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
MEDI8897 Ph2b was a randomized, parallel, quadruple-masked phase 2 prevention trial evaluating MEDI8897 versus placebo for the prevention of medically attended RSV-confirmed lower respiratory tract infection in healthy preterm infants.
| Feature | MEDI8897 Ph2b |
|---|---|
| Phase | Phase 2 |
| Condition | Respiratory Syncytial Virus Infections |
| Purpose | Prevention |
| Design | Randomized, parallel, quadruple-masked |
| Enrollment | 1453 |
| Interventions | MEDI8897 and placebo |
| Primary endpoint type | Binary |
| Primary analysis method | Poisson regression |
| Primary effect measure | Risk ratio / relative risk, reported as relative risk reduction |
| ClinicalTrials.gov | NCT02878330 |
| Trial status | Completed |
| Start | 2016-11-03 |
| Primary completion | 2018-07-17 |
| Lead sponsor | MedImmune LLC |
2. Clinical Question
The central question was whether MEDI8897 reduced the occurrence of medically attended RSV-confirmed lower respiratory tract infection compared with placebo during the period from Day 1 through Day 151 in healthy preterm infants.
Population
Healthy preterm infants enrolled in a prevention trial for respiratory syncytial virus infection.
Intervention
MEDI8897 50 mg.
Comparator
Placebo.
Primary question
Does MEDI8897 reduce medically attended RSV-confirmed LRTI from Day 1 through Day 151 compared with placebo?
3. Trial Design
Placebo
- Placebo treatment group
- Comparator for efficacy and safety
MEDI8897 50 mg
- MEDI8897 treatment group
- Evaluated for prevention of medically attended RSV-confirmed LRTI
The parallel design is important statistically because each participant contributes to one randomized treatment group rather than moving between treatments. Quadruple masking also reduces the opportunity for knowledge of treatment assignment to influence participant care, assessment, or other trial processes.
4. Endpoint Definition
| Endpoint | Time frame | Type | Definition |
|---|---|---|---|
| Number of Participants With Medically Attended Respiratory Syncytial Virus (RSV) Confirmed Lower Respiratory Tract Infection (LRTI) | From Day 1 through Day 151 | Binary | Medically attended RSV LRTI was determined from objective clinical LRTI criteria together with RSV test results from respiratory secretions using a validated RSV real-time RT-PCR assay detecting RSV A or RSV B. |
Clinical definition of LRTI
The registry definition required documented physical examination findings of rhonchi, rales, crackles, or wheeze together with at least one additional respiratory or clinical criterion.
- Increased respiratory rate at rest, with thresholds depending on age.
- Hypoxemia, defined using oxygen saturation thresholds that also depended on altitude.
- Clinical signs of severe respiratory disease.
- Dehydration secondary to inadequate oral intake because of respiratory distress, including need for intravenous fluid.
The respiratory-rate criteria were age-specific: for infants younger than 2 months, at least 60 breaths per minute; for infants 2–6 months, at least 50 breaths per minute; and for those older than 6 months through 2 years, at least 40 breaths per minute.
The oxygen-saturation criterion was less than 95% at altitudes up to and including 1800 meters or less than 92% at altitudes above 1800 meters.
5. Analysis Population and Statistical Methodology
The primary and secondary posted analyses used the intention-to-treat population. The registry defines this population as all participants randomized in the study and analyzed according to their randomized treatment group.
| Component | Registry specification |
|---|---|
| Analysis population | Intention-to-treat |
| Groups compared | Placebo vs MEDI8897 50 mg |
| Endpoint type | Binary |
| Statistical method | Poisson regression |
| Effect measure | Risk ratio / relative risk, reported as relative risk reduction |
| Hypothesis | Superiority |
| Primary time frame | Day 1 through Day 151 |
Why intention-to-treat analysis matters
Analyzing participants according to their randomized treatment group preserves the treatment comparison created by randomization. It answers the question associated with assignment to the intervention rather than restricting the analysis to participants who remained fully adherent to treatment.
For a prevention trial, this distinction is particularly useful because the estimand of interest is the difference associated with randomized treatment assignment during the specified follow-up period. Participants do not need to be perfectly compliant for randomization to remain the basis of the primary comparison.
Why Poisson regression was used
The registry identifies Poisson regression as the method used for the posted binary endpoint analyses. Although the outcome is ultimately expressed as whether a participant experienced the endpoint, Poisson regression can be used to estimate relative risks in settings where a binary outcome is analyzed through a count-model framework.
The resulting effect measure reported by the registry is relative risk reduction. This is a relative measure: it describes the difference between the groups in terms of their estimated risk rather than directly describing the absolute number of infections prevented.
A relative risk below 1 indicates lower estimated risk in the MEDI8897 group. Relative risk reduction can be expressed as 1 − RR when the effect is reported in that form.
6. Primary Result
Medically Attended RSV-Confirmed LRTI
The primary endpoint was the number of participants with medically attended RSV-confirmed LRTI from Day 1 through Day 151. The analysis used the ITT population and compared placebo with MEDI8897 50 mg using Poisson regression.
Relative risk reduction
95% CI: 52.3–81.2 · P < 0.0001
Superiority analysis in the intention-to-treat population.
| Primary endpoint | Comparison | Method | Effect estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Medically attended RSV-confirmed LRTI | Placebo vs MEDI8897 50 mg | Poisson regression | 70.1% relative risk reduction | 52.3–81.2 | <0.0001 |
The reported 70.1% relative risk reduction indicates a substantially lower estimated relative risk of the primary endpoint in the MEDI8897 group compared with placebo over Day 1 through Day 151. Equivalently, the estimated risk in the MEDI8897 group corresponds to approximately 29.9% of the placebo risk if the reported relative risk reduction is expressed as 1 − RR.
This does not mean that 70.1% of participants were protected, that 70.1% of infections were prevented in absolute terms, or that every individual participant experienced the same reduction in risk. Relative risk reduction is a group-level relative effect measure.
The 95% confidence interval of 52.3–81.2 describes uncertainty around the estimated relative risk reduction under the analysis framework. It does not describe the range of effects that individual infants would experience.
The P-value of <0.0001 addresses the statistical evidence against the null hypothesis under the specified superiority framework. It does not measure the magnitude of the treatment effect, and it should not be interpreted as the probability that the treatment effect is real.
Because the endpoint is binary and the registry reports a Poisson-regression analysis, the result should be understood as a model-based relative comparison rather than as a direct estimate of absolute risk reduction.
7. Secondary Endpoint Result
Hospitalization Due to RSV-Confirmed LRTI
The registry also posted a statistical analysis of the number of participants hospitalized due to RSV-confirmed LRTI from Day 1 through Day 151. This was a binary endpoint analyzed in the ITT population using Poisson regression.
Relative risk reduction
95% CI: 51.9–90.3 · P = 0.0002
Superiority analysis in the intention-to-treat population.
| Secondary endpoint | Comparison | Method | Effect estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Participants hospitalized due to RSV-confirmed LRTI | Placebo vs MEDI8897 50 mg | Poisson regression | 78.4% relative risk reduction | 51.9–90.3 | 0.0002 |
The reported 78.4% relative risk reduction indicates a lower estimated relative risk of hospitalization due to RSV-confirmed LRTI in the MEDI8897 group compared with placebo during Day 1 through Day 151. Expressed as a relative risk, the estimate corresponds to approximately 21.6% of the placebo risk.
The result does not mean that 78.4% of infants avoided hospitalization, nor does it provide the absolute difference in hospitalization probability between groups. An absolute risk difference would require the corresponding event risks in each treatment group.
The 95% CI of 51.9–90.3 indicates statistical uncertainty around the estimated relative risk reduction. Although the interval remains on the relative-risk-reduction side of the null, its width shows that the precise magnitude of the effect is less certain than the existence of a relative difference alone.
The P-value of 0.0002 quantifies evidence against the specified null hypothesis under the posted analysis. It does not quantify clinical importance or effect size.
Because hospitalization is a more specific event than the broader medically attended RSV-confirmed LRTI endpoint, the two results address related but distinct clinical outcomes.
8. Primary and Secondary Results Together
The two posted analyses point in the same directional pattern: the estimated relative risk was lower with MEDI8897 than with placebo for both the primary medically attended RSV-confirmed LRTI endpoint and the secondary hospitalization endpoint.
They should nevertheless be kept statistically distinct. The primary endpoint is the principal efficacy question, while hospitalization is a secondary endpoint. A consistent direction across endpoints can provide supportive evidence, but the endpoints do not become interchangeable simply because they are related.
9. Statistical Methods Explained
Why was intention-to-treat analysis used?
Question: Why analyze participants according to their randomized treatment assignment?
Answer: The ITT principle preserves the comparison established by randomization. In this trial, the registry states that the ITT population included all randomized participants and that participants were analyzed according to their randomized treatment group. This protects the primary comparison from being redefined based on treatment exposure after randomization.
Why was Poisson regression used for a binary endpoint?
Question: The primary endpoint is whether a participant experienced RSV-confirmed LRTI. Why does the registry identify Poisson regression?
Answer: Poisson regression is commonly associated with count and rate modeling, but a Poisson-model framework can also be used to estimate relative risks for binary outcomes when implemented appropriately. Here, the registry specifically identifies Poisson regression as the analysis method and relative risk reduction as the reported effect measure. The important interpretive point is that the resulting estimate is a relative risk-based comparison, not a hazard ratio.
What does a 70.1% relative risk reduction mean?
Question: How should the primary estimate be translated into plain statistical language?
Answer: A relative risk reduction of 70.1% corresponds to an estimated relative risk of approximately 0.299, because relative risk reduction is expressed as 1 minus relative risk. Thus, the estimated risk in the MEDI8897 group was approximately 29.9% of the placebo risk over the specified follow-up period.
That statement remains relative. It cannot be translated into an absolute number of infections prevented without the underlying event risks in the two treatment groups.
What does the 95% confidence interval tell us?
Question: What information is added by the interval of 52.3–81.2?
Answer: The interval quantifies uncertainty around the estimated relative risk reduction. The point estimate is 70.1%, but the statistical estimate is not exact. The interval gives a range of effect estimates compatible with the model and sampling framework used for the analysis.
Why does the P-value not measure effect size?
Question: Does P < 0.0001 mean that the treatment effect is very large?
Answer: No. A P-value measures evidence against a null hypothesis under a specified statistical model. It depends on both the magnitude of the observed effect and the amount of information available. Effect size is described by the relative risk or relative risk reduction, while the confidence interval communicates uncertainty around that effect.
Why does the analysis population matter?
Question: Why is it important that both posted efficacy analyses used the ITT population?
Answer: The treatment comparison is anchored to randomized assignment. This means the interpretation concerns the effect associated with being assigned to MEDI8897 versus placebo rather than the effect among a selected subset who completed treatment exactly as intended.
10. Understanding Relative Risk Versus Other Effect Measures
The registry reports risk ratio / relative risk as the effect measure and describes the posted estimates as relative risk reductions. This distinguishes the analysis from several other commonly encountered clinical-trial measures.
| Measure | What it describes | Used in this trial's posted analyses? |
|---|---|---|
| Risk ratio | Risk in the MEDI8897 group relative to risk in the placebo group. | Yes |
| Relative risk reduction | 1 minus the risk ratio, expressed as a percentage when reported that way. | Yes |
| Absolute risk reduction | Difference between the event risks in the two groups. | Not reported in the registry analysis information |
| Odds ratio | Ratio of the odds of the event between treatment groups. | No |
| Hazard ratio | Relative instantaneous event rate from a time-to-event model. | No |
This distinction matters because relative risk reduction, absolute risk reduction, odds ratio, and hazard ratio answer different statistical questions. A relative risk reduction cannot be interpreted as an absolute percentage-point difference.
11. Interpretation of the Confidence Intervals
Primary endpoint
Relative risk reduction: 70.1%. The 95% CI was 52.3–81.2.
Hospitalization endpoint
Relative risk reduction: 78.4%. The 95% CI was 51.9–90.3.
The primary endpoint confidence interval spans a range from 52.3% to 81.2% relative risk reduction. This means the point estimate of 70.1% should not be treated as a perfectly known quantity.
The hospitalization endpoint has a point estimate of 78.4%, with a 95% confidence interval from 51.9% to 90.3%. The interval is relatively broad because the hospitalization endpoint is a more specific event and therefore can contain less statistical information than a broader endpoint.
A confidence interval is about the uncertainty of an estimated treatment effect. It is not the probability that the true treatment effect lies inside the interval, and it is not a prediction interval for individual patients.
12. P-values and Superiority Testing
Both posted analyses are identified as superiority analyses. The primary endpoint had a P-value of <0.0001, while the hospitalization endpoint had a P-value of 0.0002.
Both P-values quantify evidence against the corresponding null hypothesis under the posted statistical analysis. Neither P-value is a measure of the size, clinical importance, or absolute benefit of MEDI8897.
Because the registry identifies the analyses as superiority analyses, the statistical question is whether the randomized groups differ in the prespecified direction of the treatment comparison rather than whether MEDI8897 is merely no worse than placebo.
13. Intention-to-Treat Analysis in Context
The ITT analysis is particularly important for a randomized prevention trial. Randomization creates the statistical basis for comparing treatment groups. Once that assignment has occurred, analyzing participants according to their randomized group maintains that basis.
| Question | ITT interpretation |
|---|---|
| Who belongs in the efficacy population? | All participants randomized in the study. |
| How are participants classified? | According to their randomized treatment group. |
| What comparison is preserved? | The comparison created by randomization. |
| What endpoint was analyzed? | Medically attended RSV-confirmed LRTI from Day 1 through Day 151 for the primary analysis. |
ITT does not eliminate every source of bias in a clinical trial, but it prevents post-randomization treatment behavior from being used to redefine the primary randomized comparison.
14. Safety
The registry reports serious adverse events by treatment arm. These data provide a separate safety perspective from the efficacy analyses.
| Treatment group | Participants with serious adverse events | At risk |
|---|---|---|
| Placebo | 81 | 479 |
| MEDI8897 50 mg | 108 | 968 |
The raw counts should be interpreted alongside the corresponding numbers at risk. The treatment groups differ in the number of participants represented in these safety data, so comparing counts alone does not provide a meaningful measure of relative safety frequency.
The serious-adverse-event figures are descriptive counts of affected participants and participants at risk. They should not be confused with the primary efficacy estimate, and they do not by themselves provide a formal comparative safety test.
15. Trial Timeline
Trial start
The registry lists November 3, 2016 as the study start date.
Primary completion
The registry lists July 17, 2018 as the primary completion date.
Final registry status
The study is listed as completed, with 1453 participants enrolled.
16. What the Primary Estimate Does — and Does Not — Mean
The reported 70.1% relative risk reduction means that the estimated risk under MEDI8897 was approximately 29.9% of the placebo risk under the posted analysis. It does not mean that 70.1% of infants were individually protected or that the absolute probability of RSV-confirmed LRTI decreased by 70.1 percentage points.
The 95% CI of 52.3–81.2 communicates uncertainty around the relative effect estimate. The point estimate is therefore best read together with its interval rather than in isolation.
The P-value of <0.0001 indicates strong statistical evidence against the null hypothesis in the posted superiority analysis. It does not establish the clinical importance of the result and does not quantify the probability that the treatment is effective.
An absolute risk reduction would answer a different question: how many fewer participants experienced the endpoint in one group than the other. The relative risk reduction alone cannot supply that absolute difference.
17. Why the Endpoint Definition Matters
The primary endpoint was not simply any RSV-positive test. It required a combination of medically attended illness, objective clinical LRTI findings, and laboratory confirmation of RSV.
This endpoint construction affects interpretation in several ways. First, laboratory confirmation helps distinguish RSV-associated disease from respiratory illness caused by other pathogens. Second, the clinical criteria require evidence of lower respiratory tract involvement rather than an isolated positive respiratory test. Third, the endpoint is limited to medically attended episodes, so it is specifically a clinical outcome occurring in the context of medical care.
Clinical component
Documented physical findings such as rhonchi, rales, crackles, or wheeze plus specified respiratory or clinical criteria.
Laboratory component
RSV testing of respiratory secretions using a validated real-time RT-PCR assay detecting RSV A or RSV B.
Care component
The endpoint is specifically medically attended RSV-confirmed LRTI.
Time component
Events are evaluated from Day 1 through Day 151.
18. Statistical Interpretation Versus Clinical Interpretation
Statistical interpretation
The posted primary analysis estimated a 70.1% relative risk reduction, with a two-sided 95% confidence interval of 52.3–81.2 and P < 0.0001.
Clinical interpretation
The endpoint was a clinically defined, medically attended RSV-confirmed LRTI outcome. The statistical result describes the relative difference in this outcome between randomized groups during the specified follow-up.
The two perspectives should remain connected but distinct. Statistical significance describes evidence under a model and hypothesis-testing framework. Clinical interpretation asks what the measured endpoint and effect mean in the context of the population, follow-up period, and event definition.
19. Important Limitations and Interpretation Issues
- Relative rather than absolute effect: the posted efficacy results are expressed as relative risk reductions. Without the corresponding event risks in both groups, the absolute risk reduction and number needed to treat cannot be derived from the reported effect measure alone.
- Endpoint specificity: the primary endpoint required both objective LRTI findings and RSV laboratory confirmation, and it was limited to medically attended illness. The result therefore pertains specifically to this endpoint definition.
- Time frame: the primary and secondary posted analyses cover Day 1 through Day 151. The reported effects should not automatically be generalized to a different follow-up period.
- Binary endpoint: the primary endpoint is a binary outcome. It should not be interpreted as a time-to-event hazard ratio or as an incidence-rate ratio.
- Model dependence: Poisson regression provides a model-based estimate. The statistical interpretation therefore depends on the model specification and assumptions underlying that analysis.
- Secondary endpoint: hospitalization due to RSV-confirmed LRTI is a secondary outcome. Its P-value should be interpreted in the context of its position as a secondary analysis rather than as if it were the primary endpoint.
- Safety counts: serious-adverse-event counts differ between groups, but the number at risk also differs. Counts alone should not be interpreted as comparative safety rates.
- Generalizability: the registry describes the study population as healthy preterm infants. Results should be interpreted in relation to that population rather than automatically extended to populations not represented by the trial.
20. Why This Trial Matters Statistically
MEDI8897 Ph2b is a useful statistical teaching case because it illustrates how a randomized prevention trial can combine a clinically specific binary endpoint with relative-risk estimation and intention-to-treat analysis.
| Concept | How it appears in MEDI8897 Ph2b |
|---|---|
| Randomization | Participants were randomized to MEDI8897 or placebo. |
| Parallel design | Participants remained in their assigned treatment group rather than crossing between interventions. |
| Quadruple masking | The trial used quadruple masking to reduce knowledge of treatment assignment. |
| Intention-to-treat analysis | Posted analyses included all randomized participants analyzed according to randomized treatment group. |
| Binary endpoint | The primary outcome recorded whether participants experienced medically attended RSV-confirmed LRTI. |
| Poisson regression | The registry identifies Poisson regression as the statistical method for the posted efficacy analyses. |
| Relative risk | The effect measure is risk ratio / relative risk. |
| Relative risk reduction | The primary estimate was reported as 70.1% relative risk reduction. |
| Confidence interval | The primary 95% confidence interval was 52.3–81.2. |
| P-value | The primary superiority analysis reported P < 0.0001. |
| Secondary endpoint | Hospitalization due to RSV-confirmed LRTI had a reported relative risk reduction of 78.4%. |
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Calculators
23. Sources
- ClinicalTrials.gov: NCT02878330 — MEDI8897 Ph2b.
- PubMed: PubMed record for PMID 38902186.
- PubMed: PubMed record for PMID 37468530.
- PubMed: PubMed record for PMID 32726528.
Continue exploring Clinical Biostats
Connect this trial's design, endpoints, and statistical methods with deeper biostatistics tutorials and practical statistical tools.
24. Record Summary
MEDI8897 Ph2b provides a clear example of randomized prevention-trial analysis using a clinically defined binary endpoint, intention-to-treat analysis, Poisson regression, and relative-risk-based effect measures. The primary analysis reported a 70.1% relative risk reduction for medically attended RSV-confirmed LRTI from Day 1 through Day 151, with a 95% confidence interval of 52.3–81.2 and P < 0.0001. A secondary analysis of hospitalization due to RSV-confirmed LRTI reported a 78.4% relative risk reduction, with a 95% confidence interval of 51.9–90.3 and P = 0.0002.
The most important statistical distinction is between relative and absolute effects. The reported relative risk reductions quantify the relative difference between randomized groups, while the confidence intervals quantify uncertainty around those estimates. The ITT population preserves the randomized comparison, and the specific clinical and laboratory endpoint definition determines exactly what outcome the reported treatment effect describes.