This page separates information reported in the ClinicalTrials.gov record from statistical interpretation. The registry record indicates that results have been posted, but the ClinicalTrials.gov record contains no formal statistical analyses.
1. Trial at a Glance
CARAVAGGIO was a completed randomized, parallel-group phase 3 treatment trial in patients with venous thromboembolism and cancer. The registry identifies two interventions, apixaban and dalteparin, and a primary efficacy endpoint of objectively confirmed recurrent venous thromboembolism over 6 months.
| Feature | CARAVAGGIO |
|---|---|
| Trial name | CARAVAGGIO |
| NCT identifier | NCT03045406 |
| Brief title | Apixaban for the Treatment of Venous Thromboembolism in Patients With Cancer |
| Therapeutic area | Hematology |
| Condition | Venous Thromboembolism |
| Phase | Phase 3 |
| Status | Completed |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 1170 |
| Interventions | Apixaban; Dalteparin |
| Lead sponsor | Fadoi Foundation, Italy |
| Sponsor type | Other |
2. Clinical Question
The registry describes a phase 3 randomized treatment trial of apixaban and dalteparin in patients with cancer and venous thromboembolism. The registered primary efficacy outcome is recurrent venous thromboembolism occurring during the study period, assessed over a 6-month time frame.
Population
Patients with cancer and venous thromboembolism.
Intervention
Apixaban.
Comparator
Dalteparin.
Primary question
How does randomized treatment with apixaban compare with dalteparin with respect to objectively confirmed recurrent venous thromboembolism over 6 months?
3. Trial Design
Apixaban
- Intervention classified as a drug.
- Randomized treatment arm.
- Serious adverse events reported for 255 affected patients among 576 at risk.
Dalteparin
- Intervention classified as a drug.
- Randomized treatment arm.
- Serious adverse events reported for 276 affected patients among 579 at risk.
4. Primary Endpoint
| Endpoint | Registry definition | Time frame | Endpoint type |
|---|---|---|---|
| Recurrent Venous Thromboembolism | Primary efficacy outcome: objectively confirmed recurrent VTE occurring during the study period, that means the composite of: proximal DVT of the lower limbs (symptomatic or unsuspected), DVT of the upper limb (symptomatic), PE (symptomatic or unsuspected) | 6 months | Other / unclear |
The endpoint is a composite outcome. A recurrent VTE event can enter the primary endpoint through more than one clinical manifestation: proximal lower-limb DVT, upper-limb DVT, or pulmonary embolism. The registry specifically distinguishes symptomatic and unsuspected events for the relevant components.
A composite endpoint combines several clinically defined event types into one outcome for the primary analysis. Statistically, this can increase the number of observed events relative to relying on a single component, but the interpretation depends on the components being defined prospectively and on understanding which component events contribute to the composite.
For CARAVAGGIO, the ClinicalTrials.gov record identifies the components of recurrent VTE but does not provide component-specific event counts or a formal statistical comparison among those components.
5. Results and Analysis Availability
The ClinicalTrials.gov record states that results have been posted and that there is one posted outcome measure. However, the registry-reported statistical-analysis field is empty:
Formal statistical analyses in the ClinicalTrials.gov record
No formal statistical analyses were posted.
This distinction is important. A registry can indicate that an outcome has been posted without providing, in the ClinicalTrials.gov record, the full statistical comparison needed to reproduce an effect estimate. The available CARAVAGGIO data therefore support identification of the endpoint and its assessment framework, but they do not support reporting an efficacy hazard ratio, risk ratio, odds ratio, confidence interval, p-value, or treatment-group event count for recurrent VTE.
What analysis would normally be appropriate?
For a recurrent VTE endpoint assessed over a defined 6-month period, an analysis could be framed as a comparison of the proportion of participants experiencing at least one objectively confirmed recurrent VTE by the specified time point. If the underlying data preserve exact event times and censoring information, a time-to-event analysis could also be used, with treatment effect summarized using an appropriate time-to-event measure.
The choice between these approaches depends on the prespecified statistical analysis plan, the exact event and censoring structure, and how competing events or multiple episodes are handled. Those details are not reported in the statistical-analysis field provided for this record. Consequently, the registry data alone do not establish which formal model or hypothesis test was used.
The precise effect measure and inferential procedure should come from the prespecified statistical analysis rather than being inferred from the endpoint label alone.
6. Serious Adverse Events
The ClinicalTrials.gov record reports serious adverse events by randomized treatment arm as affected patients divided by patients at risk.
| Safety measure | Apixaban | Dalteparin |
|---|---|---|
| Serious adverse events | 255/576 | 276/579 |
The reported safety figures describe the number of patients affected by serious adverse events relative to the number at risk in each arm. They are therefore useful for understanding the amount of serious-adverse-event information available in the registry record.
They do not by themselves establish a statistical difference between treatments. No confidence interval, hypothesis test, adjusted effect measure, or formal safety comparison is included in the ClinicalTrials.gov record.
It is also important not to treat a serious-adverse-event count as equivalent to the primary efficacy endpoint. Safety and efficacy answer different questions and require separate prespecified definitions and analyses.
7. Statistical Methodology
Randomization
Randomization is the central statistical feature identified by the registry. Participants were assigned to treatment using a randomized allocation, creating the basis for a comparative treatment-effect analysis.
The principal advantage of randomization is that, under appropriate conduct and analysis, treatment assignment is independent of baseline prognostic characteristics in expectation. This allows differences observed between treatment groups to be interpreted in relation to the randomized intervention rather than simply as differences between self-selected groups.
Randomization is designed to reduce systematic differences in prognostic factors between treatment groups. It does not eliminate sampling variation, measurement error, missing data, or problems in endpoint ascertainment.
Parallel-group analysis
The registry identifies a parallel design. In a parallel-group trial, participants remain associated with their randomized treatment group for the comparative analysis rather than acting as their own controls through crossover between randomized interventions.
This design is especially natural for estimating a treatment-group contrast because the comparison is anchored to the original randomization. The exact analysis population, handling of treatment discontinuation, and treatment of protocol deviations must nevertheless be specified in the statistical analysis plan.
Composite endpoint analysis
Recurrent VTE is defined as a composite of objectively confirmed proximal lower-limb DVT, upper-limb DVT, and PE, with symptomatic or unsuspected status incorporated according to the registry definition.
A composite endpoint should be interpreted at the level at which it was statistically tested. A significant comparison of the composite does not automatically establish that every individual component has the same magnitude of treatment effect. Conversely, a component with relatively few events may be estimated imprecisely even when the composite contains substantially more information.
Time frame and censoring
The registered time frame is 6 months. If a time-to-event analysis is used, participants who have not experienced recurrent VTE by the end of observation can contribute follow-up information without necessarily being treated as having experienced the endpoint.
This is the statistical role of censoring in survival analysis: the analysis uses the information available up to the last relevant observation. The exact censoring rules, however, are not reported in the provided statistical-analysis data.
Why the endpoint definition matters
Statistical analysis cannot compensate for an ambiguous outcome definition. The registry's definition identifies objectively confirmed recurrent VTE and specifies the clinical events that comprise the endpoint. This creates the foundation for consistent event ascertainment across the randomized groups.
8. Statistical Methods Explained
Why does randomization matter?
Randomization determines treatment assignment independently of investigators' and participants' treatment preferences. In a sufficiently large randomized trial, this tends to balance measured and unmeasured prognostic factors across groups. The resulting treatment comparison is therefore fundamentally different from an observational comparison in which patients choose or are assigned treatment for clinical reasons.
What does a parallel-group design mean statistically?
Each participant belongs to one randomized treatment group for the primary comparison. The treatment effect is therefore estimated by comparing outcomes across the groups rather than comparing repeated treatment periods within the same participant.
Why is recurrent VTE a composite endpoint?
The registry groups several objectively confirmed manifestations of recurrent venous thromboembolism into one primary efficacy outcome. A composite can provide a broader measure of treatment-related clinical events, but interpretation requires attention to the individual components and to whether the components are clinically coherent.
Why does the 6-month time frame matter?
A treatment comparison must specify when the outcome is assessed. A 6-month endpoint defines the observation window for the registered primary outcome. It prevents the primary question from becoming an undefined comparison over an arbitrary period of follow-up.
Why can't an effect estimate be inferred from the registry record?
The ClinicalTrials.gov record states that results are posted, but the formal statistical-analysis array is empty. Without the actual treatment-group recurrent-VTE event data and the prespecified inferential procedure, an effect estimate cannot be responsibly reconstructed from the ClinicalTrials.gov record.
Why is randomization not the same as statistical significance?
Randomization creates the design needed for a causal treatment comparison; statistical significance is an inferential conclusion about the observed data under a specified statistical model or test. A randomized design can therefore exist without the ClinicalTrials.gov record containing a formal p-value or confidence interval.
9. What the Missing Statistical Analysis Means
What is known
The study is randomized, parallel, phase 3, has two interventions, and has a registered primary endpoint of recurrent VTE over 6 months.
What is not reported
No formal efficacy effect estimate, confidence interval, p-value, model specification, or hypothesis-test result appears in the ClinicalTrials.gov record.
What can be interpreted
The structure of the primary endpoint and the statistical implications of randomized comparative treatment assignment can be explained.
What should not be inferred
Whether apixaban was statistically superior, non-inferior, or inferior to dalteparin cannot be determined from the registry-reported statistical-analysis field alone.
This is a useful distinction in clinical-trial statistics. A registry record is not necessarily a complete statistical report. The presence of a posted outcome measure establishes that an outcome has been reported, but a rigorous interpretation of an efficacy comparison requires the event data, estimand, analysis population, statistical method, uncertainty measure, and hypothesis-testing framework used for that comparison.
10. Analysis Populations and Missing Data
The ClinicalTrials.gov record does not specify an intention-to-treat population, per-protocol population, modified intention-to-treat population, safety population definition, or other formal analysis population for recurrent VTE.
Likewise, the ClinicalTrials.gov record does not specify a missing-data or imputation strategy. For a time-to-event or fixed-time recurrent-VTE endpoint, this can matter because participants may have incomplete observation, withdrawal, loss to follow-up, or other circumstances affecting endpoint ascertainment.
11. Multiplicity, Interim Analysis, and Other Design Features
The ClinicalTrials.gov record does not report a multiplicity strategy, interim-analysis plan, alpha-spending procedure, non-inferiority margin, crossover rule, Bayesian method, or formal hierarchy among multiple efficacy endpoints.
These omissions matter because statistical conclusions depend not only on the endpoint but also on how the trial controlled type I error and how repeated looks or multiple hypotheses were handled. None of these procedures should be inferred merely because the study was phase 3 or randomized.
| Design topic | Information in the ClinicalTrials.gov record | Interpretation |
|---|---|---|
| Randomization | Reported | Randomized allocation is a core design feature. |
| Parallel design | Reported | Two randomized treatment groups are compared. |
| Masking | None | The registry identifies the study as unmasked. |
| Multiplicity | Not reported | No multiplicity procedure can be attributed to the trial from the ClinicalTrials.gov record. |
| Interim analysis | Not reported | No interim-analysis or alpha-spending method can be inferred. |
| Non-inferiority margin | Not reported | No margin can be reported or inferred. |
| Bayesian methods | Not reported | No Bayesian analysis is identified. |
| Missing-data method | Not reported | No imputation strategy is identified. |
12. Limitations and Interpretation Issues
- Incomplete statistical-analysis information: the ClinicalTrials.gov record contains no formal statistical-analysis entries, so treatment-effect estimates and inferential results cannot be reported from the ClinicalTrials.gov record.
- Composite endpoint: recurrent VTE combines several clinically distinct events. A composite result should not automatically be interpreted as demonstrating the same effect on every component.
- Endpoint type: the registry classifies the primary endpoint as "Other / unclear." The clinical definition is available, but the statistical estimand is not specified in the ClinicalTrials.gov record.
- Analysis population: the ClinicalTrials.gov record does not define the formal efficacy analysis population.
- Censoring and follow-up: although the primary time frame is 6 months, the registry-reported statistical-analysis field does not provide detailed censoring rules or follow-up procedures.
- Multiplicity: no type-I-error strategy is provided in the ClinicalTrials.gov record.
- Safety interpretation: serious-adverse-event counts are available by arm, but no formal statistical safety comparison is reported.
- External results: this page intentionally does not import numerical efficacy results from publications because the task requires all trial numbers to come from the ClinicalTrials.gov record.
13. Why This Trial Matters Statistically
CARAVAGGIO provides a useful teaching case because the ClinicalTrials.gov record contains the core structure of a randomized phase 3 comparative trial while leaving the formal inferential analysis unspecified. That creates an opportunity to distinguish trial design, endpoint definition, statistical estimand, and reported inference rather than treating them as interchangeable.
| Concept | How it appears in CARAVAGGIO |
|---|---|
| Randomization | The allocation is explicitly recorded as randomized. |
| Parallel-group design | The registry identifies a parallel design with two interventions. |
| Composite endpoint | Recurrent VTE combines proximal lower-limb DVT, upper-limb DVT, and PE according to the registered definition. |
| Fixed time frame | The primary endpoint is assessed over 6 months. |
| Event ascertainment | The primary outcome is defined as objectively confirmed recurrent VTE. |
| Analysis specification | The registry-reported statistical-analysis field contains no formal analysis entries. |
| Safety analysis | Serious adverse events are reported as affected/at-risk counts by arm. |
| Statistical interpretation | The design supports a comparative treatment analysis, but the ClinicalTrials.gov record does not establish its formal effect estimate or p-value. |
A well-defined randomized trial does not automatically tell the reader which statistical estimand or inferential procedure was used. The correct sequence is: identify the randomized comparison, define the endpoint, identify the analysis population and estimand, determine the statistical method, and then interpret the resulting estimate with its uncertainty.
For CARAVAGGIO, the first two steps are clearly represented in the ClinicalTrials.gov record. The later inferential details are not.
14. Statistical Concepts in This Trial
Learn more about the methods and concepts used to understand this trial:
15. Related Statistical Calculators
16. Trial Timeline
Trial start
The registry records the start date of the CARAVAGGIO trial as 2017-04-13.
Primary completion
The registry records the primary completion date as 2019-06-01.
Registry status
the ClinicalTrials.gov record identifies the study status as completed and indicates that results have been posted.
17. Sources
- ClinicalTrials.gov: CARAVAGGIO — NCT03045406.
- PubMed: PubMed record for PMID 34382385.
- PubMed: PubMed record for PMID 34260816.
- PubMed: PubMed record for PMID 32223112.
Continue through the Clinical Biostats knowledge graph
Connect randomized trial design to the statistical methods used to define endpoints, compare treatment groups, quantify uncertainty, and interpret clinical evidence.
18. Record Summary
CARAVAGGIO is a completed phase 3 randomized, parallel-group treatment trial with 1170.0 enrolled participants and two interventions: apixaban and dalteparin. Its registered primary efficacy endpoint is objectively confirmed recurrent venous thromboembolism over 6 months, defined as a composite of proximal lower-limb DVT, upper-limb DVT, and PE according to the registry definition.
The ClinicalTrials.gov record also reports serious adverse events of 255/576 in the apixaban arm and 276/579 in the dalteparin arm. Importantly, however, the registry-reported statistical-analysis field contains no formal analyses. The appropriate statistical interpretation is therefore to distinguish what the randomized design and endpoint definition establish from what requires the underlying posted statistical results: treatment-effect estimates, confidence intervals, p-values, analysis populations, and the exact inferential model.