This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for A Study of LY900014 in Japanese Participants With Type 1 Diabetes Mellitus (NCT03407118), a randomized phase 1 study. The primary outcome is Pharmacokinetics (PK): Insulin Lispro Area Under the Concentration Curve Zero to 10 Hours (AUC 0-10h) Following Administration of Each Treatment Arm.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03407118 |
| Phase | Phase 1 |
| Condition | Diabetes Mellitus, Type 1 |
| Allocation | Randomized |
| Design model | Crossover |
| Masking | Double |
| Enrollment | 31 |
| Arms in the primary outcome | 2 |
| Lead sponsor | Eli Lilly and Company |
| Results posted | Yes |
2. Clinical Question
Population
Japanese participants with type 1 diabetes mellitus were studied. The registry reports enrollment of 31 and describes the primary endpoint population as randomized participants who received at least 1 dose of study drug and had evaluable pharmacokinetic parameters. Age and other participant characteristics are not reported in the registry record.
Intervention
The intervention was a single 15 U dose of LY900014, administered by subcutaneous injection. The brief summary says injections were given under the skin of the abdomen. Each treatment was given in one of the study periods.
Comparator
The comparator was a single 15 U dose of insulin lispro (Humalog), also administered subcutaneously in one of the study periods. The crossover design means the registry describes both treatments as study-period treatments rather than separate parallel groups.
Primary question
The primary question concerned pharmacokinetics: the insulin lispro area under the concentration curve from time zero to 10 hours after each treatment. The registry reports geometric means for the treatment arms and a ratio of least square means with a 95% confidence interval. It does not report a p-value for this comparison.
3. Trial Design
This was a randomized, double-masked, phase 1 crossover study of LY900014 and insulin lispro in Japanese participants with type 1 diabetes mellitus. Enrollment was 31. The registry lists the study as completed, with a start date of 2018-02-17 and a primary completion date of 2018-06-23.
Participants received a single 15 U subcutaneous dose of each treatment, with each treatment administered in one of two study periods. The brief summary specifies injection under the skin of the abdomen. Because the design was crossover, participants served as their own comparison across treatment periods; this differs from a parallel-group design in which each participant remains in one treatment group. The registry does not report the treatment sequence, the period length, or any washout interval, so those details cannot be inferred.
The primary endpoint population is described as all randomized participants who received at least 1 dose of study drug and had evaluable pharmacokinetic parameters. Posted arm counts for the primary endpoint were 31 for LY900014 and 30 for insulin lispro. These are arm-level counts in a crossover study and should not be read as counts in two independent participant groups. The registry does not report why the arm counts differ.
The study also listed a glucodynamics outcome: total amount of glucose infused over the duration of the clamp following each treatment. Results for a two-arm comparison of this other endpoint are described as missing or not estimable in the registry record. The planned study duration was described as about 3 to 6 weeks for each participant, not including screening.
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.
Period 1
| Stage | LY900014\Insulin Lispro (Humalog) | Insulin Lispro (Humalog)\LY900014 |
|---|---|---|
| STARTED | 14 | 17 |
| Received at Least 1 Dose of Study Drug | 14 | 17 |
| COMPLETED | 14 | 17 |
| NOT COMPLETED | 0 | 0 |
Period 2
| Stage | LY900014\Insulin Lispro (Humalog) | Insulin Lispro (Humalog)\LY900014 |
|---|---|---|
| STARTED | 14 | 17 |
| Received at Least 1 Dose of Study Drug | 13 | 17 |
| COMPLETED | 13 | 17 |
| NOT COMPLETED | 1 | 0 |
Reasons for not completing this period:
| Reason | LY900014\Insulin Lispro (Humalog) | Insulin Lispro (Humalog)\LY900014 |
|---|---|---|
| Withdrawal by Subject | 1 | 0 |
Analysis population for the primary outcome, as defined in the registry: All randomized participants who received at least 1 dose of study drug and had evaluable PK parameters.
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 who received at least 1 dose of study drug.
| Characteristic | Unit | All Participants |
|---|---|---|
| Participants analysed | 31 | |
| Age, Continuous | years | 39.5 (SD 11.3) |
| Sex: Female, Male: Female | Participants | 18 |
| Sex: Female, Male: Male | Participants | 13 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 0 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 31 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 0 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 0 |
| Race (NIH/OMB): Asian | Participants | 31 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 0 |
| Race (NIH/OMB): Black or African American | Participants | 0 |
| Race (NIH/OMB): White | Participants | 0 |
| Race (NIH/OMB): More than one race | Participants | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 0 |
| Region of Enrollment: Japan | Participants | 31 |
| Weight | kilograms | 59.62 (SD 8.29) |
| Height | centimeters | 161.35 (SD 7.44) |
| Body Mass Index | kilograms per meter squared | 22.85 (SD 2.44) |
6. Primary Endpoint
The registry lists the primary outcome as Pharmacokinetics (PK): Insulin Lispro Area Under the Concentration Curve Zero to 10 Hours (AUC 0-10h) Following Administration of Each Treatment Arm, measured over Predose, 5 minutes (min), 10 min, 15 min, 20 min, 25 min, 30 min, 35 min, 40 min, 45 min, 50 min, 55 min, 60 min, 70 min, 90 min, 120 min, 150 min, 180 min, 240 min, 300 min, 360 min, 420 min, 480 min, 540 min, and 600 min postdose.
| Element | Registry entry |
|---|---|
| Result type | Geometric Mean |
| Measure of spread | Geometric Coefficient of Variation |
| Unit | picomols times hour per Liter |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint was the area under the insulin lispro concentration curve from time zero to 10 hours after administration of each treatment. Area under the curve (AUC) summarizes concentration over time: it reflects both the level of measured drug in the blood and how long that concentration is present during the measurement window. The registry gives the unit as picomols times hour per Liter and identifies the summary as a geometric mean with geometric coefficient of variation.
Pharmacokinetic endpoints are used to characterize drug exposure after administration. In this study, the concentration measurements were scheduled from predose through 600 minutes postdose, with multiple measurements during that period. AUC 0-10h condenses those repeated concentration measurements into one exposure measure for each treatment, rather than treating each sampling time as a separate primary result.
The reported arm values were geometric means of 1930.0 for LY900014 and 1980.0 for insulin lispro. Their accompanying geometric coefficient of variation values were 59.0 and 20.0, respectively. The registry also reports a ratio of least square means for the treatment comparison. These summaries describe the posted endpoint; they do not by themselves establish how the concentrations varied at each individual sampling time. The precise model used to estimate the least square means is not reported in the registry record.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The posted primary endpoint values are geometric means of 1930.0 for 15 U LY900014 and 1980.0 for 15 U insulin lispro (Humalog). The registry reports a ratio of least square means of 0.967, with a 95% confidence interval from 0.803 to 1.17. The ratio is a relative-scale comparison; it is not a difference in the endpoint’s original units. The registry does not report a p-value for this analysis.
On a ratio scale, 1 represents equal estimated values for the two treatments. The reported confidence interval includes 1, so the interval is compatible with equal values as well as with values on either side of that reference. Its breadth indicates that the estimate is not highly precise: the registry’s interval allows a range of relative treatment comparisons. This description is about the statistical information in the reported interval, not a conclusion about clinical importance or whether one treatment should be used.
The arm-level geometric coefficient of variation values were 59.0 for LY900014 and 20.0 for insulin lispro. A coefficient of variation describes dispersion relative to the scale of the data; it is not a confidence interval for the treatment comparison. The different values indicate different degrees of observed relative variability in the posted arm summaries. They do not identify why variability differed, and the registry does not report a test comparing variability.
The crossover design offers a within-participant comparison, which can account for stable differences between participants when the analysis appropriately models the design. However, the analysis method and its treatment of sequence or period effects are not reported in the registry record. The reported least-square-means ratio should therefore be read as the sponsor’s posted comparison, without assuming unreported model details.
The glucodynamics outcome was total glucose infused over the duration of the clamp. A two-arm comparison for that other endpoint is identified as missing or not estimable, so no comparative result for it can be interpreted from the registry record. Nor does the pharmacokinetic estimate establish a clinical effect on blood sugar levels. It describes the specified exposure endpoint under the study conditions.
9. Statistical Methods Explained
The registry reports geometric means for the endpoint and a ratio of least square means for the treatment comparison. These statistics answer related but different questions: the arm summaries describe typical exposure on a multiplicative scale, while the ratio expresses the relative comparison estimated by the sponsor’s analysis. The registry does not report the detailed statistical model, any covariates, or the method used to construct the confidence interval.
Why use a geometric mean?
Pharmacokinetic measurements are often positively skewed: a small number of high values can make an arithmetic mean unrepresentative of the central pattern. A geometric mean is based on a logarithmic scale and is useful when differences are more naturally interpreted proportionally than additively. This is a general statistical explanation, not a claim about the distribution of the individual observations in this study. The registry identifies the endpoint summary as a geometric mean but does not provide the participant-level data needed to assess the distribution directly.
The reported geometric means were 1930.0 for LY900014 and 1980.0 for insulin lispro. The accompanying geometric coefficient of variation values were 59.0 and 20.0. A geometric coefficient of variation describes relative dispersion around a geometric mean. It is distinct from the uncertainty around an estimated treatment ratio: the former describes variability in the endpoint values, while the latter is summarized by the confidence interval for the comparison.
What does the ratio mean?
A ratio compares quantities on a relative scale. A ratio of 1 corresponds to equal quantities; a value below 1 indicates a smaller numerator relative to the denominator, and a value above 1 indicates a larger numerator relative to the denominator. This is general guidance for interpreting ratios. The registry reports the ratio of least square means as 0.967 and gives a 95% confidence interval from 0.803 to 1.17. The registry does not describe the exact coding of the contrast, so the estimate should be interpreted in the context of the treatment order stated in the posted analysis rather than by assuming additional model details.
How should the confidence interval be read?
A confidence interval conveys uncertainty around an estimate under the assumptions of the statistical method. It is not a range containing a specified proportion of individual participant values, and it does not show the probability that a particular treatment effect is true. Here, the reported 95% confidence interval includes the ratio reference value of 1. It therefore does not isolate the comparison to one side of equality. The interval also covers a range of relative values, so the point estimate alone would overstate the precision available from the registry result.
No p-value is reported for the comparison. The confidence interval can be described without inventing a p-value or claiming that a particular hypothesis test was used. The registry labels the analysis as superiority, but the test procedure, decision threshold, and any prespecified interpretation rule are not reported in the registry record.
Why does the crossover design matter?
In a crossover study, participants receive more than one treatment in separate periods. A properly specified analysis can use within-participant comparisons, reducing the influence of participant characteristics that remain stable over the periods. This is general design reasoning. It does not establish which model was fitted in this study. The registry reports randomization, double masking, and crossover structure, but treatment sequence, period effects, carryover handling, and washout details are not reported in the registry record.
The posted endpoint counts are 31 for LY900014 and 30 for insulin lispro. In a crossover design, these arm counts are treatment-period records and should not be interpreted as independent parallel groups. The registry does not explain the difference between the counts. The endpoint population required randomization, receipt of at least 1 dose, and evaluable pharmacokinetic parameters.
What can be learned from the variability summaries?
Over-dispersion refers generally to greater variability than a statistical model expects. It can make standard errors too small and confidence intervals too narrow if the model does not accommodate the extra variation. The registry gives geometric coefficient of variation summaries, but does not report a dispersion assessment or a model diagnostic. It is therefore not possible to conclude from the posted summaries alone whether over-dispersion was present or how it was handled.
The registry record describes planning illustrations based on the observed effect, a two-sided significance level of 0.05, equal allocation, and a normal approximation. Such sample-size calculations are planning illustrations, not observed power and not evidence that a study had a particular probability of detecting an effect after the fact. No planning counts are stated in the registry record used for this interpretation. The trial’s own sample-size justification and statistical plan are not reported in the registry record.
10. Analyses Reported by the Sponsor
| Method | Estimate | 95% CI | P-value |
|---|---|---|---|
| not stated | ratio of least square means 0.967 | 0.803 to 1.17 | not reported |
The registry’s own analysis reports a ratio of least square means of 0.967, with a 95% confidence interval from 0.803 to 1.17. This is the sponsor’s posted comparison of the primary pharmacokinetic endpoint. The registry does not report a p-value, detailed model specification, or an independent recalculated treatment estimate.
A recalculation from summary statistics would be an unadjusted calculation and could not necessarily reproduce a least-square-means analysis, particularly in a crossover design. The sponsor’s own analysis takes precedence when interpreting the posted comparison; the record does not flag a discrepancy. No recalculated result should be substituted for the reported estimate or used to infer details of the sponsor’s model that are not reported.
The interval includes 1, the reference value for equal ratios. It communicates uncertainty around the sponsor’s estimate and should be considered alongside the arm-level geometric means and their variability summaries. It does not by itself establish clinical importance, equivalence, or a treatment recommendation. The registry does not report a prespecified equivalence margin or a decision rule for interpreting the confidence interval.
11. Multiplicity, Missing Data and Analysis Populations
The registry lists two outcome measures: the primary pharmacokinetic endpoint and a glucodynamics endpoint. The latter is total amount of glucose infused over the duration of the clamp following each treatment. A two-arm comparison for that other endpoint is described as missing or not estimable. The registry does not give a comparative estimate for it, so the absence of a result should not be interpreted as evidence of no difference.
The record does not report whether the study used a multiplicity adjustment, how outcomes were ordered for formal testing, or whether any secondary analyses were intended to be confirmatory. It also does not report an interim analysis plan or the protocol’s statistical analysis plan. Without those details, the reported primary comparison can be described, but the overall family of planned tests and any control of false-positive error cannot be assessed.
The primary endpoint population included randomized participants who received at least 1 dose and had evaluable pharmacokinetic parameters. The posted arm counts are 31 for LY900014 and 30 for insulin lispro. The registry does not report reasons for missing or unevaluable endpoint data, nor does it describe a missing-data method or sensitivity analysis. The difference in the arm counts should not be attributed to a specific cause.
The record does not report whether the primary analysis followed intention-to-treat principles, how participants with incomplete crossover-period data were handled, or whether the analysis adjusted for treatment period or sequence. Those issues can affect interpretation in crossover studies, but the registry’s posted information does not resolve them. The result therefore supports interpretation of the stated analysis and population, not assumptions about unreported procedures.
12. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| LY900014 | 0 | 31 | 0.0% |
| Insulin Lispro (Humalog) | 0 | 30 | 0.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| LY900014 | 0 of 31 |
| Insulin Lispro (Humalog) | 0 of 30 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 5% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| LY900014 | 6 of 31 |
| Insulin Lispro (Humalog) | 7 of 30 |
Most frequent non-serious adverse event terms
| Event term | Organ system | LY900014 | Insulin Lispro (Humalog) |
|---|---|---|---|
| Headache | Nervous system disorders | 3 of 31 | 4 of 30 |
| nausea | Gastrointestinal disorders | 3 of 31 | 1 of 30 |
| vomiting | Gastrointestinal disorders | 2 of 31 | 2 of 30 |
| Back pain | Musculoskeletal and connective tissue disorders | 1 of 31 | 2 of 30 |
The registry reports no serious adverse events in either treatment group: 0 among 31 at risk for LY900014 and 0 among 30 at risk for insulin lispro (Humalog). It also reports deaths of 0 in both groups. These are the posted counts for the safety categories; they do not establish that no adverse events of any kind occurred.
For other adverse events, the registry reports 6 affected in the LY900014 group among 31 at risk and 7 affected in the insulin lispro group among 30 at risk. These are descriptive group counts. The registry does not provide event terms for these other adverse events in the information, so their nature, severity, timing, and relationship to treatment cannot be characterized from these counts alone. The list of top serious-event terms is empty.
Safety counts in a small phase 1 crossover study are limited descriptions of what was recorded during the study. They are not estimates of long-term risk and should not be treated as a statistical comparison of safety. The registry does not report a comparative safety analysis, uncertainty intervals for adverse-event counts, or further detail about how events were collected across treatment periods. The at-risk figures are reported by treatment group and should not be added together to infer a number of distinct participants.
13. What the Result Does, and Does Not, Mean
The registry result describes a specific pharmacokinetic comparison: insulin lispro area under the concentration curve from time zero to 10 hours after a single 15 U dose of LY900014 or insulin lispro in Japanese participants with type 1 diabetes. The arm-level geometric means were 1930.0 and 1980.0, and the reported ratio of least square means was 0.967 with a 95% confidence interval from 0.803 to 1.17. The interval includes 1 and spans relative values on both sides of that reference.
This result is not a direct measure of a clinical outcome. The study’s primary endpoint measures drug exposure over a defined sampling window. Although the registry describes a glucodynamics endpoint related to glucose infused during a clamp, its comparative result is missing or not estimable. The posted primary result therefore cannot be used to claim a difference in blood sugar outcomes, nor does it establish that one treatment is clinically better or worse.
The result also does not establish equivalence. Equivalence requires a prespecified margin and an analysis designed to assess whether the plausible treatment differences fall within that margin. The registry does not report such a margin or an equivalence analysis. Similarly, the ratio and confidence interval alone do not establish clinical importance; no clinical threshold for interpreting this endpoint is reported.
The design and population further define the scope of interpretation. This was a randomized, double-masked crossover study of Japanese participants with type 1 diabetes, with each treatment given as a single dose. Results from this setting do not automatically describe repeated dosing, other populations, or outcomes not measured in the study. The registry’s endpoint population was limited to randomized participants who received at least 1 dose and had evaluable pharmacokinetic parameters.
The most defensible reading is therefore narrow: the sponsor’s reported analysis estimates the relative exposure endpoint for the two study treatments, with uncertainty represented by the stated confidence interval. Details of the model, missing-data handling, and broader statistical plan are not reported in the registry record.
14. Why This Trial Matters Statistically
This study illustrates why the endpoint and design must be read together. The primary outcome is pharmacokinetic exposure, summarized as AUC from time zero to 10 hours, rather than a direct clinical outcome. The analysis therefore addresses how the concentration-time exposure compared after the study treatments, not whether participants experienced a particular clinical benefit.
The crossover structure is also statistically important. Each participant received both treatments in separate periods, making a within-participant comparison relevant in principle. Such designs can reduce the influence of stable differences between participants, but their interpretation depends on period, sequence, and carryover considerations. The registry reports randomization, double masking, and crossover structure, while those analysis details are not reported in the registry record. That distinction demonstrates why design labels alone do not fully describe an analysis.
The ratio estimate and its confidence interval offer a useful lesson in uncertainty. A point estimate gives one summary of the comparison, while the interval shows a range of values compatible with the reported analysis. Here the interval includes 1. A reader should not replace that uncertainty with a categorical claim based only on the point estimate. Nor should the absence of a reported p-value be filled in by assumption.
Finally, missing comparative results for the glucodynamics outcome illustrate the difference between an outcome being listed and a result being available for interpretation. A registry can describe multiple outcomes while providing a usable estimate for only some of them. Assessing evidence requires attention to which endpoint was analyzed, which population contributed data, and what information is absent—not simply the presence of an outcome name or an arm-level summary.
15. 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 leaves several trial-specific details unresolved. It does not report the treatment sequence, period duration, washout interval, or whether and how the analysis addressed period and carryover effects. Those details matter to interpretation of crossover comparisons, but cannot be reconstructed from the posted design label.
The primary result is reported as a ratio of least square means, but the model specification, covariates, estimation method, and confidence-interval procedure are not reported in the registry record. A p-value is also not reported. The reported geometric means and geometric coefficient of variation values do not replace those missing model details.
The primary endpoint population required receipt of at least 1 dose and evaluable pharmacokinetic parameters. The registry does not report why the primary endpoint counts were 31 for LY900014 and 30 for insulin lispro, whether any participants lacked data for a particular period, or how incomplete records were handled. No missing-data method or sensitivity analysis is reported.
The glucodynamics outcome is listed, but its two-arm comparison is reported as missing or not estimable. Consequently, the available primary pharmacokinetic result cannot address that endpoint. The registry does not explain why its comparison was unavailable.
Other limitations include the lack of reported participant characteristics beyond the stated population and condition, the absence of a reported sample-size justification or trial statistical plan, and the lack of a reported multiplicity strategy. Safety information is limited to serious-event and other-event counts, without further event descriptions in the record. These omissions constrain assessment of precision, generalizability, and analysis choices; they do not justify assumptions about what the full protocol or unposted data contained.
16. Related Tutorials
17. Related Statistical Calculators
18. Sources
- ClinicalTrials.gov: A Study of LY900014 in Japanese Participants With Type 1 Diabetes Mellitus — NCT03407118.
- Registry record last updated 2020-05-01; status last verified 2018-08.
- Results first posted 2020-05-01.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
19. Record Summary
NCT03407118 is a completed phase 1 randomized, double-masked crossover study sponsored by Eli Lilly and Company. It enrolled 31 Japanese participants with type 1 diabetes mellitus and compared single 15 U subcutaneous doses of LY900014 and insulin lispro (Humalog) in separate study periods.
The primary endpoint was insulin lispro AUC from time zero to 10 hours after each treatment. The registry reports geometric means of 1930.0 for LY900014 and 1980.0 for insulin lispro, with geometric coefficient of variation values of 59.0 and 20.0. The sponsor’s reported ratio of least square means was 0.967, with a 95% confidence interval from 0.803 to 1.17; no p-value is reported.
The registry also lists a clamp-based glucodynamics endpoint, but its comparative result is missing or not estimable. Serious adverse events and deaths were reported as 0 in both treatment groups. Treatment sequence, detailed model specifications, missing-data methods, and multiplicity procedures are not reported in the registry record.