A Trial Comparing the Efficacy and Safety of Liraglutide 1.8 mg/Day to Liraglutide 0.9 mg/Day in Japanese Subjects With Type 2 Diabetes Mellitus.: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Trial Comparing the Efficacy and Safety of Liraglutide 1.8 mg/Day to Liraglutide 0.9 mg/Day in Japanese Subjects With Type 2 Diabetes Mellitus. (NCT02505334), a randomized phase 3 study. The primary outcome is Change in Glycosylated Haemoglobin (HbA1c) (Week 26).

2. Clinical Question

The comparison concerns two liraglutide doses rather than liraglutide versus no treatment or a different medicine. The primary outcome is a continuous measure of change in HbA1c. The registry also lists outcomes concerning HbA1c response thresholds, self-measured blood glucose profiles and other measures, but the primary statistical interpretation centers on the Week 26 HbA1c endpoint. The posted record does not report a separate primary question about long-term outcomes beyond the primary endpoint timeframe.

3. Trial Design

This was a completed Phase 3, randomized, parallel-group trial with no masking. Participants were allocated to liraglutide 1.8 mg/day or liraglutide 0.9 mg/day. The registry reports enrollment of 635. The primary endpoint analysis lists 233 participants in each arm; the reason that the analyzed arm counts differ from overall enrollment is not reported in the registry record.

The two groups had different stated overall treatment-period durations. Both included a 2-week screening period, a 12-week run-in period and a 26-week treatment period. The 1.8 mg/day group also had a 26-week safety extension, producing a stated approximate total trial duration of 67 weeks; the 0.9 mg/day group had a stated approximate total duration of 41 weeks and a follow-up visit. These duration differences matter when reading safety information: the registry’s description indicates that the higher-dose arm had a longer planned observation period. The record does not provide enough information to infer how this difference affected the number or timing of adverse events.

The primary endpoint was evaluated after 26 weeks. The analysis population is described as the full analysis set, comprising all randomized subjects, and missing endpoint values were imputed using last observation carried forward. The primary analysis used ANCOVA, with treatment as a fixed effect and baseline response as a covariate. The registry lists 61 outcome measures and 60 posted secondary outcomes. The number of posted analyses is one. Further detail about the randomization procedure, allocation concealment, original sample-size justification, or protocol-level analysis plan 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.

Before assignment: Subjects received liraglutide 0.9 mg/day (starting with 0.3 mg/day with subsequent weekly dose escalation of 0.3 mg to a maximum 0.9 mg/day) during 12 weeks run-in period. Subjects with HbA1c ≥7.0% at the end of run-in period were randomised (1:1) to two treatment arms: continuing liraglutide 0.9 mg/day or dose escalation to liraglutide 1.8 mg/day.

Main Treatment Period

Reasons for not completing this period:

Extension Treatment Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: FAS, which included all randomised subjects. Missing values were imputed using the last observation carried forward (LOCF) method.

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 subjects.

6. Primary Endpoint

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

The primary endpoint is change in glycosylated haemoglobin, or HbA1c, from Week 0 to Week 26, reported as a percentage of HbA1c. HbA1c reflects glucose exposure over time and is commonly used in trials of type 2 diabetes as a continuous measure of glycemic status. A change score compares the measured value at follow-up with the participant’s baseline value; the registry’s analysis then adjusts the treatment comparison for baseline response.

Using a continuous outcome preserves information across the measured scale rather than reducing each participant to a responder or non-responder category. This can support estimation of an average difference between randomized groups. The primary analysis reports least-squares means for each arm and a treatment difference from ANCOVA. A least-squares mean is a model-adjusted estimate, rather than simply the arithmetic average of the posted arm values. The registry specifies Week 0 and Week 26 as the time points and reports the endpoint as a Week 26 change from baseline.

Interpretation remains tied to this endpoint and timeframe. It does not by itself describe every aspect of diabetes, sustained outcomes after the primary assessment, or the full safety profile. The registry’s other listed outcomes include threshold-based HbA1c response measures and self-measured blood glucose profiles, but those are distinct outcomes rather than substitutes for the primary continuous endpoint.

7. Results Posted in the Registry

• Liraglutide 0.9 mg: mean 0.17 versus -0.23 on control; difference 0.40 (95% CI 0.23 to 0.57), standardized effect d = 0.44; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary analysis used analysis of covariance, abbreviated ANCOVA. The registry describes treatment as a fixed effect and baseline response as a covariate. In general, ANCOVA estimates a treatment-group contrast while accounting for a baseline measurement related to the outcome. Including baseline can make the comparison more informative than a simple comparison of follow-up values, provided the model is appropriate and the analysis is applied as specified. The registry does not report further model details, such as additional covariates or diagnostics.

What does the adjusted contrast represent?

The reported treatment difference is -0.40, with a 95% confidence interval from -0.55 to -0.24 and p < 0.001. The sign depends on the order in which treatment groups are subtracted. In the registry’s reported contrast, the negative estimate corresponds to lower change in the 1.8 mg arm relative to the 0.9 mg arm. The posted arm least-squares means are -0.23 and 0.17, respectively. Those are adjusted model summaries, and their difference is represented by the registry’s treatment contrast. Because the contrast is adjusted, it need not be reconstructed exactly by subtracting rounded arm summaries.

A p-value is a measure of how compatible the observed data are with a specified null hypothesis under the model’s assumptions. It is not the probability that the null hypothesis is true, and it does not measure the size or importance of an effect. The registry reports p < 0.001 for the ANCOVA comparison. The recalculated unadjusted p-value is <0.001. These values come from different analyses and should not be treated as duplicate outputs from one model.

Why distinguish the recalculation from the registry analysis?

The recalculation uses the posted arm means, sample sizes and standard errors to estimate a simple unadjusted contrast. It reports 0.4 for 0.9 mg minus 1.8 mg, with an approximate 95% confidence interval from 0.234 to 0.566. The pooled standard deviation used in that calculation is approximate, as indicated in the registry-derived analysis information. It is a useful check on the scale and direction of the arm summaries, but it does not reproduce the baseline-adjusted ANCOVA. The two estimates also use reverse contrast directions, so one is positive and the other negative.

In general, an adjusted analysis can differ from an unadjusted analysis because it accounts for baseline values and may use model-specific assumptions. Posted summary values are also rounded. The record reports no discrepancy flag, but that does not make the two methods interchangeable. The sponsor’s ANCOVA is the registry’s own reported analysis and is the appropriate result to cite as such.

How should the standard errors and confidence intervals be read?

The standard error describes the estimated sampling uncertainty of a statistic, such as an arm mean. It is not the standard deviation of individual participant outcomes. Each posted arm least-squares mean has a standard error of 0.06. The registry’s interval for the treatment contrast gives a range of values compatible with the data and analysis at the stated 95% confidence level. In general, a confidence interval reflects both the estimate and its precision; a narrower interval indicates less statistical uncertainty about the estimate under the model, not less variation among individuals.

The arm-level interval limits are not reported in the registry record. The sponsor’s contrast interval is reported directly, while the recalculated interval is approximate and based on the posted summary statistics. The distinction matters because the recalculated uncertainty does not incorporate the same baseline adjustment as the ANCOVA.

How do allocation and masking affect interpretation?

Random allocation supports comparison between groups by assigning participants to treatment groups by chance, although the specific randomization procedure is not reported in the registry record. The trial is listed as having no masking. In general, an open-label design means treatment assignment is not masked from participants or study personnel. For an objective laboratory measure such as HbA1c, outcome assessment may be less directly affected by expectations than a subjective outcome, but an open-label design can still influence behavior, treatment management, reporting, and other aspects of trial conduct. The registry record does not describe steps taken to address those possibilities.

The design is parallel-group: participants were assigned to one dose group rather than crossing between both doses. The primary endpoint is measured at Week 26. The unequal stated overall durations between arms are relevant to broader follow-up and safety interpretation, while the primary outcome comparison is specified at a common Week 26 assessment.

What does the analysis population and imputation method mean?

The full analysis set is described as including all randomized subjects. Missing values were imputed with last observation carried forward, meaning a participant’s most recent observed value was carried forward to represent a later missing value in the analysis. In general, this procedure makes an assumption that the last observed measurement is an adequate substitute for an unobserved later measurement. That assumption may not reflect the participant’s actual subsequent outcome. The registry does not report the amount or pattern of missing data, or sensitivity analyses using alternative assumptions. Consequently, the impact of imputation on the estimate cannot be assessed from the registry record alone.

The record identifies one posted primary analysis and lists many outcome measures, including 60 posted secondary outcomes. It does not report a multiplicity adjustment for the secondary analyses or give a complete protocol-level statistical analysis plan. Without those details, the extent to which testing across outcomes was controlled is not reported in the registry record.

10. Confidence Intervals and Precision

The registry’s 95% confidence interval for the ANCOVA treatment difference runs from -0.55 to -0.24 around an estimate of -0.40. The interval is entirely on the negative side of zero, in keeping with the registry’s reported contrast direction and its p-value of < 0.001. It communicates the range of treatment-difference values compatible with the data and the model at the stated confidence level. It does not mean that there is a 95% probability that the true value lies within this particular interval.

The recalculated unadjusted contrast is 0.4 for the 0.9 mg arm minus the 1.8 mg arm, with an approximate 95% confidence interval from 0.234 to 0.566. This is a separate estimate with a reversed subtraction order and a different analysis method. Its interval should not be substituted for the sponsor’s ANCOVA interval. The recalculation uses an approximate pooled standard deviation from summary statistics, so its precision is approximate as well.

In general, interval width depends on the amount of information and variability in the data, along with the analysis model. The posted arm standard errors are 0.06 for both least-squares means. Those standard errors express uncertainty in the estimated means, not the spread of individual HbA1c changes. Individual responses may vary more widely, and their distribution is not described by the confidence interval for the group contrast.

The interval indicates statistical precision for the specified endpoint and analysis, but not certainty about outcomes beyond Week 26, other endpoints, or safety. The registry does not provide confidence intervals for the serious adverse event counts, and the reported primary interval cannot be applied to those safety observations.

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 illustrations express the required number of participants per arm for this continuous endpoint. They are not total event counts. In general, a time-to-event planning display may express a required total number of events, but that is not the measure used for this endpoint. The figures are planning illustrations based on the observed effect, not the trial’s original design justification; the original design assumptions are not reported in the registry record.

The stated planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms and a normal approximation. The effect to detect is set to a stated fraction of the observed effect. For planning at 100% of the observed effect, the illustration lists 83 participants per arm for 80% power and 111 participants per arm for 90% power. At 75% of the observed effect, it lists 147 participants per arm for 80% power and 196 participants per arm for 90% power. At 50% of the observed effect, it lists 330 participants per arm for 80% power and 441 participants per arm for 90% power.

In plain terms, the 80% power column describes the planning sample size associated with an 80% chance of detecting the specified effect under the assumptions; the 90% column uses a higher target chance and therefore lists larger requirements for these scenarios. Smaller effects need more participants to distinguish them from random variation under otherwise fixed assumptions. These figures are not observed power. They are hypothetical planning quantities based on the posted effect estimate and assumptions, not a statement about the probability that the completed trial’s finding is true or reproducible.

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 = 0.400; pooled standard deviation σ = 0.916.
2. n = 2 × (1.96 + 0.842)² × 0.916² / 0.400² = 82.3, rounded up to 83 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 <- 0.4                      # observed difference in means
sigma <- 0.915860251348425                     # 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(2505334)
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 = 83 (80% power), 111 (90% power)
 75% of observed effect: n per arm = 147 (80% power), 196 (90% power)
 50% of observed effect: n per arm = 330 (80% power), 441 (90% power)
power.t.test: 83.3 
Simulated power at n = 83 per arm: 0.801 (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 registry’s own primary result is the ANCOVA treatment difference of -0.40, with a 95% confidence interval from -0.55 to -0.24 and p < 0.001. It is reported as a superiority analysis. The recalculation from the posted arm summaries gives 0.4 for the contrast defined as 0.9 mg minus 1.8 mg, with an approximate 95% confidence interval from 0.234 to 0.566 and p < 0.001. The reversed subtraction order explains the opposite signs.

The recalculation is an unadjusted check based on summary statistics and an approximate pooled standard deviation. The registry analysis adjusts for baseline response using ANCOVA. It therefore does not reproduce the sponsor’s method or replace its estimate. The registry record does not flag a discrepancy between the analyses. The sponsor’s reported ANCOVA result is the primary result to use when describing the trial’s own statistical analysis; the recalculation provides a complementary, approximate comparison based on posted summaries.

14. Multiplicity, Missing Data and Analysis Populations

The registry identifies one primary analysis for the Week 26 HbA1c change endpoint. It also lists 61 outcome measures and 60 posted secondary outcomes. The available record does not describe a complete hierarchy of testing, a multiplicity adjustment, or a protocol-defined strategy for deciding which secondary results should be treated as confirmatory. Accordingly, the control of false-positive risk across multiple outcome tests is not reported in the registry record. This does not change the meaning of the primary analysis, but it is relevant when interpreting a collection of secondary findings.

The primary analysis population is the full analysis set, described as including all randomized subjects. The posted arm summaries each list 233 participants. The total enrollment is 635, and the record does not explain the relationship between enrollment and the analyzed counts. The number of participants with observed Week 26 measurements, the reasons for any missing observations, and the pattern of missingness are not reported in the registry record.

For missing primary endpoint values, the registry states that last observation carried forward was used. In general, carrying a participant’s last observed value forward treats it as a stand-in for a later unobserved outcome. This approach relies on an assumption about how the outcome would have changed after the last measurement. The record does not report sensitivity analyses that test whether the estimated treatment contrast changes under other missing-data assumptions, nor does it describe the extent of missingness. The robustness of the primary result to those assumptions therefore cannot be assessed from the record.

The registry specifies treatment as a fixed effect and baseline response as a covariate in ANCOVA. Additional model details, including diagnostics, other covariates, or handling of any protocol deviations, are not reported in the registry record. The record also does not report interim analyses, a sample-size justification, or a full statistical analysis plan. Those omissions limit assessment of how the posted analysis relates to all prespecified procedures, but do not alter the reported values themselves.

15. Safety Results

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

Deaths

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: rectal cancer (0.9%), calculus urinary (0.4%), lacunar infarction (0.4%), pancreatitis acute (0.4%), cataract (0.4%).

The registry lists serious adverse events affecting 8 of 233 participants in the liraglutide 1.8 mg group and 3 of 233 in the 0.9 mg group. It lists other adverse events affecting 99 of 233 and 65 of 233 participants, respectively. Deaths are listed as 0 in each group. These are participant counts in the registry’s adverse-event record; they are not the primary efficacy endpoint and do not describe the severity, duration, or clinical course of every event.

The serious-event terms listed among the top reported terms include Rectal cancer at 0.9%, and Calculus urinary, Lacunar infarction, Pancreatitis acute, Cataract and Large intestinal polypectomy at 0.4% each. The registry excerpt does not assign these terms to a specific treatment arm, so they should not be attributed to one dose group from these entries alone.

The arm descriptions give different overall planned durations: approximately 67 weeks for 1.8 mg/day and approximately 41 weeks for 0.9 mg/day. Event counts can be affected by observation time, but the registry record does not provide exposure-adjusted rates or enough event-level information to make such an adjustment. The counts alone do not establish a causal safety difference between doses. No medical conclusion should be drawn from these figures without fuller clinical and safety context.

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

The primary finding is a Week 26 difference in HbA1c change between two liraglutide dose groups in a randomized, open-label, parallel-group trial. The registry’s ANCOVA estimate is -0.40, with a 95% confidence interval from -0.55 to -0.24 and p < 0.001. The negative sign is tied to the registry’s treatment-contrast direction; the posted arm values show a mean change of -0.23 for 1.8 mg and 0.17 for 0.9 mg. The endpoint is a continuous laboratory measure, and the estimate describes an average group contrast rather than an individual participant’s outcome.

The result does not by itself establish the size or durability of effects after Week 26, or how outcomes varied from person to person. The registry lists additional outcomes, but the primary finding should not be treated as a result for every secondary measure. Nor does a statistically significant contrast automatically establish its clinical importance; the record does not provide a threshold for interpreting the magnitude as clinically meaningful.

The result also does not provide a complete account of safety. Serious adverse events affected 8 participants in the 1.8 mg group and 3 in the 0.9 mg group, while other adverse events affected 99 and 65 participants, respectively. These counts have no reported confidence intervals or exposure-adjusted comparison in the registry record. The trial arms also had different stated overall durations, which is relevant to reading event counts.

Finally, the analysis relies on ANCOVA and last observation carried forward for missing values. The record does not report the amount of missing data or sensitivity analyses, so the effect of alternative missing-data assumptions cannot be judged from the registry entry. The findings describe the posted statistical analysis in this study population and design; they should not be generalized beyond that evidence without additional information.

17. Why This Trial Matters Statistically

This trial is statistically informative because it compares two active doses in a randomized parallel-group design and reports both arm-level changes and a baseline-adjusted treatment contrast. The continuous HbA1c endpoint preserves information about the degree of change, while ANCOVA accounts for baseline response in estimating the group comparison. The posted confidence interval makes the uncertainty around the primary contrast visible rather than relying on a p-value alone.

The trial also illustrates why the direction and method of a contrast must be checked before comparing estimates. The registry reports -0.40, while an unadjusted recalculation gives 0.4 because it subtracts in the opposite order. The two estimates also come from adjusted and unadjusted methods. Reading the sign without knowing the contrast order could invert the interpretation; treating the estimates as identical would disregard their different statistical procedures.

Other design details affect how broadly the result can be interpreted. The study was open-label, and its two groups had different stated overall durations. Missing values were handled by last observation carried forward, but the record does not report missingness patterns or sensitivity analyses. These features do not erase the primary result, but they define important limits on interpretation and reproducibility.

Finally, the registry lists numerous secondary outcomes and adverse-event counts alongside the primary finding. Distinguishing a prespecified primary comparison from a larger set of outcomes, and separating efficacy estimates from descriptive safety counts, is central to sound statistical reading. The record supports a focused statement about Week 26 HbA1c change, while leaving several protocol and analysis details unreported.

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 registry record does not report the original sample-size justification, randomization procedure, allocation concealment, or a full statistical analysis plan. It also does not report interim analyses. These details would help assess design decisions and the relationship between planned and posted analyses, but their absence does not allow those procedures to be inferred.

The analysis lists 233 participants in each arm, while overall enrollment is 635. The record does not explain this difference or state how many participants had observed primary endpoint data. Although the full analysis set is described as including all randomized subjects, missing values were imputed using last observation carried forward. The amount and pattern of missing data, reasons for missingness, and sensitivity analyses are not reported, limiting assessment of how robust the estimate is to the imputation assumption.

The trial had no masking. The record does not describe measures to limit potential effects of treatment awareness on study conduct or participant behavior. The stated overall durations also differed between arms, with an extension period in the 1.8 mg group. This complicates interpretation of adverse-event counts because the registry does not provide exposure-adjusted rates or event timing sufficient to compare risks over equivalent observation time.

The primary endpoint is limited to Week 26 change in HbA1c. The posted estimate does not answer questions about persistence beyond that assessment or individual response patterns. The registry lists many secondary measures, but multiplicity procedures are not reported. Finally, the unadjusted recalculation depends on an approximate pooled standard deviation derived from posted summary statistics; it is not a replacement for the sponsor’s baseline-adjusted ANCOVA.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Trial Comparing the Efficacy and Safety of Liraglutide 1.8 mg/Day to Liraglutide 0.9 mg/Day in Japanese Subjects With Type 2 Diabetes Mellitus. — NCT02505334.
• Registry record last updated 2018-09-05; status last verified 2018-07.
• Results first posted 2018-06-25.
• 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

NCT02505334 is a completed Phase 3 randomized, open-label, parallel-group trial comparing liraglutide 1.8 mg/day with liraglutide 0.9 mg/day in Japanese subjects with type 2 diabetes mellitus. The registry reports enrollment of 635. The primary endpoint was change in HbA1c from Week 0 to Week 26, analyzed using ANCOVA with treatment as a fixed effect and baseline response as a covariate.

The reported treatment difference is -0.40, with a 95% confidence interval from -0.55 to -0.24 and p < 0.001. The full analysis set included all randomized subjects, and missing values were imputed using last observation carried forward. The record lists 60 posted secondary outcomes and reports serious adverse events affecting 8 participants in the 1.8 mg group and 3 in the 0.9 mg group. Important details, including the original sample-size justification and sensitivity analyses for missing data, are not reported in the registry record.