A Trial Comparing Sequential Addition of Insulin Aspart Versus Further Dose Increase With Insulin Degludec/Liraglutide in Subjects With Type 2 Diabetes Mellitus, Previously Treated With Insulin Degludec/Liraglutide and Metformin and in Need of Further Intensification: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Trial Comparing Sequential Addition of Insulin Aspart Versus Further Dose Increase With Insulin Degludec/Liraglutide in Subjects With Type 2 Diabetes Mellitus, Previously Treated With Insulin Degludec/Liraglutide and Metformin and in Need of Further Intensification (NCT02100475), a randomized phase 3 study. The primary outcome is Change From Baseline in HbA1c (Glycosylated Haemoglobin).

2. Clinical Question

3. Trial Design

NCT02100475 is listed as a phase 3, randomized, open-label trial with a parallel-group design. Randomization assigns participants to treatment groups by chance, while a parallel design means each participant follows one assigned group rather than crossing between treatments. The trial was completed, and enrollment was 31. It is described as a global trial and as an extension to NN9068-3952, NCT01952145 (DUAL™ V).

The groups were insulin degludec/liraglutide plus metformin, with 16 participants, and insulin degludec/liraglutide plus insulin aspart plus metformin, with 15 participants. The primary outcome was change from baseline in HbA1c at week 26. The registry lists body-weight change and the number of treatment-emergent confirmed hypoglycaemic episodes as secondary outcomes. Results are posted for body-weight change, but results for the hypoglycaemic-episode outcome are not reported in the registry record.

The primary outcome population was the full analysis set, described as including all randomized participants. Missing primary-outcome data were imputed using last observation carried forward. Details such as the randomization method, allocation concealment, eligibility criteria beyond the treatment history and need for intensification, dose schedules, and the original sample-size justification are not reported in the registry record. The trial is unmasked, so participants and study personnel were not blinded according to the registry classification. That design feature is relevant when interpreting outcomes that can be influenced by awareness of assigned treatment, although HbA1c is a laboratory measure.

The reported comparison is small in participant count, and the observed estimates consequently need to be read with their uncertainty in mind. The trial design establishes which groups were compared and how the endpoint was defined; it does not alone determine how precisely the treatment contrast can be estimated.

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 with type 2 diabetes mellitus who were inadequately controlled (HbA1c level ≥ 7% \[53 mmol/mol\]) on treatment with IDegLira and metformin after 26 weeks of treatment in the NN9068-3952 trial were screened. Eligible subjects were randomised in a 1:1 manner to one of the two parallel treatment groups (IDegLira or IDegLira + IAsp).

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Full analysis set (FAS) included all randomised subjects (31 subjects). Missing data 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: Baseline characteristics were presented for full analysis set (FAS = 31 subjects).

6. Primary Endpoint

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

The primary endpoint is change from baseline in HbA1c, or glycosylated haemoglobin, after 26 weeks. HbA1c is reported as a percentage of glycosylated haemoglobin. The registry describes measurement at week 0 and week 26 and identifies the endpoint statistic as a mean change with standard deviation.

Change scores express how a measured value differs from its starting value. Comparing mean change between randomized groups is one way to describe whether the groups moved differently, on average, over the treatment period. HbA1c is used as a measure of glycaemic status over time, making it relevant to a trial comparing strategies intended for further treatment intensification in type 2 diabetes. The endpoint is continuous: participants can have a range of measured values rather than being classified only as having or not having an event.

The posted arm means describe average change, not each participant's response. The registry reports a mean change of -0.43 for insulin degludec/liraglutide and -0.14 for insulin degludec/liraglutide plus insulin aspart. The negative values indicate mean decreases from baseline in both groups as the outcome is recorded. These values do not show how many participants improved, worsened, or had no change. Nor does the endpoint alone describe other outcomes, adverse events, or the overall balance of benefits and harms.

The full analysis set included all 31 randomized participants, and missing values were imputed using last observation carried forward. The extent and pattern of missing data are not reported in the registry record, so the potential influence of imputation cannot be assessed from the posted information alone.

7. Results Posted in the Registry

• IDegLira + IAsp: mean -0.14 versus -0.43 on control; difference 0.29 (95% CI -0.43 to 1.01), standardized effect d = 0.29; z-test p = 0.429.

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 From Baseline in Body Weight

• 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 deviations.
• Result:
  • IDegLira + IAsp: mean 1.50 versus 0.90 on control; difference 0.60 (95% CI -1.32 to 2.52), standardized effect d = 0.22; z-test p = 0.540.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between IDegLira + IAsp and the control. The interval gives the range of effects that remain compatible with the data.

The posted secondary results cover change in body weight, while the registry also lists treatment-emergent confirmed hypoglycaemic episodes as a secondary outcome. The overall statistical pattern is that the primary HbA1c comparison and the posted body-weight comparison are both uncertain; the registry record does not post results for the hypoglycaemic-episode outcome. Thus, the secondary information does not resolve the uncertainty in the primary result or provide a complete view across the stated outcomes.

Secondary endpoints can add context by measuring different aspects of the treatment comparison. Body weight is a continuous outcome, whereas the number of confirmed hypoglycaemic episodes concerns events. Their scales and interpretations differ, and neither can be treated as a substitute for the primary HbA1c measure. Any apparent consistency or contrast across outcomes should be interpreted cautiously, particularly when the study has few participants and estimates are imprecise.

Unadjusted secondary results carry less weight as confirmatory evidence when several outcomes are considered. The more comparisons examined, the greater the chance that at least one result may appear notable by chance. A multiplicity plan can control or otherwise address that issue, but a correction or hierarchy for secondary outcomes is not reported in the registry record. Without that information, secondary findings are best treated as exploratory or supportive descriptions, rather than independent proof of a treatment effect.

The registry lists three outcome measures and one reported analysis overall. It does not report a complete result for every listed secondary outcome, and the scope and ordering of any planned analyses are not described. These gaps mean the secondary evidence is partial. It can inform what was measured and which results were posted, but it cannot establish a comprehensive outcome profile or settle the primary comparison.

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

• Change From Baseline in Body Weight
• Number of Treatment-emergent Confirmed Hypoglycaemic Episodes

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is continuous, so the central summaries are means and standard deviations. A mean represents the average observed change in a group. A standard deviation describes how widely individual measurements tend to vary around that group mean; it is not the uncertainty of the estimated mean itself. The posted group means are -0.43 and -0.14, with standard deviations of 0.94 and 1.09, respectively. The numbers of participants contributing to the summaries are 16 and 15.

What does a mean difference represent?

A mean difference compares the average outcome in one group with the average in another. Its sign depends on the order in which groups are subtracted. In the simple change-score recalculation, the mean difference is 0.29 for insulin degludec/liraglutide plus insulin aspart relative to insulin degludec/liraglutide. Since the group means are negative changes, that positive contrast indicates a less negative average change in the group receiving insulin aspart. It does not mean that each participant had that change, or that the difference is precise enough to rule out other possible values.

There is also a sponsor-reported ANCOVA analysis of mean difference in final values. ANCOVA, or analysis of covariance, compares groups on a final measurement while statistically adjusting for one or more covariates, often including the baseline measurement. Adjustment can improve precision when baseline values predict final values, and it can account for baseline imbalance. The registry record reports a 95% confidence interval from -1.05 to 0.46 and p = 0.427 for the sponsor's analysis, but does not identify the covariates used. The reported final-value analysis and the unadjusted change-score recalculation answer related, but not identical, statistical questions. Their estimates also use a direction convention that should be understood before comparing signs.

What does a confidence interval add?

A confidence interval communicates uncertainty around an estimated contrast. In general, a 95% confidence interval is produced by a procedure that, over repeated samples under its assumptions, would cover the target parameter in 95% of such intervals. It is not a statement that the fixed parameter has a 95% probability of lying in this particular interval. A wide interval signals limited precision and leaves a broader range of values compatible with the data and model. When an interval includes zero for a difference, it includes no difference as one value consistent with the analysis, but it may also include nonzero differences.

The p-value measures how unusual the observed data, or data more extreme, would be under a specified null hypothesis and model. It is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of an effect. Here, the sponsor's p-value is 0.427. Interpreting it alongside the estimate and confidence interval is more informative than treating it as a standalone pass-or-fail threshold.

Why can small samples limit precision?

With relatively few participants, the observed mean and standard deviation can be sensitive to the particular individuals enrolled. Small samples commonly produce less precise estimates, especially when individual outcomes vary. The standard deviations in both HbA1c groups are larger than the mean difference estimate, so the group averages need to be interpreted against the spread of individual values. This comparison is descriptive; it does not by itself quantify the probability of a clinically meaningful effect.

Over-dispersion is a general concern when variability is greater than a statistical model assumes. If a model underestimates variability, its standard errors can be too small and its intervals too narrow. The registry record identifies standard deviations for the continuous endpoint, but does not report a model-based assessment of over-dispersion or diagnostics. For a continuous ANCOVA outcome, the relevant model assumptions and residual checks are not described in the registry record, so they cannot be evaluated from the posted summary alone.

How are missing values handled?

The primary analysis population is described as the full analysis set, including all randomized participants. Missing values were imputed using last observation carried forward, or LOCF. In general, LOCF replaces a missing later measurement with the participant's last observed measurement. It is a simple method that can preserve participants in an analysis, but its assumptions may be unrealistic if outcomes would have changed after the last observed value. It can also make estimated uncertainty appear smaller than it is if the imputation process is not reflected appropriately. The number of missing values, their timing, the reasons for missingness, and sensitivity analyses using other methods are not reported in the registry record.

For the mean difference and planning illustrations, distributional assumptions matter as well. The sample-size illustrations use a normal approximation, as stated in the planning description. In general, this approximates the behavior of estimates using a normal distribution. Such an approximation is more dependable when its assumptions are reasonable; with small group sizes, results can be sensitive to departures from assumptions. The registry does not report the original design assumptions or a sample-size justification, so the planning illustrations should not be mistaken for the trial's original design calculations.

Randomization supports a comparison between assigned groups by reducing systematic differences in expectation, but it cannot guarantee balance in a small trial. Open-label conduct may influence treatment behavior and outcomes that depend on reporting or clinical decisions. HbA1c is a laboratory-based measure, but the registry record does not describe masking of laboratory personnel or other procedures that might reduce assessment bias. These design and analysis details provide context for reading the point estimate, interval, and p-value together rather than treating any one statistic as conclusive.

11. Confidence Intervals and Precision

The precision of a treatment contrast is reflected most directly by its confidence interval. The sponsor-reported ANCOVA estimate is -0.3, with a 95% confidence interval from -1.05 to 0.46. The interval spans zero, so the posted analysis is compatible with no difference as well as with values on either side of zero. Its breadth indicates that the estimate does not pin down a narrow range of plausible contrasts. The reported p-value is 0.427, consistent with the interval not excluding zero under the reported analysis.

Precision is distinct from the direction of the point estimate. The point estimate is one value calculated from the observed sample; the interval expresses uncertainty around it. Participants in the trial were limited to 16 in one group and 15 in the other, while the standard deviations for HbA1c change were 0.94 and 1.09. Variation among participants and the small group sizes constrain how precisely a mean contrast can be estimated.

The unadjusted recalculation gives a mean difference of 0.29 for change from baseline. It is based on summary statistics and is not the same as the sponsor's ANCOVA estimate of final values. The registry reports no confidence limits for the arm means themselves. The adjusted analysis's interval is therefore the main posted measure of inferential precision for the primary outcome, while the raw summaries help describe the observed changes.

A confidence interval is not a range containing every participant's outcome; it concerns the estimated group contrast. Nor does an interval crossing zero prove that the groups are equivalent. A conclusion of equivalence would require a prespecified margin and an analysis designed for that question; such a margin is not reported in the registry record. The interval instead shows that the available data leave substantial uncertainty about the size and direction of the 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 illustrations translate an observed effect into approximate group sizes for future planning. For these continuous outcomes, figures are participants per arm, not total events. The calculations assume a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They use the stated fraction of the observed effect as the effect to detect. The original trial's own design assumptions and sample-size justification are not reported in the registry record.

The 80% power column gives the approximate number of participants per arm that would be required for an 80% chance of detecting the selected effect under the stated assumptions, if that effect were the true one and the model assumptions held. The 90% power column describes the corresponding planning target with a higher chance of detection. A higher target generally requires more participants. These are planning quantities, not evidence about the chance that the observed trial result is true or meaningful.

The rows use the full observed effect, 75% of the observed effect, and 50% of the observed effect. The smaller the effect a future study is designed to detect, the more participants it generally needs, because small differences are harder to distinguish from random variation. For HbA1c, planning for 100% of the observed effect gives 193 participants per arm for 80% power and 258 participants per arm for 90% power; planning for 75% gives 342 and 458, and planning for 50% gives 770 and 1,031. These figures are illustrative and depend on the effect estimate from a small trial, which may itself be uncertain.

The calculations are not observed power. Observed power is not a useful substitute for interpreting the result's estimate and confidence interval. The planning figures instead show how sample-size requirements change under stated hypothetical assumptions. They should not be read as the actual enrollment target used in NCT02100475, or as a claim that the trial was designed to detect the observed difference.

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 = 0.290; pooled standard deviation σ = 1.015.
2. n = 2 × (1.96 + 0.842)² × 1.015² / 0.290² = 192.4, rounded up to 193 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 <- 0.29                      # observed difference in means
sigma <- 1.01518471225684                     # 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(2100475)
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 = 193 (80% power), 258 (90% power)
 75% of observed effect: n per arm = 342 (80% power), 458 (90% power)
 50% of observed effect: n per arm = 770 (80% power), 1,031 (90% power)
power.t.test: 193.3 
Simulated power at n = 193 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.

14. Analyses Reported by the Sponsor

The sponsor's reported primary-outcome analysis is an ANCOVA comparison of mean final values. It gives an estimate of -0.3, a 95% confidence interval from -1.05 to 0.46, and p = 0.427. This is the registry's own inferential result for the primary comparison. It is based on an adjusted model, although the covariates included in that model are not reported in the registry record.

The simple recalculation uses posted group means for change from baseline and gives a mean difference of 0.29. It is unadjusted and uses change summaries, rather than the final-value parameter specified in the sponsor analysis. Because these approaches differ in both adjustment and outcome parameter, their estimates and signs should not be forced into exact agreement. The registry record does not flag a discrepancy between the recalculation and the sponsor result.

The recalculation can help readers understand the arithmetic implied by the posted arm-level summaries, but it cannot reproduce the participant-level ANCOVA. It does not replace the sponsor's analysis or provide the same adjustment. The sponsor result should be used when describing the registered inferential comparison, while the unadjusted contrast should be understood as a separate descriptive recalculation from summary data. Neither estimate removes the uncertainty evident in the confidence interval, and neither alone establishes equivalence or a clinically important difference.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study makes several statistical comparisons or evaluates multiple outcomes. Even if every null hypothesis is true, conducting more tests increases the chance that one or more results will appear statistically noteworthy by chance. Researchers may address this through a prespecified hierarchy, adjustment procedure, or a clearly exploratory interpretation. A multiplicity adjustment or testing hierarchy is not reported in the registry record. The number and scope of planned analyses beyond the one posted analysis are also not reported in the registry record.

The trial lists the primary HbA1c outcome and secondary outcomes for body-weight change and treatment-emergent confirmed hypoglycaemic episodes. Results are posted for body weight, while results for the hypoglycaemic-episode outcome are not reported in the registry record. Because not all listed outcomes have posted results, the available information does not permit a complete assessment of how all outcomes were analyzed or whether any multiplicity approach was used. Secondary findings should therefore be interpreted with appropriate caution rather than treated as separate confirmatory claims.

The full analysis set is described as including all randomized participants, a total of 31. The primary endpoint's missing values were imputed using last observation carried forward. This statement identifies the method, but not how many values were missing, when they were missing, why they were missing, or whether missingness differed between groups. Those details are not reported in the registry record. The numbers of participants contributing to the arm summaries are 16 and 15, matching the randomized group sizes, but that fact does not establish that every measurement was observed rather than imputed.

LOCF carries the last available measurement forward when a later value is missing. In general, this treats the last observed value as an adequate stand-in for the missing value and can be problematic if the outcome would have changed. Sensitivity analyses can assess whether conclusions depend on that assumption, but such analyses are not reported in the registry record. The original analysis plan, detailed missing-data assumptions, and any alternate analysis populations are also not reported in the registry record. These omissions limit assessment of robustness; they do not, by themselves, show that the analysis is invalid.

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: chronic myeloid leukaemia (6.7%), renal failure (6.7%), acute coronary syndrome (6.2%), coronary revascularisation (6.2%).

The safety record lists serious adverse events for 1 of 16 participants in the insulin degludec/liraglutide group and 2 of 15 participants in the group also receiving insulin aspart. It also lists other adverse events for 11 of 16 and 3 of 15 participants, respectively. These are small group counts and should be read as descriptive observations from this trial, not as precise estimates of comparative risk.

The registry lists chronic myeloid leukaemia and renal failure among the serious adverse-event terms, each at 6.7%, and acute coronary syndrome and coronary revascularisation, each at 6.2%. The posted information does not provide enough context to infer from these terms alone whether an event was caused by treatment or how the listed percentages relate to participants and event counts. Serious adverse events can occur during a trial without being caused by the assigned treatment.

Deaths are not reported in the registry record. The small enrollment and limited follow-up period of 26 weeks mean that these counts cannot establish a reliable difference in safety between groups or characterize uncommon events. The registry's safety information should be considered alongside the study's size and the absence of detailed causal assessments, not as a stand-alone determination of comparative safety.

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

The results describe a randomized comparison of two treatment strategies in 31 participants with type 2 diabetes who had previously received insulin degludec/liraglutide and metformin and were considered in need of further intensification. Over 26 weeks, the posted mean HbA1c changes were -0.43 with insulin degludec/liraglutide and -0.14 with insulin degludec/liraglutide plus insulin aspart. The sponsor's adjusted comparison of final values was -0.3, with a 95% confidence interval from -1.05 to 0.46 and p = 0.427.

Those statistics describe the observed data and uncertainty under the sponsor's analysis. They do not establish that the groups are identical, and they do not demonstrate equivalence. They also do not prove that one treatment strategy is superior. The interval is compatible with a range of contrasts, while the trial's small group sizes limit precision. The point estimate should not be read without its interval, and the p-value should not be treated as a measure of clinical importance.

The results do not show how an individual participant's HbA1c changed, nor do they provide a complete account of every outcome listed for the trial. Body-weight change is posted as a secondary outcome; results for treatment-emergent confirmed hypoglycaemic episodes are not reported in the registry record. Safety counts are also limited by the small sample. The findings therefore concern the posted outcomes and participants, rather than every possible benefit or harm.

Generalizability depends on the trial population, treatment history, study conduct, and follow-up. The registry describes an extension trial and a population already treated with insulin degludec/liraglutide and metformin. The results should not automatically be applied to people with other treatment histories, to other populations, or to time periods beyond the reported 26-week endpoint. Details needed to judge the broader applicability, including eligibility criteria and dosing regimens, are not reported in the registry record.

The most appropriate conclusion is statistical rather than clinical: the posted primary comparison does not provide a precise estimate that rules out zero difference, and the evidence does not establish equivalence or a clear between-group difference. Interpretation of the broader treatment question would require considering additional clinical context and evidence beyond these results.

18. Why This Trial Matters Statistically

This trial illustrates why a point estimate, a p-value, and a confidence interval answer different questions. The mean change summaries and recalculated contrast describe the observed group averages. The sponsor's ANCOVA estimate and p-value come from an adjusted analysis of final values. The confidence interval communicates the uncertainty around that adjusted estimate. Reading the statistics together avoids turning one threshold into the entire interpretation.

The small enrollment is especially important for understanding precision. With 16 and 15 participants, random differences between the groups can remain even after randomization, and the observed average may be sensitive to a few participants. The confidence interval shows the practical consequence: the estimated contrast is not tightly constrained. A non-significant result in a small trial should not be translated into proof of no effect, just as an observed point estimate should not be treated as a definitive treatment difference.

The trial also highlights the distinction between a posted sponsor analysis and a summary-statistic recalculation. ANCOVA of final values and a simple unadjusted comparison of change scores are related but distinct estimands and procedures. Naming the analysis makes clear why estimates may differ and why the raw summaries do not reproduce a participant-level adjusted model.

Finally, the record illustrates the limits of incomplete reporting. The primary missing-data method is named, but the amount of missingness, sensitivity analyses, multiplicity approach, and original sample-size justification are not reported in the registry record. Those details matter because they help readers assess whether the estimate is robust to assumptions and how the study was intended to answer its question. Recognizing such limits is part of statistical interpretation, not a substitute for it.

19. Limitations

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

The trial enrolled 31 participants, split between groups of 16 and 15. This limits precision and makes estimates potentially sensitive to individual observations. The primary confidence interval is broad enough to include zero and values on either side, so the posted result does not tightly determine the size or direction of the contrast.

The study was open-label. Awareness of assigned treatment can affect behavior, co-interventions, reporting, and some assessments, although the primary endpoint is a laboratory measure. The registry does not describe whether laboratory staff were masked or what other safeguards were used. The randomization process and allocation concealment are also not reported in the registry record.

Missing primary-outcome data were handled with last observation carried forward, but the number of missing measurements, their timing, reasons for missingness, and any sensitivity analyses are not reported in the registry record. The assumption that the last observed value represents a later missing value may not hold if outcomes change over time. The posted information is insufficient to assess the likely impact of that assumption.

The registry does not report the original sample-size justification, detailed analysis plan, ANCOVA covariates, or any multiplicity procedure. Those omissions make it harder to evaluate whether the analysis followed a prespecified plan and how the various outcomes were prioritized. The listed hypoglycaemic-episode outcome has no posted result, limiting assessment of the full set of stated outcomes.

The registry identifies the trial as an extension to an earlier trial, but details of the extension's eligibility criteria, treatment schedules, adherence, and participant flow are not reported in the registry record. Follow-up for the primary endpoint was 26 weeks, so these results do not describe longer-term outcomes. These limitations constrain precision, robustness assessment, and generalizability; they do not by themselves determine the direction of any treatment effect.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Trial Comparing Sequential Addition of Insulin Aspart Versus Further Dose Increase With Insulin Degludec/Liraglutide in Subjects With Type 2 Diabetes Mellitus, Previously Treated With Insulin Degludec/Liraglutide and Metformin and in Need of Further Intensification — NCT02100475.
• Registry record last updated 2017-01-20; status last verified 2016-11.
• Results first posted 2017-01-20.
• 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

NCT02100475 was a completed, randomized, open-label, parallel-group phase 3 trial with enrollment of 31. It compared insulin degludec/liraglutide plus metformin with insulin degludec/liraglutide plus insulin aspart and metformin in participants with type 2 diabetes needing further intensification.

The primary outcome was mean change from baseline in HbA1c at week 26. Mean changes were -0.43 in the 16-person comparator group and -0.14 in the 15-person insulin-aspart group. The sponsor reported an ANCOVA mean difference in final values of -0.3, with a 95% confidence interval from -1.05 to 0.46 and p = 0.427. The interval reflects substantial uncertainty and includes zero.

Body-weight change was also posted as a secondary outcome. The registry lists treatment-emergent confirmed hypoglycaemic episodes, but results for that outcome are not reported in the registry record. The primary analysis used the full analysis set and last observation carried forward for missing data; further details on missingness, multiplicity, and the original sample-size justification are not reported in the registry record.