This page separates reported trial results from statistical interpretation. Numerical results are restricted to the ClinicalTrials.gov record. Where the registry contains an endpoint or result without a formal statistical-analysis record, that distinction is made explicitly.
1. Trial at a Glance
INNOVATE was a randomized, quadruple-masked, sequential phase 2/3 trial evaluating INO-4800 for prevention of COVID-19 in adults at high risk of SARS-CoV-2 exposure. The registry reports six arms and an enrollment of 1307 participants, with primary analyses focused on phase 2 cellular and neutralizing-antibody immune responses and a phase 3 virologically confirmed COVID-19 endpoint.
| Feature | INNOVATE |
|---|---|
| Trial name | INNOVATE |
| ClinicalTrials.gov identifier | NCT04642638 |
| Phase | Phase 2/3 |
| Status | Terminated |
| Therapeutic area | Pulmonology |
| Conditions | Coronavirus Infection; Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2); COVID-19 Disease |
| Allocation | Randomized |
| Design model | Sequential |
| Masking | Quadruple |
| Primary purpose | Prevention |
| Enrollment | 1307 |
| Interventions | INO-4800; CELLECTRA® 2000; Placebo; CELLECTRA® 2000 |
| Lead sponsor | Inovio Pharmaceuticals |
| Sponsor type | Industry |
2. Clinical Question
The clinical question was whether INO-4800 could generate measurable antigen-specific cellular and neutralizing-antibody immune responses and, in phase 3, affect the occurrence of virologically confirmed COVID-19 disease in participants who were SARS-CoV-2 seronegative at baseline.
Population
Adults at high risk of SARS-CoV-2 exposure, in a trial addressing Coronavirus Infection, SARS-CoV-2, and COVID-19 Disease.
Intervention
INO-4800, administered with the CELLECTRA® 2000 device within the randomized trial structure.
Comparator
Placebo, with CELLECTRA® 2000 identified among the trial interventions.
Primary question
How did the randomized groups compare in immune-response endpoints in phase 2, and how was virologically confirmed COVID-19 disease defined for the phase 3 efficacy endpoint?
3. Trial Design
The ClinicalTrials.gov record identifies the study as a six-arm, randomized, quadruple-masked, sequential phase 2/3 prevention trial. The available trial data do not provide enough detail to reconstruct the full six-arm assignment structure, dose schedules, or randomization ratio, so those features are not inferred here.
4. Trial Arms and Intervention Structure
The registry data identify INO-4800, placebo, and the CELLECTRA® 2000 device among the trial interventions. The phase-specific serious-adverse-event records further identify six named dose groups or treatment groups.
Dose groups
- INO-4800 Dose Group 1
- INO-4800 Dose Group 2
- Serious adverse events: 4/151 and 2/147, respectively
Placebo dose groups
- Placebo Dose Group 1
- Placebo Dose Group 2
- Serious adverse events: 3/50 and 2/51, respectively
INO-4800 dose group
- INO-4800 Dose Group (2.0mg Per
- Serious adverse events: 4/601
Placebo dose group
- Placebo Dose Group
- Serious adverse events: 4/302
5. Primary Endpoints
| Endpoint | Time frame | Type | Registry definition |
|---|---|---|---|
| Phase 2: Change From Baseline in Antigen-specific Cellular Immune Response Measured by Interferon-gamma (IFN-γ) Enzyme-linked Immunospot (ELISpot) Assay | Baseline up to Week 6 | Other / unclear | Whole blood and serum samples were collected for the cellular immunology assessment. The antigen-specific cellular immune response to INO-4800 was measured in spot-forming units per million peripheral blood mononuclear cells (SFU/10^6, PBMC) using ELISpot. No samples collected after Week 6 were analyzed. |
| Phase 2: Change From Baseline in Neutralizing Antibody Response Measured by a Pseudovirus-based Neutralization Assay | Baseline up to Week 6 | Other / unclear | The immune responses to INO-4800 were measured using assays that included a pseudovirus-based neutralization assay. Immunology blood samples were collected at serial timepoints. No samples collected after Week 6 were analyzed. |
| Phase 3: Percentage of Participants (SARS-CoV-2 Seronegative at Baseline) With Virologically-confirmed COVID-19 Disease | From 14 days after completion of the 2-dose regimen up to 3 months post-dose 2 (i.e. Day 42 up to Day 126) | Binary | Participants were virologically-confirmed cases of COVID-19 if tested positive by SARS-CoV-2 RT-PCR assay, with specified COVID-19 symptoms. |
The registry identifies three primary endpoints, while the ClinicalTrials.gov record contains six formal primary analyses. Those six analyses correspond to the two phase 2 immune-response endpoints, with three pairwise comparisons for the cellular endpoint and three pairwise comparisons for the neutralizing-antibody endpoint.
6. Analysis Populations
The formal primary analyses posted on ClinicalTrials.gov for the phase 2 immune-response endpoints used a per-protocol (PP) population. The registry description states that the PP population included participants who received all doses of the trial intervention and had no key protocol violations.
| Population | Role in the ClinicalTrials.gov record |
|---|---|
| Per-protocol population | Used for the six reported primary phase 2 statistical analyses. |
| Overall number analyzed | The registry describes this as the number of participants included in the particular analysis; the registry-reported extract does not provide the complete analysis-specific counts. |
7. Results: Phase 2 Cellular Immune Response
The first primary phase 2 endpoint measured change from baseline in antigen-specific cellular immune response using IFN-γ ELISpot. The outcome unit was SFU/10^6, PBMC, and the registry reports differences in medians with two-sided non-parametric 95% confidence intervals.
INO-4800 Dose Group 1 vs Placebo Dose Group 1
Difference in median change from baseline
95% CI: 0.00 to 6.70 · Two-sided
Outcome: SFU/10^6, PBMC
The reported estimate of 6.70 is the difference in median post-baseline change from baseline between the two specified phase 2 groups. In the direction reported, the estimate indicates a higher median change in the INO-4800 Dose Group 1 comparison than in the Placebo Dose Group 1 comparison.
The estimate does not mean that every participant had a change of 6.70 SFU/10^6, nor does it describe an individual-level treatment effect. It is a comparison of group-level medians.
The two-sided 95% confidence interval extends from 0.00 to 6.70. This describes uncertainty around the estimated median difference under the analysis framework; it does not describe the range of immune responses in individual participants.
No p-value is reported in the registry statistical-analysis record. A p-value would address compatibility with a specified null hypothesis, not the magnitude or practical importance of the estimated difference. The registry also describes the analysis as non-parametric but does not name the specific statistical test.
INO-4800 Dose Group 2 vs Placebo Dose Group 2
Difference in median change from baseline
95% CI: 3.30 to 17.80 · Two-sided
Outcome: SFU/10^6, PBMC
The reported median difference of 13.30 compares the post-baseline change from baseline between INO-4800 Dose Group 2 and Placebo Dose Group 2. The positive estimate indicates a higher median change in the INO-4800 group under the direction of the reported contrast.
The confidence interval from 3.30 to 17.80 provides a range of values compatible with the estimated median difference under the stated statistical framework. It does not mean that 95% of participants had individual changes within that interval.
No p-value is reported for this analysis. A confidence interval and a p-value answer related but different questions: the interval conveys the estimated effect and its precision, whereas a p-value summarizes evidence against a particular null hypothesis.
The analysis was performed in the per-protocol population. That is important because the result applies to the participants included in that protocol-defined population rather than automatically to every randomized participant.
INO-4800 Dose Group 1 vs INO-4800 Dose Group 2
Difference in median change from baseline
95% CI: −10.00 to 0.00 · Two-sided
Outcome: SFU/10^6, PBMC
The estimate of −6.60 compares the two INO-4800 dose groups. Under the reported contrast, the negative estimate indicates a lower median change in Dose Group 1 relative to Dose Group 2.
The 95% confidence interval spans −10.00 to 0.00. Because the interval reaches 0.00, the registry-reported interval does not exclude no median difference at the corresponding two-sided confidence level.
No p-value is provided, so a formal p-value-based inference should not be reported. The estimate also should not be interpreted as evidence of a particular dose-response mechanism: a between-dose comparison is descriptive of the measured endpoint unless the broader trial design and prespecified hypotheses establish a dose-response question.
8. Results: Phase 2 Neutralizing Antibody Response
The second primary phase 2 endpoint measured change from baseline in neutralizing antibody response using a pseudovirus-based neutralization assay. The outcome unit was inhibition dilution 50 (ID50). The registry reports geometric mean fold-rise ratios for the between-group comparisons.
INO-4800 Dose Group 1 vs Placebo Dose Group 1
Geometric mean fold-rise ratio
95% CI: 1.720 to 3.632 · Two-sided
Outcome: inhibition dilution 50 (ID50)
A geometric mean fold-rise ratio of 2.50 means that the geometric mean fold rise in the INO-4800 Dose Group 1 comparison was 2.50 times the corresponding value in Placebo Dose Group 1 under the reported analysis.
This is a relative geometric measure; it does not mean that antibody titers increased by exactly 2.50 units, that every participant experienced a 2.50-fold increase, or that the ratio directly represents clinical protection against COVID-19.
The 95% confidence interval of 1.720 to 3.632 describes uncertainty around the estimated ratio. Because the interval is entirely above 1, the registry-reported confidence interval is consistent with a positive between-group geometric mean fold-rise ratio at the stated confidence level.
The registry states that the 95% confidence interval was based on a t-distribution and that the GMFR ratio was based on between-treatment-group comparisons. No p-value is reported.
INO-4800 Dose Group 2 vs Placebo Dose Group 2
Geometric mean fold-rise ratio
95% CI: 2.638 to 5.700 · Two-sided
Outcome: inhibition dilution 50 (ID50)
A ratio of 3.88 indicates that the geometric mean fold rise in the INO-4800 Dose Group 2 comparison was 3.88 times that in Placebo Dose Group 2 under the reported analysis.
The estimate is not an absolute antibody concentration and does not establish that a participant was protected from infection or disease. Immunogenicity and clinical efficacy are related questions, but they are not interchangeable endpoints.
The 95% confidence interval, 2.638 to 5.700, indicates the precision of the estimated ratio under the specified statistical framework. It does not give the range of individual participant responses.
The registry states that the confidence interval was based on a t-distribution.
Placebo Dose Group 1 vs Placebo Dose Group 2
Geometric mean fold-rise ratio
95% CI: 0.512 to 0.893 · Two-sided
Outcome: inhibition dilution 50 (ID50)
The reported ratio of 0.68 compares the geometric mean fold rises of the two placebo dose groups. Under the stated contrast, the geometric mean fold rise for Placebo Dose Group 1 was 0.68 times that for Placebo Dose Group 2.
The estimate should not be interpreted as a treatment effect because both groups are placebo groups. It is a comparison within the placebo portion of the phase 2 design.
The 95% confidence interval is 0.512 to 0.893. Because the interval is below 1, the estimated ratio is distinguishable from a ratio of 1 at the corresponding two-sided confidence level under the registry-reported analysis framework. That statistical statement does not by itself establish a clinically meaningful difference.
No p-value is provided. The analysis notes again specify that the confidence interval was based on a t-distribution.
9. Primary Results in One Table
| Primary endpoint | Comparison | Effect measure | Estimate | 95% CI |
|---|---|---|---|---|
| IFN-γ ELISpot change from baseline | INO-4800 Dose Group 1 vs Placebo Dose Group 1 | Difference in median | 6.70 | 0.00 to 6.70 |
| IFN-γ ELISpot change from baseline | INO-4800 Dose Group 2 vs Placebo Dose Group 2 | Difference in median | 13.30 | 3.30 to 17.80 |
| IFN-γ ELISpot change from baseline | INO-4800 Dose Group 1 vs INO-4800 Dose Group 2 | Difference in median | −6.60 | −10.00 to 0.00 |
| Neutralizing antibody change from baseline | INO-4800 Dose Group 1 vs Placebo Dose Group 1 | GMFR ratio | 2.50 | 1.720 to 3.632 |
| Neutralizing antibody change from baseline | INO-4800 Dose Group 2 vs Placebo Dose Group 2 | GMFR ratio | 3.88 | 2.638 to 5.700 |
| Neutralizing antibody change from baseline | Placebo Dose Group 1 vs Placebo Dose Group 2 | GMFR ratio | 0.68 | 0.512 to 0.893 |
10. Planned Analysis: Phase 3 COVID-19 Disease Endpoint
The registry reports results for the phase 3 primary endpoint, but the ClinicalTrials.gov record does not contain a formal statistical-analysis entry for this endpoint.
Registered phase 3 endpoint
Percentage of participants who were SARS-CoV-2 seronegative at baseline with virologically confirmed COVID-19 disease
Time frame: Day 42 up to Day 126
The endpoint is defined over a specified post-vaccination period: from 14 days after completion of the 2-dose regimen through 3 months after dose 2, corresponding to Day 42 through Day 126. The registry-reported definition identifies virologically confirmed cases using a positive SARS-CoV-2 RT-PCR assay together with specified COVID-19 symptoms.
For a binary disease endpoint, common effect measures include a risk ratio, risk difference, or odds ratio. In a randomized prevention trial, the choice should follow the prespecified statistical analysis plan and should be accompanied by an appropriate confidence interval. If time to confirmed disease is analyzed instead, a time-to-event method such as Kaplan-Meier estimation and a Cox model may be appropriate, but that method is not specified in the ClinicalTrials.gov record and therefore is not attributed to INNOVATE.
11. Statistical Methodology
Median differences for IFN-γ ELISpot
The cellular immune-response analyses compared post-baseline changes from baseline using differences in medians. The analysis notes describe these as non-parametric 95% confidence intervals. The specific non-parametric test or estimation procedure is not reported in the ClinicalTrials.gov record.
Unlike a mean difference, a median difference focuses on the middle observation of each distribution. This can be useful when immune-response measurements are skewed or contain extreme observations.
Geometric mean fold-rise ratios
The neutralizing-antibody analyses used geometric mean fold-rise ratios. The registry-reported analysis notes state that the 95% confidence intervals were based on a t-distribution and that the GMFR ratio was based on between-treatment-group comparisons.
A ratio of 1 represents equal geometric mean fold rise between the two comparison groups. A ratio above 1 indicates a larger geometric mean fold rise in the numerator group; a ratio below 1 indicates a smaller one.
Per-protocol analysis
The reported phase 2 analyses used a PP population consisting of participants who received all doses and had no key protocol violations. The choice changes the population being compared and therefore changes the interpretation of the estimate.
Binary disease endpoint
The phase 3 primary endpoint is binary: participants either meet the registry definition of virologically confirmed COVID-19 disease during the specified period or they do not. The ClinicalTrials.gov record identifies the endpoint and time frame but do not identify the formal statistical model used for the posted phase 3 result.
Confidence intervals
All six registry-reported primary statistical analyses report two-sided 95% confidence intervals. A confidence interval is useful because it presents both an estimated effect and the uncertainty surrounding that estimate. It should not be interpreted as a statement that 95% of individual patient responses fall within the interval.
12. Statistical Methods Explained
Why use a difference in medians for the ELISpot endpoint?
The reported ELISpot analysis compares changes from baseline using differences in medians and a non-parametric confidence interval. A median is less sensitive to extreme observations than a mean, which can be useful for biological measurements that are not well described by a symmetric distribution. The ClinicalTrials.gov record does not identify the exact non-parametric test, so no specific test is attributed to the trial.
What does a median difference of 13.30 mean?
It means that the median change from baseline in the first named group exceeded the corresponding median change in the comparison group by 13.30 SFU/10^6, PBMC under the reported contrast. It does not mean that each participant differed by 13.30, nor that the result describes the full distribution of immune responses.
What does a geometric mean fold-rise ratio of 3.88 mean?
A ratio of 3.88 means the geometric mean fold rise in the first named group was 3.88 times that in the comparison group. Because this is a ratio of geometric quantities, it is naturally interpreted multiplicatively rather than as an additive difference in ID50.
Why does the confidence interval matter?
An estimate without uncertainty can conceal how precisely the data identify the effect. For example, the 3.88 ratio has a 95% confidence interval from 2.638 to 5.700. The interval communicates the uncertainty around the ratio; it does not establish a range of effects that must occur in individual participants.
Why does a p-value not measure effect size?
A p-value is a measure of how compatible the observed data are with a specified null hypothesis under the statistical model. It does not tell the reader how large the observed effect is. Effect estimates and confidence intervals are therefore essential for understanding magnitude and precision.
Why does the phase 3 endpoint require a different analysis framework?
The phase 3 endpoint is binary rather than continuous. The primary question is whether participants experienced virologically confirmed COVID-19 disease during a defined interval. Binary outcomes are therefore summarized using event proportions and an appropriate comparative effect measure, according to the prespecified analysis plan. The registry extract does not identify the formal phase 3 statistical method.
Why does the per-protocol population matter?
A PP population excludes participants who do not meet specified adherence or protocol criteria. This can provide a clearer estimate of the effect among participants who followed the protocol, but it also means the result is not automatically equivalent to an analysis of all randomized participants. The distinction is particularly important when interpreting randomized trials.
13. Understanding the Immune-Response Results
Cellular immunity
The IFN-γ ELISpot endpoint measures antigen-specific cellular immune response using spot-forming units per million peripheral blood mononuclear cells.
Neutralizing antibodies
The pseudovirus-based neutralization endpoint is expressed in inhibition dilution 50 (ID50), with results reported as geometric mean fold-rise ratios.
Different effect scales
The cellular endpoint uses a median difference, whereas the antibody endpoint uses a geometric ratio. Their numerical estimates therefore cannot be directly compared.
Week 6 boundary
The registry definitions state that no samples collected after Week 6 were analyzed for these phase 2 immune-response endpoints.
The two phase 2 endpoints address different biological dimensions of immune response. A cellular immune-response estimate in SFU/10^6, PBMC and an antibody GMFR ratio are on different measurement scales. A larger numerical value in one analysis therefore does not imply a larger biological or clinical effect than a different numerical value in the other analysis.
14. Multiplicity and Multiple Primary Analyses
The ClinicalTrials.gov record identifies three registered primary endpoints and six formal primary statistical analyses. The six analyses arise because the two phase 2 endpoints were each examined across three pairwise comparisons.
| Endpoint family | Formal comparisons reported | Effect measure |
|---|---|---|
| IFN-γ ELISpot change from baseline | INO-4800 Dose Group 1 vs Placebo Dose Group 1; INO-4800 Dose Group 2 vs Placebo Dose Group 2; INO-4800 Dose Group 1 vs Dose Group 2 | Difference in median |
| Neutralizing antibody change from baseline | INO-4800 Dose Group 1 vs Placebo Dose Group 1; INO-4800 Dose Group 2 vs Placebo Dose Group 2; Placebo Dose Group 1 vs Placebo Dose Group 2 | GMFR ratio |
The ClinicalTrials.gov record does not state the multiplicity-adjustment procedure, familywise error rate, alpha allocation, hierarchical testing sequence, or hypothesis hierarchy for these six analyses. Consequently, the confidence intervals are reported exactly as provided, but no additional claim about multiplicity control is made.
15. Safety Results
The ClinicalTrials.gov record reports serious adverse events by arm as affected participants divided by participants at risk. These counts are reported separately for the phase 2 and phase 3 groups.
| Study group | Serious adverse events | Affected / at risk |
|---|---|---|
| Phase 2: INO-4800 Dose Group 1 | Serious adverse events | 4 / 151 |
| Phase 2: INO-4800 Dose Group 2 | Serious adverse events | 2 / 147 |
| Phase 2: Placebo Dose Group 1 | Serious adverse events | 3 / 50 |
| Phase 2: Placebo Dose Group 2 | Serious adverse events | 2 / 51 |
| Phase 3: INO-4800 Dose Group (2.0mg Per | Serious adverse events | 4 / 601 |
| Phase 3: Placebo Dose Group | Serious adverse events | 4 / 302 |
The affected/at-risk counts describe the number of participants with serious adverse events relative to the number at risk in each listed group. They are safety counts rather than efficacy estimates, and they should not be combined across phase 2 and phase 3 without a prespecified definition of the population and exposure period.
16. Randomization and Blinding
Randomization is a central design feature of INNOVATE. Random assignment helps create comparable groups at baseline, so observed between-group differences can be interpreted in the context of a randomized treatment comparison rather than a purely observational comparison.
The trial is described as quadruple-masked. Masking can reduce the opportunity for knowledge of assignment to influence participant behavior, clinical assessment, treatment administration, or other aspects of the trial. The ClinicalTrials.gov record does not specify the four masked roles, so they are not identified here.
Randomization protects comparison
Random assignment helps balance known and unknown prognostic factors in expectation, supporting causal interpretation of treatment-group differences.
Masking protects measurement
Masking can reduce bias that could arise when participants or study personnel know which intervention was assigned.
17. Timing of Endpoint Assessment
Trial start
The ClinicalTrials.gov record identifies November 30, 2020 as the study start date.
Phase 2 immune-response window
Both registered phase 2 primary immune-response endpoints use a baseline-up-to-Week-6 time frame. No samples collected after Week 6 were analyzed for these endpoints.
Phase 3 disease window
The registered COVID-19 disease endpoint begins 14 days after completion of the 2-dose regimen and extends to 3 months post-dose 2.
Primary completion
The ClinicalTrials.gov record identifies September 13, 2022 as the primary completion date.
The different assessment windows illustrate an important statistical principle: an endpoint is defined not only by what is measured, but also by when it is measured. A Week 6 immune-response comparison and a Day 42-to-Day 126 disease endpoint answer different questions and should not be conflated.
18. What the Effect Measures Do — and Do Not — Mean
A median difference such as 13.30 summarizes how the middle values of two change-from-baseline distributions differ. It does not imply that every participant experienced that difference and does not describe the variability of individual responses.
A ratio such as 3.88 describes a multiplicative comparison of geometric mean fold rises. It does not mean that antibody measurements increased by 3.88 absolute units, nor does it by itself establish clinical protection.
A 95% confidence interval describes statistical uncertainty around an estimated effect under the specified analysis framework. It does not represent the distribution of individual participant responses.
No p-values are posted on ClinicalTrials.gov for the six formal primary analyses in the ClinicalTrials.gov record.
19. Primary Endpoint Comparison: Statistical Reading
| Question | What the ClinicalTrials.gov record allows us to say |
|---|---|
| Was the phase 2 cellular response compared statistically? | Yes. Three median-difference comparisons are reported, each with a two-sided 95% confidence interval. |
| Was the phase 2 neutralizing-antibody response compared statistically? | Yes. Three geometric mean fold-rise ratio comparisons are reported, each with a two-sided 95% confidence interval. |
| Are p-values reported? | No p-values are reported in the six statistical-analysis records. |
| Is a formal phase 3 treatment-effect estimate reported? | No formal statistical-analysis record for the phase 3 primary endpoint appears in the registry-reported six analysis records. |
| Can a clinical efficacy conclusion be inferred from immune-response estimates alone? | No. Immune-response measures and virologically confirmed COVID-19 disease are distinct endpoints. |
20. Important Limitations and Interpretation Issues
- Per-protocol analysis: the six reported phase 2 analyses use a PP population. This can limit direct generalization to all randomized participants.
- Incomplete phase 3 statistical detail: the ClinicalTrials.gov record identifies the phase 3 endpoint and state that results are posted, but do not provide a formal statistical-analysis record for it.
- Multiplicity: six formal primary analyses are reported across the phase 2 endpoint families, but the extract does not specify an adjustment strategy or hypothesis hierarchy.
- Method not fully specified: the cellular analysis is described as non-parametric, but the exact test is not reported in the ClinicalTrials.gov record.
- Arm structure: the trial has six arms, but the ClinicalTrials.gov record does not provide a complete six-arm assignment table or randomization ratio.
- Safety detail: serious-adverse-event counts are available, but the ClinicalTrials.gov record does not provide a full adverse-event profile or a formal comparative safety analysis.
- Immune response versus clinical outcome: immunogenicity results describe biological responses; they do not substitute for direct analysis of the registered COVID-19 disease endpoint.
- Time-frame restriction: the phase 2 primary immune-response definitions explicitly restrict analyzed samples to baseline through Week 6.
- Terminated status: the registry status in the ClinicalTrials.gov record is TERMINATED. The statistical meaning of individual posted results should therefore be distinguished from any assumption about completion of every planned efficacy analysis.
21. Why This Trial Matters Statistically
INNOVATE is a useful statistical teaching case because it places two different classes of endpoint in the same randomized phase 2/3 program: continuous immunogenicity measures in phase 2 and a binary clinical disease endpoint in phase 3.
| Concept | How it appears in INNOVATE |
|---|---|
| Randomization | The trial uses randomized allocation. |
| Blinding | The registry describes the study as quadruple-masked. |
| Sequential design | The design model is listed as sequential. |
| Continuous endpoint | Phase 2 immune-response changes are continuous measurements. |
| Median difference | IFN-γ ELISpot changes are compared using differences in medians. |
| Geometric ratio | Neutralizing-antibody results are expressed using geometric mean fold-rise ratios. |
| Confidence intervals | All six registry-reported formal primary analyses report two-sided 95% confidence intervals. |
| Per-protocol analysis | The six registry-reported phase 2 analyses use a PP population. |
| Binary endpoint | The phase 3 endpoint is the percentage of baseline-seronegative participants with virologically confirmed COVID-19 disease. |
| Endpoint timing | Phase 2 uses baseline through Week 6; phase 3 uses Day 42 through Day 126. |
| Multiplicity | Six formal primary analyses are reported across the two phase 2 immune-response endpoints. |
| Safety comparison | Serious-adverse-event counts are reported separately by phase and arm. |
22. Related Tutorials
Learn more about the methods used in this trial:
23. Related Calculators
24. Sources
- ClinicalTrials.gov: INNOVATE (NCT04642638).
Continue through Clinical Biostats
Explore statistical tutorials and calculators related to randomized clinical-trial design, confidence intervals, endpoint analysis, and immunogenicity comparisons.
25. Record Summary
INNOVATE illustrates how statistical interpretation changes with endpoint type. Its phase 2 primary analyses use per-protocol populations and compare immune-response changes using median differences for IFN-γ ELISpot and geometric mean fold-rise ratios for neutralizing antibodies. The six registry-reported formal analyses all report two-sided 95% confidence intervals, but no p-values are provided. The phase 3 primary endpoint is a binary clinical disease outcome measured from Day 42 through Day 126 in participants who were SARS-CoV-2 seronegative at baseline, while the ClinicalTrials.gov record does not provide a formal statistical comparison for that endpoint.
The most important statistical lesson is that an immune-response estimate and a clinical disease endpoint answer different questions. A median difference or geometric mean ratio describes a biological measurement; it does not by itself establish a reduction in COVID-19 disease. Likewise, a confidence interval communicates uncertainty around an estimate but does not replace knowledge of the endpoint definition, analysis population, timing, multiplicity structure, and prespecified statistical method.