A Study to Assess the Safety, Tolerability and Glucose-Lowering Efficacy of MK-0893 in Participants With Type 2 Diabetes Mellitus (MK-0893-005): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Assess the Safety, Tolerability and Glucose-Lowering Efficacy of MK-0893 in Participants With Type 2 Diabetes Mellitus (MK-0893-005) (NCT02004886), a randomized phase 2 study. The primary outcome is Change From Baseline in 24-hour Weighted Mean Glucose (WMG) at Week 4.

2. Clinical Question

This was a Phase 2 study with randomized allocation and a parallel-group design. It was described as multi-center, double-blind, double-dummy and placebo-controlled. These features support comparison among the assigned groups while limiting the influence of expectations about treatment assignment. The trial also included a metformin group, offering a second active-treatment comparison, although the stated primary hypothesis focused on MK-0893 versus placebo.

The primary outcome represents glucose measurements across a full 24-hour period rather than a single fasting measurement. It was assessed at baseline and Week 4. The registry record lists secondary measures of fasting glucose, fructosamine, C-peptide, insulin, post-prandial glucose excursion and glucose-related area under the curve measures. Numeric results for all those secondary outcomes are not reported in the registry record analyzed here, so the breadth of the outcome list should not be mistaken for evidence about their findings.

3. Trial Design

This completed Phase 2 trial used randomized allocation and a parallel model. The official study description identifies it as a multi-center, double-blind, double-dummy, placebo-controlled study. Double-dummy procedures use matching placebos so that participants in different treatment groups can follow a similar tablet-taking arrangement even when the active medicines differ. The record lists MK-0893, metformin and placebo interventions, with separate treatment groups for the two MK-0893 doses, metformin and placebo.

The assigned regimens were MK-0893 40 mg once daily after a 160 mg loading dose, MK-0893 120 mg once daily after a 500 mg loading dose on Day 1, metformin titrated across the treatment period, and placebo. The metformin schedule began at 500 mg twice daily from Day 1 to Day 6, changed to 1000 mg in the morning and 500 mg in the evening from Day 7 to Day 13, and listed 1000 mg twice daily from Day 14 to Day 28. Matching placebo tablets were used with active regimens as described in the arm descriptions.

The registry record lists enrollment of 74. The posted efficacy results use a Completers Population rather than describing all randomized participants as the analysis population. To qualify, a participant had to take at least one dose of study therapy and have both a baseline measurement and a post-randomization measurement in the treatment period at Week 4. The numbers contributing to the primary results are reported by group in the posted estimates.

The study began on 2006-08-11, had a primary completion date of 2007-02-07, and is listed as completed. Details of the randomization sequence, allocation concealment, site count, treatment adherence, and protocol deviations are not reported in the registry record. The record also does not report the planned sample-size justification, any interim analyses, or a full statistical analysis plan. Those omissions limit what can be inferred about how trial conduct and analysis decisions were specified in advance.

The design allowed comparisons of each MK-0893 dose and metformin with placebo for the primary endpoint. It does not, in the posted analyses, establish a direct statistical comparison between the two MK-0893 doses or between MK-0893 and metformin. A numerical difference between group estimates is not itself a test of that difference.

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: Completers Population was used for all efficacy analyses, and required that a participant took at least one dose of study therapy, had a baseline measurement, and had a post-randomization measurement in the treatment period at Week 4.

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 From Baseline in 24-hour Weighted Mean Glucose (WMG) at Week 4, measured over Baseline and Week 4.

The primary endpoint was change from baseline in 24-hour weighted mean glucose (WMG) at Week 4, measured in mg/dL. WMG was calculated by averaging multiple plasma glucose measurements over a 24-hour period. Blood sampling was scheduled before meals, at several post-meal times, and at midnight, 3 AM, and the following morning at 6:30 AM and 7:30 AM. This repeated sampling captures glucose at different points in the daily cycle, rather than relying on one measurement.

A change-from-baseline endpoint expresses how the measured outcome at follow-up differs from the participant's starting value. In this record, negative group estimates correspond to decreases in WMG from baseline. The reported group values are least-squares means with 95% confidence intervals. Least-squares means are model-based estimates, which can account for terms included in the analysis model; the registry record does not report the complete model specification.

WMG was selected as the primary measure because the stated study question concerned glucose lowering over a 24-hour period. A daily average can summarize the repeated plasma glucose measurements in a single endpoint, while retaining information across several times of day. It does not describe every aspect of glucose control, and a change in this measure alone does not establish what happened to other outcomes or to participants' health more broadly.

The endpoint was measured at baseline and Week 4. Longer-term outcome results are not reported in the registry record. Consequently, the posted primary result addresses the measured short-term change, not whether the observed pattern continued beyond the specified assessment.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Several statistical ideas help interpret the reported estimates. The registry describes group results as least-squares means with 95% confidence intervals and reports ANCOVA comparisons with placebo. A confidence interval represents the uncertainty of an estimate under the statistical model and its assumptions. It is not a range containing a fixed proportion of individual participants, and it does not account for sources of uncertainty that the model does not address.

What does a change from baseline represent?

A change-from-baseline outcome compares a follow-up measurement with the starting measurement. Here, the endpoint is the change in 24-hour WMG from baseline to Week 4. Negative values indicate a decrease in the measure. This framing focuses on movement over the treatment period, but it remains important to distinguish a group's estimated average change from the experience of any one participant.

The group estimates are reported as least-squares means. In general, a least-squares mean is an estimate from a statistical model rather than a simple unadjusted average. It may account for covariates included in that model. The registry record does not report the complete model specification, so the precise covariate adjustments and other modeling details cannot be assessed from the record.

How should the ANCOVA comparisons be read?

ANCOVA is a general method for comparing groups while incorporating one or more continuous covariates. In trials with baseline and follow-up measures, it is commonly used to compare follow-up outcomes while accounting for baseline measurements, but the exact use of covariates in this trial is not fully reported. The registry labels its comparisons as ANCOVA and gives an estimate, confidence interval and p-value for each active-treatment comparison with placebo.

The reported contrast for MK-0893 120 mg is labeled a median difference in final values, while the contrasts for MK-0893 40 mg and metformin are labeled mean differences in final values. These are not identical summary parameters. The differing labels should be retained when interpreting the estimates rather than treating all three as the same kind of average. The registry record does not explain the reason for this difference in reporting.

What do the p-values and intervals say?

A p-value evaluates how compatible the observed data are with a specified null hypothesis under the statistical model. It is not the probability that the null hypothesis is true, nor does it measure the size or clinical importance of a result. The reported p-values for the three primary comparisons are each p < 0.001. Their corresponding confidence intervals lie below zero, indicating estimates in the direction of lower final WMG in the active groups relative to placebo as parameterized in the registry.

A confidence interval gives a range of values reasonably compatible with the observed data and the model at its stated confidence level. Its width reflects statistical precision: wider intervals indicate greater uncertainty about the exact contrast, while narrower intervals indicate greater precision. Neither the interval nor the p-value resolves issues such as missing data, analysis-population selection, outcome measurement, or multiplicity.

How do recalculations differ from the sponsor analyses?

Recalculated comparisons are unadjusted calculations from posted summary statistics; they do not reproduce an ANCOVA model when its participant-level data and complete specification are unavailable. For the posted fasting C-peptide comparisons, the recalculation uses group summaries and an approximate pooled standard deviation. That approximation and the unadjusted approach mean these results are a separate analytic check, not a replacement for the sponsor's reported model-based analysis. The registry does not provide recalculated primary endpoint contrasts in the posted results.

An unadjusted recalculation can help illustrate how summaries relate, but it may not match an adjusted analysis. Differences can arise from baseline adjustment, model choice, the analysis population, rounding, and assumptions used to reconstruct variability. A recalculated p-value should therefore not be presented as if it were the sponsor's test result.

What do the planning illustrations mean?

Sample-size illustrations based on an observed effect are planning scenarios, not observed power. The stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect set to a stated fraction of the observed effect. Such illustrations indicate how a future study's size might vary under those assumptions. They do not show the probability that this completed trial detected an effect, and they are not evidence that the observed effect will recur.

General statistical caution is warranted when planning from an observed effect: estimates from a study can be imprecise, and using them directly for planning may produce an overly optimistic target. The registry record does not report that these illustrations were the trial's original design basis. The trial's own sample-size justification is not reported in the registry record.

10. Analyses Reported by the Sponsor

The registry reports the sponsor's ANCOVA analyses for the primary comparisons. The estimates are -25.9 mg/dL for MK-0893 40 mg versus placebo, -53.6 mg/dL for MK-0893 120 mg versus placebo, and -26.0 mg/dL for metformin 2000 mg versus placebo. The first and third are labeled mean differences in final values; the second is labeled a median difference in final values. Each has a 95% confidence interval below zero and p < 0.001. These reported analyses are the primary source for interpreting the trial's stated comparisons.

The separate recalculations described for fasting C-peptide are unadjusted estimates reconstructed from posted group summaries, not a replication of the sponsor's ANCOVA. They use approximate variability and yield unadjusted comparisons. The sponsor's reported analyses and these recalculations answer related but not identical statistical questions. A difference between them would not, by itself, demonstrate an error in either analysis; it can reflect different methods and assumptions. The registry discrepancy flag is false.

No recalculated primary endpoint comparison is reported in the registry record analyzed. Accordingly, there is no recalculation to set beside the sponsor's primary ANCOVA estimates. The relevant interpretation of the primary endpoint rests on the sponsor's own reported comparisons, while the recalculation material for the secondary endpoint should be treated as a supplementary, method-dependent check rather than as an alternative trial result.

11. Multiplicity, Missing Data and Analysis Populations

The registry reports three primary active-treatment comparisons with placebo, along with several secondary outcomes. The primary comparisons each have p < 0.001, but the record does not report a multiplicity adjustment, a testing hierarchy, or a prespecified method for controlling the overall chance of false-positive findings across comparisons. Without that information, the p-values should be read as reported for the individual comparisons rather than as evidence that the full family of tests was controlled at a specified error rate.

Multiplicity matters because repeated statistical testing creates more opportunities to observe a small p-value by chance, even when no true effect exists for any comparison. This concern is especially relevant when a study includes multiple treatment groups and multiple endpoints. It does not negate the reported estimates; it affects how confidently a collection of results can be interpreted as confirmatory. Whether an adjustment was performed is not reported in the registry record.

The efficacy population is described as a Completers Population. It required at least one dose of study therapy, a baseline measurement, and a post-randomization measurement during the treatment period at Week 4. This definition excludes participants without the specified outcome data from the efficacy analysis, but the record does not give details of missing observations, reasons for missingness, or how missing data were handled. Imputation methods, sensitivity analyses, and a comparison of the Completers Population with all randomized participants are not reported in the registry record.

These details matter because missingness can affect both the precision and the validity of an estimate. A completer-based analysis may be representative if missing data are unrelated to outcome after accounting for relevant factors, but that assumption cannot be assessed without further information. Other assumptions about the statistical model, endpoint distributions, and treatment adherence are also not fully described.

The record does not report interim analyses, stopping rules, a detailed statistical analysis plan, or the sample-size justification. Their absence from the registry record should not be interpreted as proof that they did not exist; their status is not reported in the registry record. These gaps limit evaluation of whether outcome selection, analysis choices, and multiplicity procedures were specified before results were examined.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

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's serious adverse event figures list zero affected participants in each group: MK-0893 40 mg, MK-0893 120 mg, metformin 2000 mg, and placebo. The numbers at risk for this serious-event measure are reported as 19, 18, 18 and 19, respectively. These figures describe what is recorded for the trial and its reported safety population; they do not establish that serious events are impossible or that risks are equivalent across groups.

The record also lists participants affected by other, non-serious adverse events: 3 for MK-0893 40 mg, 7 for MK-0893 120 mg, 9 for metformin 2000 mg and 8 for placebo. These counts are distinct from the serious-event figures and should not be combined with them. The registry does not provide the serious adverse event terms in its top-term list. Death counts are not reported in the registry record.

Safety counts from a trial of this size and duration have limited ability to identify uncommon events or characterize risks over longer exposure. The record does not provide event-level details, severity and relatedness assessments, or information about timing and recurrence. Accordingly, the figures are a descriptive account of recorded events, not a complete comparison of safety profiles and not a basis for treatment advice.

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

The primary results show that the estimated 24-hour WMG decreased from baseline in each treatment group by Week 4. The sponsor's reported ANCOVA comparisons show lower final values for each active group than for placebo, with the reported confidence intervals below zero and p-values less than 0.001. The largest estimated decrease and the most negative reported contrast were in the MK-0893 120 mg group. These are statements about the posted results and the specific endpoint and analysis reported.

The results do not establish that one MK-0893 dose is statistically superior to the other: the registry reports comparisons against placebo, not a direct dose-to-dose test. They also do not establish that MK-0893 is superior to metformin, because the reported analyses do not compare those groups directly. Similar-looking estimates are not a formal test of equivalence or non-inferiority.

The result concerns the Week 4 assessment in a Completers Population. It does not show whether the change persisted after Week 4, whether other secondary outcomes followed the same pattern, or what the results would have been for participants without qualifying follow-up measurements. Numeric results for several listed secondary outcomes are not reported in the registry record.

A statistically clear contrast is not by itself a measure of clinical importance. The record does not report a threshold for a clinically important change in WMG, nor does it provide enough information to connect this endpoint change to longer-term outcomes. A group estimate also does not describe the range of individual responses. The confidence intervals quantify uncertainty under the analysis model, but cannot remove limitations related to population selection, missing data, model assumptions or multiplicity.

Thus, the posted data support an interpretation of short-term differences in the measured glucose endpoint between active groups and placebo under the reported analyses. They do not on their own establish a broader treatment benefit, a long-term effect, or a complete assessment of safety. No medical advice follows from these statistical results alone.

14. Why This Trial Matters Statistically

This trial illustrates why a clinical trial result should be read at several levels. The group changes describe what happened to the measured outcome within each assigned group; the between-group contrasts address the comparative question; and the confidence intervals indicate how precisely those contrasts were estimated. Here, the active-versus-placebo comparisons are reported directly, which is more informative than inferring a difference from separate group estimates alone.

The record also highlights how endpoint definition shapes interpretation. A 24-hour weighted mean glucose measure aggregates samples collected at multiple times, making it distinct from fasting glucose or a single post-meal measure. Results on one measure cannot automatically be generalized to the other listed outcomes. The availability of a broad outcome list does not substitute for numeric results or a clear account of how multiple tests were handled.

Differences in reported comparison parameters deserve attention as well. The 40 mg and metformin comparisons are labeled mean differences in final values, whereas the 120 mg comparison is labeled a median difference. Even when all estimates point in the same direction, unlike summary parameters should not be treated as interchangeable. Analysts should retain those labels and avoid making an unsupported direct comparison across doses.

Finally, the trial demonstrates the limits of interpreting a completer analysis without fuller missing-data information. The efficacy population has a specific follow-up requirement, but the record does not describe what happened to participants who did not meet it or how missing outcomes were handled. Statistical conclusions are strongest when the estimand, analysis population, model, missing-data approach and multiplicity strategy are all transparent. Those elements determine what question the reported estimate answers and how far it can be generalized.

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.

Several trial-specific limitations constrain interpretation. The efficacy results use a Completers Population defined by treatment exposure and availability of baseline and Week 4 measurements. The registry record does not report how many randomized participants lacked those measurements, why data were missing, or whether their characteristics differed from those included. The effect of that selection on the estimates cannot therefore be assessed.

The record does not describe the complete ANCOVA model, including all covariates, handling of deviations from model assumptions, or sensitivity analyses. It also does not explain why the reported 120 mg comparison is labeled a median difference while the other two primary comparisons are labeled mean differences. The figures can be interpreted according to their labels, but direct comparison among unlike summary parameters requires caution.

Multiplicity procedures are not reported, despite multiple active-treatment comparisons and a list of secondary outcomes. The numeric results for most of the listed secondary measures are not reported in the registry record, limiting assessment of consistency across outcomes. The record also does not report a detailed statistical analysis plan, sample-size justification, interim analyses, or stopping rules.

The primary assessment is at Week 4, and longer-term results are not reported in the registry record. The available safety summary is limited to event counts; event terms, timing, severity, relatedness, and death counts are not reported in the registry record. These constraints limit assessment of duration and safety detail. These are limitations of what can be concluded from the registry record, not proof that additional information was never collected or analyzed.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Study to Assess the Safety, Tolerability and Glucose-Lowering Efficacy of MK-0893 in Participants With Type 2 Diabetes Mellitus (MK-0893-005) — NCT02004886.
• Registry record last updated 2018-09-05; status last verified 2018-08.
• Results first posted 2014-03-18.
• 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

This completed Phase 2 randomized, double-blind, double-dummy, parallel-group study enrolled 74 participants with type 2 diabetes mellitus. It compared MK-0893 40 mg, MK-0893 120 mg, metformin 2000 mg and placebo. The primary endpoint was change from baseline in 24-hour weighted mean glucose at Week 4, analyzed in a Completers Population.

All four groups had estimated decreases in WMG. The registry reports ANCOVA comparisons of each active group with placebo; each reported confidence interval lies below zero and each p-value is p < 0.001. A direct comparison between the two MK-0893 doses or between MK-0893 and metformin is not reported. The registry lists eight secondary outcomes, but numeric results for most are not reported in the registry record analyzed.

The efficacy analysis population, unreported multiplicity and missing-data procedures, and limited duration of the primary assessment constrain interpretation. Serious adverse event counts were zero in all listed groups; other adverse event counts are also reported, while death counts are not reported in the registry record. The posted findings describe short-term statistical results for the specified endpoint and population.