Comparing Efficacy and Safety of Thrice Daily Versus Twice Daily NovoMix® 30 (Biphasic Insulin Aspart 30) in Subjects With Type 2 Diabetes Inadequately Controlled With Basal Insulin: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Comparing Efficacy and Safety of Thrice Daily Versus Twice Daily NovoMix® 30 (Biphasic Insulin Aspart 30) in Subjects With Type 2 Diabetes Inadequately Controlled With Basal Insulin (NCT02582242), a randomized phase 4 study. The primary outcome is Change in Glycosylated Haemoglobin (HbA1c).

2. Clinical Question

The trial was randomized and parallel-group, with no masking. It enrolled 437 participants. The primary outcome was change in HbA1c from baseline to week 24. The posted group means were reductions of 1.66 percentage points with thrice-daily dosing and 1.52 percentage points with twice-daily dosing. The sponsor’s reported repeated-measures analysis estimated a treatment difference of -0.09 percentage points, with a 95% confidence interval from -0.23 to 0.06 and p = 0.260. These figures describe the result under the analysis and contrast definition recorded in the registry; the uncertainty interval includes values on either side of zero.

Several secondary measures address other aspects of glucose control and hypoglycaemia, and safety outcomes are also posted. Those outcomes provide context, but the primary question remains the between-group comparison of change in HbA1c at week 24. The registry record does not report enough detail to infer additional clinical objectives beyond the stated efficacy and safety comparison.

3. Trial Design

This was a phase 4, randomized, open-label, parallel-group trial comparing biphasic insulin aspart 30 administered three times daily with the same preparation administered twice daily. The official title describes the study as a 24-week, multinational, multicentre, treat-to-target trial. The brief summary says it was conducted in Asia. The registry reports an enrollment of 437, with 220 participants in the thrice-daily group and 217 in the twice-daily group for the primary outcome.

The study is recorded as completed. Its recorded start date is 2015-10-19, and its primary completion date is 2017-03-18. The primary endpoint was change from baseline in HbA1c, evaluated at week 24. The analysis population was the full analysis set: randomized participants who were dosed and had any post-randomization data. The outcome description states that missing data were imputed by last observation carried forward.

Randomization and a parallel-group structure support a comparison between the assigned schedules, while the absence of masking means that participants and trial personnel were not blinded to treatment assignment. The registry record does not report the random-sequence generation procedure, allocation concealment, site count, country list, or detailed eligibility criteria. It also does not report the specific dose-adjustment targets and rules implied by the treat-to-target description. These details should not be inferred from the trial title alone.

The primary analysis reported by the sponsor used a mixed model for repeated measurements and gave a treatment difference at week 24. The record does not provide the model’s covariates, covariance structure, exact handling of missing observations within that model, or other implementation details. A separate unadjusted calculation from the posted arm-level means and standard deviations is useful for understanding the summaries, but it is not a substitute for the sponsor’s analysis.

The registry record lists 18 outcome measures and one reported analysis. It lists 17 posted secondary outcomes. The original design’s sample-size assumptions and justification are not reported in the registry record. No conclusion about the planned power or the existence of interim analyses can be drawn from the information recorded.

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.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: FAS included all randomised subjects who were dosed and had any post randomisation data.

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) included all randomised subjects who were dosed and had any post randomisation data.

6. Primary Endpoint

The registry lists the primary outcome as Change in Glycosylated Haemoglobin (HbA1c), measured over Week 0, Week 24.

The primary endpoint is the change in glycosylated haemoglobin, or HbA1c, from baseline to week 24, expressed as a percentage of HbA1c. HbA1c is a continuous measure used to assess average glycaemia over time. A change score describes how the measured value at follow-up differs from the baseline value; a negative change in these results represents a decrease from baseline.

The posted mean change was -1.66 in the thrice-daily group and -1.52 in the twice-daily group. The corresponding standard deviations were 1.0 and 0.94. A standard deviation describes variability among participant measurements within a group; it is not the uncertainty interval for the difference between group means. The group sizes for this outcome were 220 and 217.

Comparing changes between randomized groups is intended to address whether the average change differed between the assigned dosing schedules over the recorded follow-up. A continuous endpoint retains information about the amount of change rather than reducing the result to whether each participant crossed a threshold. The registry also lists secondary outcomes based on HbA1c thresholds, hypoglycaemia, fasting plasma glucose, and self-measured glucose profiles, but these do not replace the designated primary endpoint.

The outcome description says missing data were imputed using last observation carried forward. This means the analysis used a participant’s last recorded observation in place of a missing later value, as described in the registry. The record does not report further details about the imputation process or sensitivity analyses. The endpoint therefore needs to be read alongside both the stated analysis population and the missing-data method, rather than as a measurement available for every randomized participant at week 24.

7. Results Posted in the Registry

• Biphasic Insulin Aspart 30 (Twice Daily): mean -1.52 versus -1.66 on control; difference 0.14 (95% CI -0.04 to 0.32), standardized effect d = 0.14; z-test p = 0.132.

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 FPG by Central Laboratory Analysis

• 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:
  • Biphasic Insulin Aspart 30 (Twice Daily): mean -1.07 versus -1.34 on control; difference 0.27 (95% CI -0.27 to 0.81), standardized effect d = 0.09; z-test p = 0.327.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Biphasic Insulin Aspart 30 (Twice Daily) and the control. The interval gives the range of effects that remain compatible with the data.

7-point SMPG Profiles: Change From Baseline in Mean of 2-hour PPG Over 3 Main Meals (Breakfast, Lunch and Main Evening Meal)

• 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:
  • Biphasic Insulin Aspart 30 (Twice Daily): mean -3.77 versus -3.98 on control; difference 0.21 (95% CI -0.45 to 0.87), standardized effect d = 0.06; z-test p = 0.535.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Biphasic Insulin Aspart 30 (Twice Daily) and the control. The interval gives the range of effects that remain compatible with the data.

7-point SMPG Profiles: Change From Baseline in Mean of PPG Increment Over 3 Main Meals (Breakfast, Lunch and Main Evening Meal)

• 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:
  • Biphasic Insulin Aspart 30 (Twice Daily): mean -1.22 versus -0.96 on control; difference -0.26 (95% CI -0.75 to 0.23), standardized effect d = -0.10; z-test p = 0.297.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Biphasic Insulin Aspart 30 (Twice Daily) and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes span threshold-based HbA1c measures, hypoglycaemia outcomes, fasting plasma glucose, and self-measured glucose profiles. The registry lists 17 posted secondary outcomes, although detailed results are not available in the record for every listed outcome in the registry record described. The available continuous secondary summaries generally show changes in both groups and unadjusted between-group intervals that include zero. This broad pattern does not point to a clear separation between dosing schedules in the posted continuous summaries.

These measures complement the primary HbA1c result by describing other glucose measures and safety-related outcomes. They do not alter the fact that the primary endpoint is change in HbA1c at week 24, analyzed with the sponsor’s repeated-measures model. Secondary results can add context, but they should not be combined into a single claim about an overall treatment effect without an analysis plan that specifies how the outcomes are related.

Unadjusted secondary analyses carry less weight because multiple outcomes create multiple opportunities for apparently notable results. If many tests are conducted and no multiplicity procedure is reported, a small p-value on any one result may be less persuasive than it would be for a single prespecified primary outcome. The registry record does not report a multiplicity strategy, so the posted secondary comparisons should be interpreted cautiously and as supportive or exploratory evidence rather than as independent confirmation.

The secondary outcome list includes proportions achieving HbA1c below 7.0%, with and without specified hypoglycaemic episodes, and counts classified under different definitions. The record also lists fasting plasma glucose and profile-based measures. The presence of these outcomes indicates that the trial assessed several dimensions, but the registry does not provide a basis for assuming that every listed measure was analyzed with the same model or that all results were adjusted for multiple testing.

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

• Proportion of Subjects Achieving HbA1c Below 7.0%
• Proportion of Subjects Achieving HbA1c Below 7.0% Without Severe Hypoglycaemic Episodes.
• Proportion of Subjects Achieving HbA1c Below 7.0% Without Severe or Blood Glucose (BG) Confirmed Hypoglycaemic Episodes (According to the Novo Nordisk Classification)
• Number of Treatment Emergent Hypoglycaemic Episodes Classified According to the American Diabetes Association (ADA) Definition
• Number of Treatment Emergent Hypoglycaemic Episodes Classified According to Novo Nordisk Definition
• Change From Baseline in FPG by Central Laboratory Analysis
• 7-point SMPG Profile
• 7-point SMPG Profiles: Change From Baseline in 2-hour PPG at Individual Meal (Breakfast, Lunch and Main Evening Meal)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is continuous, so its group summaries are means and standard deviations. A mean is the average change across participants in a group. A standard deviation describes how spread out individual changes are around that group average. These are different quantities: the standard deviation reflects variation among participants, whereas a standard error or confidence interval concerns uncertainty in an estimated mean or contrast.

How to read the group contrast

The primary arm summaries are -1.66 for thrice-daily treatment and -1.52 for twice-daily treatment. The unadjusted recalculation reports the difference as twice-daily minus thrice-daily, giving 0.14. The sign of a contrast depends on which group is subtracted from which; reversing the order reverses the sign without changing the underlying separation between the group summaries. The sponsor’s reported treatment difference is -0.09, but the registry record does not specify enough about the contrast coding and model to map its sign directly onto that recalculation.

The recalculated contrast is based on posted summary statistics. It is not the same as fitting a model to participant-level measurements. The sponsor reported a mixed model for repeated measurements, a method designed for data measured repeatedly over time. In general, repeated-measures models can use the pattern of observations across visits to estimate a treatment contrast at a specified time. Their estimates can depend on the model specification and assumptions about the relationship among repeated observations. The registry record does not report the detailed model formula, covariance structure, or covariate set, so those features cannot be described more specifically.

Uncertainty and hypothesis testing

A confidence interval gives a range of estimates compatible with the data and the stated analysis under its assumptions. The sponsor’s 95% confidence interval for the treatment difference is -0.23 to 0.06. The unadjusted recalculation gives -0.042 to 0.322. Because each interval includes zero, neither result rules out a zero difference at the corresponding confidence level. An interval is not a guarantee that the true value lies within its endpoints, nor is it a measure of the probability that the treatment effect is exactly zero.

A p-value measures how unusual the data, or more extreme data, would be under a specified null hypothesis and analysis model. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The sponsor reported p = 0.260. The unadjusted calculation reports p = 0.132. They are results from different analytical approaches and should not be treated as interchangeable. The sponsor’s analysis is the registered report of the primary statistical comparison; the recalculation helps interpret the posted summary statistics.

Repeated measures and missing observations

The outcome description states that missing data were imputed with last observation carried forward. In general, this approach substitutes a participant’s most recent observed value for a missing later value. It creates an analysis value at a later time from an earlier observation and thus relies on assumptions about how the outcome would have evolved after the last measurement. Those assumptions can be consequential if missingness is related to participant outcomes or if the outcome would have changed after the last observation. The registry record does not report the amount or pattern of missingness, the reason for missing observations, or sensitivity analyses that might examine alternative assumptions.

The primary analysis population is the full analysis set, defined in the record as randomized participants who were dosed and had any post-randomization data. This is not described as every randomized participant without qualification. The reported group counts for the primary outcome are 220 and 217. The record does not report further detail on exclusions from the full analysis set or how they were distributed.

Why variability matters

General statistical explanation: greater participant-to-participant variability makes a group mean less precise when the group size is unchanged. The precision of a between-group estimate also depends on the number of participants in each group and on the analysis used. Here, the posted standard deviations for HbA1c change are 1.0 and 0.94. The recalculation uses the observed group summaries to estimate a standard error and interval for an unadjusted contrast. It does not reproduce the sponsor’s model-based uncertainty estimate.

Over-dispersion is a general concern in analyses of counts when observed variability exceeds what a simple count model assumes. It can lead to uncertainty estimates that are too small if not accommodated. The primary endpoint is a continuous mean, not a count, so the listed primary result should not be described as a count analysis. Some secondary outcomes concern numbers of hypoglycaemic episodes, but the registry record does not provide the methods or results needed to assess over-dispersion for those outcomes.

Multiple outcomes and interpretation

The registry lists 18 outcome measures and 17 posted secondary outcomes. When many outcomes are tested, the chance of at least one apparently notable result can increase if each is assessed independently without adjustment. Multiplicity methods can control this issue, but the registry record does not report whether such a procedure was planned or used. The secondary continuous recalculations are unadjusted and should therefore be treated as exploratory evidence rather than as a collection of independent confirmatory tests.

The original sample-size justification, planned effect size, and design assumptions are not reported in the registry record. The separate planning illustrations based on observed effects are not evidence of the trial’s original design or of its ability to detect the observed result. Statistical power is a property of a specified design and assumed effect, not a retrospective probability that a completed study detected a real effect.

11. Confidence Intervals and Precision

Precision concerns how tightly an estimate is determined by the data and analysis. The sponsor-reported treatment difference at week 24 is -0.09, with a 95% confidence interval from -0.23 to 0.06. The interval crosses zero and includes possible differences on either side. Its endpoints show that the analysis does not pinpoint a single treatment difference; a range remains compatible with the data under the model’s assumptions.

The unadjusted recalculation from the posted arm-level summaries estimates a difference of 0.14, with a 95% confidence interval from -0.042 to 0.322. That interval also crosses zero. The sponsor and recalculated intervals come from different methods and estimates, so their widths should not be used as though they were two versions of the same model. The registry record does not provide enough detail about the mixed model to reproduce its result from participant-level data.

Confidence intervals are more informative than a p-value alone because they show both the estimate and a range of values compatible with the analysis. An interval that includes zero does not prove equality or absence of an effect. It indicates that the data, analyzed in that way, do not distinguish zero from the other values within the interval at the stated confidence level. The interval also does not account automatically for unreported sources of uncertainty such as the choice of missing-data assumptions or the handling of multiple outcomes.

The arm-level standard deviations, 1.0 and 0.94, describe variability in individual HbA1c changes. They do not show the precision of the treatment contrast by themselves. Group sizes, the relationship among repeated measurements, and the fitted model also affect the uncertainty of a contrast. The sponsor’s 95% interval is the reported model-based interval, while the recalculated interval is an unadjusted one based on summaries.

Accordingly, the result supports a cautious description: the estimated difference is modest in the context of the observed group changes, but its precise direction and size are uncertain. The interval does not establish equivalence, and the registry does not report an equivalence margin or a non-inferiority margin.

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 express participants per arm for the stated continuous-outcome comparisons. They are not total event counts. The planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each illustration uses a 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 describes the number of participants per arm that a hypothetical study would need, under those assumptions, to have an 80% chance of detecting the specified effect if that effect were present. The 90% power column gives the corresponding planning number for a 90% chance. These are design calculations for future planning, not estimates of what happened in the completed trial.

The illustrations are shown for planning at 100%, 75%, and 50% of the observed effect. The smaller the effect one wishes to detect, the more participants are generally needed to distinguish it from random variation under otherwise unchanged assumptions. For example, planning for 75% or 50% of the observed effect requires more participants per arm than planning for the full observed effect. The figures therefore illustrate how sensitive sample-size needs are to the assumed target effect.

These calculations are not observed power. They use the observed effect as a planning input and do not tell whether the trial was adequately designed, because the trial’s original target effect and assumptions are not reported. Nor do they replace interpretation of the confidence interval: interval width and the range of effects compatible with the analysis are more useful for describing the precision of the completed result.

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 = 0.140; pooled standard deviation σ = 0.971.
2. n = 2 × (1.96 + 0.842)² × 0.971² / 0.140² = 754.6, rounded up to 755 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 <- 0.14                      # observed difference in means
sigma <- 0.970670582283374                     # 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(2582242)
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 = 755 (80% power), 1,011 (90% power)
 75% of observed effect: n per arm = 1,342 (80% power), 1,796 (90% power)
 50% of observed effect: n per arm = 3,019 (80% power), 4,041 (90% power)
power.t.test: 755.6 
Simulated power at n = 755 per arm: 0.800 (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 analysis used a mixed model for repeated measurements and estimated a treatment difference at week 24 of -0.09. Its 95% confidence interval was -0.23 to 0.06, and the reported p-value was 0.260. This is the registry’s own reported analysis for the primary comparison.

A separate recalculation from the posted arm means and standard deviations gives an unadjusted difference of 0.14, with a 95% confidence interval from -0.042 to 0.322 and p = 0.132. That calculation compares the endpoint summaries without fitting the repeated-measures model. Its result can differ from a model-based analysis because the analytical methods and the information used are not the same. In addition, the registry does not state the sponsor model’s contrast coding in enough detail to account fully for the difference in signed estimates.

The recalculation is useful as a transparent check on the posted summary statistics, but it should not replace the sponsor’s analysis. The registry record does not indicate a discrepancy flag for this result. The two approaches both produce confidence intervals that include zero, but they are not identical estimates and should be kept distinct when describing the evidence.

The model’s covariates, covariance structure, and detailed missing-data implementation are not reported in the registry record. Consequently, the sponsor result cannot be independently reproduced from the information recorded. The most appropriate reading is to report the sponsor’s estimate as the primary analysis and label the summary-statistic calculation as an unadjusted recalculation.

15. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when several outcomes, contrasts, or statistical tests are considered. With more tests, the chance of observing at least one apparently unusual result can increase even when there is no corresponding underlying difference. Common approaches include prespecifying a testing hierarchy or adjusting significance thresholds, but the registry record does not report whether a multiplicity procedure was planned or applied.

The registry identifies a primary HbA1c outcome and lists 18 outcome measures, including 17 posted secondary outcomes. The primary endpoint therefore has a distinct role in interpretation, while the secondary outcomes provide additional context. Without a reported multiplicity strategy, secondary p-values and intervals should not be treated as though each were a separate confirmatory test with the same evidential status as the primary comparison.

The outcome description says missing data for the primary endpoint were imputed using last observation carried forward. In general, this approach carries a participant’s latest observed value forward when a later measurement is missing. The resulting analysis depends on assumptions about the unobserved outcome after the last recorded measurement. The registry record does not report the amount of missing data, the reasons for missingness, patterns by treatment group, or sensitivity analyses using alternative assumptions. It also does not provide enough detail to determine exactly how the reported repeated-measures model incorporated imputed values.

The full analysis set included randomized participants who were dosed and had any post-randomization data. The record does not report further details on how many randomized participants were excluded from this population or why. The posted primary outcome groups include 220 and 217 participants. These registry entries support describing the analysis population as recorded, but not assuming that it included every randomized participant or that missingness was identical across groups.

The registry record does not report interim analyses, the original sample-size justification, or the planned effect size underlying the trial’s design. Those omissions prevent reconstruction of the trial’s original statistical planning. They also mean that no claim should be made about whether interim monitoring or a particular approach to multiplicity was used.

16. 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: coronary artery disease (0.9%), diabetic neuropathy (0.9%), cerebral infarction (0.5%), diabetes mellitus inadequate control (0.5%), diabetic foot (0.5%).

The registry reports serious adverse events in 12 of 220 participants in the thrice-daily group and 9 of 217 in the twice-daily group. Other, non-serious adverse events were reported in 34 of 220 and 36 of 217 participants, respectively. There were 0 deaths in each group. These are participant counts reported for the groups at risk; they are not exposure-adjusted rates, and the registry record does not provide further detail on duration of exposure for these calculations.

The listed serious adverse-event terms include coronary artery disease and diabetic neuropathy, each reported as 0.9%, and cerebral infarction, diabetes mellitus inadequate control, diabetic foot, and diabetic retinopathy, each reported as 0.5%. The registry record does not give further information in the registry record about event timing, severity grading, relatedness assessments, or whether an individual could contribute to more than one term.

These figures describe what was recorded during the trial, not a definitive comparison of safety between schedules. The counts are limited in size, and the record does not report a statistical comparison for these adverse-event summaries. They should be read alongside the study duration, group sizes, and the limits of the registry’s safety detail. No inference about safety beyond these posted observations is warranted.

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

The primary result describes average HbA1c change over 24 weeks in the trial’s full analysis set. Mean change was -1.66 with thrice-daily dosing and -1.52 with twice-daily dosing. The sponsor’s mixed model estimated a treatment difference of -0.09, with a 95% confidence interval from -0.23 to 0.06 and p = 0.260. The interval includes zero, so the reported analysis does not establish a difference between schedules.

That does not establish that the schedules are equivalent. Equivalence requires a prespecified margin and an analysis designed to test whether the interval lies within that margin. The registry record does not report an equivalence or non-inferiority margin. A non-significant p-value is not proof of no effect; it indicates that the analysis did not provide strong evidence against the specified null hypothesis.

The result also does not mean that every participant’s HbA1c changed by the group mean. Individual responses vary, as reflected in the group standard deviations. The average contrast cannot describe the distribution of individual outcomes, and it cannot be used to predict an individual participant’s response. The study’s open-label design and the stated missing-data approach are additional parts of the context in which the result was generated.

The secondary outcomes and safety counts broaden the description of the trial but do not transform the primary result into a definitive conclusion across all measures. The registry does not report a multiplicity plan, detailed analyses for every secondary outcome, or extensive model details. Unadjusted secondary comparisons are therefore less conclusive than a clearly prespecified primary analysis.

In short, the posted results provide estimates with uncertainty for one randomized comparison over the study period. They do not demonstrate that one schedule is superior, that the schedules have identical effects, or that the trial’s findings apply to populations or follow-up periods not described in the registry record. The figures should be understood as statistical evidence from this particular trial, not as individual medical guidance.

18. Why This Trial Matters Statistically

This trial illustrates why interpreting a clinical comparison requires more than looking at the direction of two group means. Both mean HbA1c changes were negative, but the question of interest is the between-group contrast and its uncertainty. A difference in group summaries is not, by itself, evidence that the schedules differ. Confidence intervals and the method used to estimate the contrast are central to understanding what the data can support.

It also illustrates the distinction between a sponsor-reported model and an unadjusted recalculation from group summaries. The repeated-measures model and the recalculation use different approaches, and the registry does not describe the model in enough detail to reproduce it fully. Reporting the methods alongside the estimates helps readers distinguish the primary analysis from a simple summary-statistic check.

The trial’s endpoint and analysis population matter as well. The primary outcome is a continuous change score, and the stated full analysis set includes randomized participants who were dosed and had any post-randomization data. Missing data were handled using last observation carried forward according to the outcome description. Those definitions shape what the estimated result represents.

Finally, the many secondary outcomes create a reason to consider multiplicity and the weight assigned to exploratory results. The registry record does not report a multiplicity procedure or the original sample-size rationale. Recognizing these gaps prevents secondary findings and planning illustrations from being mistaken for confirmatory evidence or for the trial’s original design calculations.

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 registry record gives a concise account of the trial but leaves important methodological details unreported. It does not specify the random-sequence generation method, allocation concealment, site count, country list, detailed eligibility criteria, or the treatment-adjustment rules used in the treat-to-target design. These omissions limit assessment of how participants entered the trial and how closely the treatment procedures can be understood from the record alone.

The study was open label. This means treatment assignment was not masked, which can matter for outcomes or decisions influenced by participant or investigator knowledge. HbA1c is an objective laboratory measure, but the record does not describe blinding of laboratory personnel or other outcome-assessment procedures, so no assumptions about those details should be made.

The primary outcome description specifies last observation carried forward for missing data, but the amount and pattern of missingness are not reported. The record does not describe sensitivity analyses or provide enough detail to determine how the imputation related to the mixed model. Model covariates, covariance structure, and contrast coding are also unreported, limiting independent reproduction and explanation of the sponsor estimate.

The record lists many secondary outcomes, but it does not report a multiplicity strategy. The original sample-size justification, planned effect size, and interim analysis information are also not reported. As a result, planning illustrations based on observed effects cannot be used to infer the trial’s original design aims, and secondary findings should be interpreted with caution.

Finally, the primary follow-up is 24 weeks, and the posted adverse-event counts describe events recorded over the trial. The registry entries do not provide detailed exposure time, event attribution, or longer-term follow-up. The findings therefore describe the recorded study period and population rather than effects beyond them.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Comparing Efficacy and Safety of Thrice Daily Versus Twice Daily NovoMix® 30 (Biphasic Insulin Aspart 30) in Subjects With Type 2 Diabetes Inadequately Controlled With Basal Insulin — NCT02582242.
• Registry record last updated 2019-06-11; status last verified 2019-05.
• Results first posted 2018-05-29.
• 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

NCT02582242 is a completed phase 4 randomized, open-label, parallel-group trial comparing biphasic insulin aspart 30 given three times daily with the same treatment given twice daily. The official title describes a 24-week, multinational, multicentre, treat-to-target study in people with type 2 diabetes inadequately controlled with basal insulin. The trial enrolled 437 participants and was conducted in Asia.

The primary outcome was change in HbA1c from baseline to week 24 in the full analysis set. Posted mean changes were -1.66 and -1.52. The sponsor-reported mixed model estimated a treatment difference of -0.09, with a 95% confidence interval from -0.23 to 0.06 and p = 0.260. An unadjusted recalculation from the posted summaries gives a difference of 0.14, with a 95% confidence interval from -0.042 to 0.322.

Both intervals include zero. The findings do not establish a difference or equivalence between schedules. The registry record does not report several details needed for fuller assessment, including the original sample-size justification, multiplicity procedure, and detailed model specification.