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
CLEVER was a randomized, double-blind, parallel-group phase 2/3 trial evaluating clesrovimab for prevention of respiratory syncytial virus infection in infants. The registry reports 3632 enrolled participants, two study arms, eight registered primary endpoints, and posted statistical analyses for six primary endpoints.
| Feature | CLEVER |
|---|---|
| Trial name | CLEVER |
| Brief title | Efficacy and Safety of Clesrovimab (MK-1654) in Infants (MK-1654-004) |
| Phase | 2/3 |
| Status | Completed |
| Therapeutic area | Infectious Disease |
| Condition | Respiratory Syncytial Virus Infection |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Prevention |
| Enrollment | 3632 |
| Lead sponsor | Merck Sharp & Dohme LLC |
2. Clinical Question
The central clinical question was whether clesrovimab could reduce respiratory syncytial virus-associated medically attended lower respiratory infection while maintaining an acceptable safety profile compared with placebo in infants.
Population
Infants enrolled in a randomized prevention trial for respiratory syncytial virus infection.
Intervention
Clesrovimab 105 mg.
Comparator
Placebo.
Primary question
Does clesrovimab reduce RSV-associated medically attended lower respiratory infection and how do prespecified safety outcomes compare between groups?
3. Trial Design
Clesrovimab 105 mg
- Biological intervention
- Evaluated against placebo
- Primary efficacy comparison for RSV Season 1
- Safety outcomes assessed across prespecified postdose windows
Placebo
- Placebo comparator
- Used as the reference group for efficacy estimates
- Included in the same prespecified safety assessments
- Serious adverse events reported by RSV season
4. Timeline
Trial start
The registry records April 7, 2021 as the study start date.
Primary completion
The registry records July 9, 2024 as the primary completion date.
Completed
The ClinicalTrials.gov record lists the study status as completed and reports results.
5. Endpoints
The registry lists eight primary endpoints. Most are binary outcomes defined over specific postdose windows. The primary efficacy endpoint focuses on RSV-associated medically attended lower respiratory infection; the other primary endpoints are safety outcomes.
| Primary endpoint | Time frame | Type |
|---|---|---|
| RSV Season 1: Number of Participants With RSV-Associated Medically Attended Lower Respiratory Infection (MALRI) | Day 1 (postdose) to Day 150 | Binary |
| RSV Season 1: Percentage of Participants With Solicited Injection-Site Adverse Events | Day 1 (postdose) to Day 5 | Binary |
| RSV Season 1: Percentage of Participants With Fever | Day 1 (postdose) to Day 5 | Binary |
| RSV Season 1: Percentage of Participants With Solicited Systemic AEs | Day 1 (postdose) to Day 5 | Binary |
| RSV Season 1: Percentage of Participants With Anaphylaxis/Hypersensitivity AE of Special Interest (AESI) | Day 1 (postdose) to Day 42 | Binary |
| RSV Season 1: Percentage of Participants With Rash AESI | Day 1 (postdose) to Day 42 | Binary |
| RSV Season 1: Percentage of Participants With ≥1 Nonserious AE | Day 1 (postdose) to Day 42 | Binary |
| Percentage of Participants With Serious Adverse Events (SAEs) | RSV Season 1: Day 1 to Day 365; RSV Season 2: Day 366 to Day 515 | Binary |
How MALRI was defined
Outpatient and inpatient MALRI was defined by the presence of cough or difficulty breathing, together with at least one of wheezing, chest wall in-drawing or retraction, rales or crackles, hypoxemia, tachypnea, or dehydration due to respiratory symptoms, together with an RSV-positive reverse transcriptase polymerase chain reaction nasopharyngeal sample.
For RSV Season 1, the protocol reported outpatient and inpatient RSV-associated MALRI together. This is important statistically because the endpoint represents a prespecified composite clinical category rather than a laboratory result alone.
Safety endpoint definitions
An adverse event was defined as any untoward medical occurrence in a clinical study participant temporally associated with study intervention, whether or not considered related to the intervention.
Solicited injection-site adverse events included redness or erythema, swelling, and pain or tenderness. Solicited systemic adverse events included irritability, somnolence or drowsiness, and appetite lost.
Fever was defined as a rectal temperature ≥102.2°F (≥39.0°C) or axillary temperature ≥101.7°F (≥38.7°C).
The anaphylaxis/hypersensitivity AESI included anaphylaxis, angioedema, bronchospasm, drug hypersensitivity, dyspnea, hypersensitivity, dysphonia, and wheezing. The rash AESI included acute generalized exanthematous pustulosis, drug eruption, drug reaction with eosinophilia and systemic symptoms, erythema multiforme, generalized rash of exfoliative nature, Stevens-Johnson syndrome, toxic epidermal necrolysis, and urticaria.
6. Statistical Methodology
Binary endpoint analysis
The primary efficacy endpoint and the reported secondary efficacy endpoints are binary outcomes: participants either experienced the specified event during the defined time window or did not. The registry reports exact methods for some binary comparisons and modified Poisson regression with robust variance for efficacy estimation.
The reported efficacy estimate therefore expresses the estimated relative reduction in risk associated with clesrovimab compared with placebo. A value of 60.4 corresponds to an estimated relative risk of 0.396 when interpreted through the registry's stated formula.
Modified Poisson regression with robust variance
For the primary MALRI analysis, the registry states that a modified Poisson regression with robust variance was used to generate the efficacy estimate and 95% confidence interval. The model included region, gestational age, and age at randomization as covariates.
The same covariate-adjusted modified Poisson approach was reported for the secondary endpoints of RSV-associated hospitalization and RSV-associated MALRI. This approach directly targets a relative-risk framework for binary outcomes rather than using an odds ratio as the primary effect measure.
Exact analysis
The primary MALRI analysis was reported with an exact method, and the registry normalizes this as an exact or Clopper-Pearson binomial approach. The analysis notes specify that the one-sided p-value was estimated using an exact method.
Score-based confidence intervals for safety proportions
For the reported primary safety endpoints, the registry identifies the Miettinen-Nurminen method for estimating the difference between treatment-group percentages and its 95% confidence interval. This is a score-based approach for comparing two proportions.
| Endpoint family | Reported method | Primary effect measure |
|---|---|---|
| RSV-associated MALRI | Exact method plus modified Poisson regression with robust variance | Efficacy estimate, defined as 1 − RR |
| RSV-associated hospitalization | Exact method plus modified Poisson regression with robust variance | Efficacy estimate, defined as 1 − RR |
| RSV-associated MALRI through Day 180 | Modified Poisson regression with robust variance | Efficacy estimate, defined as 1 − RR |
| Solicited injection-site AEs | Miettinen-Nurminen | Estimated percentage difference |
| Fever | Miettinen-Nurminen | Estimated percentage difference |
| Solicited systemic AEs | Miettinen-Nurminen | Estimated percentage difference |
| Anaphylaxis/hypersensitivity AESI | Miettinen-Nurminen | Estimated percentage difference |
| Rash AESI | Miettinen-Nurminen | Estimated percentage difference |
7. Results
The registry reports formal statistical analyses for six primary endpoints and two secondary efficacy endpoints. The reported primary results below use the registry's stated estimates, confidence intervals, p-values, analysis populations, and time frames.
Primary efficacy endpoint: RSV-associated MALRI
RSV Season 1 MALRI efficacy estimate
95% CI: 44.1–71.9 · P < 0.001
Time frame: Day 1 (postdose) to Day 150
The analysis population consisted of all randomized participants who received a dose of study treatment and were protocol compliant. The comparison was clesrovimab 105 mg versus placebo in RSV Season 1.
The reported efficacy estimate of 60.4 is defined in the registry analysis as 1 − relative risk. In that framework, it represents an estimated relative reduction in the risk of RSV-associated MALRI during the specified Day 1 to Day 150 period.
It does not mean that 60.4% of infants were protected, that 60.4% of participants avoided every form of RSV illness, or that the absolute risk difference between groups was 60.4 percentage points. The estimate is a relative measure derived from the treatment-group risks.
The 95% confidence interval of 44.1–71.9 describes uncertainty around the estimated efficacy under the analysis framework. The interval is about the population-level effect estimate; it is not a prediction interval for an individual infant.
The P < 0.001 value addresses the prespecified statistical evidence against the relevant null hypothesis under the reported testing framework. A p-value does not measure the size of the treatment effect, its clinical importance, or the probability that the treatment works.
The analysis also used covariate adjustment for region, gestational age, and age at randomization. The reported one-sided p-value and two-sided 95% confidence interval should therefore be read together rather than treating the p-value and interval as interchangeable summaries.
Primary safety endpoint: Solicited injection-site adverse events
Estimated percentage difference
95% CI: −2.6–1.8
Time frame: Day 1 (postdose) to Day 5
The safety analysis population included randomized participants who received a dose of protocol-specified study treatment. Per protocol, participants with protocol deviation were excluded from the safety analysis population. The reported comparison was based on the Miettinen-Nurminen method.
The estimated percentage difference of −0.4 represents the estimated difference in the percentage of participants with solicited injection-site adverse events between the clesrovimab 105 mg and placebo groups, using the registry's treatment-group comparison.
The estimate is close to zero, but the confidence interval of −2.6 to 1.8 spans zero. The interval therefore includes both a modest lower percentage and a modest higher percentage in the clesrovimab group.
This is a precision statement, not a proof that the two groups are identical. The width of the interval reflects the uncertainty around the difference. No p-value is reported for this endpoint in the statistical analysis record.
Primary safety endpoint: Fever
Estimated percentage difference
95% CI: −1.4–−0.0
Time frame: Day 1 (postdose) to Day 5
The analysis population included randomized participants who received a dose of protocol-specified study treatment and had temperature data available. Participants with protocol deviation were excluded according to the protocol. The Miettinen-Nurminen method was used for the estimated percentage difference and 95% confidence interval.
The estimated percentage difference of −0.6 indicates that the observed percentage of participants meeting the registry definition of fever was estimated to be lower in the clesrovimab group than in the placebo group by 0.6 percentage points under this analysis.
The 95% confidence interval extends from −1.4 to −0.0. Because the upper endpoint is reported as −0.0, the result should be read exactly as reported rather than converting it into a more precise value.
This estimate concerns the prespecified fever definition and the Day 1 to Day 5 period. It does not describe fever outside that window or establish an effect on broader clinical outcomes.
Primary safety endpoint: Solicited systemic adverse events
Estimated percentage difference
95% CI: −4.3–2.1
Time frame: Day 1 (postdose) to Day 5
The estimated percentage difference of −1.1 describes the difference between the clesrovimab and placebo groups for solicited systemic adverse events during the specified five-day period.
The confidence interval of −4.3 to 2.1 includes zero and therefore is compatible with a range of possible percentage differences in either direction. The estimate itself is not a probability and should not be interpreted as the probability of an adverse event.
The endpoint includes irritability, somnolence or drowsiness, and appetite lost. The analysis therefore summarizes a prespecified collection of solicited symptoms rather than all adverse events occurring after dosing.
Primary safety endpoint: Anaphylaxis/hypersensitivity AESI
Estimated percentage difference
95% CI: −0.3–0.2
Time frame: Day 1 (postdose) to Day 42
The estimated percentage difference is 0.0, with a 95% confidence interval of −0.3 to 0.2. The point estimate is exactly zero, while the interval shows that the underlying population-level percentage difference remains uncertain within the range supported by the analysis.
The endpoint covers a prespecified set of anaphylaxis and hypersensitivity events and is assessed over Day 1 through Day 42. It should therefore not be generalized to every possible adverse event or to periods outside the registered time frame.
Primary safety endpoint: Rash AESI
Estimated percentage difference
95% CI: −0.4–0.5
Time frame: Day 1 (postdose) to Day 42
The estimated percentage difference of 0.1 is close to zero. The 95% confidence interval of −0.4 to 0.5 includes zero and therefore encompasses both a small lower and a small higher percentage in the clesrovimab group relative to placebo.
As with the other safety comparisons, the confidence interval describes uncertainty around the between-group percentage difference. It does not establish equivalence, because an interval containing zero is not by itself an equivalence test.
Primary endpoints without a posted formal statistical analysis
The registry lists two additional primary endpoints: RSV Season 1 percentage of participants with at least one nonserious adverse event from Day 1 to Day 42, and percentage of participants with serious adverse events across RSV Season 1 and RSV Season 2. The ClinicalTrials.gov record reports these outcomes as primary endpoints but does not post a formal statistical analysis for them.
8. Secondary Endpoint Results
RSV-associated hospitalization
Efficacy estimate
95% CI: 66.6–92.6 · P < 0.001
Time frame: Day 1 (postdose) to Day 150
The analysis population included all randomized participants who received a dose of study treatment and were protocol compliant. A modified Poisson regression with robust variance generated the efficacy estimate and 95% confidence interval, with region, gestational age, and age at randomization included as covariates. The registry also identifies an exact method for this analysis.
The reported 84.2 efficacy estimate represents 1 − relative risk under the registry's stated framework. It therefore describes an estimated relative reduction in RSV-associated hospitalization over Day 1 to Day 150, rather than an absolute reduction in the percentage of hospitalized participants.
The 95% confidence interval of 66.6–92.6 indicates that the estimated relative effect is subject to statistical uncertainty, while remaining within a range of substantial relative risk reduction under the model used.
The P < 0.001 value provides evidence against the relevant null hypothesis under the reported statistical testing framework. It does not quantify the magnitude of benefit or the probability that an individual participant will avoid hospitalization.
RSV-associated MALRI through Day 180
Efficacy estimate
95% CI: 43.3–71.1
Time frame: Day 1 (postdose) to Day 180
This secondary endpoint used a modified Poisson regression with robust variance. The analysis population consisted of randomized participants who received a dose of study treatment and were protocol compliant. Region, gestational age, and age at randomization were included as covariates.
The efficacy estimate of 59.5 represents the registry's 1 − relative risk measure for RSV-associated MALRI through Day 180. The longer time frame differs from the primary MALRI endpoint, which was assessed through Day 150.
The 95% confidence interval of 43.3–71.1 describes uncertainty around the relative efficacy estimate. Because the time frame and endpoint definition must be considered together, this estimate should not be treated as interchangeable with the primary Day 150 result.
No p-value is reported for this secondary analysis in the statistical analysis record. A confidence interval and a p-value provide related but distinct information; neither alone describes the complete clinical meaning of an endpoint.
| Secondary endpoint | Estimate | 95% CI | P-value |
|---|---|---|---|
| RSV-associated hospitalization, Day 1 to Day 150 | 84.2 | 66.6–92.6 | <0.001 |
| RSV-associated MALRI, Day 1 to Day 180 | 59.5 | 43.3–71.1 | Not reported |
9. Safety Results
The registry provides serious adverse-event counts by treatment group and RSV season. These are reported as affected participants over the corresponding participants at risk.
| RSV season | Treatment group | Participants with serious AEs | Participants at risk |
|---|---|---|---|
| RSV Season 1 | Clesrovimab 105 mg | 282 | 2409 |
| RSV Season 1 | Placebo | 151 | 1202 |
| RSV Season 2 | Clesrovimab 105 mg | 32 | 1017 |
| RSV Season 2 | Placebo | 20 | 501 |
The registry defines an SAE as an untoward medical occurrence resulting in death, being life-threatening, requiring inpatient hospitalization or prolonging an existing hospitalization, resulting in persistent or significant disability or incapacity, constituting a congenital anomaly or birth defect, or representing another important medical event.
10. Analysis Populations
The registry specifies different analysis populations according to endpoint purpose. This distinction is statistically important because the population included in an efficacy analysis is not necessarily the same population used for a safety endpoint.
| Analysis population | Definition in the registry | Endpoints using it |
|---|---|---|
| Efficacy population | All randomized participants who received a dose of study treatment and were protocol compliant. | Primary MALRI; RSV-associated hospitalization; RSV-associated MALRI through Day 180 |
| Safety population | All randomized participants who received a dose of protocol-specified study treatment. For the reported safety analyses, participants with protocol deviation were excluded. | Solicited injection-site AEs; fever; solicited systemic AEs; anaphylaxis/hypersensitivity AESI; rash AESI |
| Safety population with temperature data | Randomized participants who received a dose of protocol-specified study treatment and had temperature data available, with protocol deviations excluded. | Fever |
Analysis-population definitions matter because treatment-effect estimates describe the participants actually included in the calculation. In randomized trials, excluding participants after randomization can alter the balance created by randomization, which is one reason the exact analysis population should always accompany an efficacy estimate.
11. Statistical Methods Explained
Why use a modified Poisson regression?
The key efficacy outcomes are binary: a participant either experienced the specified RSV-associated event during the defined time window or did not. The modified Poisson model with robust variance provides a direct framework for estimating relative risk rather than relying on an odds ratio. In CLEVER, the registry states that the model generated the efficacy estimate as 1 − RR and adjusted for region, gestational age, and age at randomization.
What does an efficacy estimate of 60.4 mean?
The registry defines the efficacy estimate as 1 − relative risk. Thus, 60.4 represents an estimated relative reduction in risk of the specified endpoint during the stated period. It is not a percentage of individuals who necessarily benefit and it is not an absolute percentage-point difference.
Why was an exact method used?
Binary clinical-trial outcomes can be analyzed using exact methods when the observed event information is represented directly through binomial probabilities. The registry identifies an exact method for the primary MALRI endpoint and reports a one-sided p-value from that exact method. Exact methods can be particularly useful when event counts are limited because they do not rely on the same large-sample approximations as ordinary asymptotic procedures.
What does a Miettinen-Nurminen confidence interval estimate?
For the reported safety endpoints, the Miettinen-Nurminen method was used to estimate the difference between treatment-group percentages and construct a 95% confidence interval. The resulting interval describes uncertainty around the between-group percentage difference. An interval that includes zero is compatible with no difference, but it does not by itself establish equivalence.
Why does the analysis adjust for region, gestational age, and age at randomization?
The modified Poisson models for the reported efficacy analyses included these covariates. Covariate adjustment can account for prespecified participant or design characteristics and can improve precision when those characteristics are associated with outcome risk. The adjusted estimate should therefore be interpreted as the model-based comparison after accounting for the listed covariates.
Why is the analysis population stated for every endpoint?
The analysis population determines which participants contribute to the estimate. CLEVER uses a protocol-compliant efficacy population for the reported efficacy analyses and a safety population for the reported safety comparisons. Those populations answer somewhat different statistical questions and should not be silently substituted for one another.
12. Confidence Intervals and P-values
The CLEVER registry illustrates why a treatment effect should be presented with both its estimate and its uncertainty. The primary MALRI efficacy estimate is 60.4 with a 95% confidence interval of 44.1–71.9. The hospitalization efficacy estimate is 84.2 with a 95% confidence interval of 66.6–92.6.
Confidence interval
A confidence interval describes statistical uncertainty around the estimated population effect under the specified analysis framework. A narrower interval indicates greater precision than a wider interval, all else equal.
P-value
A p-value measures compatibility between the observed data and a specified null hypothesis under the statistical model and testing procedure. It does not measure effect size or clinical importance.
The distinction is especially important for the safety endpoints. For example, the estimated percentage difference for solicited systemic adverse events is −1.1, with a 95% confidence interval of −4.3–2.1. The interval includes zero, so the reported estimate is compatible with both a modest decrease and a modest increase in the percentage experiencing the endpoint.
13. One-Sided and Two-Sided Inference
The primary MALRI analysis is unusual in that the registry reports a one-sided p-value while the efficacy estimate is accompanied by a two-sided 95% confidence interval. These are different inferential quantities and should be interpreted according to their stated sides.
A one-sided test allocates the relevant type I error to one direction of the treatment effect. A two-sided confidence interval allows uncertainty in both directions around the estimate. The presence of a one-sided p-value therefore does not turn the reported confidence interval into a one-sided interval.
14. Primary Endpoint Structure
CLEVER has eight registered primary endpoints, spanning efficacy and safety. Six have formal statistical analyses posted in the registry; two are listed as primary outcomes without a corresponding formal statistical analysis in the statistical-analysis record.
| Primary endpoint | Formal analysis posted | Reported inferential result |
|---|---|---|
| RSV-associated MALRI, Day 1–150 | Yes | 60.4; 95% CI 44.1–71.9; P < 0.001 |
| Solicited injection-site AEs, Day 1–5 | Yes | −0.4; 95% CI −2.6–1.8 |
| Fever, Day 1–5 | Yes | −0.6; 95% CI −1.4–−0.0 |
| Solicited systemic AEs, Day 1–5 | Yes | −1.1; 95% CI −4.3–2.1 |
| Anaphylaxis/hypersensitivity AESI, Day 1–42 | Yes | 0.0; 95% CI −0.3–0.2 |
| Rash AESI, Day 1–42 | Yes | 0.1; 95% CI −0.4–0.5 |
| ≥1 nonserious AE, Day 1–42 | No formal analysis posted | Registry reports endpoint without formal statistical analysis |
| Serious adverse events, Season 1 and Season 2 | No formal analysis posted | Registry reports endpoint and affected/at-risk counts |
This structure matters because the eight endpoints do not all answer the same question. The MALRI endpoint addresses prevention of clinically attended RSV-associated lower respiratory infection, whereas the other primary endpoints address prespecified tolerability and safety outcomes over shorter or longer observation windows.
15. Multiplicity and Multiple Primary Endpoints
Multiple primary endpoints create an important statistical-design issue: several hypothesis tests or estimands may contribute to the overall interpretation of a trial. The registry identifies eight primary endpoints, but the statistical-analysis record does not state a single unified multiplicity-adjustment procedure across those endpoints.
The reported p-values and confidence intervals should therefore be interpreted in the context of their individual analysis records and the prespecified trial framework. A small p-value for one endpoint does not, by itself, describe the evidence across all primary endpoints.
16. Safety Analysis and Clinical Interpretation
The safety endpoints use short, prespecified observation windows for solicited events and longer windows for serious adverse events. This separation is useful because adverse-event ascertainment is inherently time-dependent: an event observed from Day 1 to Day 5 answers a different question from an SAE observed across RSV Season 1 and RSV Season 2.
| Safety domain | Observation window | Statistical approach |
|---|---|---|
| Injection-site adverse events | Day 1–5 | Miettinen-Nurminen percentage difference |
| Fever | Day 1–5 | Miettinen-Nurminen percentage difference |
| Solicited systemic adverse events | Day 1–5 | Miettinen-Nurminen percentage difference |
| Anaphylaxis/hypersensitivity AESI | Day 1–42 | Miettinen-Nurminen percentage difference |
| Rash AESI | Day 1–42 | Miettinen-Nurminen percentage difference |
| ≥1 nonserious AE | Day 1–42 | Endpoint reported without formal posted analysis |
| Serious adverse events | Day 1–365 and Day 366–515 | Counts reported by RSV season and treatment group |
The reported confidence intervals for the five formally analyzed safety endpoints are centered near zero. That pattern describes the estimated treatment-group differences for those particular endpoints and windows; it should not be generalized to every possible adverse event.
17. What the Efficacy Estimate Does — and Does Not — Mean
The primary MALRI efficacy estimate of 60.4 is based on the registry's definition 1 − RR. In practical statistical terms, it represents an estimated relative reduction in the risk of the specified RSV-associated MALRI endpoint during Day 1 through Day 150.
It does not mean that 60.4% of all participants avoided RSV infection, that 60.4% of participants were protected from every RSV-related outcome, or that the absolute risk fell by 60.4 percentage points.
The 95% CI of 44.1–71.9 describes uncertainty around the estimated efficacy. It gives a more complete picture than the point estimate alone because it communicates the precision of the treatment-effect estimate under the stated analysis.
The primary MALRI analysis has P < 0.001. This indicates strong statistical evidence against the relevant null hypothesis under the reported one-sided testing framework, but the magnitude of the treatment effect is conveyed by the efficacy estimate and its confidence interval, not by the p-value.
18. Comparing the Day 150 and Day 180 MALRI Analyses
The registry contains two closely related efficacy analyses of RSV-associated MALRI with different time frames. The primary endpoint covers Day 1 through Day 150, while the secondary endpoint covers Day 1 through Day 180.
| Feature | Primary MALRI analysis | Secondary MALRI analysis |
|---|---|---|
| Time frame | Day 1 to Day 150 | Day 1 to Day 180 |
| Analysis population | Randomized, dosed, protocol-compliant | Randomized, dosed, protocol-compliant |
| Effect measure | Efficacy estimate = 1 − RR | Efficacy estimate = 1 − RR |
| Estimate | 60.4 | 59.5 |
| 95% CI | 44.1–71.9 | 43.3–71.1 |
| P-value | <0.001 | Not reported |
The estimates are numerically close, but they should not be treated as duplicate measurements. The endpoint windows differ, and the secondary analysis has a different statistical-analysis record. Time frame is part of the definition of a clinical endpoint and therefore part of the interpretation of its effect estimate.
19. Covariate Adjustment
The reported modified Poisson analyses included region, gestational age, and age at randomization. Covariate adjustment changes the statistical model used to estimate the treatment effect while retaining the randomized treatment comparison.
The registry specifies these covariates for the modified Poisson regression used to generate the efficacy estimate and confidence interval.
Adjustment is not equivalent to changing the treatment assignment. The randomized groups remain the treatment comparison of interest; the covariates are incorporated into the model to account for specified participant or design characteristics.
20. Exact Methods and Rare Safety Events
Several safety endpoints concern specific adverse-event categories that may occur relatively infrequently. The registry uses a score-based method for these percentage differences and an exact method for the primary efficacy analysis.
For binary data, the choice of confidence-interval method can materially affect interval coverage, particularly when event probabilities are low or sample sizes are unbalanced. The Miettinen-Nurminen approach used here is specifically designed for confidence intervals for differences in two binomial proportions.
Importantly, the absence of a large numerical difference does not mean that the underlying rates are known with unlimited precision. The confidence interval remains essential to understanding what the observed data can support.
21. Limitations
- Multiple primary endpoints: eight endpoints are registered as primary, spanning efficacy and safety. The statistical-analysis record does not state one overall multiplicity procedure across all eight.
- Different analysis populations: efficacy analyses use randomized, dosed, protocol-compliant participants, whereas safety analyses use specified safety populations. These populations should not be conflated.
- Different observation windows: endpoints range from Day 1–5 to RSV Season 2 follow-up through Day 515. Results therefore answer different time-specific questions.
- Binary endpoints: the reported efficacy and safety outcomes reduce clinical experience to whether prespecified events occurred during a defined period.
- Model dependence: adjusted efficacy estimates depend on the modified Poisson regression specification and its included covariates.
- Interpretation of confidence intervals: a confidence interval containing zero for a percentage difference does not establish equivalence or identical event rates.
- Endpoint-specific inference: a result for one RSV outcome or safety category should not be generalized to all RSV outcomes or all adverse events.
- Registry-level reporting: the statistical record does not provide formal analyses for every registered primary endpoint, so the inferential detail varies across endpoints.
22. Why This Trial Matters Statistically
CLEVER is a useful teaching case because it combines randomized prevention-trial design with several different approaches to binary clinical outcomes. The primary efficacy analysis uses an exact method and an adjusted modified Poisson regression, while the formally analyzed safety endpoints use score-based confidence intervals for differences in proportions.
| Concept | How it appears in CLEVER |
|---|---|
| Randomization | Randomized, parallel-group design |
| Blinding | Double-masked study |
| Binary outcomes | RSV-associated MALRI, hospitalization, fever, adverse events, and AESIs |
| Relative risk | Underlying measure for the reported efficacy estimates |
| Risk difference | Reported percentage-difference measure for safety endpoints |
| Exact methods | Used for the primary MALRI analysis and reported hospitalization analysis |
| Poisson regression | Modified Poisson regression with robust variance for efficacy estimation |
| Covariate adjustment | Region, gestational age, and age at randomization |
| Score-based confidence intervals | Miettinen-Nurminen method for reported safety comparisons |
| One-sided testing | One-sided p-value reported for primary MALRI efficacy |
| Analysis populations | Distinct efficacy and safety populations |
| Time-window definition | Endpoints assessed over prespecified postdose and RSV-season periods |
23. Related Tutorials
Learn more about the methods used in this trial:
24. Related Statistical Calculators
25. Sources
- ClinicalTrials.gov: NCT04767373.
- PubMed: PubMed record 40961446.
- PubMed: PubMed record 40673795.
- PubMed: PubMed record 34774184.
Continue through Clinical Biostats
Connect clinical-trial results with deeper tutorials and statistical calculators for the methods used in randomized clinical research.
26. Record Summary
CLEVER illustrates how a prevention trial can combine efficacy and safety endpoints with distinct statistical estimands and analysis populations. The primary RSV-associated MALRI analysis reports an efficacy estimate of 60.4 with a 95% CI of 44.1–71.9 and P < 0.001, while the formally analyzed safety endpoints report percentage differences with Miettinen-Nurminen confidence intervals. The secondary hospitalization analysis reports an efficacy estimate of 84.2 and the Day 1 to Day 180 MALRI analysis reports 59.5.
The statistical interpretation depends on keeping the endpoint definition, time window, analysis population, effect measure, covariate adjustment, and inferential procedure together. That framework is especially important when a trial contains multiple primary endpoints and when efficacy and safety outcomes are analyzed using different statistical methods.