Glycemic Efficacy and Renal Safety Study of Dapagliflozin in Subjects With Type 2 Diabetes Mellitus and Moderate Renal Impairment: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Glycemic Efficacy and Renal Safety Study of Dapagliflozin in Subjects With Type 2 Diabetes Mellitus and Moderate Renal Impairment (NCT00663260), a randomized phase 2, phase 3 study. The primary outcome is Adjusted Mean Change From Baseline in Hemoglobin A1C (HbA1c) at Week 24 (Last Observation Carried Forward [LOCF].

2. Clinical Question

The clinical question is about a measured change in glycemic control in a specific trial population, not about every person with type 2 diabetes or every level of renal function. The registry record describes the study purpose as evaluating glycemic efficacy and renal safety in participants with moderate renal impairment and inadequate glycemic control. It also identifies pharmacokinetics and pharmacodynamics among the broader study objectives. The primary posted outcome, however, is the Week 24 HbA1c change.

The posted results compare each dose separately with placebo. The key statistical question is whether the estimated between-group changes are sufficiently distinct from no difference, and how uncertain those estimates are. The reported confidence intervals and p-values are relevant to that question, but they do not by themselves establish the clinical importance of a difference. The registry record does not report a threshold defining a clinically important HbA1c change.

3. Trial Design

This was a randomized, double-blind, placebo-controlled, parallel-group Phase 2/3 trial sponsored by AstraZeneca. The registry record lists enrollment of 631 and identifies the study as completed. The recorded start was 2008-06, and primary completion was 2009-12. These are registry dates, not a description of each participant’s follow-up.

The three listed groups were placebo, dapagliflozin 5 mg, and dapagliflozin 10 mg. The primary outcome results were based on 82 participants in the placebo group, 83 in the 5 mg group, and 82 in the 10 mg group. Those endpoint analysis counts should not be read as the overall enrollment or as the number randomized to each group: the primary outcome population required receipt of study medication and nonmissing baseline and Week 24 HbA1c values under the stated last-observation-carried-forward approach.

Random allocation is intended to make treatment groups comparable at entry on average, while double masking is intended to limit the influence of treatment knowledge on participants’ experiences and outcome assessment. A parallel-group design assigns participants to a group that is compared with the other groups over the trial period. These are general explanations of the design terms; the registry record does not report the randomization sequence, allocation concealment procedures, or the people covered by masking.

The official title identifies the trial as multicenter, but the registry record does not report the number or locations of centers. It also does not report the original design assumptions, planned group allocation ratio, or sample-size justification. Enrollment of 631 is a recorded trial characteristic and should not be taken as evidence of the reasoning used to select that target.

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: All randomized participants who received study medication and had nonmissing HbA1c values at baseline and Week 24 (LOCF)

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 participants who received at least 1 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 (Last Observation Carried Forward [LOCF], measured over From Baseline to Week 24.

The primary endpoint was adjusted mean change from baseline in hemoglobin A1c (HbA1c) at Week 24, reported as a percentage of hemoglobin. HbA1c is a continuous laboratory measure, so the outcome is expressed as a mean change rather than as a count of participants meeting a threshold. The registry describes measurement by a central laboratory and identifies Week 24 as the endpoint time point.

Change from baseline expresses the Week 24 measurement relative to a participant’s starting measurement. Comparing mean changes between randomized groups helps address whether the average outcome differed between assigned treatments and placebo. The registry calls the endpoint adjusted, and its reported group comparisons use ANCOVA. The particular covariates and adjustment specification are not reported in the registry record.

The analysis excluded data collected after rescue medication. Baseline was defined using the last assessment before the first dose, with a date-based fallback when assessment or dose time was unavailable. The registry lists HbA1c assessments across qualification and lead-in periods, then on Day 1 and at multiple visits through Week 24. For the primary analysis population, Week 24 values used last observation carried forward (LOCF). LOCF carries a participant’s last available observation forward to represent the endpoint; it is an imputation choice and relies on assumptions about what an unobserved later value would have been. The registry record does not report a sensitivity analysis addressing those assumptions.

7. Results Posted in the Registry

• Dapagliflozin 5 mg: mean -0.41 versus -0.32 on control; difference -0.09 (95% CI -0.56 to 0.38), standardized effect d = -0.06; z-test p = 0.708.
• Dapagliflozin 10 mg: mean -0.44 versus -0.32 on control; difference -0.12 (95% CI -0.59 to 0.35), standardized effect d = -0.08; z-test p = 0.619.

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 (FPG) at Week 24 (Last Observation Carried Forward [LOCF])

• 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:
  • Dapagliflozin 5 mg: mean -5.20 versus 8.40 on control; difference -13.60 (95% CI -40.17 to 12.97), standardized effect d = -0.16; z-test p = 0.316.
  • Dapagliflozin 10 mg: mean -0.60 versus 8.40 on control; difference -9.00 (95% CI -35.53 to 17.53), standardized effect d = -0.10; z-test p = 0.506.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Dapagliflozin 5 mg and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Dapagliflozin 10 mg and the control. The interval gives the range of effects that remain compatible with the data.

Adjusted Mean Change From Baseline in Total Body Weight (kg) at Week 24 (Last Observation Carried Forward [LOCF])

• 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:
  • Dapagliflozin 5 mg: mean -1.54 versus 0.27 on control; difference -1.81 (95% CI -3.15 to -0.47), standardized effect d = -0.41; z-test p = 0.008.
  • Dapagliflozin 10 mg: mean -1.89 versus 0.27 on control; difference -2.16 (95% CI -3.49 to -0.83), standardized effect d = -0.49; z-test p = 0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Dapagliflozin 5 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Dapagliflozin 10 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted secondary outcomes cover fasting plasma glucose (FPG) and total body weight at Week 24. Across these measures, the group patterns are not uniform in their strength or statistical precision. The posted comparisons show that the secondary outcomes can yield a different pattern from the primary HbA1c analyses, so they add context rather than simply repeating the primary result.

FPG and body weight are separate continuous outcomes with their own measurement scales, variability, analysis populations, and reported comparisons. Their direction and size cannot be directly equated with an HbA1c difference. The registry reports ANCOVA analyses for these outcomes as well as the group summaries. As with the primary endpoint, the registry record does not give the covariates and full model specification for those ANCOVA analyses.

The unadjusted recalculations for secondary outcomes are exploratory readings of posted summary statistics. They should carry less weight than a prespecified primary outcome because multiple outcomes and comparisons create more opportunities for apparently notable results by chance. The registry record does not report a multiplicity adjustment across the secondary outcomes or a hierarchy that would control how they are interpreted. A nominally small unadjusted p-value therefore does not, by itself, establish a confirmatory finding.

The secondary results can be considered alongside the primary outcome to assess whether the overall pattern is consistent, but they do not erase the uncertainty in the primary comparisons. Nor do the secondary outcomes answer the trial’s other stated objectives concerning renal safety, pharmacokinetics, or pharmacodynamics. Those questions require their own reported outcomes and analyses; they are not settled by FPG or body-weight results.

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

• Adjusted Mean Change From Baseline in Fasting Plasma Glucose (FPG) at Week 24 (Last Observation Carried Forward [LOCF])
• Adjusted Mean Change From Baseline in Total Body Weight (kg) at Week 24 (Last Observation Carried Forward [LOCF])

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous: each participant contributes an HbA1c measurement, and the group-level result is expressed as a mean change. The primary results list means and standard errors. A mean describes the group’s average observed change; a standard error describes the statistical uncertainty around the estimated mean, rather than the spread of individual participant values. Standard errors are useful for inference about means, but they are not a description of the range of responses among individuals.

What does the randomized comparison estimate?

The treatment comparison is a difference between group means, not a comparison of whether either group changed from baseline in isolation. A negative mean difference in the posted analyses indicates a more negative average change in the dapagliflozin group than in placebo, given the comparison’s stated ordering. Randomization supports interpreting the contrast as a comparison of assigned groups, subject to the trial’s conduct, analysis population, and missing-data choices. It does not make every observed group difference necessarily causal or clinically important.

The recalculated comparisons are simple unadjusted calculations from the posted summary statistics. They compare the listed group means and use the available group-level precision information to quantify uncertainty. Because they are not the sponsor’s fitted ANCOVA model, they cannot reproduce model adjustments that are not contained in the summary statistics. The recalculated p-values are consequently descriptive checks, not replacements for the registry’s reported inferential analyses.

What does ANCOVA add?

Analysis of covariance (ANCOVA) compares outcomes between groups while adjusting for specified covariates. In a baseline-adjusted setting, ANCOVA can account for baseline outcome values and may improve precision when the adjustment variables are appropriate and the model is correctly specified. This is a general explanation of ANCOVA. Although the registry identifies ANCOVA as the sponsor’s method, the covariates, model details, and exact analysis specification are not reported in the registry record. The posted ANCOVA estimates should therefore be presented as reported, without inventing details about how the adjustment was performed.

The registry labels the reported contrast “Mean Difference (Final Values),” while the primary outcome is described as adjusted mean change from baseline. That wording reflects the registry’s reporting labels; it is not enough to infer additional model details. In particular, no unreported adjustment variables or transformations should be assumed.

How should a p-value be read?

A p-value measures how unusual data at least as incompatible with a specified null hypothesis would be under the assumptions of the analysis. It is not the probability that the null hypothesis is true, and it does not measure the size or importance of an effect. The reported primary p-values are 0.561 and 0.435. Read alongside the confidence intervals, they indicate that these analyses do not provide strong evidence against a zero between-group difference. The registry record does not state a multiplicity adjustment for the primary comparisons, so these p-values should not be treated as if a particular family-wise error procedure had been documented.

What is the role of the analysis population?

The primary analysis population included randomized participants who received study medication and had nonmissing baseline and Week 24 HbA1c values using LOCF. This is narrower than a description of all randomized participants. Requiring endpoint information and receipt of study medication can affect which participants contribute to the estimate. The registry record does not report a complete participant flow for this endpoint or enough detail to compare the primary analysis with an intention-to-treat analysis.

Last observation carried forward fills a missing Week 24 value using the participant’s last available observation. This allows a stated analysis population to include participants without a directly observed Week 24 value, but it treats the carried-forward observation as representative of the endpoint. That assumption may not reflect the participant’s subsequent course. General missing-data methods differ in their assumptions; the registry record does not report sensitivity analyses or other missing-data approaches for the primary outcome.

Why can dispersion matter?

Participants’ outcomes vary, and the amount of variation affects how precisely a group mean or between-group contrast can be estimated. Over-dispersion is a general concern when observed variability exceeds what an assumed model accommodates; if not addressed, standard errors and tests can be misleading. The posted primary outcome is continuous and reports standard errors, while the sponsor reports ANCOVA confidence intervals. The registry record does not identify an over-dispersion issue or a special method to address one. No such issue should be inferred from the reported estimates alone.

All statistical interpretations are conditional on the analysis definitions, model assumptions, and outcome availability. The registry record does not report a protocol-level statistical analysis plan, the full model specification, or details of interim analyses. Their absence from the registry record is not evidence that a procedure was or was not used elsewhere; it limits what can be concluded from the registry record itself.

11. Confidence Intervals and Precision

A confidence interval gives a range of estimates compatible with the data under the analysis method and its assumptions. It communicates uncertainty around an estimated contrast; it is not a guarantee that the true effect lies in the interval, nor is it a range containing a stated proportion of individual participant outcomes. The sponsor reports 95% confidence intervals for both primary comparisons.

For dapagliflozin 5 mg versus placebo, the sponsor’s estimate was -0.08, with a 95% confidence interval from -0.37 to 0.20. For dapagliflozin 10 mg versus placebo, the estimate was -0.11, with a 95% confidence interval from -0.40 to 0.17. Both intervals cross zero, meaning the estimates compatible with the data include no between-group difference as well as differences in either direction. This is consistent with the reported p-values of 0.561 and 0.435.

The intervals span a range of possible effects, so the point estimates alone would overstate precision. The registry record does not give a threshold for a clinically important difference. Without such a benchmark, the intervals can be described statistically, but not classified against a trial-defined clinical margin. The registry record also does not report a noninferiority or equivalence margin, so crossing zero should not be interpreted as proof that treatments are equivalent.

The recalculated intervals use unadjusted summary-statistic methods, whereas the sponsor’s intervals come from ANCOVA. They are therefore not interchangeable. Differences in estimates or interval widths can arise from the analysis method and adjustment, and the registry record does not provide enough model detail to attribute any difference to a specific covariate or modeling choice. The sponsor’s intervals are the relevant intervals for the sponsor-reported ANCOVA analyses.

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 planning illustrations translate an observed effect into hypothetical sample-size requirements under stated assumptions. For the primary HbA1c comparisons, the figures are participants per arm, not total participants across both groups and not total events. A row based on 100% of the observed effect uses the observed comparison as the planning target; rows using 75% or 50% of that effect consider smaller effects. For example, the primary comparison planning figures include 4,627 participants per arm for 80% power and 6,194 participants per arm for 90% power when planning for 100% of the observed 5 mg comparison. For the 10 mg comparison, the corresponding figures are 2,598 and 3,477 participants per arm.

The stated planning assumptions 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 80% power column describes a hypothetical design that would have an 80% chance of rejecting the null under the assumptions if the target effect were true. The 90% column describes the analogous design with a 90% chance. These are planning probabilities, not probabilities that the treatment works or that a particular result is true.

Smaller target effects require more participants to distinguish them from random variation under the same assumptions. The listed figures are not the trial’s original design justification: the original sample-size assumptions are not reported in the registry record. These calculations are planning illustrations based on observed effects, not observed power. They should not be read as evidence about the probability that the completed trial detected an effect.

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 = -0.090; pooled standard deviation σ = 1.545.
2. n = 2 × (1.96 + 0.842)² × 1.545² / -0.090² = 4,626.3, rounded up to 4,627 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 <- -0.09                      # observed difference in means
sigma <- 1.54503869842535                     # 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(663260)
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 = 4,627 (80% power), 6,194 (90% power)
 75% of observed effect: n per arm = 8,225 (80% power), 11,011 (90% power)
 50% of observed effect: n per arm = 18,506 (80% power), 24,774 (90% power)
power.t.test: 4627.2 
Simulated power at n = 4,627 per arm: 0.794 (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 registry reports sponsor analyses using ANCOVA for both primary treatment-placebo comparisons. The estimate for dapagliflozin 5 mg was -0.08, with a 95% confidence interval from -0.37 to 0.20 and p = 0.561. The estimate for dapagliflozin 10 mg was -0.11, with a 95% confidence interval from -0.40 to 0.17 and p = 0.435. These are the sponsor’s reported inferential results.

The recalculated estimates from posted group summaries are -0.09 for 5 mg and -0.12 for 10 mg, with unadjusted p-values of 0.708 and 0.619. These calculations are simple, unadjusted comparisons and are not a reconstruction of the ANCOVA model. Their purpose is to offer a check using the group-level summaries, while the sponsor’s analysis remains distinct because it uses the reported ANCOVA method.

The registry record does not flag a discrepancy between the reported and recalculated analyses. That does not make the methods identical. The estimates and p-values should be interpreted in the context of their respective approaches, and the sponsor’s reported ANCOVA results should not be replaced by the unadjusted recalculations. The covariates and other model details needed to reproduce the sponsor analysis are not reported in the registry record.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when several hypotheses are tested: even if each individual test uses a conventional significance threshold, testing multiple outcomes or comparisons increases the chance of at least one apparently notable result under a collection of null hypotheses. This is a general statistical concern, not evidence that a particular reported result is false. The registry record reports two primary treatment-placebo comparisons and two secondary outcomes, but does not report whether or how multiplicity was controlled.

The primary p-values are 0.561 and 0.435, and both sponsor-reported confidence intervals include zero. The secondary outcomes have their own comparisons, but the registry record does not identify a testing hierarchy, adjustment procedure, or confirmatory status that would clarify how their results were intended to be interpreted together. Unadjusted secondary p-values therefore warrant caution, especially when considered alongside multiple outcomes and dose comparisons.

Missing data also matter. The primary analysis used participants who received study medication and had nonmissing baseline and Week 24 HbA1c values, with last observation carried forward for the endpoint. Data after rescue medication were excluded. These choices define which measurements contributed and how some endpoint values were handled. LOCF is a particular imputation approach, not a guarantee that the carried-forward value matches what would have been measured at Week 24.

The registry record does not report the number of participants with missing primary endpoint values, reasons for missingness, treatment discontinuations, or sensitivity analyses under alternative missing-data assumptions. It also does not report whether the primary comparison was analyzed in other populations. Thus the robustness of the primary result to missing-data assumptions cannot be assessed from the registry record alone.

The registered primary analysis population is not simply described as all randomized participants: it requires medication receipt and qualifying HbA1c measurements. The record does not report the original statistical analysis plan, detailed analysis-population definitions beyond the posted primary description, or interim analyses. Those items are not reported in the registry record, so no protocol-specific conclusions about them are warranted.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: acute myocardial infarction (2.4%), hypoglycaemia (2.4%), cardiac failure (1.2%), cardiomyopathy (1.2%), cellulitis (1.2%).

The registry reports serious adverse event counts for the placebo, dapagliflozin 5 mg, and dapagliflozin 10 mg groups. The numbers at risk for the serious-event summary were 84, 83, and 85, respectively; the corresponding numbers with serious adverse events were 9, 7, and 12. The record also lists other adverse event counts: 39 among 84 at risk with placebo, 45 among 83 at risk with dapagliflozin 5 mg, and 42 among 85 at risk with dapagliflozin 10 mg.

These are counts within the registry’s reported groups, not adjusted comparisons or evidence that a treatment caused an event. The serious-event terms listed include acute myocardial infarction and hypoglycaemia at 2.4% each, and cardiac failure, cardiomyopathy, cellulitis, and pneumonia at 1.2% each. The registry record does not provide enough detail to attribute those events to treatment or compare risks statistically.

Deaths are not reported in the registry record. Safety counts should be interpreted alongside the trial’s population, follow-up, event definitions, and completeness of reporting; further details on those points are not reported in the registry record.

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

The primary results describe average HbA1c change at Week 24 in the trial’s stated analysis population. The sponsor’s ANCOVA comparisons estimate small negative differences for both dapagliflozin doses relative to placebo, but both 95% confidence intervals include zero and both reported p-values are above conventional significance thresholds. The data therefore do not provide clear statistical evidence of a between-group difference in these reported analyses. This is not the same as proving that the groups are identical.

The interval estimates retain uncertainty about the size and direction of the underlying difference. Since the registry record does not state a clinically important difference threshold, it is not possible to classify the estimates against a trial-defined threshold. No equivalence or noninferiority margin is reported, so the results should not be described as demonstrating equivalence or noninferiority.

The results are specific to participants with type 2 diabetes and moderate renal impairment who met the primary analysis population definition, and to the Week 24 outcome analyzed using the stated LOCF approach and exclusion of post-rescue data. They do not establish what would happen in people with a different clinical profile, under a different treatment context, or over a different follow-up period. Those broader questions are not answered by the posted primary comparison.

The trial also had stated objectives beyond HbA1c, including renal safety, pharmacokinetics, and pharmacodynamics. The primary glycemic result should not be used as a stand-in for those outcomes. The posted secondary FPG and body-weight outcomes add outcome-specific information, but multiplicity and analysis details limit their confirmatory interpretation. The serious adverse event counts are descriptive and do not by themselves establish treatment causation or comparative safety.

These statistical findings are not medical advice. They describe the registry’s reported estimates, uncertainty, and analysis choices, and do not determine an individual treatment decision.

18. Why This Trial Matters Statistically

This trial is statistically instructive because it presents a randomized, double-blind, placebo-controlled comparison in a population defined by both type 2 diabetes and moderate renal impairment. The primary endpoint is continuous, and its reported treatment contrasts are accompanied by confidence intervals and p-values. Together, those features make it possible to distinguish the direction of the estimated group differences from the uncertainty around them.

The record also illustrates why within-group change and between-group effect are different questions. Each group’s mean HbA1c change is negative, but the primary treatment question concerns whether the changes differ between a dapagliflozin group and placebo. A pattern of change within every arm does not alone identify an intervention-specific difference.

Another important lesson is that sponsor-reported adjusted analyses and simple unadjusted recalculations answer related but nonidentical statistical questions. The ANCOVA estimates should be read as the reported trial analyses; summary-based recalculations can provide a transparent check but cannot recover an undocumented model specification. Reporting both kinds of estimates makes the distinction between model-based inference and unadjusted summary comparisons consequential rather than merely technical.

Finally, the registry exposes limits on interpretation: the primary population uses specific outcome-availability and rescue-medication rules, secondary outcomes raise multiplicity considerations, and original design assumptions and several analysis details are not reported in the registry record. These gaps show why effect estimates, uncertainty, analysis populations, and documentation all matter when evaluating a clinical trial. A p-value alone cannot convey the full strength or limitations of the evidence.

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 primary analysis population is restricted to randomized participants who received study medication and had nonmissing baseline and Week 24 HbA1c values under the LOCF definition. The registry record does not report the full participant flow or the number and reasons for missing measurements, making it difficult to assess how selection into this analysis may affect interpretation. Excluding data after rescue medication also means the primary estimate does not use those later measurements.

LOCF assumes that the last available measurement is a suitable stand-in for the Week 24 value. The registry record does not report sensitivity analyses using alternative missing-data assumptions, so the stability of the result under other approaches cannot be assessed. The exact ANCOVA covariates and model specification are also not reported, limiting independent reproduction of the sponsor’s adjusted estimates.

The registry record does not report the original sample-size justification or design assumptions, multiplicity adjustment, or interim analyses. It also does not provide a trial-defined clinically important HbA1c difference or equivalence margin. Accordingly, the planning illustrations based on observed effects cannot be mistaken for the original design rationale, and the confidence intervals cannot be judged against a stated clinical threshold.

The primary outcome is measured at Week 24 and does not, by itself, answer questions about longer-term outcomes. The posted serious adverse event counts are descriptive, and deaths are not reported in the registry record. Further details about adverse event ascertainment and follow-up are not reported in the registry record. These limitations constrain conclusions about durability, comparative safety, and the generalizability of the reported estimates.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Glycemic Efficacy and Renal Safety Study of Dapagliflozin in Subjects With Type 2 Diabetes Mellitus and Moderate Renal Impairment — NCT00663260.
• Registry record last updated 2017-02-10; status last verified 2016-12.
• Results first posted 2017-02-10.
• 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, randomized, double-blind, placebo-controlled, parallel-group Phase 2/3 trial sponsored by AstraZeneca, with enrollment of 631. It studied dapagliflozin 5 mg and 10 mg against placebo in people with type 2 diabetes and moderate renal impairment.

The primary outcome was adjusted mean change in HbA1c at Week 24. Sponsor-reported ANCOVA estimates were -0.08 for 5 mg versus placebo and -0.11 for 10 mg versus placebo; both 95% confidence intervals included zero. The registry also reports secondary FPG and body-weight outcomes and adverse event counts.

Interpretation is limited by the defined primary analysis population, LOCF handling, exclusion of post-rescue data, and statistical details not reported in the registry record, including original design assumptions and multiplicity procedures.