Study of an Investigational Drug for the Prevention of Thrombosis-related Events Following Hip Replacement Surgery (ADVANCE-3): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of an Investigational Drug for the Prevention of Thrombosis-related Events Following Hip Replacement Surgery (ADVANCE-3) (NCT00423319), a randomized phase 3 study. The primary outcome is Rate of Composite of Adjudicated Venous Thromboembolic Event (VTE)-Related (Pulmonary Embolism and Symptomatic and Asymptomatic Deep Vein Thrombosis[DVT]) and All-cause Death During the Intended Treatment Period.

2. Clinical Question

This is a comparison of assigned treatment strategies in a postoperative setting, not an isolated comparison of individual clot types. The endpoint combines events identified through clinical assessment and a planned venogram, along with deaths. That design makes the primary result a measure of the composite as defined in the record; it does not by itself establish the separate effect on every component. The question is specific to participants having elective hip replacement and to the stated treatment period. Whether results apply to other procedures or populations is not reported in the registry record.

3. Trial Design

The registry identifies ADVANCE-3, NCT00423319, as a Phase 3 randomized, double-blind, active-controlled, parallel-group, multi-center study. Its official title describes an evaluation of apixaban versus injectable enoxaparin in people undergoing elective total hip replacement. Enrollment is listed as 5407, and the study status is completed. The registry lists a start of 2007-03 and primary completion in 2009-09.

The two treatment groups used matching placebo to support blinding despite different routes of administration. The apixaban regimen was 2.5 mg twice daily orally with an enoxaparin-placebo injection once daily. The enoxaparin regimen was 40 mg once daily subcutaneously with apixaban-placebo tablets twice daily. The record identifies random allocation and double masking, but details such as the method used to generate or conceal the allocation sequence are not reported in the registry record.

The primary endpoint was evaluated during the intended treatment period. For participants who received study drug, that period ended at the later of 2 days after the last dose or 38 days after the first dose; for randomized participants who did not receive study drug, it ended 38 days after randomization. The endpoint description states that a bilateral ascending contrast venogram was to be obtained on Day 35 (± 3). Further details about adherence, treatment discontinuation, or the distribution of follow-up time are not reported in the registry record.

There are 14 listed outcome measures and 12 analyses. Secondary outcomes cover additional venous thromboembolic events, deaths, pulmonary embolism and deep vein thrombosis categories, bleeding, serious adverse events, and neurologic adverse events. These entries do not establish which secondary results were prespecified as confirmatory or how they were controlled statistically. The registry record does not report the full protocol 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: A total of 5765 subjects were enrolled, and 5407 were randomized to double-blind study drug.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All randomized participants who, during the Intended Treatment Period, had an adjudicated and evaluable bilateral venogram, had an adjudicated venous thromboembolic event, or died of any cause.

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 Rate of Composite of Adjudicated Venous Thromboembolic Event (VTE)-Related (Pulmonary Embolism and Symptomatic and Asymptomatic Deep Vein Thrombosis[DVT]) and All-cause Death During the Intended Treatment Period, measured over Day 1 (first dose of study drug) to later of 2 days after last dose or 38 days after first dose.

The primary outcome is a composite event rate for adjudicated venous thromboembolic events and all-cause death during the intended treatment period. The specified thromboembolic events include pulmonary embolism and symptomatic or asymptomatic deep vein thrombosis. The record defines the event rate as the number of events divided by the number of patients evaluated and reports the outcome as a percentage of events or patients evaluated.

Use of a composite can bring several clinically relevant event types into one prespecified summary, allowing the comparison to address the combined burden represented by those components. But a composite result does not tell whether each component contributed similarly. A finding for the combined endpoint cannot be attributed to pulmonary embolism, deep vein thrombosis, or death separately without component-specific results. Such component details are listed among secondary outcomes, but the registry information considered for the primary comparison does not provide their numerical results.

The record describes a mandatory bilateral ascending contrast venogram planned for Day 35 (± 3), alongside adjudication of symptomatic events and deaths. This means the endpoint includes events found through scheduled imaging as well as clinically apparent events. The primary population includes randomized participants with an adjudicated and evaluable venogram, an adjudicated venous thromboembolic event, or death during the intended treatment period. It is therefore important to distinguish the endpoint population from all randomized participants and from the safety populations. The registry does not report detailed rules for handling unevaluable venograms or other missing endpoint information.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a Farrington-Manning test for the primary comparison and expresses the analysis as both a risk ratio and a risk difference. These statistics describe the same two-group comparison on different scales. A risk ratio compares event rates proportionally; a risk difference expresses their absolute separation on the outcome scale. The scales answer related but non-identical questions, so a result should be interpreted with attention to which measure is being reported.

What does a confidence interval communicate?

A confidence interval is a range of values generated by a statistical procedure that, under its assumptions and repeated use, has the stated coverage. In a single trial, it communicates uncertainty about the estimated comparison. Narrower intervals generally indicate greater precision, while wider intervals leave more plausible values compatible with the data and method. An interval is not a range containing a specified proportion of individual participant outcomes, and it is not a guarantee about the true effect.

For a risk ratio, the null value is one, representing equal rates on the ratio scale. For a risk difference, the null value is zero, representing no absolute difference. A confidence interval that excludes the relevant null value is commonly associated with evidence against that null in a two-sided test at a corresponding level, subject to the method and rounding. In this registry entry, the reported risk-ratio interval excludes one. The risk-difference interval is reported as -3.54 to 1.50, which includes zero, while its p-value is reported as < 0.001. Those two pieces of information appear inconsistent if they refer to the same contrast and inferential procedure. The record offers no explanation, and the figures should be presented as posted rather than altered.

Why report more than one effect scale?

Relative measures can communicate how rates compare proportionally, while absolute measures describe the difference in event rates on the original scale. Relative estimates can appear substantial even when the absolute event rates are low, so viewing both scales can make the result easier to assess. Neither scale alone describes all aspects of the outcome. Here, the endpoint is a composite and includes both imaging-detected and symptomatic thrombosis, pulmonary embolism, and all-cause death. The meaning of either measure depends on that definition and the evaluated population.

What does the p-value mean?

A p-value is calculated under a null hypothesis and assumptions about the data and analysis. It describes how unusual results at least as incompatible with the null would be if those assumptions and the null were correct. It does not give the probability that chance alone caused the result, the probability that a treatment hypothesis is true, or the magnitude of an effect. A small p-value can coexist with uncertainty about the practical importance of a difference, and interpretation also depends on how many outcomes and analyses were examined.

The reported method is the Farrington-Manning test. In general, this is a method for comparing proportions that can be used to test hypotheses about a risk difference, including settings where a nonzero margin is specified. The registry record does not report the full test specification, such as any margin used, nor does it give enough detail to reconstruct all analysis choices. The stated method and displayed estimates should therefore be understood as the registry’s reported analysis, not as evidence of unreported design features.

Why can event-rate analyses require care?

This endpoint is described as a rate of events divided by patients evaluated, and it is a composite of several types of outcomes. Such a summary can be useful for a defined comparison, but it may obscure variation among components. A scheduled venogram can identify asymptomatic deep vein thrombosis that would not necessarily be detected through symptoms alone. Differences in evaluability or ascertainment can also affect the set of participants contributing to an analysis. The record defines the endpoint population but does not provide detailed missing-data procedures or enough information to assess how unevaluable assessments were handled.

In general, over-dispersion matters when a statistical model assumes less variability than the data exhibit: standard errors may then be too small and intervals too narrow if that extra variation is ignored. This is a general statistical consideration, not a reported feature of this trial’s analysis. The registry does not report an over-dispersion assessment or a model-based count analysis for this endpoint. The reported Farrington-Manning method should not be conflated with a count model without additional documentation.

What can the planning illustrations tell readers?

Sample-size calculations are planning illustrations based on assumptions, not observed properties of a completed trial. The stated illustration assumptions use a two-sided significance level 0.05, equal allocation, and a normal approximation, with the target effect set as a fraction of the observed effect. For time-to-event outcomes, the stated convention is to report a total event count; for other outcomes, it is participants per arm for 80% or 90% power. Such calculations depend on the assumed effect and method. They are not observed power, do not establish the trial’s original planning rationale, and cannot substitute for the confidence intervals or outcome data. The original sample-size justification is not reported in the registry record.

10. Analyses Reported by the Sponsor

The registry lists the Farrington-Manning test as the sponsor-reported analysis for the primary comparison, with a risk ratio of 0.36, a 95% confidence interval from 0.22 to 0.54, and p < 0.001. It also lists a risk difference of -2.47, a 95% confidence interval from -3.54 to 1.50, and p < 0.001. The risk-difference interval and p-value appear internally inconsistent, and the registry record does not explain the mismatch. They should be retained as reported rather than repaired by inference.

The registry does not flag a discrepancy between the posted result and the simple unadjusted recalculation. That does not resolve the apparent inconsistency between the posted risk-difference interval and p-value. The recalculation is a descriptive check based on posted summary statistics, not a replacement for the sponsor’s inferential analysis or its full analysis dataset. No alternative adjusted result or additional analysis specification is reported in the registry record. The reported analysis and the limits of the posted detail should therefore remain distinct in interpretation.

11. Multiplicity, Missing Data and Analysis Populations

The record lists 14 outcome measures and 12 analyses, including several secondary comparisons concerning types of thrombosis, death, bleeding, serious adverse events, and neurologic events. Examining multiple outcomes or analyses raises a multiplicity question: as the number of statistical tests increases, the chance of at least one apparently unusual result can increase if each test is interpreted without adjustment. This general issue does not establish that any specific secondary finding is false, but it affects how confirmatory claims should be framed.

Whether a multiplicity adjustment was planned, which outcomes were designated primary or key secondary for testing, and whether a hierarchical testing sequence was used are not reported in the registry record. It is therefore not possible to infer from the count of posted outcomes whether family-wise error, false discovery, or another criterion was controlled. The primary comparison is identified, but the full prespecified testing strategy is not available in this record.

The primary population is defined around an evaluable venogram, an adjudicated venous thromboembolic event, or death during the intended treatment period. The number evaluated in each primary endpoint group is posted, but the registry does not describe a detailed missing-data strategy, reasons for unevaluable assessments, sensitivity analyses, or imputation methods. The effect of missing or unevaluable endpoint information therefore cannot be quantified from the registry entry alone.

Similarly, the record does not fully specify whether the primary comparison followed an intention-to-treat principle, a modified intention-to-treat definition, or another named analysis population. The endpoint description identifies the qualifying population, but does not provide the complete analysis-set rules. Details about interim analyses, stopping rules, and sample-size justification are also not reported in the registry record. Without those details, readers should avoid assumptions about monitoring, prespecification, or the treatment of participants who did not receive study drug.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: deep vein thrombosis (0.7%), dislocation of joint prosthesis (0.4%), joint dislocation (0.3%), pulmonary embolism (0.3%), device dislocation (0.3%).

The registry reports serious adverse events among 2673 participants at risk in the apixaban group and 2659 participants at risk in the enoxaparin group. Serious adverse events affected 184 participants in the apixaban group and 172 in the enoxaparin group. The record also reports other adverse events among the same respective at-risk groups: 1260 and 1315 participants. These counts are not the primary endpoint denominators and should not be combined with the efficacy analysis.

The most frequent serious adverse event terms listed include deep vein thrombosis at 0.7%, dislocation of joint prosthesis at 0.4%, joint dislocation at 0.3%, pulmonary embolism at 0.3%, device dislocation at 0.3%, and wound infection at 0.3%. The registry excerpt does not provide group-specific values for each of these listed terms, nor does it provide a full account of severity, timing, relatedness, or duration. The death counts are not reported.

Adverse-event counts are descriptive safety information. They do not, by themselves, establish that an event was caused by a study drug, and differences in counts should not be treated as adjusted comparative effects. The registry lists bleeding outcomes among its secondary measures, but numerical results for those outcomes are not reported in the registry record used for this interpretation.

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

The posted primary comparison describes fewer observed composite events in the apixaban group than in the enoxaparin group within the endpoint population and intended treatment period. The relative estimate is 0.36, with a 95% confidence interval from 0.22 to 0.54, and the reported p-value is < 0.001. The arm-level event-rate estimates are 1.39% and 3.86%, respectively. These results pertain to the specific composite, population definition, and observation window in the registry record.

The result does not isolate the contribution of each endpoint component. Because the composite includes adjudicated pulmonary embolism, symptomatic and asymptomatic deep vein thrombosis, and all-cause death, the overall comparison cannot establish a separate result for each. Nor does it establish effects beyond the intended treatment period, in people undergoing other procedures, or in populations not described by this trial. Such generalizability questions require evidence beyond the posted primary result.

The risk-difference estimate and interval are also reported, but the interval from -3.54 to 1.50 includes zero while the associated p-value is < 0.001. That mismatch limits interpretation of the absolute-scale inferential result unless clarified. It should not be resolved by assuming a typographical correction. The relative-scale result is internally more coherent as posted, but remains subject to the endpoint and analysis-population limitations.

Statistical evidence is not the same as a complete assessment of benefit and harm. The registry lists serious adverse events, bleeding, and other outcomes, but the available figures do not provide a full comparative safety analysis. The record also does not report enough detail about missing-data methods, multiplicity control, or the full analysis plan to assess all inferential choices. These limitations do not erase the posted result; they define what the result can support on its own.

No treatment recommendation follows from these statistics. They are a report of the comparison recorded for this trial, not a substitute for clinical judgment or a statement about an individual’s likely outcome. The registry record should be read as one source of trial evidence, with careful attention to the study population, endpoint construction, follow-up window, and uncertainty.

14. Why This Trial Matters Statistically

This trial illustrates how endpoint construction shapes a statistical comparison. Its primary outcome combines venous thromboembolic events with all-cause death and includes both symptomatic and asymptomatic deep vein thrombosis. A scheduled bilateral venogram was part of ascertainment, so the endpoint is not limited to events that brought a participant to clinical attention. Readers need to understand that definition before interpreting the rate or applying it to another setting.

The registry also demonstrates why reporting both relative and absolute measures can be useful. The risk ratio describes a proportional comparison, while the risk difference expresses an absolute comparison. Both can contribute to interpretation, but the posted risk-difference interval and p-value do not align in the usual way. This makes transparent reporting and internal consistency essential: a small p-value should not distract from checking whether estimates, intervals, and tests tell a coherent statistical story.

There is also a clear distinction between randomized enrollment, the primary endpoint population, and the safety population. The registry reports 5407 enrolled participants, endpoint analysis denominators of 1949.0 and 1917.0, and safety at-risk counts of 2673 and 2659. These figures answer different questions and should not be substituted for one another. A reader who treats them as interchangeable could misunderstand which participants contribute to each reported statistic.

Finally, the entry has many outcome measures and analyses, while details about multiplicity, missing data, and the full analysis plan are not reported. This is a reminder that a reported primary result is only one part of statistical appraisal. Estimates, uncertainty, endpoint definitions, analysis populations, and reporting completeness all matter when judging what a trial result can establish.

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 give the complete protocol or statistical analysis plan. Allocation concealment, detailed eligibility criteria, adherence, treatment discontinuation, and the rules for all analysis populations are not reported. The primary endpoint population is defined, but the record does not fully explain the number of participants who lacked an evaluable venogram, why assessments were unevaluable, or how missing endpoint data were handled.

The posted risk-difference confidence interval, -3.54 to 1.50, includes zero, whereas its associated p-value is < 0.001. No explanation is given for this inconsistency. It limits confidence in interpreting that absolute-scale test result and should be resolved from a more complete source before treating the interval and p-value as a coherent pair. The risk-ratio estimate and interval are reported separately.

The endpoint is a composite, so its overall rate does not show the individual contribution of each component. The registry names secondary outcome categories but does not provide the numerical component results in this record. A venogram-based outcome also depends on imaging assessment and adjudication; the registry description gives the intended process but not detailed completion or adjudication data.

Multiplicity control, interim analyses, stopping rules, and the original sample-size justification are not reported in the registry record. The planning illustrations use stated assumptions and the observed effect, so they cannot be read as the trial’s original design rationale or as observed power. The available safety information includes group-level counts and selected serious adverse event terms, but not complete event-level detail or a numerical comparison for each listed safety outcome. These gaps constrain conclusions about robustness, secondary findings, and the balance of outcomes.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Study of an Investigational Drug for the Prevention of Thrombosis-related Events Following Hip Replacement Surgery (ADVANCE-3) — NCT00423319.
• Registry record last updated 2014-05-14; status last verified 2014-04.
• Results first posted 2014-05-14.
• 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

NCT00423319 is a completed Phase 3 randomized, double-blind, active-controlled, parallel-group, multi-center study of apixaban and enoxaparin in elective total hip replacement. Enrollment is listed as 5407. The primary endpoint was an adjudicated composite of venous thromboembolic events and all-cause death during the intended treatment period.

The posted event rates were 1.39% for apixaban and 3.86% for enoxaparin. The reported risk ratio was 0.36, with a 95% confidence interval from 0.22 to 0.54 and p < 0.001. The reported risk-difference interval includes zero despite its p-value of < 0.001, an inconsistency not explained in the registry record. The entry also lists secondary efficacy and safety outcomes, but detailed multiplicity and missing-data methods are not reported.