To Compare the Effect of Liraglutide When Given Together With Metformin With the Effect of Metformin Given Alone and With the Effect of Glimepiride and Metformin Given Together: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for To Compare the Effect of Liraglutide When Given Together With Metformin With the Effect of Metformin Given Alone and With the Effect of Glimepiride and Metformin Given Together (NCT00318461), a randomized phase 3 study. The primary outcome is Change in Glycosylated A1c (HbA1c) at Week 26.

2. Clinical Question

The clinical question is about comparative change in a continuous measure of glycaemic control, rather than a count of events or a time-to-event outcome. The randomized, double-blind period lasted 26 weeks and was followed by an open-label extension. The primary endpoint is tied to the randomized period, so it does not by itself answer whether between-group patterns persisted through the full extension.

The trial record identifies the study as conducted across Europe, Oceania, Africa, Asia, and South America. It does not report a detailed participant eligibility description beyond the diabetes condition in the information summarized in the registry record. Likewise, the registry record does not report a stated clinically important difference for HbA1c. Statistical differences should therefore be read as estimates of group contrasts, not as a direct measure of clinical importance.

3. Trial Design

This was a phase 3, randomized, double-blind, parallel-group trial with enrollment of 1091. Its primary randomized period lasted 26 weeks, followed by an 18 months open-label extension; the brief summary describes the full trial duration as 2 years (104 weeks). The study is listed as completed, with the primary completion recorded as 2007-05.

Five groups appear in the arm descriptions: three liraglutide-plus-metformin dose groups, metformin monotherapy, and metformin plus glimepiride. Metformin was listed at 1.5-2.0 g/day in each arm using it, while the listed glimepiride dose in the combination group was 4 mg/day. Placebos are specified in the descriptions for the masked treatment components. These details describe the assigned regimens; the registry record does not provide a fuller account of dose titration, adherence, or treatment changes.

The primary outcome was the change in HbA1c from week 0 to week 26, reported as a least-squares mean with standard error. The analysis population was the intention-to-treat set defined in the registry record as all randomized participants exposed to at least one dose of study products, using last observation carried forward. Arm-level primary endpoint data are available for 239.0, 232.0, 236.0, 120.0, and 234.0 participants across the listed groups. The record's adverse-event denominators differ slightly from these outcome-analysis counts.

The randomized, double-blind comparison is the most direct setting for comparing assigned groups, while the later open-label extension has a different masking context. The registry record reports outcomes at week 26 and week 104 for several secondary measures, but does not report a complete account of follow-up, withdrawals, or how treatment exposure evolved during the extension. The original trial's detailed statistical plan and design assumptions are not reported in the registry record.

4. Participant Flow and Accounting

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

Before assignment: Eligible subjects discontinued their oral anti-diabetic drug treatment and commenced a 3-week period of forced titration of metformin followed by a 3-week maintenance period. Subjects on current metformin therapy could go through a modified titration period or advance directly to the 3-week maintenance period at the discretion of the investigator.

Double-Blind, 6 Months

Reasons for not completing this period:

Open-Label Extension, 18 Months

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intention to treat (ITT) analysis set using LOCF (Last Observation Carried Forward) is all randomised subjects who had been exposed to at least one dose of the study products.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Change in Glycosylated A1c (HbA1c) at Week 26, measured over week 0, week 26.

The primary endpoint measures change in glycosylated haemoglobin, HbA1c, from baseline at week 0 to week 26. It is reported in percentage points of total HbA1c, with lower values representing a decrease from the participant's baseline. The registry record presents least-squares mean changes for each group and standard errors describing uncertainty around those arm estimates.

HbA1c is used in clinical trials as a continuous measure related to glycaemic control over time. As a continuous outcome, it can capture graded changes across participants rather than reducing each person's result to a binary category. Comparing group-level changes can help describe how the assigned regimens differ on this measure over the randomized period.

The registry record specifies that the primary endpoint was measured from week 0 through week 26, the end of randomization. Its analysis set included randomized participants who received at least one dose, with last observation carried forward used for missing values. That approach carries a previously observed value forward when a later measurement is unavailable; it does not make the missing measurement known. The missing-data assumptions and any sensitivity analyses are not reported in the registry record.

HbA1c change does not represent every aspect of diabetes care or every possible benefit and harm. It also does not, by itself, establish how an observed difference should be judged clinically. The endpoint is one outcome in this trial, interpreted alongside the other posted measures, the analysis method, and the uncertainty in the estimates.

7. Results Posted in the Registry

• Lira 1.2 + Met: mean -0.97 versus -0.69 on control; difference -0.28 (95% CI -0.47 to -0.09), standardized effect d = -0.26; z-test p = 0.005.
• Lira 1.8 + Met: mean -1.00 versus -0.69 on control; difference -0.31 (95% CI -0.50 to -0.12), standardized effect d = -0.29; z-test p = 0.002.
• Met Mono: mean 0.09 versus -0.69 on control; difference 0.78 (95% CI 0.56 to 1.00), standardized effect d = 0.74; z-test p = <0.001.
• Met + Glim: mean -0.98 versus -0.69 on control; difference -0.29 (95% CI -0.48 to -0.10), standardized effect d = -0.27; z-test p = 0.003.

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.

Change in Body Weight at Week 26

• 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:
  • Lira 1.2 + Met: mean -2.58 versus -1.78 on control; difference -0.80 (95% CI -1.45 to -0.15), standardized effect d = -0.22; z-test p = 0.016.
  • Lira 1.8 + Met: mean -2.79 versus -1.78 on control; difference -1.01 (95% CI -1.65 to -0.37), standardized effect d = -0.28; z-test p = 0.002.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Lira 1.2 + Met 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 Lira 1.8 + Met than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change in Body Weight at Week 104

• 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:
  • Lira 1.2 + Met: mean -3.03 versus -2.07 on control; difference -0.96 (95% CI -1.75 to -0.17), standardized effect d = -0.22; z-test p = 0.017.
  • Lira 1.8 + Met: mean -2.91 versus -2.07 on control; difference -0.84 (95% CI -1.62 to -0.06), standardized effect d = -0.19; z-test p = 0.034.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Lira 1.2 + Met 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 Lira 1.8 + Met than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change in Fasting Plasma Glucose (FPG) at Week 26

• 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:
  • Lira 1.2 + Met: mean -1.63 versus -1.13 on control; difference -0.50 (95% CI -0.93 to -0.07), standardized effect d = -0.21; z-test p = 0.023.
  • Lira 1.8 + Met: mean -1.68 versus -1.13 on control; difference -0.55 (95% CI -0.97 to -0.13), standardized effect d = -0.24; z-test p = 0.010.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Lira 1.2 + Met 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 Lira 1.8 + Met than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The posted secondary measures cover body weight, fasting plasma glucose, and glucose-profile outcomes at week 26 and week 104. For the listed week-26 outcomes, the liraglutide-plus-metformin groups generally have negative arm-level changes in weight and fasting plasma glucose, while metformin plus glimepiride has a positive weight change in the posted summary. At week 104, the displayed weight changes remain negative for the liraglutide combinations and metformin monotherapy, while the metformin-plus-glimepiride estimate is positive. These broad patterns describe the arm estimates; they do not establish the precision or statistical status of every secondary comparison.

The pattern is directionally consistent with the primary outcome in that the displayed liraglutide-plus-metformin groups have negative mean changes across several reported measures. However, different outcomes have different units and meanings, and the secondary endpoints cannot be treated as independent confirmations of the primary result. The open-label extension also differs from the randomized, double-blind period, so week-104 findings have a different context from the primary week-26 endpoint.

The registry record lists 14 outcome measures and 74 analyses, while the posted secondary outcome summaries include a range of endpoints and time points. The full multiplicity plan is not reported in the registry record. When many outcomes and comparisons are examined, some apparently notable results can arise by chance even when there is no corresponding underlying effect. Unadjusted secondary results therefore carry less weight than a prespecified, multiplicity-controlled primary analysis.

Secondary results are best read as descriptive evidence about the broader outcome pattern, with attention to the particular analysis and its uncertainty. The registry record does not report sponsor-reported comparison statistics for every secondary measure, so arm-level changes should not be mistaken for formal adjusted contrasts. Nor should the secondary outcomes be combined into a single overall treatment conclusion.

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

• Change in Body Weight at Week 26
• Change in Body Weight at Week 104
• Change in Fasting Plasma Glucose (FPG) at Week 26
• Change in Fasting Plasma Glucose (FPG) at Week 104
• Change in Mean Prandial Increments of Plasma Glucose Based on Self-measured 7-point Plasma Glucose Profiles at Week 26
• Change in Mean Prandial Increments of Plasma Glucose Based on Self-measured 7-point Plasma Glucose Profiles at Week 104
• Change in Mean Post Prandial Plasma Glucose Based on Self-measured 7-point Plasma Glucose Profiles at Week 26
• Change in Mean Post Prandial Plasma Glucose Based on Self-measured 7-point Plasma Glucose Profiles at Week 104

9. Statistical Interpretation

10. Statistical Methods Explained

Several statistical features affect how the results should be read. The primary outcome is continuous, and the registry reports least-squares means, standard errors, between-group contrasts, confidence intervals, and p-values. These quantities answer related but distinct questions: arm estimates describe average change in each group, while a contrast estimates the difference between groups. A confidence interval describes the range of values reasonably compatible with a particular estimate under its model and assumptions. A p-value describes compatibility with a specified null hypothesis, not the probability that the null hypothesis is true.

What is a least-squares mean?

General explanation: a least-squares mean is a model-based group mean, often adjusted for variables included in the analysis model. The registry record identifies the primary arm estimates as least-squares means and identifies the sponsor's reported comparisons as ANCOVA estimates. It does not list the covariates used in ANCOVA, the model specification, or the model diagnostics. The reported adjusted estimates therefore cannot be fully reconstructed from the record's arm means alone.

The arm-level values are accompanied by standard errors. A standard error reflects uncertainty in an estimated mean; it is not the spread of individual participant values. The simple recalculations use arm means and approximate standard deviations inferred from those standard errors, as indicated in their calculation notes. They are therefore not substitutes for the sponsor's ANCOVA model or its adjusted estimates.

What is ANCOVA?

General explanation: analysis of covariance models an outcome while accounting for one or more covariates. In a randomized trial, adjustment can improve precision or account for baseline outcome values when those variables are included appropriately. The registry record reports ANCOVA for the sponsor's contrasts, but does not report the full covariate set, exact model formula, handling of model assumptions, or any adjustment factors. Those details are not reported in the registry record.

Why distinguish recalculated and sponsor-reported results?

The simple unadjusted re-calculations use posted arm summary statistics to form pairwise comparisons. The sponsor's own analysis uses ANCOVA and reports estimated treatment differences with confidence intervals and p-values. Since the registry record flags a discrepancy, the sponsor's analysis takes precedence. The calculations can still help illustrate the difference between a direct summary-statistic contrast and a model-based contrast, but they should not be presented as the trial's formal analysis.

The reported primary endpoint population used an intention-to-treat label, but the record's definition is specific: randomized participants exposed to at least one dose. This is not necessarily the same as every randomized participant regardless of treatment exposure. Last observation carried forward (LOCF) was used for missing primary endpoint data. General explanation: LOCF fills an unavailable later measurement with the participant's last observed value. It can be problematic if the outcome would have changed after that observation or if missingness is related to outcomes. The registry record does not report sensitivity analyses that assess alternative missing-data assumptions.

What do the p-values and test labels mean?

The registry reports ANCOVA estimates and 95% confidence intervals, and labels the analyses as non-inferiority or equivalence. General explanation: non-inferiority asks whether an intervention is not unacceptably worse than a comparator according to a prespecified margin; equivalence asks whether the difference lies within prespecified bounds. The margin, direction of testing, hypotheses, and multiplicity procedure are not reported in the registry record. As a result, the test labels and p-values should be quoted as reported, without inferring the precise decision rule or treating every p-value as a conventional test of no difference.

The simple recalculations are explicitly unadjusted. Their p-values do not account for the multiple comparisons and outcomes represented in the record, unless a method for doing so is separately reported. A small p-value is not a measure of effect size or clinical importance. The size and direction of an estimate, its interval, and the analysis that produced it all matter.

Why can over-dispersion matter?

General explanation: over-dispersion occurs when the observed variability is greater than a model assumes. It is especially relevant to some count or binary data models; for continuous outcomes, the broader point is that uncertainty estimates depend on a suitable model for variability. The primary endpoint is continuous and reported with standard errors. The registry record does not report an over-dispersion analysis, so none should be inferred for this outcome.

The record includes 74 analyses across 14 outcome measures. This breadth increases the importance of knowing whether testing was prioritized or adjusted for multiple comparisons. A multiplicity plan is not reported in the registry record. Secondary endpoints and exploratory contrasts therefore need cautious interpretation, particularly when their p-values are unadjusted.

11. Confidence Intervals and Precision

The confidence intervals help show how precisely the group contrasts are estimated. For the simple recalculation comparing liraglutide 1.2 mg/day plus metformin with the 0.6 mg/day combination group, the mean difference was -0.28 and the 95% confidence interval was -0.474 to -0.086. For the 1.8 mg/day comparison with that reference group, the estimated difference was -0.31, with a 95% confidence interval from -0.504 to -0.116. These intervals are based on unadjusted re-calculations and should not be mistaken for the sponsor's ANCOVA intervals.

For the sponsor-reported comparison between liraglutide 1.8 mg/day plus metformin and metformin alone, the estimated treatment difference was -1.09, with a 95% confidence interval from -1.30 to -0.88. For the comparison with metformin plus glimepiride, the estimate was -0.02, with a 95% confidence interval from -0.19 to 0.15. The intervals express uncertainty in the estimated contrasts under the reported ANCOVA analysis. They are not ranges containing a fixed percentage of individual participant changes.

General explanation: a confidence interval is narrower when the estimate is more precise, and wider when it is less precise. Its interpretation depends on the model and analysis population used. It does not account automatically for missing-data uncertainty, model misspecification, unreported design choices, or multiple testing. Nor does an interval alone define what difference would matter clinically; the registry record does not report a clinical importance threshold for the primary outcome.

The sponsor-reported intervals and simple recalculation intervals differ because the analyses are not identical. The discrepancy is explicitly flagged in the registry record, so the sponsor's analysis takes precedence. Precision should be judged within each analysis rather than by treating all intervals as interchangeable. The arm-level standard errors are also not confidence intervals; they quantify uncertainty in each arm's least-squares mean rather than directly describing a between-group contrast.

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 figures are planning illustrations based on the observed effect, not the trial's original design calculation. For these continuous-outcome comparisons, each figure is the number of participants per arm. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect equal to the listed fraction of the observed effect. The original trial's own sample-size assumptions and justification are not reported in the registry record.

The 80% power column gives the planned number of participants per arm under those assumptions for an 80% chance of detecting the specified effect if it is truly present. The 90% power column gives the corresponding number per arm for a 90% chance. These are prospective planning concepts; neither column reports the power actually achieved in the trial. This is not observed power.

Rows planning for 100% of the observed effect use the full estimated effect as the target. Rows planning for 75% or 50% use smaller target effects. Smaller effects are harder to distinguish from random variation, so they require more participants under the same assumptions. The displayed planning counts can therefore rise as the target fraction becomes smaller. This is a general feature of sample-size planning, not evidence that the original trial was designed around these recalculations.

The planning illustrations are based on unadjusted summary-statistic comparisons. They do not incorporate the sponsor's ANCOVA model, a multiplicity adjustment, attrition, the LOCF approach, or any original design details not reported in the registry record. Their purpose is to show how hypothetical group sizes vary with the assumed effect and power target, not to recommend a sample size for another study.

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 = -0.280; pooled standard deviation σ = 1.074.
2. n = 2 × (1.96 + 0.842)² × 1.074² / -0.280² = 231.1, rounded up to 232 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 <- -0.28                      # observed difference in means
sigma <- 1.07433978337826                     # 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(318461)
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 = 232 (80% power), 310 (90% power)
 75% of observed effect: n per arm = 411 (80% power), 551 (90% power)
 50% of observed effect: n per arm = 925 (80% power), 1,238 (90% power)
power.t.test: 232.1 
Simulated power at n = 232 per arm: 0.802 (20000 replicates)

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

14. Analyses Reported by the Sponsor

The registry record includes sponsor-reported ANCOVA estimates for pairwise comparisons and simple unadjusted recalculations based on posted arm-level summary statistics. The recalculations are not identical to the sponsor's analysis: they use summary means and standard errors, with an approximate standard deviation, and do not reproduce the ANCOVA adjustment.

The discrepancy flag means the sponsor's own analysis takes precedence when describing the trial's formal results. For example, the sponsor-reported estimate comparing liraglutide 1.8 mg/day plus metformin with metformin alone is -1.09, with a 95% confidence interval from -1.30 to -0.88. The recalculated estimate for a different contrast, 1.8 mg/day versus 0.6 mg/day liraglutide with metformin, is -0.31, with a 95% confidence interval from -0.504 to -0.116. Those contrasts differ in comparator as well as method and should not be set side by side as if they were the same question.

Comparisons with the same nominal groups may also differ between the recalculation and sponsor result because the latter is model-based and may account for covariates not available in the summary statistics. The registry record does not report all ANCOVA details, so the exact source of each numerical difference cannot be determined from the posted information. The recalculated results are useful as simple checks and illustrations, but they do not replace the sponsor's adjusted analyses.

15. Multiplicity, Missing Data and Analysis Populations

The registry record contains multiple outcomes, time points, groups, and analyses. It lists 14 outcome measures and 74 analyses. Testing many comparisons can increase the chance of obtaining small p-values by chance, even if each individual test uses a conventional threshold. Statistical methods address this issue through prespecified testing sequences, adjusted significance thresholds, or other procedures, but the multiplicity plan for this trial is not reported in the registry record.

The record identifies the primary endpoint as HbA1c change at week 26, but does not provide a complete hierarchy of hypotheses or explain how the primary comparisons were prioritized. It also does not report whether the secondary endpoint analyses were confirmatory or exploratory, or whether their p-values were adjusted. The unadjusted p-values attached to the summary-statistic recalculations should thus be considered descriptive rather than interpreted as multiplicity-controlled evidence.

The primary analysis used an intention-to-treat analysis set defined as randomized participants exposed to at least one dose of study products. Missing measurements were handled using last observation carried forward. The registry record does not report the amount or pattern of missing primary endpoint data, reasons for missingness, or sensitivity analyses using alternative assumptions. LOCF can make the analysis depend on the assumption that a participant's outcome would remain at the last observed value after measurement stopped; that assumption cannot be checked from the posted summary alone.

The record does not report whether additional analysis populations were used, such as a per-protocol population, or how conclusions were compared across populations. It also does not report the full covariate specification for ANCOVA, the non-inferiority or equivalence margin, interim analyses, or a detailed statistical analysis plan. These are limits on interpretation, not evidence that a particular procedure was or was not carried out.

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: ankle fracture (1.7%), angina pectoris (0.8%), musculoskeletal pain (0.8%), acute myocardial infarction (0.8%), myocardial infarction (0.8%).

The adverse-event summary reports serious adverse events affecting 36 of 242 participants in the liraglutide 0.6 mg/day plus metformin group, 25 of 240 in the 1.2 mg/day group, 16 of 242 in the 1.8 mg/day group, 9 of 121 receiving metformin alone, and 24 of 242 receiving metformin plus glimepiride. These are participant counts with the corresponding at-risk denominators, not event rates calculated for comparison in this discussion.

The record also reports other adverse events affecting 132 of 242, 144 of 240, 158 of 242, 44 of 121, and 128 of 242 participants in those same groups, respectively. The registry record lists serious adverse-event terms including ankle fracture at 1.7%, angina pectoris at 0.8%, musculoskeletal pain at 0.8%, acute myocardial infarction at 0.8%, myocardial infarction at 0.8%, and atrial fibrillation at 0.8%. The record does not specify the denominator or provide enough detail in this summary to interpret those percentages as comparative risks.

Deaths are not reported in the registry record. These safety summaries should be read as descriptive counts and listed terms. The posted information does not establish causal attribution, comparative risk, severity distribution, or the completeness of event follow-up, and it should not be used alone to draw a safety conclusion.

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

The posted results describe average changes in HbA1c and other continuous outcomes among trial groups. They show that the group estimates were not identical, and the sponsor's ANCOVA analyses report differences for several comparisons. Confidence intervals indicate uncertainty around those estimated contrasts, while the standard errors describe uncertainty around arm-level estimates.

These statistics do not, by themselves, establish the clinical importance of a difference. The registry record does not report a threshold for a clinically meaningful HbA1c change or a full account of how the formal non-inferiority or equivalence conclusions were defined. Statistical significance is not the same as clinical importance, and a confidence interval is not a guarantee that the true value lies within its endpoints.

The evidence also has boundaries. The primary endpoint concerns change through week 26 during the randomized period. Secondary results at week 104 come from a later open-label extension, which has a different design context. The record's LOCF approach for missing primary measurements and the absence of reported sensitivity analyses leave uncertainty about how missing data could affect the estimates. The sponsor's ANCOVA specification and multiplicity strategy are not fully described.

The comparisons concern the regimens and participants in this trial, not every possible dose, population, or treatment context. The registry record describes a broad multinational setting, but does not provide enough detail to establish how representative the enrolled participants were of any particular clinical population. The results should therefore be understood as trial-specific estimates, with the stated analysis methods and limitations, rather than as a complete account of comparative benefits and harms.

No single statistic resolves all these questions. The arm means, adjusted contrasts, confidence intervals, secondary outcomes, and adverse-event counts each convey different information. A careful reading keeps those measures distinct and avoids turning a statistical result into medical advice.

18. Why This Trial Matters Statistically

This trial illustrates why the analysis behind a reported treatment difference matters. A direct calculation from arm-level summaries and an ANCOVA model-based estimate may not match, even when they concern similar groups. The registry record flags such a discrepancy, making it important to distinguish exploratory recalculations from the sponsor's reported analysis and to give precedence to the latter.

It also shows how an outcome can be presented at multiple levels. Arm-level least-squares means describe the average change in each group, while a treatment contrast focuses on a difference between groups. Standard errors and confidence intervals communicate different kinds of uncertainty, and neither should be confused with the variation among individual participants.

The trial includes a primary endpoint, multiple secondary outcomes, several treatment groups, and more than one time point. This makes multiplicity a practical concern: without a reported analysis hierarchy or adjustment plan, secondary p-values should not be read as if each were an isolated confirmatory test. Likewise, an unreported non-inferiority margin leaves important context missing for understanding the stated analysis labels.

Finally, the study distinguishes a randomized, double-blind period from an open-label extension. That distinction helps readers avoid treating longer-term observations as if they came from an unchanged design. The registry record is informative about reported estimates, but it also demonstrates why transparent reporting of model specifications, missing-data methods, and testing plans is essential for evaluating precision and strength of 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.

Several trial-specific limitations affect interpretation. The record identifies a randomized, double-blind period followed by an open-label extension, but does not provide detailed information about treatment exposure, adherence, changes in therapy, or retention during the extension. Results at week 104 therefore should not be assumed to have the same interpretive basis as the primary endpoint at week 26.

The primary analysis used last observation carried forward for missing measurements. The record does not report the extent or pattern of missing data or sensitivity analyses using other assumptions. This makes it difficult to judge how dependent the primary estimates are on the imputation approach.

The sponsor-reported analyses are ANCOVA estimates, but the registry record does not describe the model's covariates, assumptions, or diagnostics. It labels the comparisons as non-inferiority or equivalence but does not report the margin or full testing framework. The multiplicity strategy for the numerous analyses is also not reported.

The registry record's participant counts vary across posted outcome and safety summaries. It does not explain those differences in the information. The broad geographic coverage is described, but detailed eligibility criteria and participant characteristics are not reported in this summary. This limits assessment of applicability to particular groups.

Finally, posted arm-level summaries cannot reproduce a fully adjusted model or resolve the flagged discrepancy between simple recalculations and sponsor analyses. The recalculations are useful only as unadjusted illustrations. The sponsor's own analysis takes precedence for the formal comparisons, while unreported methods and missing-data details remain sources of uncertainty.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: To Compare the Effect of Liraglutide When Given Together With Metformin With the Effect of Metformin Given Alone and With the Effect of Glimepiride and Metformin Given Together — NCT00318461.
• Registry record last updated 2017-03-07; status last verified 2017-01.
• Results first posted 2010-03-12.
• 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 phase 3 randomized, double-blind, parallel-group trial comparing three doses of liraglutide added to metformin with metformin alone and metformin plus glimepiride. The primary endpoint was change in HbA1c from baseline to week 26, analyzed in an intention-to-treat set defined as randomized participants exposed to at least one dose, using last observation carried forward for missing measurements.

Arm-level least-squares mean changes and sponsor-reported ANCOVA comparisons are posted, along with secondary outcomes and adverse-event summaries. A discrepancy is flagged between simple unadjusted recalculations and the sponsor's analysis; the sponsor's analysis takes precedence. The record does not report the full multiplicity plan, ANCOVA specification, non-inferiority or equivalence margin, or missing-data sensitivity analyses.

The trial's randomized period was followed by an open-label extension. Primary and secondary findings should be interpreted in light of that design distinction, the analysis methods, and the uncertainty in the posted estimates.