | Day 7, Day 14, Day 21, Day 28, and Day 60 clinical-status analyses |
| Odds ratio | Effect measure for the logistic-regression clinical-status analyses |
| Cochran-Mantel-Haenszel test | Several categorical and fixed-horizon clinical outcomes |
| Weighted percentage difference | Mortality, ventilation, respiratory-failure, and discharge-related categorical outcomes |
| Linear regression | Duration of mechanical ventilation among participants ventilated at baseline |
| Restricted analysis populations | Ventilation outcomes separated by baseline mechanical-ventilation status |
| Missing endpoint data | Two RDV+TCZ participants excluded from each listed clinical-status analysis because of missing post-baseline data |
24. Statistical Methods Explained in Practice
Time-to-event endpoints require more than a simple event rate
Suppose two participants both eventually experience hospital discharge, but one is discharged much earlier than the other. A simple binary endpoint at a fixed day treats both according to whether the event occurred by that day, while a time-to-event endpoint uses the timing of the event. This is why the REMDACTA primary endpoint is analyzed using a survival-analysis method rather than a simple comparison of proportions.
The primary endpoint is not the same as "length of stay"
The registry defines the primary endpoint as time from randomization to hospital discharge or "Ready for Discharge" subject to specified clinical-status, readmission, death, and censoring rules. Therefore, it should not be casually relabeled as average length of hospitalization. The registered endpoint has a particular event definition and analysis framework.
Why can a confidence interval cross 1 for an HR?
For ratios such as hazard ratios and odds ratios, 1 represents equality between groups. A confidence interval that crosses 1 includes both directions of a relative difference. For example, the primary CI of 0.78–1.19 includes values below 1 and above 1.
Why can a percentage difference cross zero?
For an absolute or weighted percentage difference, 0 represents no difference. A confidence interval such as -7.8 to 5.2 therefore includes both a negative and positive difference. This is conceptually different from an HR or OR, where the no-difference value is 1.
Why should Day 28 and Day 60 results not be pooled casually?
Day 28 and Day 60 represent different follow-up horizons. A participant can have one status at Day 28 and a different status at Day 60. Combining the two as though they were repeated measurements of one identical endpoint would discard the distinction built into the registry definitions.
Why do baseline ventilation restrictions matter?
An endpoint such as initiation of mechanical ventilation after baseline cannot be meaningfully applied in the same way to someone already receiving mechanical ventilation at baseline. Conversely, an endpoint concerning being alive and free of respiratory failure among participants requiring ventilation at baseline is specifically designed for that starting population. Restricting the analysis population makes the statistical question match the clinical state being studied.
25. Related Tutorials
Learn more about the methods used in this trial:
26. Related Calculators
27. Sources
- ClinicalTrials.gov: REMDACTA, NCT04409262.
Continue through the Clinical Biostats statistical library
Connect this trial's design and analysis methods to deeper tutorials and statistical calculators for clinical research.
28. Record Summary
REMDACTA provides a useful example of how a phase 3 randomized clinical trial can require several statistical frameworks within the same study. Its primary endpoint is a time-to-event outcome analyzed with a log-rank test and reported as a hazard ratio, while secondary endpoints use logistic regression with odds ratios, Cochran-Mantel-Haenszel methods with weighted percentage differences, and linear regression with a mean difference.
The primary result was an HR of 0.965 with a 95% CI of 0.78–1.19 and P = 0.7414 for time to hospital discharge or "Ready for Discharge" up to Day 28. The statistical interpretation depends not only on that point estimate but also on the confidence interval, the treatment-group ordering, the endpoint definition, the mITT analysis population, and the censoring rule specified in the registry.
The secondary results reinforce a central principle of clinical-trial statistics: an endpoint cannot be interpreted independently of its scale and statistical model. Hazard ratios, odds ratios, weighted percentage differences, and mean differences answer different questions, and each requires its own interpretation of uncertainty.