Treatment of Diabetes in Patients With Systolic Heart Failure: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Treatment of Diabetes in Patients With Systolic Heart Failure (NCT02920918), a randomized phase 4 study. The primary outcome is Change From Baseline Aerobic Exercise Capacity at 12 Weeks.

2. Clinical Question

The registry describes the study as examining treatment of diabetes in people with systolic heart failure. Its brief description names peak oxygen consumption and the minute ventilation to carbon dioxide production slope as primary measures of aerobic exercise capacity and ventilatory efficiency. The posted primary outcome result is for change in peak oxygen consumption. A comparative result for the ventilatory slope is not reported in the registry record.

The clinical question is therefore specific: it concerns the measured change in exercise capacity over the stated treatment period, in the enrolled population and under an active comparison. It does not, by itself, answer broader questions about other outcomes, longer-term effects, or people outside the population enrolled in this trial.

3. Trial Design

This was a Phase 4, randomized, parallel-group, active-control study sponsored by Virginia Commonwealth University. The registry lists triple masking. The study was completed and enrolled 36 participants. Both groups received an oral pill daily for 12 weeks: canagliflozin 100 mg or sitagliptin 100 mg. Since both groups received an active treatment, the comparison estimates the difference between these treatment groups rather than a difference from no active treatment.

The primary outcome described in the posted results is change from baseline in aerobic exercise capacity at 12 weeks. Peak oxygen consumption was measured by maximal cardiopulmonary exercise test and expressed in mL/kg/min. The endpoint analysis used available cases: 16 participants in the canagliflozin group and 18 in the sitagliptin group. The registry reports group means and standard deviations for the changes.

The registry also describes blood pressure, body water content, body composition, cardiac function, and diet as secondary endpoints. Although the record lists two outcome measures, no secondary outcome results are posted in the information available for this analysis. The trial’s original sample-size rationale, detailed allocation procedures, and fuller protocol description are not reported in the registry record. The dates of the study are recorded as 2016-10 for the start and 2018-09 for primary completion; the status is completed.

Randomization and masking are features of the design, but they do not remove the need to consider the small analyzed groups, missing endpoint data, and the precision of the estimate. Nor do these design labels alone establish how allocation was generated, how masking was maintained, or how missing values were handled; those details are 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.

Overall Study

Analysis population for the primary outcome, as defined in the registry: Unadjusted p values were reported throughout, with statistical significance set at the 2-tailed 0.05 level. Only cases with available data used to compute the primary endpoint will be included in the analyses (16 subjects for canagliflozin group and 18 subjects for sitagliptin group).

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 Change From Baseline Aerobic Exercise Capacity at 12 Weeks, measured over baseline and 12 weeks.

The posted primary endpoint is change from baseline in peak oxygen consumption at 12 weeks. Peak oxygen consumption, often written as VO2, is the highest oxygen consumption measured during a maximal cardiopulmonary exercise test. It is used as a measure of aerobic exercise capacity: it reflects the ability to take in, transport, and use oxygen during exertion. The registry expresses this measure in mL/kg/min.

Using change from baseline focuses the comparison on how the measured value changed over the treatment period rather than on the 12-week value alone. In the posted results, the mean change was 0.67 for canagliflozin and -0.53 for sitagliptin. The comparison therefore concerns the difference between those average changes, not the difference between individual participants’ responses or a broader measure of health.

As a general statistical point, a continuous endpoint retains information about the size and direction of measured changes. Group means are useful for describing the average, while the standard deviation describes how spread out the participant-level changes are within a group. Neither summary reveals the full distribution of individual outcomes. Peak oxygen consumption is a measure of exercise capacity, but it should not be treated as a direct measure of every aspect of heart failure, diabetes, daily functioning, or long-term health.

The registry’s brief description also names the VE/VCO2 slope as a primary measure. A result for that measure is not reported in the registry record, so the interpretation of the posted primary result is limited to peak oxygen consumption.

7. Results Posted in the Registry

• Sitagliptin: mean -0.53 versus 0.67 on control; difference -1.20 (95% CI -2.51 to 0.11), standardized effect d = -0.62; z-test p = 0.072.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary measure is continuous, meaning it can take a range of numerical values rather than placing participants into categories. The reported group summaries are means and standard deviations. A mean describes the average observed change in a group. A standard deviation describes how dispersed individual changes are around that average; it is not the uncertainty in the mean itself. The registry lists the standard deviations as 2.1 for canagliflozin and 1.75 for sitagliptin.

What does the mean difference represent?

The recalculated comparison subtracts the canagliflozin group mean from the sitagliptin group mean. The resulting mean difference is -1.2 mL/kg/min. A negative sign follows from that stated order of subtraction: the sitagliptin mean change was lower than the canagliflozin mean change. If the subtraction order were reversed, the sign would also reverse, while the underlying contrast would be the same. The estimate is unadjusted: it compares the posted group summaries without adjustment for other variables.

A mean difference describes the contrast between averages, not the range of individual responses. It also does not show whether the distribution in either group is symmetric or whether particular participants had unusually large changes. Those details are not reported in the registry record. The pooled standard deviation and standardized effect size are available as recalculated summary measures, but they are not necessary to understand the clinical-scale mean difference and are not a substitute for the original measurement units.

How should the p-values be read?

A p-value describes how compatible the observed data are with a specified null hypothesis under the assumptions of the test. A smaller p-value indicates that data at least as incompatible with that null hypothesis would be less common if the null model and its assumptions held. A p-value is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect.

The registry states that unadjusted p-values were reported throughout and that statistical significance was set at the two-tailed 0.05 level. The recalculated unadjusted comparison gives p = 0.072. The sponsor’s reported ANOVA gives p = 0.083. Both exceed the stated threshold. The registry does not provide enough detail about the ANOVA to reconstruct its model or to determine whether it used additional covariates or other adjustments. The reported ANOVA result should therefore be described as reported, without attributing unlisted modeling choices to it.

Why report a confidence interval?

A confidence interval expresses uncertainty about an estimated quantity by giving a range of values compatible with the data under the calculation’s assumptions. It is useful alongside a point estimate because it conveys how precisely the comparison has been estimated. The recalculated interval for the mean difference includes zero. That means the interval includes both a value of no difference and values on either side of it; it does not identify a single precise effect from the posted data. The interval’s confidence level is not reported in the registry record, and the sponsor’s ANOVA result has no accompanying interval in the posted results.

In general, interval width depends on the amount and variability of the data as well as the calculation method. Larger standard errors generally correspond to wider intervals. A confidence interval is not a range containing a known proportion of individual participants’ outcomes; it concerns uncertainty in an estimated group-level quantity. It also does not account for every possible source of bias or uncertainty, including missing data, limitations in the design, and unreported analytic decisions.

Why can over-dispersion matter?

As a general statistical concept, over-dispersion occurs when observations vary more than a model’s assumptions allow. If a method assumes less variability than the data contain, its standard errors may be too small and its intervals too narrow, making results appear more precise than they are. For this continuous endpoint, the registry reports standard deviations for each group, but it does not report a diagnostic assessment of model assumptions. Over-dispersion is therefore a general consideration, not a documented finding in this trial.

More broadly, the validity of a statistical comparison depends on assumptions about the design and data, not just on the arithmetic of group means. Randomized allocation supports a between-group comparison, while masking may limit some influences on conduct or assessment. However, detailed information on allocation implementation, outcome assessment procedures, model checks, and handling of missing values is not reported in the registry record. The available summaries cannot verify these details.

10. Confidence Intervals and Precision

The recalculated interval for the unadjusted mean difference spans values below and above zero. The lower endpoint is -2.509; the upper endpoint is positive. Because zero lies within the interval, the posted data are compatible with no average difference as well as with differences in either direction under the interval calculation. The registry does not report the confidence level for this recalculated interval, so a confidence percentage should not be inferred from the posted information.

The point estimate is -1.2 mL/kg/min, but it is not a precise single value for the underlying group contrast. The interval indicates uncertainty around that estimate. The sponsor’s reported ANOVA p-value is 0.083, but the registry provides no interval for the sponsor’s analysis. A p-value alone does not show the range of effect sizes that are compatible with the data.

The group standard deviations are 2.1 for canagliflozin and 1.75 for sitagliptin, while the endpoint analysis includes 16 and 18 participants, respectively. These summaries help explain why uncertainty matters: the estimated group means are based on limited numbers of observations with participant-to-participant variation. The registry does not provide participant-level data or additional precision measures that would allow a fuller assessment of the outcome distributions.

In general, a confidence interval communicates sampling uncertainty under the assumptions used to calculate it. It does not guarantee that the true effect lies within its limits, and it cannot correct for bias or missing information. Here, the interval is best read together with the small analyzed groups, the unadjusted nature of the recalculation, and the absence of a sponsor-reported interval. Its inclusion of zero means the data do not pin down the direction of the group difference with high precision.

11. 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 planning figures are participants per arm, not total events. They are illustrations based on the observed effect, with equal allocation between two arms, a two-sided significance level of 0.05, and a normal approximation. Each planning row uses the stated fraction of the observed effect as the effect to detect. The original trial’s own sample-size assumptions and justification are not reported in the registry record.

The 80% power column gives the number of participants per arm in a hypothetical design that would have an 80% probability of detecting the specified effect at the stated significance level, under the planning assumptions. The 90% power column gives the corresponding per-arm planning number for a 90% probability. These are design calculations, not probabilities that the trial result is correct. For the full observed effect, the illustrations give 41 participants per arm for 80% power and 54 for 90% power. For planning at 75% of the observed effect, they give 72 and 96 participants per arm. For planning at 50% of the observed effect, they give 162 and 216 participants per arm.

Smaller effects are harder to distinguish from random variation, so the planning calculations require more participants when the assumed effect is reduced. These values depend on the stated assumptions and should not be treated as a reconstruction of the trial’s original planning process. The registry record does not report that process. This is not observed power.

12. 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, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -1.200; pooled standard deviation σ = 1.922.
2. n = 2 × (1.96 + 0.842)² × 1.922² / -1.200² = 40.3, rounded up to 41 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
delta <- -1.2                      # observed difference in means
sigma <- 1.92201473069277                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^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 (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(2920918)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 41 (80% power), 54 (90% power)
 75% of observed effect: n per arm = 72 (80% power), 96 (90% power)
 50% of observed effect: n per arm = 162 (80% power), 216 (90% power)
power.t.test: 41.3 
Simulated power at n = 41 per arm: 0.800 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-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.

13. Analyses Reported by the Sponsor

The registry reports a sponsor analysis using ANOVA, with p = 0.083 for the comparison between canagliflozin and sitagliptin. The registry does not provide an ANOVA effect estimate, confidence interval, test statistic, degrees of freedom, or further model details. Its stated significance threshold was two-tailed 0.05, and the reported p-value is above that threshold.

The recalculated comparison uses the posted group means and standard deviations to estimate an unadjusted mean difference of -1.2 mL/kg/min, with p = 0.072. This recalculation is a simple comparison of the available summary statistics. It is not a replacement for the sponsor’s analysis and does not reproduce undocumented aspects of the ANOVA.

The two p-values are close but arise from analyses described differently in the registry. They should be reported with their methods rather than blended into a single result. Neither is below the stated 0.05 threshold. The registry record does not report details that would explain the difference between the analyses, so no explanation for the distinct p-values can be established from the record alone. Both results should be interpreted alongside the estimated group means, the uncertainty of the recalculated contrast, and the limitations of the available-case analysis.

14. Multiplicity, Missing Data and Analysis Populations

The registry states that unadjusted p-values were reported and that the two-tailed significance threshold was 0.05. The study description names peak oxygen consumption and the VE/VCO2 slope as primary endpoints, and it lists several secondary measures, including blood pressure, body water content, body composition, cardiac function, and diet. The posted primary result available for this analysis concerns peak oxygen consumption. A result for the other endpoint and secondary measures is not reported in the registry record.

When multiple outcomes or comparisons are tested, the chance of at least one apparently unusual result can increase. Statistical methods sometimes address this multiplicity by adjusting significance thresholds or by specifying a hierarchy of outcomes. Whether such a strategy was planned or used is not reported in the registry record. The posted statement that p-values were unadjusted does not establish whether any broader multiplicity strategy was planned elsewhere in the protocol.

The endpoint description says that only cases with available data were included in the primary endpoint analysis: 16 participants in the canagliflozin group and 18 in the sitagliptin group. The record does not explain why other enrolled participants were not included in that analysis, whether data were missing for all participants excluded from it, or how any missing measurements were handled. It also does not report an intention-to-treat analysis, imputation, sensitivity analysis, or a comparison of included and excluded participants. The observed-case estimate should therefore be understood as based on those with available endpoint data, without assuming how missingness may have affected the result.

The registry does not provide further details about the analysis population, prespecified analysis plan, interim analyses, or handling of protocol deviations. These items are not reported in the registry record. The available information is sufficient to describe the posted endpoint summaries and analyses, but not to verify all decisions that may have shaped the statistical interpretation.

15. 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 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: occlusion superficial femoral artery (5.9%), heart failure (5.9%), dizziness and acute kidney injury (5.9%), influenza b (5.3%).

The registry lists serious adverse events affecting 3 of 17 participants at risk in the canagliflozin group and 1 of 19 participants at risk in the sitagliptin group. It lists other adverse events affecting 7 of 17 and 6 of 19 participants, respectively. Deaths are reported as 0 in each group. These are counts of participants affected as recorded in the adverse-event summary, not a statistical comparison of treatment groups.

The serious-event terms listed include occlusion superficial femoral artery, heart failure, dizziness and acute kidney injury, and influenza B. The registry gives percentages beside these terms, but does not provide enough context in the posted information to interpret them as comparative risks. The listed terms and counts do not establish that an event was caused by a study treatment. The registry record does not report a formal between-group safety analysis or uncertainty intervals for these figures.

These counts should be read in light of the small numbers at risk and the limited duration of treatment, which was 12 weeks. A difference in raw counts is descriptive; it does not establish a reliable treatment-group difference. The registry’s adverse-event summary also cannot answer questions about events not listed or about safety beyond the recorded study period.

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

The posted primary endpoint results describe average change in peak oxygen consumption over 12 weeks among participants with available endpoint data. The mean change was 0.67 mL/kg/min with canagliflozin and -0.53 mL/kg/min with sitagliptin. The recalculated unadjusted difference was -1.2 mL/kg/min for sitagliptin minus canagliflozin. The recalculated p-value was 0.072, while the sponsor’s reported ANOVA p-value was 0.083. Both are above the stated two-tailed 0.05 threshold.

These results do not establish that there was no difference between groups. The recalculated interval includes zero and extends to values on either side, so the direction and size of the underlying average contrast remain uncertain. At the same time, the observed estimates are not proof of a treatment effect. The p-values do not express the probability that either treatment is better, nor do they measure clinical importance.

The result addresses a specific exercise-capacity measure, in a small randomized active-control trial, over the stated treatment period. It does not answer whether either treatment changes the VE/VCO2 slope, because a result for that measure is not reported in the registry record. The listed secondary outcomes likewise have no posted results available in this record. The data do not establish longer-term outcomes or apply automatically to people with different characteristics from those enrolled.

In general, evidence from a trial should be considered in light of its estimate, uncertainty, design, and completeness of reporting. Here, the available-case endpoint analysis and lack of detail about missing data limit what can be inferred. The appropriate reading is descriptive and bounded: the registry reports these group means and statistical tests, while substantial uncertainty remains about the precise group contrast and about outcomes beyond the posted endpoint.

17. Why This Trial Matters Statistically

This trial illustrates why statistical interpretation should not stop at whether a p-value falls on one side of a threshold. The registry reports two forms of evidence for the primary comparison: an ANOVA p-value from the sponsor and a recalculated unadjusted mean difference with its interval and p-value. The p-values are above the stated 0.05 threshold, while the point estimate and interval help convey the direction and uncertainty of the measured contrast.

The endpoint is continuous, so the group means retain information about the size and direction of change in a way that a simple count of participants improving or worsening would not. At the same time, a mean is only one summary of the data. The standard deviations show variation, and the small analyzed groups make precision an important part of interpretation. Reporting both the estimated difference and its uncertainty prevents a non-significant test from being mistaken for proof of no difference.

The record also demonstrates the importance of distinguishing the trial’s observed sample from hypothetical design planning. Planning numbers based on the observed effect can illustrate how sample-size needs change when the target effect becomes smaller, but they do not explain why the completed trial enrolled its actual number of participants. That original rationale is not reported in the registry record.

Finally, transparent reporting matters alongside statistical calculation. The available-case population is described, but reasons for missing endpoint data and details of the sponsor’s analysis are not. The registry names other outcomes, yet results for them are not reported. These limitations shape what can be concluded from the primary result and show why estimates, uncertainty, analysis populations, and completeness of outcome reporting should be read together.

18. 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 endpoint analysis includes 16 participants in the canagliflozin group and 18 in the sitagliptin group, although enrollment was 36. The registry does not report why the analyzed endpoint counts differ from the enrollment total or how participants without available endpoint data were handled. This limits assessment of whether the available-case comparison could differ from an analysis of all randomized participants.

The posted recalculation is unadjusted, while the sponsor’s analysis is described as ANOVA. Model details for the sponsor analysis are not reported, and its effect estimate and interval are not posted. The registry also does not report diagnostic checks, detailed allocation methods, or a full account of prespecified analyses. As a result, the sponsor’s test cannot be fully reconstructed from the registry information.

The study description names two primary endpoints, but a comparative result for the VE/VCO2 slope is not reported. The record also lists secondary measures without posting their results. The handling of multiple outcomes and any multiplicity strategy are not reported in the registry record. These omissions limit interpretation beyond the peak oxygen consumption result.

The analyzed groups are small, and treatment lasted 12 weeks. The interval for the recalculated contrast includes zero, indicating substantial uncertainty about the group difference. The registry does not provide participant-level data or a confidence interval for the sponsor’s ANOVA result. The serious adverse event summaries are also descriptive, with no reported comparative statistical analysis. These trial-specific limitations mean that the posted statistics should be interpreted as evidence about the recorded comparison and period, not as a complete account of all relevant outcomes.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Treatment of Diabetes in Patients With Systolic Heart Failure — NCT02920918.
• Registry record last updated 2019-10-29; status last verified 2019-10.
• Results first posted 2019-10-29.
• 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.

22. Record Summary

NCT02920918 is a completed, randomized, parallel-group, triple-masked Phase 4 active-control study sponsored by Virginia Commonwealth University. It enrolled 36 participants with systolic heart failure and type 2 diabetes. The groups received canagliflozin 100 mg or sitagliptin 100 mg orally once daily for 12 weeks.

The posted primary result concerns change in peak oxygen consumption at 12 weeks, measured by maximal cardiopulmonary exercise test. Available endpoint data were reported for 16 participants in the canagliflozin group and 18 in the sitagliptin group. The mean changes were 0.67 and -0.53 mL/kg/min, respectively. The recalculated unadjusted difference was -1.2 mL/kg/min, with p = 0.072; the sponsor’s reported ANOVA p-value was 0.083.

The registry lists serious adverse events affecting 3 of 17 participants at risk with canagliflozin and 1 of 19 with sitagliptin, and reports 0 deaths in each group. Results for the other named primary endpoint and listed secondary outcomes are not reported in the registry record.