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
TEMPO3:4 was a randomized, parallel, phase 3 trial evaluating tolvaptan versus placebo in autosomal dominant polycystic kidney disease (ADPKD). The study was completed after beginning in January 2007, with primary completion in January 2012.
| Feature | TEMPO3:4 |
|---|---|
| Trial name | TEMPO3:4 |
| ClinicalTrials.gov identifier | NCT00428948 |
| Phase | Phase 3 |
| Condition | Polycystic Kidney Disease, Autosomal Dominant |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 1,445 |
| Interventions | Tolvaptan versus placebo |
| Start | January 2007 |
| Primary completion | January 2012 |
| Lead sponsor | Otsuka Pharmaceutical Development & Commercialization, Inc. |
2. Clinical Question
The central statistical question was whether treatment with tolvaptan produced a different change in total kidney volume over the first 36 months than placebo in participants with ADPKD.
Population
Participants enrolled in the phase 3 TEMPO3:4 study for autosomal dominant polycystic kidney disease.
Intervention
Tolvaptan.
Comparator
Placebo.
Primary question
Does tolvaptan differ from placebo in percentage change per year in total kidney volume from baseline to Month 36?
3. Trial Design
Tolvaptan
- Study intervention: tolvaptan
- Compared with placebo across the registered endpoints
Placebo
- Study intervention: placebo
- Reference group for the randomized comparisons
Imaging assessment for the primary endpoint
Total kidney volume was assessed using T1-weighted magnetic resonance images collected at each study site and sent to a central reviewing facility. At the central facility, blinded radiologists used proprietary software to measure the volume of both kidneys.
4. Endpoints
The registry lists one primary endpoint and six secondary endpoints with posted statistical analyses. The endpoints span longitudinal continuous outcomes, recurrent clinical events, renal pain, blood pressure, and a categorical treatment-related outcome.
| Role | Endpoint | Time frame | Endpoint type |
|---|---|---|---|
| Primary | Percentage Change Per Year in Total Kidney Volume From Baseline to Month 36 | Baseline to Month 36 | Binary |
| Secondary | Number of ADPKD Clinical Progression Events Per 100 Follow-up Years From Baseline to Month 36 | Baseline to Month 36 | Time-to-event |
| Secondary | Change in Renal Function Per Year From Week 3 to Month 36 | Week 3 to Month 36 | Continuous |
| Secondary | Change in Mean Arterial Blood Pressure Per Year in Non-hypertensive Participants From Baseline to Month 36 | Baseline to Month 36 | Binary |
| Secondary | Area Under the Concentration-time Curve of Change in Renal Pain From Baseline to Month 36 | At screening, Baseline, Day 1, every 4 months up to month 36/early tremination (ET), follow-up visit 1 and 2 | Continuous |
| Secondary | Number of Hypertensive Events Per 100 Follow-up Years in Non-hypertensive Participants From Baseline to Month 36 | Baseline to Month 36 | Time-to-event |
| Secondary | Percentage of Participants With a Clinically Sustained Decrease of Blood Pressure Leading to a Sustained Reduction in Antihypertensive Therapy From Baseline to Month 36 | Baseline to Month 36 | Binary |
5. Statistical Methodology
Primary endpoint: longitudinal mixed-effects model
The primary posted analysis compared the percentage change per year in total kidney volume from baseline to Month 36 using a linear mixed model. The analysis population was the intent-to-treat population defined as all randomized participants who had baseline and post-baseline observations of total kidney volume. Only participants with available data were included in the analysis.
The 95% two-sided confidence interval was −3.269 to −2.147, with P < 0.0001. The null hypothesis was that there was no difference between treatment groups in percentage change per year in total kidney volume.
The registry notes that the variance of the random-effect intercept was zero and resulted in a non-positive, definite variance-covariance matrix for the random effects. This is an important model-fitting detail: the planned mixed-model framework encountered a boundary estimate for one component of the random-effects structure.
Recurrent-event analysis
Two secondary endpoints counted clinical events over follow-up years: ADPKD clinical progression events and hypertensive events among non-hypertensive participants. These were analyzed using recurrent-event methods, with hazard ratio as the reported effect measure.
The registry reports hazard ratios for these analyses rather than simple proportions of participants with at least one event.
Renal-function slope analysis
Change in renal function per year from Week 3 to Month 36 was analyzed using a mixed-model approach. The reported effect was the difference in slope, derived from testing the time-by-treatment interaction in a linear mixed model in which both intercept and slope were fixed and random effects.
Mean arterial blood pressure slope
Change in mean arterial blood pressure per year among non-hypertensive participants was also evaluated using a mixed-model analysis. The treatment effect was expressed as a difference in slope, with the registry specifying placebo minus tolvaptan for this endpoint.
ANCOVA for renal pain
The area under the concentration-time curve of change in renal pain was analyzed using ANCOVA. The analysis included baseline renal pain as a covariate and incorporated treatment and baseline stratification factor interaction. The resulting effect measure was a mean difference in final values.
Cochran-Mantel-Haenszel analysis
The percentage of participants with a clinically sustained decrease in blood pressure leading to a sustained reduction in antihypertensive therapy was analyzed with a Cochran-Mantel-Haenszel test. The effect measure was a risk ratio comparing tolvaptan with placebo.
6. Primary Result: Total Kidney Volume
The primary endpoint was the percentage change per year in total kidney volume from baseline to Month 36. The posted analysis used a linear mixed model and reported a mean difference in final values between tolvaptan and placebo.
Mean difference in percentage change per year
95% CI: −3.269 to −2.147 · P < 0.0001
Superiority hypothesis; two-sided 95% confidence interval.
The estimated mean difference of −2.708 indicates that the percentage change per year in total kidney volume was estimated to be 2.708 percentage points lower in the tolvaptan group than in the placebo group, under the analysis framework reported by the registry.
The negative sign is meaningful because the effect was defined as a treatment-group difference. It does not mean that total kidney volume itself decreased by 2.708%; rather, it describes the between-group difference in the endpoint defined as percentage change per year.
The 95% confidence interval from −3.269 to −2.147 quantifies uncertainty around the estimated mean difference. Because the entire interval is below zero, the interval is consistent with a treatment-group difference favoring a lower percentage change per year in total kidney volume.
The P-value of <0.0001 addresses the null hypothesis of no difference. It does not measure the magnitude or clinical importance of the treatment effect. The estimate and confidence interval provide the information about magnitude and precision.
The analysis should also be interpreted in light of the model structure and the registry's note that the variance of the random-effect intercept was zero. That boundary estimate affected the random-effects covariance structure and is an important feature of the reported model fit.
7. Secondary Results
ADPKD clinical progression events
The number of ADPKD clinical progression events per 100 follow-up years from baseline to Month 36 was analyzed with a recurrent-event analysis in the intent-to-treat population.
Hazard ratio for recurrent clinical progression events
95% CI: 0.775–0.965 · P = 0.0095
Tolvaptan divided by placebo.
An HR of 0.865 means that the estimated hazard associated with recurrent ADPKD clinical progression events was approximately 86.5% of the corresponding hazard in the placebo group under the reported recurrent-event analysis. Expressed as a simple relative comparison, 1 − 0.865 = 0.135, so the estimate corresponds to an approximately 13.5% lower estimated hazard.
This is a relative hazard, not a 13.5% reduction in the number of participants experiencing an event and not a 13.5% absolute reduction in event probability. Because recurrent events are involved, the analysis concerns repeated clinical progression events over follow-up rather than only the first event.
The 95% CI of 0.775–0.965 describes statistical uncertainty around the estimated hazard ratio. It remains below 1. The P-value of 0.0095 tests the null hypothesis of no difference; it does not quantify the size of the treatment effect.
As with other hazard-based measures, interpretation depends on the event-analysis framework and censoring structure. The registry identifies this as a recurrent-event analysis rather than a conventional first-event Kaplan-Meier analysis.
Renal function per year
Change in renal function per year from Week 3 to Month 36 was evaluated among randomized participants with at least 4 months of follow-up and available data.
Difference in renal-function slope
95% CI: 0.597–1.357 · P < 0.0001
Derived from the time-by-treatment interaction in a linear mixed model.
The estimated difference in slope was 0.977, meaning the modeled annual change in renal function differed by 0.977 units between the treatment groups in the direction represented by the registry's treatment comparison.
This is a slope difference, not a single post-treatment measurement and not a hazard ratio. The analysis uses repeated measurements over time, so the treatment effect is represented through the time-by-treatment relationship.
The 95% CI of 0.597–1.357 gives the uncertainty around that slope difference. The P-value of <0.0001 tests the null hypothesis of no difference in the annual change in renal function; it does not indicate that the effect is clinically large simply because the P-value is small.
Mean arterial blood pressure in non-hypertensive participants
Mean arterial blood pressure per year from baseline to Month 36 was evaluated in non-hypertensive participants with at least 4 months of follow-up and available data. The registry specifies the effect as placebo minus tolvaptan.
Difference in blood-pressure slope
95% CI: −1.059–0.566 · P = 0.5520
Placebo minus tolvaptan.
Because the effect is defined as placebo minus tolvaptan, the negative point estimate indicates a slightly lower modeled annual slope in the placebo group relative to tolvaptan under that coding convention.
The confidence interval, −1.059 to 0.566, crosses zero. Thus the posted estimate is compatible with differences in either direction as well as with little or no between-group difference.
The P-value of 0.5520 does not measure the size of the observed difference. It is evidence from this test against the specified null hypothesis. The confidence interval is more informative for understanding the range of differences compatible with the analysis.
Renal pain
The area under the concentration-time curve of change in renal pain was analyzed with ANCOVA, adjusting for baseline renal pain and incorporating treatment and baseline stratification factor interaction.
Mean difference in renal pain AUC
95% CI: −0.20–0.03 · P = 0.1604
ANCOVA-adjusted comparison of tolvaptan versus placebo.
The estimated mean difference was −0.08. The negative value indicates a lower adjusted final-value measure in the tolvaptan group relative to placebo under the registry's comparison.
The 95% CI of −0.20 to 0.03 includes zero, so the data are compatible with a small difference in either direction. The P-value of 0.1604 is a test against the null hypothesis of no difference; it is not a measure of the magnitude of the observed estimate.
Because this was an ANCOVA, baseline renal pain was explicitly included as a covariate. This is different from simply comparing unadjusted final measurements.
Hypertensive events in non-hypertensive participants
Hypertensive events per 100 follow-up years from baseline to Month 36 were evaluated among non-hypertensive participants using recurrent-event analysis.
Hazard ratio for recurrent hypertensive events
95% CI: 0.805–1.233 · P = 0.9704
The estimated hazard ratio of 0.996 is very close to 1. Under the reported recurrent-event model, this corresponds to an estimated hazard in the tolvaptan group that was approximately 99.6% of the placebo-group hazard.
The 95% CI of 0.805–1.233 crosses 1 and includes materially different relative hazard values in both directions. The P-value of 0.9704 provides little evidence against the null hypothesis of no difference in recurrent hypertensive events.
The analysis concerns recurrent events per follow-up time rather than simply the percentage of participants who ever developed hypertension.
Sustained reduction in antihypertensive therapy
The percentage of participants with a clinically sustained decrease of blood pressure leading to a sustained reduction in antihypertensive therapy from baseline to Month 36 was analyzed using a Cochran-Mantel-Haenszel test.
Risk ratio for the sustained blood-pressure outcome
95% CI: 0.602–2.017 · P = 0.7532
Tolvaptan divided by placebo.
A risk ratio of 1.100 means that the observed risk of the specified sustained blood-pressure outcome was estimated to be 1.100 times that in the placebo group under the registry's analysis.
This is a relative risk measure, not an absolute percentage-point difference. The 95% CI of 0.602–2.017 is wide and includes 1, indicating substantial uncertainty about the relative effect. The P-value of 0.7532 tests the null hypothesis of no difference and does not measure the clinical size of the observed risk ratio.
8. Results Summary
The seven posted analyses show how different endpoint structures led to different statistical methods and effect measures. The primary endpoint used a mixed-effects model, while secondary outcomes used recurrent-event analysis, mixed models, ANCOVA, and a Cochran-Mantel-Haenszel test.
| Endpoint | Method | Effect | 95% CI | P-value |
|---|---|---|---|---|
| Percentage change per year in total kidney volume | Linear mixed model | Mean difference −2.708 | −3.269 to −2.147 | <0.0001 |
| ADPKD clinical progression events | Recurrent event analysis | HR 0.865 | 0.775–0.965 | 0.0095 |
| Change in renal function per year | Mixed-model analysis | Slope difference 0.977 | 0.597–1.357 | <0.0001 |
| Mean arterial blood pressure per year in non-hypertensive participants | Mixed-model analysis | Slope difference −0.246 | −1.059–0.566 | 0.5520 |
| Renal pain AUC | ANCOVA | Mean difference −0.08 | −0.20–0.03 | 0.1604 |
| Hypertensive events in non-hypertensive participants | Recurrent event analysis | HR 0.996 | 0.805–1.233 | 0.9704 |
| Sustained decrease in blood pressure leading to reduced antihypertensive therapy | Cochran-Mantel-Haenszel | RR 1.100 | 0.602–2.017 | 0.7532 |
9. Statistical Methods Explained
Why was a mixed-effects model used for the primary kidney-volume endpoint?
Total kidney volume was measured longitudinally, with observations available at multiple points from baseline through Month 36. A linear mixed model is designed for repeated measurements because it models the relationship between the outcome and time while accounting for correlation among observations from the same participant.
The registry specifically reports a linear mixed model for the primary analysis. It also notes that the variance of the random-effect intercept was zero, resulting in a non-positive, definite variance-covariance matrix for the random effects. This indicates that the fitted random-effects structure reached a boundary rather than behaving like a conventional model with a clearly positive random-intercept variance.
What does a mean difference of −2.708 mean?
The mean difference is a direct comparison on the scale of the primary endpoint. Because the analysis compares tolvaptan with placebo, −2.708 indicates a lower percentage change per year in total kidney volume for tolvaptan by 2.708 percentage points on that endpoint scale.
It should not be translated into a hazard ratio or risk ratio. Each effect measure has its own interpretation and statistical scale.
Why is the ADPKD clinical progression endpoint analyzed differently?
The endpoint counts clinical progression events per 100 follow-up years and is explicitly classified as a time-to-event outcome. Participants can contribute more than one event, so a recurrent-event framework is more appropriate than a simple binary comparison of whether each participant ever experienced an event.
What does an HR of 0.865 mean?
An HR of 0.865 indicates an estimated event hazard 13.5% lower for tolvaptan relative to placebo under the reported model, because 1 − 0.865 = 0.135. It does not mean that 13.5% fewer participants necessarily experienced an event, and it does not represent an absolute risk reduction.
Why was ANCOVA used for renal pain?
The renal-pain analysis incorporated baseline renal pain as a covariate. ANCOVA can improve precision when baseline values are prognostically related to the outcome by accounting for baseline differences within a prespecified regression framework rather than relying only on unadjusted final measurements.
Why does the confidence interval matter more than the P-value for effect size?
A P-value summarizes evidence against a specified null hypothesis. It does not tell the reader how large the treatment effect is. The confidence interval provides an interval estimate of the effect and therefore shows both the direction and precision of the estimate. For example, the renal-pain estimate of −0.08 has a 95% CI from −0.20 to 0.03, while the primary kidney-volume estimate of −2.708 has a 95% CI from −3.269 to −2.147.
10. Analysis Populations and Missing Data
The registry identifies the intent-to-treat population as the basis for the posted analyses, but the exact available-data requirements differ by endpoint.
| Endpoint group | Analysis population |
|---|---|
| Primary kidney-volume endpoint | All randomized participants who had baseline and post-baseline observations of total kidney volume; only participants with available data were included. |
| ADPKD clinical progression events | All randomized participants; only participants with available data were included. |
| Renal function | All randomized participants with at least 4 months of follow-up; only participants with available data were included. |
| Mean arterial blood pressure | All randomized participants with at least 4 months of follow-up; only non-hypertensive participants with available data were included. |
| Renal pain | All randomized participants with at least 4 months of follow-up; only participants with available data were included. |
| Hypertensive events | All randomized participants with at least 4 months of follow-up; only non-hypertensive participants with available data were included. |
| Antihypertensive therapy outcome | All randomized participants with at least 4 months of follow-up; only participants taking antihypertensive medication at baseline with available data were included. |
The registry describes these analyses as intent-to-treat analyses while also specifying endpoint-specific availability criteria. Therefore, "intent-to-treat" should not be read as meaning that every randomized participant necessarily contributed an observed value to every endpoint.
11. Stratification and Covariate Adjustment
The posted statistical analyses identify stratified analysis as a concept for the renal-pain endpoint and explicitly state that baseline renal pain was included as a covariate. The analysis was derived from ANCOVA with treatment, baseline stratification factor interaction, and baseline renal pain.
Covariate adjustment
Baseline renal pain was included directly in the renal-pain ANCOVA rather than treating the baseline measurement as irrelevant to the final analysis.
Stratified structure
The renal-pain analysis incorporated the interaction between treatment and the baseline stratification factor specified in the registry analysis description.
Covariate adjustment can improve statistical precision when baseline measurements are associated with the outcome. It does not turn an observational comparison into a randomized one; in this trial, the treatment comparison itself comes from randomized allocation.
12. Recurrent Events Versus First Events
The ADPKD clinical progression and hypertensive-event endpoints illustrate an important distinction in time-to-event analysis. A conventional survival analysis often focuses on the time to a participant's first event. A recurrent-event analysis instead recognizes that multiple events can occur during follow-up.
| Feature | First-event analysis | Recurrent-event analysis |
|---|---|---|
| Primary event concept | Time to first event | Repeated events over follow-up |
| Subsequent events | Usually no longer contribute as new primary events | Can contribute additional information according to the model |
| Effect measure in TEMPO3:4 | Not the posted method for these endpoints | Hazard ratio |
| Interpretation | Relative timing of first events | Relative event hazard under the recurrent-event framework |
This distinction matters because an HR from a recurrent-event analysis should not automatically be interpreted as if it came from an ordinary first-event Kaplan-Meier analysis.
13. Confidence Intervals and Statistical Precision
The posted analyses illustrate how confidence-interval width varies substantially across endpoints.
The visual above is only a conceptual display of the posted point estimates, not a forest plot and not a graphical representation of their confidence intervals. The numerical confidence intervals should be read from the results table because the three measures are on different statistical scales.
| Endpoint | Point estimate | 95% CI | What the interval indicates |
|---|---|---|---|
| Total kidney volume | −2.708 | −3.269 to −2.147 | The estimated mean difference remains below zero across the interval. |
| Clinical progression events | HR 0.865 | 0.775–0.965 | The interval remains below 1. |
| Renal function slope | 0.977 | 0.597–1.357 | The interval remains above zero. |
| Mean arterial blood pressure slope | −0.246 | −1.059–0.566 | The interval crosses zero. |
| Renal pain | −0.08 | −0.20–0.03 | The interval crosses zero. |
| Hypertensive events | HR 0.996 | 0.805–1.233 | The interval crosses 1. |
| Antihypertensive therapy outcome | RR 1.100 | 0.602–2.017 | The interval crosses 1 and is relatively wide. |
14. P-values: What They Do and Do Not Tell You
The seven posted analyses use two-sided P-values to test their respective null hypotheses. Those P-values range from <0.0001 to 0.9704.
P-value < 0.0001
The primary kidney-volume and renal-function analyses both report P < 0.0001. This provides strong statistical evidence against their respective null hypotheses, but the P-value itself is not a measure of effect magnitude.
P-value 0.9704
The hypertensive-event analysis reports P = 0.9704. The point estimate and confidence interval should still be examined because the P-value alone does not describe the range of plausible effects.
A P-value is conditional on the statistical model, the endpoint definition, the analysis population, and the hypothesis being tested. It should therefore be read together with the effect estimate and confidence interval.
15. Safety Results
The registry reports serious adverse events by treatment arm. The affected and at-risk counts are shown below.
| Safety measure | Tolvaptan | Placebo |
|---|---|---|
| Serious adverse events | 177 / 961 | 95 / 483 |
The registry reports the affected and at-risk counts rather than a formal comparative safety analysis in the posted statistical-analyses data summarized here. Safety should therefore be interpreted descriptively from these counts rather than by assigning an unreported hypothesis test or effect estimate.
16. Statistical Interpretation of the Primary Endpoint
The primary mean difference of −2.708 is a treatment-group difference on the scale of percentage change per year in total kidney volume. Its direction indicates a lower estimated annual percentage change in the tolvaptan group relative to placebo.
The estimate does not mean that every participant had a 2.708% difference in kidney-volume change, nor does it mean that kidney volume fell by 2.708%. It is a population-level between-group estimate from the specified longitudinal model.
The 95% CI from −3.269 to −2.147 indicates the statistical precision of the estimated treatment difference. The interval is entirely below zero, so the direction of the estimated difference is consistent throughout the reported interval.
The P-value of <0.0001 addresses the null hypothesis of no treatment difference. It does not quantify how important the treatment difference is and should not be substituted for the effect estimate or confidence interval.
The analysis is based on repeated observations and a linear mixed model. Interpretation therefore depends on the model's specification for time, treatment, and random effects, as well as the available observations contributing to the analysis.
17. Statistical Interpretation of the Secondary Endpoints
| Endpoint | Key interpretation |
|---|---|
| Clinical progression events | HR 0.865 corresponds to an estimated 13.5% lower hazard under the reported recurrent-event comparison, with a 95% CI of 0.775–0.965. |
| Renal function slope | A slope difference of 0.977 indicates a between-group difference in the modeled annual change, with a 95% CI of 0.597–1.357. |
| Mean arterial blood pressure slope | The placebo-minus-tolvaptan estimate was −0.246, with a 95% CI of −1.059–0.566. |
| Renal pain AUC | The adjusted mean difference was −0.08, with a 95% CI of −0.20–0.03. |
| Hypertensive events | HR 0.996 indicates an estimated recurrent-event hazard very close to the placebo reference, with a 95% CI of 0.805–1.233. |
| Antihypertensive therapy outcome | RR 1.100 indicates a relative risk 1.100 times that of placebo under the reported analysis, with a wide 95% CI of 0.602–2.017. |
The endpoints are not interchangeable. A mean difference, slope difference, hazard ratio, and risk ratio describe different quantities. Comparing their numerical values directly would not be statistically meaningful.
18. What the Registry Does and Does Not Report
The ClinicalTrials.gov record reports the trial design, seven posted outcome analyses, effect estimates, confidence intervals, P-values, analysis populations, and selected method descriptions. It does not provide, in the information summarized here, a complete baseline-characteristics table, Kaplan-Meier survival estimates, subgroup forest plots, or a detailed multiplicity framework.
That distinction is important because statistical interpretation should follow the quantities actually defined by the endpoint and analysis. For example, the clinical progression endpoint has a recurrent-event hazard ratio, whereas the primary kidney-volume endpoint has a longitudinal mean difference.
19. Limitations
- Endpoint-specific populations: the posted analyses apply different follow-up and availability criteria. Some secondary endpoints are restricted to non-hypertensive participants or participants taking antihypertensive medication at baseline.
- Available-data requirement: the registry states that only participants with available data were included in the posted analyses. The information provided does not specify a single imputation strategy across all endpoints.
- Mixed-model boundary estimate: the primary analysis notes that the random-effect intercept variance was zero, resulting in a non-positive, definite variance-covariance matrix for the random effects.
- Recurrent-event interpretation: hazard ratios for recurrent events should not be interpreted as ordinary first-event hazard ratios without considering the recurrent-event framework.
- Different effect measures: mean differences, slope differences, hazard ratios, and risk ratios operate on different scales and should not be compared numerically as if they represented the same estimand.
- Safety comparison: the posted safety information summarized here consists of affected/at-risk serious-adverse-event counts by arm rather than a formal comparative safety analysis.
- Registry detail: the ClinicalTrials.gov record provides statistical-analysis summaries but not every component of a full statistical analysis plan.
20. Why This Trial Matters Statistically
TEMPO3:4 is a useful statistical teaching case because the registry contains several distinct analytical structures within the same randomized comparison. The primary endpoint is longitudinal and uses a mixed-effects model; clinical progression and hypertensive events are recurrent time-to-event outcomes; renal function and blood pressure are evaluated through longitudinal slopes; renal pain uses covariate-adjusted ANCOVA; and the sustained antihypertensive-therapy outcome uses a Cochran-Mantel-Haenszel test.
| Concept | How it appears in TEMPO3:4 |
|---|---|
| Randomization | Randomized, parallel-group phase 3 comparison of tolvaptan and placebo. |
| Blinding | Quadruple masking. |
| Intention-to-treat analysis | Identified as the analysis population for the posted endpoint analyses, with endpoint-specific availability criteria. |
| Mixed-effects model | Used for the primary kidney-volume endpoint and renal-function and blood-pressure slope analyses. |
| Longitudinal analysis | Repeated measurements over time are incorporated into mixed-model analyses. |
| Recurrent-event analysis | Used for ADPKD clinical progression and hypertensive events per 100 follow-up years. |
| Hazard ratio | Used for the two recurrent-event analyses. |
| ANCOVA | Used for the renal-pain area-under-the-curve endpoint with baseline renal pain as a covariate. |
| Cochran-Mantel-Haenszel test | Used for the sustained decrease in blood pressure leading to reduced antihypertensive therapy. |
| Risk ratio | Reported for the categorical antihypertensive-therapy outcome. |
| Confidence intervals | Posted as two-sided 95% intervals for all seven statistical analyses. |
| Superiority testing | All seven posted statistical analyses are identified with a superiority hypothesis type. |
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Statistical Calculators
23. Sources
- ClinicalTrials.gov: TEMPO3:4, NCT00428948.
- PubMed: PMID 39356039.
- PubMed: PMID 38857379.
- PubMed: PMID 38414126.
- PubMed: PMID 37986188.
- PubMed: PMID 37349694.
Continue through the Clinical Biostats knowledge graph
Connect the endpoints and methods used in TEMPO3:4 to deeper statistical tutorials and analysis tools.
24. Record Summary
TEMPO3:4 provides a broad example of clinical-trial statistical methodology within a single randomized phase 3 study. The primary analysis used a linear mixed model for percentage change per year in total kidney volume, while secondary analyses applied recurrent-event methods, longitudinal slope models, ANCOVA, and the Cochran-Mantel-Haenszel test. The reported effects include mean differences, slope differences, hazard ratios, and a risk ratio, each requiring its own interpretation.
The central statistical result was a mean difference of −2.708 in percentage change per year in total kidney volume, with a two-sided 95% confidence interval of −3.269 to −2.147 and P < 0.0001. The secondary analyses provide additional perspectives on clinical progression events, renal function, blood pressure, renal pain, hypertensive events, and sustained reduction in antihypertensive therapy.