A Study in Healthy Men to Test Whether Food Affects the Amount of BI 894416 in the Blood: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study in Healthy Men to Test Whether Food Affects the Amount of BI 894416 in the Blood (NCT03871595), a randomized phase 1 study. The primary outcome is Area Under the Concentration-time Curve of BI 894416 in Plasma Over the Time Interval From 0 to the Last Quantifiable Data Point (AUC0-tz).

2. Clinical Question

This is a pharmacokinetic comparison: it concerns the amount of BI 894416 measured in plasma over time, not a clinical outcome or a comparison of health benefits. The registry describes the study as investigating relative bioavailability after a high-fat, high-calorie breakfast compared with administration in the fasting state. The two conditions used the same dose and tablet formulation, while meal status differed.

The crossover design means that the comparison was organized around two treatment conditions within a two-period, two-sequence study. The registry identifies the study as randomized and open-label. It does not report the sequence allocation details, period-specific results, or the interval between periods. The posted comparison therefore gives a concise estimate of relative exposure but does not, by itself, describe every feature of the conduct or analysis.

3. Trial Design

This was a completed phase 1 randomized, open-label, two-period, two-sequence crossover study in healthy male subjects. The registry lists enrollment as 14. BI 894416 was administered in both conditions as a single oral 30 mg dose, using three 10 mg tablets and 240 mL of water. In the test condition, dosing followed a high-fat, high-calorie meal. In the reference condition, dosing followed an overnight fast of at least 10 hours.

A crossover study compares treatments within participants rather than assigning each participant permanently to only one condition. The registry’s arm counts are 14 for the fed condition and 14 for the fasted condition. Those counts describe the two treatment conditions; they should not be read as evidence that the study enrolled two separate groups of 14 people. The record does not give a participant-level flow or explain whether every enrolled subject completed both periods.

The primary endpoint was AUC0-tz, the area under the plasma concentration-time curve from time zero to the last quantifiable data point. Blood sampling was scheduled within 3 hours before dosing and then at 15 minutes, 30 minutes, 45 minutes, 1 hour, 1.5 hours, 2 hours, 2.5 hours, 3 hours, 4 hours, 5 hours, 6 hours, 8 hours, 10 hours, 12 hours, 24 hours, 34 hours, and 48 hours after administration. The record describes these as planned times relative to the start of dosing.

The study began on 2019-03-20, reached primary completion on 2019-04-24, and is listed as completed. The sponsor was Boehringer Ingelheim. The registry identifies one secondary outcome, AUC from time zero extrapolated to infinity. Details of any treatment washout, sequence-specific results, or additional operational procedures 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.

Before assignment: All subjects were screened for eligibility prior to participation in the trial. Subjects attended a specialist site which ensured that they (the subjects) strictly met all inclusion and none of the exclusion criteria. Subjects were not to be allocated to a treatment group if any of the entry criteria were violated.

Treatment Period 1 (P1) + Wash-out

Treatment Period 2 (P2)

Analysis population for the primary outcome, as defined in the registry: The pharmacokinetic set included all subjects who received at least one dose of trial medication and provided at least one PK endpoint that was defined as primary and that was not excluded due to relevant protocol deviations or due to PK non-evaluability.

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: The treated set included all subjects who received at least one dose of trial medication.

6. Primary Endpoint

The registry lists the primary outcome as Area Under the Concentration-time Curve of BI 894416 in Plasma Over the Time Interval From 0 to the Last Quantifiable Data Point (AUC0-tz), measured over Within 3 hours before and then 15 minutes (min), 30 min, 45 min, and 1, 1.5, 2, 2.5, 3, 4, 5, 6, 8, 10, 12, 24, 34 and 48 hours after drug administration..

AUC0-tz measures the total plasma concentration of BI 894416 accumulated over the observed sampling interval, from time zero until the last quantifiable data point. It combines information across the concentration-time profile rather than relying on a single blood concentration at one time. Its unit is nanomole (nmol) multiplied by hour (h) per liter (L).

Pharmacokinetic AUC is used to describe systemic exposure. In this study, comparing AUC after the fed and fasted conditions addresses whether meal status was associated with a difference in measured exposure after the same dose. AUC0-tz ends at the last quantifiable measurement, so it is tied to the observed profile and its quantifiable endpoint. The registry also lists AUC from time zero extrapolated to infinity as a secondary outcome, which concerns a different time span; its numerical results are not reported in the record.

The primary endpoint is reported as a geometric mean. The registry lists standard error as the dispersion measure, but numerical arm-level standard errors are not reported. The fed geometric mean was 4437.0 nmol*h/L, and the fasted geometric mean was 4459.0 nmol*h/L. The primary analysis reports the adjusted geometric mean ratio of test to reference, rather than treating the two arm values as independent samples. That ratio is the main posted summary of relative exposure.

AUC is a pharmacokinetic measure, not a direct measure of clinical benefit, harm, or long-term outcome. Its interpretation is limited to the exposure question, study conditions, participants, and observation period recorded for this trial.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry’s primary endpoint is AUC0-tz, reported as a geometric mean, and the main comparison is an adjusted geometric mean ratio for fed versus fasted dosing. The reported ratio is 99.51%, with a 90% confidence interval from 95.36% to 103.83%. The record does not identify the statistical model, the adjustment terms, or the estimation procedure. The explanation below distinguishes what those reported statistics generally mean from the analysis details that are not documented.

What does AUC measure?

In general, area under a concentration-time curve summarizes exposure across multiple measurements over time. AUC0-tz covers the interval from dosing time zero through the last quantifiable sample. It is not a single concentration and does not describe the entire concentration profile by itself: two profiles can have similar AUC while differing in their timing or shape. The scheduled blood draws extended through 48 hours after administration, with the registry also specifying a pre-dose sampling window. The endpoint description defines the endpoint by the last quantifiable data point, not simply by the last scheduled sampling time.

The primary endpoint uses a geometric mean. In general, geometric means are useful for positive measurements that are analyzed on a multiplicative scale, where relative differences are often more informative than absolute differences. A ratio of geometric means expresses one condition relative to another. Here, the comparison is identified as test divided by reference, meaning fed relative to fasted. The posted estimate is 99.51%.

How should the confidence interval be read?

In general, a confidence interval conveys uncertainty around an estimated quantity under the assumptions of the statistical analysis. The 90% interval reported for the ratio runs from 95.36% to 103.83%. It does not mean that 90% of individual participants’ exposures fall inside that range, nor does it give a range of observed individual values. It is an interval for the estimated comparison.

In repeated use of a statistical procedure under its assumptions, a confidence interval at a stated confidence level is designed to cover the underlying quantity at that level. For a particular study, the interval is interpreted as a set of values reasonably compatible with the data and analysis, while recognizing that the method’s assumptions matter. The registry does not report enough analysis detail to evaluate those assumptions fully.

A ratio can be compared with 100% as a descriptive reference for equal values on the ratio scale. But a confidence interval that includes or crosses that reference does not by itself establish equivalence. A formal equivalence assessment requires prespecified acceptable bounds and an appropriate analysis plan. Such bounds and a decision rule are not reported in the registry record. No p-value is reported for the adjusted ratio.

Why does the crossover design matter?

In general, a crossover design can make a treatment comparison more informative by allowing each participant to contribute observations under both conditions. Measurements from the same participant are often related, and an analysis can account for that within-person structure. This differs from a parallel-group analysis that treats each arm as a separate set of participants. The trial is identified as a randomized two-period, two-sequence crossover, but the registry does not state the model used to account for participant, period, or sequence effects. It is therefore not possible to characterize the exact treatment-effect estimation method from the registry entry alone.

Over-dispersion, in general, occurs when observed variability is greater than a model assumes. If variability is understated, uncertainty estimates can be too narrow; if variability is represented appropriately, interval estimates can reflect the data more accurately. The record does not report a dispersion comparison or evidence of over-dispersion, and no conclusion about it can be drawn for this study. The endpoint’s listed dispersion measure is standard error, but arm-level numerical standard errors are not reported.

What can be inferred from the posted analysis?

The reported adjusted ratio and its confidence interval provide a point estimate and an uncertainty interval for relative exposure. They do not disclose the analysis code, model formula, transformations, treatment of missing samples, or handling of protocol deviations. Those details are not reported in the registry record. A reader can describe the posted estimate precisely but cannot reproduce the full inferential analysis from the summary values alone.

The registry identifies the pharmacokinetic set as subjects who received at least one dose and provided at least one primary pharmacokinetic endpoint that was not excluded because of relevant protocol deviations or pharmacokinetic non-evaluability. This definition identifies the intended analysis population, but the number meeting each part of the definition, exclusions, and missing observations are not separately described. The listed count of 14 per condition does not resolve those details.

Finally, the study lists a secondary AUC endpoint extrapolated to infinity, but its numerical results are not reported. The registry lists 3 outcome measures and 3 analyses, but the statistical plan, multiplicity handling, and any interim analyses are not reported in the registry record. Without those details, the primary comparison should be interpreted as the result posted for the stated endpoint, rather than as a complete account of all planned analyses or decision rules.

10. Analyses Reported by the Sponsor

The registry’s posted comparison is an adjusted geometric mean ratio of fed to fasted AUC0-tz. The ratio is 99.51%, and the 90% confidence interval is 95.36% to 103.83%. The registry also reports arm geometric means of 4437.0 nmol*h/L for fed administration and 4459.0 nmol*h/L for fasted administration. These are the numerical results available for interpreting the primary exposure comparison.

The registry does not identify the statistical method used to produce the adjustment, nor does it provide the model specification or adjustment variables. A separate unadjusted ratio or a separately labeled re-calculation is not reported in the registry record. Thus, there is no second numerical comparison to place alongside the sponsor-posted estimate. The reported adjusted ratio and interval should be described as the registry’s analysis, without attributing an undocumented method to it.

The record does not flag a discrepancy between a sponsor analysis and an independent recalculation. That absence does not supply missing model details or establish that the underlying analysis can be reconstructed from the summary data. The posted estimate remains the relevant reported comparison, with its interpretation limited by the unavailable analysis specification and the absence of a reported equivalence margin or p-value.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists the primary AUC0-tz endpoint and one secondary outcome: AUC from time zero extrapolated to infinity. It also lists 3 outcome measures and 3 analyses. The numerical results for the secondary endpoint are not reported in the registry record. The record does not specify how the additional outcomes or analyses were related to the primary question, which results were considered confirmatory, or whether any multiplicity adjustment was applied.

In general, multiplicity arises when several outcomes, comparisons, time points, or analyses are examined. Without a stated strategy, readers cannot tell whether the reported uncertainty accounts for the number of opportunities to find a notable result. The registry does not report a multiplicity plan, prespecified hierarchy, or adjustment method. No claim about whether multiplicity was controlled is warranted from the posted information.

The primary endpoint description defines the pharmacokinetic set as subjects who received at least one dose and provided at least one eligible primary pharmacokinetic endpoint, excluding endpoints affected by relevant protocol deviations or pharmacokinetic non-evaluability. The record does not provide a detailed participant flow, the number excluded, the reasons for any exclusions, or counts of missing concentration measurements. It also does not describe an imputation method or other missing-data procedure. These matters are not reported in the registry record.

The listed enrollment is 14, and the condition counts are 14 for fed and 14 for fasted administration. In a crossover study, those condition counts do not establish how many participants completed both periods or how incomplete sequences were handled. The record does not report whether the primary analysis included every enrolled participant or whether any participant’s data were omitted. The endpoint-specific population definition helps clarify eligibility for the pharmacokinetic analysis, but it does not replace a full account of exclusions and missingness.

The registry also does not report an interim-analysis plan, a stopping rule, or the detailed statistical analysis plan. Their presence or absence in trial conduct cannot be inferred from the posted summary. These gaps matter when interpreting the scope of the reported interval and whether it reflects all planned analyses and decision procedures.

12. 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 non-serious adverse event terms

The registry reports no serious adverse events in either treatment condition: 0 of 14 for fed administration and 0 of 14 for fasted administration. No deaths are reported in either condition. For other adverse events, 1 of 14 subjects is listed as affected in the fed condition and 3 of 14 in the fasted condition. These are the posted counts for the two treatment conditions.

The crossover design means that the condition counts should not be treated as observations from two independent groups. The record does not identify the subjects who experienced other adverse events, whether any individual had events under both conditions, or the event terms and severity details. The list of top serious adverse event terms is empty, consistent with the reported zero serious events, but detailed non-serious event descriptions are not included in the posted information.

Zero serious events in these small condition counts describes what was observed in this trial; it does not establish that serious events cannot occur. The registry does not report enough detail to compare event timing, duration, relationship to treatment, or the statistical uncertainty around the counts. These figures should be read as descriptive safety reporting for this study, not as a broad estimate of risk.

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

The central result is a reported adjusted geometric mean ratio of 99.51% for fed versus fasted BI 894416 AUC0-tz, with a 90% confidence interval from 95.36% to 103.83%. The arm geometric means were 4437.0 nmol*h/L after fed administration and 4459.0 nmol*h/L after fasted administration. Taken together, these figures describe relative plasma exposure under the two conditions in this study.

The result does not establish a clinical effect. AUC0-tz measures pharmacokinetic exposure, not symptom change, disease control, or another patient outcome. The registry’s population was healthy male subjects, and the dose was a single 30 mg administration. The data do not directly answer questions about other populations, repeated dosing, different doses, or meal conditions other than the stated high-fat, high-calorie meal and overnight fast of at least 10 hours.

The result also does not, on its own, establish equivalence or bioequivalence. The registry does not report an acceptance range, a prespecified decision rule, or a p-value for the primary comparison. The confidence interval communicates statistical uncertainty for the ratio, but its interpretation against any formal criterion depends on criteria that are not available in the registry record.

The study’s small enrollment and crossover structure are relevant context. Both conditions have 14 subjects listed, but subject-level data, within-person variability, period and sequence results, and the full model are not reported. The reader therefore cannot independently reconstruct the estimate or assess the influence of individual observations and design effects from the summary values alone.

The most defensible reading is narrow: the registry reports an estimated relative exposure close to 100% on the ratio scale, with the stated interval, for the described conditions and analysis population. It does not support broader conclusions about clinical outcomes or treatment use, and it does not resolve analysis details that the registry does not report.

14. Why This Trial Matters Statistically

This trial illustrates why a pharmacokinetic comparison is not reducible to two arm averages. The main result is a ratio of geometric means, paired with a confidence interval. The ratio gives a relative comparison of exposure, while the interval communicates uncertainty. Reporting both is more informative than presenting only the point estimate, particularly when the scientific question concerns how exposure under one condition relates to exposure under another.

The crossover design also highlights an important statistical distinction. Each participant can contribute information under both fed and fasted conditions, so the comparison can reflect within-person as well as between-person variation. The design is identified in the registry, but the exact model and its treatment of participant, period, and sequence effects are not described. This makes the trial useful for seeing both the value of a crossover comparison and the importance of reporting its analysis in enough detail for readers to understand how dependence between observations was handled.

The confidence interval provides a second lesson: an estimate near 100% is not, by itself, a formal equivalence conclusion. A prespecified margin and analysis rule are needed to interpret a ratio against an equivalence objective. Those details are not reported. The result can still be described as an estimate of relative exposure, but the statistical conclusion should not be made stronger than the available criteria allow.

Finally, the record distinguishes endpoint definition from analysis transparency. It specifies the AUC interval, sampling schedule, and pharmacokinetic-set criteria, yet does not report the model, missing-data handling, multiplicity strategy, or secondary endpoint values. This contrast is a practical reminder that clear measurement definitions and complete inferential reporting are separate components of a statistical account.

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 entry leaves important analysis details unspecified. Although the primary comparison is an adjusted geometric mean ratio, the model, adjustment factors, estimation method, and treatment of within-subject correlation are not reported. The reported 90% confidence interval can be quoted, but the analysis assumptions cannot be independently assessed from the summary alone. Arm-level standard errors are also not numerically reported.

The record does not report a prespecified equivalence margin or decision rule, so the ratio and interval cannot be classified against a stated formal criterion. No p-value is reported for the comparison. The registry also does not report a detailed statistical analysis plan, multiplicity strategy, or interim-analysis plan.

Participant accounting is limited. Enrollment is 14 and both treatment conditions list 14 subjects, but the record does not describe completion by period, exclusions, reasons for exclusion, or missing concentration measurements. The primary pharmacokinetic population is defined, yet the number of subjects meeting that definition is not separately stated. No method for handling missing data is reported.

The trial is small and is limited to healthy male subjects receiving a single 30 mg dose under the stated meal and fasting conditions. The registry does not provide individual concentration profiles, subject-level AUC values, sequence-specific results, or period effects. The secondary AUC endpoint extrapolated to infinity is listed, but its numerical result is not reported. These limitations constrain reconstruction of the analysis and the scope of interpretation; they do not change the reported estimate, but they limit what can be concluded beyond it.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study in Healthy Men to Test Whether Food Affects the Amount of BI 894416 in the Blood — NCT03871595.
• Registry record last updated 2023-08-14; status last verified 2022-09.
• Results first posted 2023-08-14.
• 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.

19. Record Summary

NCT03871595 is a completed phase 1 randomized, open-label crossover study sponsored by Boehringer Ingelheim. It enrolled 14 healthy male subjects and compared a single oral 30 mg dose of BI 894416 after a high-fat, high-calorie meal with the same dose after an overnight fast of at least 10 hours.

The primary endpoint was plasma AUC0-tz. The registry reports geometric means of 4437.0 nmol*h/L for fed administration and 4459.0 nmol*h/L for fasted administration, with an adjusted test-to-reference geometric mean ratio of 99.51% and a 90% confidence interval from 95.36% to 103.83%. AUC extrapolated to infinity is listed as a secondary outcome, but its result is not reported.

No serious adverse events or deaths are reported in either condition. The registry does not report the statistical model, equivalence criteria, p-value, multiplicity plan, or detailed missing-data handling. The result is therefore best read as a focused estimate of relative exposure under the study conditions.