Bioequivalence Study in Healthy Subjects, 2*5 mg Tablets Rivaroxaban Versus 1*10 mg Tablet Rivaroxaban: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Bioequivalence Study in Healthy Subjects, 2*5 mg Tablets Rivaroxaban Versus 1*10 mg Tablet Rivaroxaban (NCT01436526), a randomized phase 1 study. The primary outcome is Area Under the Plasma Concentration Versus Time Curve From Time Zero to Infinity After Single Dose (AUC) Incl. Bioequivalence (BE) Evaluation.

2. Clinical Question

This is a formulation comparison, not a comparison between rivaroxaban and an untreated or placebo group. Both regimens contain the same drug, and each study sequence includes both formulations. The crossover arrangement allows the comparison to use within-participant pharmacokinetic observations, while randomization determines which formulation is administered first. The study also listed safety and tolerability as secondary objectives. Its findings concern measured exposure and safety observations in the study setting; they do not directly answer questions about clinical outcomes in people undergoing surgery.

3. Trial Design

The registry classifies this as a Phase 1, randomized, open-label, two-way crossover study in healthy subjects. The planned comparison was between a single oral dose of rivaroxaban as 2*5 mg tablets and a single oral dose as 1*10 mg tablet, both given under fasting conditions. One sequence received the 2*5 mg formulation first and the 1*10 mg formulation second; the other sequence received them in the reverse order. A 7-day washout separated the periods.

The crossover structure matters for interpreting the reported groups. The formulations are not parallel treatment arms in which different participants receive only one regimen. The sequence descriptions indicate that participants were intended to receive both regimens, with order randomized. Accordingly, the arm-level pharmacokinetic counts should not be added to derive a number of distinct participants. For the primary pharmacokinetic endpoint, the registry identifies N=26 as valid for analysis. It separately reports 28 enrolled; a participant-flow explanation for the difference is not reported in the registry record.

The registry describes the study as single-dose and non-placebo-controlled. The study was to be conducted at one center in Germany, according to the brief summary. Its primary objective was bioequivalence assessment using pharmacokinetic data. Blood sampling was specified at multiple times from 0 min through 72 hours post administration, and the primary endpoint was AUC from time zero to infinity. Secondary measures included dose-normalized AUC and maximum concentration, mean residence time, time to maximum concentration, and terminal half-life.

Open-label status means masking was not used. The record does not report the randomization procedure, sequence allocation ratio, handling of period or sequence effects, or detailed analysis population rules. It also does not report a sample-size justification. The study’s short-term, controlled pharmacokinetic setting should be kept distinct from longer-term clinical effectiveness or safety questions.

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: 37 participants screened; 9 participants were screening failures; 28 participants were included in the study. Safety analysis: 27 individuals were analyzed in the group with 2\*5mg, 27 individuals were analyzed in the group with 1\*10mg.

Period 1

Period 2

Analysis population for the primary outcome, as defined in the registry: N=26 valid for pharmacokinetic analysis

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.

6. Primary Endpoint

The registry lists the primary outcome as Area Under the Plasma Concentration Versus Time Curve From Time Zero to Infinity After Single Dose (AUC) Incl. Bioequivalence (BE) Evaluation, measured over 0 min, 15 min, 30 min, 45 min, 1 hour, 1.5 hours, 2 hours, 2.5 hours, 3 hours, 4 hours, 6 hours, 8 hours, 12 hours, 15 hours, 24 hours, 36 hours, 48 hours and 72 hours post administration.

The primary endpoint is the area under the plasma concentration-versus-time curve from time zero to infinity after a single dose, reported in µg*hr/L. AUC represents systemic drug exposure: concentrations measured in blood samples are considered across the sampling period to estimate the area under the concentration-time profile. The registry lists sampling times from 0 min to 72 hours post administration and describes the primary summary as a geometric mean with geometric coefficient of variation.

For a formulation comparison, AUC is useful because it summarizes overall exposure rather than relying on a concentration measured at a single moment. The registry’s primary endpoint also specifies an evaluation of bioequivalence. In general, a ratio-based comparison expresses exposure under one formulation relative to exposure under the other. A ratio near the reference value can indicate similar estimated exposure, but a bioequivalence conclusion also depends on the prespecified acceptance limits and the uncertainty around the estimate. The acceptance limits are not reported in the registry record, so the numerical estimate alone cannot establish whether the protocol’s bioequivalence criteria were met.

The registry reports geometric mean AUC values of 1374.0 for 2*5 mg and 1268.0 for 1*10 mg, each with geometric coefficient of variation. Those descriptive summaries characterize the observed pharmacokinetic distributions; they are not, by themselves, the adjusted treatment comparison. The reported inferential analysis is an ANOVA estimate of the least-squares mean ratio with a confidence interval. Interpretation should therefore distinguish the descriptive arm summaries from that model-based comparison.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the crossover design contribute?

In a crossover study, participants receive more than one treatment condition in separate periods. Here, randomized sequences determined whether 2*5 mg or 1*10 mg was given first, and each sequence included the other formulation in the next period after a 7-day washout. This design permits a within-participant comparison: each participant can contribute observations under both formulations. In general, using within-participant information can reduce the influence of stable differences between people, because the comparison is not based only on contrasts between separate groups. It does not remove the need to consider period, sequence, carryover, measurement, and model assumptions.

The washout is intended to separate treatment periods, but the registry record does not report an analysis or diagnostic establishing whether carryover was absent. The detailed statistical model is not described beyond the reported ANOVA method. The design description alone should not be used to infer which period or sequence terms were included in that model.

What do geometric means and coefficients of variation describe?

The registry identifies the endpoint parameter as a geometric mean and the dispersion measure as a geometric coefficient of variation. In general, a geometric mean summarizes values on a multiplicative scale and is often used for measurements that are positively valued and vary proportionally. A coefficient of variation describes relative variability rather than variability in the original measurement units. These descriptive statistics characterize the observed AUC values for each formulation; they do not provide the full participant-level distribution or the adjusted comparison.

The posted arm summaries are 1374.0 with geometric coefficient of variation 29.3 for 2*5 mg, and 1268.0 with geometric coefficient of variation 30.7 for 1*10 mg. The registry reports N=26 valid for pharmacokinetic analysis. It does not provide participant-level observations, so the summaries cannot be used to reconstruct the within-participant pattern or independently verify the fitted model.

How should the least-squares mean ratio be read?

The reported ANOVA analysis gives a least-squares mean ratio, expressed as a percentage. In general, a least-squares mean is a model-based estimate that may account for terms included in the analysis model. A ratio compares the estimated endpoint under one condition with the estimate under another. The registry reports the ratio as 108.35%, with a 90% confidence interval from 101.59% to 115.57%. The precise treatment ordering and model specification should be interpreted according to the registry’s stated group ordering and analysis description; further model details are not reported in the registry record.

A confidence interval conveys uncertainty around an estimated quantity under the model and its assumptions. A narrower interval indicates greater precision than a wider interval for the same kind of estimate, but precision does not answer whether a prespecified clinical or regulatory criterion has been met. For bioequivalence, the interval is assessed against protocol-defined limits. Those limits are not reported in the registry record. The interval should not be judged against unstated bounds.

What does the p-value add?

The reported p-value is 0.044. In general, a p-value is calculated relative to a specified null hypothesis and statistical model; it is not the probability that the null hypothesis is true, nor a measure of the size or importance of an effect. The registry labels the analysis as non-inferiority or equivalence, but does not state the exact null hypothesis, margin, or decision procedure. Consequently, the estimate and confidence interval are more directly interpretable as descriptions of relative exposure and uncertainty than the p-value is as a stand-alone verdict.

The study’s planning significance level and other planning assumptions are described in the registry’s planning material, but no planning calculation results are included in the endpoint details. Such sample-size illustrations are planning exercises rather than observed power, and they should not be confused with the observed AUC estimate or its confidence interval.

What assumptions remain difficult to assess?

Statistical inference depends on the model being suitable for the data, the endpoint being measured consistently, and the analysis population being defined appropriately. In general, over-dispersion means that variability is greater than a chosen model would expect; ignoring it can make uncertainty estimates too optimistic. The registry gives geometric coefficients of variation but does not report a diagnostic assessment of model fit, distributional assumptions, outlier handling, or sensitivity analyses. There is therefore not enough detail in the registry record to evaluate those aspects independently.

The registry reports eight outcome measures and three analyses, while five secondary outcomes are listed. It does not describe a multiplicity adjustment or a hierarchy for interpreting multiple analyses. Without a stated plan, the reported p-value should not be read as if it were necessarily adjusted for all outcome and analysis opportunities. Nor does the record explain how missing or invalid pharmacokinetic observations were handled. These omissions do not change the values that are reported, but they limit a full assessment of the analysis strategy and its robustness.

10. Analyses Reported by the Sponsor

The registry’s inferential result is the sponsor-reported ANOVA comparison: a least-squares mean ratio of 108.35%, with a 90% confidence interval from 101.59% to 115.57% and p = 0.044. The record does not flag a discrepancy between the posted result and a recalculation. The sponsor’s analysis is therefore the relevant reported model-based comparison; it should not be replaced by arithmetic based on the separate geometric mean summaries.

Simple unadjusted recalculations, when available, are not equivalent to the sponsor’s crossover analysis. They may not account for the design or the analysis model. In this case, no separate recalculated result is available to interpret. The arm-level geometric means describe the observed AUC summaries, whereas the ANOVA output is the reported inferential comparison.

The sponsor analysis still cannot be used to verify the bioequivalence decision without the protocol’s acceptance limits and full decision rule. Those details are not reported in the registry record. The point estimate, confidence interval, and p-value should be read as the reported statistical result, not as a substitute for missing protocol criteria.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists eight outcome measures and reports three analyses; five secondary outcomes are identified. The primary AUC endpoint is explicitly described as including bioequivalence evaluation. The registry record does not report how the primary endpoint and secondary outcomes were ordered for formal inference, whether any multiplicity adjustment was planned, or how the three analyses relate to the full set of outcomes.

Multiplicity arises when several hypotheses or outcomes are examined: the chance of at least one apparently unusual result can change as the number of analyses increases. In general, a prespecified testing hierarchy or adjustment can clarify how evidence across multiple analyses is to be interpreted. No such method is reported in the registry record. The p-value for the AUC analysis should therefore be understood as the value attached to that reported analysis, not assumed to be adjusted across all endpoints or analyses.

The primary pharmacokinetic population is identified as N=26 valid for analysis, while the registry reports 28 enrolled. The record does not explain the participant flow, reasons observations may not have been valid, or whether exclusions occurred before or after treatment assignment. It also does not report a missing-data method, rules for incomplete concentration profiles, or sensitivity analyses for the primary endpoint. This leaves uncertainty about the handling of unavailable or unusable measurements.

The safety summaries use 27 at risk for each formulation, whereas the primary pharmacokinetic analysis population is N=26 valid. Because the study is crossover, these treatment-specific safety denominators should not be added to infer a number of distinct participants. The registry does not explain the relationship between the safety and pharmacokinetic analysis populations or whether safety collection covered both periods in the same way.

These reporting gaps limit assessment of whether exclusions, missing measurements, or multiple testing could affect interpretation. They do not establish that the analysis was inappropriate; the relevant methods are simply not reported in the registry record.

12. Safety Results

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

Other (non-serious) adverse events

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

Most frequent non-serious adverse event terms

The registry reports no serious adverse events among the 27 participants at risk for 2*5 mg and no serious adverse events among the 27 at risk for 1*10 mg. It also reports six other adverse events in each formulation group. These are event counts, not a comparison adjusted for the crossover design, exposure duration, or repeated observations. The record does not provide event terms for the listed non-serious events in the registry record used for this explanation.

In a crossover study, the treatment-specific safety groups may reflect exposure periods rather than independent sets of participants. The denominators should not be added to claim a total number of unique people. The registry does not explain whether the other adverse-event counts represent participants with events or events themselves, nor does it provide timing, severity, relatedness, or treatment-period detail in the reported summary. The record also does not report a death count; deaths are not reported in the registry record.

Zero reported serious adverse events in these treatment groups is a description of the posted observations, not proof that serious harms cannot occur or that the regimens have equivalent safety. The study was a single-dose pharmacokinetic comparison in healthy subjects, and the registry summary does not support conclusions about uncommon or longer-term events.

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

The result addresses a focused pharmacokinetic question: how AUC after a single fasted dose of rivaroxaban compares when administered as 2*5 mg tablets or as 1*10 mg tablet. The registry reports geometric mean AUC summaries and an ANOVA-based least-squares mean ratio with a 90% confidence interval. These quantities describe relative systemic exposure and the uncertainty of the estimated comparison in the analyzed study population.

The result does not, by itself, establish that the formulations meet the study’s bioequivalence criteria. The registry states that bioequivalence was the primary objective, but the acceptance limits and complete decision rule are not reported in the registry record. The estimate and interval cannot be judged against criteria that are not stated. The reported p-value also cannot independently supply those missing criteria.

The comparison is between two tablet presentations of the same drug, rather than between a drug and placebo or no treatment. Both were given as single doses under fasting conditions. The study population was healthy subjects, and the endpoint was a pharmacokinetic measure. It therefore does not directly establish comparative clinical outcomes in people undergoing knee or hip replacement surgery, or in other patient populations.

The safety counts describe events recorded during this study, but the small, short-duration exposure described in the record cannot settle questions about rare or delayed adverse events. The absence of reported serious adverse events is limited to the recorded study observations and should not be interpreted as a guarantee of safety.

In practical statistical terms, the record offers a model-based exposure estimate and a measure of its precision, but leaves important interpretive details unavailable: protocol acceptance limits, full model specification, multiplicity handling, and missing-data procedures. These limitations define what can be concluded from the posted results without implying that unreported procedures were or were not performed.

14. Why This Trial Matters Statistically

This study illustrates why bioequivalence is not assessed by looking only at two descriptive means. The geometric mean AUC values are useful summaries, but a formulation comparison also requires an inferential estimate and its uncertainty. The reported least-squares mean ratio and confidence interval provide that comparative perspective under the sponsor’s ANOVA analysis.

The randomized crossover design is statistically relevant because both formulations were administered across the study sequences. In general, within-participant comparisons can account for stable individual differences in exposure, which is valuable in pharmacokinetic research. At the same time, interpretation depends on period and sequence considerations, washout adequacy, analysis details, and handling of unusable measurements. The registry describes the design and washout but does not report all of those analytical details.

The result also demonstrates the importance of decision criteria. A ratio and confidence interval can be reported precisely, yet a bioequivalence conclusion depends on comparing the interval with prespecified limits. Those limits are not reported in the registry record. Statistical reporting is most interpretable when the estimated effect, its uncertainty, and the decision rule are all available together.

Finally, the setting defines the scope of the evidence. The endpoint concerns drug exposure after single doses in healthy subjects under fasting conditions. That is a narrower question than comparative clinical benefit or longer-term safety. Keeping the pharmacokinetic inference separate from broader clinical questions avoids overstating what this design and endpoint can answer.

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 record does not report the specific bioequivalence margins or acceptance criteria, despite identifying bioequivalence as the primary objective. Without those criteria, the reported ratio and 90% confidence interval cannot be used to independently verify whether the study met its planned decision rule.

The pharmacokinetic analysis population is given as N=26 valid, while 28 were enrolled. The record does not describe the participant flow or why the valid analysis population differs from enrollment. It does not report how missing or invalid concentration measurements were handled, how AUC extrapolation was performed, or whether sensitivity analyses were conducted. These details are relevant to assessing the reliability of an exposure estimate from sampled concentration profiles.

The statistical model is identified as ANOVA, but its terms and assumptions are not described. The registry does not report how sequence, period, or potential carryover were handled, whether outliers were assessed, or whether model diagnostics were performed. Randomization and a washout period are part of the design, but do not alone document these analytic choices.

The registry lists multiple outcomes and analyses without reporting a multiplicity strategy or inferential hierarchy. The p-value should therefore be interpreted only as reported for its analysis, without assuming adjustment across outcomes. A sample-size justification is also not reported in the registry record.

External interpretation is limited by the single-dose, fasting, healthy-subject setting and the pharmacokinetic focus. The study does not directly answer clinical outcome questions in surgical patients. Safety interpretation is constrained by the reported event summaries and lack of detailed event information. The record does not report deaths, and does not provide enough detail to assess rare or longer-term harms.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Bioequivalence Study in Healthy Subjects, 2*5 mg Tablets Rivaroxaban Versus 1*10 mg Tablet Rivaroxaban — NCT01436526.
• Registry record last updated 2015-04-22; status last verified 2015-04.
• Results first posted 2011-12-22.
• 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

This completed Phase 1 study compared single fasted doses of rivaroxaban given as 2*5 mg tablets and 1*10 mg tablet in a randomized, open-label, two-way crossover design. The registry reports 28 enrolled and identifies N=26 as valid for primary pharmacokinetic analysis. The primary endpoint was AUC from time zero to infinity.

The reported geometric mean AUC values were 1374.0 µg*hr/L for 2*5 mg and 1268.0 µg*hr/L for 1*10 mg. The ANOVA least-squares mean ratio was 108.35%, with a 90% confidence interval from 101.59% to 115.57% and p = 0.044. The registry does not report the bioequivalence acceptance limits, so the decision cannot be independently verified from these results alone.

No serious adverse events were reported in either treatment group; six other adverse events were recorded in each group. The registry does not report deaths in the available summary.