Dapagliflozin Evaluation to Improve the LIVEs of Patients With PReserved Ejection Fraction Heart Failure.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Dapagliflozin Evaluation to Improve the LIVEs of Patients With PReserved Ejection Fraction Heart Failure. (NCT03619213), a randomized phase 3 study. The primary outcome is Subjects Included in the Composite Endpoint of CV Death, Hospitalization Due to Heart Failure or Urgent Visit Due to Heart Failure..

2. Clinical Question

The trial’s primary question concerns a composite clinical outcome rather than a single event type. It combines cardiovascular death with worsening heart failure events that result in hospitalization or an urgent visit. The registry record also identifies the study as event-driven and describes a broader composite involving cardiovascular death or heart failure events. The primary outcome entry specifically names the composite of cardiovascular death, hospitalization due to heart failure, or urgent visit due to heart failure.

This distinction matters when interpreting the result: the primary comparison addresses the combined endpoint, not the separate effect on each component. Separate cardiovascular-death and all-cause-mortality outcomes are also posted, but they are secondary outcomes and should not be treated as interchangeable with the primary composite. The trial therefore asks whether the groups differed on the specified combined endpoint over the reported follow-up, not whether every individual component changed in the same way.

3. Trial Design

This was an international, multicentre, randomized, double-blind, placebo-controlled phase III study with a parallel-group design. The registry lists quadruple masking. Participants were assigned in a 1:1 ratio to dapagliflozin or matching placebo. Dapagliflozin was given at 10 mg once daily in addition to background regional standard-of-care therapy, including treatment to control co-morbidities. The condition listed is heart failure with preserved ejection fraction.

The registry reports an enrollment of 6263 participants. The posted primary outcome counts are 3131 participants in the dapagliflozin group and 3132 in the placebo group. The trial status is completed. Its recorded start date is 2018-08-27, and its primary completion date is 2022-03-27. The primary endpoint timeframe is up to 42.1 months; this is the stated outcome timeframe and should not be interpreted as a uniform follow-up duration for every participant.

The primary analysis is described as covering all randomized participants in the full analysis set, including events occurring on or before the primary analysis censoring date. The censoring date itself is not reported in the registry record. The sponsor’s posted primary analysis used regression and Cox methods and reported a hazard ratio. Further detail on the model specification, covariate adjustment, handling of censoring, or assumptions is not reported in the registry record.

The registry identifies five secondary outcomes, including heart-failure events within the composite, a subgroup defined by left ventricular ejection fraction, KCCQ total symptom score, cardiovascular death, and all-cause mortality. Results are posted for three secondary outcomes in the record. The methods used for each secondary analysis differ, so their estimates should be interpreted according to the outcome and analysis reported rather than as a single uniform set of measures.

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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set

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 Subjects Included in the Composite Endpoint of CV Death, Hospitalization Due to Heart Failure or Urgent Visit Due to Heart Failure., measured over Up to 42.1 months.

The primary endpoint counts participants included in a composite of cardiovascular death, hospitalization due to heart failure, or an urgent visit due to heart failure. A composite endpoint brings several clinically relevant events into one outcome. In general, this can capture more than one way a condition may worsen and can provide a broader comparison than relying on a single component alone. It also means that the result belongs to the composite as defined; it does not establish that every component contributed equally or showed the same pattern.

The registry describes the primary endpoint as a dual primary efficacy endpoint and states that it was analyzed in the full analysis set. Events occurring on or before the primary analysis censoring date were included. The endpoint timeframe is up to 42.1 months. The posted participant counts are 512 of 3131 in the dapagliflozin arm and 610 of 3132 in the placebo arm. The sponsor’s reported time-to-event analysis expresses the comparison as a hazard ratio, while the unadjusted recalculation expresses it as a risk difference. These are different statistical summaries, and each answers the comparison using its own scale.

Composite outcomes are useful when multiple event types matter to the trial question, but interpretation requires attention to the exact component definitions. The registry does not report the separate primary-composite contribution of each component in the outcome entry. Separate secondary cardiovascular-death results are posted, but they do not replace a component-by-component account of the primary endpoint.

7. Results Posted in the Registry

• Dapa 10 mg: 16.4% (512/3131) versus 19.5% (610/3132) on control; risk difference -3.1 percentage points (95% CI -5.0 to -1.2), risk ratio 0.84 (0.75–0.93); unadjusted two-proportion z-test p = 0.001.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Change From Baseline in the KCCQ Total Symptom Score at 8 Months

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Dapa 10 mg: mean 8.30 versus 5.20 on control; difference 3.10 (95% CI 1.52 to 4.68), standardized effect d = 0.15; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Dapa 10 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Subjects Included in the Endpoint of Cardiovascular Death

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Dapa 10 mg: 7.4% (231/3131) versus 8.3% (261/3132) on control; risk difference -1.0 percentage points (95% CI -2.3 to 0.4), risk ratio 0.89 (0.75–1.05); unadjusted two-proportion z-test p = 0.160.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Dapa 10 mg and the control. The interval gives the range of effects that remain compatible with the data.

Subjects Included in the Endpoint of All-cause Mortality

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Dapa 10 mg: 15.9% (497/3131) versus 16.8% (526/3132) on control; risk difference -0.9 percentage points (95% CI -2.8 to 0.9), risk ratio 0.95 (0.84–1.06); unadjusted two-proportion z-test p = 0.324.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Dapa 10 mg and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes broaden the questions beyond the primary composite, covering symptom scores, cardiovascular death, all-cause mortality, and heart-failure event definitions, including a subgroup outcome. The posted results do not show one uniform pattern across these different outcome types. The KCCQ analysis uses a symptom-score measure, while the mortality outcomes use time-to-event methods. Their estimates and uncertainty therefore should not be collapsed into one overall secondary conclusion.

As a group, the secondary analyses provide context for the primary result, but they do not redefine it. The primary endpoint combines cardiovascular death and worsening heart failure events; separate secondary outcomes may help describe particular aspects of that clinical picture, yet they are not a substitute for a complete component breakdown of the primary endpoint. Results for all five listed secondary outcomes are not reported in the registry record, so the full secondary-outcome pattern cannot be assessed from the posted results.

Unadjusted secondary p-values carry less weight when many outcomes or analyses are considered, because the chance of at least one apparently notable finding can increase as the number of comparisons grows. The registry record does not report a multiplicity adjustment strategy for these secondary analyses. Consequently, individual secondary results are best treated as supportive or exploratory rather than as independent confirmatory findings. That qualification applies even when a reported result has a small p-value: the context includes the number of outcomes, analysis choices, and whether the outcome was designated for confirmatory inference.

There is no basis in the registry record to infer a hidden hierarchy among the secondary endpoints or to assume how their multiplicity was handled. The posted secondary results should be read separately, with their measures, methods, and confidence intervals kept in view, and with less inferential weight than the primary analysis.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Events Included in the Composite Endpoint of CV Death or Recurrent Heart Failure Event (Hospitalization Due to Heart Failure or Urgent Heart Failure Visit)
• Events Included in the Composite Endpoint of CV Death or Recurrent Heart Failure Event (Hospitalization Due to Heart Failure or Urgent Heart Failure Visit) for LVEF <60% Subpopulation
• Change From Baseline in the KCCQ Total Symptom Score at 8 Months
• Subjects Included in the Endpoint of Cardiovascular Death
• Subjects Included in the Endpoint of All-cause Mortality

9. Statistical Interpretation

10. Statistical Methods Explained

The central statistical distinction in this trial is between the sponsor’s reported time-to-event analysis and the simple unadjusted recalculation from the posted counts. Both compare randomized groups, but they summarize the data differently. The sponsor reports a hazard ratio from regression and Cox methods. The recalculation reports a risk difference based on the recorded participant counts. Neither measure should be substituted for the other.

What does a hazard ratio describe?

In general, a hazard ratio compares the rate at which events occur over follow-up between groups, using a time-to-event model. A value below 1 indicates a lower estimated event hazard in the intervention group relative to the comparator under the model’s coding. The sponsor reports 0.82 for the primary composite, with a 95% confidence interval from 0.73 to 0.92. A hazard ratio is not a ratio of the posted event counts and does not directly state the difference in the probability that an individual participant experiences an event. Interpretation depends on the model and its assumptions, including how time and censoring are handled.

The registry reports the analysis method as regression and Cox, but detailed model specifications are not reported in the registry record. This includes whether and how covariates were included, the exact handling of censoring, and the diagnostics or checks used for model assumptions. Those details matter for reproducing or fully evaluating a time-to-event analysis.

What does the risk difference describe?

In general, a risk difference is the difference between the proportions with an event in two groups over a specified observation period. The recalculation uses the posted counts and denominators and reports -0.031 for the primary outcome, with a 95% confidence interval from -0.050 to -0.012. The negative sign reflects fewer participants counted with the event in the dapagliflozin group in this subtraction direction. This is a simple unadjusted summary of the posted group counts. It does not account for event timing in the way a time-to-event model does.

The recalculated interval describes uncertainty around that unadjusted difference under its calculation. It should not be presented as a replacement for the sponsor’s model-based result. The registry indicates no discrepancy between the posted count-based values and the reported analysis in the sense monitored for this page, but the statistical estimands still differ.

How should a confidence interval be read?

In general, a confidence interval gives a range of values compatible with the observed data under the analysis and its assumptions. A narrower interval indicates greater precision on that scale; a wider interval indicates more uncertainty. For a hazard ratio, the reference value of no relative difference is 1. For a risk difference, it is 0. If an interval includes its reference value, the data do not rule out no difference at the corresponding confidence level using that analysis. If it does not include the reference value, the interval indicates evidence against that value under the analysis, while still leaving uncertainty about the size of the effect.

A p-value gives a measure of how unusual data at least as extreme as those observed would be under 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 the result. The sponsor’s primary p-value is 0.001; the unadjusted recalculation reports p = 0.001. Their closeness should not obscure that they arise from different approaches.

Why does the outcome definition matter?

The primary outcome is a composite of cardiovascular death and worsening heart failure events. Composite outcomes can increase the range of events represented in a trial’s main question, but a combined estimate does not show the separate effect on each component. Components may differ in frequency, severity, or how they are ascertained. The registry’s primary outcome entry does not provide a breakdown of the primary composite by component. The separately posted cardiovascular-death outcome is informative about that endpoint but cannot be used to infer the unreported component breakdown of the primary analysis.

For the KCCQ total symptom score, the posted outcome is continuous and is reported as a mean and standard deviation in each group. In general, a mean difference compares group averages, while the standard deviation describes variation among individual scores. The sponsor also reports a Win Ratio analysis for this secondary outcome. A Win Ratio is a distinct comparative summary, not another expression of the mean difference; its interpretation depends on the analysis hierarchy and rules, which are not reported in the registry record.

What do the planning illustrations mean?

In general, sample-size calculations estimate how many participants or events a future study might need under stated assumptions to have a chosen probability of detecting a specified effect at a selected significance level. The planning illustrations use the observed effect or a stated fraction of it, a two-sided significance level of 0.05, equal allocation, and a normal approximation. They are not the trial’s original design calculation. The original design assumptions and any protocol-specific sample-size justification are not reported in the registry record.

These planning calculations should be read as conditional illustrations, not as confirmation that a design would achieve the stated result. A smaller target effect generally requires more information, often through more participants or events, to distinguish it from no difference. The illustration does not establish whether those assumptions fit a future population or analysis.

11. Confidence Intervals and Precision

Precision is reflected in the range of values around an estimate, not just in the estimate itself or its p-value. For the sponsor’s primary hazard ratio of 0.82, the 95% confidence interval is 0.73 to 0.92. This interval lies below the no-relative-difference value of 1, while still allowing a range of possible effect sizes. The unadjusted risk-difference estimate is -0.031, and its 95% confidence interval runs from -0.050 to -0.012. That interval lies below the no-difference value of 0. These intervals are on different scales and arise from different analyses.

The secondary cardiovascular-death estimate is a hazard ratio of 0.88, with a 95% confidence interval from 0.74 to 1.05. The all-cause-mortality hazard ratio is 0.94, with a 95% confidence interval from 0.83 to 1.07. Each interval includes 1, so the posted analyses leave uncertainty that includes no relative difference for those outcomes. This does not establish that the groups are equivalent; it means the estimates are not precise enough to rule out the reference value using those analyses.

For the KCCQ mean difference, the estimate is 3.1, with a 95% confidence interval from 1.517 to 4.683. The interval describes uncertainty in the group mean contrast under the recalculation. The sponsor’s Win Ratio result is a separate analysis with its own reported interval. Confidence intervals should be interpreted alongside the outcome definition, model, and analysis population; they do not correct for unreported design details or for multiple testing.

In general, greater precision is associated with a narrower interval, but a confidence interval does not describe the range of individual participant responses. The registry record does not provide a single precision summary that can resolve all uncertainty about endpoint components, subgroup findings, or the analysis assumptions.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size illustrations translate an observed estimate, or a stated fraction of that estimate, into a hypothetical planning requirement. For binary outcomes, the entries are participants per arm. For time-to-event planning tables, entries are the total number of events. The assumptions listed for these illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the effect to detect is the stated fraction of the observed effect. The original trial’s design assumptions are not reported in the registry record.

The 80% power column describes a planning scenario in which the assumed study would have an 80% probability of detecting the target effect at the stated significance level, if the assumptions hold. The 90% power column describes the corresponding scenario with a 90% probability. These are design probabilities under a hypothetical calculation, not guarantees about a future study’s result. The calculations use the observed effect as an input, or a fraction of it, rather than a separately established clinically important difference.

The rows for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect show how the planning requirement changes as the target effect is reduced. Smaller effects need more participants or events to be detected with the same planned power under the stated assumptions. For example, the primary endpoint illustration gives 2,365 participants per arm for 80% power at 100% of the observed effect and 9,780 participants per arm for 80% power at 50% of the observed effect. These figures are planning illustrations, not observed power.

The illustrations do not reproduce the trial’s own design rationale, and they should not be read as a recommendation for a future trial without considering its population, endpoint, follow-up, analysis, and design assumptions. The trial’s sample-size justification is not reported in the registry record.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.195; treatment proportion p₁ = 0.164; average p̄ = 0.179.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5423 = 1.0629.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5419 = 0.4560.
4. n = (1.0629 + 0.4560)² / (-0.0312)² = 2,364.5, rounded up to 2,365 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.194763729246488                          # control proportion
p1 <- 0.163526030022357                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(3619213)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 2,365 (80% power), 3,165 (90% power)
 75% of observed effect: n per arm = 4,276 (80% power), 5,724 (90% power)
 50% of observed effect: n per arm = 9,780 (80% power), 13,093 (90% power)
power.prop.test: 2364.5 
Simulated power at n = 2,365 per arm: 0.808 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The sponsor’s primary analysis is reported as regression and Cox methods, with a hazard ratio of 0.82, a 95% confidence interval from 0.73 to 0.92, and p < 0.001. This is the trial’s posted model-based analysis of the primary time-to-event outcome. It incorporates the timing of events and censoring according to the sponsor’s analysis, although the detailed model specification is not reported in the registry record.

The unadjusted recalculation uses the arm-level participant counts and reports a risk difference of -0.031, with a 95% confidence interval from -0.050 to -0.012 and p = 0.001. It is a simple comparison of the posted counts and does not reproduce the sponsor’s time-to-event model. The two analyses point in the same direction, but their estimates have different meanings and scales. Their numerical values are therefore not expected to match.

The recalculation is useful as a transparent check on the posted summary counts, but it does not supersede the sponsor’s analysis for the trial’s stated time-to-event question. The registry marks no discrepancy between the posted primary results and the recalculation. The posted sponsor result remains the relevant reported analysis, while the recalculation offers an unadjusted view of the group counts. Neither result removes uncertainty about unreported model details or the individual components of the composite endpoint.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study evaluates more than one outcome, subgroup, time point, or analysis. In general, the more statistical comparisons made, the greater the opportunity for some results to appear notable by chance, even if no corresponding underlying differences exist. Researchers may address this through a prespecified testing hierarchy, adjustment of significance thresholds, or other strategies. The statistical plan for multiplicity, including whether such a strategy was used for the secondary outcomes, is not reported in the registry record.

The registry lists five secondary outcomes and reports analyses for three of them in the posted results. The primary endpoint is identified as a dual primary efficacy endpoint, but the registry entry does not explain any testing hierarchy or multiplicity procedure. Without those details, secondary p-values should not be interpreted as though each were a separate, fully protected confirmatory test. The status of each secondary analysis as confirmatory or exploratory is not reported in the registry record.

The primary analysis is described as using the full analysis set, including events on or before the primary analysis censoring date. This indicates the stated analysis population, but it does not explain every detail of the handling of missing outcome data, withdrawals, or censoring. Specific missing-data methods, sensitivity analyses, and the number of participants with incomplete data are not reported in the registry record. The censoring date is also not reported.

These omissions limit the ability to assess how robust the reported estimates are to alternative assumptions about missingness, censoring, and multiple comparisons. They do not negate the reported results, but they constrain how fully an independent reader can reconstruct the analysis or distinguish confirmatory from exploratory evidence. A registry description is not a substitute for a complete statistical analysis plan.

16. 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 2% 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 (12.1%), covid-19 (5.9%), pneumonia (3.6%), covid-19 pneumonia (2.8%), cardiac failure congestive (2.6%).

The registry reports serious adverse events affecting 1454 of 3126 participants in the dapagliflozin safety group and 1508 of 3127 participants in the placebo safety group. Other adverse events are reported for 464 of 3126 and 485 of 3127 participants, respectively. The listed deaths are 510 in the dapagliflozin group and 533 in the placebo group. These are posted counts with the stated safety denominators; the registry entry does not provide an adjusted comparison or establish that events were caused by assigned treatment.

Among the listed serious adverse-event terms, the record gives Cardiac failure at 12.1%, Covid-19 at 5.9%, Pneumonia at 3.6%, Covid-19 pneumonia at 2.8%, Cardiac failure congestive at 2.6%, and Ischaemic stroke at 2.2%. These are coded event terms and should not automatically be interpreted as mutually exclusive categories or as treatment-related events. The registry does not report the attribution criteria or a detailed event-level breakdown in this entry.

Safety counts are descriptive and depend on the registry’s definitions, ascertainment, and follow-up. Without further details, they do not by themselves establish comparative risk or causation. The record’s safety summaries should be considered alongside the trial’s methods and the context of its participants, rather than read as an individualized prediction.

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

The reported primary analysis indicates a difference between randomized groups on the composite of cardiovascular death, hospitalization due to heart failure, or urgent visit due to heart failure during the stated follow-up. The sponsor’s hazard ratio and the unadjusted risk-difference recalculation both favor the dapagliflozin group on their respective scales, and their reported confidence intervals exclude their corresponding no-difference values. This is a statement about the posted group-level statistical results for the defined primary outcome.

It does not mean that every participant in one group had a better outcome, nor does it predict what will happen to a particular person. It also does not establish that dapagliflozin changed each component of the composite in the same way. The separate component-specific information is incomplete in the registry record, and the primary result cannot be used to determine which component contributed most.

The secondary analyses are not uniformly conclusive, and they carry less inferential weight in the absence of reported multiplicity procedures. In particular, the posted mortality analyses have confidence intervals that include no relative difference. This does not prove there is no difference; it indicates that the analyses leave a range of possibilities that includes the reference value.

The trial’s results apply to the randomized population and analysis described in the registry. The entry does not report all details needed to assess subgroup consistency, missing-data assumptions, model specification, or the original sample-size rationale. Statistical evidence should therefore be kept distinct from claims about clinical importance, generalizability to other populations, or the balance of benefits and harms for an individual. The registry findings are a description of trial results, not medical advice.

18. Why This Trial Matters Statistically

This trial is statistically informative because it combines random assignment, a placebo comparator, masking, and a prespecified primary composite in a large phase III study. Randomization supports a structured comparison of the groups, while masking can reduce some sources of bias in treatment delivery and outcome assessment. The registry describes an international, multicentre study and identifies the full analysis set as the primary analysis population.

The record also illustrates why outcome scale and analysis method matter. The sponsor’s Cox analysis reports a hazard ratio, which uses event timing, while the count-based recalculation reports a risk difference. Both can help describe a randomized comparison, but they answer the question in different ways. Keeping the measures distinct helps prevent an unadjusted count comparison from being mistaken for the trial’s model-based time-to-event result.

The composite endpoint offers a broad measure of cardiovascular death and worsening heart failure events, but it also makes component-level interpretation important. A statistically supported composite does not reveal the contribution of each event type without the corresponding breakdown. The separate mortality results and symptom-score analysis add context but do not resolve that limitation.

Finally, the record highlights how confidence intervals, secondary-outcome multiplicity, and sample-size planning affect interpretation. The primary interval describes a range of plausible estimates under its analysis; secondary results require caution; and the planning figures are illustrative rather than evidence of observed power or of the original design rationale. Together, these considerations show why a result should be read as an estimate with assumptions and limits, not as a single definitive number.

19. 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 entry does not include the full statistical analysis plan, detailed Cox model specification, covariate set, diagnostics, or the precise censoring rules. It identifies the full analysis set and states that events up to the primary analysis censoring date were included, but does not give that date. These omissions limit independent reconstruction of the sponsor’s primary analysis.

The primary outcome is composite, while a component-level breakdown for the primary analysis is not reported in the registry record. This restricts interpretation of how cardiovascular death, hospitalization, and urgent visits contributed to the overall result. The posted secondary cardiovascular-death outcome provides a separate estimate but does not supply the missing breakdown of the primary composite.

The record lists five secondary outcomes, but results for all of them are not reported. It also does not describe a multiplicity procedure, testing hierarchy, or whether the secondary outcomes were intended as confirmatory. The interpretation of individual secondary findings is therefore limited. Details of missing-data handling, withdrawals, sensitivity analyses, and outcome ascertainment are also not reported in the registry record.

Safety results are counts and coded terms; the entry does not provide detailed attribution or an adjusted comparative safety analysis. The sample-size illustrations are based on observed effects and stated general assumptions, not the trial’s original design rationale. These limitations do not erase the reported estimates, but they narrow what can be concluded from the registry record alone.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Dapagliflozin Evaluation to Improve the LIVEs of Patients With PReserved Ejection Fraction Heart Failure. — NCT03619213.
• Registry record last updated 2023-07-11; status last verified 2023-07.
• Results first posted 2023-07-11.
• 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.

23. Record Summary

The registry record describes a completed, international, randomized, double-blind, placebo-controlled phase III trial of dapagliflozin 10 mg once daily in HFpEF, added to background regional standard-of-care therapy. Enrollment was 6263. The primary endpoint was the composite of cardiovascular death, hospitalization due to heart failure, or urgent visit due to heart failure, analyzed in the full analysis set.

The sponsor reported a Cox hazard ratio of 0.82, with a 95% confidence interval from 0.73 to 0.92 and p < 0.001. The unadjusted count-based recalculation points in the same direction but is a different measure. Secondary findings are mixed across outcome types, and the record does not report results for every listed secondary outcome or the multiplicity plan. Safety summaries are descriptive counts. Missing analysis details and the absence of a primary component breakdown limit interpretation.