ACTION ON COVID-19: A Study to Test Whether BI 764198 Helps Lung Health of People Hospitalised With COVID-19: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for ACTION ON COVID-19: A Study to Test Whether BI 764198 Helps Lung Health of People Hospitalised With COVID-19 (NCT04604184), a randomized phase 2 study. The primary outcome is Percentage of Patients Alive and Free of Mechanical Ventilation.

2. Clinical Question

The clinical question combines survival with respiratory support status. A patient had to be alive and not receiving mechanical ventilation to meet the primary outcome definition. The result therefore concerns this combined status at the specified assessment time, rather than survival alone, the duration of ventilation, or all aspects of recovery.

The posted comparison is between BI 764198 and placebo in the trial population described by the registry. It does not establish how the result would apply to people outside those eligibility characteristics or to other settings. The registry record does not report a reason for the trial’s terminated status.

3. Trial Design

ACTION ON COVID-19 was a phase 2, randomised, double-blind, parallel-group study of BI 764198 and placebo in people hospitalised with COVID-19. The registry lists an enrollment of 133. Participants were assigned by chance to receive BI 764198 capsules or matching placebo capsules; the summary describes one capsule per day. The planned time in the study was about a month.

The registry identifies the primary endpoint as the percentage of patients alive and free of mechanical ventilation at Day 29. Secondary outcomes included being alive and discharged without oxygen use, a composite involving in-hospital mortality or intensive care unit admission or mechanical ventilation, time to clinical improvement or discharge-related criteria, ventilator-free days, and mortality at specified follow-up days. Numerical results are posted for some, but not all, of those secondary outcomes.

The study began on 2020-11-03, and the primary completion date is 2021-03-24. Its status is listed as terminated. The reason for termination, the original statistical design assumptions, and the planned sample-size justification are not reported in the registry record. The listed enrollment should therefore not be read as evidence of the intended or achieved power for the primary comparison.

The primary endpoint description refers to a Full Analysis Set of patients in the randomised set who received at least one dose of trial drug. It reports that this set comprised 129 patients (85.4%). It also states that one patient without a baseline creatinine value was excluded from the statistical analysis, with N=128 specified in that sentence. The outcome arms list 65 patients in the BI 764198 group and 64 in the placebo group. These registry descriptions do not give a complete account of how every analysis population and denominator was determined.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: All subjects were screened for eligibility prior to participation in the trial. Subjects attended a specialist site which ensured that they (the subjects) strictly met all inclusion and none of the exclusion criteria. Subjects were not to be allocated to a treatment group if any of the entry criteria were violated.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full Analysis Set (FAS): FAS included all patients (pts) in randomized set (RS) who received at least 1 dose of trial drug, comprised 129 pts (85.4%). Of all pts randomised, number (%) of pts not treated was similar in BI 764198 (2/67 pts \[3.0%\]) and placebo group (2/66 pts \[3.0%\]). The 4 pts excluded from FAS discontinued trial before receiving treatment for following reasons: pt's withdrawal (n=2), meeting an exclusion criterion (n=1), and clinical worsening (n=1).

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Full Analysis Set (FAS): FAS included all patients (pts) in randomized set (RS) who received at least 1 dose of trial drug, comprised 129 pts (85.4%). Of all pts randomised, number (%) of pts not treated was similar in BI 764198 (2/67 pts \[3.0%\]) and placebo group (2/66 pts \[3.0%\]). The 4 pts excluded from FAS discontinued trial before receiving treatment for following reasons: pt's withdrawal (n=2), meeting an exclusion criterion (n=1), and clinical worsening (n=1).

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Patients Alive and Free of Mechanical Ventilation, measured over At Day 29.

The primary endpoint is the percentage of patients alive and free of mechanical ventilation at Day 29. It is a binary outcome: each patient either met this combined status at the assessment time or did not. The endpoint brings together survival and a marker of severe respiratory support, so it reflects more than one aspect of a patient’s status at that time.

As a general statistical rationale, a combined endpoint can represent an outcome that matters clinically while remaining straightforward to report as a proportion. Its interpretation depends on the individual components and the assessment time: the result does not show which component accounted for a patient not meeting the endpoint, nor does it describe the course of illness before Day 29.

The registry reports 83.1% for the BI 764198 group and 87.5% for placebo. The unadjusted recalculation expresses the comparison as a risk difference of -0.044, with a 95% confidence interval from -0.166 to 0.078. A negative risk difference means the observed percentage meeting the endpoint was lower in the BI 764198 group in this comparison. The interval includes zero, so the estimate is not precise enough to distinguish the observed direction from no difference under the stated analysis.

The endpoint is a snapshot at Day 29. It should not be interpreted as a measure of all recovery outcomes, or as a direct summary of adverse events. Those questions require separate outcome definitions and analyses.

7. Results Posted in the Registry

• BI 764198 Treatment Group: 83.1% (54/65) versus 87.5% (56/64) on control; risk difference -4.4 percentage points (95% CI -16.6 to 7.8), risk ratio 0.95 (0.82–1.10); unadjusted two-proportion z-test p = 0.478.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Percentage of Patients Alive and Discharged Free of Oxygen Use

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • BI 764198 Treatment Group: 75.4% (49/65) versus 82.8% (53/64) on control; risk difference -7.4 percentage points (95% CI -21.4 to 6.5), risk ratio 0.91 (0.76–1.09); unadjusted two-proportion z-test p = 0.300.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between BI 764198 Treatment Group and the control. The interval gives the range of effects that remain compatible with the data.

Percentage of Patients With Occurrence of Any Component of Composite: In-hospital Mortality or Intensive Care Unit (ICU) Admission or Mechanical Ventilation

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • BI 764198 Treatment Group: 26.2% (17/65) versus 23.4% (15/64) on control; risk difference 2.7 percentage points (95% CI -12.2 to 17.6), risk ratio 1.12 (0.61–2.04); unadjusted two-proportion z-test p = 0.721.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between BI 764198 Treatment Group and the control. The interval gives the range of effects that remain compatible with the data.

Number of Ventilator Free Days

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • BI 764198 Treatment Group: mean 24.12 versus 25.08 on control; difference -0.96 (95% CI -3.84 to 1.92), standardized effect d = -0.12; z-test p = 0.513.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between BI 764198 Treatment Group and the control. The interval gives the range of effects that remain compatible with the data.

The secondary results cover oxygen-free discharge, a composite involving in-hospital mortality or intensive care unit admission or mechanical ventilation, and ventilator-free days. Their estimates do not show a consistent pattern of separation between groups, and the reported confidence intervals include no difference. They are related to the primary endpoint because they describe survival, respiratory support, or recovery, but they are not identical measures and use different outcome definitions.

The primary result concerns being alive and free of mechanical ventilation at Day 29. Oxygen-free discharge adds a discharge and oxygen-use status, while the composite captures whether any listed severe event occurred. Ventilator-free days summarizes a duration-related outcome. Agreement across distinct outcomes can sometimes add context, but these endpoints do not measure the same clinical state and should not be collapsed into a single conclusion.

The registry lists time to clinical improvement or discharge-related criteria and mortality at Day 15, 29, 60 and 90 among the secondary outcomes. Numerical results for those listed outcomes are not reported in the registry record. The posted record therefore does not permit a complete numerical comparison across every secondary outcome category.

Secondary results generally carry less evidential weight than the primary result when many outcomes are examined and no multiplicity strategy is reported. As a general statistical issue, repeated testing can make nominally notable results more likely to arise by chance across a collection of outcomes. The registry does not report a multiplicity adjustment or a hierarchy for interpreting these analyses. The secondary estimates are best read as supporting context, not as independent confirmation or a substitute for the primary endpoint.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Percentage of Patients Alive and Discharged Free of Oxygen Use
• Percentage of Patients With Occurrence of Any Component of Composite: In-hospital Mortality or Intensive Care Unit (ICU) Admission or Mechanical Ventilation
• Time to Clinical Improvement of at Least 2 Points (From Randomisation) on the World Health Organization Clinical Progression Scale, Discharge From the Hospital, or Considered Fit for Discharge
• Number of Ventilator Free Days
• Percentage of Mortality at Day 15, 29, 60 and 90

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is a percentage, so its basic comparison is between the proportions of patients meeting the endpoint in the two groups. The registry reports the percentages directly, while the recalculated comparison uses a risk difference: the proportion in the BI 764198 group minus the proportion in the placebo group. A negative risk difference indicates a lower observed proportion in the BI 764198 group; a positive value would indicate a higher one. This is a direct, absolute comparison of proportions, not a relative ratio.

The recalculated primary risk difference is -0.044. Its 95% confidence interval runs from -0.166 to 0.078. As a general explanation, a confidence interval describes the range of estimates compatible with the observed data under the statistical procedure and its assumptions. An interval that spans zero includes no difference as one compatible value, as well as values on either side. It does not show that every value in the interval is equally likely, and it is not a guarantee that the true effect lies within the interval.

The unadjusted p-value for this recalculation is 0.478. As a general statistical concept, a p-value describes how unusual data at least as incompatible with a specified null model would be if that model were correct, under the test assumptions. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The posted comparison is not statistically conclusive on its own; interpretation should also consider the estimate, its interval, design, and the set of outcomes examined.

What does an unadjusted comparison mean?

The recalculated risk difference is unadjusted: it compares the posted group summaries without incorporating covariate adjustment from a model. Such a comparison is transparent and directly tied to the observed proportions, but it may differ from a model-based estimate. The registry’s sponsor-reported primary analysis is described as logistic regression with a risk difference as the reported parameter. The covariates, model specification, and any other adjustment details are not reported in the registry record. The reported regression estimate should therefore be read as a separate sponsor analysis, not as a duplicate of the simple unadjusted calculation.

For the continuous secondary outcome, ventilator-free days, the group summaries are means and standard deviations. A mean difference compares the average number of days in the groups. The recalculated mean difference is -0.96, with a 95% confidence interval from -3.839 to 1.919 and an unadjusted p-value of 0.513. The sponsor-reported analysis is described as ANCOVA and reports a mean difference of -0.80, with a 95% confidence interval from -3.47 to 1.87 and p=0.553. The registry does not report the covariates or model details for that analysis. A mean difference is interpreted in the outcome’s units, which are days, while its interval indicates the precision of the estimate.

Why can over-dispersion matter?

As a general explanation, over-dispersion occurs when variation in the data is greater than a model assumes. If a statistical model understates that variation, its standard errors may be too small and its confidence intervals too narrow. The posted primary outcome is a binary percentage, and the registry does not report an over-dispersion assessment for it. Over-dispersion should not be assumed to explain the observed results. For the continuous ventilator-free-days endpoint, the registry reports standard deviations, but does not report an over-dispersion analysis.

What does the analysis population describe?

The primary description defines the Full Analysis Set as patients in the randomised set who received at least one dose of trial drug, and reports 129 patients (85.4%). It also says one patient without a baseline creatinine value was excluded from the statistical analysis and specifies N=128 in that sentence. The outcome groups list 65 and 64 patients. These statements refer to analysis populations and denominators, but the registry description is incomplete and does not fully reconcile them. Details of missing outcomes, imputation, and sensitivity analyses are not reported in the registry record.

For secondary binary outcomes, the same general ideas apply: the group percentages describe observed proportions, and risk differences describe absolute contrasts. Secondary comparisons are exploratory in the absence of a reported multiplicity strategy. The registry records 9 analyses, and the methods for controlling the chance of misleading findings across analyses are not reported. The numerical estimates and intervals should therefore be considered alongside the primary endpoint, rather than as isolated tests that independently settle the study question.

11. Confidence Intervals and Precision

The primary unadjusted risk difference is -0.044, with a 95% confidence interval from -0.166 to 0.078. This interval spans zero and includes estimates in either direction. Its width indicates that the data do not pin down the primary between-group difference closely. The confidence interval is compatible with a lower observed percentage in the BI 764198 group, little or no difference, or a higher percentage, subject to the analysis assumptions.

The registry’s sponsor-reported primary analysis has a 95% confidence interval from -16.08 to 5.30 for its reported risk difference of -5.39. It also spans zero. Because this result comes from logistic regression and the recalculation is unadjusted, the two intervals describe different analyses. Details that would explain any difference in modelling are not reported in the registry record.

As a general statistical explanation, precision depends on the information in the data and the variability of the outcome, as well as the analysis method. A confidence interval communicates uncertainty about an estimate; it does not determine whether an effect would matter clinically. The registry record does not report a clinically important difference threshold for the primary endpoint, so the results cannot be judged against a specified threshold of that kind.

The observed group percentages are based on the analysis denominators reported for the endpoint. The primary description also gives a Full Analysis Set count and a separate statement about a patient excluded for missing baseline creatinine. The relationship among these analysis counts is not fully explained. That lack of detail, together with the interval’s breadth, is relevant when judging how much certainty to place in the estimated contrast.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size planning figures are illustrations based on the observed effect, not the trial’s original design calculation. For binary and other non-time-to-event outcomes, the figures are participants per arm. The assumptions listed are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The planning effect is the stated fraction of the observed effect. The trial’s own original sample-size assumptions and justification are not reported in the registry record.

The 80% power column gives the planned number of participants per arm that the illustration associates with an 80% chance of detecting the specified effect under those assumptions, if that effect were the true one. The 90% power column gives the corresponding planning number for a 90% chance. These are conventional design probabilities, not probabilities that the treatment is effective. The displayed figures are not observed power.

Rows describing planning for 100%, 75%, or 50% of the observed effect use progressively smaller assumed effects. Smaller effects are harder to distinguish from random variation, so the illustration requires more participants per arm to reach the same planned power. This relationship is a general feature of sample-size planning, not a claim about the trial’s original intended enrollment.

The primary outcome’s planning illustrations use the observed primary effect as their starting point. They should not be read as a recommendation or as evidence that the trial should have enrolled a particular number. Because the observed effect is itself uncertain, planning based on it is only a reference point; a different assumed effect would produce different planning requirements. The figures do not account for design or analysis features that are not listed in the registry assumptions.

For time-to-event outcomes, the stated convention is total number of events, using the Schoenfeld method. That convention does not apply to the posted primary endpoint, which is a percentage outcome. The planning display is useful for understanding how assumptions about effect size and power relate to required sample size, but it cannot establish the study’s actual power or explain why the trial was terminated.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.875; treatment proportion p₁ = 0.831; average p̄ = 0.853.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5009 = 0.9818.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5000 = 0.4208.
4. n = (0.9818 + 0.4208)² / (-0.0442)² = 1,005.6, rounded up to 1,006 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.875                          # control proportion
p1 <- 0.830769230769231                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(4604184)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 1,006 (80% power), 1,346 (90% power)
 75% of observed effect: n per arm = 1,733 (80% power), 2,319 (90% power)
 50% of observed effect: n per arm = 3,772 (80% power), 5,049 (90% power)
power.prop.test: 1005.6 
Simulated power at n = 1,006 per arm: 0.799 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The sponsor-reported primary analysis is described as logistic regression, with the risk difference reported as -5.39. Its 95% confidence interval is -16.08 to 5.30, and the reported p-value is 0.323. The interval includes zero. The registry does not report which covariates, if any, were included, how the risk difference was derived from the model, or other details needed to reproduce the analysis.

The simple recalculation from the posted group summaries is unadjusted. It estimates a risk difference of -0.044, with a 95% confidence interval from -0.166 to 0.078 and p=0.478. The sponsor-reported and recalculated results are not identical because they are not the same analysis. Model specification, analysis population, and other analytic details can affect an estimate, but the registry record does not provide enough information to identify the reason for the difference in this trial.

Both analyses have confidence intervals that include zero, and neither provides a precise estimate of the direction of the primary contrast. The sponsor’s analysis is the result reported in the registry as the sponsor analysis; the recalculation is a simple comparison of the posted summaries. They should be kept distinct rather than combined into a single estimate or treated as interchangeable evidence.

15. Multiplicity, Missing Data and Analysis Populations

The registry record lists 9 analyses, but does not report a multiplicity strategy, such as an adjustment to account for testing across several outcomes or a prespecified order for interpreting them. As a general statistical explanation, examining multiple outcomes can increase the chance of obtaining an apparently notable result by chance. Without a reported strategy, unadjusted secondary p-values should not be read as if each were the only analysis planned.

Missing-data methods are not reported in the registry record. The primary endpoint description states that one patient without a baseline creatinine value was excluded from the statistical analysis, and gives N=128 in that sentence. It also defines the Full Analysis Set as randomised patients who received at least one dose of trial drug, reporting 129 patients (85.4%). The endpoint groups list 65 patients in the BI 764198 group and 64 in placebo. The record does not fully explain how these counts relate to one another, or report the handling of missing outcome values, imputation, or sensitivity analyses.

The listed population is therefore important context for interpreting the percentages and model results. A result based on an analysis subset can differ from one based on all randomised participants, particularly when the reason for missingness is related to outcome status. The registry record does not provide enough information to assess whether missingness could have influenced the estimates.

The original statistical analysis plan, including whether interim analyses were planned or performed, is not reported in the registry record. The sample-size illustration uses stated assumptions, but those are not the trial’s original design assumptions. The reason for termination is also not reported. These omissions limit assessment of whether the posted analyses followed the original plan and whether the available results represent all intended analyses.

16. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: acute kidney injury (7.8%), respiratory failure (7.8%), covid-19 pneumonia (6.2%), septic shock (4.7%), acute respiratory distress syndrome (4.7%).

The registry reports serious adverse events affecting 19 patients in the BI 764198 group and 17 in the placebo group. Other adverse events affected 10 and 12 patients, respectively. Deaths are listed for 11 patients in the BI 764198 group and 5 in the placebo group. These are counts reported for the groups; they do not by themselves show whether an event was caused by the study drug or how the event counts relate to one another.

The most frequently listed serious event terms include acute kidney injury and respiratory failure, each reported as 7.8%, followed by COVID-19 pneumonia at 6.2%, septic shock and acute respiratory distress syndrome at 4.7%, and acute respiratory failure at 4.6%. These percentages describe the reported terms, not a comparison adjusted for differences in exposure or baseline risk. The registry does not provide enough detail in these figures to assess causality or the timing and clinical course of every event.

Safety counts need to be interpreted alongside the trial’s size, follow-up, event definitions, and analysis population. The registry record does not report a detailed comparative safety analysis or explain the relationship between serious events, other events, and deaths. The figures describe what was recorded; they do not establish that the treatment caused or prevented any particular event.

17. What the Result Does, and Does Not, Mean

The primary result describes the proportion of patients alive and free of mechanical ventilation at Day 29. The reported percentages were 83.1% with BI 764198 and 87.5% with placebo. The unadjusted recalculated risk difference is -0.044, and its 95% confidence interval includes zero. The sponsor-reported analysis also has an interval that includes zero. Taken together, these results do not provide a precise estimate of a difference between the groups on this endpoint.

This does not prove that the groups had identical outcomes. A result with an interval spanning zero can be compatible with no difference and with effects in either direction. The interval’s width matters: a range of plausible effects remains, so the data do not sharply identify the true contrast. The registry does not report a threshold for a clinically important difference, which also limits assessment of what size of effect would matter.

The endpoint is specific to survival and mechanical ventilation status at Day 29. It does not by itself measure all symptoms, recovery, duration of hospitalisation, or long-term outcomes. Secondary outcomes offer additional perspectives, but numerical results are not posted for every listed secondary category, and no multiplicity strategy is reported.

The trial is listed as terminated, with no reason reported in the registry record. The record does not explain whether termination affected follow-up or the analysis plan. The presented comparisons therefore need to be interpreted with the stated analysis populations, missing-data uncertainty, and available methods in mind.

These statistics describe the trial record; they do not constitute medical advice and do not establish that BI 764198 works or does not work. In particular, failure to demonstrate a clear difference in this record is not proof of no effect. The findings are limited to the studied population, endpoint definitions, follow-up, and analyses that are reported.

18. Why This Trial Matters Statistically

This trial illustrates why an estimate and its precision should be read together. The primary percentages describe the observed group outcomes, while the risk difference gives an absolute contrast. The confidence interval shows that the contrast remains uncertain and includes values on either side of no difference. A single point estimate would conceal that uncertainty.

It also illustrates how analyses of the same endpoint can differ. The unadjusted recalculation compares the posted group summaries directly, while the sponsor-reported analysis uses logistic regression. Because the registry does not report the model details, readers cannot fully assess why the estimates differ or reproduce the sponsor’s analysis. Recognising the distinction between an unadjusted calculation and a model-based result is essential to fair interpretation.

The registry’s outcome set spans survival, respiratory support, discharge status, and ventilator-free days. These outcomes provide complementary descriptions, but multiple testing makes it important to know whether the analysis plan accounted for the number of comparisons. No multiplicity approach is reported, and numerical results are unavailable for some listed secondary outcomes.

Finally, analysis populations and missing-data handling matter. The registry describes a Full Analysis Set, a separate exclusion related to baseline creatinine, and outcome-specific group counts without fully reconciling them. The missing detail limits confidence in how completely the posted comparisons represent the intended analysis population. These are practical examples of why transparent reporting of the estimate, interval, analysis method, populations, and prespecified plan is central to biostatistical interpretation.

19. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

Several limitations are specific to the available record. The trial is listed as terminated, but the reason is not reported in the registry record. It is therefore not possible to determine from the record whether termination was related to recruitment, operational issues, safety, efficacy, or another consideration, and no such explanation should be inferred.

The primary population description is incomplete. It reports a Full Analysis Set of 129 patients (85.4%) and separately states that one patient without a baseline creatinine value was excluded from statistical analysis, giving N=128 in that sentence. Outcome denominators are listed by arm, but the record does not reconcile all analysis counts or explain missing outcome handling. Missing-data sensitivity analyses are not reported.

The sponsor-reported regression analysis lacks model details, including covariates and other specifications. The registry does not report the original sample-size justification, the trial’s own design assumptions, or whether interim analyses were planned or conducted. The planning illustration is not a substitute for those design details.

There is no reported multiplicity strategy, and numerical results are not reported for every secondary outcome category listed. Safety counts are posted, but the available summary does not establish event causality or provide a complete comparative safety analysis. These reporting gaps constrain assessment of the robustness, completeness, and generalisability of the results.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: ACTION ON COVID-19: A Study to Test Whether BI 764198 Helps Lung Health of People Hospitalised With COVID-19 — NCT04604184.
• Registry record last updated 2022-04-11; status last verified 2022-04.
• Results first posted 2022-04-11.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

23. Record Summary

ACTION ON COVID-19 (NCT04604184) was a phase 2 randomised, double-blind, parallel-group study comparing BI 764198 with placebo in adults hospitalised with COVID-19. The registry lists enrollment of 133, a start date of 2020-11-03, primary completion on 2021-03-24, and status as terminated.

The primary endpoint was the percentage alive and free of mechanical ventilation at Day 29: 83.1% in the BI 764198 group and 87.5% in the placebo group. The unadjusted recalculated risk difference was -0.044, with a 95% confidence interval from -0.166 to 0.078. The sponsor-reported regression analysis also had a confidence interval that included zero.

Secondary and safety results add context, but the record does not report all listed secondary outcomes, a multiplicity strategy, full missing-data methods, or the reason for termination. The available results should be interpreted as estimates with uncertainty, not as definitive evidence for or against an effect.