The Effect of Rivaroxaban in Sickle Cell Disease: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for The Effect of Rivaroxaban in Sickle Cell Disease (NCT02072668), a randomized phase 2 study. The primary outcome is Change From Baseline to 4 Weeks in Soluble Vascular Cell Adhesion Molecule-1 (VCAM-1).

2. Clinical Question

The clinical rationale recorded in the registry is that factor Xa inhibition might reduce inflammation, coagulation, and endothelial cell activation and improve microvascular blood flow. These are study hypotheses, not findings established by the design itself. The primary endpoint specifically measures soluble VCAM-1, one marker of endothelial activation; it does not directly measure every process named in the hypothesis.

The record also lists secondary measures, including inflammatory markers and sICAM, as well as TH1. It describes laser Doppler velocimetry of post-occlusive reactive hyperemia as the approach for assessing microvascular blood flow. The posted primary result therefore addresses a defined biomarker question within a broader set of biological questions. It should not be read as a direct assessment of every clinical consequence of sickle cell disease.

3. Trial Design

This was a phase 2 randomized, quadruple-masked crossover study sponsored by the University of North Carolina, Chapel Hill. The registry reports enrollment of 14 and status as completed. The record lists sickle cell anemia and sickle cell-beta0-thalassemia as conditions.

In a crossover design, participants receive more than one treatment in sequence, so each participant can contribute information under both treatment conditions. This can allow a comparison within participants rather than relying only on differences between separate treatment groups. The study record describes two sequences: rivaroxaban followed by placebo, and placebo followed by rivaroxaban. Each treatment period lasted 4 weeks, with a 2-week washout between periods.

The record describes quadruple masking but does not specify in the listed information which roles were masked. It also does not report details of the randomization procedure, the allocation ratio, or the planned handling of treatment sequence and period effects. These details matter for understanding how treatment order was assigned and incorporated into analysis, but should not be inferred from the design label alone.

The primary endpoint was change from baseline to 4 weeks in soluble VCAM-1. The registry reports arm-level results for 13 participants under each treatment condition, and says that the primary analysis population included only participants who completed both interventions. Those summaries do not show whether each participant’s results were paired across treatment periods. A crossover comparison depends on that within-participant structure, so the arm summaries alone are not enough to reproduce a paired treatment estimate.

The study also listed secondary biomarker outcomes and other outcome measures. The registry record does not report a rationale for the enrollment target, a sample-size justification, interim analyses, or a detailed statistical analysis plan.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: 15 subjects signed informed consent and were successfully screened. One subject withdrew during Baseline and prior to the first intervention and data from this individual are included in the baseline characteristics. One subject entered the second intervention period but was lost to follow up before receiving the intervention.

First Intervention

Washout (2 Weeks)

Second Intervention

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Data reported only for those participants who completed both interventions.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: All participants who had a screening visit

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline to 4 Weeks in Soluble Vascular Cell Adhesion Molecule-1 (VCAM-1), measured over Baseline, 4 weeks.

Soluble VCAM-1 is a circulating marker associated with endothelial cell activation. The registry describes an assay for soluble VCAM-1 using a commercially available enzyme-linked immunosorbent assay (ELISA). The endpoint is the change from baseline to 4 weeks, measured in pg/mL and summarized as a mean with a 95% confidence interval.

Using a change score focuses the analysis on how the measured marker differs from its baseline value during a treatment period, rather than reporting only a measurement at the end of that period. In general, an endpoint of this kind can provide a quantitative way to assess whether a biological marker shifts during treatment. A confidence interval expresses the uncertainty around an estimated mean; it is not a range containing all individual participants’ changes.

VCAM-1 was relevant to the stated study hypothesis because endothelial activation was one of the processes the study aimed to assess. It is still a biomarker endpoint, not a direct measure of symptoms, complications, or long-term clinical outcomes. Results for this marker cannot by themselves establish effects on the other biological measures named in the study rationale.

The registry identifies the endpoint population as participants who completed both interventions. It does not report participant-level measurements, the distribution of change scores, or a paired treatment contrast with its confidence interval. The posted arm summaries therefore describe mean changes for each treatment condition, but do not provide all the information needed to assess a crossover contrast independently.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the crossover design change?

In a parallel-group study, participants generally receive one assigned treatment, and the analysis compares outcomes between treatment groups. In a crossover study, participants receive treatments in sequence. The comparison can use each participant as their own control, which can account for stable differences between participants. The statistical analysis must preserve this pairing: the two measurements from a participant are related, not independent observations.

That relationship matters when estimating uncertainty. A paired analysis uses information about the within-participant differences. Separate means and confidence intervals for rivaroxaban and placebo do not reveal the covariance or the distribution of those paired differences. The record gives arm-level means, intervals, and participant counts, but not the participant-level data or the paired contrast. A simple unpaired calculation from these summaries would not recreate the crossover analysis.

What do the confidence intervals describe?

A confidence interval communicates the precision of an estimated quantity under the assumptions of the method used. The registry reports a 95% confidence interval for the mean change under each treatment condition. These intervals describe uncertainty around the corresponding arm-level means; they are not intervals for the difference between treatments, and they do not describe the range of individual responses.

The reported intervals are broad and include values in both directions relative to no change. That pattern is compatible with considerable uncertainty about the mean change within each condition. It does not, by itself, tell whether the treatment contrast is statistically or clinically important. To answer that question, the relevant quantity is a treatment contrast with its own uncertainty estimate, preferably from an analysis that accounts for the crossover pairing and study periods.

How should the sponsor-reported test be read?

The registry names a Student t-test followed by ANOVA. A t-test generally evaluates a contrast in means under assumptions about the data and the design. ANOVA is a framework for comparing means and can accommodate factors in a model, depending on how it is specified. The registry record does not provide enough detail to determine exactly how these methods were applied, whether the test was paired, or which terms were included in the ANOVA.

For a crossover trial, period and treatment sequence can matter because outcomes may change over time, and treatment order is part of the design. A suitable analysis needs to reflect the repeated measurements and the way participants received the interventions. The record does not report the model specification, assumptions checks, or treatment of period and sequence. Therefore, the method name alone does not establish whether every feature of the design was incorporated.

A p-value describes how compatible the observed test statistic is with a null model under the analysis assumptions. It is not the probability that the null hypothesis is true, nor does it measure the probability that a result will replicate. A p-value also does not quantify effect size or show that an absence of statistical evidence is evidence of no effect. Interpretation should be paired with the estimated treatment contrast and its confidence interval; those comparative quantities are not reported in the registry result.

Why is a summary-statistic re-calculation unavailable?

A re-calculation from published summary statistics is useful only when those summaries contain the information needed for the intended comparison. Here, the endpoint is continuous and the design is crossover. The posted means and confidence intervals are separated by treatment condition, but no paired difference, standard deviation of paired differences, or participant-level data are reported. The registry also flags the two-arm continuous comparison as not estimable from the available values.

As a result, an unadjusted independent-groups calculation would impose a structure that does not match the crossover design. It could produce a numerical answer, but that answer would not reconstruct the sponsor’s paired analysis and could misstate its uncertainty. No such re-calculated treatment contrast should be presented as if it were the study’s estimate.

What does multiplicity mean in this study?

Multiplicity arises when a study evaluates several outcomes or makes several statistical comparisons. When many tests are considered, the chance of at least one apparently notable result can increase if each test is interpreted without regard to the others. Statistical approaches can address this in different ways, but the appropriate choice depends on the study aims and analysis plan.

The registry lists 17 outcome measures, 11 analyses, and 15 posted secondary outcomes. These counts are not explained further in the record. It does not report whether a multiplicity adjustment was planned or applied, whether the primary endpoint had a prespecified testing hierarchy, or how secondary findings should be interpreted. No correction or hierarchy should be assumed. The primary endpoint and secondary outcomes should therefore be distinguished, and secondary results should be treated cautiously unless their analysis plan is available.

What assumptions and analyses are not documented?

General considerations for continuous outcomes include the distribution of observations, the influence of outliers, and whether the chosen model adequately represents variation and dependence. For crossover data, the analysis also needs to address repeated observations and may need to account for period or sequence. These are general statistical considerations, not claims that any particular assumption was violated in this study.

The registry record does not report diagnostics, missing-data methods, sensitivity analyses, or a detailed account of the analysis population beyond the statement that primary data were reported only for participants who completed both interventions. The record also does not describe interim analyses or the rationale for the planned enrollment. Without those details, readers cannot fully assess model adequacy, robustness, or the handling of incomplete participation.

10. Analyses Reported by the Sponsor

The registry reports the sponsor’s comparison as a Student t-test followed by ANOVA. It provides a p-value but does not list a treatment-effect estimate or a confidence interval for the comparison. The arm-level means and confidence intervals are not enough to reconstruct the paired treatment contrast in this crossover study.

The summary-statistic re-calculation is therefore not estimable from the posted continuous-outcome values for a two-arm comparison. The issue is not that a different result was calculated and disagreed with the sponsor; the paired information needed for a valid re-calculation is absent. Treating the arm summaries as if they came from independent groups would not account for the crossover structure.

The registry record does not flag a discrepancy between the sponsor’s analysis and a re-calculation. The sponsor’s reported test is consequently the only comparative analysis available in the record, although its interpretation is limited by the lack of a reported comparative estimate, confidence interval, and model details. The sponsor-reported analysis should not be replaced with an unsupported summary-based calculation, nor should its p-value be taken as proof of equivalence or absence of an effect.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists multiple outcome measures, including the primary VCAM-1 endpoint and secondary biomarkers of inflammation and endothelial activation. It also reports 17 outcome measures, 11 analyses, and 15 posted secondary outcomes. The record does not explain the difference between these counts or identify a complete analysis hierarchy.

When several outcomes or comparisons are tested, multiplicity can make isolated findings harder to interpret. A prespecified plan may identify a primary endpoint, define an order for testing, or use an adjustment. The registry record does not report whether any such approach was specified or used. No multiplicity correction should therefore be inferred from the number of posted outcomes or from the named test alone.

The primary endpoint population is described as participants who completed both interventions. The record does not give a full account of the analysis population for each secondary outcome, the reasons for any missing measurements, or the methods used to handle missing data. It also does not report a sensitivity analysis comparing alternative assumptions about incomplete data. A completer-based result may not represent everyone enrolled if completion is related to outcome, but the registry does not provide information to assess whether that concern affected this study.

The record does not report interim analyses, a sample-size justification, or a detailed statistical analysis plan. It is therefore not possible to determine from the registry whether the outcome selection, analysis population, or multiplicity approach followed a prespecified plan. These omissions do not establish that the analysis was inappropriate; they limit independent assessment of how the results were selected and interpreted.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: painful crisis with bacteremia (7.7%), pneumonia (7.1%).

The registry reports serious adverse events for the rivaroxaban and placebo treatment conditions. For rivaroxaban, 14 participants were listed as at risk for serious adverse events and 1 was affected; for placebo, 13 were at risk and 1 was affected. Deaths were reported as 0 in each condition. These are descriptive counts from the registry record and do not establish whether an event was caused by a treatment.

For other adverse events, the record lists 14 participants at risk and 8 affected under rivaroxaban, and 13 at risk and 7 affected under placebo. The record does not provide event-level detail in the listed summary sufficient to assess severity, timing, recurrence, or participant overlap across events. The group counts should not be interpreted as a randomized comparison of risk without the corresponding analysis and definitions.

The top serious-event terms are listed as painful crisis with bacteremia at 7.7% and pneumonia at 7.1%. The registry summary does not identify those terms by treatment condition in the listed information, so they should not be assigned to one condition. Adverse-event reporting describes observed events during the study; it does not by itself establish comparative safety or causality.

13. What the Result Does, and Does Not, Mean

The registry documents a randomized, quadruple-masked crossover study of rivaroxaban and placebo in sickle cell disease, with soluble VCAM-1 change as the primary endpoint. The posted arm summaries show mean changes of 40.9 pg/mL during rivaroxaban and 10.7 pg/mL during placebo, with broad 95% confidence intervals. The sponsor also reported a statistical test, but the record does not give a comparative effect estimate and confidence interval that can be independently checked from the summary data.

These results do not establish that rivaroxaban changes soluble VCAM-1 relative to placebo. They also do not establish that the treatments are equivalent or that a difference is absent. The uncertainty in the arm-level means, the lack of a reported paired treatment contrast, and the missing model details constrain what can be concluded from the registry result.

The endpoint is a biomarker intended to reflect endothelial activation. It is not a direct measure of clinical benefit, symptoms, or long-term complications. The study rationale included other biological processes and microvascular blood flow, but the primary endpoint alone cannot answer all those questions. Secondary measures require their own results and careful interpretation, particularly given the number of outcomes and the absence of a reported multiplicity plan.

The adverse-event counts are descriptive and do not show that an event was caused by either treatment. Likewise, the registry status of completed describes the study record, not the completeness or strength of evidence for every outcome.

Overall, the posted information is best understood as a small phase 2 biomarker study with an imprecise primary endpoint summary and an incompletely documented crossover analysis. It can inform discussion of the study’s measured outcomes, but it does not settle broader questions about treatment effects in sickle cell disease.

14. Why This Trial Matters Statistically

This study raises several statistical issues that are important when reading clinical research. The crossover design means the treatment comparison depends on within-participant data, not merely on separate arm-level means. That distinction matters because paired measurements can yield a different estimate of uncertainty from an analysis that treats treatment conditions as independent groups.

The registry reports mean changes and confidence intervals for each condition, but not the paired contrast. This illustrates why readers need to distinguish descriptive summaries from the estimate that answers the trial’s comparative question. A p-value without a reported effect estimate and confidence interval is also limited: it does not communicate the size or precision of a treatment difference.

The trial also assesses several biological outcomes. That breadth can be valuable for exploring mechanisms, but it raises questions about prespecification and multiplicity. Without a reported analysis plan, it is not possible to tell how the primary and secondary findings were prioritized or whether testing across outcomes was addressed.

Finally, the primary result is restricted to participants who completed both interventions. That population definition is important in a crossover study, where both treatment periods contribute to the endpoint. The registry does not describe missing-data handling or the characteristics of participants without complete endpoint data. These details help determine how precisely the findings apply beyond the analyzed participants.

For biostatisticians and clinical researchers, the record is a reminder to align the interpretation with the design, population, and level of detail actually reported. The trial’s result is not just a pair of means: its meaning depends on the crossover contrast, its uncertainty, the handling of multiple outcomes, and the completeness of the analysis report.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry reports arm-level means and confidence intervals for the primary endpoint but does not provide the paired treatment difference or its confidence interval. Because participants received both treatments, the within-participant relationship is central to a crossover comparison. Without paired summaries or participant-level results, an independent re-calculation cannot reproduce that comparison.

The record names a Student t-test followed by ANOVA but does not specify the model terms, whether the t-test was paired, or how treatment period and sequence were addressed. It does not report assessment of carryover, model diagnostics, sensitivity analyses, or the assumptions used for the primary analysis. These are reporting gaps; they do not demonstrate that the analysis failed to account for those issues.

The primary endpoint results were reported only for participants who completed both interventions. The registry does not provide reasons for incomplete participation, an account of missing endpoint measurements, or a method for addressing missing data. The extent to which the completer population differs from those enrolled cannot be evaluated from the record.

The study included multiple outcome measures, but the record does not report a multiplicity strategy or a full prespecified hierarchy. It also does not give the sample-size justification, interim-analysis information, or a detailed statistical analysis plan. Those omissions make it difficult to judge whether the study was designed to estimate the primary contrast with a particular precision or how secondary results were intended to be used.

Finally, the primary outcome is a biomarker rather than a clinical endpoint, and the posted summary does not establish effects on clinical outcomes. The small enrollment and completer-only primary population further limit precision and generalizability. These limitations should be considered alongside the reported mean changes and confidence intervals rather than treated as evidence for a particular treatment effect.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: The Effect of Rivaroxaban in Sickle Cell Disease — NCT02072668.
• Registry record last updated 2020-04-13; status last verified 2020-03.
• Results first posted 2020-04-13.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT02072668, titled The Effect of Rivaroxaban in Sickle Cell Disease, is listed as a completed phase 2 randomized, quadruple-masked crossover study sponsored by the University of North Carolina, Chapel Hill. The registry reports enrollment of 14 and includes sickle cell anemia and sickle cell-beta0-thalassemia as conditions.

Participants received rivaroxaban 20 mg per day and placebo for 4 weeks each, separated by a 2-week washout. The primary endpoint was change from baseline to 4 weeks in soluble VCAM-1, reported in pg/mL as a mean with a 95% confidence interval. The registry states that primary data were reported only for participants who completed both interventions.

The posted endpoint summaries give mean changes of 40.9 pg/mL for rivaroxaban and 10.7 pg/mL for placebo. The record names a Student t-test followed by ANOVA but does not report a comparative estimate and confidence interval or enough paired information for an independent summary-statistic re-calculation. Secondary outcomes and adverse events are also listed, while details of multiplicity, missing-data handling, and the statistical analysis plan are not reported in the registry record.