This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical results on this page are restricted to the registry-reported COMET-PEAK trial data.
1. Trial at a Glance
COMET-PEAK was a randomized, parallel, open-label phase 2 study in non-hospitalized participants with mild to moderate COVID-19. The ClinicalTrials.gov record reports an enrollment of 354 participants and evaluate second-generation VIR-7831 material through safety, viral pharmacodynamic, and pharmacokinetic endpoints.
| Feature | COMET-PEAK |
|---|---|
| Trial name | COMET-PEAK |
| NCT identifier | NCT04779879 |
| Phase | Phase 2 |
| Therapeutic area | Infectious Disease |
| Condition | Covid19 |
| Population | Non-hospitalized participants with mild to moderate COVID-19 |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 354.0 |
| Results posted | Yes |
| Outcome measures posted | 127 |
| Statistical analyses posted | 12 |
| Primary analyses | 2 |
| Methods reported in analyses | ANCOVA; ANOVA |
| Primary effect measure | Ratio of geometric least squares mean |
| Lead sponsor | Vir Biotechnology, Inc. |
2. Clinical Question
The statistical question was not simply whether a treatment was associated with an outcome. The design separates several related questions: whether second-generation VIR-7831 material was safe and tolerated, how its administration affected SARS-CoV-2 viral-load exposure over time, and how pharmacokinetic exposure compared across intravenous and intramuscular dose levels.
Population
Non-hospitalized participants with mild to moderate COVID-19.
Intervention
Sotrovimab (Gen2), including 500 mg IV, 500 mg IM, and 250 mg IM regimens across Parts B and C.
Comparator
The formal statistical comparisons in the analyses posted on ClinicalTrials.gov are between specified Gen2 IV and IM dose groups rather than a placebo-controlled efficacy comparison.
Primary questions
How do the specified Gen2 administration groups compare in viral-load AUC from Day 1 to Day 8, and what safety and tolerability outcomes were evaluated?
3. Trial Design
the ClinicalTrials.gov record describes the study as randomized, parallel, and unmasked. The registered intervention field identifies sotrovimab (Gen1) and sotrovimab (Gen2), while the formal statistical analyses focus on second-generation VIR-7831 material.
Sotrovimab Gen1 / Gen2
- Sotrovimab Gen1: 500 mg IV
- Sotrovimab Gen2: 500 mg IV
- Safety and tolerability endpoints through Day 29
Second-generation material
- Part B: 500 mg IV and 500 mg IM
- Part C: 500 mg IV and 250 mg IM
- Viral pharmacodynamic and pharmacokinetic comparisons
4. Randomization, Stratification, and Analysis Populations
Randomization was part of the registered design. For the Part C viral-load analyses, the ANCOVA model additionally included the randomization stratification factor of prior exposure to an authorized or approved SARS-CoV-2 vaccine.
| Analysis population | Definition / role |
|---|---|
| Viral Pharmacodynamic Population | For Part B, participants in the Safety Population who had a Baseline (Day 1) quantifiable viral load assessed by qRT-PCR from nasopharyngeal swabs. For the reported analyses, participants with the required viral-load data and covariate information were analyzed. |
| Part C Viral Pharmacodynamic Population | Participants with data available at the specified time points without missing covariate information. |
| Pharmacokinetic Population | Participants with data available at the specified pharmacokinetic data points; individual analyses use the population definitions reported in the registry extract. |
| Safety Population | The population referenced by the Part A safety endpoint definition and the Part B/C serious-adverse-event counts. |
5. Primary Endpoints
| Endpoint | Time frame | Definition / assessment | Statistical method |
|---|---|---|---|
| Part A: Number of Participants With All Adverse Events (AEs) and Serious Adverse Events (SAEs) Through Day 29 | Up to Day 29 | An adverse event is an untoward medical occurrence temporally associated with use of a study intervention, whether or not considered related to the intervention. A serious adverse event is defined according to the registry's seriousness criteria. | Binary endpoint; the ClinicalTrials.gov record does not provide a formal comparative statistical analysis. |
| Part A: Number of Participants With Adverse Events of Special Interest (AESI) Through Day 29 | Up to Day 29 | AESIs were infusion-related reactions including hypersensitivity, events related to antibody-dependent enhancement, and events related to immunogenicity. | Binary endpoint; the ClinicalTrials.gov record does not provide a formal comparative statistical analysis. |
| Part A: Number of Participants With Worst-case Post Baseline Abnormal Electrocardiogram (ECG) Findings Through Day 29 | Up to Day 29 | Twelve-lead ECGs were recorded after rest. Clinically significant abnormal findings were determined by investigator clinical judgement. | Binary endpoint; the ClinicalTrials.gov record does not provide a formal comparative statistical analysis. |
| Part A: Number of Participants With Disease Progression Events (Disease-Related Events) Through Day 29 | Up to Day 29 | Disease-related events were adverse events related to expected progression, signs, or symptoms of COVID-19 that met the registered definition. | Binary endpoint; the ClinicalTrials.gov record does not provide a formal comparative statistical analysis. |
| Part B: Mean Area Under the Curve (AUC) of SARS-CoV-2 Viral Load From Day 1 to Day 8 (AUCD1-8) | Day 1 to Day 8 | AUC of SARS-CoV-2 viral load measured by qRT-PCR from Day 1 to Day 8 in nasopharyngeal swab samples. | ANCOVA adjusted for treatment and Baseline logarithm (base 10) viral load. |
| Part C: Mean AUC of SARS-CoV-2 Viral Load From Day 1 to Day 8 (AUCD1-8) | Day 1 to Day 8 | AUC of SARS-CoV-2 viral load measured by qRT-PCR from Day 1 to Day 8 in nasopharyngeal swab samples. | ANCOVA adjusted for treatment, Baseline logarithm (base10) viral load, and randomization stratification factor. |
6. Statistical Methodology
The formal primary viral-load analyses use analysis of covariance (ANCOVA). This is a linear-model framework that compares treatment groups while adjusting for baseline viral load. Because the reported effect measure is a ratio of geometric least squares means, the analysis is naturally interpreted on a multiplicative scale.
Part B ANCOVA
The Part B model included treatment and Baseline logarithm (base 10) viral load.
The reported comparison was Part B sotrovimab Gen2 500 mg IV versus Part B sotrovimab Gen2 500 mg IM.
Part C ANCOVA
The Part C model included treatment, Baseline logarithm (base 10) viral load, and the randomization stratification factor for prior exposure to an authorized or approved SARS-CoV-2 vaccine.
The reported comparison was Part C sotrovimab Gen2 500 mg IV versus Part C sotrovimab Gen2 250 mg IM.
The secondary pharmacokinetic analyses use both ANCOVA and ANOVA. ANCOVA was used for drug bioavailability comparisons with treatment and baseline weight as covariates. ANOVA was used for dose-proportionality analyses in the ClinicalTrials.gov record, with treatment dose as a covariate for each parameter of interest.
7. Primary Results: Part B Viral-Load AUC
Part B · AUCD1-8
Ratio of geometric least squares mean · 90% two-sided CI: 0.98–1.09
500 mg IV versus 500 mg IM · Day 1 to Day 8
| Feature | Reported result |
|---|---|
| Outcome | Mean Area Under the Curve (AUC) of SARS-CoV-2 Viral Load From Day 1 to Day 8 (AUCD1-8) |
| Unit | Day*log10 copies per milliliter (mL) |
| Groups | Part B Sotrovimab Gen2 500 mg IV vs Part B Sotrovimab Gen2 500 mg IM |
| Analysis population | Viral Pharmacodynamic Population |
| Method | ANCOVA |
| Effect measure | Ratio of geometric least squares mean |
| Estimate | 1.04 |
| Confidence interval | 90% two-sided CI: 0.98–1.09 |
| Covariates | Treatment and Baseline logarithm (base 10) viral load |
A ratio of geometric least squares means of 1.04 indicates that the model-adjusted geometric mean AUC for the first-listed group, 500 mg IV, was estimated to be 1.04 times that of 500 mg IM under the specified ANCOVA model. On a simple multiplicative interpretation, this is approximately 4% higher estimated AUC.
The estimate is not a percentage change in an individual participant's viral load, and it does not describe the probability that one participant will have a higher viral load than another. It is a group-level adjusted comparison of the AUC outcome.
The 90% two-sided confidence interval of 0.98–1.09 describes the uncertainty around the estimated ratio. It spans values below and above 1, so the registry-reported interval is compatible with modest differences in either direction as well as a ratio close to equality.
No p-value is reported for this analysis in the ClinicalTrials.gov record. A p-value would address a specified null-hypothesis test; it would not itself quantify the magnitude or practical importance of the observed ratio.
Because the endpoint is an integrated viral-load measure, interpretation depends on the measurement schedule and the construction of the AUC. The registry extract does not provide an underlying participant-level dataset, so the precision of the estimate cannot be independently reconstructed beyond the reported confidence interval.
8. Primary Results: Part C Viral-Load AUC
Part C · AUCD1-8
Ratio of geometric least squares mean · 90% two-sided CI: 0.94–1.11
500 mg IV versus 250 mg IM · Day 1 to Day 8
| Feature | Reported result |
|---|---|
| Outcome | Mean AUC of SARS-CoV-2 Viral Load From Day 1 to Day 8 (AUCD1-8) |
| Unit | Day*log10 copies/mL |
| Groups | Part C Sotrovimab Gen2 500 mg IV vs Part C Sotrovimab Gen2 250 mg IM |
| Analysis population | Viral Pharmacodynamic Population |
| Method | ANCOVA |
| Effect measure | Ratio of geometric least squares mean |
| Estimate | 1.02 |
| Confidence interval | 90% two-sided CI: 0.94–1.11 |
| Covariates | Treatment, Baseline logarithm (base10) viral load, and prior exposure to an authorized or approved SARS-CoV-2 vaccine as the randomization stratification factor |
A ratio of 1.02 means that the model-adjusted geometric mean AUC for the first-listed group, 500 mg IV, was estimated to be 1.02 times that of 250 mg IM. In multiplicative terms, the point estimate is approximately 2% higher.
This estimate should not be interpreted as evidence that every participant had a 2% difference in viral load. The outcome is an AUC and the estimate is adjusted for baseline viral load and the specified randomization stratification factor.
The 90% two-sided confidence interval of 0.94–1.11 spans 1.00. It therefore includes the possibility of essentially equal adjusted geometric mean AUCs as well as modest differences in either direction.
The registry-reported analysis does not report a p-value, and the hypothesis type is recorded as "Other / not stated." Consequently, the confidence interval and point estimate provide the principal quantitative description available from the registry-reported statistical analysis.
The inclusion of the vaccine-exposure stratification factor is an important design detail. It preserves information from the randomization structure in the adjusted analysis rather than treating that stratification factor as irrelevant after randomization.
9. Secondary Viral Pharmacodynamic Results
The ClinicalTrials.gov record also report six secondary viral-load AUC analyses. These use the same general ANCOVA framework and extend the observation window to Day 5 or Day 11.
| Endpoint | Comparison | Estimate | 90% two-sided CI | Method |
|---|---|---|---|---|
| Part B AUCD1-5 | 500 mg IV vs 500 mg IM | 1.05 | 1.00–1.11 | ANCOVA |
| Part B AUCD1-11 | 500 mg IV vs 500 mg IM | 1.02 | 0.97–1.07 | ANCOVA |
| Part C AUCD1-5 | 500 mg IV vs 250 mg IM | 1.01 | 0.93–1.09 | ANCOVA |
| Part C AUCD1-11 | 500 mg IV vs 250 mg IM | 1.02 | 0.94–1.10 | ANCOVA |
The Part B Day 1-to-Day 5 estimate was 1.05 with a 90% two-sided confidence interval of 1.00–1.11. The Part B Day 1-to-Day 11 estimate was 1.02 with a 90% two-sided confidence interval of 0.97–1.07. Both analyses adjusted for treatment and Baseline logarithm (base 10) viral load.
For Part C, the Day 1-to-Day 5 estimate was 1.01 with a 90% two-sided confidence interval of 0.93–1.09, while the Day 1-to-Day 11 estimate was 1.02 with a 90% two-sided confidence interval of 0.94–1.10. These analyses included treatment, baseline logarithm (base10) viral load, and the vaccine-exposure randomization stratification factor.
10. Secondary Pharmacokinetic Results
The pharmacokinetic analyses shift the statistical question from viral pharmacodynamics to exposure to VIR-7831 Gen2. The principal effect measure remains a ratio of geometric least squares means, which is appropriate for multiplicative pharmacokinetic comparisons.
| Outcome | Comparison | Method | Estimate | 90% two-sided CI |
|---|---|---|---|---|
| Dose-normalized AUCinf | 500 mg IM vs 500 mg IV | ANCOVA | 0.66 | 0.48–0.89 |
| Dose-normalized AUCinf | 250 mg IM vs 500 mg IV | ANCOVA | 0.58 | 0.43–0.79 |
| Dose-normalized AUCinf | 500 mg IM vs 250 mg IM | ANOVA | 1.14 | 0.67–1.95 |
| Dose-normalized AUClast | 500 mg IM vs 250 mg IM | ANOVA | 1.06 | 0.69–1.62 |
| Dose-normalized AUCD1-D29 | 500 mg IM vs 250 mg IM | ANOVA | 1.11 | 0.68–1.82 |
| Dose-normalized Cmax | 500 mg IM vs 250 mg IM | ANOVA | 1.28 | 0.77–2.12 |
Bioavailability comparisons
For the AUCinf comparisons involving intravenous administration, the registry states that drug bioavailability was analyzed using ANCOVA with treatment and weight at Baseline as covariates.
500 mg IM vs 500 mg IV
Ratio of geometric least squares mean: 0.66; 90% two-sided CI 0.48–0.89.
250 mg IM vs 500 mg IV
Ratio of geometric least squares mean: 0.58; 90% two-sided CI 0.43–0.79.
A ratio below 1 indicates a lower dose-normalized geometric mean AUC for the first-listed intramuscular comparison group relative to the intravenous reference in these analyses. The confidence intervals quantify uncertainty around those ratios; they are not probabilities that the true ratio lies inside the interval.
Dose proportionality across the two IM dose levels
The registry used ANOVA for the comparisons between the 250 mg IM and 500 mg IM dose levels. The analysis notes state that treatment dose (250 mg, 500 mg) was included as a covariate for each parameter of interest.
| PK parameter | 500 mg IM vs 250 mg IM | 90% two-sided CI |
|---|---|---|
| AUCinf | 1.14 | 0.67–1.95 |
| AUClast | 1.06 | 0.69–1.62 |
| AUCD1-D29 | 1.11 | 0.68–1.82 |
| Cmax | 1.28 | 0.77–2.12 |
These are ratio estimates, not raw concentration differences. In particular, a Cmax ratio of 1.28 represents an estimated 1.28-fold geometric mean ratio for the specified comparison. It does not mean that Cmax increased by 28% for every participant, nor does it by itself establish a formal dose-proportionality conclusion.
11. Safety
The ClinicalTrials.gov record provides serious adverse-event counts by the specific dose groups. These are affected participants divided by the stated number at risk.
| Study part | Treatment group | Serious adverse events |
|---|---|---|
| Part A | Sotrovimab Gen1: 500 mg IV | 0/8 |
| Part A | Sotrovimab Gen2: 500 mg IV | 0/22 |
| Part B | Sotrovimab Gen2: 500 mg IV | 1/84 |
| Part B | Sotrovimab Gen2: 500 mg IM | 2/82 |
| Part C | Sotrovimab Gen2: 500 mg IV | 2/79 |
| Part C | Sotrovimab Gen2: 250 mg IM | 3/78 |
The registry's Part A safety endpoints include all AEs and SAEs, AESIs, worst-case post-baseline abnormal ECG findings, and disease progression events through Day 29. The ClinicalTrials.gov record does not provide corresponding arm-level numerical results for those four Part A endpoints beyond the serious-adverse-event counts listed above.
12. Statistical Methods Explained
Why was ANCOVA used for the viral-load AUC?
ANCOVA allows the analysis to compare treatment groups while accounting for baseline viral load. Here the baseline covariate is the logarithm (base 10) of viral load. Adjusting for an important baseline measurement can improve precision and helps distinguish the treatment-group comparison from baseline differences in the outcome measure.
What does a ratio of geometric means of 1.04 mean?
It means the estimated geometric least squares mean for the first-listed group is 1.04 times that of the comparison group under the specified model. The ratio is multiplicative: 1.00 represents equal geometric means, while values above or below 1 represent higher or lower estimated geometric means for the first-listed group.
Why use a geometric rather than arithmetic comparison?
Viral-load and pharmacokinetic measurements can naturally be interpreted on a multiplicative scale. A ratio of geometric means expresses the relative exposure or outcome level directly. The analyses posted on ClinicalTrials.gov report the ratio rather than an arithmetic difference, so interpretation should remain on that scale.
What does the 90% confidence interval tell us?
The interval expresses statistical uncertainty around the estimated ratio under the analysis model. For example, the Part C primary estimate of 1.02 has a 90% two-sided confidence interval of 0.94–1.11. The interval therefore encompasses 1.00, meaning that the registry-reported interval includes equality as well as modest differences in either direction.
Why was a stratification factor included in the Part C model?
The Part C ANCOVA included prior exposure to an authorized or approved SARS-CoV-2 vaccine as the randomization stratification factor. Including a factor used in randomization helps preserve the planned structure of the comparison and adjusts the model for that categorical design feature.
What is the distinction between ANCOVA and ANOVA here?
Both are linear-model approaches. The registry-reported ANCOVA analyses explicitly incorporate covariates such as baseline viral load or baseline weight. The registry-reported ANOVA analyses address dose-level comparisons using treatment dose in the model. The choice therefore reflects the specific pharmacodynamic or pharmacokinetic question being analyzed.
Does a ratio near 1 prove the two treatments are equivalent?
No. A ratio close to 1 is a descriptive result. Formal equivalence or non-inferiority requires a prespecified margin and a corresponding hypothesis-testing framework. The registry-reported COMET-PEAK analyses identify the hypothesis type as "Other / not stated" and do not provide an equivalence or non-inferiority margin.
13. Interpreting the Statistical Evidence
COMET-PEAK is statistically interesting because the trial moves across several related scales of evidence. Safety endpoints are binary counts. Viral pharmacodynamics use an integrated continuous AUC outcome. Pharmacokinetic endpoints use dose-normalized exposure measures such as AUCinf, AUClast, AUCD1-D29, and Cmax. The analysis consequently changes from descriptive event counting to covariate-adjusted linear models.
A particularly important distinction is between precision and effect size. The point estimate describes the estimated ratio, while the confidence interval describes its uncertainty. A narrow interval around a ratio near 1 would provide a different statistical picture from a very wide interval around the same point estimate. Several of the pharmacokinetic intervals are comparatively broad, so the interval—not just the point estimate—is essential to interpretation.
The absence of reported p-values in the registry-reported primary analyses also matters. A p-value cannot be inferred from a confidence interval without knowing the precise hypothesis, confidence-level convention, model, and testing framework. The appropriate interpretation therefore uses the reported ratio and confidence interval directly rather than manufacturing a p-value.
14. Planned and Posted Safety Analyses
The registry identifies four Part A safety-related primary endpoints as having results posted: all adverse events and serious adverse events through Day 29, adverse events of special interest through Day 29, worst-case post-baseline abnormal ECG findings through Day 29, and disease progression events through Day 29.
These are binary participant-level endpoints. A typical comparative analysis of such an endpoint could describe event proportions by group and, where prespecified, estimate a risk difference, risk ratio, odds ratio, or an exact confidence interval. However, the registry-reported COMET-PEAK statistical-analyses data do not provide such formal comparative estimates for these Part A endpoints. The numerical safety information that is reported is limited to serious adverse events by dose group.
15. Multiplicity, Interim Analysis, and Missing Data
The ClinicalTrials.gov record does not describe a multiplicity-adjustment procedure, interim-analysis boundary, or formal alpha-spending strategy. They also do not specify a multiple-testing hierarchy across the six registered primary endpoints.
Missing-data handling is partly visible through the analysis-population definitions. For the Part C viral pharmacodynamic analyses, only participants with data available at the specified time points and without missing covariate information were analyzed. This is important because an analysis restricted to complete information can have a different estimand and precision than an analysis using an explicit missing-data model.
The ClinicalTrials.gov record does not specify an imputation method for missing viral-load measurements. Therefore, no assumption such as last observation carried forward, multiple imputation, or a particular missing-at-random model should be attributed to this analysis.
16. Non-Inferiority, Equivalence, and the Meaning of 1.00
The primary viral-load estimates are ratios, making 1.00 the natural reference value for equality of geometric means. But equality as a reference point is not the same thing as a formal equivalence hypothesis.
Reference point for a ratio
A ratio below 1 indicates a lower estimated geometric mean for the first-listed group; a ratio above 1 indicates a higher estimated geometric mean.
For a formal equivalence or non-inferiority analysis, the clinically acceptable difference would be represented by a prespecified margin. The registry-reported COMET-PEAK data do not report such a margin or state that the primary analyses were designed as equivalence or non-inferiority tests. It would therefore be inappropriate to label the confidence intervals as demonstrating equivalence solely because they are near 1.00.
17. Pharmacokinetic Interpretation
The pharmacokinetic results illustrate why dose normalization and route of administration matter. The registry-reported analysis compares 500 mg IM and 250 mg IM with a 500 mg IV reference for dose-normalized AUCinf, and it separately evaluates the two IM dose levels for several PK parameters.
For the 500 mg IM versus 500 mg IV AUCinf comparison, the reported ratio is 0.66 with a 90% two-sided confidence interval of 0.48–0.89. For 250 mg IM versus 500 mg IV, the ratio is 0.58 with a 90% two-sided confidence interval of 0.43–0.79. These estimates are explicitly described in the registry analysis notes as drug-bioavailability analyses adjusted for treatment and baseline weight.
The two-IM-dose analyses are different. Their estimates are 1.14 for AUCinf, 1.06 for AUClast, 1.11 for AUCD1-D29, and 1.28 for Cmax, with confidence intervals that are wider than the corresponding primary viral-load intervals. These results demonstrate the importance of considering uncertainty separately for each PK parameter rather than assuming that one exposure measure automatically determines all others.
18. Limitations
- Registry-level evidence: This analysis is limited to the numerical and methodological information contained in the ClinicalTrials.gov-derived trial data.
- No participant-level data: Individual viral-load trajectories, censoring patterns, residual distributions, and influence diagnostics cannot be independently assessed.
- Limited formal hypothesis information: The analyses posted on ClinicalTrials.gov record the hypothesis type as "Other / not stated" and do not provide p-values for the primary viral-load comparisons.
- No reported equivalence or non-inferiority margin: Ratios near 1 should not be converted into an equivalence or non-inferiority conclusion.
- Missing-data details are incomplete: The analysis populations identify participants with available data, but the registry-reported extract does not specify a formal imputation strategy.
- Multiplicity information is incomplete: The ClinicalTrials.gov record does not describe adjustment across the multiple primary and secondary outcomes.
- Safety comparison is descriptive here: Serious-adverse-event counts are provided, but comparative risk estimates and confidence intervals are not.
- Generalizability: The registered population is specifically non-hospitalized participants with mild to moderate COVID-19, so the statistical results should be interpreted within that study population.
19. Why This Trial Matters Statistically
COMET-PEAK provides a compact example of how biostatistical methods change with the scientific question. The same randomized study contains binary safety outcomes, continuous viral pharmacodynamic endpoints, and continuous pharmacokinetic endpoints. Treating all of these outcomes with the same statistical method would obscure important differences in their measurement and interpretation.
The viral-load analyses demonstrate the practical role of ANCOVA. Baseline viral load is not merely a descriptive characteristic; it becomes an explicit model covariate. The Part C analysis goes one step further by incorporating the vaccine-exposure stratification factor used in randomization.
The pharmacokinetic analyses demonstrate another important principle: route and dose comparisons can require a multiplicative effect measure. Ratios of geometric means make it possible to express exposure differences on a relative scale, while the confidence intervals show how much uncertainty remains around those estimates.
Finally, the trial illustrates why a statistical result should not be reduced to a single number. The estimate, confidence interval, analysis population, covariates, comparison direction, endpoint definition, and hypothesis framework all contribute to the meaning of the result.
The statistical takeaway
The most informative reading of COMET-PEAK combines the reported ratio, its 90% confidence interval, the ANCOVA or ANOVA model, the analysis population, and the exact comparison being made.
20. Related Tutorials
Learn more about the methods used in this trial:
21. Related Calculators
Explore statistical tools connected to the methods used in this analysis:
22. Sources
- ClinicalTrials.gov: COMET-PEAK (NCT04779879) — official trial registry record.
- PubMed: PMID 41284195 — linked publication identified in the ClinicalTrials.gov record.
- PubMed: PMID 34473343 — linked publication identified in the ClinicalTrials.gov record.
Continue the statistical pathway
Use the related tutorials and calculators to explore ANCOVA, ANOVA, confidence intervals, covariate adjustment, randomization, and stratified analysis in greater depth.
23. Record Summary
COMET-PEAK provides a useful statistical case study in randomized pharmacodynamic and pharmacokinetic analysis. Its primary viral-load analyses use ANCOVA with baseline viral load adjustment, with an additional vaccine-exposure stratification factor in Part C. The reported effect measure is a ratio of geometric least squares means, allowing the comparisons to be interpreted on a multiplicative scale.
The Part B Day 1-to-Day 8 viral-load AUC comparison produced a ratio of 1.04 with a 90% two-sided confidence interval of 0.98–1.09. The corresponding Part C comparison produced a ratio of 1.02 with a 90% two-sided confidence interval of 0.94–1.11. Secondary viral-load and pharmacokinetic analyses extend the same statistical framework across additional time windows, routes, dose levels, and PK parameters.