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
REMAP-CAP is a randomized phase 3 factorial trial designed as a multifactorial adaptive platform for community-acquired pneumonia. The registry lists community-acquired pneumonia, influenza, and COVID-19 as conditions and describes a treatment-purpose study with no masking.
| Feature | REMAP-CAP |
|---|---|
| Phase | Phase 3 |
| Conditions | Community-acquired Pneumonia, Influenza, COVID-19 |
| Design | Randomized, factorial |
| Allocation | Randomized |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 20,000 |
| Number of arms | 18 |
| Lead sponsor | UMC Utrecht |
| Sponsor type | Other |
| Status | Recruiting |
| Start date | April 11, 2016 |
| Primary completion date | February 2026 |
| ClinicalTrials.gov | NCT02735707 |
2. Clinical Question
The central statistical question is how multiple treatment strategies can be evaluated efficiently within a randomized, multifactorial platform in patients with community-acquired pneumonia, influenza, and COVID-19.
Population
The registry identifies community-acquired pneumonia, influenza, and COVID-19 as the conditions under study, with a planned enrollment of 20,000 participants.
Intervention structure
The intervention list spans multiple therapeutic domains, including antibacterial treatment, macrolides, corticosteroids, antivirals, immune modulation, anticoagulation, antiplatelet therapy, endothelial modulation, renin-angiotensin system strategies, and mechanical ventilation strategies.
Comparator structure
The registry includes no-intervention or standard-care strategies alongside active treatment strategies in several domains, allowing treatment strategies to be evaluated within a multifactorial platform.
Primary questions
The registered primary endpoints ask whether treatment strategies are associated with differences in all-cause mortality by Day 90 and, for patients with suspected or proven COVID-19 pandemic infection, days alive and not receiving organ support in ICU by Day 21.
3. Trial Design
REMAP-CAP is registered as a randomized, factorial phase 3 study with no masking and a treatment primary purpose. The defining statistical feature is its multifactorial platform structure: multiple treatment domains are evaluated within a single overarching randomized trial rather than treating each question as an entirely separate conventional trial.
Why a factorial platform changes the statistical problem
In a conventional two-arm trial, the primary comparison can often be represented as one treatment versus one comparator. A factorial platform instead contains multiple treatment questions operating within a common trial infrastructure. The statistical unit of interpretation therefore becomes the particular treatment factor and its relevant comparison, rather than simply the total number of arms.
This distinction matters because the number of listed interventions is much larger than the number of registered arms. The intervention list contains multiple strategies within domains such as antibacterial therapy, corticosteroid therapy, antiviral therapy, immune modulation, anticoagulation, antiplatelet therapy, and respiratory support. The factorial structure is what allows those treatment questions to coexist within one platform.
4. Intervention Domains
The registry lists a broad set of interventions. For statistical interpretation, these can be organized into treatment domains without treating the individual interventions as though they were one conventional sequence of treatment arms.
| Domain represented in the intervention list | Examples listed in the registry |
|---|---|
| Antibacterial therapy | Ceftriaxone; moxifloxacin or levofloxacin; piperacillin-tazobactam; ceftaroline; amoxicillin-clavulanate |
| Macrolide strategy | Standard course macrolide; extended course macrolide |
| Corticosteroid strategy | No systemic corticosteroid; fixed-duration hydrocortisone; shock-dependent hydrocortisone; fixed-duration higher-dose hydrocortisone; fixed-duration dexamethasone |
| Influenza antiviral strategies | No antiviral agent for influenza; five-days oseltamivir; ten-days oseltamivir; baloxavir marboxil; oseltamivir plus baloxavir strategies |
| COVID-19 antiviral strategies | No antiviral agent for COVID-19; lopinavir/ritonavir; nirmatrelvir/ritonavir; remdesivir; nirmatrelvir/ritonavir plus remdesivir |
| Immune modulation | No immune modulation for COVID-19; interferon beta-1a; anakinra; tocilizumab; sarilumab; baricitinib; no immune modulator for influenza |
| Anticoagulation | Local standard venous thromboprophylaxis; therapeutic dose anticoagulation; conventional low-dose thromboprophylaxis; intermediate-dose thromboprophylaxis; continuation of therapeutic-dose anticoagulation |
| Convalescent plasma | No immunoglobulin; convalescent plasma; delayed administration of convalescent plasma |
| Other therapeutic domains | Vitamin C; aspirin; P2Y12 inhibitor; simvastatin; eritoran; apremilast; imatinib; renin-angiotensin system strategies; cysteamine |
| Mechanical ventilation | Clinician-preferred mechanical ventilation strategy; protocolised mechanical ventilation strategy |
5. Primary Endpoints
| Endpoint | Registry time frame | Registry description |
|---|---|---|
| All-cause mortality | Day 90 | All-cause mortality measured at Day 90. |
| Days alive and not receiving organ support in ICU | Day 21 | Primary end-point for patients with suspected or proven COVID-19 pandemic infection. |
These two endpoints have different statistical structures. All-cause mortality is a binary status at a specified time point if analysed simply as mortality by Day 90. The days-alive-and-free-of-organ-support endpoint is a bounded outcome that incorporates both survival and time spent without organ support, making its distribution fundamentally different from a conventional binary endpoint.
6. Planned Analysis
No formal statistical analyses are posted in the ClinicalTrials.gov record. The registry identifies the outcomes that are to be measured, but it does not provide numerical treatment-effect estimates, confidence intervals, or p-values for these primary endpoints.
All-cause mortality at Day 90
For a fixed-time mortality endpoint, a natural statistical analysis compares the proportion of patients who have died by Day 90 between the relevant randomized treatment strategies. Depending on the prespecified analysis plan, this can be expressed using a risk ratio, risk difference, odds ratio, or a regression model for binary outcome data. The key estimand is the difference in mortality attributable to randomized treatment assignment at the specified Day 90 horizon.
Days alive and not receiving organ support in ICU at Day 21
This endpoint combines several clinical states into a single outcome. Patients who die early cannot accumulate additional days alive without organ support, while survivors may accumulate more days according to the amount of time they remain alive and free of organ support through Day 21. A suitable analysis therefore needs to respect the bounded and potentially non-normal distribution of the endpoint rather than automatically treating it as an ordinary continuous measurement.
7. Statistical Methodology
Randomization as the basis for causal comparison
Randomization is the central design feature supporting causal interpretation. If treatment assignment is randomized as specified, differences in outcomes between randomized groups can be interpreted as treatment-assignment effects under the assumptions of the trial and its estimand. This is distinct from an observational comparison in which treatment choice may be associated with prognosis.
Factorial analysis
Factorial designs evaluate more than one treatment factor within the same experimental platform. Their efficiency comes from sharing trial infrastructure and, when assumptions permit, information across treatment questions. The analysis must nevertheless identify the specific factor comparison being estimated.
The relevant contrast depends on the outcome, estimand, randomization structure, and the treatment factor being evaluated. A factorial design does not make every intervention comparison interchangeable.
Adaptive platform structure
An adaptive platform can permit treatment questions to evolve within a continuing trial infrastructure. From a statistical perspective, this creates a need to distinguish the treatment comparison being evaluated, the information available when decisions are made, and the error-control framework governing repeated or adaptive decisions.
Fixed-time mortality
For mortality measured at Day 90, the analysis can be framed around the probability of death by a prespecified time. If all participants have a defined Day 90 status, the endpoint can be analysed using standard binary-outcome methods. If time-to-death information is also analysed, survival methods can instead use the timing of death and censoring.
Composite or hierarchical interpretation of days alive and free of organ support
The Day 21 endpoint combines survival with absence of organ support. This means that a treatment can affect the endpoint through mortality, through duration of organ support among survivors, or through both. Its interpretation is therefore different from simply reporting mortality or mean ICU duration alone.
8. Statistical Methods Explained
Why use a factorial design?
A factorial design allows several treatment factors to be evaluated within one randomized platform. The principal statistical advantage is efficiency: participants and trial infrastructure can contribute to more than one treatment question. The analysis must still preserve the distinction between the separate treatment factors.
What does randomized allocation contribute?
Randomization makes treatment assignment independent of baseline prognostic characteristics in expectation. Consequently, systematic differences in outcomes between randomized groups can be attributed to treatment assignment more directly than in an observational comparison.
Why is all-cause mortality at Day 90 a useful endpoint?
All-cause mortality is objectively defined and does not require adjudication of a particular cause of death for the primary classification. The Day 90 time point also gives the analysis a prespecified clinical horizon. The resulting estimand is mortality status by that time rather than an unspecified long-term survival measure.
Why is days alive and not receiving organ support different from mortality?
Mortality records whether a patient is alive at the specified time. Days alive and not receiving organ support contains additional information about the course of critical illness among survivors. Two treatment groups could have similar mortality but differ in the amount of time survivors spend without organ support.
Why does a factorial platform require careful multiplicity thinking?
When multiple treatment questions are evaluated within one platform, the number of statistical comparisons can increase. If each hypothesis is tested independently at the same nominal significance level, the probability of at least one false-positive finding can increase. Adaptive platforms therefore require a prespecified framework for the hypotheses, decision rules, and error control relevant to each treatment domain.
Why can the number of arms be misleading?
The registry lists 18 arms, but it also lists many interventions and treatment strategies. In a multifactorial platform, the statistical question is not necessarily an omnibus comparison of all arms. The scientifically relevant contrast may concern a particular treatment factor against its corresponding comparator while accounting for the factorial structure.
9. Endpoint-Specific Statistical Considerations
| Endpoint | Statistical structure | Key interpretation issue |
|---|---|---|
| All-cause mortality, Day 90 | Fixed-time binary outcome if analysed as mortality status at Day 90 | Effect estimate refers specifically to mortality by Day 90. |
| Days alive and not receiving organ support in ICU, Day 21 | Bounded outcome incorporating survival and organ-support status | Distribution may be skewed or concentrated at boundary values. |
Binary mortality versus time-to-event survival
A Day 90 mortality analysis and a time-to-death analysis answer related but different questions. A fixed-time analysis asks whether mortality differs by Day 90. A survival analysis uses the timing of deaths and can account for censoring. Neither should be described as though it were the other.
Bounded outcomes require distributional attention
Days alive and free of organ support cannot take arbitrary real-valued observations. The endpoint has natural boundaries and may have many observations concentrated at particular values, especially when patients die or remain free of organ support for much of the observation period. Methods should therefore be chosen with the observed endpoint distribution and the prespecified estimand in mind.
10. Factorial Design and Interaction
One of the most important statistical questions in a factorial platform is whether the effect of one treatment factor depends on another treatment factor. This is an interaction question.
Here, βAB represents the interaction term. A substantial interaction would indicate that the estimated effect associated with one factor changes according to the level of another factor.
The existence of multiple factors does not automatically establish an interaction. An interaction must be evaluated statistically and interpreted in the context of the prespecified design and the amount of information available for the relevant combinations.
11. Adaptive Platform Statistical Considerations
The registry title identifies REMAP-CAP as a multifactorial adaptive platform trial. Adaptive designs can modify aspects of a study in response to accumulating information while retaining a common trial infrastructure.
Efficiency
A continuing platform can evaluate multiple treatment strategies without requiring an entirely separate trial infrastructure for every treatment question.
Information sharing
Factorial and platform structures can allow information from the broader trial to contribute to treatment comparisons when the statistical assumptions and randomization structure support that use.
Decision rules
Adaptive decisions require prespecified statistical rules so that treatment changes do not turn an accumulating sequence of analyses into an uncontrolled collection of tests.
Interpretation
A treatment effect should always be tied to the specific randomized contrast and analysis population from which it was estimated.
12. Multiplicity
Multiplicity is especially important in a multifactorial platform because multiple treatment questions can be evaluated within the same overarching trial. Each additional hypothesis creates another opportunity for a chance finding if the testing structure is not appropriately controlled.
| Multiplicity source | Why it matters |
|---|---|
| Multiple treatment factors | Each factor can generate a separate treatment hypothesis. |
| Multiple endpoints | All-cause mortality and days alive and not receiving organ support are distinct primary endpoints. |
| Multiple treatment strategies | The intervention list contains numerous active and comparator strategies. |
| Adaptive decisions | Repeated decisions based on accumulating information can affect the statistical error structure. |
| Interaction analyses | Testing whether treatment effects differ across factors adds additional hypotheses. |
A nominal p-value is meaningful only in the context of the testing procedure that produced it. In a platform trial, the relevant statistical framework may therefore be more complex than applying an isolated 0.05 threshold to every observed comparison.
13. Stratification and Covariate Adjustment
The registry information identifies randomized allocation and a factorial design but does not provide a list of stratification factors or a prespecified covariate-adjustment model. Those details are important because the exact estimand and standard-error calculation depend on how the randomization and analysis models are specified.
14. Missing Data and Censoring
The registry identifies fixed follow-up horizons of Day 21 and Day 90 but does not specify a missing-data or imputation strategy. For Day 90 mortality, missing vital-status information can affect the classification of the endpoint. For days alive and not receiving organ support, missing follow-up can affect the number of days that can be assigned.
These issues are distinct from ordinary right censoring in a time-to-event analysis. Missing outcome information requires assumptions or rules for determining how incomplete observations contribute to the analysis. The validity of those assumptions should be evaluated rather than treating missingness as statistically neutral.
15. Safety and Adverse Events
The registry information identifies numerous therapeutic strategies but does not provide posted serious-adverse-event counts by treatment arm. Consequently, there are no arm-specific serious-adverse-event estimates to summarize here.
In a platform with many treatment domains, safety analysis has an additional structural challenge: adverse events need to be linked to the relevant treatment exposure and comparison. A single overall adverse-event rate across the entire platform would not necessarily answer the safety question for any particular treatment factor.
Treatment-specific safety
Safety should generally be interpreted in relation to the treatment comparison that generated the exposure.
Exposure matters
Differences in treatment exposure, duration, and clinical severity can affect observed adverse-event frequencies.
Serious events
Serious adverse events are clinically important outcomes, but their interpretation requires event definitions and arm-specific counts.
Platform complexity
Multiple simultaneous treatment questions make it especially important to identify which treatment contrast a safety estimate represents.
16. Trial Timeline
Trial start
The ClinicalTrials.gov record lists April 11, 2016 as the study start date.
Randomized factorial platform
The registry describes the study as a randomized, phase 3, factorial trial with no masking and a treatment primary purpose.
Primary completion
The registry lists February 2026 as the primary completion date.
Recruiting
The ClinicalTrials.gov record lists the study status as recruiting.
17. What the Primary Endpoints Do — and Do Not — Mean
All-cause mortality at Day 90 is a direct measure of whether a patient has died by the specified time point. A treatment effect for this endpoint would describe a difference in mortality by Day 90. It would not by itself describe the timing of deaths after Day 90 or the quality and duration of recovery among survivors.
This endpoint captures more than survival alone. A patient contributes days according to being alive and not receiving organ support in ICU through Day 21. Its interpretation therefore reflects both mortality and the clinical course among patients who remain alive.
A future treatment-effect estimate should be accompanied by an appropriate confidence interval. The point estimate alone describes the observed effect estimate, while the confidence interval describes statistical uncertainty under the analysis framework used.
18. Why the Absence of a Posted Effect Estimate Matters
The ClinicalTrials.gov record contains the registered primary endpoints and their time frames but no posted statistical analyses. Consequently, the central statistical information currently available from the registry is the trial architecture rather than a numerical estimate of treatment effect.
This distinction is important for interpretation. Knowing that a trial is randomized and factorial explains how causal comparisons can be constructed. It does not establish the direction or magnitude of any particular treatment effect. Those conclusions require the corresponding analyzed outcome data.
19. Limitations and Interpretation Issues
- No posted statistical analyses: the registry does not provide numerical treatment-effect estimates, confidence intervals, or p-values for the registered primary endpoints.
- Multifactorial structure: the trial contains multiple treatment questions, so a simple all-arm comparison would not necessarily correspond to the scientific question for any individual treatment factor.
- Multiplicity: multiple treatment factors and endpoints create a complex hypothesis-testing environment requiring appropriate error control.
- Adaptive platform: adaptive decision-making requires careful prespecification of decision rules and statistical operating characteristics.
- Endpoint heterogeneity: Day 90 mortality and days alive and not receiving organ support in ICU by Day 21 have different statistical structures and should not be interpreted as interchangeable outcomes.
- Missing outcome information: the registry information does not specify a missing-data or imputation strategy for the registered primary endpoints.
- Safety interpretation: arm-specific serious-adverse-event results are not posted in the registry information summarized here.
- Registry status: the ClinicalTrials.gov record lists the study as recruiting while also listing February 2026 as the primary completion date.
- Generalizability: applicability depends on the enrolled population, treatment domain, clinical setting, and specific randomized comparison being considered.
20. Why This Trial Matters Statistically
REMAP-CAP is a useful teaching case because it illustrates how clinical-trial statistics change when the study moves beyond a single fixed two-arm comparison. The factorial adaptive platform structure creates a setting in which trial efficiency, multiple treatment questions, endpoint selection, interaction, multiplicity, and adaptive decision-making all become part of the statistical problem.
| Concept | How it appears in REMAP-CAP |
|---|---|
| Randomization | The registry specifies randomized allocation. |
| Factorial design | The registry specifies a factorial design model. |
| Adaptive platform | The trial title identifies it as a multifactorial adaptive platform trial. |
| Multiple treatment questions | The intervention list contains numerous treatment strategies across several therapeutic domains. |
| Multiple endpoints | All-cause mortality at Day 90 and days alive and not receiving organ support in ICU at Day 21 are registered as primary endpoints. |
| Time-to-event thinking | Mortality can be studied using survival methods when the timing of death and censoring are available. |
| Bounded outcomes | Days alive and free of organ support is constrained by the Day 21 horizon and incorporates survival status. |
| Interaction | Factorial trials require consideration of whether treatment effects depend on other treatment factors. |
| Multiplicity | Multiple factors and endpoints create multiple statistical hypotheses. |
| Missing data | Incomplete follow-up can affect fixed-time and composite clinical outcomes. |
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Calculators
23. Sources
- ClinicalTrials.gov: NCT02735707 — REMAP-CAP.
Continue through Clinical Biostats
Explore statistical tutorials, calculators, and additional clinical-trial analyses.
24. Record Summary
REMAP-CAP is registered as a phase 3 randomized factorial trial with a multifactorial adaptive platform structure, no masking, and a treatment primary purpose. The registry lists 20,000 participants, 18 arms, and a broad set of treatment strategies spanning antibacterial therapy, corticosteroids, antivirals, immune modulation, anticoagulation, antiplatelet therapy, respiratory support, and other therapeutic domains. Its registered primary endpoints are all-cause mortality at Day 90 and, for patients with suspected or proven COVID-19 pandemic infection, days alive and not receiving organ support in ICU at Day 21.
From a statistical perspective, the central teaching points are the distinction between a factorial platform and a conventional multi-arm trial, the importance of defining the treatment-specific estimand, the different outcome structures of mortality and days alive without organ support, and the need to account for multiplicity and adaptive decision-making. The ClinicalTrials.gov record currently provides the trial design and registered endpoint framework but no posted numerical statistical analyses for the primary endpoints.