Efficacy and Safety of BI 10773 in Combination With Insulin in Patients With Type 2 Diabetes: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Efficacy and Safety of BI 10773 in Combination With Insulin in Patients With Type 2 Diabetes (NCT01011868), a randomized phase 2 study. The primary outcome is Change From Baseline in Glycosylated Haemoglobin A1c (HbA1c) After 18 Weeks of Treatment.

2. Clinical Question

The clinical question concerns a continuous laboratory measure rather than a binary outcome. The treatment contrasts therefore describe average HbA1c change between groups; they do not describe the response of every participant. The registry also lists longer-term outcomes and response thresholds, but the primary endpoint addressed in the posted results is the 18-week change.

3. Trial Design

The registry describes a Phase IIb, randomized, double-blind, placebo-controlled, parallel-group study. It lists 494 enrolled participants and a planned treatment period of 78 weeks. The trial status is completed. The registry dates are 2009-11 for study start and 2012-05 for primary completion.

Participants were assigned to placebo, empagliflozin 10 mg, or empagliflozin 25 mg, taken once daily alongside basal insulin. The registry describes basal insulin as glargine, detemir, or NPH, with or without concomitant metformin and/or sulfonylurea therapy. Double masking means that treatment assignment was intended to be concealed from participants and relevant study personnel; the registry record does not report further operational details about masking.

The primary outcome was change from baseline in HbA1c after 18 weeks. The registry also lists outcomes at later treatment visits, including HbA1c, fasting plasma glucose, insulin dose, body weight, relative HbA1c response, and attainment of an HbA1c under treatment below 7.0%. Eight secondary outcomes are posted, while the record lists 10 outcome measures and 40 analyses. Those counts indicate that the registry contains multiple outcomes and analyses, but they do not by themselves establish how the study controlled the overall chance of false-positive findings.

The posted primary arm summaries use 125 participants for placebo, 132 for empagliflozin 10 mg, and 117 for empagliflozin 25 mg. These are the endpoint analysis counts, not a replacement for the overall enrollment figure. The registry describes the primary population as FAS18 completers included: patients who did not discontinue prematurely before Week 18, completed the required minimum treatment duration, and had an on-treatment HbA1c value in the Week 18 window. Further detail on allocation procedures, concealment, the original sample-size justification, and the full protocol is not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

After Week 18 Follow-up

Reasons for not completing this period:

After Week 78 Follow-up

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: FAS18-completers-included FAS patients not prematurely discontinue prior to Week 18, completed required minimum treatment duration, and had an on treatment HbA1c value within Week 18 time window. Values after start of antidiabetic rescue therapy were set to missing and last observation carried forward (LOCF-18) was used for imputation. (LOCF-18)

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) - including all randomized and treated patients who had a baseline HbA1c value. Treatment assignment as randomized.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Glycosylated Haemoglobin A1c (HbA1c) After 18 Weeks of Treatment, measured over Baseline and 18 weeks.

The primary endpoint is change from baseline in glycosylated haemoglobin A1c (HbA1c) after 18 weeks of treatment, expressed as percentage of HbA1c. It is a continuous outcome summarized by a mean and standard error for each group. Change from baseline represents the difference between a participant’s HbA1c at the starting assessment and the later assessment, then the group summary describes the average change among participants included in the analysis.

HbA1c is used as a measure of glycaemic control because it reflects glucose exposure over time rather than only the value at a single measurement. As a continuous endpoint, it can capture the direction and magnitude of average change, rather than reducing each participant’s result to a responder/non-responder category. That can make differences between group averages informative, while still not showing how much individual responses vary unless the distribution is also described.

The registry’s primary analysis population excludes patients who discontinued before Week 18 and requires an on-treatment value within the Week 18 window. It also states that values after antidiabetic rescue therapy were set to missing and that last observation was carried forward. The endpoint therefore reflects the stated population and missing-data handling, not necessarily every randomized participant’s outcome. Further details of the analysis procedure are not reported in the registry record.

7. Results Posted in the Registry

• Empagliflozin 10 mg: mean -0.58 versus 0.03 on control; difference -0.61 (95% CI -0.80 to -0.42), standardized effect d = -0.77; z-test p = <0.001.
• Empagliflozin 25 mg: mean -0.75 versus 0.03 on control; difference -0.78 (95% CI -0.99 to -0.57), standardized effect d = -0.95; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

What does a mean difference represent?

A mean difference compares the average endpoint change in one treatment group with the average change in the comparator group. Here, the unadjusted recalculation uses the posted mean change for each group. A negative contrast means that the empagliflozin group’s mean change is lower than placebo’s mean change; in this endpoint, that corresponds to a greater average decrease from baseline. It remains a comparison of group averages, not a measure of how every participant responded.

The recalculated contrasts are -0.61 for empagliflozin 10 mg versus placebo and -0.78 for empagliflozin 25 mg versus placebo. They are simple, unadjusted comparisons derived from the posted summary statistics. The registry’s sponsor analyses instead use ANCOVA and report adjusted mean differences. Adjustment can account for model terms included in the analysis; the specific covariates and model details are not reported in the registry record. The adjusted and unadjusted estimates consequently need not be identical.

What does ANCOVA do?

ANCOVA, or analysis of covariance, compares group outcomes while adjusting for one or more covariates in a statistical model. In clinical trials, adjustment is often used to account for baseline values or other prespecified factors, which can improve precision when those variables explain outcome variation. That is a general description of the method, not a statement about which covariates were included in this trial. The registry identifies ANCOVA as the sponsor’s method and reports adjusted mean differences, but the detailed model specification is not reported in the registry record.

The sponsor’s reported adjusted contrasts are -0.56 for empagliflozin 10 mg and -0.70 for empagliflozin 25 mg, each compared with placebo. The reported confidence intervals are labelled 97.5%, and the p-values are < 0.001. The recalculated p-values, in contrast, are unadjusted and listed as <0.001. Differences in adjustment and analysis specification can produce different estimates and inferential results; the sponsor analysis is the trial’s reported analysis.

How should the standard errors be understood?

A standard error describes the estimated uncertainty in a group mean, rather than the spread of individual participant values. The posted standard errors are 0.07 for placebo, 0.07 for empagliflozin 10 mg, and 0.08 for empagliflozin 25 mg. The recalculation uses these posted standard errors with the group summaries to estimate contrasts and intervals. Standard errors should not be interpreted as ranges containing most individual outcomes; that role is closer to a standard deviation or a distributional summary.

The pooled standard deviations reported for the recalculations are 0.794 for the 10 mg comparison and 0.824 for the 25 mg comparison. A pooled standard deviation combines estimates of within-group variation for a two-group comparison under a common-variance approach. It gives a scale for variability in the outcome, but does not describe the adjusted ANCOVA model. The standardized effect estimates are -0.768 and -0.947, respectively; these express the mean contrasts in relation to the pooled variability and are not additional clinical endpoints.

What does a p-value tell the reader?

A p-value measures how incompatible the observed data are with a specified null hypothesis, assuming the statistical model and its assumptions are appropriate. A small p-value is evidence against that null model, but it is not the probability that the null hypothesis is true. It does not state the probability that a finding is due to chance, quantify clinical importance, or account automatically for testing many outcomes.

The sponsor reports p < 0.001 for each adjusted comparison. The recalculated unadjusted p-values are <0.001 for both. The exact inferential procedures behind the recalculation are not the same as the sponsor’s adjusted ANCOVA analysis. The record does not report enough detail to reconstruct every modeling choice or the trial’s approach to multiplicity.

Why does over-dispersion matter in some analyses?

Over-dispersion is a general issue for count outcomes when observed variation exceeds what a chosen probability model expects. A model that assumes too little variation can produce standard errors that are too small and intervals that are too narrow. The primary HbA1c endpoint is continuous, and the posted primary comparisons use means; over-dispersion is not the central issue for these calculations. It can matter in other types of outcomes, such as counts, but the registry record does not report a count-model analysis for this primary endpoint.

The analysis population and missing-data rule also shape interpretation. The primary population is limited to FAS patients meeting the stated Week 18 completion and measurement criteria. Values after rescue therapy were set to missing, with last observation carried forward described in the record. Last observation carried forward uses an earlier observed measurement in place of a later missing value; as a general method, it can rely on assumptions that may not hold for every participant. Sensitivity analyses and further details of missing-data handling are not reported in the registry record.

10. Confidence Intervals and Precision

A confidence interval gives a range of values compatible with the data and analysis model, reflecting uncertainty around an estimated contrast. A narrower interval indicates greater precision under that model; an interval does not guarantee that the true effect lies within it. Its interpretation depends on the analysis assumptions and on how the interval was constructed.

For the sponsor’s ANCOVA comparison of empagliflozin 10 mg with placebo, the adjusted mean difference is -0.56 and the reported 97.5% confidence interval is -0.78 to -0.33. For empagliflozin 25 mg versus placebo, the adjusted mean difference is -0.70 and the reported 97.5% confidence interval is -0.93 to -0.47. Both intervals lie below zero, so the reported adjusted contrasts are consistent with lower mean change in the empagliflozin group than in placebo under the sponsor’s model.

The unadjusted recalculated estimates are -0.61, with an interval from -0.804 to -0.416, and -0.78, with an interval from -0.988 to -0.572. The confidence level for these recalculated intervals is not reported in the registry record. They are not interchangeable with the sponsor’s 97.5% intervals because they arise from unadjusted calculations rather than the reported ANCOVA model.

Precision depends in part on the amount of information in the analysis, the variability of the outcome, and the analysis method. The primary analysis counts are 125, 132, and 117, while the posted standard errors range from 0.07 to 0.08. These summaries give context for the reported uncertainty but do not resolve questions about representativeness, missing outcomes, or the consequences of testing multiple outcomes. The intervals describe uncertainty about group-average contrasts, not individual participant outcomes.

11. Planning a Similar Trial: Sample Size

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

The planning estimates are participants per arm for the two-group comparisons, not total event counts. They are illustrations based on the observed effects and do not state the original trial’s design assumptions, which are not reported in the registry record. 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 gives the illustrated number of participants per arm for a design that would have an 80% chance of detecting the specified effect under those assumptions if that effect were the true one. The 90% power column gives the corresponding per-arm number for a higher chance of detection under the same assumptions. These are design calculations, not estimates of how likely the completed trial was to detect its actual effect.

For the 10 mg comparison, the estimates are 27 and 36 participants per arm for 80% and 90% power when planning for 100% of the observed effect; 48 and 64 when planning for 75% of the observed effect; and 107 and 143 when planning for 50% of the observed effect. For the 25 mg comparison, the corresponding figures are 18 and 24, then 32 and 42, then 71 and 94.

Smaller effects require more participants to detect under otherwise unchanged assumptions. The planning figures are illustrations only, not observed power, and they do not establish the sample size actually chosen for the trial or its original rationale.

12. How the Sample Size Is Derived

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

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -0.610; pooled standard deviation σ = 0.794.
2. n = 2 × (1.96 + 0.842)² × 0.794² / -0.610² = 26.6, rounded up to 27 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 <- -0.61                      # observed difference in means
sigma <- 0.793801476638665                     # 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(1011868)
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 = 27 (80% power), 36 (90% power)
 75% of observed effect: n per arm = 48 (80% power), 64 (90% power)
 50% of observed effect: n per arm = 107 (80% power), 143 (90% power)
power.t.test: 27.6 
Simulated power at n = 27 per arm: 0.796 (20000 replicates)

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

13. Analyses Reported by the Sponsor

The sponsor’s own primary analysis is reported as ANCOVA, with adjusted mean differences of -0.56 for empagliflozin 10 mg versus placebo and -0.70 for empagliflozin 25 mg versus placebo. The reported 97.5% confidence intervals are -0.78 to -0.33 and -0.93 to -0.47, respectively; both p-values are < 0.001.

The recalculated contrasts use the posted group means and standard errors without adjustment. They are -0.61 for the 10 mg comparison and -0.78 for the 25 mg comparison, with unadjusted p-values of <0.001 for both. The recalculated estimates are not substitutes for the sponsor’s model-based estimates: they use a different analysis approach, and the registry does not report the sponsor model’s full specification.

The record does not flag a discrepancy between the recalculated and sponsor-reported findings. The adjusted and unadjusted estimates point in the same direction, but the sponsor’s analysis remains the trial’s reported analysis. The posted results do not provide enough detail to reproduce the ANCOVA model independently.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists multiple outcomes and analyses: 10 outcome measures, 40 analyses, and eight posted secondary outcomes. Multiple testing creates a general statistical concern because the chance of at least one nominally significant result can increase as more hypotheses are tested. The registry record does not report whether the trial used a prespecified multiplicity procedure, how the primary comparisons were ordered, or whether the reported p-values were adjusted for multiple comparisons. The sponsor’s confidence intervals are labelled 97.5%, but the reason for that confidence level and its relationship to any multiplicity strategy are not reported in the registry record.

The primary analysis population is described as FAS18 completers included. It consists of FAS patients who did not discontinue prematurely before Week 18, completed the required minimum treatment duration, and had an on-treatment HbA1c value in the Week 18 time window. This population definition differs from an analysis that necessarily includes every randomized participant. The registry record does not provide a full account of the numbers excluded from this endpoint population or reasons for missing endpoint measurements.

The record states that values after the start of antidiabetic rescue therapy were set to missing and that last observation was carried forward. This rule affects how outcomes after rescue treatment contribute to the analysis. Last observation carried forward is a general imputation method that carries a previously observed value forward when a later value is missing. Its implications depend on the pattern and reasons for missingness and on whether the earlier measurement remains a suitable representation of the later outcome. Sensitivity analyses, alternative imputation methods, and further details of the missing-data plan are not reported in the registry record.

The registry does not report the original sample-size justification, interim analyses, or the full statistical analysis plan. It also does not report enough information to assess all assumptions behind the sponsor’s ANCOVA model or the handling of every secondary endpoint. These omissions limit independent evaluation of the analysis plan and the extent to which the posted primary finding was protected against selective analysis or multiplicity.

15. 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: thyroid cancer (1.3%), syncope (1.2%), bradycardia (1.2%), coronary artery disease (1.2%), osteoarthritis (1.2%).

The posted serious adverse event counts are 28 in each group: placebo, empagliflozin 10 mg, and empagliflozin 25 mg. The corresponding numbers at risk are 170, 169, and 155. These are counts of participants affected as listed in the registry, paired with the group sizes at risk; they should not be read as counts of distinct event episodes or as rates calculated for comparison. No rate or comparative risk estimate is reported.

For other adverse events, the posted counts are 115 for placebo, 110 for empagliflozin 10 mg, and 95 for empagliflozin 25 mg, with 170, 169, and 155 at risk, respectively. The registry lists thyroid cancer at 1.3%, and syncope, bradycardia, coronary artery disease, osteoarthritis, and road traffic accident at 1.2% each among the top serious adverse event terms. The record does not identify a treatment group for each listed term in the registry record, so those percentages should not be attributed to a particular arm.

Deaths are not reported in the registry record. These safety summaries do not establish whether a treatment caused an event, and the counts alone do not show event severity beyond the serious-event classification or account for exposure time. They should be interpreted as descriptive posted counts rather than as definitive comparative safety estimates.

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

The primary results describe average HbA1c change over the 18-week treatment period among participants meeting the registry’s endpoint-population criteria. The placebo mean change is 0.03, compared with -0.58 for empagliflozin 10 mg and -0.75 for empagliflozin 25 mg. The sponsor-reported adjusted comparisons and the unadjusted recalculations both indicate lower mean change relative to placebo for each empagliflozin dose in this analysis.

These estimates do not mean that every participant’s HbA1c decreased, nor do they show the full distribution of individual changes. A difference between group means is not a prediction for a particular participant. The results also do not establish how the outcome would change beyond the measured period, in populations with different characteristics, or under treatment conditions not described in the registry.

Statistical significance and clinical importance are distinct. The reported p-values indicate evidence against a null comparison under the analyses used, while the estimated differences and intervals describe magnitude and precision of average contrasts. The registry does not provide a stated threshold for interpreting the size of the primary mean difference as clinically important. No conclusion about clinical importance should therefore be inferred solely from the p-values.

The record also lists secondary measures, including later HbA1c assessments, fasting plasma glucose, basal insulin dose, body weight, and categorical HbA1c responses. Numerical results for these outcomes are not reported in the registry record for this analysis. The primary result cannot stand in for those other outcomes, and the posted safety counts do not establish causation or provide a complete assessment of safety.

In short, the posted primary endpoint data show group-average differences at the specified time point, with sponsor-adjusted estimates and unadjusted recalculations of those contrasts. They do not by themselves establish treatment effects for every participant, long-term results, or conclusions beyond the studied population and analysis.

17. Why This Trial Matters Statistically

This trial illustrates why the analysis population is part of the endpoint definition. The primary result is not simply a comparison of all enrolled participants: it is based on FAS patients meeting the Week 18 completion and measurement criteria. That distinction matters when interpreting a mean and its uncertainty, especially when outcomes after rescue therapy are set to missing and last observation is carried forward.

It also shows how adjusted and unadjusted analyses can produce related but non-identical estimates. The sponsor’s ANCOVA comparisons are adjusted, while the recalculations use posted group summaries without adjustment. Readers should distinguish the trial’s reported model-based findings from a simple check based on summary statistics, rather than treating the two as interchangeable.

The study reports a continuous primary outcome with group means, standard errors, adjusted differences, intervals, and p-values. Each summary answers a different question: means describe group-level change, contrasts quantify between-group separation, intervals express uncertainty, and p-values assess compatibility with a null comparison under a model. None alone communicates individual variation or clinical importance.

Finally, the registry contains multiple outcomes and analyses, while its record does not report the multiplicity plan, original sample-size justification, or full statistical analysis plan. This highlights the value of prespecified methods and complete reporting for interpreting a trial’s evidential strength. The reported primary comparisons are informative, but their interpretation remains bounded by the population definition, the available analysis details, and the outcomes actually reported.

18. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The record does not report the original sample-size justification, so the enrollment of 494 cannot be assessed against a stated target effect or assumed variability. The separate planning illustrations based on observed effects do not recover the trial’s original design rationale. Interim analyses, the full protocol, and the complete statistical analysis plan are also not reported in the registry record.

The primary endpoint population is restricted to FAS patients who met the Week 18 completion and measurement criteria. The record does not provide enough detail in the registry record to assess how this population compares with all randomized participants or to determine the full pattern of missing outcomes. The stated handling of values after rescue therapy and last observation carried forward also leaves uncertainty about the sensitivity of the findings to alternative missing-data approaches; sensitivity results are not reported in the registry record.

The sponsor’s ANCOVA model is identified, but its covariates and other specification details are not reported. The unadjusted recalculations cannot replace that model. The record lists 40 analyses and multiple outcomes but does not report a multiplicity procedure. Numerical results for the listed secondary outcomes are not reported in the registry record used for this analysis, limiting assessment beyond the primary endpoint.

Safety information is presented as counts and selected serious-event terms. Deaths are not reported in the registry record, and the registry entries do not provide exposure-adjusted rates, treatment attribution for each listed serious-event term, or comparative estimates with uncertainty. These limitations mean that the posted counts should be read descriptively. They do not support causal attribution or a comprehensive comparison of safety.

These limitations concern the information available in the registry record; they do not establish that the trial did not collect or analyze additional information.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Efficacy and Safety of BI 10773 in Combination With Insulin in Patients With Type 2 Diabetes — NCT01011868.
• Registry record last updated 2014-09-30; status last verified 2014-09.
• Results first posted 2014-09-30.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

NCT01011868 is a completed, randomized, double-blind, parallel-group Phase IIb study of empagliflozin, listed as BI 10773, added to basal insulin in patients with type 2 diabetes and insufficient glycaemic control. The registry lists 494 enrolled participants and a planned treatment period of 78 weeks.

The primary endpoint was change from baseline in HbA1c after 18 weeks. The sponsor reported adjusted ANCOVA mean differences of -0.56 for empagliflozin 10 mg and -0.70 for empagliflozin 25 mg versus placebo, with 97.5% confidence intervals of -0.78 to -0.33 and -0.93 to -0.47, respectively.

The registry also reports serious adverse event counts of 28 in each group. The original sample-size justification, multiplicity approach, full model details, and deaths are not reported in the registry record.