Safety and Efficacy of Dapagliflozin in Triple Therapy to Treat Subjects With Type 2 Diabetes: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Safety and Efficacy of Dapagliflozin in Triple Therapy to Treat Subjects With Type 2 Diabetes (NCT01646320), a randomized phase 3 study. The primary outcome is Adjusted Mean Change From Baseline in Hemoglobin A1C (HbA1c) at Week 24.

2. Clinical Question

The primary outcome is a continuous measure, expressed as percentage of glycosylated hemoglobin. The registry's reported treatment comparison is a mean difference from a longitudinal repeated measures analysis. The analysis excluded data collected after rescue medication, according to the endpoint description. The central statistical question is about the estimated contrast between the treatment groups and the precision of that contrast, not whether a particular change is clinically important for an individual. Clinical importance thresholds are not reported in the registry record.

3. Trial Design

This was a randomized, double-blind, placebo-controlled, parallel-group phase 3 trial. Its official description identifies a multicenter design. Participants were assigned to dapagliflozin added to saxagliptin and metformin immediate release, or matching placebo added to the same background medicines. The registry lists an enrollment of 320 and a completed status. It gives a start date of 2012-09 and a primary completion date of 2014-08.

The primary outcome was adjusted mean change in HbA1c from baseline to Week 24. The registry describes measurements during qualification and lead-in periods and on Day 1 and Weeks 6, 12, 18, and 24 in the double-blind period. Baseline was defined using the last assessment before the first dose of double-blind study medication, with a date-based fallback when assessment or dosing time was unavailable. These rules define the baseline value used in the analysis; they do not by themselves describe all details of the statistical model.

The primary endpoint population included randomized participants who received at least one dose of double-blind study medication during the double-blind treatment period. The posted primary arm summaries each include 158 participants. The registry does not explain how those analysis counts relate to the enrollment figure or describe the reasons any enrolled participants may not appear in the primary summaries. Details such as the original allocation ratio, planned sample-size justification, and trial-specific interim analyses are not reported in the registry record.

Four secondary outcomes are posted: fasting plasma glucose, 120-minute postprandial glucose, body weight, and the percentage achieving HbA1c below 7.0% at Week 24 using last observation carried forward. The listed quantitative secondary results cover three of these measures; a result for the glycemic-response percentage is not included in the posted outcome results.

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: Arm1: Dapagliflozin (10 mg) + Saxagliptin + Metformin IR Arm 2: Placebo + Saxagliptin + Metformin IR

Short-term Period (Day 1 to Week 24)

Reasons for not completing this period:

Long-term Period (Weeks 24 to 52)

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The Randomized Subjects Data Set consists of all randomized subjects who received at least one dose of double-blind study medication during the double-blind treatment period.

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: All randomized subjects who received at least one dose of study medication

6. Primary Endpoint

The registry lists the primary outcome as Adjusted Mean Change From Baseline in Hemoglobin A1C (HbA1c) at Week 24, measured over From Baseline to Week 24.

The primary endpoint is adjusted mean change from baseline in HbA1c at Week 24, measured as a percentage of hemoglobin by a central laboratory. HbA1c reflects glycosylated hemoglobin and is used in diabetes trials as a measure of glycemic status over time. As a continuous outcome, it allows the analysis to compare average changes across groups rather than only counting participants who cross a threshold.

Change from baseline focuses the comparison on how the measured value shifted during treatment relative to each participant's starting assessment. The registry defines baseline as the last assessment before the first dose of double-blind medication, using the date of dosing where the relevant times were unavailable. Measurements were scheduled during qualification and lead-in and at Day 1 and Weeks 6, 12, 18, and 24. The endpoint is specifically the adjusted mean change at Week 24, not an average of all measurements across the entire period.

The posted endpoint description says that observations after rescue medication were excluded from this analysis. It also identifies the analysis population as randomized participants who received at least one dose of double-blind study medication. Further details of the adjustment model, including the exact covariates and handling of all missing assessments, are not reported in the registry record. HbA1c is a surrogate measure; the posted endpoint does not directly measure clinical events or establish the importance of a change for an individual participant.

7. Results Posted in the Registry

• Pla+Saxa+Met: mean -0.10 versus -0.82 on control; difference 0.72 (95% CI 0.53 to 0.91), standardized effect d = 0.82; 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.

Adjusted Mean Change From Baseline in Fasting Plasma Glucose at Week 24

• 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 errors.
• Result:
  • Pla+Saxa+Met: mean -5.30 versus -32.70 on control; difference 27.40 (95% CI 19.37 to 35.43), standardized effect d = 0.75; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Pla+Saxa+Met than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Adjusted Mean Change From Baseline in 120-minute Postprandial Glucose (PPG) at Week 24

• 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 errors.
• Result:
  • Pla+Saxa+Met: mean -38.00 versus -73.50 on control; difference 35.50 (95% CI 24.19 to 46.81), standardized effect d = 0.75; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Pla+Saxa+Met than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Adjusted Mean Change From Baseline in Body Weight at Week 24

• 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 errors.
• Result:
  • Pla+Saxa+Met: mean -0.41 versus -1.91 on control; difference 1.50 (95% CI 0.88 to 2.12), standardized effect d = 0.53; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Pla+Saxa+Met than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted quantitative secondary outcomes show a consistent direction of estimated change across fasting plasma glucose, 120-minute postprandial glucose, and body weight: each group mean changed more in the dapagliflozin add-on group than in the placebo add-on group under the posted change scale. The registry's reported repeated measures comparisons for these outcomes are also in the direction of a greater reduction with the dapagliflozin combination. This broad pattern aligns with the primary HbA1c comparison in showing group differences across several glycemic measures, while body weight provides a distinct measure rather than another glycemic endpoint.

These secondary outcomes add context to the primary result, but they do not replace it. They assess different measures, have different available analysis counts in some cases, and may have different levels of precision. The registry lists a fourth secondary outcome, the percentage achieving HbA1c below 7.0% at Week 24 using last observation carried forward, but a corresponding result is not reported in the posted outcomes.

The secondary recalculations are unadjusted, and the registry does not report a multiplicity procedure for interpreting the full set of secondary comparisons. In general, when several outcomes or comparisons are examined, the chance of at least one apparently notable result can increase unless the analysis plan accounts for the multiple tests. Without a reported adjustment or hierarchy, these secondary results should be treated as supportive descriptions rather than as equally strong confirmatory evidence. The registry record does not report whether a prespecified testing sequence or other multiplicity control was used.

Differences in analysis counts also matter when considering the pattern: the fasting plasma glucose summaries include 158 and 157 participants, while postprandial glucose includes 134 and 132. Reasons for these varying counts are not reported in the registry record. They may affect precision and comparability, so the direction of the overall pattern should not be mistaken for identical evidence strength across endpoints.

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

• Adjusted Mean Change From Baseline in Fasting Plasma Glucose at Week 24
• Adjusted Mean Change From Baseline in 120-minute Postprandial Glucose (PPG) at Week 24
• Adjusted Mean Change From Baseline in Body Weight at Week 24
• Percentage of Subjects Achieving a Therapeutic Glycemic Response (Hemoglobin A1c [HbA1C]) <7.0% at Week 24 (Last Observation Carried Forward [LOCF])

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous: it records HbA1c change on a numerical scale. The registry posts adjusted means and standard errors for each arm, and a separate reported group contrast from a longitudinal repeated measures analysis. The arm summaries describe the estimated average change within each group; the contrast describes the estimated difference between groups. These quantities answer different questions and should not be mistaken for two estimates of the same arm-level value.

What does a mean difference describe?

A mean difference is a subtraction in a specified order. For the primary endpoint, the posted recalculation defines the contrast as the placebo combination minus the dapagliflozin combination, giving a positive estimate. The sponsor's reported contrast uses the opposite direction, giving a negative estimate. The sign conveys which group is subtracted from which; it does not mean the underlying findings conflict. When comparing results, check the group order and the outcome scale before interpreting the sign.

In general, an adjusted mean is a model-based estimate that accounts for factors included in the statistical model. The registry identifies the primary method as longitudinal repeated measures analysis and labels the endpoint summary as a least-squares mean, but the full model specification and adjustment factors are not reported in the registry record. The method name indicates repeated measurements were analyzed together; it does not disclose every assumption, covariance structure, or model diagnostic.

Why use repeated measurements?

In general, a longitudinal model can use measurements taken at several times to estimate patterns over follow-up while accounting for the fact that measurements from the same participant are related. This differs from treating every observation as unrelated. The registry lists HbA1c assessments at multiple visits and identifies a longitudinal repeated measures analysis for the reported contrast. The exact contribution of each visit to the Week 24 estimate, and the model's detailed handling of incomplete follow-up, are not reported in the registry record.

The endpoint description says that data after rescue medication were excluded from the primary analysis. That is an analysis rule for those observations, not a complete account of the missing-data strategy. The registry does not report whether other missing assessments were imputed, modeled under particular assumptions, or assessed through sensitivity analyses. Consequently, the posted estimate should be interpreted with the stated exclusion in mind, without assuming a particular unreported procedure.

What does the standard error tell the reader?

In general, a standard error describes the estimated sampling variability of an estimate, such as an adjusted mean. It is not the spread of individual participant measurements. A standard error is useful for understanding how precisely an estimated average is determined, while a confidence interval gives a range of values compatible with the analysis under its assumptions. The arm-level standard errors are posted for the primary and quantitative secondary endpoints. The sponsor's model-based confidence interval for the group contrast is the more direct precision summary for the primary comparison.

The displayed recalculated intervals are based on simple unadjusted calculations from posted arm summaries. For these summaries, the spread used in the calculation is approximated from the arm standard errors. Such a reconstruction does not reproduce a longitudinal model, its adjustments, or its use of repeated observations. It provides a transparent approximate contrast from the listed summary statistics, but it is not a substitute for the sponsor's analysis. The registry marks no discrepancy between the recalculation and the reported result.

How should p-values be read?

In general, a p-value measures how incompatible the observed data are with a specified null model, assuming that model and the analysis assumptions hold. It does not report the probability that the treatment has no effect, the probability that the result arose by chance in an everyday sense, or the size of an effect. The primary sponsor-reported value is p < 0.001. The posted unadjusted recalculation gives p < 0.001. Those labels arise from different analyses and should not be treated as competing trial analyses; the registry's own repeated measures analysis is the primary reported comparison.

All inferential conclusions are conditional on the analysis population, endpoint definition, model assumptions, and handling of observations. The registry provides the primary population definition and rescue-medication exclusion, but not the complete statistical analysis plan. This limits the ability to independently assess model diagnostics, prespecification, missing-data assumptions, or the full set of analyses considered.

11. Confidence Intervals and Precision

The sponsor's primary estimate is a mean difference of -0.72, with a 95% confidence interval from -0.91 to -0.53. The interval expresses uncertainty about the group contrast under the reported longitudinal repeated measures analysis. In general, a 95% confidence interval is produced by a procedure that, over repeated use under its assumptions, would contain the underlying parameter 95% of the time. It is not a statement that there is a 95% probability that this particular fixed parameter lies within this particular interval.

The interval is entirely below zero in the sponsor's reported direction. This indicates that the estimated contrast is not close to zero according to that interval, while its endpoints show a range of plausible values under the model. It does not tell whether that range is clinically important: the registry record gives no threshold for a clinically meaningful HbA1c difference. Precision and importance are separate considerations.

The simple recalculation reports 0.72 with a 95% confidence interval from 0.527 to 0.913, in the reverse contrast direction. Its decimal detail reflects the recalculation and should not be read as greater methodological precision than the sponsor's model-based estimate. The two confidence intervals are not identical, because one is derived from posted arm summaries and the other comes from the reported longitudinal repeated measures analysis. The sponsor's analysis takes precedence.

In general, an interval's width reflects both the variability in the data and the amount of information available, as well as the analysis method. Standard errors posted for adjusted arm means help describe uncertainty in those means, but they are not individual-level variability measures. The registry does not report enough model detail to independently reconstruct the sponsor's interval or assess all sources of uncertainty. Confidence intervals also do not account for unreported analysis choices or multiplicity unless those are incorporated into the method.

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 display is a planning illustration based on the observed effect, not a record of the trial's original design calculations. Its figures are participants per arm, not total events. The assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation; each planning scenario targets the stated fraction of the observed effect. The trial's own sample-size justification and original design assumptions are not reported in the registry record.

The 80% power column describes the number of participants per arm estimated to give an 80% chance of detecting the specified effect under those planning assumptions, if that effect is the true one and the statistical model assumptions apply. The 90% power column uses a higher target chance of detection under the same scenario structure. These are planning probabilities, not guarantees that a future trial will produce a statistically significant result.

Rows for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect show how the planning estimate changes when the assumed target effect changes. Smaller effects need more participants to attain the same target power, all else equal. The curve conveys that relationship across effect assumptions; it does not forecast an outcome for a future study or reveal the original enrollment rationale. These figures are not observed power: they do not measure the probability that this completed trial detected its effect after the results were known.

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

Worked calculation for this trial

1. Observed difference Δ = 0.720; pooled standard deviation σ = 0.874.
2. n = 2 × (1.96 + 0.842)² × 0.874² / 0.720² = 23.1, rounded up to 24 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 <- 0.72                      # observed difference in means
sigma <- 0.873674699187289                     # 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(1646320)
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 = 24 (80% power), 31 (90% power)
 75% of observed effect: n per arm = 42 (80% power), 56 (90% power)
 50% of observed effect: n per arm = 93 (80% power), 124 (90% power)
power.t.test: 24.1 
Simulated power at n = 24 per arm: 0.802 (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.

14. Analyses Reported by the Sponsor

The sponsor's reported primary analysis is a longitudinal repeated measures analysis, with a mean difference in final values of -0.72, a 95% confidence interval from -0.91 to -0.53, and p < 0.001. This is the trial's own reported analysis and is the result to prioritize when describing its primary comparison.

The recalculation uses the posted arm-level means and standard errors to form a simple unadjusted contrast. It reports 0.72, with a 95% confidence interval from 0.527 to 0.913 and p < 0.001, using placebo minus dapagliflozin rather than the sponsor's reverse ordering. The opposite signs reflect the contrast direction. The approximate calculation does not reproduce the repeated measures model, its adjustments, or the use of repeated observations.

The registry does not flag a discrepancy between the recalculation and the sponsor analysis. Their estimates are consistent in direction after accounting for the reversed group order, and their intervals are compatible in magnitude. Differences in the interval endpoints and p-value presentation should be understood in light of the distinct methods and rounding. The recalculation is a secondary descriptive check, not an alternative to the sponsor's model-based result.

15. Multiplicity, Missing Data and Analysis Populations

The registry reports a primary endpoint and several secondary outcomes, but a multiplicity strategy is not reported in the registry record. Multiplicity concerns arise when statistical tests are conducted for more than one outcome or contrast: interpreting every nominal p-value as if it were the only planned test can overstate the strength of evidence across the entire family of analyses. Without details of any testing hierarchy, adjustment, or other control, the secondary comparisons are best considered supportive and descriptive rather than independent confirmatory findings.

The primary endpoint description states that measurements after rescue medication were excluded from the analysis. Baseline has a specific definition tied to the last assessment before the first double-blind dose, with a date-based fallback when times are unavailable. These details help define the endpoint and analysis set, but do not reveal how all other missing Week 24 or intermediate measurements were handled. A missing-data method, assumptions about missingness, and sensitivity analyses are not reported in the registry record.

The primary analysis population consists of randomized participants who received at least one dose of double-blind medication during the double-blind treatment period. This is not described as an analysis of every randomized participant regardless of treatment receipt. The registry does not report a full participant flow, reasons for any exclusions from endpoint analyses, or whether additional analysis populations were used. It also does not report the complete statistical analysis plan, model covariates, or diagnostic assessments.

These reporting gaps do not invalidate the posted estimates, but they constrain independent evaluation of robustness. The reported confidence interval and p-value are conditional on the sponsor's model and analysis decisions. Readers cannot determine from the registry record alone whether the findings remain similar under alternative missing-data assumptions, how all planned tests were organized, or whether interim analyses affected the testing framework.

16. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 2% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: thrombocytopenia (0.6%), acute myocardial infarction (0.6%), angina unstable (0.6%), atrial fibrillation (0.6%), cardiac failure (0.6%).

The registry posts serious adverse event counts for 160 participants at risk in each group: 7 affected participants in the dapagliflozin, saxagliptin, and metformin group, and 4 in the placebo, saxagliptin, and metformin group. These are counts of participants with a serious adverse event, not counts of all event episodes. The record also lists other adverse events affecting 103 participants and 112 participants, respectively, among 160 at risk in each group. Those figures should be read as participant counts, not as severity-weighted summaries or evidence of causation.

Several serious event terms are listed with a frequency of 0.6% each: thrombocytopenia, acute myocardial infarction, angina unstable, atrial fibrillation, cardiac failure, and cardiac failure acute. The registry record does not provide enough detail to infer how these listed terms relate to the total serious-event counts or to attribute events to treatment. Death counts are not reported in the registry record.

These descriptive safety counts do not establish that a regimen caused or prevented an event. The record's duration and event summaries also do not characterize all possible safety outcomes. Comparative interpretation should account for the limited counts and the absence of further detail on event definitions, timing, or attribution.

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

The posted results describe average changes over the Week 24 endpoint period among the participants included in the respective analyses. For the primary outcome, the sponsor's longitudinal repeated measures analysis estimates a group difference in HbA1c change, with a confidence interval that is entirely on one side of zero under its stated direction. The statistical result is evidence about a group-level comparison under the model; it is not a prediction of the response of any particular participant.

A mean can conceal variation among individuals. The registry's arm-level averages do not show how many participants had a larger, smaller, or opposite change, and the posted result does not establish that every participant experienced the average pattern. It also does not determine the clinical importance of the observed contrast, because a meaningful-change threshold is not reported. Statistical evidence and clinical relevance are related questions, but they are not interchangeable.

The endpoint is a laboratory measure, and the analyses concern the stated follow-up and population. They do not establish outcomes beyond Week 24, effects on unreported clinical endpoints, or long-term safety. The results also depend on the primary population definition and the exclusion of data after rescue medication. The registry record does not provide the complete missing-data strategy or enough detail to assess all model assumptions independently.

The trial compares two treatment strategies that share saxagliptin and metformin, differing by dapagliflozin versus matching placebo. Accordingly, the reported contrast pertains to adding dapagliflozin within that background regimen and in the studied population. It should not be generalized automatically to other populations, treatment combinations, durations, or clinical outcomes. No medical advice or individual treatment conclusion follows from these group summaries alone.

18. Why This Trial Matters Statistically

This trial illustrates why a statistical result should be read through its endpoint, analysis population, contrast direction, and method rather than by a p-value alone. The primary outcome is a continuous measure, and the registry provides both adjusted arm summaries and a sponsor-reported longitudinal repeated measures comparison. Understanding that the reported contrast and the simple recalculation use opposite group orders prevents a sign difference from being mistaken for a disagreement about the direction of the estimate.

It also shows the distinction between a model-based trial result and a reconstruction from summary statistics. The sponsor's analysis uses the stated repeated measures method; the recalculation is an approximate unadjusted comparison derived from the posted summaries. The latter is useful for checking scale and direction but cannot reproduce the original model. Giving precedence to the reported trial analysis while treating the reconstruction as supplementary is an important statistical reading practice.

The secondary results illustrate another general issue: several outcomes may point in a similar direction without carrying equal confirmatory weight. The registry does not report a multiplicity plan, and some secondary analysis counts differ. A coherent pattern is informative context, but it does not remove uncertainty about how multiple comparisons were handled or why outcome availability varied.

Finally, the sample-size illustrations show how planning assumptions depend on the target effect and desired power, while remaining separate from the trial's original design rationale. The registry does not report that rationale. Keeping planning calculations distinct from observed results, and distinguishing precision from clinical importance, helps readers avoid overinterpreting either the estimates or their statistical significance.

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.

Several trial-specific reporting limitations constrain interpretation. The registry identifies the primary analysis as longitudinal repeated measures but does not report the full model specification, including adjustment factors and covariance assumptions. The sponsor's result is therefore available for interpretation, but its analysis cannot be fully reconstructed from the registry record alone.

The primary endpoint excludes measurements after rescue medication. The registry does not report a complete strategy for other missing data, sensitivity analyses, or the reasons for differing participant counts across secondary outcomes. It is not possible to determine from the posted descriptions how those differences arose or whether alternative missing-data assumptions would materially change the estimates.

The registry record does not report the original sample-size justification, trial-specific interim analyses, or a multiplicity plan. The planning illustrations based on observed effects cannot fill those gaps. The record also does not provide a result for the listed percentage achieving HbA1c below 7.0% outcome in the posted results.

Safety reporting is limited to the event counts and terms listed. Death counts, event timing, attribution, and further details needed to interpret the listed serious events are not reported in the registry record. Finally, the principal efficacy assessment is at Week 24; the posted findings do not establish longer-term effects or outcomes beyond the measures reported.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Safety and Efficacy of Dapagliflozin in Triple Therapy to Treat Subjects With Type 2 Diabetes — NCT01646320.
• Registry record last updated 2016-06-22; status last verified 2016-05.
• Results first posted 2016-04-11.
• 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

This completed, randomized, double-blind, parallel-group phase 3 trial compared dapagliflozin 10 mg added to saxagliptin and metformin immediate release with matching placebo added to the same background therapy in participants with type 2 diabetes and inadequate glycemic control. The primary endpoint was adjusted mean change in HbA1c from baseline to Week 24.

The sponsor's longitudinal repeated measures analysis reported a mean difference of -0.72, with a 95% confidence interval from -0.91 to -0.53 and p < 0.001. The recalculation from arm summaries expresses the contrast in the reverse direction and is approximate; it does not replace the sponsor's analysis. Posted quantitative secondary outcomes show a similar direction of group differences, while multiplicity and several analysis details are not reported in the registry record.