Non-inferiority Study to Compare the Efficacy and Safety of Mylan's Insulin Glargine With Lantus® in Type 1 Diabetes Mellitus Patients (INSTRIDE 1): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Non-inferiority Study to Compare the Efficacy and Safety of Mylan's Insulin Glargine With Lantus® in Type 1 Diabetes Mellitus Patients (INSTRIDE 1) (NCT02227862), a randomized phase 3 study. The primary outcome is Change in HbA1c From Baseline to 24 Weeks.

2. Clinical Question

The registry record describes a comparison of Mylan's insulin glargine with Lantus® for people with type 1 diabetes. Both treatments were given once daily alongside mealtime insulin lispro. The trial's stated primary question was whether Mylan's insulin glargine was non-inferior to Lantus® based on change in HbA1c from baseline to 24 weeks.

HbA1c is a continuous measure of average blood glucose over time. The primary comparison therefore concerns average change, not the proportion of participants who reached a particular target. A non-inferiority question is not answered simply by finding no statistically significant difference: it requires a prespecified margin defining the largest acceptable disadvantage, along with an analysis that evaluates the result against that margin. Because the margin and its analysis are not reported in the registry record, the posted summary can describe the estimated contrast and its uncertainty but cannot, by itself, establish that the non-inferiority criterion was met.

3. Trial Design

This was a phase 3, randomized, open-label, parallel-group clinical trial sponsored by Mylan Inc. The listed condition was type 1 diabetes. The registry reports an enrollment of 558 and a completed study. The recorded start was 2014-06, and primary completion was 2016-07.

Participants were assigned to Mylan's insulin glargine or Lantus®. Both treatment descriptions include insulin lispro, and the brief summary specifies once-daily glargine in combination with mealtime lispro. In the posted primary outcome, the Mylan's insulin glargine arm has 280.0 participants and the Lantus® arm has 277.0. Those are the group sizes reported for the primary outcome; the record does not explain the difference between these analysis counts and the overall enrollment.

Randomization is intended to make treatment groups comparable, on average, with respect to factors that might affect outcomes. That is a general property of randomized designs, not evidence that every measured or unmeasured characteristic was balanced in this particular trial. The registry record does not report the randomization sequence, allocation concealment procedures, or baseline characteristics in the material summarized for the primary result.

The study was open-label, meaning masking was not used. Participants and study personnel could therefore know which treatment was assigned. HbA1c is a laboratory measure, but the registry record does not describe laboratory blinding, outcome-assessor blinding, or procedures used to limit the influence of treatment knowledge on other outcomes.

The primary endpoint was change in HbA1c from baseline to 24 weeks. Secondary outcomes include HbA1c summaries, fasting plasma glucose, self-monitored blood glucose profiles, insulin dose, hypoglycemic events, local and systemic reactions, and insulin antibody binding. The record lists 13 outcome measures and 12 posted secondary outcomes. The sponsor's original sample-size rationale, detailed statistical analysis plan, analysis populations, and rules for handling missing outcome data are not reported in the registry record.

4. Participant Flow and Accounting

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

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

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

6. Primary Endpoint

The registry lists the primary outcome as Change in HbA1c From Baseline to 24 Weeks, measured over 24 weeks.

The primary endpoint is change in HbA1c from baseline to 24 weeks, measured in percent. It is reported as a least-squares mean for each treatment arm, with a standard error. HbA1c is a continuous laboratory measure used to summarize average blood glucose over time. Examining change from baseline focuses the comparison on how the measure moved during the study rather than only on its value at the endpoint.

A continuous measure preserves information about the size and direction of changes across participants. It can therefore provide a comparison of average change without converting the result into a yes-or-no outcome based on an unreported threshold. The least-squares means are model-based means; the registry record does not describe the model or its covariates, so the exact adjustment represented by those means is not reported in the registry record.

The posted mean changes are 0.14 for Mylan's insulin glargine and 0.11 for Lantus®. The comparison is expressed as Lantus® relative to Mylan's insulin glargine, with an estimated mean difference of -0.03. This indicates a numerically lower mean change in the Lantus® arm under that contrast convention. It does not alone establish a meaningful treatment difference or answer the non-inferiority question. Non-inferiority depends on the size and direction of the confidence interval relative to a prespecified margin, and that margin is not reported in the registry record.

7. Results Posted in the Registry

• Lantus®: mean 0.11 versus 0.14 on control; difference -0.03 (95% CI -0.18 to 0.12), standardized effect d = -0.03; z-test p = 0.694.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is continuous: each participant's HbA1c change can take a range of values rather than falling into only an event or no-event category. The registry reports a least-squares mean and standard error for each arm. A mean describes the central average change, while a standard error describes uncertainty in the estimated mean. It is not the same as the spread of individual participant outcomes. The registry record does not describe the model that produced the least-squares means, so its covariates and other modeling details are not reported in the registry record.

What does the mean difference represent?

The posted comparison uses Lantus® minus Mylan's insulin glargine. Its estimate is -0.03. This is the estimated difference between the arm means under that direction of subtraction. A negative estimate means the reported mean change is numerically lower in the Lantus® group than in the Mylan's insulin glargine group. The estimate concerns an average; it does not imply that all participants experienced a change in the same direction or by the same amount.

How should the standard error be understood?

Each arm's standard error is 0.054. In general, a standard error describes the precision of an estimated mean: smaller standard errors indicate less sampling uncertainty in the mean estimate, all else being equal. The standard error is not a measure of how much individual participants differ from one another. No standard deviations for the outcome are reported with the arm means. The pooled standard deviation used in the recalculation is approximate, and its value is not needed to state the primary contrast.

What does the confidence interval add?

A confidence interval gives a range of estimates compatible with the observed data under the calculation's assumptions. It is more informative than the point estimate alone because it makes uncertainty visible. The posted interval for the contrast includes zero and includes negative and positive values. Zero represents no difference in mean change under the stated contrast. The interval therefore does not isolate a single direction of difference. In general, a confidence interval is not a range that guarantees where the true value lies, and it does not account for design or analysis problems that are not represented in the calculation.

What does the p-value test?

The posted unadjusted p-value is 0.694. In general, a p-value describes how unusual data at least as discrepant as those observed would be under a specified null hypothesis and statistical model. It is not the probability that the null hypothesis is true, and it is not the probability that the finding occurred by chance. The registry record does not state a sponsor hypothesis-testing procedure for this posted comparison. The unadjusted value should not be treated as a test of the non-inferiority objective, because that objective requires a specified margin and an appropriate comparison with that margin.

Why is a non-inferiority margin necessary?

For a non-inferiority study, investigators define in advance how much worse the experimental treatment could be, at most, while still meeting the study's criterion. This boundary is the non-inferiority margin. The confidence interval is then assessed against the boundary in the direction that represents disadvantage. A result can have a non-significant test for difference and still fail to establish non-inferiority; conversely, the relevant conclusion is not determined solely by whether the interval includes zero. The margin and the sponsor's analysis are not reported in the registry record, so this criterion cannot be evaluated from the posted summary.

What does recalculation mean?

The posted contrast and interval are labeled as recalculated, meaning they are unadjusted calculations from the summary statistics rather than the sponsor's own analysis. They use the reported means and standard errors and a normal approximation. This approach is useful for a transparent, approximate comparison when detailed model output is unavailable. It does not reproduce an analysis that might account for baseline values, stratification, repeated measures, or other prespecified features. Those analysis details are not reported in the registry record. The sponsor's original analysis is not represented by this recalculation.

Why can over-dispersion matter in general?

Over-dispersion is a general concern when the variability in data is greater than a statistical model assumes. If a model underestimates variability, its standard errors can be too small and its confidence intervals too narrow. If variability is greater than expected, an analysis designed for a simpler variance structure may be less reliable. The primary endpoint is continuous, and the registry record does not report an over-dispersion assessment or a variance model for the sponsor analysis. Over-dispersion should therefore be understood as a general statistical consideration, not as a reported problem in this trial.

How should secondary outcomes be distinguished from the primary endpoint?

The registry lists outcomes involving fasting plasma glucose, self-monitored blood glucose, insulin dose, hypoglycemic events, local and systemic reactions, and insulin antibody binding, among other HbA1c summaries. These outcomes can describe other aspects of the trial, but they do not replace the stated primary endpoint. The record lists 13 outcome measures and 12 posted secondary outcomes. It does not report a multiplicity adjustment or a hierarchy for interpreting those outcomes. Without such details, secondary findings should be treated as additional descriptive or exploratory evidence rather than as independent confirmation of the primary result.

10. Confidence Intervals and Precision

The estimate of the mean difference is -0.03, with a confidence interval that runs from a negative to a positive value and includes zero. This is the central precision message: the point estimate is close to zero, but the uncertainty allows for contrasts in either direction. The posted interval should be considered alongside the point estimate rather than treated as a secondary detail.

A confidence interval describes uncertainty in an estimated contrast under the calculation's assumptions. It does not describe the full range of individual participant outcomes, and it is not a guarantee that the true effect lies within its limits. Its interpretation also depends on the statistical model and analysis population. The registry record does not describe the sponsor's model, its handling of missing data, or the population used for the primary analysis.

The standard errors for both reported arm means are 0.054. These quantify sampling uncertainty in the arm means, not individual variability. The recalculated contrast uses an approximate pooled standard deviation, so it should be read as an unadjusted illustration rather than a reconstruction of the sponsor's full analysis. The unadjusted p-value of 0.694 is consistent with the interval's inclusion of zero, but neither statistic establishes equivalence.

For the trial's non-inferiority objective, precision must be judged relative to the prespecified margin. A narrow interval can still be inconclusive if it crosses that margin, while an interval can support non-inferiority only if it falls within the acceptable boundary in the relevant direction. Since the margin is not reported in the registry record, the interval's adequacy for that purpose cannot be determined from the posted result.

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 figures are illustrative calculations based on the observed effect, not the trial's original design assumptions. For this continuous-outcome comparison, the listed figures are participants per arm, not total participants or total events. The assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The target effect is set at the stated fraction of the observed effect. The trial's own original sample-size assumptions and justification are not reported in the registry record.

The 80% power column gives the estimated participants per arm for a hypothetical study to have an 80% chance of detecting the specified effect under those planning assumptions, if that effect and the assumptions were correct. The 90% power column gives the corresponding planning count for a 90% chance. Power is a design calculation, not a measure of how likely the trial's result is true. These figures are not observed power.

The rows consider planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. The planning counts rise as the effect being targeted becomes smaller: a smaller signal is harder to distinguish from random variation, so more participants are needed under the same assumptions. The listed counts are very large relative to the group sizes reported for the primary outcome, but they should not be used to infer that the trial was underpowered for its original non-inferiority objective. The illustration targets fractions of an observed effect and does not use a non-inferiority margin.

These calculations assume a simple normal-approximation comparison and equal allocation. The registry record does not report whether the trial's original design used those assumptions, or whether its own planning considered other design features. The planning values are best read as sensitivity illustrations about detecting the selected effect sizes, not as a retrospective verdict on the trial design.

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 = -0.030; pooled standard deviation σ = 0.901.
2. n = 2 × (1.96 + 0.842)² × 0.901² / -0.030² = 14,165.1, rounded up to 14,166 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 <- -0.03                      # observed difference in means
sigma <- 0.901182358484254                     # 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(2227862)
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 = 14,166 (80% power), 18,964 (90% power)
 75% of observed effect: n per arm = 25,183 (80% power), 33,713 (90% power)
 50% of observed effect: n per arm = 56,661 (80% power), 75,853 (90% power)
power.t.test: 14166.1 
Simulated power at n = 14,166 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. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary HbA1c endpoint and 12 posted secondary outcomes. Secondary measures include additional HbA1c summaries, fasting plasma glucose, self-monitored blood glucose profiles, insulin dose, hypoglycemic events, reactions, and insulin antibody binding. The record does not report a multiplicity adjustment, a testing hierarchy, or rules for deciding which secondary outcomes were confirmatory. When many outcomes are examined, the chance of at least one apparently notable result can increase. This is a general statistical issue, not evidence that any particular finding in this trial is spurious.

The record lists zero analysis records. It therefore does not provide a detailed account of the sponsor's analyses, including the model for the primary endpoint, the statistical testing procedure, or whether analyses were adjusted for baseline or other factors. The primary values are reported as least-squares means, but the model details are not reported in the registry record. The posted comparison is an unadjusted recalculation from summary statistics and does not substitute for a full prespecified analysis.

The analysis population for the primary result is not reported in the registry record. The record does not say whether the reported means include all randomized participants, participants who received treatment, participants with an evaluable endpoint, or another defined group. It also does not report how participants who discontinued treatment or did not provide a 24-week measurement were handled.

Missing data can affect an estimate when the probability of having an observed outcome is related to a participant's outcome or treatment experience. General approaches include explicit assumptions about the missingness process and sensitivity analyses to examine how conclusions change under alternative assumptions. The registry record does not report the amount of missing primary-outcome data, the method used to address it, or sensitivity analyses. These gaps limit assessment of how robust the posted mean difference is to incomplete observations.

Because the non-inferiority margin, sponsor's primary analysis, analysis population, multiplicity strategy, and missing-data procedures are not reported in the registry record, the posted summary supports a descriptive reading of the estimate and its interval, but not a complete audit of the planned inferential framework.

14. 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 1% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hypoglycaemia (3.6%), generalized tonic-clonic seizure (0.7%), goitre (0.7%), acute kidney injury (0.7%), hypertension (0.4%).

The serious adverse event figures are counts reported alongside the number at risk. For Mylan's insulin glargine, 18 serious adverse events are reported among 280 participants at risk; for Lantus®, 22 are reported among 278 participants at risk. The registry also reports 2 deaths in the Mylan's insulin glargine group and 1 in the Lantus® group. These figures are descriptive counts, not proof of a treatment-related difference. No comparison test or adjusted analysis of these safety figures is reported in the registry record.

The record separately lists other adverse events among 225 participants at risk in the Mylan's insulin glargine group and 239 in the Lantus® group, with 225 and 239 affected, respectively. Those denominators differ from the serious-event denominators. The record does not explain the reason for the different risk counts or provide enough detail to infer exposure time or the rules used to count events and affected participants.

Serious-event terms include hypoglycaemia, generalized tonic-clonic seizure, goitre, acute kidney injury, hypertension, and malignant oligodendroglioma, with percentages listed for each term. The record does not assign these listed percentages to treatment arms in the summarized information. Counts and terms should therefore be read as reported safety descriptions, not as causal findings or comparative risk estimates.

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

The posted primary result describes average HbA1c change over 24 weeks in two randomized treatment groups, each receiving glargine with mealtime insulin lispro. The arm means are 0.14 for Mylan's insulin glargine and 0.11 for Lantus®. The estimated contrast, Lantus® minus Mylan's insulin glargine, is -0.03. The confidence interval includes zero and extends in both directions, indicating that the estimated difference is uncertain.

These data do not establish that the treatments are equivalent. They also do not establish that one treatment is superior, or that no difference exists. A non-inferiority conclusion requires a prespecified margin and an analysis testing the estimate against that boundary. Neither the margin nor the sponsor's inferential analysis is reported in the registry record. The unadjusted p-value of 0.694 is not a substitute for that evaluation.

The result concerns group averages. It does not describe how many individual participants had an increase or decrease, nor how responses varied across individuals. The registry record does not report the distribution of individual changes or the analysis population and missing-data handling needed to assess the completeness of the endpoint comparison.

The safety counts likewise describe events recorded in the trial; they do not establish that a treatment caused an event or that the groups differ in safety. Differences in the number at risk, event definitions, follow-up exposure, and attribution can matter in interpreting safety comparisons. Details of these features and a statistical comparison are not reported in the registry record.

The results apply to the study's described setting: an open-label randomized comparison in people with type 1 diabetes, with both glargine regimens used alongside insulin lispro. The registry summary does not establish how the findings apply to other populations, treatment settings, or regimens. The most defensible interpretation is therefore limited: the posted summary gives a small estimated difference in mean HbA1c change, with uncertainty spanning both directions, while leaving the trial's formal non-inferiority conclusion unresolved.

16. Why This Trial Matters Statistically

This trial illustrates why a non-inferiority question requires more than a conventional test for a difference. The primary estimate is -0.03 and its confidence interval includes zero. That pattern can look like little apparent separation between group means, but it does not by itself demonstrate that the treatments are close enough to satisfy a non-inferiority standard. The clinically acceptable boundary must be set in advance and evaluated directly.

It also illustrates the distinction between a point estimate and its precision. The observed mean changes, 0.14 and 0.11, are close numerically, yet the interval includes plausible contrasts in either direction. Readers need both the estimated difference and its uncertainty to understand what the result can support. A p-value of 0.694 alone would not convey the range of effects compatible with the data.

The registry record provides enough summary information for an approximate unadjusted recalculation, but not the sponsor's full analysis framework. Least-squares means are reported without the model details, and the analysis population and missing-data procedures are not reported. The recalculation can aid statistical interpretation while remaining distinct from a prespecified trial analysis.

Finally, planning illustrations based on fractions of the observed effect are not equivalent to the original design justification for a non-inferiority trial. They address hypothetical detection of selected effect sizes under stated assumptions, whereas a non-inferiority design depends on its margin and analysis plan. Keeping those questions separate prevents a planning illustration from being mistaken for evidence about the trial's actual design adequacy.

17. 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 non-inferiority margin, so the primary objective cannot be evaluated against its defining criterion. It also does not report the sponsor's primary inferential analysis. The posted comparison is an unadjusted recalculation from summary statistics, and the pooled standard deviation used for that calculation is approximate.

The model behind the least-squares means is not described. The record does not identify the primary analysis population, explain the difference between the overall enrollment and the group sizes used for the primary outcome, or report how missing 24-week HbA1c measurements were handled. The extent of missing primary-outcome data and any sensitivity analyses are not reported in the registry record.

The study was open-label. The registry does not describe whether laboratory personnel or outcome assessors were masked, or how awareness of treatment assignment was managed in outcome assessment. Randomization is reported, but allocation concealment and baseline characteristics are not detailed in the summarized record.

For secondary outcomes, the record does not report a multiplicity strategy or testing hierarchy. For safety, it does not provide a comparative statistical analysis, event-exposure information, or the details needed to interpret the different numbers at risk across outcome groupings. The listed serious-event percentages are not assigned to treatment arms in the summarized information.

These reporting gaps do not show that the trial's analyses were inappropriate; they limit what can be concluded from the registry summary. In particular, they prevent a complete assessment of non-inferiority, the sensitivity of the result to missing data, and comparative safety.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Non-inferiority Study to Compare the Efficacy and Safety of Mylan's Insulin Glargine With Lantus® in Type 1 Diabetes Mellitus Patients (INSTRIDE 1) — NCT02227862.
• Registry record last updated 2022-03-03; status last verified 2022-03.
• Results first posted 2020-03-09.
• 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.

21. Record Summary

NCT02227862 is a completed phase 3 randomized, open-label, parallel-group trial sponsored by Mylan Inc. It compared Mylan's insulin glargine once daily with Lantus® once daily in people with type 1 diabetes, with insulin lispro used at mealtimes. The registry reports enrollment of 558, a study start of 2014-06, and primary completion in 2016-07.

The primary endpoint was change in HbA1c from baseline to 24 weeks. The reported least-squares mean changes were 0.14 and 0.11, respectively, with an estimated Lantus®-minus-Mylan difference of -0.03 and an unadjusted p-value of 0.694. The confidence interval includes zero and values in either direction. The non-inferiority margin and sponsor's primary analysis are not reported in the registry record, so the posted summary alone does not establish the formal non-inferiority conclusion.

Safety counts and multiple secondary outcomes are also listed, but the registry summary does not report a complete comparative safety analysis, analysis populations, or missing-data procedures.