This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial facts on this page are restricted to the ClinicalTrials.gov record. Because the ClinicalTrials.gov record contains no formal statistical-analysis results, the results sections distinguish reported endpoint status from statistical interpretation.
1. Trial at a Glance
Mosaico was a completed phase 3 randomized, parallel-group, triple-masked prevention trial enrolling 3900 participants. The study evaluated Ad26.Mos4.HIV together with Clade C and Mosaic gp140 HIV bivalent vaccine against placebo for prevention of HIV-1 infection.
| Feature | Mosaico |
|---|---|
| Trial name | Mosaico |
| Phase | Phase 3 |
| Status | Completed |
| Therapeutic area | Infectious Disease |
| Condition | Healthy |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Triple |
| Primary purpose | Prevention |
| Enrollment | 3900 |
| Arms | 2 |
| Primary endpoints | 2 binary endpoints |
| Lead sponsor | Janssen Vaccines & Prevention B.V. |
| Sponsor type | Industry |
| Start | 2019-10-31 |
| Primary completion | 2023-08-10 |
2. Clinical Question
The central statistical question is whether the randomized vaccine regimen affects the occurrence of confirmed HIV-1 infection during the registered follow-up periods, compared with placebo.
Population
Healthy cis-gender men and transgender individuals who have sex with cis-gender men and/or transgender individuals, as specified by the trial's brief title.
Intervention
Ad26.Mos4.HIV together with Clade C and Mosaic gp140 HIV bivalent vaccine.
Comparator
Placebo.
Primary question
How many participants developed confirmed HIV-1 infection between Month 7 and Month 24, and between Month 7 and Month 30, in the per-protocol set?
3. Trial Design
Ad26.Mos4.HIV + Clade C and Mosaic gp140
- Ad26.Mos4.HIV
- Clade C and Mosaic gp140 HIV bivalent vaccine
- Serious adverse events: 82/1942 affected/at risk
Placebo
- Placebo
- Randomized parallel comparator group
- Serious adverse events: 77/1945 affected/at risk
The ClinicalTrials.gov record describes a randomized, parallel-group, triple-masked prevention study. Those design features are statistically important because randomization establishes the framework for comparing the occurrence of infection between groups, while masking is intended to reduce opportunities for knowledge of assignment to influence study conduct or assessment.
What triple masking contributes
The registry identifies the study as triple-masked but does not provide, in the ClinicalTrials.gov record, the identities of the three masked parties. The statistical implication is nevertheless important: masking is a design safeguard intended to reduce information-related bias. It should not be confused with randomization. Randomization addresses allocation of participants to treatment groups; masking addresses knowledge of that allocation.
4. Endpoints
| Registered primary endpoint | Time frame | Endpoint type |
|---|---|---|
| Number of Participants With Confirmed Human Immunodeficiency Virus (HIV)-1 Infections Diagnosed Between the Month 7 and Month 24 Visits (Per-protocol [PP] Set) | From Month 7 up to Month 24 | Binary |
| Number of Participants With a Confirmed Human Immunodeficiency Virus (HIV)-1 Infections Diagnosed Between the Month 7 and Month 30 Visits (PP Set) | From Month 7 up to Month 30 | Binary |
For both endpoints, the registry definition states that the number of participants with confirmed HIV-1 infections was reported and that the data represent the cumulative incidence of HIV-1 infections. Both endpoints use the per-protocol set.
5. Statistical Methodology
Binary endpoint analysis
The two registered primary endpoints are binary outcomes: for each participant in the specified per-protocol analysis set, the relevant outcome is whether a confirmed HIV-1 infection was diagnosed during the stated follow-up interval.
For a randomized comparison of two groups, a natural statistical framework for such an endpoint is a comparison of infection proportions. Depending on the prespecified statistical analysis plan, this can be expressed using a risk ratio, risk difference, odds ratio, or a hypothesis test comparing the underlying proportions. The ClinicalTrials.gov record does not identify which formal comparison was used, so this page does not attribute a particular test or effect measure to Mosaico.
For two randomized groups, common comparative measures include the risk difference, risk ratio, and odds ratio. The selected measure should follow the prespecified analysis plan rather than being chosen after looking at the observed results.
Cumulative incidence
The registry explicitly describes the endpoint data as representing cumulative incidence of HIV-1 infections. In this context, cumulative incidence refers to the proportion of the relevant analysis population that has experienced the specified infection event during the defined observation interval.
Because the two primary endpoints use different upper follow-up boundaries, they answer related but distinct questions: one concerns confirmed infections from Month 7 up to Month 24, while the other concerns confirmed infections from Month 7 up to Month 30.
Per-protocol analysis
Both primary endpoints are explicitly registered for the per-protocol set. This matters because the per-protocol population is not necessarily identical to the randomized population. A per-protocol analysis generally restricts attention to participants who satisfy prespecified criteria for adherence, eligibility, follow-up, or other protocol requirements.
Confidence intervals
A formal binary-outcome analysis would ordinarily accompany an estimated treatment contrast with a confidence interval. The interval quantifies statistical uncertainty around the estimated contrast under the chosen statistical model and sampling framework. The registry-reported Mosaico statistical-analysis data contain no confidence intervals, so none are reported here.
Hypothesis testing
A confirmatory binary endpoint analysis would ordinarily specify a null hypothesis and an alternative hypothesis before examining the data. For example, a null hypothesis could state that the two groups have equal infection risk over the specified interval. A p-value from such a test would quantify the compatibility of the observed data with that null model; it would not itself quantify the size or clinical importance of a treatment effect.
6. Results Status and Planned Analysis
ClinicalTrials.gov reports that results are posted and that both registered primary endpoints have results posted for them. However, the ClinicalTrials.gov record contains no formal statistical analyses. Consequently, the available information establishes that endpoint results were reported but does not provide a formal treatment-effect estimate, confidence interval, or p-value that can be reproduced on this page.
| Primary endpoint | Registry result status | Formal statistical analysis reported? |
|---|---|---|
| Confirmed HIV-1 infections from Month 7 up to Month 24, PP set | Results posted | No statistical-analysis record reported |
| Confirmed HIV-1 infections from Month 7 up to Month 30, PP set | Results posted | No statistical-analysis record reported |
How these endpoints would typically be analysed
For each primary endpoint, the first step would be to construct a two-by-two table containing the number of participants with and without confirmed HIV-1 infection in each randomized group within the defined per-protocol analysis set. The primary comparison could then be expressed through an appropriate measure of relative or absolute risk, accompanied by a confidence interval and a prespecified hypothesis test.
Month 24 analysis
The analysis would compare the occurrence of confirmed HIV-1 infection from Month 7 up to Month 24 in the per-protocol set.
Month 30 analysis
The analysis would compare the occurrence of confirmed HIV-1 infection from Month 7 up to Month 30 in the per-protocol set.
7. Clinical Biostats Interpretation of the Endpoint Structure
The primary endpoint counts participants with a confirmed HIV-1 infection during a specified follow-up interval. It therefore represents an event occurrence measure at the participant level rather than a continuous measure such as a laboratory value.
The endpoint does not by itself describe the biological mechanism of protection, the severity of infection, or outcomes after infection. It records whether the specified confirmed infection event occurred during the registered interval.
An infection count from Month 7 up to Month 24 is not interchangeable with an infection count from Month 7 up to Month 30. Extending observation can change the number of participants who experience an event and therefore changes the estimand being described.
Both primary endpoints are defined on the per-protocol set. The resulting estimate therefore describes the specified per-protocol population rather than automatically representing the entire enrolled population of 3900 participants.
8. Statistical Methods Explained
Why is this a binary endpoint?
Each primary endpoint asks whether a participant had a confirmed HIV-1 infection during a defined interval. That creates a two-category outcome for the relevant analysis population: infection occurred or it did not occur. The registry consequently identifies the primary endpoint type as binary.
Why is cumulative incidence relevant?
The registry explicitly states that the infection data represent cumulative incidence. Cumulative incidence is useful when the scientific question concerns the proportion of a defined population that experiences an event over a specified period. It incorporates the time window into the definition of the outcome.
Why is the per-protocol set important?
The primary endpoints are registered in the PP set rather than simply being described for all enrolled participants. This means that the denominator used for the formal endpoint analysis depends on the protocol-defined criteria for membership in that set. Without those criteria, the enrollment number of 3900 should not automatically be treated as the denominator for either primary endpoint.
What would a risk ratio mean?
If a risk ratio were the prespecified effect measure, it would compare the probability of confirmed infection in the intervention group with the probability in the comparator group over the specified interval. A ratio below 1 would indicate a lower estimated infection probability in the intervention group, while a ratio above 1 would indicate a higher estimated probability. This is a general explanation of the measure, not a claim that a risk ratio was the formal Mosaico analysis.
What would a risk difference mean?
A risk difference would subtract the infection probability in the comparator group from that in the intervention group. Unlike a relative measure, it is expressed on an absolute probability scale. It can therefore communicate the absolute separation between groups, but its interpretation depends on the baseline infection probability.
Why does a p-value not measure effect size?
A p-value is tied to a null hypothesis and the statistical model used to calculate it. It does not tell the reader how large the treatment effect is. Effect size should be described using an appropriate measure such as a risk difference or risk ratio, together with its confidence interval and the relevant absolute event frequencies.
9. Randomization and Blinding
The registry identifies Mosaico as randomized and triple-masked. These two features address different sources of potential bias.
| Design feature | Statistical role |
|---|---|
| Randomization | Creates the treatment-assignment mechanism that supports comparison of outcomes between groups. |
| Parallel design | Participants remain assigned to one of the two study groups rather than switching between randomized interventions as part of the primary design. |
| Triple masking | Reduces the opportunity for knowledge of treatment assignment to influence aspects of trial conduct or assessment. |
| Prevention purpose | The outcome of interest is incident HIV-1 infection rather than treatment of an established disease. |
Randomization does not guarantee that every observed characteristic will be numerically identical between groups. Its statistical purpose is to make treatment assignment independent of baseline prognostic factors in expectation under the randomization mechanism. Masking provides a separate safeguard against information-related influences.
10. Safety Results
The ClinicalTrials.gov record reports serious adverse events by arm as affected participants divided by participants at risk.
| Safety measure | Ad26.Mos4.HIV + Clade C and Mosaic gp140 | Placebo |
|---|---|---|
| Serious adverse events | 82/1942 | 77/1945 |
The denominators posted on ClinicalTrials.gov for this safety measure are 1942 and 1945, respectively. These are the reported participants at risk for the serious-adverse-event measure and should not be substituted for the enrollment total of 3900 when interpreting this safety result.
11. Safety and Event Denominators
One of the most important statistical disciplines in reading a trial record is to keep denominators attached to the endpoint they belong to.
Enrollment denominator
The trial enrolled 3900 participants. This is the overall enrollment figure reported by the registry data.
Safety denominators
The serious-adverse-event data use 1942 and 1945 as participants at risk in the two groups.
Primary efficacy denominator
The registered primary endpoints use the per-protocol set. Its size is not reported in the ClinicalTrials.gov record.
Why this matters
Using an inappropriate denominator can produce a misleading infection rate, safety rate, or treatment comparison even when the numerator is correctly copied.
12. What Can and Cannot Be Concluded From the registry Results Record
The ClinicalTrials.gov record establishes several important facts about the Mosaico results record: results are posted; both registered primary endpoints have results posted; and the endpoint definitions concern confirmed HIV-1 infections in the per-protocol set over specified follow-up intervals.
They do not, however, supply formal statistical-analysis records. Therefore this page cannot responsibly report an observed risk ratio, risk difference, odds ratio, confidence interval, p-value, or hypothesis-test conclusion for either primary endpoint.
| Information | Available in the ClinicalTrials.gov record? |
|---|---|
| Trial enrollment | Yes — 3900 |
| Number of arms | Yes — 2 |
| Randomized design | Yes |
| Triple masking | Yes |
| Primary endpoint definitions | Yes |
| Endpoint time frames | Yes |
| Endpoint type | Yes — Binary |
| Results posted status | Yes |
| Formal statistical-analysis records | No |
| Primary endpoint effect estimates | No |
| Primary endpoint confidence intervals | No |
| Primary endpoint p-values | No |
| Serious adverse events by arm | Yes |
13. Trial Timeline
Trial start
The registry profile lists 2019-10-31 as the study start date.
Randomized prevention study
The study is described as a randomized, parallel, triple-masked phase 3 trial with 3900 enrolled participants and 2 arms.
First registered primary endpoint window
The first primary endpoint counts confirmed HIV-1 infections diagnosed from Month 7 up to Month 24 in the per-protocol set.
Second registered primary endpoint window
The second primary endpoint counts confirmed HIV-1 infections diagnosed from Month 7 up to Month 30 in the per-protocol set.
Primary completion
The registry profile lists 2023-08-10 as the primary completion date.
14. Statistical Interpretation of a Binary Prevention Endpoint
An absolute risk difference answers: how much did the probability of confirmed infection differ between groups over the specified interval? It is directly tied to the event probability in each group.
A relative measure such as a risk ratio answers: how does the infection probability in one group compare proportionally with the other? Relative and absolute measures can communicate different aspects of the same comparison.
A confidence interval around an effect estimate indicates how precisely the study estimates the comparative quantity under the selected statistical framework. A narrow interval generally provides more precise estimation than a wide interval, although precision is not the same as clinical importance.
A p-value addresses compatibility with a specified null hypothesis. It does not replace an effect estimate, and it should not be interpreted as the probability that the null hypothesis is true.
15. Missing Data and Analysis Population
The ClinicalTrials.gov record identifies the primary endpoints as per-protocol outcomes but does not provide the detailed rules governing protocol deviations, missing assessments, loss to follow-up, or construction of the PP set.
Those details are important because a per-protocol analysis can be sensitive to which participants are excluded. If exclusion from the PP set is related to treatment adherence, follow-up, or participant characteristics, the resulting analysis population can differ systematically from the randomized population.
16. Multiplicity and the Two Primary Endpoints
Mosaico has two registered primary endpoints. Both address confirmed HIV-1 infection, but they use different follow-up windows: Month 7 up to Month 24 and Month 7 up to Month 30.
When a trial has more than one primary endpoint, the prespecified testing strategy becomes important. A formal confirmatory analysis may require an alpha-allocation or multiplicity procedure so that the overall type I error is controlled across the primary hypotheses. The registry-reported Mosaico data do not specify such a procedure.
| Endpoint | Question addressed | Multiplicity information reported? |
|---|---|---|
| Month 7 to Month 24 | Number of participants with confirmed HIV-1 infection during the first registered primary window | No |
| Month 7 to Month 30 | Number of participants with confirmed HIV-1 infection during the second registered primary window | No |
Accordingly, this page does not assign confirmatory significance to either endpoint or attempt to reconstruct a multiplicity-adjusted p-value.
17. Interim Analysis, Bayesian Methods, Crossover, and Non-Inferiority
The ClinicalTrials.gov record does not identify an interim-analysis procedure, Bayesian method, crossover strategy, or non-inferiority margin. These design topics therefore should not be attributed to Mosaico.
Interim analysis
No interim-analysis method is reported in the ClinicalTrials.gov record. No alpha-spending rule or stopping boundary is reported here.
Bayesian methods
No Bayesian analysis is identified in the ClinicalTrials.gov record.
Crossover
The registry-reported design data do not identify a crossover component. The trial is classified as a parallel design.
Non-inferiority
No non-inferiority margin or non-inferiority hypothesis is reported. The page therefore does not apply non-inferiority logic to the primary endpoints.
18. Why This Trial Matters Statistically
Mosaico provides a useful teaching case because its primary endpoint structure differs from the time-to-event framework often encountered in oncology trials. The registered primary outcomes are binary infection events within defined follow-up windows, and the registry explicitly identifies cumulative incidence as the meaning of the reported data.
| Concept | How it appears in Mosaico |
|---|---|
| Randomization | The study is randomized, providing the core treatment-assignment framework for comparative inference. |
| Parallel design | The trial has 2 parallel study arms. |
| Triple masking | The study is identified as triple-masked. |
| Binary endpoint | Both primary endpoints are classified as binary. |
| Cumulative incidence | The registry describes the primary infection data as cumulative incidence of HIV-1 infections. |
| Per-protocol analysis | Both primary endpoints are defined in the PP set. |
| Multiple primary endpoints | Two primary infection endpoints use different follow-up windows. |
| Denominator discipline | The primary endpoints use a PP population, while serious adverse-event results use specified participants at risk. |
| Safety analysis | Serious adverse events are reported separately for the two groups. |
| Result transparency | Results are posted, but the ClinicalTrials.gov record contains no formal comparative estimates. |
19. Limitations
- No formal statistical-analysis records reported: no formal statistical analyses were posted, so treatment-effect estimates, confidence intervals, and p-values cannot be reported from the ClinicalTrials.gov record.
- Per-protocol primary endpoints: both registered primary endpoints are defined in the PP set. The ClinicalTrials.gov record does not describe how that population was constructed.
- Unknown multiplicity strategy: two primary endpoints are registered, but the ClinicalTrials.gov record does not identify the method used to control type I error across them.
- Unknown missing-data strategy: the ClinicalTrials.gov record does not identify rules for missing assessments, loss to follow-up, or protocol deviations.
- Unknown formal effect measure: the registry data identify the endpoints as binary but do not identify whether the formal analysis used a risk difference, risk ratio, odds ratio, or another comparative statistic.
- Safety denominators differ from total enrollment: the serious-adverse-event data use 1942 and 1945 as participants at risk rather than the overall enrollment figure of 3900.
- Limited endpoint information: the ClinicalTrials.gov record does not include subgroup results, baseline characteristics, duration-of-follow-up summaries beyond the registered endpoint windows, or formal efficacy estimates.
- Results-versus-analysis distinction: the registry indicates that results are posted, but a posted outcome result is not the same as a registry-reported formal statistical-analysis record.
20. How to Read the Mosaico Results Correctly
Start with the estimand
The registered question concerns confirmed HIV-1 infection from Month 7 through a specified follow-up boundary in the PP set.
Then identify the population
The primary endpoint population is the PP set, not automatically all 3900 enrolled participants.
Then identify the effect measure
A risk difference, risk ratio, odds ratio, or another measure answers a different quantitative question. The prespecified analysis should determine which is appropriate.
Finally assess uncertainty
A treatment-effect estimate should be interpreted alongside its confidence interval and prespecified hypothesis-testing framework.
This sequence prevents a common statistical error: jumping directly from the number of infections to a conclusion without first establishing the population, time window, denominator, estimand, and uncertainty framework.
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Statistical Calculators
23. Sources
- ClinicalTrials.gov: Mosaico, NCT03964415.
- Linked publication: PubMed record for PMID 41314741.
Continue through the Clinical Biostats statistical learning pathway
Connect randomized trial design and binary prevention endpoints to deeper tutorials and statistical calculation tools.
24. Record Summary
Mosaico is a completed phase 3 randomized, parallel, triple-masked prevention trial with 3900 enrolled participants and 2 study arms. Its two registered primary endpoints are binary measures of confirmed HIV-1 infection in the per-protocol set, evaluated from Month 7 up to Month 24 and from Month 7 up to Month 30. The registry describes these outcomes as cumulative incidence of HIV-1 infections and reports that results are posted.
The ClinicalTrials.gov record, however, contains no formal statistical analyses. The appropriate statistical framework for these endpoints would ordinarily begin with comparison of binary infection risks over the prespecified intervals, followed by an appropriate effect measure, confidence interval, and hypothesis-testing procedure defined in advance.
The safety record posted on ClinicalTrials.gov for the two groups reports serious adverse events of 82/1942 in the Ad26.Mos4.HIV plus Clade C and Mosaic gp140 group and 77/1945 in the placebo group. These safety denominators are distinct from the overall enrollment and from the unspecified size of the per-protocol efficacy population.