Efficacy and Safety Study With Empagliflozin (BI 10773) vs. Placebo as add-on to Metformin or Metformin Plus Sulfonylurea Over 24 Weeks in Patients With Type 2 Diabetes: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Efficacy and Safety Study With Empagliflozin (BI 10773) vs. Placebo as add-on to Metformin or Metformin Plus Sulfonylurea Over 24 Weeks in Patients With Type 2 Diabetes (NCT01159600), a randomized phase 3 study. The primary outcome is HbA1c Change From Baseline.

2. Clinical Question

These details define a treatment-effect question for a particular population and follow-up period. They do not establish how results would apply to people outside the stated eligibility context or to different treatment durations. The registry record identifies a Phase III, randomized, double-blind, parallel-group study and describes the objective as evaluating efficacy, safety and tolerability. The result should therefore be read as an account of the posted outcomes and analyses, rather than as a general conclusion about all uses of the treatment.

3. Trial Design

The registry record describes a Phase III, randomized, double-blind, placebo-controlled, parallel-group trial. Enrollment was 1504, and the study was completed. The intervention period in the trial title and primary outcome timeframe was 24 weeks. The listed population had type 2 diabetes with insufficient glycaemic control despite metformin alone or metformin combined with a sulfonylurea.

In the randomized groups, participants received BI 10773 at 10 mg or 25 mg once daily, or matching placebo, alongside their background medication. The outcomes are separated according to whether the background treatment was metformin alone or metformin plus sulfonylurea. This separation matters when interpreting the mean changes: the registry explicitly says the randomized-group means were adjusted separately for these two background regimens. The two 25 mg open-label groups are also listed, but their results are descriptive and do not have the same randomized, blinded comparison structure.

The primary outcome description identifies the full analysis set and says treatment assignment was analyzed as randomized. The open-label analysis used an open-label set that included patients entered in the 25 mg open-label treatment arm. The registry record does not report further detail about the randomization procedure, allocation concealment, eligibility criteria, or the full composition of the analysis populations. It also does not report the trial’s original sample-size assumptions or the detailed statistical analysis plan.

Several intervention entries in the registry refer to matching placebos and BI 10773, while the arm descriptions identify the dose and background treatment for the outcome results. For interpretation, the clearest structure is the set of reported arm labels: metformin with placebo or active treatment, and metformin plus sulfonylurea with placebo or active treatment, alongside the separate open-label arms. The open-label comparisons should not be treated as equivalent to randomized treatment effects.

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.

Only the first 6 of 8 groups are shown.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set. Treatment assignment as randomised. Open-label analysis: Open-label set which included all patients entered in the empa 25mg open-label treatment arm. Values after start of antidiabetic rescue therapy were set to missing and last observation carried forward (LOCF) was used for imputation of missing values.

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: Full analysis set (FAS), which included all randomised patients treated with at least one dose of study drug with a baseline HbA1c value. Treatment assignment as randomised. Open label set which included all patients entered in the empa 25mg open-label arm.

Only the first 6 of 9 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as HbA1c Change From Baseline, measured over Baseline and 24 weeks.

The primary endpoint is change from baseline in HbA1c after 24 weeks, reported as a mean and measured in percentage of HbA1c. The registry record gives the timeframe as baseline and 24 weeks. A change score describes how the endpoint value at follow-up differs from its baseline value; a negative mean change in these results represents a decrease from baseline. The endpoint description says randomized-group means were adjusted separately for metformin alone and metformin plus sulfonylurea background medication. The open-label group means were descriptive.

In general, HbA1c is used as a measure of longer-term glycaemic status, so a change from baseline offers a way to compare the average follow-up pattern across treatment groups. A group mean describes the center of the observed or adjusted distribution; it does not show how every participant changed. The standard errors listed with the group means convey uncertainty in estimating those means, rather than the spread of individual participant values.

The primary result should be interpreted through the comparison between randomized active-treatment and matching-placebo groups within each background-medication regimen. The baseline-to-follow-up mean in an active arm alone is not, by itself, the randomized treatment contrast. The registry record also notes that values after antidiabetic rescue therapy were set to missing and that last observation carried forward was used, but the description ends before fully specifying this procedure. Further details about the endpoint’s handling of missing observations are not reported in the registry record.

7. Results Posted in the Registry

• Met: Empa 10mg: mean -0.70 versus -0.13 on control; difference -0.57 (95% CI -0.71 to -0.43), standardized effect d = -0.78; z-test p = <0.001.
• Met: Empa 25mg: mean -0.77 versus -0.13 on control; difference -0.64 (95% CI -0.78 to -0.50), standardized effect d = -0.88; z-test p = <0.001.
• Met: Empa 25mg Open Label: mean -2.78 versus -0.13 on control; difference -2.65 (95% CI -3.07 to -2.23), standardized effect d = -2.48; z-test p = <0.001.
• Met+SU: Placebo: mean -0.17 versus -0.13 on control; difference -0.04 (95% CI -0.18 to 0.10), standardized effect d = -0.05; z-test p = 0.572.

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.

Body Weight Change From Baseline

• 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:
  • Met: Empa 10mg: mean -2.08 versus -0.45 on control; difference -1.63 (95% CI -2.10 to -1.16), standardized effect d = -0.66; z-test p = <0.001.
  • Met: Empa 25mg: mean -2.46 versus -0.45 on control; difference -2.01 (95% CI -2.48 to -1.54), standardized effect d = -0.82; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Met: Empa 10mg 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 Met: Empa 25mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Mean Daily Plasma Glucose (MDG) Change From Baseline

• 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:
  • Met: Empa 10mg: mean -9.64 versus -1.99 on control; difference -7.65 (95% CI -13.03 to -2.27), standardized effect d = -0.33; z-test p = 0.005.
  • Met: Empa 25mg: mean -14.36 versus -1.99 on control; difference -12.37 (95% CI -17.75 to -6.99), standardized effect d = -0.54; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Met: Empa 10mg 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 Met: Empa 25mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted secondary outcomes cover body-weight change from baseline and mean daily plasma glucose change from baseline, each assessed at baseline and 24 weeks. Across these measures, the randomized active-treatment arms generally have more negative posted mean changes than the corresponding placebo arms. The pattern is broadly consistent with the direction of the primary endpoint results, but the secondary endpoints describe different measures and should be interpreted on their own terms. The open-label means are also descriptive rather than randomized treatment contrasts.

The sponsor reports ANCOVA comparisons for selected secondary contrasts, while recalculated estimates are also available for some comparisons. As with the primary endpoint, these analyses are not identical: a recalculation from summary means and standard errors is unadjusted and should not be confused with a sponsor-adjusted result. The registry record flags a discrepancy for the primary outcome; it does not provide a general explanation for differences between sponsor models and recalculations across all outcomes.

In general, multiple outcomes and multiple treatment comparisons raise the chance that some findings will appear statistically notable by chance if each is assessed separately without accounting for the family of tests. The registry record does not report whether multiplicity was controlled, or how. For that reason, secondary findings without a stated multiplicity approach carry less confirmatory weight than a clearly prespecified primary analysis with a documented testing strategy. The registry record identifies the primary endpoint, but details of the broader testing hierarchy are not reported.

The secondary results can therefore be read as complementary descriptions of the trial’s outcome pattern, not as independent proof that the primary finding is larger, smaller or more certain. Their interpretation depends on their own estimates and uncertainty, the analysis methods, and the missing-data approach. They do not establish effects beyond the measured follow-up or resolve questions about differences between background regimens or doses.

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

• Body Weight Change From Baseline
• Mean Daily Plasma Glucose (MDG) Change From Baseline

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous, so the registry reports mean changes and mean differences. A mean summarizes the average change for a group; it does not describe the full distribution or guarantee that each participant had a change close to that average. The arm-level results include a standard error. In general, a standard error describes the precision of an estimated mean, while a standard deviation describes variation among individual observations. The registry record identifies standard error as the dispersion measure for the posted primary means.

The sponsor’s reported method for the listed primary comparisons is ANCOVA, or analysis of covariance. In general, ANCOVA estimates a between-group difference while adjusting for covariates included in the model. The primary outcome description specifically says that randomized-group means were adjusted separately for the two background-medication regimens. However, the full covariate list, model specification and handling of other design features are not reported in the registry record. It would therefore be inappropriate to infer which additional variables were included or how all modeling decisions were made.

The sponsor’s ANCOVA results and the recalculated results have different roles. The recalculations are simple, unadjusted comparisons derived from the posted summary statistics. They contrast the group means using the arm-level standard errors and calculate an interval and p-value under the stated recalculation approach. They do not reproduce the sponsor’s covariate-adjusted model. Because the registry record flags a discrepancy, the sponsor’s own analysis takes precedence for the reported treatment comparisons. The recalculation can help show how the summary means relate arithmetically, but it cannot resolve a difference in analytical methods.

In general, a confidence interval gives a range of estimates compatible with the data and the assumptions of the method at a specified confidence level. A narrow interval indicates greater precision than a wide interval, all else equal. The sponsor reports 97.5% confidence intervals for the primary ANCOVA comparisons. The recalculated intervals should be understood as separate unadjusted estimates, not as replacements for those sponsor intervals. No inference should be made that the two interval types must match exactly.

The p-values in the sponsor’s reported primary analyses are based on ANCOVA. In general, a p-value is calculated relative to a null hypothesis and the model assumptions; it is not the probability that the null hypothesis is true. The registry record does not report the complete testing hierarchy, whether the comparisons were confirmatory or exploratory, or how multiplicity across outcomes and dose comparisons was addressed. A p-value should consequently be read alongside the estimate, interval, outcome definition and reporting context.

Several additional analytic choices affect interpretation. The registry record says treatment assignment was analyzed as randomized for the full analysis set. For the open-label arm, it describes an open-label set that included all patients entered in the 25 mg open-label arm and says the mean is descriptive. Values after antidiabetic rescue therapy were set to missing, and last observation carried forward was used for the incomplete statement that follows. The complete missing-data procedure is not reported in the registry record. Other details, including the handling of withdrawals and missing baseline or follow-up measures, are also not reported.

In general, last observation carried forward replaces a missing later measurement with an earlier observed value. This approach can make assumptions about how outcomes would have evolved after the last observation, and those assumptions may not match actual later changes. Its use should be evaluated in the context of the amount and reasons for missingness, which are not reported in the registry record. The available description also does not establish whether additional sensitivity analyses were conducted.

The planning illustrations use a different calculation from either the sponsor’s ANCOVA or the trial’s original design. They use the observed effect or a stated fraction of it, a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation. The listed figures are participants per arm for these continuous outcomes. In general, a planning calculation estimates a sample size under chosen assumptions; it is not evidence about the operating characteristics of the original trial unless those assumptions match its actual design. The original design assumptions are not reported in the registry record.

In general, over-dispersion is relevant when variability is greater than a model assumes, particularly in count or rate outcomes. The primary endpoint is a continuous mean, and the registry does not report an over-dispersion model. The appropriate statistical issue for this endpoint is whether the chosen continuous-outcome model and its assumptions fit the observed data; those diagnostics are not reported in the registry record. There is likewise no basis in the registry record to infer interim analyses or other unreported statistical procedures.

11. Confidence Intervals and Precision

Precision is conveyed by the uncertainty around an estimated group mean or between-group difference. In general, a confidence interval gives a range of estimates compatible with the data and the analysis assumptions at its stated confidence level. Wider intervals indicate less precision than narrower intervals, all else equal. The sponsor reports 97.5% confidence intervals for the primary ANCOVA comparisons, while the recalculated intervals are separate unadjusted estimates from posted summary statistics.

For the sponsor-reported metformin-only comparisons, the 10 mg estimate is -0.57 with a 97.5% confidence interval from -0.72 to -0.42; the 25 mg estimate is -0.64 with an interval from -0.79 to -0.48. For metformin plus sulfonylurea, the sponsor reports -0.64 with an interval from -0.79 to -0.49 for 10 mg, and -0.59 with an interval from -0.74 to -0.44 for 25 mg. These intervals are all on the negative side of zero. Their width shows that the estimates are not exact; each interval retains a range of plausible values under the sponsor’s model.

The recalculated intervals do not always match the sponsor’s intervals or point estimates. That difference is consistent with the distinction between a summary-statistic recalculation and an adjusted ANCOVA analysis, and the registry record flags a discrepancy. The sponsor’s analysis takes precedence. The recalculated interval should not be used to overrule the sponsor estimate or treated as though it arose from the same model.

In general, a confidence interval is not a guarantee that the true effect lies within its endpoints, nor does a narrow interval alone establish importance. Precision also depends on the analysis population, outcome handling and model assumptions. The registry record does not report all details needed to assess those assumptions, including full missing-data sensitivity analyses. The intervals are therefore useful measures of statistical uncertainty, but they do not remove the limitations of the posted analysis description.

12. Planning a Similar Trial: Sample Size

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

The sample-size illustrations give participants per arm for these continuous outcomes. They are not total-event counts: the registry’s planning note distinguishes these figures from time-to-event calculations, which would report total events. The planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms and a normal approximation. Each planning scenario uses the stated fraction of the observed effect as the effect to detect.

The 80% power column means the illustrated number of participants per arm under those assumptions for an 80% chance of detecting the specified effect if that effect is the true one. The 90% power column gives the corresponding number per arm for a 90% chance under the same assumptions. These are planning probabilities conditional on the calculation’s model and chosen effect; they are not probabilities that the trial result is true. The illustrations are not observed power.

Rows that plan for 75% or 50% of the observed effect use a smaller target effect than the row based on 100% of the observed effect. Smaller effects are harder to distinguish from sampling variability, so the illustrations generally require more participants per arm as the target effect becomes smaller. This relationship is a feature of the planning calculation, not a claim about what effect the trial would actually produce in a future study.

The figures were computed from the observed effects and do not report the trial’s original design assumptions. The original sample-size justification and any design assumptions used by the trial are not reported in the registry record. These illustrations are therefore best read as hypothetical planning aids for the stated contrasts, not as evidence that the trial was designed around these inputs or that its achieved power can be inferred from them.

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 = -0.570; pooled standard deviation σ = 0.728.
2. n = 2 × (1.96 + 0.842)² × 0.728² / -0.570² = 25.6, rounded up to 26 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 <- -0.57                      # observed difference in means
sigma <- 0.72821439736578                     # 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(1159600)
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 = 26 (80% power), 35 (90% power)
 75% of observed effect: n per arm = 46 (80% power), 61 (90% power)
 50% of observed effect: n per arm = 103 (80% power), 138 (90% power)
power.t.test: 26.6 
Simulated power at n = 26 per arm: 0.791 (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 record includes sponsor-reported ANCOVA estimates for the randomized primary comparisons. Those analyses report mean differences, 97.5% confidence intervals and p-values. The recalculated results use posted group means and standard errors to form simple unadjusted comparisons. They are not a reconstruction of the ANCOVA model and do not necessarily use the same adjustments or analysis details.

The discrepancy flag is important because some recalculated estimates differ from the sponsor-reported estimates. For example, the sponsor reports a mean difference of -0.64 for the metformin-plus-sulfonylurea 10 mg comparison, while the recalculation is -0.69. For the corresponding 25 mg comparison, the sponsor reports -0.59 and the recalculation is -0.64. The registry record does not provide enough information to explain these differences fully. The sponsor’s own analysis takes precedence, so its estimates and intervals are the primary source for interpreting the reported randomized comparisons.

Recalculation remains useful as a transparent check on the relationship between posted summary statistics, but its role is limited. A difference between estimates should not be taken as proof of an error in either analysis without the complete model details and underlying data. The appropriate interpretation is to distinguish the adjusted sponsor analysis from the unadjusted recalculation, report the discrepancy, and avoid combining their intervals or p-values as if they came from a single method.

15. Multiplicity, Missing Data and Analysis Populations

The registry record identifies HbA1c change from baseline at 24 weeks as the primary outcome and lists body-weight change and mean daily plasma glucose change as secondary outcomes. It reports four sponsor ANCOVA comparisons for the primary endpoint and selected sponsor comparisons for secondary endpoints. The record does not report a complete testing hierarchy, whether multiplicity adjustment was used, or how the family of outcomes and treatment contrasts was handled. Without those details, the p-values should be interpreted as reported tests rather than as evidence that the overall error rate across all analyses was controlled.

The primary outcome description identifies a full analysis set and says treatment assignment was as randomized. It also describes a separate open-label set that included patients entered in the 25 mg open-label arm. The randomized-group means were adjusted separately for the two background-medication groups, while open-label means were descriptive. The registry record does not report further definitions of these populations, the number of participants contributing to each analysis beyond the arm-level outcome counts, or all reasons for exclusion from an analysis.

For missing outcome data, the primary description says values after antidiabetic rescue therapy were set to missing and last observation carried forward was used. The sentence describing the procedure is incomplete in the registry record, so its full implementation is not reported. The record does not report the amount or pattern of missing data, reasons for other missing observations, or whether sensitivity analyses assessed the impact of the missing-data assumptions.

In general, missing-data methods require assumptions about outcomes that were not observed. Last observation carried forward uses an earlier value in place of a later missing one; it can fail to represent what would have been observed at follow-up if the outcome continued to change. The available registry description is not sufficient to determine how sensitive the estimates are to that assumption. Details of interim analyses, if any, and the trial’s full statistical analysis plan are not reported in the registry record.

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: injury (1.4%), herpes zoster (1.0%), prostate cancer (1.0%), acute myocardial infarction (1.0%), angina pectoris (1.0%).

The registry record posts counts of participants at risk and counts affected by serious adverse events for the listed arms. Among randomized arms, the serious-event counts are 7 of 206 for metformin with placebo, 7 of 217 for metformin with 10 mg, and 5 of 214 for metformin with 25 mg. For metformin plus sulfonylurea, the counts are 14 of 225 for placebo, 11 of 224 for 10 mg and 1 of 217 for 25 mg. Open-label counts are 1 of 69 and 5 of 101 for the metformin and metformin-plus-sulfonylurea groups, respectively. These are descriptive counts, not adjusted comparisons.

The record also lists serious adverse-event terms and percentages, including Injury at 1.4%, and Herpes zoster, Prostate cancer, Acute myocardial infarction, Angina pectoris and Lumbar vertebral fracture at 1.0% each. The registry record does not provide sufficient context in this listing to attribute those terms to a particular comparison or to infer comparative risks from them. Death counts are not reported in the registry record.

In general, safety counts should be interpreted with attention to exposure, follow-up, event definitions and how events were collected. The posted counts alone do not establish whether observed differences are due to treatment, chance or other factors. The record does not report detailed exposure-adjusted analyses or enough information to make a broader safety assessment from these figures.

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

The posted primary results describe average changes in HbA1c over 24 weeks and sponsor-reported adjusted differences between randomized treatment and matching placebo within each background-medication group. The sponsor estimates are negative, their reported 97.5% confidence intervals are also negative, and the reported p-values are < 0.001. These results express the direction and statistical precision of the reported comparisons under the sponsor’s ANCOVA analysis. They do not imply that every participant had the same change or that the average applies to people outside the trial population.

The result does not, by itself, establish whether one dose is superior to the other, whether effects differ by background regimen, or how the findings would persist beyond the 24-week timeframe. Those questions require appropriate direct comparisons and longer-term evidence; such results are not reported in the registry record. Nor should the open-label group means be treated as randomized evidence. They are descriptive results from a separate analysis set and do not have the same blinded placebo-controlled comparison.

Statistical significance is not the same as clinical importance. The registry record does not provide a threshold for interpreting a clinically important change in HbA1c, so the reported estimates should not be labeled clinically meaningful or unimportant on these data alone. Confidence intervals describe uncertainty in the estimated difference under an analysis; they do not capture every source of uncertainty, including incomplete reporting of model details and missing-data handling.

The result also does not settle safety. Serious adverse events are listed as counts, but the available safety information is descriptive and lacks the detail needed to infer comparative risk. Overall, the posted results support a statistical description of the measured outcomes under the study’s stated conditions. They do not provide medical advice or a universal conclusion about treatment.

18. Why This Trial Matters Statistically

This trial illustrates why the analysis method matters when interpreting a reported treatment contrast. The registry presents sponsor ANCOVA estimates alongside group means and standard errors from which unadjusted comparisons can be recalculated. The discrepancy flag makes clear that these approaches do not produce identical results in every comparison. Keeping the adjusted sponsor analysis distinct from the summary-statistic recalculation prevents a numerical check from being mistaken for the trial’s principal model.

The trial also separates participants by background medication. Reporting the randomized comparisons within metformin-only and metformin-plus-sulfonylurea groups preserves that structure, rather than pooling results across clinically distinct regimens without justification. At the same time, separate estimates do not establish a difference between regimens; that requires a direct analysis, which is not reported in the registry record.

Open-label results offer another useful statistical distinction. Descriptive means can add context about observed outcomes, but they do not carry the same causal interpretation as blinded randomized comparisons. Treating them as equivalent would ignore differences in study design and analysis population.

Finally, the record highlights the limits of inference from a registry-level summary. Confidence intervals and p-values help characterize statistical uncertainty, but their interpretation depends on analysis choices, missing-data assumptions and multiplicity handling. Several of those details are not reported. The planning illustrations also demonstrate how sample-size needs depend on the effect targeted and on the assumptions chosen; they are not evidence of the original trial’s design or observed power. These distinctions are central to reading clinical trial results precisely.

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 posted follow-up for the primary endpoint is 24 weeks. The registry record does not report longer-term results for these outcomes, so the results do not describe what happened after that timeframe. The open-label arms have descriptive means and do not provide the same randomized comparison as the placebo-controlled groups. Their results should not be used to infer a randomized treatment effect.

The primary outcome description gives some analysis details but is incomplete at the point where it describes missing observations. It states that values after antidiabetic rescue therapy were set to missing and last observation carried forward was used, while the complete implementation is not reported. The amount and pattern of missing data, reasons for missingness and sensitivity analyses are also not reported in the registry record.

The sponsor reports ANCOVA analyses, but the full model specification and covariates are not reported. The record flags a discrepancy between the sponsor’s analysis and simple unadjusted recalculations from posted summary statistics; it does not explain the difference fully. The sponsor analysis takes precedence, but the incomplete detail limits independent assessment of the model and the cause of the discrepancy.

The record does not report whether multiplicity was controlled across outcomes and comparisons, or provide the full statistical analysis plan, original sample-size justification or details of interim analyses. The safety results are counts, without detailed comparative analyses or death counts. These reporting gaps limit how far the outcome and safety findings can be extended beyond the specific posted analyses.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Efficacy and Safety Study With Empagliflozin (BI 10773) vs. Placebo as add-on to Metformin or Metformin Plus Sulfonylurea Over 24 Weeks in Patients With Type 2 Diabetes — NCT01159600.
• Registry record last updated 2014-06-17; status last verified 2014-05.
• Results first posted 2014-06-17.
• 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

NCT01159600 is a completed Phase III randomized, double-blind, parallel-group study of BI 10773 (empagliflozin) added to metformin alone or metformin plus sulfonylurea. Enrollment was 1504, and the treatment period was 24 weeks. The primary endpoint was change from baseline in HbA1c.

The sponsor reports ANCOVA comparisons of active treatment with matching placebo, separately by background medication. The registry also posts arm-level means, selected secondary outcomes and serious adverse-event counts. Open-label results are descriptive. A discrepancy is flagged between the sponsor’s analyses and unadjusted recalculations; the sponsor’s own analysis takes precedence. Details of multiplicity handling, the complete missing-data procedure and the original sample-size justification are not reported in the registry record.