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
SOLVE-TAVI was a randomized factorial trial evaluating two separate treatment dimensions in patients with aortic stenosis undergoing transcatheter aortic valve implantation: valve type and anesthesia strategy. The factorial structure allows these two comparisons to be studied within the same overall randomized population.
| Feature | SOLVE-TAVI |
|---|---|
| Trial acronym | SOLVE-TAVI |
| Clinical condition | Aortic Stenosis |
| Design | Randomized, factorial, unmasked |
| Allocation | RANDOMIZED |
| Design model | FACTORIAL |
| Masking | NONE |
| Primary purpose | TREATMENT |
| Enrollment | 444 |
| Number of arms | 4 |
| Trial phase | NA |
| Registry status | UNKNOWN |
| Start date | 2016-04 |
| Primary completion date | 2019-02-01 |
| Lead sponsor | Helios Health Institute GmbH |
2. Clinical Question
The trial addresses two treatment questions within patients with aortic stenosis undergoing TAVI. The first concerns the type of prosthetic valve; the second concerns the anesthesia strategy used during the procedure.
Population
Participants with aortic stenosis enrolled in a randomized TAVI trial.
Valve intervention
CoreValve Evolut R, a self-expandable device.
Valve comparator
Edwards Sapien 3, a balloon-expandable device.
Anesthesia comparison
Local anesthesia with conscious sedation versus general anesthesia.
The factorial structure is statistically important because the valve comparison and anesthesia comparison represent two distinct randomized factors. Rather than treating the four arms simply as four unrelated groups, a factorial analysis can evaluate the main effect of each factor while also considering whether the effect of one factor depends on the level of the other.
3. Trial Design
CoreValve Evolut R
- Device intervention
- Second-generation self-expandable valve
Edwards Sapien 3
- Device comparator
- Balloon-expandable valve
Local anesthesia with conscious sedation
- Procedure intervention
- Local anesthesia with conscious sedation
General anesthesia
- Procedure comparator
- General anesthesia
4. Understanding the Factorial Structure
A two-factor factorial trial crosses the levels of two interventions. Here, one factor is valve type and the other is anesthesia strategy. Conceptually, the four resulting combinations can be represented as a two-by-two treatment structure.
| Local anesthesia with conscious sedation | General anesthesia | |
|---|---|---|
| CoreValve Evolut R | Valve + local anesthesia | Valve + general anesthesia |
| Edwards Sapien 3 | Valve + local anesthesia | Valve + general anesthesia |
The statistical advantage of this design is efficiency. A single randomized population can provide information about two treatment factors. For example, the valve comparison can be examined across the anesthesia assignments, while the anesthesia comparison can be examined across the valve assignments.
5. Primary Endpoints
The registry identifies two primary endpoints, one concerning treatment-emergent adverse events associated with the device and one concerning treatment-emergent adverse events associated with anesthesia. Both are assessed at 30 day follow-up.
| Endpoint | Time frame | Registry definition |
|---|---|---|
| Incidence of Treatment-Emergent Adverse Events (Device) | 30 day follow-up | Number of participants with adverse events related to the treatment at 30-day follow-up consisting of all-cause mortality, stroke, moderate or severe prosthetic valve regurgitation, and permanent pacemaker implantation |
| Incidence of Treatment-Emergent Adverse Events (Anesthesia) | 30 day follow-up | Number of participants with adverse events related to the treatment at 30-day follow-up consisting of all-cause mortality, stroke, myocardial infarction, infection requiring antibiotic treatment, and acute kidney injury |
Device-related primary endpoint
The first endpoint counts participants experiencing an adverse event related to treatment at 30-day follow-up, with the endpoint consisting of all-cause mortality, stroke, moderate or severe prosthetic valve regurgitation, and permanent pacemaker implantation.
Anesthesia-related primary endpoint
The second endpoint similarly counts participants with adverse events related to treatment at 30-day follow-up, but the component events are all-cause mortality, stroke, myocardial infarction, infection requiring antibiotic treatment, and acute kidney injury.
6. Planned Analysis
The registry identifies the two 30-day participant-level adverse-event composites as primary endpoints, but no formal statistical analyses are posted in the ClinicalTrials.gov record. Consequently, the registry does not provide numerical treatment-effect estimates, confidence intervals, p-values, or event counts for these primary endpoints.
How the device endpoint would typically be analysed
For a binary 30-day composite endpoint defined as the presence or absence of at least one qualifying event, the primary comparison would typically begin with the proportion of participants experiencing the composite in the relevant randomized groups. A risk difference, risk ratio, or odds ratio could describe the relative and absolute contrast, with a confidence interval quantifying precision.
Because the trial has a factorial design, the central analysis would ordinarily distinguish the main effect of valve type from the main effect of anesthesia strategy. A factorial regression model can include both factors simultaneously and can also include a valve-by-anesthesia interaction term.
How the anesthesia endpoint would typically be analysed
The anesthesia composite is also a binary 30-day endpoint. The same general framework applies: estimate the proportion of participants with at least one qualifying event, compare randomized groups, and quantify uncertainty with a confidence interval.
The factorial structure means that the anesthesia comparison should not automatically be interpreted as a simple comparison of one of the four arms against another. The principal question is whether the probability of the anesthesia-related composite differs according to anesthesia assignment, while accounting for the crossed valve factor.
For a 30-day participant-level composite, Y = 1 indicates that the participant experienced at least one qualifying component event by the endpoint time.
Interaction analysis
A factorial model can test whether the effect of one factor changes according to the level of the other factor. In this trial, that means asking whether the valve comparison differs between local anesthesia with conscious sedation and general anesthesia, or conversely whether the anesthesia comparison differs between the two valve types.
An interaction term is conceptually different from comparing the four arm-specific percentages informally. The formal question is whether the treatment effects differ from one another beyond what would be expected from sampling variation.
7. Statistical Methodology
Factorial analysis
The defining statistical feature of SOLVE-TAVI is its factorial design. With two treatment factors, the analysis can estimate the effect associated with each factor while using the full randomized population efficiently.
The interaction term asks whether the effect of one factor changes according to the assigned level of the other factor.
If the interaction is small relative to the precision of the estimates, the main effects provide a compact description of the two treatment questions. If there is evidence of interaction, the four treatment combinations become more important because a single average main effect can conceal differences between combinations.
Binary outcome analysis
Both primary endpoints are defined as the number of participants with qualifying adverse events at 30-day follow-up. This makes the natural statistical scale binary at the participant level: each participant either meets the composite endpoint definition or does not.
Possible effect measures include the risk difference, risk ratio, and odds ratio. These measures are not interchangeable. The risk difference expresses the absolute difference in event probability, the risk ratio expresses a relative probability, and the odds ratio compares odds rather than probabilities.
Confidence intervals
A confidence interval would accompany an estimated treatment effect to show its statistical precision. A narrow interval indicates greater precision around the estimated effect, while a wide interval indicates greater uncertainty.
The interval should be interpreted together with the effect measure itself. For a risk difference, for example, the clinically relevant scale is percentage points; for a risk ratio, the reference value is 1; and for an odds ratio, the reference value is also 1.
Hypothesis testing
A p-value would ordinarily quantify the compatibility of the observed data with a specified null hypothesis under the chosen statistical model. It does not measure the size or clinical importance of an effect. An effect estimate and its confidence interval are therefore essential parts of the statistical description.
Intention-to-treat principle
Because the trial is randomized, an intention-to-treat analysis would ordinarily preserve assignment as the basis for the primary treatment comparison. This maintains the comparison created by randomization rather than redefining treatment groups according to subsequent exposure or adherence.
8. Statistical Methods Explained
Why use a factorial design?
A factorial design allows two treatment factors to be investigated in one randomized population. In SOLVE-TAVI, the factors are valve type and anesthesia strategy. This can be more statistically efficient than conducting two entirely separate randomized trials, provided the design and analysis appropriately account for the two factors.
What is a main effect?
A main effect is the average effect of one treatment factor across the levels of the other factor. For example, the main effect of valve type would compare CoreValve Evolut R with Edwards Sapien 3 while averaging across the anesthesia assignments in the factorial framework.
What is an interaction?
An interaction occurs when the effect of one factor changes according to the level of the other factor. If the difference between the two valve types were substantially different under local anesthesia with conscious sedation than under general anesthesia, that pattern would represent a valve-by-anesthesia interaction.
Why does the composite endpoint matter?
The device endpoint combines all-cause mortality, stroke, moderate or severe prosthetic valve regurgitation, and permanent pacemaker implantation. These events can have different clinical meanings and frequencies. A statistically significant composite therefore does not automatically imply that every individual component changed to the same extent.
What does a 30-day binary endpoint measure?
It measures whether a participant met the composite definition by the specified follow-up time. It is different from a time-to-event endpoint because the registry definition focuses on the number of participants with adverse events at 30-day follow-up rather than specifying time from randomization to the first event.
Why should the four arms not automatically be treated as unrelated groups?
The four groups arise from the crossing of two randomized factors. A four-group analysis can be useful descriptively, but the factorial question is usually more structured: estimate the valve effect, estimate the anesthesia effect, and evaluate whether the two effects interact.
What would a confidence interval add?
A confidence interval would show the precision of the estimated difference or ratio. Two analyses can have similar point estimates but substantially different uncertainty. The interval therefore provides information that a p-value alone cannot provide.
9. Endpoint-Specific Statistical Considerations
| Statistical issue | Relevance to SOLVE-TAVI |
|---|---|
| Binary outcome | Both primary endpoints are defined by the number of participants experiencing qualifying adverse events at 30-day follow-up. |
| Composite endpoint | Each primary endpoint combines multiple clinically distinct adverse events. |
| Factorial design | Valve type and anesthesia strategy are the two treatment dimensions. |
| Main effects | Permit separate estimation of the valve and anesthesia treatment effects. |
| Interaction | Tests whether the effect of one factor depends on the level of the other. |
| Randomization | Provides the foundation for comparing outcomes according to assigned treatment. |
| 30-day follow-up | Defines the registry time frame for both primary endpoints. |
10. Why Composite Endpoints Require Careful Interpretation
Composite endpoints are attractive because they can capture several clinically relevant events in a single outcome and may increase the total number of participants meeting the endpoint definition. But the interpretation depends on the component events.
Different clinical importance
All-cause mortality, stroke, prosthetic valve regurgitation, and pacemaker implantation are not interchangeable clinical outcomes.
Different frequencies
If one component occurs much more frequently than the others, the overall composite can be driven disproportionately by that component.
Direction of effects
The most informative composite analysis considers whether the component effects point in compatible directions rather than assuming they are identical.
Participant-level counting
A participant-level composite generally counts the participant once if at least one qualifying component occurs.
The same principle applies to the anesthesia endpoint, which combines all-cause mortality, stroke, myocardial infarction, infection requiring antibiotic treatment, and acute kidney injury.
11. Statistical Interpretation of the Factorial Design
The factorial structure creates an important distinction between arm comparisons and factor comparisons.
| Question | Statistical interpretation |
|---|---|
| Does valve type matter? | Compare the outcome across the CoreValve Evolut R versus Edwards Sapien 3 assignments while accounting for the anesthesia factor. |
| Does anesthesia strategy matter? | Compare the outcome across local anesthesia with conscious sedation versus general anesthesia while accounting for the valve factor. |
| Does the valve effect depend on anesthesia? | Evaluate the valve-by-anesthesia interaction. |
| Does anesthesia effect depend on valve? | The same interaction term addresses this reciprocal question. |
| What do the four arms show? | They describe the four combinations of the two randomized treatment factors. |
This distinction is central to interpreting a factorial clinical trial. If only the four individual arm comparisons are considered, the efficiency and scientific structure of the factorial design can be lost. Conversely, reducing everything to two main effects without examining interaction can conceal a meaningful dependence between the factors.
12. Multiplicity in a Factorial Trial
SOLVE-TAVI has two registered primary endpoints and two treatment factors. This creates several layers of statistical comparison: the device-related endpoint, the anesthesia-related endpoint, the valve factor, the anesthesia factor, and potentially their interaction.
The registry information identifies the primary endpoints and factorial design but does not provide a posted multiplicity strategy for these comparisons. Consequently, the statistical interpretation should distinguish the prespecified endpoint structure from any additional exploratory comparisons that might be conducted.
When several confirmatory hypotheses are tested, the overall type I error should be controlled according to a prespecified testing strategy.
A factorial design does not by itself eliminate multiplicity. Instead, it provides an efficient framework for studying multiple treatment factors. The confirmatory interpretation still depends on which hypotheses were designated primary and how simultaneous testing was handled.
13. What the Registry Reports — and What It Does Not
The ClinicalTrials.gov record identifies the trial design, population condition, four interventions, enrollment, dates, sponsor, and two primary 30-day endpoints. It does not post statistical estimates for those endpoints.
| Information category | Registry record |
|---|---|
| Trial acronym | SOLVE-TAVI |
| NCT identifier | NCT02737150 |
| Enrollment | 444 |
| Number of arms | 4 |
| Design model | FACTORIAL |
| Allocation | RANDOMIZED |
| Masking | NONE |
| Primary endpoints | 2 registered primary endpoints |
| Primary endpoint time frame | 30 day follow-up |
| Posted statistical analyses | None |
| Numerical primary endpoint estimates | Not reported in the ClinicalTrials.gov record |
| Primary endpoint confidence intervals | Not reported in the ClinicalTrials.gov record |
| Primary endpoint p-values | Not reported in the ClinicalTrials.gov record |
14. Important Limitations and Interpretation Issues
- No posted statistical estimates: the ClinicalTrials.gov record does not provide event counts, effect estimates, confidence intervals, or p-values for the two primary endpoints.
- Composite endpoints: each primary endpoint combines several adverse events with potentially different clinical importance and frequencies.
- Factorial interaction: a main-effect interpretation can be incomplete if the effect of valve type depends on anesthesia strategy, or vice versa.
- Multiplicity: the presence of two primary endpoints and multiple factorial hypotheses makes the confirmatory testing strategy important for interpretation.
- Unmasked design: the registry specifies no masking. For procedural and anesthesia interventions, lack of masking can affect some aspects of outcome ascertainment, particularly for outcomes requiring clinical judgment.
- 30-day horizon: the registered primary endpoints are assessed at 30-day follow-up, so they describe early outcomes rather than the full longer-term course of aortic valve treatment.
- Absence of posted results: without numerical results in the registry, the statistical magnitude and precision of the primary treatment effects cannot be characterized from the registry record.
15. Why This Trial Matters Statistically
SOLVE-TAVI is a useful teaching example because it combines randomized treatment assignment with a factorial design and participant-level composite safety or efficacy-style endpoints. The statistical questions are therefore broader than a simple comparison of two percentages.
| Concept | How it appears in SOLVE-TAVI |
|---|---|
| Randomization | The registry identifies the allocation as RANDOMIZED. |
| Factorial design | Two treatment dimensions are crossed: valve type and anesthesia strategy. |
| Four-arm structure | The registry specifies four arms. |
| Binary endpoint | The primary endpoints count participants with qualifying adverse events at 30-day follow-up. |
| Composite endpoint | Each primary endpoint combines multiple adverse-event components. |
| Main effects | The factorial structure supports separate questions about valve and anesthesia strategy. |
| Interaction | The design permits assessment of whether one treatment factor modifies the effect of the other. |
| Confidence intervals | Would quantify precision around estimated treatment effects if numerical results were reported. |
| Multiplicity | Multiple primary endpoints and factorial hypotheses require careful confirmatory interpretation. |
| 30-day follow-up | Defines the time frame for both registered primary endpoints. |
The trial also illustrates why statistical design cannot be separated from endpoint definition. A factorial design determines how treatment factors can be compared, while the composite endpoint determines what event is actually being compared. The final interpretation depends on both layers.
16. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The registry specifies a randomized factorial design with two primary participant-level adverse-event endpoints assessed at 30-day follow-up. The appropriate analysis would distinguish valve and anesthesia main effects and consider their interaction.
Clinical interpretation
The clinical questions concern two aspects of TAVI treatment: the choice between CoreValve Evolut R and Edwards Sapien 3, and the choice between local anesthesia with conscious sedation and general anesthesia.
These two perspectives should remain distinct. The statistical design determines how the treatment comparisons are estimated; the clinical interpretation determines what those treatment comparisons mean in the context of aortic stenosis and TAVI.
17. A Practical Analysis Framework
If complete numerical results were available, a statistically coherent analysis would proceed in several stages.
Apply the 30-day composite definitions
Determine whether each participant meets the registry definition for the device-related and anesthesia-related treatment-emergent adverse-event endpoints.
Analyze according to randomized assignment
Maintain the factorial treatment structure when estimating the treatment effects.
Separate valve and anesthesia effects
Estimate the treatment effect associated with each factor rather than relying only on unrelated pairwise comparisons of the four arms.
Ask whether the factors modify one another
Assess whether the valve effect changes according to anesthesia strategy or whether the anesthesia effect changes according to valve type.
Report effect estimates and confidence intervals
Present absolute and relative measures on an interpretable scale, together with their statistical precision.
18. Four Arms, Two Factors, Several Questions
One of the most useful ways to understand the design is to separate the number of arms from the number of scientific factors. There are 4 arms, but the registry describes two sets of interventions: two valve options and two anesthesia options.
CoreValve Evolut R versus Edwards Sapien 3
Local anesthesia with conscious sedation versus general anesthesia
This structure makes the trial especially useful for understanding why factorial trials should be analysed according to their treatment factors rather than being reduced immediately to a collection of pairwise arm comparisons.
19. Sources
- ClinicalTrials.gov: NCT02737150 — SOLVE-TAVI.
Continue through the Clinical Biostats statistical library
Explore related statistical concepts and calculators that can help explain factorial clinical-trial designs and their analysis.
20. Related Tutorials
Learn more about the methods used in this trial:
21. Related Calculators
22. Record Summary
SOLVE-TAVI is a randomized factorial trial in aortic stenosis that evaluates two treatment dimensions within the TAVI setting: CoreValve Evolut R versus Edwards Sapien 3 and local anesthesia with conscious sedation versus general anesthesia. The registry specifies 444 participants, 4 arms, randomized allocation, no masking, and two primary treatment-emergent adverse-event endpoints assessed at 30 day follow-up.
The principal statistical lesson is the factorial structure. The four treatment combinations should be understood as the crossing of two factors, allowing separate questions about valve type and anesthesia strategy while also raising the possibility of a treatment interaction. The two primary endpoints are participant-level composite outcomes, so their interpretation requires attention to both the overall composite and the individual component events.