This page separates reported trial results from statistical interpretation. Numerical results on this page are restricted to the ClinicalTrials.gov record. The registry provides formal p-values for the posted analyses, but the ClinicalTrials.gov record does not provide treatment-effect estimates or confidence intervals for those comparisons.
1. Trial at a Glance
ACTIV-4a was a randomized, unmasked, phase 4 treatment trial in COVID-19. The registry describes a sequential design with 8 arms and 3591 participants enrolled. The posted primary endpoint was 21 Day Organ Support (Respiratory or Vasopressor) Free Days, measured through the first 21 days after study enrollment.
| Feature | ACTIV-4a |
|---|---|
| Trial name | ACTIV-4a |
| Brief title | Accelerating COVID-19 Therapeutic Interventions and Vaccines 4 ACUTE |
| Phase | Phase 4 |
| Status | Completed |
| Condition | Covid19 |
| Allocation | Randomized |
| Design model | Sequential |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 3591 |
| Arms | 8 |
| Primary endpoint | 21 Day Organ Support (Respiratory or Vasopressor) Free Days |
| Results posted | Yes |
| Outcome measures posted | 11 |
| Statistical analyses posted | 43 |
| Hypothesis type reported | Superiority |
| ClinicalTrials.gov | NCT04505774 |
2. Clinical Question
The statistical question was whether randomized comparisons involving therapeutic versus prophylactic anticoagulation, anticoagulation combined with a P2Y12 inhibitor, an SGLT2 inhibitor, or crizanlizumab were associated with differences in the number of days alive and free of specified organ support during the first 21 days after enrollment.
Population
Participants enrolled in ACTIV-4a with the registry-listed condition of Covid19.
Interventions
The registry lists therapeutic heparin, prophylactic heparin, P2Y12, Crizanlizumab Injection, and SGLT2 inhibitor interventions.
Comparator structure
The posted analyses include therapeutic-dose versus prophylactic-dose anticoagulation, corresponding anticoagulation comparisons with P2Y12 inhibition, and standard-of-care comparisons involving SGLT2 inhibitor or crizanlizumab.
Primary question
For each registered primary comparison, does the intervention produce a difference in 21 Day Organ Support (Respiratory or Vasopressor) Free Days?
3. Trial Design
The ClinicalTrials.gov record identifies the study as randomized, sequential, and unmasked. They do not provide a detailed allocation ratio, randomization strata, or a prespecified block-randomization scheme.
Therapeutic versus prophylactic dose
- Therapeutic Dose Anticoagulation
- Prophylactic Dose Anticoagulation
Anticoagulation plus P2Y12
- Therapeutic Dose Anticoagulation + P2Y12 Inhibitor
- Prophylactic Dose Anticoagulation + P2Y12 Inhibitor
Standard of care with SGLT2 inhibitor
- Standard of Care + SGLT2 Inhibitor
- Standard of Care (SGLT2)
Standard of care with crizanlizumab
- Standard of Care + Crizanlizumab
- Standard of Care (Criza)
4. Endpoints
| Endpoint | Time frame | Definition / registry description | Endpoint type |
|---|---|---|---|
| 21 Day Organ Support (Respiratory or Vasopressor) Free Days | 21 days from study enrollment | The number of days that a patient is alive and free of organ support through the first 21 days after trial entry. Organ support is defined as receipt of non-invasive mechanical ventilation, high flow nasal canula oxygen, mechanical ventilation, or vasopressor therapy, with death at any time during the index hospitalization assigned -1 days. | Continuous |
| Death Within 28 Days | 28 days from enrollment | Death within the specified 28-day period. | Binary |
| Acute Kidney Injury | 90 days from enrollment | Acute kidney injury endpoint as posted in the registry. | Binary |
| Major Thrombotic Event or in Hospital Death | 28 days | Composite binary endpoint. | Binary |
| Any Thrombotic Event or in Hospital Death | 28 days | Composite binary endpoint. | Binary |
| Any Renal Replacement Therapy | 28 days | Occurrence of renal replacement therapy. | Binary |
| Days Free of Organ Support and Renal Replacement Therapy | 28 days | Number of days free of the specified support and renal replacement therapy. | Continuous |
| Ventilator Free Days up to Day 28 | 28 days | Number of ventilator-free days through day 28. | Continuous |
| Days Free of Vasopressors | 28 days | Number of days free of vasopressors. | Continuous |
| Progression to Intubation or Death | 28 days | Occurrence of intubation or death. | Binary |
| Survival Until Discharge | 28 days | Time-to-event endpoint recorded in the registry analysis. | Time-to-event |
The primary endpoint is a composite-style continuous measure of organ-support-free survival over a fixed 21-day window. Its construction incorporates both survival and the absence of specified forms of respiratory or hemodynamic support.
5. Statistical Methodology
The registry reports two principal statistical methods for ACTIV-4a: t-tests for continuous outcomes and chi-squared tests for categorical outcomes. The ClinicalTrials.gov record identifies the primary analyses as two-sided t-tests under a superiority hypothesis.
| Outcome structure | Example in ACTIV-4a | Registry method |
|---|---|---|
| Continuous | 21 Day Organ Support (Respiratory or Vasopressor) Free Days | t-test, 2 sided |
| Continuous | Days Free of Organ Support and Renal Replacement Therapy | t-test, 2 sided |
| Continuous | Ventilator Free Days up to Day 28 | t-test, 2 sided |
| Continuous | Days Free of Vasopressors | t-test, 2 sided |
| Binary | Death Within 28 Days | Chi-squared |
| Binary | Major Thrombotic Event or in Hospital Death | Chi-squared |
| Binary | Progression to Intubation or Death | Chi-squared |
| Time-to-event | Survival Until Discharge | Chi-squared in the posted analysis |
Two-sided t-tests for continuous endpoints
For continuous outcomes, the registry identifies a two-sided t-test. Conceptually, the comparison asks whether the average outcome differs between the two randomized groups being compared.
Here, μ1 and μ2 represent the population means of the continuous endpoint in the two randomized groups.
Chi-squared tests for binary outcomes
Binary outcomes such as death within 28 days and progression to intubation or death were analyzed with chi-squared tests. The basic question is whether the observed proportions differ between the randomized groups.
The chi-squared test evaluates the compatibility of the observed treatment-group counts with the null hypothesis of equal event proportions.
Superiority hypotheses
The analyses posted on ClinicalTrials.gov identify the hypothesis type as superiority. That means the statistical objective was to test for a difference between the compared groups rather than to demonstrate that an intervention was no worse than a prespecified non-inferiority margin.
Missing data
Several secondary analyses explicitly state that numbers are different from participant flow due to missing data. The registry analysis text also identifies multiple imputation / missing data as an additional concept for the reported binary and continuous secondary analyses.
Endpoint-specific analysis populations
The registry provides an especially important limitation for the 90-day Acute Kidney Injury analyses: analysis was not completed on the Therapeutic Dose Anticoagulation or Prophylactic Dose Anticoagulation arms because the outcome was not collected in that protocol. The ClinicalTrials.gov record also state that the numbers differ from participant flow because of missing data.
6. Results: Primary Endpoint
ClinicalTrials.gov reports four formal statistical analyses for the primary endpoint, corresponding to four intervention comparisons. The endpoint is 21 Day Organ Support (Respiratory or Vasopressor) Free Days, measured 21 days from study enrollment.
| Comparison | Method | Hypothesis | P-value |
|---|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | t-test, 2 sided | Superiority | .6778 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | t-test, 2 sided | Superiority | .6292 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | t-test, 2 sided | Superiority | .6858 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | t-test, 2 sided | Superiority | .1351 |
The ClinicalTrials.gov record reports p-values but do not provide the corresponding mean differences, group means, standard errors, or confidence intervals. Consequently, this page does not reconstruct or infer an effect estimate from the p-values.
Therapeutic-dose anticoagulation vs prophylactic-dose anticoagulation
Primary endpoint comparison
Two-sided t-test for 21 Day Organ Support (Respiratory or Vasopressor) Free Days.
Hypothesis type: superiority.
The reported p-value of .6778 is the probability, under the specified null framework and statistical model, of obtaining a result at least as discrepant from the null as the observed test statistic, using the registry's two-sided t-test. It is not an estimate of the size of the treatment effect.
The registry analysis does not report the mean difference or a confidence interval for this comparison. Therefore, the magnitude and precision of the between-group difference cannot be determined from the ClinicalTrials.gov record alone.
A p-value also does not establish that the two treatment groups are clinically identical. A result can be statistically compatible with no difference while still being consistent with a range of effect sizes that would need to be quantified using the underlying group-level data and confidence interval.
Therapeutic-dose anticoagulation plus P2Y12 vs prophylactic-dose anticoagulation plus P2Y12
Primary endpoint comparison
Two-sided t-test for 21 Day Organ Support (Respiratory or Vasopressor) Free Days.
Hypothesis type: superiority.
The reported p-value of .6292 comes from the registry's two-sided t-test comparing 21-day organ-support-free days between the two anticoagulation strategies when used with P2Y12 inhibition.
The ClinicalTrials.gov record does not report the underlying group means, mean difference, standard error, or confidence interval. Thus, the p-value alone cannot establish how large the observed difference was or how precisely it was estimated.
Because this is a superiority analysis, the statistical question concerns evidence of a difference. It should not be reinterpreted as a formal non-inferiority or equivalence result.
Standard of care plus SGLT2 inhibitor vs standard of care
Primary endpoint comparison
Two-sided t-test for 21 Day Organ Support (Respiratory or Vasopressor) Free Days.
Hypothesis type: superiority.
The registry reports a p-value of .6858 for the two-sided t-test. This quantifies the statistical evidence against the null hypothesis within the reported testing framework; it does not quantify the magnitude of any treatment effect.
No confidence interval or treatment-effect estimate is included in the ClinicalTrials.gov record. The precision of the comparison therefore cannot be evaluated from this p-value alone.
In particular, a p-value should not be converted into a percentage benefit or percentage harm. Effect size and uncertainty require the underlying outcome summaries or individual-participant data.
Standard of care plus crizanlizumab vs standard of care
Primary endpoint comparison
Two-sided t-test for 21 Day Organ Support (Respiratory or Vasopressor) Free Days.
Hypothesis type: superiority.
The reported p-value of .1351 comes from the registry's two-sided t-test. It is a measure of statistical evidence under the null hypothesis, not a measure of clinical importance or effect size.
The registry analysis does not provide a confidence interval or numerical difference between the two groups. Consequently, the precision and practical magnitude of the observed comparison cannot be assessed from the reported p-value alone.
As with the other primary comparisons, the interpretation should also recognize that multiple primary comparisons were posted. The ClinicalTrials.gov record does not provide a multiplicity-adjustment procedure for combining these four primary tests into a single familywise-error statement.
7. Secondary Endpoint Results
The registry contains numerous secondary statistical analyses. The following table reports the posted p-values exactly as reported in the registry. The statistical-analysis records generally do not provide corresponding effect estimates or confidence intervals.
Death Within 28 Days
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | Chi-squared | .1959 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | Chi-squared | 0.9496 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | Chi-squared | .9902 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | Chi-squared | .0814 |
The analysis population note for these comparisons states that the numbers are different from participant flow due to missing data, and the analysis text identifies multiple imputation / missing data as an analysis concept.
Acute Kidney Injury
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | Chi-squared | 1.0 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | Chi-squared | 1 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | Chi-squared | 1 |
Major Thrombotic Event or in Hospital Death
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | Chi-squared | .0577 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | Chi-squared | .5015 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | Chi-squared | .5892 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | Chi-squared | .1714 |
Any Thrombotic Event or in Hospital Death
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | Chi-squared | .0732 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | Chi-squared | .9018 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | Chi-squared | .5075 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | Chi-squared | .1450 |
Any Renal Replacement Therapy
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | Chi-squared | .8837 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | Chi-squared | 1 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | Chi-squared | 1 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | Chi-squared | .5343 |
Days Free of Organ Support and Renal Replacement Therapy
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | t-test, 2 sided | .6636 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | t-test, 2 sided | .5878 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | t-test, 2 sided | .7148 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | t-test, 2 sided | .1117 |
Ventilator Free Days up to Day 28
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | t-test, 2 sided | .6780 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | t-test, 2 sided | .6638 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | t-test, 2 sided | .7645 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | t-test, 2 sided | 0.0987 |
Days Free of Vasopressors
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | t-test, 2 sided | .4016 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | t-test, 2 sided | .5815 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | t-test, 2 sided | .9085 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | t-test, 2 sided | .0820 |
Progression to Intubation or Death
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | Chi-squared | .3135 |
| Therapeutic Dose Anticoagulation + P2Y12 Inhibitor vs Prophylactic Dose Anticoagulation + P2Y12 Inhibitor | Chi-squared | .9581 |
| Standard of Care + SGLT2 Inhibitor vs Standard of Care (SGLT2) | Chi-squared | .9661 |
| Standard of Care + Crizanlizumab vs Standard of Care (Criza) | Chi-squared | .1073 |
Survival Until Discharge
| Comparison | Method | P-value |
|---|---|---|
| Therapeutic Dose Anticoagulation vs Prophylactic Dose Anticoagulation | Chi-squared | .3575 |
Although the registry classifies Survival Until Discharge as a time-to-event endpoint, the ClinicalTrials.gov record reports a chi-squared method rather than a Kaplan-Meier/log-rank or Cox-model analysis. This page therefore reports the registry method as recorded rather than substituting a different analysis.
8. Safety Results
The ClinicalTrials.gov record provides serious adverse-event counts by arm. These figures are reported as affected participants over participants at risk where the registry supplies both numbers.
| Arm | Serious adverse events |
|---|---|
| Therapeutic Dose Anticoagulation | 46/569 |
| Prophylactic Dose Anticoagulation | 57/514 |
| Therapeutic Dose Anticoagulation + P2Y12 | 151/772 |
| Prophylactic Dose Anticoagulation + P2Y1 | 133/739 |
| Standard of Care + SGLT2 Inhibitor | 20/287 |
| Standard of Care (sglt2) | 33/288 |
| Standard of Care + Crizanlizumab | 2 |
These serious-adverse-event figures are descriptive safety information. They are not equivalent to the efficacy p-values reported for the trial endpoints, and the ClinicalTrials.gov record does not provide formal between-arm confidence intervals or hypothesis tests for these serious-adverse-event counts.
9. Statistical Methods Explained
Why was a t-test used for the primary endpoint?
The registry classifies 21 Day Organ Support (Respiratory or Vasopressor) Free Days as a continuous endpoint and reports a two-sided t-test for each primary comparison. A t-test provides a direct comparison of the average continuous outcome between two groups. Its use is therefore aligned with the endpoint's continuous representation in the registry.
The important qualification is that the primary endpoint is not an ordinary measurement such as blood pressure or body weight. It is a bounded, clinically constructed number of days that incorporates death and organ-support status. The statistical properties of such an endpoint deserve attention, particularly when distributions are skewed or concentrated at particular values.
What does a two-sided p-value tell us?
A two-sided p-value evaluates evidence against equality in either direction. For example, the primary comparison with p = .6778 was not designed to test only whether therapeutic-dose anticoagulation increased organ-support-free days; the reported two-sided framework considers departures in either direction.
The p-value does not tell us how large the observed difference was. It also does not provide the probability that the null hypothesis is true.
Why were chi-squared tests used for binary outcomes?
Death Within 28 Days, Acute Kidney Injury, Major Thrombotic Event or in Hospital Death, and Progression to Intubation or Death are binary outcomes. A chi-squared test compares the observed event and non-event counts across groups under a null hypothesis of equal proportions.
For a binary endpoint, however, the most informative treatment-effect presentation generally includes an absolute risk difference, a relative risk or odds ratio, and a confidence interval. Those effect estimates are not reported in the ClinicalTrials.gov record, so they are not inferred from the p-values.
How should multiple imputation be understood here?
The registry analysis text for many secondary endpoints explicitly identifies multiple imputation / missing data. This indicates that missing observations were an important part of the statistical analysis context.
Multiple imputation does not simply replace every missing value with one arbitrary number. In general, it creates multiple plausible completed datasets, analyzes each one, and combines the resulting estimates so that uncertainty associated with missingness can be incorporated into inference.
Why is randomization important for these comparisons?
Randomization assigns participants to treatment groups according to a prespecified allocation mechanism rather than allowing treatment choice to determine group membership. In principle, this balances measured and unmeasured baseline factors in expectation and provides the foundation for causal comparisons of the randomized strategies.
Randomization does not eliminate missing data, treatment deviations, or measurement limitations. Those issues still affect the analysis and interpretation of the resulting estimates.
Should Survival Until Discharge have a survival-analysis method?
The registry classifies Survival Until Discharge as a time-to-event endpoint but reports a chi-squared method for the registry-reported analysis. A conventional time-to-event analysis might instead use Kaplan-Meier estimation and a log-rank comparison when censoring and event timing are central to the estimand. Because the ClinicalTrials.gov record explicitly identify chi-squared as the reported method, this page preserves that registry method rather than replacing it with an inferred alternative.
10. Missing Data and Analysis Populations
Missingness is an explicit part of the statistical record for many secondary endpoints. The registry repeatedly states that numbers are different from participant flow due to missing data.
Participant flow is not the analysis denominator
A randomized trial can begin with a fixed enrollment total while a particular endpoint has fewer evaluable observations because of missing measurements or other endpoint-specific availability.
Multiple imputation is identified
The registry analysis text identifies multiple imputation / missing data for numerous secondary analyses.
Missingness can affect precision
When fewer observations contribute to an endpoint, the effective information available for estimation may be lower than the overall enrollment number suggests.
Endpoint-specific availability matters
Acute Kidney Injury was not collected for the therapeutic-dose and prophylactic-dose anticoagulation arms in this protocol, so those arms were not analyzed for that endpoint.
11. Understanding the Primary Endpoint
21 Day Organ Support (Respiratory or Vasopressor) Free Days is designed to capture more than one dimension of clinical outcome. A participant contributes days on which they are alive and free of the specified forms of organ support. Organ support includes non-invasive mechanical ventilation, high flow nasal canula oxygen, mechanical ventilation, or vasopressor therapy.
The exact registry definition makes mortality part of the endpoint construction rather than treating death as an unrelated secondary event.
This structure has an important statistical implication: the outcome distribution can be quite different from the approximately symmetric continuous measurements for which a textbook t-test is often introduced. Values are constrained by the 21-day observation window, and the special assignment of -1 for death creates a distinct lower-end category.
The correct interpretation of a reported t-test therefore depends not only on the formal test but also on the underlying distribution of the endpoint. The ClinicalTrials.gov record does not provide the group-level distribution, means, variances, or graphical summaries needed to assess those features directly.
12. P-values and Effect Sizes
ACTIV-4a is a useful illustration of why a p-value should never be treated as a substitute for an effect estimate.
| Quantity | What it answers | What the registry-reported ACTIV-4a data provide |
|---|---|---|
| P-value | How compatible the observed test result is with the null hypothesis under the specified testing framework. | Reported for the posted analyses. |
| Effect estimate | How large the between-group difference or association is. | Not provided for the statistical analyses reported. |
| Confidence interval | How precisely the effect is estimated under the specified model and sampling framework. | Not provided for the statistical analyses reported. |
| Clinical importance | Whether the size and direction of an effect matter in practice. | Cannot be determined from p-values alone. |
For example, the four primary p-values are .6778, .6292, .6858, and .1351. These values tell us about the reported hypothesis tests, but they do not tell us the corresponding differences in organ-support-free days.
For the same reason, it would be inappropriate to translate any of these p-values into a percentage increase or decrease in clinical outcome. That requires an effect estimate.
13. Multiple Comparisons
The registry reports four separate primary-endpoint analyses, corresponding to four intervention comparisons. Each is identified as a superiority hypothesis and uses a two-sided t-test.
| Primary comparison | Endpoint | Test | P-value |
|---|---|---|---|
| Therapeutic vs prophylactic anticoagulation | 21 Day Organ Support Free Days | Two-sided t-test | .6778 |
| Therapeutic vs prophylactic anticoagulation + P2Y12 | 21 Day Organ Support Free Days | Two-sided t-test | .6292 |
| Standard of care + SGLT2 vs standard of care | 21 Day Organ Support Free Days | Two-sided t-test | .6858 |
| Standard of care + crizanlizumab vs standard of care | 21 Day Organ Support Free Days | Two-sided t-test | .1351 |
This distinction is important. When several hypotheses are tested, the probability of obtaining at least one small p-value can exceed the nominal level associated with any one test if no adjustment is made. The appropriate interpretation depends on the prespecified trial testing strategy, which is not reported here.
14. Non-Inferiority, Crossover, Factorial Design, and Bayesian Methods
The ClinicalTrials.gov record identifies the primary hypotheses as superiority. No non-inferiority margin is provided, so non-inferiority logic is not applied to the primary endpoint.
| Design topic | What the ClinicalTrials.gov record shows |
|---|---|
| Non-inferiority margin | Not provided. Primary hypotheses are identified as superiority. |
| Crossover | Not reported in the ClinicalTrials.gov record. |
| Factorial design | Not identified in the ClinicalTrials.gov record. |
| Interim analysis | Not described in the ClinicalTrials.gov record. |
| Bayesian methods | Not identified in the ClinicalTrials.gov record. |
| Stratification | Not provided in the ClinicalTrials.gov record. |
These omissions are deliberate. A statistical analysis page should distinguish between a method that is documented in the available record and a method that might be plausible based on common clinical-trial practice.
15. What the Secondary Results Do and Do Not Establish
The secondary analyses cover several clinically different outcome domains: mortality, thrombotic events, renal outcomes, organ-support duration, ventilator-free days, vasopressor-free days, intubation or death, and survival until discharge.
Different endpoints answer different questions
Death within 28 days is a binary mortality endpoint, whereas ventilator-free days and vasopressor-free days quantify time without specified forms of support.
Composite endpoints require component awareness
Major thrombotic event or in-hospital death combines two clinically distinct events into one binary outcome.
Missing data affect comparisons
The registry notes endpoint-specific missing data for many secondary analyses, so enrollment should not automatically be interpreted as the denominator for every analysis.
P-values do not rank endpoints
A smaller p-value does not make one endpoint more clinically important than another and cannot be used to rank the interventions.
16. Limitations
- Effect estimates are not reported: the ClinicalTrials.gov record contains p-values but do not provide corresponding treatment-effect estimates or confidence intervals for the primary comparisons.
- Multiple primary comparisons: four primary statistical analyses are reported, but the ClinicalTrials.gov record does not specify how multiplicity was handled across those comparisons.
- Endpoint distribution: the primary endpoint is a bounded, constructed number of days that incorporates death and organ-support status. The ClinicalTrials.gov record does not provide enough information to assess its empirical distribution.
- Missing data: the registry states that numbers differ from participant flow because of missing data for many secondary endpoints.
- Acute Kidney Injury availability: the registry text states that the Therapeutic Dose Anticoagulation and Prophylactic Dose Anticoagulation arms were not analyzed for Acute Kidney Injury because the endpoint was not collected in that protocol.
- Time-to-event method: Survival Until Discharge is identified as a time-to-event endpoint, but the registry analysis uses a chi-squared method. The underlying analytic rationale is not provided.
- Safety denominator: the serious-adverse-event value for Standard of Care + Crizanlizumab is reported as 2 without a denominator.
- Limited design detail: the ClinicalTrials.gov record does not provide a randomization ratio, stratification factors, interim-analysis plan, Bayesian methods, crossover information, or a detailed prespecified multiplicity strategy.
- Registry-level reporting: this analysis is constrained to the ClinicalTrials.gov record and therefore does not reconstruct analyses from external publications.
17. Why This Trial Matters Statistically
ACTIV-4a is a useful teaching case because its registry record places several important statistical concepts in one randomized clinical-trial framework: continuous organ-support outcomes, binary clinical outcomes, missing-data methods, multiple treatment comparisons, and the distinction between p-values and effect estimates.
| Concept | How it appears in ACTIV-4a |
|---|---|
| Randomization | The study is registered as randomized. |
| Sequential design | The registry identifies a sequential design model with 8 arms. |
| Continuous endpoint | 21 Day Organ Support (Respiratory or Vasopressor) Free Days is analyzed using a two-sided t-test. |
| Binary endpoints | Death, thrombotic outcomes, renal replacement therapy, and progression to intubation or death are analyzed using chi-squared tests. |
| Missing data | The registry repeatedly notes that analysis numbers differ from participant flow because of missing data. |
| Multiple imputation | Multiple imputation / missing data is identified in the analysis text for many secondary outcomes. |
| Superiority testing | The statistical analyses posted on ClinicalTrials.gov identify superiority as the hypothesis type. |
| Multiple comparisons | Four primary analyses are reported for the same primary endpoint across different intervention comparisons. |
| Composite endpoint | The primary endpoint incorporates survival and freedom from specified organ support, with death assigned -1 days. |
| Time-to-event endpoint | Survival Until Discharge is classified as time-to-event, while the posted method is chi-squared. |
18. A Statistical Reading of the Primary Endpoint
The registry provides four formal two-sided t-test p-values for the primary endpoint. These are the statistical outputs that can be reported directly from the ClinicalTrials.gov record.
They do not provide the numerical treatment difference, confidence interval, clinical magnitude, or probability that either treatment strategy is superior. Those conclusions require information beyond the p-values.
A confidence interval around a mean difference would show the range of effect estimates compatible with the statistical model and data. Without that interval, the precision of the estimated treatment difference cannot be evaluated from the registry analysis.
The primary endpoint integrates survival and organ-support status over a fixed 21-day period. Its interpretation therefore differs from a conventional continuously measured physiologic variable.
19. Results in Context
The statistical record contains a broad set of analyses rather than a single comparison. The primary endpoint was tested separately for four intervention comparisons, while secondary endpoints examined mortality, thrombosis, renal outcomes, organ-support duration, ventilator-free days, vasopressor-free days, progression to intubation or death, and survival until discharge.
This breadth creates an important interpretive distinction: each p-value answers a specific statistical question about a specific endpoint and treatment comparison. It should not be treated as a universal measure of the trial's overall success or failure.
For educational interpretation, the strongest approach is therefore to preserve the hierarchy of the registered endpoint, identify the analysis method, report the exact statistical output, and then state what information remains unavailable. That approach prevents a p-value from being mistaken for an effect estimate.
20. Sources
- ClinicalTrials.gov: ACTIV-4a, NCT04505774.
- PubMed: PMID 42089155.
- PubMed: PMID 39250922.
- PubMed: PMID 37979717.
- PubMed: PMID 37489818.
- PubMed: PMID 37356038.
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Statistical Calculators
Continue through Clinical Biostats
Explore statistical tutorials and calculators covering the methods that recur across randomized clinical trials, including categorical tests, continuous-data comparisons, missing-data methods, and time-to-event analysis.
23. Record Summary
ACTIV-4a provides a useful statistical case study in how a large randomized trial can combine a constructed continuous primary endpoint with multiple binary and clinical-support outcomes. The registry reports four primary analyses of 21 Day Organ Support (Respiratory or Vasopressor) Free Days using two-sided t-tests, alongside numerous secondary analyses using t-tests and chi-squared tests.
The ClinicalTrials.gov record is strongest when interpreted at the level actually reported: the exact p-values, endpoint definitions, comparison groups, and analysis methods. The ClinicalTrials.gov record does not include the corresponding effect estimates or confidence intervals, so those quantities should not be inferred from the p-values. Similarly, the presence of four primary comparisons means that a complete multiplicity interpretation would require the prespecified error-control strategy, which is not included in the ClinicalTrials.gov record.
The trial is particularly instructive for understanding the distinction between statistical testing and effect estimation. A p-value addresses compatibility with a null hypothesis under a specified testing framework; an effect estimate describes the size and direction of a difference; and a confidence interval describes uncertainty around that estimate. Keeping those three concepts separate is essential when interpreting randomized clinical-trial results.