A Study to Test the Effect of Empagliflozin in Patients Who Are in Hospital for Acute Heart Failure: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Test the Effect of Empagliflozin in Patients Who Are in Hospital for Acute Heart Failure (NCT04157751), a randomized phase 3 study. The primary outcome is Percentage of Pairwise Comparisons With Wins of Clinical Benefit, a Composite of Death, Number of Heart Failure Events (HFEs), Time to the First HFE and ≥5-point Difference in CfB in KCCQ-TSS After 90 Days of Treatment.

2. Clinical Question

The clinical question is broader than whether a single event occurred. The composite seeks to order outcomes across several dimensions, giving priority to death and heart-failure events before considering symptoms measured by the Kansas City Cardiomyopathy Questionnaire-Total Symptom Score (KCCQ-TSS). The registry says comparisons were made between patients in the two randomized groups within strata. The score ranges from 0 to 100, with higher scores reflecting a better outcome.

As a general statistical point, a ranked composite can incorporate outcomes that differ in type and importance. Its interpretation depends on the sequence of criteria and the rules for ties. It does not, by itself, show how much each component contributed to an overall comparison. The registry record does not report component-specific primary-outcome results or a numerical estimate for the overall win-based endpoint.

3. Trial Design

EMPULSE is a completed, phase 3, randomized, double-blind, parallel-group trial sponsored by Boehringer Ingelheim. Its official title describes it as multicentre. The registry lists enrollment as 530 and records the study dates as 2020-05-18 to 2021-05-28 for study start and primary completion, respectively. The trial compared once-daily empagliflozin 10 mg with placebo in patients hospitalized for acute heart failure.

The registry description says participants were still in hospital at the beginning of the study and were stabilized before treatment initiation. The primary endpoint covers up to 90 days. The description characterizes participation as about 3 months, with later hospital visits and a phone call, but the registry record does not report a detailed visit schedule in the registry record for the outcomes. It also does not report treatment adherence or the number of participants who completed follow-up.

Random allocation and double masking are design features intended to reduce systematic differences between treatment groups and bias in outcome assessment. In general, randomization supports a fair comparison by balancing measured and unmeasured characteristics on average; it does not guarantee identical groups. Double masking can limit the influence of treatment knowledge on care or measurement, though the registry record does not report the specific masking procedures or whether masking was assessed.

The registry lists 11 outcome measures and 5 analyses. The listed secondary measures include KCCQ-TSS changes, NTproBNP, days alive and out of hospital, heart-failure events, hospitalization for heart failure, and a renal composite. These outcome listings describe what was assessed, not the results. Numerical estimates for these outcomes are not reported in the registry record. The detailed statistical analysis plan, including model specifications and handling of missing observations, is also not reported in the registry record.

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: All subjects were screened for eligibility prior to participation in the trial. Subjects attended a specialist site which ensured that they (the subjects) strictly met all inclusion and none of the exclusion criteria. Subjects were not to be allocated to a treatment group if any of the entry criteria were violated.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Randomised Set (RS), including all randomised patients.

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: Randomised Set (RS), including all randomised patients.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Pairwise Comparisons With Wins of Clinical Benefit, a Composite of Death, Number of Heart Failure Events (HFEs), Time to the First HFE and ≥5-point Difference in CfB in KCCQ-TSS After 90 Days of Treatment, measured over Up to 90 days. For KCCQ-TSS: at baseline and at day 90..

The primary endpoint is the percentage of pairwise comparisons classified as wins for clinical benefit. The registry describes comparing randomized patients in the empagliflozin group with randomized patients in the placebo group within strata. For each pair, the comparison proceeds sequentially: death, number of heart-failure events, time to the first heart-failure event, and then change from baseline in KCCQ-TSS at day 90. The sequence stops when one patient has an advantage. If no earlier criterion distinguishes the pair, the symptom-score criterion can determine a win when the difference reaches the stated threshold of 5 points; a difference below 5 is treated as tied.

The endpoint combines outcomes that represent survival, clinical events, timing, and reported symptoms. As a general statistical explanation, this type of ordered composite can use information from several domains while expressing priority through the sequence of comparisons. It may be useful when a single measure would not capture the range of clinical outcomes considered relevant. However, a single overall win percentage does not automatically explain the contribution of each component. Interpretation requires the comparison result and, ideally, component-level context.

The registry record gives the endpoint definition but does not report a numerical win percentage, a between-group contrast, or an uncertainty interval. Therefore, the magnitude and precision of the primary comparison cannot be assessed from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes a randomized, double-blind, parallel-group design and a primary endpoint based on pairwise wins. The analysis details behind the endpoint are not reported in the registry record. The following explanations are general statistical context and should not be read as a claim that a particular unreported method was used.

What does a pairwise win measure?

In general, a pairwise win approach compares an outcome for a participant in one group with an outcome for a participant in the other group. The comparison rules establish which outcome is preferable and how to handle ties. Here, the registry’s stated hierarchy begins with death, then considers the number of heart-failure events, then time to the first event, and finally KCCQ-TSS change from baseline at day 90. The endpoint is the percentage of pairwise comparisons classified as wins for clinical benefit.

This structure differs from comparing only the average of a continuous measure or the proportion experiencing a single event. The hierarchy puts more emphasis on earlier criteria: later criteria matter only when earlier criteria do not distinguish the pair. Consequently, the overall result would reflect both outcomes and the ordering rule. The registry record does not report how many comparisons were wins for either group, how many were ties, or the numerical result after the comparisons were combined.

Why does the hierarchy matter?

As a general explanation, a hierarchy makes explicit that some outcomes are treated as more important than others for the comparison. It can prevent a favorable result on a less serious measure from overriding an unfavorable result on a higher-priority measure within a pair. But an overall endpoint still compresses multiple domains into one summary. Component-level findings are useful for understanding what drives that summary; they are not reported for the primary endpoint in the registry record.

The endpoint uses KCCQ-TSS, which the registry says ranges from 0 to 100, with higher scores reflecting a better outcome. Its stated rule treats a difference of at least 5 points as a win at that stage and a difference below 5 as a tie. In general, a threshold can make a comparison rule more interpretable by defining what difference counts for the endpoint. It does not reveal the distribution of observed scores or the number of pairs decided at that stage.

What does randomization contribute?

In general, random assignment supports comparison by reducing systematic allocation differences in expectation. It does not ensure that the groups are identical in a particular trial, and it does not remove uncertainty from the results. The registry says patients were compared within strata, but the specific strata and the procedures used to account for them are not reported in the registry record.

Double masking is also part of the recorded design. In general, masking can reduce the possibility that knowledge of assigned treatment influences care, reporting, or assessment. The registry record does not report how masking was maintained, whether any unmasking occurred, or whether outcome assessment was independently reviewed. Those details should not be inferred from the design label alone.

How should estimates and uncertainty be read?

In general, an effect estimate describes the observed contrast under a specified analysis, and a confidence interval communicates the range of values reasonably compatible with the data and model assumptions. A confidence interval’s width is informative about precision; it is not a guarantee that the true value lies within a particular realized interval. A p-value, if reported, describes how compatible the data are with a null hypothesis under the statistical model. It does not state the probability that a treatment is effective, nor does it measure the size or clinical importance of an effect.

No primary-outcome estimate, confidence interval, or p-value is reported in the registry record. Thus, these concepts explain what would be needed for an assessment, not what EMPULSE found. The same limitation applies to secondary outcomes: the registry record lists them but does not give numerical results in the registry record.

What can sample-size planning illustrate?

In general, sample-size calculations are planning exercises based on assumptions about the effect to detect, variability or event information, allocation, and statistical error rates. The planning description associated with this trial refers to a two-sided significance level of 0.05, equal allocation, and a normal approximation, with an effect defined as a stated fraction of the observed effect. For time-to-event planning, it describes total events; for other outcomes, it describes participants per arm.

Such planning figures are not observed power and do not establish the probability that the trial detected an effect after the fact. The numerical planning illustrations and the observed effect used for them are not reported in the registry record. No trial-specific sample-size justification or design assumptions beyond the planning description are reported in the registry record.

What remains unknown about the analysis?

The registry record does not report the statistical model, variance estimation, confidence-interval method, missing-data procedure, sensitivity analyses, or interim-analysis plan. It also does not report whether analyses were adjusted for baseline characteristics or how multiplicity across outcomes was handled. These are important parts of statistical interpretation, particularly for a composite endpoint and a list of secondary measures. Their absence should be treated as unavailable information, not as evidence that a particular method was or was not used.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 11 outcome measures and 5 analyses. It identifies a primary endpoint and also names secondary measures covering symptoms, a biomarker, days alive and out of hospital, heart-failure events, hospitalization, and a renal composite. The registry record does not report how many statistical tests were planned across these outcomes, how they were ordered for inference, or whether any adjustment for multiple testing was used.

As a general statistical point, assessing multiple outcomes can increase the chance of at least one apparently unusual result when all are examined without accounting for multiplicity. Prespecified hierarchies, adjusted testing procedures, or a clear distinction between confirmatory and exploratory analyses can help readers understand the strength of evidence. The presence of several listed outcomes alone does not show which approach was used. The registry record does not report the multiplicity strategy.

Missing-data handling is also not reported in the registry record. The primary endpoint includes KCCQ-TSS at day 90, but the record does not describe how unavailable scores, incomplete follow-up, or other missing observations were handled. In general, missingness can affect estimates when participants with missing measurements differ from those with observed measurements. Statistical approaches make assumptions about those differences, and sensitivity analyses can examine how conclusions change under alternative assumptions. No such procedures are reported.

The primary population is the Randomised Set, defined as including all randomized patients. This identifies the population named for the primary endpoint, but it does not give the analysis details needed to determine how every randomized participant contributed to each component or how unavailable component data were treated. The registry record also does not report a per-protocol population or other analysis-population definitions.

The primary outcome was not reduced to a single comparison in the registry data available for this analysis because it has several categories or time points. Accordingly, there is no single primary-outcome estimate to interpret. The record does not report an analysis plan that would clarify the intended handling of the hierarchical endpoint, missing data, or multiplicity.

11. 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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: cardiac failure (14.0%), acute kidney injury (7.2%), cardiac failure congestive (4.2%), ventricular tachycardia (3.1%), cardiac arrest (1.9%).

The registry reports serious adverse event counts for placebo and 10 mg empagliflozin, along with counts for other adverse events and deaths. For placebo, the listed serious adverse event figures are 115 participants affected among 264 at risk; the other adverse event figures are 19 among 264 at risk, and deaths are listed as 22. For 10 mg empagliflozin, the corresponding figures are 84 among 260 at risk for serious adverse events, 20 among 260 at risk for other adverse events, and 11 deaths.

These are descriptive group entries. They should not be converted into a comparative claim without the relevant analysis and uncertainty. In general, adverse-event counts depend on how events were defined, collected, and followed, as well as on the population and time period represented. The registry record does not report all those details in the registry record, nor does it provide a comparative estimate or confidence interval for these figures.

The registry lists serious adverse event terms including cardiac failure at 14.0%, acute kidney injury at 7.2%, cardiac failure congestive at 4.2%, ventricular tachycardia at 3.1%, cardiac arrest at 1.9%, and hypotension at 1.9%. These are listed term-specific percentages; the registry record does not report their denominator or derivation in the registry record. A term listing does not establish that an event was caused by treatment. The death counts are reported separately and should not be added to other adverse-event counts.

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

The registry record describes a randomized comparison of empagliflozin and placebo in patients hospitalized for acute heart failure who had been stabilized. It specifies a hierarchical primary endpoint intended to compare clinical benefit through 90 days. This information explains the question the trial was designed to address, but it does not reveal the numerical answer.

The primary endpoint’s design combines death, heart-failure events, their timing, and KCCQ-TSS change. In general, a composite can capture more than one aspect of outcome, and a hierarchy can express priorities among them. At the same time, the combined result can be difficult to interpret without its component results. The registry record does not report a primary win percentage, a between-group estimate, a confidence interval, or a significance test. It also does not report component-level numerical primary results.

The available safety entries describe counts of participants with serious or other adverse events and deaths in each group. They are not, by themselves, an adjusted comparison and do not establish causation. The listed serious-event terms and percentages likewise should not be interpreted as proof that a treatment caused or prevented a particular event. The registry record does not report a comparative safety estimate or its uncertainty in the registry record.

As a general statistical principle, evidence about an effect depends on both the estimated contrast and its precision, along with the analysis assumptions. A non-significant result, if reported, would not alone establish equivalence or absence of an effect. Conversely, a statistically detectable difference would not alone establish that the difference is important to patients. No such primary-outcome test result is reported.

The trial’s completed status and its enrollment listing of 530 are design and the registry entries, not outcome findings. Neither establishes whether the groups differed on the primary endpoint. The statistically defensible interpretation is limited to the endpoint definition, trial design, outcome listings, and descriptive safety counts. Claims about benefit, lack of benefit, or comparative risk cannot be drawn from the primary-outcome information reported in the registry record.

13. Why This Trial Matters Statistically

EMPULSE is statistically notable for defining clinical benefit as an ordered comparison across survival, heart-failure events, event timing, and symptoms rather than as a single event or score. The hierarchy makes the endpoint’s priorities explicit: death is considered before event counts, event counts before timing, and symptom-score change after those criteria. That design raises important interpretive questions about how often each stage determines a comparison and how the overall result relates to individual outcomes.

The trial also illustrates why endpoint definitions and endpoint results must be kept separate. A detailed definition can show how outcomes were intended to be judged, but it cannot substitute for an estimate of the observed treatment contrast. The registry record describes the primary endpoint but does not report its numerical result or uncertainty. Without those quantities, readers cannot assess the direction, magnitude, or precision of the primary comparison.

Randomization and double masking are important design features for reducing bias in a treatment comparison. Their presence does not eliminate uncertainty or answer questions about missing data, analysis choices, or multiplicity. The registry lists multiple secondary outcomes, making clear that the trial examined several domains, but the registry record does not report a numerical result for those outcomes in the registry record.

For biostatisticians and clinical researchers, the record is a reminder that composite endpoints require transparent reporting of both the overall summary and its components. An overall win-based measure may be informative, but its interpretation depends on the hierarchy, tie rules, analysis population, and uncertainty assessment. Here, the endpoint rules are described, while many analysis details and the primary numerical result are not reported in the registry record.

14. 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 main limitation for statistical interpretation is that the registry record describes the primary endpoint without reporting its numerical result. There is no estimate, confidence interval, or significance test in the registry record. As a result, the treatment contrast’s direction, magnitude, and precision cannot be assessed from this record.

The primary measure is a hierarchical composite. Although the sequence and tie rule are described, the registry record does not report how many pairwise comparisons were decided at each stage, the number of wins or ties, or component-level results. Without those details, readers cannot determine which outcome domains contributed most to an overall comparison or assess how the composite relates to its individual components.

The record lists secondary outcomes but does not report their numerical results in the registry record. It also does not report the statistical analysis plan, the method used to estimate uncertainty, missing-data procedures, sensitivity analyses, or multiplicity handling. These gaps limit assessment of the robustness and confirmatory status of any outcome findings. They should be treated as information not reported in the registry record, rather than filled with assumptions.

The safety entries are descriptive counts and term-specific percentages. The registry entries available do not include comparative safety estimates or uncertainty, and the denominator or derivation of the listed term-specific percentages is not reported in the registry record. These entries therefore cannot establish a treatment-related difference.

The enrollment listing and completed status provide context but do not establish the number analyzed for each outcome or the completeness of follow-up. The registry record does not report those details in the registry record. These limitations concern what can be concluded from the registry record; they do not establish that additional information is absent from every other source.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: A Study to Test the Effect of Empagliflozin in Patients Who Are in Hospital for Acute Heart Failure — NCT04157751.
• Registry record last updated 2022-07-19; status last verified 2022-06.
• Results first posted 2022-06-06.
• 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.

18. Record Summary

EMPULSE is a completed, phase 3, randomized, double-blind, parallel-group trial sponsored by Boehringer Ingelheim. It compared once-daily oral empagliflozin 10 mg with placebo in stabilized patients hospitalized for acute heart failure. The registry lists enrollment as 530 and a study period from 2020-05-18 to 2021-05-28.

The primary endpoint is the percentage of pairwise comparisons favoring clinical benefit through 90 days, using a hierarchy of death, heart-failure events, time to first event, and KCCQ-TSS change. The primary population is the Randomised Set, defined as all randomized patients. Secondary measures include symptoms, NTproBNP, days alive and out of hospital, heart-failure outcomes, and a renal composite.

The registry record does not report a numerical primary-outcome estimate or its uncertainty. It lists serious and other adverse-event counts, deaths, and several serious-event terms, but these entries do not provide a comparative treatment effect. Analysis details, missing-data procedures, and multiplicity handling are not reported in the registry record.