This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
VRV12 was a randomized, parallel-group, quadruple-masked phase 3 prevention trial comparing purified vero rabies vaccine, serum free, VRVg-2 with two reference rabies vaccines: Verorab and Imovax Rabies. The trial enrolled children and adults for a primary pre-exposure vaccination series and evaluated a single booster dose in a subset of adults.
| Feature | VRV12 |
|---|---|
| Phase | Phase 3 |
| Condition | Rabies; Healthy Volunteers |
| Primary purpose | Prevention |
| Design | Randomized, parallel-group, quadruple-masked |
| Enrollment | 1,708 |
| Study groups | 6 |
| Primary endpoint | Percentage of participants with RVNA titer ≥0.5 IU/mL at Day 42 post-vaccination |
| Primary endpoint type | Binary |
| Primary effect measure | Risk difference / percentage difference |
| Hypothesis framework | Non-inferiority; superiority for specified secondary immunogenicity objectives |
| ClinicalTrials.gov | NCT04127786 |
| Lead sponsor | Sanofi Pasteur, a Sanofi Company |
2. Clinical Question
The primary clinical question was whether the immune response produced by VRVg-2 was non-inferior to the responses produced by the reference rabies vaccines in the prespecified primary-series comparisons. The principal immunogenicity endpoint was the percentage of participants achieving an RVNA titer of at least 0.5 IU/mL at Day 42 after vaccination.
Population
Children and adults participating as healthy volunteers in a rabies pre-exposure vaccination study.
Intervention
Purified vero rabies vaccine, serum free, VRVg-2.
Comparators
Purified inactivated rabies vaccine, Verorab®, and purified inactivated rabies vaccine, Imovax® Rabies.
Primary question
Is the percentage achieving RVNA ≥0.5 IU/mL with VRVg-2 sufficiently close to the reference-vaccine percentage to satisfy the prespecified non-inferiority criterion?
3. Trial Design
VRVg-2
- Purified vero rabies vaccine, serum free, VRVg-2.
- Group 1: Cohort 1.
- Group 4: Cohort 2.
Verorab®
- Purified inactivated rabies vaccine, Verorab®.
- Group 2: Cohort 1.
- Group 5: Cohort 2.
Imovax® Rabies
- Purified inactivated rabies vaccine, Imovax® Rabies.
- Group 3: Cohort 1.
- Group 6: Cohort 2.
Subset of adults
- A single booster dose was evaluated in a subset of adults.
- The registry separately reports serious adverse events for the booster phase.
4. Trial Timeline
Study start
The VRV12 study began.
Primary completion
The registry reports primary study completion on March 25, 2020.
Statistical results posted
ClinicalTrials.gov contains posted outcome results and 12 statistical analyses, including four analyses of the registered primary endpoint.
5. Primary Endpoint
| Endpoint | Definition | Time frame | Type |
|---|---|---|---|
| Primary Series Cohort 1: Percentage of Participants With RVNA Titer ≥0.5 IU/mL | Percentage of participants with a rabies virus neutralizing antibody titer greater than or equal to 0.5 IU/mL. RVNA was assessed using the Rapid Fluorescent Focus Inhibition Test (RFFIT) assay method. | Day 42 (post-vaccination) | Binary |
The registered primary endpoint is a binary responder endpoint. Each participant is classified according to whether the Day 42 RVNA titer reaches the prespecified threshold of 0.5 IU/mL. The treatment comparison is therefore expressed as a difference between percentages rather than as a ratio or time-to-event measure.
6. Statistical Methodology
Per-protocol analysis set
The primary endpoint analyses were performed on the per-protocol analysis set (PPAS) for Day 42. The registry defines this population as participants who received at least 1 dose of study vaccine in the primary series and had no relevant protocol deviation before Day 42. The number analyzed for a specified category represents participants with available data for that category.
A risk difference of 0 means that the observed percentages are the same. A positive value means the observed percentage was higher in the VRVg-2 group; a negative value means it was lower.
Non-inferiority framework
The registry specifies that non-inferiority was demonstrated when the lower limit of the two-sided 95% confidence interval for the difference in percentages was greater than -5% at Day 42 for the primary-series comparisons of VRVg-2 with Verorab or Imovax Rabies.
Prespecified non-inferiority criterion
The margin defines how much lower the VRVg-2 responder percentage could be, under the prespecified confidence-interval criterion, while still satisfying the non-inferiority objective.
Binary endpoint analysis
The registry does not report a named statistical method for the posted primary analyses. It reports percentage differences, two-sided 95% confidence intervals, and the non-inferiority decision rule. For a binary endpoint of this type, the central statistical task is estimation of the between-group difference in response percentages together with its uncertainty.
Superiority sequencing
For the specified secondary immunogenicity objective involving VRVg-2, the registry states that, after non-inferiority was reached for VRVg-2 versus the comparator vaccines at Day 42, superiority was assessed using an overall observed responder percentage of at least 99% in the VRVg-2 group together with a lower limit of the 95% confidence interval of at least 97%.
Fixed-sequence multiplicity control
The registry states that the secondary immunogenicity outcome measures were evaluated sequentially following a fixed-sequence method. This is a multiplicity strategy in which later hypotheses are interpreted within a prespecified sequence rather than treating every comparison as an independent hypothesis test.
7. Primary Results
The registry reports four formal analyses of the primary endpoint. All four use the Day 42 PPAS and a two-sided 95% confidence interval for the percentage difference. The analyses are separated by age group and reference vaccine.
VRVg-2 vs Verorab® — Pediatric participants
Risk difference in RVNA response
95% CI: -1.4 to 4.3
Non-inferiority margin: -5%
Analysis population: Day 42 per-protocol analysis set
The reported percentage difference was 0, meaning the observed percentage of pediatric participants reaching an RVNA titer of at least 0.5 IU/mL was the same in the VRVg-2 and Verorab groups for this analysis.
The 95% confidence interval extends from -1.4 to 4.3. In the context of the prespecified non-inferiority framework, the lower confidence limit is above the -5% margin. The interval therefore provides the precision information needed for the registry's non-inferiority criterion.
The confidence interval does not say that individual participants' immune responses fall between -1.4% and 4.3%. It describes uncertainty around the estimated difference between the two group-level percentages. The risk difference also does not measure the magnitude of an individual participant's antibody response; it concerns the proportion crossing the prespecified threshold.
No p-value is reported for this primary analysis. In a non-inferiority design, the confidence interval relative to the prespecified margin is particularly important because the question is whether the treatment difference excludes an unacceptably large disadvantage, rather than simply whether a conventional null hypothesis of zero difference is rejected.
VRVg-2 vs Verorab® — Adult participants
Risk difference in RVNA response
95% CI: -0.6 to 6.4
Non-inferiority margin: -5%
Analysis population: Day 42 per-protocol analysis set
The estimated percentage difference was 1.2%, indicating that the observed percentage of adult participants reaching the RVNA threshold was 1.2 percentage points higher in the VRVg-2 group than in the Verorab group in this analysis.
The 95% confidence interval ranges from -0.6 to 6.4. The interval includes zero, so the estimate is compatible with a small disadvantage, essentially no difference, or a higher response percentage for VRVg-2. Importantly, the lower limit remains above the prespecified -5% non-inferiority margin.
The width of the interval also illustrates why a point estimate alone is insufficient. An estimate of 1.2% does not imply that the true percentage difference is exactly 1.2%; the confidence interval quantifies uncertainty around that estimate under the analysis framework.
No p-value is reported for this analysis. The non-inferiority question is evaluated using the confidence interval and the prespecified margin rather than by treating a p-value for equality as the primary decision criterion.
VRVg-2 vs Imovax Rabies® — Pediatric participants
Risk difference in RVNA response
95% CI: -1.4 to 4.4
Non-inferiority margin: -5%
Analysis population: Day 42 per-protocol analysis set
The estimated percentage difference was 0. Thus, the observed percentage of pediatric participants achieving an RVNA titer of at least 0.5 IU/mL was the same in the VRVg-2 and Imovax Rabies groups for this analysis.
The 95% confidence interval extends from -1.4 to 4.4. Because the lower limit is above the -5% non-inferiority margin, the confidence interval satisfies the registry's stated non-inferiority criterion.
The upper confidence limit of 4.4% indicates that the data are also compatible with a modestly higher responder percentage for VRVg-2. That possibility should not be confused with a formal superiority conclusion: non-inferiority and superiority are different statistical questions, and the registry's superiority framework is separately specified for the relevant secondary endpoint.
No p-value is reported for this comparison. The principal inferential quantity is the percentage difference and its confidence interval relative to the non-inferiority margin.
VRVg-2 vs Imovax Rabies® — Adult participants
Risk difference in RVNA response
95% CI: -1.5 to 4.6
Non-inferiority margin: -5%
Analysis population: Day 42 per-protocol analysis set
The estimated percentage difference was 0, indicating equal observed responder percentages for VRVg-2 and Imovax Rabies in the adult analysis.
The 95% confidence interval ranges from -1.5 to 4.6. The interval includes zero, but its lower limit remains above the prespecified -5% non-inferiority margin. Thus, the interval is compatible with the registry's stated non-inferiority criterion.
As with the other primary comparisons, the interval is about uncertainty in the group-level percentage difference rather than uncertainty about an individual person's antibody titer. The analysis also uses the PPAS, so its interpretation is conditional on the population defined by receipt of at least 1 dose and absence of relevant protocol deviations before Day 42.
No p-value is reported. A non-inferiority conclusion is not equivalent to proving the two vaccines are identical; it establishes that the estimated difference is sufficiently bounded away from the prespecified unacceptable disadvantage under the stated confidence-interval rule.
| Primary comparison | Population | Risk difference | 95% CI | NI margin |
|---|---|---|---|---|
| VRVg-2 vs Verorab® | Pediatric | 0 | -1.4 to 4.3 | -5% |
| VRVg-2 vs Verorab® | Adult | 1.2% | -0.6 to 6.4 | -5% |
| VRVg-2 vs Imovax Rabies® | Pediatric | 0 | -1.4 to 4.4 | -5% |
| VRVg-2 vs Imovax Rabies® | Adult | 0 | -1.5 to 4.6 | -5% |
8. Secondary Endpoint Results
The registry reports several secondary immunogenicity analyses involving the percentage of participants with RVNA titers of at least 0.5 IU/mL. These analyses extend the primary comparison to Day 28, pooled populations, and a comparison of a 2-dose VRVg-2 regimen with a 3-dose Imovax Rabies regimen.
VRVg-2 — Day 42 overall response
Percentage achieving RVNA ≥0.5 IU/mL
95% CI: 99.3 to 100
Analysis population: Day 42 PPAS
Hypothesis type: Superiority
The registry states that this superiority objective was evaluated after the non-inferiority objectives for VRVg-2 versus comparator vaccines were reached at Day 42. The stated superiority criterion required an overall observed percentage of at least 99% and a lower limit of the 95% confidence interval of at least 97%.
The reported estimate of 100% means that all participants with available data in the specified analysis category were classified as having an RVNA titer of at least 0.5 IU/mL. The 95% confidence interval ranges from 99.3 to 100, providing a quantitative measure of uncertainty around the observed percentage.
The lower confidence limit of 99.3% is above the registry's stated 97% threshold for the superiority objective. This is a threshold-based interpretation rather than a comparison of VRVg-2 against a separate treatment group's response percentage.
The result does not mean that every vaccinated person necessarily has the same antibody concentration, because the endpoint is binary: it records whether the RVNA titer crosses 0.5 IU/mL. It also does not imply that the underlying continuous antibody titers are all equal.
Pooled VRVg-2 vs pooled Verorab® — Pediatric
Day 28 risk difference
95% CI: -1.4 to 4.3
Non-inferiority margin: -5%
This analysis compared pooled Groups 1 and 4 receiving VRVg-2 with pooled Groups 2 and 5 receiving Verorab® in pediatric participants. The analysis was performed on the Day 28 PPAS.
Pooled VRVg-2 vs pooled Verorab® — Adult
Day 28 risk difference
95% CI: -1.9 to 2.7
Non-inferiority margin: -5%
The adult pooled comparison estimated a percentage difference of -0.2%. The confidence interval extends from -1.9 to 2.7, and its lower limit remains above the registry's -5% non-inferiority margin.
Pooled VRVg-2 vs pooled Imovax Rabies® — Pediatric
Day 28 risk difference
95% CI: -1.4 to 4.5
Non-inferiority margin: -5%
The pooled pediatric comparison of VRVg-2 with Imovax Rabies® produced an estimated percentage difference of 0, with a 95% confidence interval from -1.4 to 4.5.
Pooled VRVg-2 vs pooled Imovax Rabies® — Adult
Day 28 risk difference
95% CI: -0.5 to 5.3
Non-inferiority margin: -5%
The adult pooled comparison produced an estimated percentage difference of 1.8%, with a 95% confidence interval from -0.5 to 5.3. The lower confidence limit is above the -5% non-inferiority margin specified for the Day 28 pooled comparison.
Two-dose VRVg-2 at Day 28 vs three-dose Imovax Rabies® at Day 42 — Pediatric
Cross-regimen risk difference
95% CI: -1.4 to 4.4
Non-inferiority margin: -10%
The registry describes an additional non-inferiority comparison between pooled Groups 1 and 4 receiving VRVg-2 at Day 28 and Cohort 1 Group 3 receiving Imovax Rabies® at Day 42. For pediatric participants, the estimate was 0 with a 95% confidence interval of -1.4 to 4.4. The stated non-inferiority margin for this cross-regimen comparison was -10%.
Two-dose VRVg-2 at Day 28 vs three-dose Imovax Rabies® at Day 42 — Adult
Cross-regimen risk difference
95% CI: -3.1 to 3.0
Non-inferiority margin: -10%
For adults, the corresponding estimate was -1.7%, with a 95% confidence interval from -3.1 to 3. The lower limit is above the stated -10% non-inferiority margin.
Imovax Rabies® — Day 28 vs Day 42
Within-vaccine percentage difference
95% CI: -4.4 to 1.3
Non-inferiority margin: -10%
The registry also reports a non-inferiority analysis within the Imovax Rabies® group comparing Day 28 with Day 42. The estimated percentage difference was -1.3%, with a 95% confidence interval from -4.4 to 1.3. The stated non-inferiority margin was -10%.
| Secondary analysis | Estimate | 95% CI | Hypothesis | Margin |
|---|---|---|---|---|
| VRVg-2 vs pooled Verorab®, pediatric, Day 28 | 0 | -1.4 to 4.3 | Non-inferiority | -5% |
| VRVg-2 vs pooled Verorab®, adult, Day 28 | -0.2% | -1.9 to 2.7 | Non-inferiority | -5% |
| VRVg-2 vs pooled Imovax®, pediatric, Day 28 | 0 | -1.4 to 4.5 | Non-inferiority | -5% |
| VRVg-2 vs pooled Imovax®, adult, Day 28 | 1.8% | -0.5 to 5.3 | Non-inferiority | -5% |
| VRVg-2 Day 28 vs Imovax® Day 42, pediatric | 0 | -1.4 to 4.4 | Non-inferiority | -10% |
| VRVg-2 Day 28 vs Imovax® Day 42, adult | -1.7% | -3.1 to 3.0 | Non-inferiority | -10% |
| Imovax® Day 28 vs Day 42 | -1.3% | -4.4 to 1.3 | Non-inferiority | -10% |
9. Primary Analysis Population: Why Per-Protocol Matters
The primary endpoint was analyzed using the Day 42 per-protocol analysis set rather than simply all randomized participants. The registry definition requires receipt of at least 1 dose in the primary series and no relevant protocol deviation before Day 42.
Why use a PPAS?
For non-inferiority, protocol adherence can be important because substantial deviations may make two treatments appear more similar by weakening the treatment contrast.
What does PPAS exclude?
Participants with relevant protocol deviations before Day 42 are excluded from this defined analysis population.
Available-data rule
The registry defines the number analyzed as participants with available data for each specified category.
Interpretive caution
A per-protocol estimate is conditional on protocol adherence and is not interchangeable with an all-randomized analysis.
This distinction is particularly important in non-inferiority trials. In a superiority trial, protocol deviations generally tend to dilute differences between groups and make it harder to detect superiority. In a non-inferiority trial, dilution can instead make treatments appear more similar and therefore potentially make a non-inferiority conclusion easier to obtain. For that reason, the definition and integrity of the PPAS are central to interpretation.
10. Non-Inferiority Logic
The most important statistical feature of VRV12 is that its principal comparisons were framed around a prespecified non-inferiority margin. The registry uses the difference in responder percentages and compares the lower confidence limit with the margin.
For the cross-regimen comparisons involving 2-dose VRVg-2 at Day 28 versus 3-dose Imovax Rabies at Day 42, the registry specifies a -10% margin instead.
Consider the adult VRVg-2 versus Verorab comparison. The estimated difference was 1.2%, while the lower confidence limit was -0.6%. Even though the confidence interval includes zero, its lower bound remains substantially above -5%. This illustrates a central point in non-inferiority testing: an interval can include zero while still supporting non-inferiority.
Conversely, non-inferiority does not mean the two vaccines have identical immunogenicity. It means the observed data are sufficiently precise, under the prespecified analysis framework, to exclude a difference worse than the non-inferiority margin.
11. Statistical Methods Explained
Why was a risk difference used?
The primary endpoint is binary: each participant either has an RVNA titer of at least 0.5 IU/mL or does not. A risk difference expresses the comparison directly in percentage points, making the non-inferiority margin naturally interpretable on the same scale.
Why is the non-inferiority margin more important than a p-value for the primary question?
The primary question is whether VRVg-2 is not unacceptably worse than the reference vaccine. The registry defines that boundary as -5% for the primary Day 42 comparisons. The relevant evidence is therefore whether the lower limit of the 95% confidence interval remains above that boundary. A conventional p-value for equality would answer a different question.
What does a risk difference of 0 mean?
A risk difference of 0 means the observed percentages meeting the RVNA threshold were equal in the two compared groups. It does not mean that every participant had the same antibody titer or that the underlying continuous antibody distributions were identical.
Why can a confidence interval include zero while still supporting non-inferiority?
Zero represents no difference, whereas the non-inferiority margin represents the largest disadvantage considered acceptable under the trial design. For example, an interval from -0.6% to 6.4% includes zero but does not extend down to -5%. The interval therefore remains compatible with non-inferiority even though it does not exclude equality.
Why does the per-protocol population matter?
Non-inferiority can be sensitive to treatment dilution caused by protocol deviations. The registry therefore defines the primary analysis using participants who received at least 1 dose and had no relevant protocol deviation before Day 42. The resulting estimate answers the non-inferiority question in that defined analysis population.
Why was a fixed-sequence approach used for secondary immunogenicity outcomes?
The registry states that secondary immunogenicity outcome measures were evaluated sequentially following a fixed-sequence method. Such a hierarchy can control multiplicity by allowing subsequent formal interpretations only after earlier objectives in the sequence have been addressed.
Why are the Day 28 and Day 42 comparisons not interchangeable?
They represent different time points and, in some analyses, different vaccine-dose regimens. A comparison of VRVg-2 at Day 28 with Imovax Rabies at Day 42 is therefore not simply a comparison of the same endpoint at two common time points; it is a cross-regimen comparison with a separately specified -10% non-inferiority margin.
12. Confidence Intervals: Reading the Primary Results
The four primary confidence intervals can be read directly against the -5% non-inferiority margin.
The visual comparison is useful because the decision boundary is not zero. All four lower confidence limits lie above -5%. The point estimates themselves range from 0 to 1.2%, but the confidence intervals are what provide the precision assessment needed for the stated non-inferiority framework.
13. Multiplicity and Sequential Testing
VRV12 contains multiple immunogenicity objectives. The registry specifically identifies multiplicity adjustment as a concept in the secondary analysis and states that the secondary immunogenicity outcomes were evaluated sequentially following a fixed-sequence method.
| Feature | Registry-supported interpretation |
|---|---|
| Primary endpoint | Day 42 RVNA response in Primary Series Cohort 1 |
| Primary hypotheses | Non-inferiority |
| Secondary hypothesis types | Non-inferiority and superiority |
| Multiplicity concept | Secondary immunogenicity measures evaluated sequentially |
| Sequence | Fixed-sequence method |
A fixed sequence is statistically different from conducting many unrelated tests at the same significance level. The sequence creates an order in which formal inferential objectives are considered. This matters because a collection of confidence intervals and hypothesis tests should not automatically be interpreted as if each were an independent confirmatory claim.
14. Serious Adverse Events
The registry reports serious adverse events by study group for both the primary series and the booster phase. These counts are presented as affected participants over participants at risk.
| Phase | Group | Vaccine | Serious AEs | At risk |
|---|---|---|---|---|
| Primary Series: Cohort 1 | Group 1 | VRVg-2 | 10 | 607 |
| Primary Series: Cohort 1 | Group 2 | Verorab | 3 | 202 |
| Primary Series: Cohort 1 | Group 3 | Imovax | 5 | 200 |
| Primary Series: Cohort 2 | Group 4 | VRVg-2 | 2 | 419 |
| Primary Series: Cohort 2 | Group 5 | Verorab | 0 | 139 |
| Primary Series: Cohort 2 | Group 6 | Imovax | 1 | 139 |
| Booster Phase: Cohort 1 | Group 1 | VRVg-2 | 2 | 94 |
| Booster Phase: Cohort 1 | Group 2 | VRVg-2 | 0 | 31 |
| Booster Phase: Cohort 1 | Group 3 | VRVg-2 | 1 | 32 |
| Booster Phase: Cohort 2 | Group 4 | VRVg-2 | 3 | 128 |
| Booster Phase: Cohort 2 | Group 5 | VRVg-2 | 0 | 41 |
| Booster Phase: Cohort 2 | Group 6 | VRVg-2 | 1 | 43 |
The raw counts illustrate why safety comparisons should be interpreted using both the number of affected participants and the corresponding number at risk. A count of zero in a smaller group and a larger count in a larger group do not have the same denominator context.
Primary-series Group 1
10 serious adverse events were reported among 607 participants at risk in Cohort 1 Group 1.
Primary-series Group 4
2 serious adverse events were reported among 419 participants at risk in Cohort 2 Group 4.
Verorab Cohort 2
No serious adverse events were reported among 139 participants at risk in Group 5.
Imovax Cohort 2
1 serious adverse event was reported among 139 participants at risk in Group 6.
15. Booster Phase: Statistical Perspective
The trial included a single booster-dose evaluation in a subset of adults. The registry reports serious adverse events for this booster phase, but the posted statistical analyses in the ClinicalTrials.gov record focus on the primary-series immunogenicity endpoint and secondary immunogenicity comparisons.
The booster-phase safety counts therefore provide descriptive safety information rather than a posted comparative efficacy analysis. The denominators vary by group, ranging from 31 to 128 participants at risk in the reported booster groups.
| Booster group | Serious AEs | At risk |
|---|---|---|
| Cohort 1 Group 1 | 2 | 94 |
| Cohort 1 Group 2 | 0 | 31 |
| Cohort 1 Group 3 | 1 | 32 |
| Cohort 2 Group 4 | 3 | 128 |
| Cohort 2 Group 5 | 0 | 41 |
| Cohort 2 Group 6 | 1 | 43 |
16. What the Risk Difference Does — and Does Not — Mean
A risk difference of 1.2% means that the observed percentage of participants meeting the binary RVNA threshold was 1.2 percentage points higher in the VRVg-2 group than in the comparator group for the specified analysis.
It does not mean that an individual's antibody titer was 1.2% higher, nor does it quantify the magnitude of the continuous antibody response among participants who crossed the threshold.
For the adult VRVg-2 versus Verorab comparison, the estimate of 1.2% has a 95% confidence interval of -0.6 to 6.4. The interval communicates that the estimate is uncertain and is compatible with a small negative difference as well as a positive difference.
Zero represents no observed difference. The non-inferiority margin represents the maximum disadvantage considered acceptable under the trial design. Consequently, an interval may cross zero and still satisfy non-inferiority if its lower limit remains above the prespecified margin.
The endpoint classifies participants using an RVNA threshold of 0.5 IU/mL. Two participants with very different antibody titers can receive the same binary classification if both are above the threshold. The binary endpoint therefore answers a threshold-response question rather than a continuous antibody-concentration question.
17. Limitations and Interpretation Issues
- Per-protocol analysis: the primary analyses use the Day 42 PPAS, so the estimates are specific to the population meeting the registry's protocol-adherence definition.
- Binary endpoint: the primary endpoint records whether RVNA reaches 0.5 IU/mL and does not describe the full continuous distribution of antibody titers.
- Confidence-interval interpretation: confidence intervals quantify uncertainty around the group-level percentage difference; they do not describe the range of individual treatment effects.
- Non-inferiority margin: the -5% margin is a design parameter. Its clinical acceptability is a substantive trial-design judgment and should not be confused with a statistical estimate from the observed data.
- Cross-time comparisons: the analyses comparing VRVg-2 at Day 28 with Imovax Rabies at Day 42 involve different assessment times and different regimen structures, so they should not be interpreted as ordinary same-time treatment comparisons.
- Multiplicity: the registry identifies a fixed-sequence approach for secondary immunogenicity outcomes, so individual secondary analyses should be interpreted within that sequence rather than as unrelated tests.
- Booster subset: the booster phase represents a subset of adults and has smaller reported denominators than the primary enrollment population.
- Method reporting: the registry does not identify a named the statistical method for the posted percentage-difference analyses. The inferential framework is instead described through the effect measure, confidence interval, hypothesis type, and non-inferiority margin.
18. Why This Trial Matters Statistically
VRV12 is a useful teaching example because its central statistical question is different from the familiar superiority question used in many randomized trials. The objective is not simply to demonstrate that one vaccine produces a larger response percentage than another. The primary framework asks whether the observed difference can exclude a prespecified clinically unacceptable loss of response.
| Concept | How it appears in VRV12 |
|---|---|
| Randomization | Participants were randomized across six study groups. |
| Blinding | The registry describes the study as quadruple-masked. |
| Binary endpoint | RVNA titer ≥0.5 IU/mL at specified post-vaccination time points. |
| Risk difference | Primary treatment effect expressed as a percentage difference. |
| Non-inferiority | Primary comparisons use a -5% margin. |
| Per-protocol analysis | Primary Day 42 analyses use the PPAS. |
| Confidence interval | Two-sided 95% CIs are compared with the non-inferiority margin. |
| Multiplicity | Secondary immunogenicity outcomes were evaluated sequentially. |
| Superiority | A specified secondary VRVg-2 objective used a threshold-based superiority criterion after non-inferiority. |
| Repeated time points | RVNA was assessed at Day 0, Day 28, and Day 42 in specified analyses. |
| Cross-regimen comparison | Two-dose VRVg-2 at Day 28 was compared with three-dose Imovax Rabies at Day 42 using a -10% margin. |
19. Statistical Interpretation of the Overall Evidence
The primary analyses consistently produced lower confidence limits above the registry's prespecified -5% non-inferiority margin. The four primary point estimates were 0, 1.2%, 0, and 0%, respectively, while the corresponding lower confidence limits were -1.4%, -0.6%, -1.4%, and -1.5%.
That pattern is statistically important because the point estimate is not the principal decision boundary in a non-inferiority design. The critical comparison is between the lower confidence limit and the margin. Each primary interval remains above -5%.
The secondary analyses provide additional information at Day 28 and in pooled populations. Their estimates range from -1.7% to 1.8% for the reported non-inferiority comparisons, with lower confidence limits remaining above their specified margins of either -5% or -10%.
The separate superiority objective for VRVg-2 reports an observed percentage of 100% with a 95% confidence interval of 99.3 to 100. The registry's stated superiority criterion required an observed percentage of at least 99% and a lower confidence limit of at least 97% after the non-inferiority objective had been reached.
These findings should be read as a collection of endpoint-specific statistical statements rather than as a single universal measure of vaccine performance. The primary endpoint is binary, the main inferential framework is non-inferiority, and the analysis population is per-protocol. Those design features determine what the estimates can legitimately establish.
20. Related Tutorials
Learn more about the methods used in this trial:
21. Related Calculators
22. Sources
- ClinicalTrials.gov: NCT04127786.
- Linked publication: PubMed record for PMID 39587931.
Continue through the Clinical Biostats statistical library
Explore related biostatistical tutorials, calculators, and clinical trial analyses.
23. Record Summary
VRV12 provides a clear example of non-inferiority methodology applied to a binary immunogenicity endpoint. The trial compared VRVg-2 with two reference rabies vaccines, used a Day 42 per-protocol analysis for its primary endpoint, expressed treatment effects as percentage differences, and evaluated two-sided 95% confidence intervals against a prespecified -5% non-inferiority margin.
The four primary analyses produced percentage differences of 0, 1.2%, 0, and 0%, with lower confidence limits of -1.4%, -0.6%, -1.4%, and -1.5%, respectively. Each lower limit is above the registry's -5% margin. Secondary analyses extended the framework to pooled populations, Day 28 assessments, and cross-regimen comparisons using a -10% margin.
Statistically, the most important lesson is that a non-inferiority conclusion depends on the prespecified margin, the confidence interval, and the analysis population—not simply on whether the point estimate is positive or negative and not on whether a conventional p-value crosses 0.05. The VRV12 registry record also illustrates how fixed-sequence multiplicity control and per-protocol analysis shape the interpretation of a modern vaccine trial.