Phase III Study of AZD7442 for Treatment of COVID-19 in Outpatient Adults: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Phase III Study of AZD7442 for Treatment of COVID-19 in Outpatient Adults (NCT04723394), a randomized phase 3 study. The primary outcome is A Composite of Either Severe COVID-19 or Death From Any Cause Through Day 29.

2. Clinical Question

3. Trial Design

NCT04723394 is a completed Phase III randomized study of AZD7442 for treatment of COVID-19 in non-hospitalized adults. Its official title describes a double-blind, placebo-controlled, multicenter study; the registry fields also list triple masking and a parallel design. Participants were assigned to AZD7442 or saline placebo. The arm descriptions state a 1:1 allocation and describe the planned arms as up to approximately 850 participants each, within a planned total of up to approximately 1700. The registry enrollment field records 910.

The study start date is 2021-01-28, and primary completion is recorded as 2021-08-21. The primary endpoint is a count of participants experiencing either severe COVID-19 or death from any cause through Day 29. The primary timeframe is described as Baseline (Day 1) and Day 29. The registry reports the endpoint analysis in the modified Full Analysis Set, defined as randomized participants who received study treatment within 7 days of symptom onset and were not hospitalized at baseline for isolation purposes. It reports population sizes of 413 for AZD7442 and 421 for placebo. The analyzed endpoint counts are recorded as 410 and 419, respectively; the reason for the difference is not reported in the registry record.

The registry also lists a secondary composite of death from any cause or hospitalization for COVID-19 complications or sequelae through Day 169. The record does not report the original sample-size justification or the assumptions used for the trial's design. The planning illustrations accompanying the posted results are based on the observed effect and are not a substitute for those original design assumptions.

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: In TACKLE, at the first visit, ie, the enrollment visit 1, participants were evaluated regarding the protocol mandated inclusion and exclusion criteria. After enrolment, eligible participants were randomized at a 1: 1 ratio to receive a single dose of study provided AZD7442 (AZD8895 + AZD1061) or placebo.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Modified Full Analysis Set (mFAS) - All randomized participants who received IMP ≤ 7 days from symptom onset and were not hospitalized at baseline (≤ Day 1) for isolation purpose. For AZD7442, N = 413; Placebo, N = 421. For the analysis of the primary endpoint, only participants within this population who were assessed for the primary endpoint were included in the analysis.

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 - All participants randomized and dosed with IMP

6. Primary Endpoint

The registry lists the primary outcome as A Composite of Either Severe COVID-19 or Death From Any Cause Through Day 29, measured over Baseline (Day 1) and Day 29.

The primary endpoint counts participants with either severe COVID-19 or death from any cause through Day 29. A composite endpoint combines specified events into one outcome: a participant is counted as having an event if either component occurs. The registry's severe COVID-19 definition requires at least one of pneumonia or hypoxemia features, together with a WHO Clinical Progression Scale score of 5 or higher. The listed pneumonia features include fever, cough, tachypnea, or dyspnea, and lung infiltrates. The hypoxemia definition includes SpO2 below 90% in room air and/or severe respiratory distress.

Using a composite can capture more than one clinically serious outcome within a stated follow-up period, but the combined result does not by itself show which component contributed most to the observed events. The posted primary result is for the composite; separate component-specific primary estimates are not reported in the registry record. Death is included regardless of cause, while the severe COVID-19 component uses the registry's clinical criteria.

The endpoint is assessed from Baseline (Day 1) through Day 29. Its interpretation therefore concerns event occurrence over that defined period and in the stated modified Full Analysis Set, rather than outcomes beyond that window or in every population of adults with COVID-19.

7. Results Posted in the Registry

• AZD7442: 4.4% (18/410) versus 8.8% (37/419) on control; risk difference -4.4 percentage points (95% CI -7.8 to -1.1), risk ratio 0.50 (0.29–0.86); unadjusted two-proportion z-test p = 0.010.

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.

A Composite of Death From Any Cause or Hospitalization for COVID-19 Complications or Sequelae Through Day 169

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • AZD7442: 5.0% (20/404) versus 9.7% (40/411) on control; risk difference -4.8 percentage points (95% CI -8.3 to -1.2), risk ratio 0.51 (0.30–0.85); unadjusted two-proportion z-test p = 0.009.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically lower for AZD7442 than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The registry lists one secondary endpoint: a composite of death from any cause or hospitalization for COVID-19 complications or sequelae through Day 169. The posted secondary analysis has the same general direction as the primary comparison: the reported event risk is lower in the AZD7442 group than in the placebo group, and the unadjusted interval for the risk difference excludes zero. This is a broad pattern consistent in direction with the primary result, but it concerns a distinct composite and a longer follow-up period.

The secondary endpoint should not be treated as a repeat measurement of the primary endpoint. Its components include hospitalization for COVID-19 complications or sequelae, and its stated timeframe extends through Day 169 rather than Day 29. A participant-level composite can also be influenced by its individual components in different ways. Component-specific results are not reported in the registry record, so the overall secondary comparison cannot establish which component accounts for the observed pattern.

The registry reports one secondary analysis and its Cochran-Mantel-Haenszel result. The posted unadjusted result is also an analysis of that endpoint, but it does not replace the sponsor-reported method. Unadjusted secondary findings generally carry less weight than a prespecified primary result when multiple outcomes are considered, because testing more outcomes can increase the chance of at least one apparently unusual result. Whether or how multiplicity was controlled for this trial is not reported in the registry record. The secondary result is therefore best read as supportive context, not as an independent confirmation with a documented multiplicity-adjusted interpretation.

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

• A Composite of Death From Any Cause or Hospitalization for COVID-19 Complications or Sequelae Through Day 169

9. Statistical Interpretation

10. Statistical Methods Explained

What does the event count represent?

The primary outcome is binary at the participant level: a participant either experienced the composite event through Day 29 or did not. The posted group values are counts of participants with an event, accompanied by the number analyzed in each arm. A count is useful for seeing how many events were observed, while an event risk expresses those events relative to the analyzed group size. The registry does not provide separate primary counts for the severe COVID-19 and death components.

What does the risk difference show?

The unadjusted risk difference is -0.044, with a 95% confidence interval from -0.078 to -0.011. It is calculated as the AZD7442 event risk minus the placebo event risk. On this scale, zero represents equal observed risks; a negative value means the observed risk was lower in the AZD7442 arm. The interval communicates uncertainty around that estimate. It is an unadjusted comparison derived from the posted summary statistics, rather than the sponsor's reported analysis.

How do the relative measures differ?

The posted unadjusted risk ratio is 0.497, with a 95% confidence interval from 0.288 to 0.859. A risk ratio divides one group's risk by the other group's risk, so a value of one represents equal risks. The odds ratio, 0.474 with a 95% confidence interval from 0.265 to 0.847, compares event odds rather than event risks. Risk differences are expressed on an absolute scale; risk ratios and odds ratios are relative scales. Those scales can tell different aspects of the same comparison, and the odds ratio should not be casually described as a risk ratio.

What is the sponsor-reported method?

The registry reports a Cochran-Mantel-Haenszel analysis with relative risk reduction as the parameter. Its estimate is 50.38, the 95% confidence interval is 14.38 to 71.25, and p = 0.010. In general, a Cochran-Mantel-Haenszel procedure combines information across strata. The strata, any stratification variables, and the detailed implementation are not reported in the registry record, so the reported method name alone does not establish which factors were used or how they entered the analysis. The sponsor result should be interpreted using its own relative measure and reported method, rather than substituted for the unadjusted risk-difference calculation.

What does a p-value say?

In general, a p-value describes how compatible the observed data are with a specified null model, assuming that model and the analysis assumptions hold. 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 sponsor reports p = 0.010 for the primary analysis; the unadjusted recalculation reports p = 0.01. These values belong to their respective analyses and should not be treated as proof of certainty or as a measure of treatment benefit.

Why can over-dispersion matter?

General explanation: over-dispersion means that observed variability is greater than a model expects. It matters when an analysis relies on a variance model, because understated variability can make intervals too narrow and p-values too small. The posted primary endpoint is a participant count for a binary composite, and the registry does not report an over-dispersion parameter or a model-based count analysis. Over-dispersion is therefore a general statistical concept, not a stated feature of this trial's primary analysis.

Overall, the posted unadjusted calculations are simple comparisons from summary event counts and analyzed group sizes. The registry's reported analysis is the Cochran-Mantel-Haenszel relative risk reduction analysis. They address the same primary comparison but use different statistical summaries, so interpretation should preserve that distinction.

11. Confidence Intervals and Precision

Confidence intervals describe the uncertainty around an estimated comparison under the analysis used. In general, a confidence interval gives a range of values compatible with the observed data and the method's assumptions; it does not mean that every value in the interval is equally likely, or that the true value has a stated probability of lying in that particular interval.

For the unadjusted primary risk difference, the estimate is -0.044 and the 95% confidence interval runs from -0.078 to -0.011. The interval is entirely below zero, the value corresponding to no absolute difference. Its width also shows that the data do not identify one exact risk difference: a range of negative estimates remains compatible with the calculation. The unadjusted risk-ratio interval is 0.288 to 0.859, and the unadjusted odds-ratio interval is 0.265 to 0.847. For those relative measures, one is the value corresponding to equal group risks or odds, respectively.

The sponsor's 95% confidence interval for relative risk reduction is 14.38 to 71.25. This uses a different measure and analysis from the unadjusted risk-difference interval, so the endpoints should not be compared as if they were on one scale. The intervals support a direction of difference in their respective analyses, while leaving uncertainty about its magnitude.

Precision depends in part on the number of observed events and the analysis method. The primary analysis records 18 events in the AZD7442 arm and 37 in the placebo arm. The registry does not report a separate precision analysis for the two components of the composite. Accordingly, the reported intervals quantify uncertainty for the combined endpoint, not for severe COVID-19 and death considered separately.

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 planning figures are participants per arm, not total events. They are illustrations computed from the observed effect, not the trial's original design justification. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each row uses the stated fraction of the observed effect as the effect to detect. The original trial design assumptions and sample-size justification are not reported in the registry record.

Read each row as a planning scenario: it specifies the fraction of the observed effect used for planning, then gives the number of participants per arm for 80% power and the number of participants per arm for 90% power. Power is the planned probability of detecting an effect of the specified size under the stated assumptions, if that effect is present. The 90% power column requires more participants per arm than the 80% column for each displayed scenario.

The rows also illustrate that smaller effects need more participants. Planning for 100% of the observed primary effect gives 491 participants per arm for 80% power and 656 for 90% power. Planning for 75% gives 941 and 1,259 participants per arm, while planning for 50% gives 2,268 and 3,036 participants per arm. The curve represents the same planning relationship across assumed effect sizes; it is not a record of enrollment or observed outcomes. These figures are planning illustrations, not observed power.

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 = 0.088; treatment proportion p₁ = 0.044; average p̄ = 0.066.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.3514 = 0.6887.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.3500 = 0.2945.
4. n = (0.6887 + 0.2945)² / (-0.0444)² = 490.3, rounded up to 491 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 <- 0.0883054892601432                          # control proportion
p1 <- 0.0439024390243902                          # 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(4723394)
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 = 491 (80% power), 656 (90% power)
 75% of observed effect: n per arm = 941 (80% power), 1,259 (90% power)
 50% of observed effect: n per arm = 2,268 (80% power), 3,036 (90% power)
power.prop.test: 490.3 
Simulated power at n = 491 per arm: 0.807 (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 uses the Cochran-Mantel-Haenszel method and reports relative risk reduction of 50.38, with a 95% confidence interval from 14.38 to 71.25 and p = 0.010. It is identified in the registry as a superiority analysis. This is the trial's reported analysis and should be distinguished from the unadjusted re-calculation based on the posted group counts.

The re-calculation expresses the comparison as a risk difference of -0.044, with a 95% confidence interval from -0.078 to -0.011, and also reports relative measures. Risk difference and relative risk reduction use different scales; the figures are not expected to be numerically interchangeable. The registry record does not give enough detail about the sponsor analysis strata or other implementation choices to reproduce that method fully from the posted summary statistics.

The discrepancy flag is false, so the record does not identify a discrepancy between the posted recalculation and sponsor analysis. That does not make their estimands or methods identical. For interpretation of the trial's formal reported result, the sponsor analysis remains the relevant reported analysis; the unadjusted calculations are supplementary descriptions derived from the displayed counts.

15. Multiplicity, Missing Data and Analysis Populations

The registry reports a primary superiority analysis and one secondary endpoint analysis. It gives the sponsor's Cochran-Mantel-Haenszel method, the relative risk reduction estimates, confidence intervals, and p-values. It does not report a multiplicity-control procedure, a testing hierarchy, or whether the primary and secondary hypotheses were considered together for confirmatory inference. The registry record also does not report whether any other outcomes were tested but not posted. Without those details, the posted p-values should be read as results for their individual analyses, not assumed to have been adjusted for the number of outcomes considered.

The primary endpoint population is described as the modified Full Analysis Set, including randomized participants who received study treatment within 7 days of symptom onset and were not hospitalized at baseline for isolation purposes. The record gives population sizes of 413 for AZD7442 and 421 for placebo. The primary endpoint analysis counts are recorded as 410 and 419. How those counts relate to the larger stated population sizes, and the reasons for any exclusions from the endpoint analysis, are not reported in the registry record.

The registry does not report the handling of missing endpoint data, the assumptions used for any imputation, or sensitivity analyses for missing outcomes. It also does not report a detailed statistical analysis plan, interim analyses, or the original sample-size justification. These omissions limit assessment of how robust the posted estimates are to alternative analysis choices. They do not show that data were missing or that a particular method was used; they indicate only that these details are not reported in the registry record.

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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: covid-19 pneumonia (8.2%), covid-19 (2.0%), arrhythmia (0.4%), abortion spontaneous (0.4%), cholecystitis chronic (0.4%).

For serious adverse events, the registry records 46 affected participants among 452 at risk in the AZD7442 group and 65 among 451 at risk in the placebo group. It separately reports 8 deaths in each group. These safety counts are not the primary efficacy endpoint counts and should be interpreted in their own reporting context. The registry does not provide an adjusted comparison or a causal attribution for these figures in the information.

The listed serious-event terms include COVID-19 pneumonia at 8.2% and COVID-19 at 2.0%. Arrhythmia, abortion spontaneous, chronic cholecystitis, and acute myocardial infarction are each listed at 0.4%. The registry also reports other adverse events affecting 230 participants among 452 at risk in the AZD7442 group and 224 among 451 at risk in the placebo group.

These figures describe reported events, not proof that treatment caused or prevented any particular event. The registry record does not provide further detail on event severity grading, attribution procedures, timing, or statistical comparisons for these safety summaries. Differences in raw counts should therefore not be interpreted as causal differences without that context.

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

The posted primary data show fewer observed Day 29 composite events in the AZD7442 group than in the placebo group, with an unadjusted risk difference below zero and a sponsor-reported relative risk reduction above zero. Both analyses report confidence intervals that exclude their respective null values. The record therefore contains statistical evidence of a difference for the specified primary comparison under the reported analyses.

That finding has a defined scope. It concerns non-hospitalized adults with COVID-19 in the stated modified Full Analysis Set, the composite of severe COVID-19 or death from any cause, and follow-up through Day 29. It does not establish separate effects for severe COVID-19 and death, because component-specific primary estimates are not reported. It does not establish the result for other populations, other timeframes, or outcomes beyond those recorded.

The point estimates are not exact measures of an effect that can be assumed to recur unchanged. Confidence intervals show uncertainty, and the unadjusted and sponsor analyses summarize the data using different measures and methods. The registry's secondary endpoint addresses death or hospitalization for COVID-19 complications or sequelae through Day 169; it is related to the primary question but is not the same endpoint.

Statistical evidence of a difference is not itself a judgment about clinical importance. The registry record does not provide a threshold for a clinically important effect or information that would support such a judgment. Nor do the posted results establish that every participant would have the same outcome or that the observed estimates apply outside the studied population. The results should be read as trial-specific estimates, with their stated uncertainty and limitations.

18. Why This Trial Matters Statistically

This trial illustrates why the estimand and the statistical scale matter. The primary comparison can be expressed as an absolute risk difference, a relative risk, an odds ratio, or the sponsor's relative risk reduction. Each describes the group contrast differently. Reporting these measures alongside the event counts helps distinguish the number of observed events from the size and direction of the estimated comparison.

It also shows how a composite endpoint shapes interpretation. The primary result combines severe COVID-19 and death through Day 29. A combined estimate can summarize the prespecified event definition, but it cannot tell readers from the overall result alone how much each component contributed. The secondary composite asks a related but different question over a longer timeframe. Keeping these endpoints distinct avoids treating a coherent direction across outcomes as proof that they measure the same thing.

The comparison also demonstrates the role of analysis methods and precision. The sponsor's Cochran-Mantel-Haenszel result and the unadjusted re-calculation are not identical procedures or scales. Confidence intervals convey uncertainty that a single point estimate or p-value cannot. Finally, the planning illustrations show how required participant numbers increase when the planning effect is made smaller, while making clear that those illustrations do not recover the trial's original design rationale.

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.

The registry record leaves several trial-specific details unresolved. Although it describes the primary endpoint population and gives its arm-specific sizes, the endpoint analysis counts are smaller; the reasons are not reported. The approach to missing data and any sensitivity analyses are also not reported, limiting assessment of how alternative assumptions might affect the estimates.

The registry names the Cochran-Mantel-Haenszel method but does not report the strata, stratification factors, or full analysis implementation. It does not report a multiplicity procedure, the original sample-size justification, interim analyses, or the trial's detailed statistical analysis plan. The planning illustrations based on the observed effect cannot fill those gaps, since they are not the original design assumptions.

The primary endpoint is a composite, and separate primary results for severe COVID-19 and death are not reported. The registry also does not provide enough detail in the posted safety summaries to assess event attribution, timing, or statistical comparisons. While the study is described as randomized and masked, the registry record does not report additional details about adherence, protocol deviations, or the success of masking. These are limits of what the record reports, not evidence that any particular problem occurred.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Phase III Study of AZD7442 for Treatment of COVID-19 in Outpatient Adults — NCT04723394.
• Registry record last updated 2023-07-05; status last verified 2023-06.
• Results first posted 2022-12-20.
• 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

NCT04723394 is a completed Phase III randomized, placebo-controlled study of AZD7442 in non-hospitalized adults with COVID-19. The registry records enrollment of 910 and primary completion on 2021-08-21. The primary endpoint is the composite of severe COVID-19 or death from any cause through Day 29 in the stated modified Full Analysis Set.

The posted primary event counts are 18 in the AZD7442 group and 37 in the placebo group. The sponsor-reported Cochran-Mantel-Haenszel analysis reports a relative risk reduction of 50.38, with a 95% confidence interval from 14.38 to 71.25 and p = 0.010. The unadjusted risk-difference estimate is -0.044, with a 95% confidence interval from -0.078 to -0.011.

The registry also reports a secondary composite through Day 169 and adverse-event counts. Important details, including missing-data handling, multiplicity procedures, and original sample-size assumptions, are not reported in the registry record.