A Study of LY900014 Compared to Insulin Lispro (Humalog) in Adults With Type 1 Diabetes: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study of LY900014 Compared to Insulin Lispro (Humalog) in Adults With Type 1 Diabetes (NCT03952130), a randomized phase 3 study. The primary outcome is Change From Baseline in Hemoglobin A1c (HbA1c).

2. Clinical Question

HbA1c is used as a measure of average plasma glucose over a prolonged period. The primary result is a comparison of modeled changes, rather than a comparison of the groups’ final HbA1c values alone. Both groups received mealtime insulin alongside one of two basal insulin options, so the comparison concerns the mealtime treatments in that treatment context. The registry describes this as a Phase 3, randomized, double-blind study with parallel groups. Its stated purpose included assessing safety and effectiveness. The reported primary result addresses HbA1c change; it does not by itself answer every safety or effectiveness question.

3. Trial Design

The study is registered as NCT03952130 and is titled “A Study of LY900014 Compared to Insulin Lispro (Humalog) in Adults With Type 1 Diabetes.” The official title describes a prospective, randomized, double-blind comparison of LY900014 and insulin lispro, each combined with insulin glargine or insulin degludec, in adults with type 1 diabetes. The registry lists the phase as Phase 3, the allocation as randomized, the masking as double, and the model as parallel. Enrollment is listed as 354. The recorded study status is completed.

The intervention groups received mealtime LY900014 or mealtime insulin lispro. Each was administered subcutaneously 0–2 minutes before meals, with once-daily subcutaneous basal insulin glargine or insulin degludec. The record does not specify in the summarized information how participants were assigned between the two basal insulin options or whether those options were balanced between groups. The study start is listed as 2019-05-29 and primary completion as 2022-01-10.

The primary endpoint was change from baseline in HbA1c, with the timeframe recorded as baseline and Week 26. The primary analysis population was randomized participants with a baseline and at least one post-baseline HbA1c measurement. The registry reports least-squares means from a mixed model repeated measures analysis and classifies the comparison as non-inferiority. Its original sample-size justification, the non-inferiority margin, details of any interim analyses, and the full statistical analysis plan are not reported in the registry record. Those missing details matter when interpreting the intended design and the formal basis for a non-inferiority conclusion.

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: All randomized participants with baseline and at least one post-baseline HbA1c 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: All randomized participants.

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Hemoglobin A1c (HbA1c), measured over Baseline, Week 26.

The primary endpoint is change from baseline in hemoglobin A1c (HbA1c), reported as a percentage of HbA1c, with measurements at baseline and Week 26. HbA1c is the glycosylated fraction of hemoglobin A. It is used to identify average plasma glucose concentration over prolonged periods, rather than the glucose level at a single moment. A change endpoint focuses on how the measure moved relative to each participant’s baseline, while the group comparison estimates whether that movement differed between treatments.

The registry reports least-squares means from a mixed model repeated measures model. Such means are model-based estimates, not simply the arithmetic averages of the observed values. The listed model includes baseline HbA1c and several design or treatment terms, including treatment, time, and their interaction. The record’s analysis population requires baseline data and at least one post-baseline HbA1c value; it is not described as every randomized participant regardless of data availability.

HbA1c is useful for this question because it reflects glucose exposure over a prolonged period, matching the study’s interest in longer-term glycemic control. It does not capture every dimension of glucose patterns or safety, and the primary endpoint alone cannot establish conclusions about the secondary outcomes. The registry’s primary result is classified as non-inferiority, but the margin and the decision rule are not reported in the registry record. Consequently, the numerical change and its uncertainty can be described, while the formal non-inferiority determination cannot be independently assessed from these details alone.

7. Results Posted in the Registry

• LY900014: mean -0.21 versus -0.28 on control; difference 0.07 (95% CI -0.11 to 0.25), standardized effect d = 0.09; z-test p = 0.436.

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.

1-hour Postprandial Glucose (PPG) Excursion During Mixed-Meal Tolerance Test (MMTT)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • LY900014: mean 85.00 versus 102.80 on control; difference -17.80 (95% CI -29.40 to -6.20), standardized effect d = -0.35; z-test p = 0.003.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for LY900014 than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

2-hour PPG Excursion During MMTT

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • LY900014: mean 116.70 versus 142.20 on control; difference -25.50 (95% CI -40.91 to -10.09), standardized effect d = -0.38; z-test p = 0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for LY900014 than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Rate of Severe Hypoglycemia

• Endpoint type: Event rate: events per unit of follow-up, compared as a rate ratio.
• Test: Poisson rate ratio, based on the estimated number of events and follow-up time.
• Result:
  • LY900014: 5.87 versus 6.88 (Events per 100 participant years) on control; rate ratio 0.85 (95% CI 0.26–2.76, approximate, Poisson model); p ≈ 0.791.
• Interpretation:
  • The 95% confidence interval includes no difference (a ratio of 1), so these data do not show a statistically significant difference between LY900014 and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes span post-meal glucose excursions, hypoglycemia, glucose-related measures, daily insulin dose, and HbA1c thresholds. Across the posted results, the two mixed-meal postprandial glucose comparisons favor lower excursions in the LY900014 group, while the severe-hypoglycemia rate comparison is uncertain. Taken together, these outcomes add context about glucose patterns and safety, but they do not replace the primary HbA1c comparison.

The secondary findings address distinct questions. A meal-test excursion reflects a response during a specified test, whereas severe hypoglycemia is an event-rate outcome over the stated study period. Other secondary measures have their own scales, timeframes, and data availability. It would be inappropriate to treat a direction of effect on one measure as proof of a corresponding effect on every other measure. Nor does the overall pattern resolve the primary non-inferiority question, because the non-inferiority margin is not reported in the registry record.

The registry lists 27 analyses across nine outcome measures and posts eight secondary outcomes. When many outcomes or comparisons are examined, some results can appear statistically notable by chance. A multiplicity adjustment, a testing hierarchy, or a clear distinction between confirmatory and exploratory secondary tests is not reported in the registry record. Accordingly, individual unadjusted secondary results carry less evidential weight than a prespecified primary result and should be interpreted as supportive or exploratory rather than as a collection of independent confirmations.

The analyses also do not necessarily use the same participants or methods. The posted numbers contributing to each endpoint vary, and the severe-hypoglycemia outcome is reported as a rate rather than a mean. The registry record does not provide detailed missing-data handling or analysis-population definitions for every secondary result. This means the pattern is useful for understanding the range of outcomes examined, while remaining limited by uncertainty, multiple testing, and outcome-specific data availability.

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

• 1-hour Postprandial Glucose (PPG) Excursion During Mixed-Meal Tolerance Test (MMTT)
• 2-hour PPG Excursion During MMTT
• Rate of Severe Hypoglycemia
• Rate of Documented Symptomatic Post Meal Hypoglycemia
• Change From Baseline in 1,5-Anhydroglucitol (1,5-AG)
• Change From Baseline in 10-Point Self-Monitoring Blood Glucose (SMBG) Values
• Change From Baseline in Daily Insulin Dose
• Percentage of Participants With HbA1c <7% and ≤6.5%

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is a continuous measure: change in HbA1c. The registry reports a least-squares mean for each treatment group and a least-squares mean difference between groups. A least-squares mean is an estimate from a statistical model that accounts for the terms included in that model. It can differ from a raw arithmetic average because it reflects the model specification and the available data. The registry identifies a mixed model repeated measures analysis and lists baseline, pooled country, study-entry bolus stratum, basal-insulin stratum, treatment, time, and treatment-by-time terms.

What does the treatment difference represent?

The reported difference of 0.07 compares the modeled change for LY900014 with the modeled change for insulin lispro. Positive values mean the estimated change is higher in the LY900014 group under that subtraction order; negative values mean it is lower. Since the endpoint is change, the negative group estimates indicate decreases from baseline. The difference is not a percentage change between treatments, nor is it a comparison of final values alone.

In general, a confidence interval describes a range of estimates compatible with the data and the analysis assumptions at the stated confidence level. It conveys uncertainty around the estimate; it does not say that a particular proportion of participants had outcomes within that range. The sponsor-reported interval for the primary result runs from -0.11 to 0.24. Because it spans zero, it is compatible with a range of directions for the treatment difference. For a non-inferiority question, interpretation also depends on the prespecified margin. That margin is not reported in the registry record, so the interval cannot be compared with it.

Why is the registry analysis different from a simple recalculation?

The recalculation uses the posted group estimates and standard errors to approximate an unadjusted comparison. Its interval and p-value are not the same kind of result as an analysis from the mixed model. Model-based adjustment, the repeated-measures structure, and the analysis population all affect the sponsor’s estimate and uncertainty. The recalculation is useful as a transparent check on the direction and rough scale of the result, but it does not reproduce the complete sponsor analysis. The registry record does not report a discrepancy between the recalculation and the reported primary result.

The model description identifies terms used in the primary analysis, but it does not supply the full model specification or all implementation details. For example, the registry record does not report the covariance structure, detailed missing-data assumptions, or sensitivity analyses in the information summarized for this analysis. Those omissions limit the ability to reproduce or fully audit the model from the registry description alone.

How should the secondary analyses be read?

The registry posts secondary findings on one-hour and two-hour postprandial glucose excursions during a mixed-meal tolerance test, severe and symptomatic post-meal hypoglycemia, 1,5-anhydroglucitol, self-monitoring blood glucose values, daily insulin dose, and HbA1c thresholds. Different outcomes use different scales and analysis methods. The postprandial glucose outcomes are continuous modeled comparisons; severe hypoglycemia is expressed as an event rate per participant-time. These are not interchangeable measures, and a result on one does not automatically establish a result on another.

In general, multiplicity arises when several hypotheses are tested: the chance of at least one apparently notable result can increase across a collection of tests. The registry lists nine outcome measures, eight posted secondary outcomes, and 27 analyses. A multiplicity adjustment or a prespecified hierarchy for interpreting these analyses is not reported in the registry record. The posted secondary p-values should therefore not be treated as though each were the sole confirmatory test. They are informative as part of the overall pattern, but carry less weight than a primary analysis defined in advance.

In general, over-dispersion means that outcome variability exceeds what a simple model assumes. For event-rate data, this can make a basic Poisson model’s uncertainty too narrow if the model does not account for extra variation. The registry’s recalculated severe-hypoglycemia comparison uses a Poisson approximation, while the reported analysis uses an empirical method. Those methods need not give identical intervals or p-values. The posted event-rate estimates are descriptive of the recorded timeframe; the small event totals and broad reported interval make the relative-rate estimate imprecise.

Across outcomes, interpretation also depends on which participants contributed data, how missing observations were handled, and how repeated measurements were modeled. The registry reports the primary analysis population and the primary model description, but detailed missing-data procedures and the analysis populations for every secondary outcome are not reported in the registry record. The estimates should be understood within those limits rather than as complete information about every randomized participant or every time point.

11. Confidence Intervals and Precision

The primary comparison has a sponsor-reported estimate of 0.07 and a 95% confidence interval from -0.11 to 0.24. This interval communicates uncertainty in the estimated difference in HbA1c change. It includes zero, as well as negative and positive differences, so the data do not locate the group difference precisely enough to establish its direction from the interval alone. Its interpretation for non-inferiority additionally requires the margin, which is not reported in the registry record.

The recalculated interval is -0.106 to 0.246. It is close to the reported interval, but comes from an unadjusted calculation using posted summary statistics and an approximate pooled standard deviation. It should not be mistaken for the model-based sponsor result. In general, a narrower interval indicates greater precision and a wider interval indicates more uncertainty, given the analysis assumptions. Precision is distinct from whether an estimate is clinically important or whether a formal decision threshold has been met.

For the postprandial glucose outcomes, the reported intervals are entirely below zero for the LY900014-minus-insulin-lispro differences, while the severe-hypoglycemia relative-rate interval is broad and includes values on both sides of no rate difference. These patterns describe the precision of those particular estimates; they do not resolve the primary non-inferiority question or establish an overall benefit-risk conclusion. The interval for the primary endpoint and the different intervals for secondary endpoints concern different outcomes and should not be combined as if they measured the same effect.

In general, confidence intervals rely on the statistical model and the data included in the analysis. The primary analysis used a repeated-measures model, and its result pertains to participants meeting the registry’s stated analysis-population definition. The record does not provide all model details or a full account of missing-data sensitivity analyses. As a result, the intervals are best read as uncertainty under the reported analyses, not as a guarantee that all sources of uncertainty have been captured.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The sample-size figures are planning illustrations based on the observed effect, not the trial’s original design calculations. For continuous outcomes, the entries are participants per arm. For time-to-event calculations, the entries are total events; the severe-hypoglycemia illustration is based on an event-rate comparison. The assumptions listed for these illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect.

The 80% power column gives the planned participants per arm, or total events where applicable, for a design intended to have an 80% chance of detecting the specified effect under those assumptions if that effect is real. The 90% column gives the corresponding planning amount for a 90% chance. These are hypothetical design quantities, not observed trial outcomes. This is not observed power.

Rows for 100% of the observed effect use the full estimated effect as the planning target; rows for 75% and 50% use smaller targets. Smaller effects are harder to distinguish from random variation, so they require more participants or events to reach the same planned power under the stated assumptions. The observed effect is itself uncertain, particularly when its confidence interval is broad, so these illustrations should not be read as precise forecasts for a future study.

The original sample-size assumptions and justification are not reported in the registry record. The planning figures therefore do not establish what assumptions guided the completed trial or whether the trial was designed around the observed effect. They are comparisons for considering hypothetical future designs, not evidence about the probability that the completed study’s result is true.

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 = 0.070; pooled standard deviation σ = 0.811.
2. n = 2 × (1.96 + 0.842)² × 0.811² / 0.070² = 2,105.2, rounded up to 2,106 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 <- 0.07                      # observed difference in means
sigma <- 0.810643129836192                     # 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(3952130)
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 = 2,106 (80% power), 2,819 (90% power)
 75% of observed effect: n per arm = 3,743 (80% power), 5,011 (90% power)
 50% of observed effect: n per arm = 8,421 (80% power), 11,274 (90% power)
power.t.test: 2106.2 
Simulated power at n = 2,106 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.

14. Analyses Reported by the Sponsor

The registry reports the primary comparison as a least-squares mean difference of 0.07, with a 95% confidence interval from -0.11 to 0.24, and classifies it as non-inferiority. The sponsor-reported primary p-value is not reported in the registry record. The summary also gives group least-squares mean changes of -0.28 for insulin lispro and -0.21 for LY900014.

The recalculated estimate is 0.07, with a 95% interval from -0.106 to 0.246 and an unadjusted p-value of 0.436. This is a simple calculation from the posted group summaries, not a reconstruction of the sponsor’s mixed model. It uses an approximate pooled standard deviation and does not reproduce the full adjustment or repeated-measures analysis. The small differences between the reported and recalculated interval limits reflect that distinction and their rounding, rather than a reported discrepancy.

The registry record does not flag a discrepancy. The reported analysis remains the appropriate result to cite for the registry’s specified primary analysis, while the recalculation is an auxiliary check on the summary data. The formal non-inferiority interpretation cannot be independently evaluated without the margin and decision rule, which are not reported in the registry record. The recalculated p-value should not be substituted for an unreported sponsor p-value or treated as the basis for a non-inferiority decision.

15. Multiplicity, Missing Data and Analysis Populations

The primary analysis population is described as all randomized participants with baseline HbA1c data and at least one post-baseline HbA1c result. That definition is narrower than an analysis that includes every randomized participant regardless of whether follow-up endpoint data are available. The registry reports a mixed model repeated measures analysis, but the detailed missing-data assumptions, imputation procedures, and sensitivity analyses are not reported in the registry record.

The record lists 27 analyses and nine outcome measures, with eight secondary outcomes posted. Testing several endpoints or contrasts creates a multiplicity issue: without an adjustment or prespecified testing strategy, the chance of at least one nominally notable result across the set may be greater than for a single test. A multiplicity adjustment, hierarchy, or other strategy for controlling interpretation across the secondary analyses is not reported in the registry record. The secondary p-values should therefore be read as individual, unadjusted evidence rather than as a family of confirmatory tests.

For the primary result, the registry identifies the analysis as non-inferiority, but the margin and decision rule are not reported. The original sample-size justification and the registry’s detailed statistical analysis plan are also not reported. These omissions make it difficult to determine the intended threshold for the primary conclusion, how the study was planned to address missing outcomes, and whether all analyses followed a prespecified plan.

These limits do not invalidate the estimates that are posted. They define what can be concluded from the registry record: the reported model-based primary comparison and the descriptive secondary findings can be discussed, but the full robustness of the analyses, the handling of missing data, and the control of false-positive findings across outcomes cannot be independently assessed from the information reported.

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: hypoglycaemia (2.2%), abortion missed (1.3%), haemorrhagic ovarian cyst (1.3%), arteriosclerosis coronary artery (0.6%), abdominal pain (0.6%).

The registry reports serious adverse events for the insulin lispro group among 178 participants at risk and for the LY900014 group among 176 participants at risk. Serious adverse events affected 12 participants in the insulin lispro group and 13 in the LY900014 group. Other adverse events affected 29 participants among 178 at risk in the insulin lispro group and 28 among 176 at risk in the LY900014 group. The registry reports zero deaths in each group.

These are counts of participants affected, not event rates adjusted for time at risk, and the counts alone do not establish whether an event was caused by treatment. The listed serious-event terms include hypoglycaemia at 2.2%, abortion missed at 1.3%, haemorrhagic ovarian cyst at 1.3%, arteriosclerosis coronary artery at 0.6%, abdominal pain at 0.6%, and clavicle fracture at 0.6%. The record excerpt does not give enough detail to interpret these terms by treatment group or to assess their clinical context.

Safety comparisons require attention to definitions, exposure duration, event severity, and the number of people monitored. The figures describe what the registry posted; they do not establish equivalence of safety profiles or rule out uncommon harms. The registry record does not report further detail about event relatedness, timing, or treatment discontinuation due to adverse events.

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

The primary estimate describes a small positive difference in modeled HbA1c change for LY900014 relative to insulin lispro, with a confidence interval that includes both negative and positive values. This is an estimate with uncertainty, not a direct statement about every participant. The result does not establish that the treatments have identical effects, and the interval alone does not determine whether a non-inferiority criterion was met.

The registry labels the primary analysis as non-inferiority, but the margin and decision rule needed to assess that claim are not reported in the registry record. The posted result can therefore be described accurately as a reported non-inferiority comparison with a difference estimate and confidence interval; a formal conclusion about meeting the criterion cannot be independently drawn from the available details. The registry also does not report the primary sponsor p-value.

The secondary results broaden the picture by addressing post-meal glucose measures and hypoglycemia as well as other glycemic outcomes. They do not all measure the same aspect of treatment, and multiple unadjusted tests make individual secondary findings less definitive. Safety counts likewise describe recorded events rather than proving causation or excluding rare harms. A result on one outcome should not be generalized to every outcome or to longer periods than the recorded timeframe.

In general, statistical significance, clinical importance, and a decision about treatment are different questions. Statistical estimates describe patterns in the analyzed data under assumptions; clinical importance requires context beyond a p-value or a point estimate. This trial’s registry results support a careful account of the Week 26 comparison and its uncertainty, but do not alone settle broader questions about long-term effects, individual outcomes, or treatment choice. No medical recommendation follows from these registry statistics.

18. Why This Trial Matters Statistically

This trial illustrates why a point estimate should be read together with its confidence interval and the question the analysis was designed to answer. A result labeled non-inferiority cannot be judged from the label alone: the margin and decision rule are essential to interpreting whether the estimate is sufficiently bounded. Their absence from the registry record limits independent assessment of the formal primary claim.

It also shows the difference between a sponsor’s model-based result and a simple recalculation from posted summaries. The latter can serve as a check, but it does not recreate repeated-measures modeling, adjustment, or all analysis details. Treating those results as interchangeable would overstate what the recalculation establishes.

The range of secondary outcomes highlights a further statistical issue. Different endpoints describe different aspects of glucose measures and safety, while a large set of analyses raises multiplicity concerns. Without a reported strategy for interpreting the set, isolated secondary p-values deserve restraint. Event-rate outcomes also require attention to event counts and the method used to quantify uncertainty.

Finally, the planning illustrations make clear that sample-size requirements depend on the effect a future study aims to detect and on design assumptions. They are not observed power and do not reveal the original rationale for this completed trial. These distinctions help readers separate the registry’s reported findings from assumptions, recalculations, and broader conclusions that the posted information cannot support.

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 does not report the non-inferiority margin or decision rule, so the formal primary non-inferiority conclusion cannot be independently assessed. The sponsor-reported primary p-value is not reported either. The original sample-size justification, detailed statistical analysis plan, and information about interim analyses are also not reported in the registry record.

The primary endpoint timeframe is baseline to Week 26. The result therefore describes that measurement period and does not establish longer-term patterns. The primary analysis population requires baseline data and at least one post-baseline HbA1c measurement; the registry does not provide a full account of missing-data handling or sensitivity analyses. The analysis may not represent every randomized participant’s endpoint data.

The model description names several covariates and treatment terms but does not report all implementation details, such as the covariance structure. The recalculated primary interval and p-value rely on summary-statistic approximations and are not substitutes for the mixed model. Secondary outcomes involve multiple analyses, and a multiplicity strategy is not reported. The severe-hypoglycemia recalculation uses a Poisson approximation, whereas the registry reports an empirical method, so the two approaches can yield different uncertainty estimates.

Safety information is limited to the posted counts, selected serious-event terms, and deaths. The registry excerpt does not report additional context such as event relatedness or timing. These limitations call for a narrow interpretation of the available estimates rather than treating them as a complete account of efficacy, safety, or the full trial analysis.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study of LY900014 Compared to Insulin Lispro (Humalog) in Adults With Type 1 Diabetes — NCT03952130.
• Registry record last updated 2023-02-08; status last verified 2023-02.
• Results first posted 2023-02-08.
• 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

NCT03952130 is a completed, Phase 3, randomized, double-blind, parallel-group study in adults with type 1 diabetes. The registry lists enrollment as 354 and compares mealtime LY900014 with mealtime insulin lispro, each combined with once-daily insulin glargine or insulin degludec. The primary endpoint is change in HbA1c from baseline to Week 26.

The reported primary least-squares mean difference is 0.07, with a 95% confidence interval from -0.11 to 0.24. The registry classifies the comparison as non-inferiority, but the margin and decision rule are not reported in the registry record. Secondary outcomes cover postprandial glucose, hypoglycemia, and other glycemic measures. Their interpretation is limited by multiple analyses and missing detail about multiplicity and missing-data procedures. The posted results describe the stated analyses and timeframes; they do not by themselves settle broader or individual-level conclusions.