RENOIR: Complete Statistical Analysis of RSVpreF in Lower Respiratory Tract Illness

1. Trial at a Glance

RENOIR is a randomized, parallel-group, triple-masked phase 3 prevention study evaluating RSVpreF in adults with lower respiratory tract illness as the condition of interest. The registry reports 38,861 participants, 9 arms, and RSVpreF and placebo as the biological interventions.

2. Clinical Question

The central statistical question is whether vaccination with RSVpreF reduces the occurrence of a first episode of RSV-associated lower respiratory tract illness during the first RSV season, while the study also evaluates prompted reactions, adverse events, serious adverse events, newly diagnosed chronic medical conditions, and longer-term immunogenicity after revaccination.

3. Trial Design

The 9-arm structure is important when interpreting the registry as a whole. The registry identifies RSVpreF and placebo as the biological interventions, but the information available here does not specify the allocation ratio or provide a participant count for each of the 9 arms. Those quantities should therefore not be inferred from the total enrollment.

4. Endpoints

The registry contains a broad endpoint structure spanning first-season efficacy, solicited local and systemic reactions, adverse events, serious adverse events, newly diagnosed chronic medical conditions, antibody neutralizing titers, and revaccination assessments across SSA, SSB, and SSC.

Main efficacy and safety endpoints

SSA endpoints

SSB endpoints

SSC endpoints: Year 3 and Year 4

5. Statistical Methodology

The registry establishes the randomized, parallel, triple-masked structure and defines the endpoints, but no formal statistical analyses are posted in the registry. Consequently, the appropriate statistical framework is best understood by matching each endpoint to the type of data it generates.

First RSV-associated LRTI: time-to-first-event analysis

The main efficacy endpoint counts the first episode of RSV-associated lower respiratory tract illness from Day 15 after vaccination through the end of season 1 visit. Because participants can have different follow-up histories and because the endpoint is explicitly defined around the first event, a time-to-first-event framework is a natural statistical approach.

For a randomized prevention study, the most important distinction is between the estimand and the statistical model. The scientific target is the effect of RSVpreF on the occurrence of the qualifying first event. A Cox model, for example, would estimate a relative hazard, whereas a cumulative-incidence or fixed-time risk comparison would describe an absolute probability difference or ratio at a specified time.

Proportions of prompted reactions

Local and systemic reactions are binary participant-level outcomes within a fixed seven-day window. The primary descriptive quantity is therefore a proportion: the number of participants reporting at least one qualifying event divided by the relevant number of participants evaluated for that endpoint.

A comparative analysis would ordinarily report the proportions in the randomized groups together with a treatment-group contrast and confidence interval. Depending on the prespecified analysis plan, the contrast could be expressed as a risk difference, risk ratio, odds ratio, or another measure. The registry information does not specify which comparative model or confidence-interval method was selected.

Adverse events and serious adverse events

AE and SAE endpoints are also participant-level proportions, but their observation windows differ substantially. An AE is assessed within 1 month after vaccination or revaccination, whereas SAE endpoints extend throughout much longer periods. This distinction affects both exposure time and the interpretation of event proportions.

For safety, the clinically relevant unit is often the participant rather than the total number of events. A participant experiencing several events is still one participant reporting the binary endpoint "proportion of participants reporting AE." Event-level summaries answer a different question and should not be substituted for participant-level proportions.

Neutralizing titers and geometric means

The immunogenicity endpoints use RSV A and RSV B neutralizing titers expressed as GMTs, GMFRs, geometric mean NT ratios, or GMIRs. These are inherently multiplicative quantities. A logarithmic transformation is therefore a natural statistical framework because differences on the log scale correspond to ratios on the original scale.

A ratio of geometric means of 1 indicates equal geometric means between the comparison groups. A ratio above 1 indicates a higher geometric mean in the numerator group, while a ratio below 1 indicates a lower geometric mean. That interpretation is distinct from an absolute difference in mean titers.

6. Planned Analysis

Primary efficacy endpoint

The main efficacy endpoint is the number of first episodes of RSV-associated lower respiratory tract illness (LRTI-RSV) in the first RSV season, assessed from Day 15 after vaccination until the end of season 1 visit, an average of 6 months.

For this type of endpoint, an appropriate analysis would generally begin by defining the analysis population, the precise time origin, the event definition, censoring rules, and the treatment contrast before examining the data. A time-to-first-event analysis can account for participants who have not experienced LRTI-RSV by the end of observation, while a fixed-time risk analysis can directly describe the cumulative probability of an event over a specified period.

The registry description contains two formulations of the LRTI-RSV symptom threshold: one specifies 2 or more lower respiratory signs/symptoms lasting more than 1 day, while the repeated description specifies 3 or more. Both descriptions also require RT-PCR-confirmed RSV infection within 7 days of ARI symptom onset. This wording difference is important for statistical analysis because changing the event definition changes the set of participants classified as having the endpoint.

Solicited local and systemic reactions

The prompted local and systemic reaction endpoints are binary outcomes collected within 7 days after vaccination or revaccination. The natural primary summary is the proportion of participants reporting at least one qualifying reaction, with severity categories described separately.

For redness and swelling, the registry defines 1 measuring device unit as 0.5 cm and classifies redness and swelling as mild at 2.5 to 5.0 cm, moderate at >5.0 to 10.0 cm, and severe at >10 cm. Pain is classified according to interference with daily activity. These ordinal severity definitions should be retained when analysing severity rather than collapsing all grades into a single continuous measurement.

AE, SAE, and NDCMC endpoints

The AE endpoint is evaluated within 1 month after vaccination or revaccination, up to 35 days. SAE and NDCMC endpoints extend over substantially longer observation periods. The appropriate analysis therefore needs to preserve the stated observation window rather than treating all safety endpoints as though they had the same follow-up.

Immunogenicity endpoints

SSA and SSB measure RSV A and RSV B neutralizing titers before revaccination and at 1, 6, 12, and 18 months after revaccination. The registry specifies GMTs and GMFRs. SSC extends the immunogenicity assessment to geometric mean NT ratios and GMIRs following revaccination in Year 3 and Year 4.

For positive titer measurements, log transformation is generally useful because it stabilizes multiplicative variation and allows comparisons to be expressed as ratios after back-transformation. A repeated-measures framework can also be appropriate when the same participants contribute measurements at multiple time points, provided the analysis plan specifies how within-participant correlation and missing visits are handled.

7. Statistical Methods Explained

Why is the first LRTI-RSV episode a time-to-event endpoint?

The endpoint is defined as the first episode during a specified observation period. Participants can therefore differ in whether and when they experience the event. Time-to-event methods preserve the timing information and can accommodate participants whose follow-up ends without a qualifying episode.

What does a hazard ratio mean if one is used?

A hazard ratio compares estimated instantaneous event rates between groups over the modelled follow-up. An HR below 1 would indicate a lower estimated event hazard in the RSVpreF group relative to the comparator under the fitted model. It would not mean that the same percentage of every participant's individual risk was reduced by that amount.

A Cox model also requires attention to the proportional-hazards assumption if a single hazard ratio is used as the summary treatment effect. If hazards vary substantially in relative terms over time, a single HR may obscure clinically relevant changes in the treatment effect.

Why are prompted reactions analysed as proportions?

The registry defines these endpoints as the proportion of participants reporting a reaction within 7 days. This makes the participant-level binary outcome the natural starting point: each participant either does or does not meet the specified endpoint during the window.

The percentage alone, however, does not quantify uncertainty. A confidence interval around the proportion or around a treatment-group contrast shows how precisely the observed rate has been estimated.

Why use geometric means for neutralizing titers?

Neutralizing titers are expressed on a multiplicative scale. A participant with a titer twice another participant's titer represents a ratio, not a fixed additive increment. The geometric mean therefore describes the central tendency more naturally than the arithmetic mean for many such distributions.

What does a geometric mean ratio mean?

A geometric mean ratio compares two geometric means multiplicatively. A ratio of 1 represents equality. A ratio of 1.5 means the numerator group's geometric mean is 50% higher, while a ratio of 0.67 means it is approximately one-third lower. The ratio should not be interpreted as a percentage of participants responding unless the endpoint itself is defined that way.

Why do the Year 3 and Year 4 immunogenicity endpoints need separate interpretation?

The registry defines distinct SSC endpoints for Year 3 and Year 4, with separate revaccination time frames and different comparison histories. Combining them into a single longitudinal estimate would require an explicit statistical model and prespecified handling of repeated observations, treatment history, and missing measurements.

Why is the LRTI-RSV definition itself statistically important?

Endpoint definitions determine who is counted as an event. The registry contains both a 2-or-more and a 3-or-more lower-respiratory-signs/symptoms formulation in its description. That distinction is not cosmetic: it can change the numerator of the efficacy endpoint and therefore alter the estimated treatment effect.

8. Immunogenicity Analysis

The immunogenicity component is structurally different from the main efficacy endpoint. Rather than counting clinical illness episodes, it measures RSV A and RSV B neutralizing titers and expresses them through geometric summaries and ratios.

The registry defines a neutralizing titer as the interpolated reciprocal of the serum dilution resulting in a 50% reduction in the number of viral focus forming units compared with the control without test serum. This definition gives the assay result a quantitative scale suitable for geometric statistical summaries.

GMT versus GMFR

A geometric mean titer describes the central tendency of measured neutralizing titers at a particular assessment. A geometric mean fold rise instead describes multiplicative change relative to an earlier measurement. They answer different questions: GMT describes the level, while GMFR describes change.

9. Safety Analysis

Safety is evaluated through prompted local reactions, prompted systemic events, AEs, SAEs, and NDCMCs. The registry applies these measures over different windows following initial vaccination and revaccination.

The safety endpoint definitions also illustrate why "adverse event" and "serious adverse event" are not interchangeable. The registry defines an AE broadly as any untoward medical occurrence after receipt of investigational product without regard to causal relationship. An SAE is an AE associated with specified serious outcomes or deemed significant for another reason.

NDCMC is different again: the registry defines it as a disease or medical condition not previously identified that is expected to be persistent or otherwise long-lasting in its effects. Statistical interpretation should therefore respect the distinct clinical definitions rather than treating all three categories as different severity levels of the same endpoint.

10. Understanding the Main Efficacy Estimand

11. Interpreting the Immunogenicity Measures

Neutralizing titers require a different interpretive vocabulary from clinical-event endpoints. The principal quantities in RENOIR are GMT, GMFR, geometric mean NT ratio, and GMIR.

An important statistical distinction is that an immunogenicity ratio is a laboratory measurement comparison, whereas the first-season LRTI-RSV endpoint is a clinical event outcome. A change in one should not automatically be translated into a numerical change in the other without an explicit validated relationship.

12. Longitudinal Structure

The registry's endpoint architecture extends well beyond the first RSV season. SSA and SSB include immunogenicity and safety assessments after revaccination, while SSC includes Year 3 and Year 4 immunogenicity and safety endpoints.

This longitudinal structure creates a fundamental distinction between the main first-season efficacy question and later immunogenicity or safety questions. They use different outcome scales, different observation windows, and potentially different analysis populations.

13. Missing Data and Follow-Up

Longitudinal prevention studies require explicit handling of participants who do not complete every assessment. For the first-event efficacy endpoint, censoring is particularly important because participants who remain event-free contribute information until the point at which follow-up ends or the endpoint observation period closes.

For repeated neutralizing-titer measurements, missingness can occur at individual time points even when a participant remains in the study. An appropriate longitudinal analysis would need to specify whether available observations are analysed directly, whether repeated-measures modelling is used, and what assumptions govern the relationship between missingness and observed outcomes.

14. Multiplicity and Multiple Endpoints

RENOIR has a large endpoint structure. The registry includes a main efficacy endpoint plus prompted reaction, AE, SAE, and NDCMC endpoints, multiple immunogenicity measurements, and separate Year 3 and Year 4 assessments.

When a study evaluates many endpoints, statistical interpretation must distinguish between descriptive measurement and confirmatory hypothesis testing. If several formal hypotheses are tested, the overall probability of at least one false-positive conclusion can increase unless multiplicity is addressed through a prespecified strategy.

Without a posted statistical analysis plan specifying the multiplicity strategy, it is not appropriate to assign confirmatory significance to individual endpoints merely because a nominal p-value might later be available.

15. Randomization and Masking

Randomization is the central design mechanism supporting a causal comparison between RSVpreF and placebo. By assigning participants to study groups rather than allowing treatment choice to determine assignment, randomization helps balance measured and unmeasured prognostic factors in expectation.

The registry identifies the allocation as RANDOMIZED, the design model as PARALLEL, and the masking as TRIPLE. Triple masking is particularly relevant in a study containing subjective or participant-recorded outcomes such as pain, fatigue, headache, and other prompted reactions because knowledge of treatment assignment can potentially influence reporting or assessment.

The 9-arm design means that the complete statistical structure is more complex than a simple two-arm trial. The registry information available here identifies the interventions but does not provide the allocation ratio or sample size for each arm. Those details are necessary before calculating arm-specific precision or reconstructing comparative power.

16. Why Event Definitions Matter

Clinical trial statistics begin with the endpoint definition. A sophisticated statistical model cannot repair ambiguity in the underlying outcome classification.

The registry's repeated description uses both a threshold of 2 or more and a threshold of 3 or more lower respiratory signs/symptoms. For a formal analysis, the prespecified final endpoint definition would need to resolve which threshold governs the primary analysis.

17. Clinical Interpretation vs Statistical Interpretation

18. Important Limitations and Interpretation Issues

• No posted statistical analyses: the registry contains endpoint definitions but no posted effect estimates, confidence intervals, or p-values for these endpoints.
• Endpoint-definition discrepancy: the main LRTI-RSV description contains both a 2-or-more and a 3-or-more lower-respiratory-signs/symptoms formulation.
• Arm-level information: the registry identifies 9 arms but does not specify the participant count or allocation ratio for each arm in the information available here.
• Multiplicity: the endpoint structure contains numerous clinical, safety, and immunogenicity outcomes, so confirmatory interpretation requires a prespecified multiplicity framework.
• Different follow-up periods: seven-day reaction endpoints, 35-day AE endpoints, approximately 18- to 30-month safety endpoints, and Year 3 and Year 4 endpoints should not be treated as if they represented the same observation period.
• Repeated measurements: immunogenicity assessments at multiple time points create within-participant correlation that should be accounted for in a formal longitudinal analysis.
• Censoring: a first-event analysis requires explicit rules for participants who do not experience an event during observation or whose follow-up ends early.
• Hazard-ratio assumptions: if a Cox model is used for the efficacy endpoint, proportional hazards should be assessed rather than assumed automatically.
• Immunogenicity versus clinical efficacy: higher neutralizing titers or geometric ratios are laboratory outcomes and should not be treated as direct substitutes for clinical event reduction.
• Study status: the registry lists the study as active, not recruiting, with primary completion scheduled for 2027-01-18.

19. Why This Trial Matters Statistically

RENOIR is a useful statistical teaching case because it combines several different endpoint architectures within a single randomized prevention study. The main efficacy question is event-based, while much of the safety analysis is binary and the immunogenicity program is longitudinal and multiplicative.

20. Statistical Concepts in This Trial

The main statistical lesson from RENOIR is that different endpoint types require different estimands and analytical tools. A first RSV-associated LRTI episode is not analysed in the same way as a seven-day solicited reaction, and neither is equivalent to a neutralizing-titer ratio.

21. Sources

• ClinicalTrials.gov: RENOIR, NCT05035212.
• PubMed: PubMed record 39928572.
• PubMed: PubMed record 37018468.

22. Record Summary

RENOIR is a phase 3 randomized, parallel, triple-masked prevention study of RSVpreF in adults with lower respiratory tract illness. The registry reports an enrollment of 38,861 participants and 9 arms, with RSVpreF and placebo identified as the biological interventions.

Statistically, the study is notable for its combination of a first-event clinical efficacy endpoint, fixed-window participant-level safety outcomes, and longitudinal immunogenicity measurements. The main LRTI-RSV endpoint is assessed from Day 15 after vaccination through the end of season 1 visit, an average of 6 months, and requires RT-PCR-confirmed RSV A and/or RSV B-associated illness together with the specified respiratory-symptom criteria.

The immunogenicity program uses GMTs, GMFRs, geometric mean NT ratios, and GMIRs across SSA, SSB, and SSC assessments. These measures require multiplicative rather than purely additive interpretation and are statistically distinct from clinical-event efficacy measures.

The registry also defines safety outcomes across markedly different observation periods, including prompted reactions within 7 days, AEs within 1 month or up to 35 days, and longer-term SAE and NDCMC assessments. SSC extends the study's revaccination assessments through Year 3 and Year 4.