Sitagliptin Comparative Study in Patients With Type 2 Diabetes (0431-049): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Sitagliptin Comparative Study in Patients With Type 2 Diabetes (0431-049) (NCT00449930), a randomized phase 3 study. The primary outcome is Change From Baseline in Hemoglobin A1c (HbA1c) at Week 24.

2. Clinical Question

3. Trial Design

This was a Phase 3, randomized, double-blind, parallel-group study. Its official title describes a multicenter comparison of sitagliptin with metformin in patients with type 2 diabetes mellitus and inadequate glycemic control. The registry record lists an enrollment of 1050 and identifies the study as completed. The recorded start date is 2007-03-01, and the primary completion date is 2008-07-25.

The two interventions were sitagliptin phosphate and metformin hydrochloride. The primary results identify the arms as sitagliptin 100 mg and metformin. The reported per-protocol analysis included 455 participants in the sitagliptin group and 439 in the metformin group. Those analysis populations are not the same thing as the overall enrollment: the endpoint definition required measurements at both baseline and Week 24 and excluded patients with major protocol violations.

Randomization and double masking are stated in the registry record, as is a parallel-group model. Further information on the randomization procedure, allocation concealment, treatment adherence beyond the protocol-violation example, and follow-up procedures is not reported in the registry record. The record also does not report the treatment duration as a separate design detail; the primary endpoint timeframe is baseline and 24 weeks.

The posted primary outcome is change from baseline in HbA1c at Week 24, summarized as a least-squares mean with a 95% confidence interval for each arm. A separate reported analysis gives a net mean difference and is categorized as non-inferiority or equivalence. The margin required to interpret that framework is not reported in the registry record. Thus, the design and endpoint can be described, but the full protocol and statistical rationale cannot be reconstructed from the registry results alone.

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: Patients 18-78 years of age with Type 2 diabetes mellitus (T2DM) who were not on anti-hyperglycemic agents for at least 4 months (16 weeks) and with a hemoglobin A1c (HbA1c) of ≥ 6.5 and ≤9.0%

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: The per protocol population required that a patient had measurements both at baseline and at Week 24, and did not have any major protocol violations (e.g. drug compliance \<85%, addition of prohibited antihyperglycemic agent, incorrect double-blind study medication). No missing data were imputed.

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 Hemoglobin A1c (HbA1c) at Week 24, measured over Baseline and 24 weeks.

The primary endpoint is change from baseline in hemoglobin A1c (HbA1c) at Week 24, measured in percent. The registry defines the change as the Week 24 HbA1c percent minus the Week 0 HbA1c percent. A negative change therefore represents a lower endpoint value than baseline, while the reported estimate remains a change score rather than an absolute Week 24 value.

HbA1c is the study’s measure of glycemic control, and the primary question concerns how that continuous measure changed during the study. As a general statistical point, using a continuous endpoint preserves the scale of the measurement and allows the average change in each group to be estimated. Least-squares means are model-based group estimates; the specific model used for this analysis is not reported in the registry record.

The endpoint estimate applies to the per-protocol population, which required both baseline and Week 24 measurements and excluded major protocol violations. The registry states that no missing data were imputed. This definition matters for interpretation: the reported group estimates describe the eligible per-protocol observations, not necessarily every randomized participant. The registry record does not report a separate rationale for choosing this endpoint or explain how missing Week 24 measurements affected the analyzed population beyond the stated eligibility requirement.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is a change in HbA1c measured in percent. For a continuous outcome, a group mean describes the average value of the measure among the analyzed participants, while a mean change describes the average within-person difference between the specified time points. The registry reports least-squares means, rather than identifying them simply as raw arithmetic averages. Least-squares means are generally model-based estimates adjusted according to a statistical model. The model and any covariates used are not reported in the registry record.

What do the arm estimates describe?

The sitagliptin arm’s least-squares mean change is -0.43 percent, with a 95% confidence interval from -0.48 to -0.38. The metformin arm’s estimate is -0.57 percent, with a 95% confidence interval from -0.62 to -0.51. Each estimate refers to change from baseline to Week 24 in the per-protocol population. The interval around an estimate communicates sampling uncertainty under the analysis method. It does not describe the range of changes for individual participants.

The reported 95% confidence interval is a frequentist interval: in general, the method is designed so that intervals constructed repeatedly under the same assumptions would cover the target parameter at the stated rate. That general interpretation does not mean there is a 95% probability that this particular fixed parameter lies within the observed interval. The interval also cannot account for possible bias from exclusions, missing observations, or model misspecification unless those are explicitly addressed in the analysis.

How should the treatment contrast be read?

The registry reports a net mean difference of 0.14 with a 95% confidence interval from 0.06 to 0.21. This is the direct reported comparison between the treatment groups. The meaning of a positive value depends on the contrast’s sign convention; the registry record does not define that convention in the details. The arm estimates and the contrast should therefore be read as separately reported results rather than using one to reconstruct the other.

The analysis is categorized as non-inferiority or equivalence. These are distinct inferential aims from simply testing whether two group means differ. In general, a non-inferiority analysis asks whether a treatment is not worse than a comparator by more than a prespecified margin. An equivalence analysis asks whether the difference is contained within prespecified bounds in both directions. The margin or bounds are central to interpreting either design. They are not reported in the registry record, so the category label alone does not establish the result against a decision threshold.

What can a confidence interval say about precision?

A confidence interval shows a range of estimates compatible with the observed data and statistical assumptions. A narrower interval generally indicates greater precision for the quantity estimated; a wider interval indicates more uncertainty. Precision is not the same as clinical importance, and an interval does not say whether a difference matters to patients. The registry record supplies no threshold for clinical importance and no non-inferiority or equivalence margin.

The model used to calculate the least-squares means and the contrast is not reported. Other relevant details, including the variance assumptions and any adjustment for baseline values, are also not reported in the registry record. Without those details, an independent reproduction of the sponsor analysis is not possible from the posted summaries alone.

Why does the analysis population matter?

The primary analysis was per protocol. It required baseline and Week 24 measurements and excluded patients with major protocol violations. The registry gives examples: drug compliance <85%, addition of a prohibited antihyperglycemic agent, and incorrect double-blind study medication. No missing data were imputed. An analysis restricted in this way estimates the endpoint among participants meeting those requirements; it does not automatically answer the same question as an analysis of everyone randomized.

As a general statistical concept, missing-data methods and analysis-population definitions can affect both the estimate and its uncertainty. Requiring observed measurements at both time points means participants without those measurements do not contribute to the reported endpoint estimate. The registry does not report the number or reasons for missing endpoint measurements, nor does it describe sensitivity analyses for missing data.

What do planning illustrations mean?

Planning sample sizes are not observed trial results and should not be interpreted as observed power. The planning assumptions described for this analysis use the observed effect, a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation; the target effect is specified as a fraction of the observed effect. Such calculations are illustrations of the sample size needed under those assumptions, not evidence that the trial was designed using those assumptions. The trial’s own sample-size justification is not reported in the registry record.

These assumptions simplify the planning calculation. In general, a sample-size calculation depends on the effect size being targeted, variability, significance level, desired power, and design assumptions. If those assumptions differ from the actual trial design or analysis, the resulting planning illustration may not match the trial’s original rationale. No sample-size justification, interim analysis plan, or full statistical analysis plan is reported in the registry record.

10. Analyses Reported by the Sponsor

The registry’s reported inferential comparison is a net mean difference of 0.14, with a 95% confidence interval from 0.06 to 0.21. The analysis is labeled non-inferiority or equivalence, but the margin or bounds needed to interpret that framework are not reported in the registry record. A p-value is also not reported for this comparison. The label and interval should therefore be read as the sponsor’s posted analysis, not as a standalone declaration that a particular criterion was met.

No separate recalculated comparison is reported in the registry record. The arm-level estimates are least-squares means with their own confidence intervals, and the registry’s direct comparison remains the appropriate posted contrast to describe. Recalculating a contrast from arm summaries would not necessarily reproduce an adjusted analysis, particularly when the model and covariance information are not available. No discrepancy between a sponsor result and a recalculated result is reported.

The sponsor’s method for the comparison is not stated in the registry results. This limits independent assessment of how the interval was produced and how the non-inferiority or equivalence framework was applied. The sponsor estimate can be presented as reported, while noting that the decision margin, analysis details, and p-value are not reported in the registry record.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 5 outcome measures and 5 analyses, and identifies 4 secondary outcomes. The secondary outcomes listed concern patients reporting one or more episodes of diarrhea, nausea, abdominal pain, or vomiting. The registry record does not report whether the analyses were adjusted for multiple comparisons, how outcomes were prioritized beyond naming the primary endpoint, or whether a formal testing sequence was used.

Multiplicity arises when several outcomes or comparisons are tested: in general, conducting more tests can increase the chance of at least one false-positive finding if each is assessed without an appropriate strategy. Whether and how to control that risk depends on the analysis plan and the intended claims. Since that plan is not reported in the registry record, the secondary outcomes should be treated as separately posted results rather than evidence of a stated multiplicity-controlled testing strategy.

For the primary endpoint, the registry specifies a per-protocol population requiring measurements at baseline and Week 24 and no major protocol violations. It gives examples of major violations, including drug compliance <85%, addition of a prohibited antihyperglycemic agent, and incorrect double-blind study medication. The record states that no missing data were imputed. It does not report how many participants lacked endpoint measurements, why measurements were missing, or whether any sensitivity analysis assessed the consequences of missingness.

The record does not report an intention-to-treat analysis, a modified intention-to-treat analysis, or estimates for other analysis populations. Nor does it describe interim analyses or stopping rules. These omissions do not imply that such procedures were absent from the full study documentation; they mean that the relevant details are not reported in the registry record. The listed analysis population and missing-data approach are the limits that can be described from the posted endpoint information.

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

The result describes average change in HbA1c from baseline to Week 24 among participants who met the per-protocol requirements. The least-squares mean estimates are -0.43 percent for sitagliptin 100 mg and -0.57 percent for metformin. The registry’s direct comparison is a net mean difference of 0.14, with a 95% confidence interval from 0.06 to 0.21. These statistics summarize group-level estimates and their uncertainty; they do not describe every participant’s response.

The study is categorized as a non-inferiority or equivalence analysis, but the relevant margin or bounds are not reported in the registry record. Without that decision threshold, it is not possible to determine from the posted estimate and interval whether the sponsor’s prespecified criterion was satisfied. The comparison’s p-value, statistical model, and detailed analysis plan are also not reported. The posted results can therefore be described, but the full inferential claim cannot be independently evaluated from these details alone.

The analysis population is another boundary on interpretation. It excludes patients without measurements at both time points and those with major protocol violations, and no missing data were imputed. The record does not report how the findings compare with an analysis including all randomized participants or how sensitive the estimate is to missing outcomes or protocol deviations.

These results do not establish effects beyond the measured endpoint and timeframe, do not establish individual outcomes, and do not provide a basis for treatment decisions. They are a registry-reported summary of one randomized comparison, with specific population and analysis limitations. The registry also reports adverse-event counts, but exposure denominators are not listed for those results, limiting interpretation of event rates.

13. Why This Trial Matters Statistically

This trial illustrates why the endpoint estimate and the direct treatment contrast answer related but different questions. The least-squares means describe change within each treatment group, while the net mean difference is the registry’s reported comparison between groups. The comparison is the relevant estimate for the stated non-inferiority or equivalence analysis; interpreting it requires the margin or bounds, which are not reported in the registry record.

It also demonstrates the importance of reading an estimate together with its analysis population. The primary results use a per-protocol population defined by measurements at baseline and Week 24 and the absence of major protocol violations. The registry says no missing data were imputed. Those details affect the population to which the posted estimates directly apply and distinguish them from results based on all randomized participants.

Confidence intervals make uncertainty visible, but they cannot answer every question about the result. The intervals quantify uncertainty under the methods used, while the registry does not report the model details, the decision margin, or the analysis plan needed for fuller assessment. A non-inferiority or equivalence label is not a substitute for the margin and decision rule.

The record also presents a setting where a registry summary is informative but incomplete. It provides the main outcome estimates, analysis population, and selected adverse-event information, while leaving key design and statistical details unreported. For clinical researchers and biostatisticians, separating what the posted estimates establish from what depends on unreported protocol information is essential to a careful interpretation.

14. 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 of the posted primary result. First, the analysis is per protocol rather than an analysis explicitly identified as including all randomized participants. Participants needed measurements at baseline and Week 24 and could not have major protocol violations. This restriction may make the analyzed population differ from the full randomized population. The registry record does not report a parallel estimate for another analysis population.

Second, the registry states that no missing data were imputed, but does not report the amount or reasons for missing endpoint data. Without those details or sensitivity analyses, the effect of missing measurements on the estimate cannot be assessed from the registry record. The model used to calculate the least-squares means and net mean difference is also not reported, limiting independent reproduction and evaluation of modeling assumptions.

Third, although the comparison is categorized as non-inferiority or equivalence, the margin or equivalence bounds are not reported. The registry does not provide a p-value for the comparison or the sponsor’s statistical method. Consequently, the reported net mean difference and confidence interval cannot be fully assessed against the intended decision criterion.

Finally, the registry reports adverse-event counts without the corresponding at-risk denominators in the posted adverse-event information. This prevents those counts from being interpreted as comparative risks from the available details. The record lists serious-event counts of 10 for sitagliptin 100 mg and 8 for metformin, and other adverse-event counts of 19 and 57, respectively. It also lists selected serious-event terms with percentages, including Any Cardiac disorders at 0.6% and Any Infections and infestations at 0.4%. These entries do not resolve the missing denominator or provide a complete account of safety analyses.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Sitagliptin Comparative Study in Patients With Type 2 Diabetes (0431-049) — NCT00449930.
• Registry record last updated 2017-05-11; status last verified 2017-04.
• Results first posted 2009-08-11.
• 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.

18. Record Summary

The registry describes a completed, randomized, double-blind, parallel-group Phase 3 study of sitagliptin versus metformin in patients with type 2 diabetes and inadequate glycemic control. Enrollment was 1050. The recorded study start date was 2007-03-01, and primary completion was 2008-07-25.

The primary endpoint was change from baseline in HbA1c at Week 24. In the per-protocol population, the least-squares mean change was -0.43 percent for sitagliptin 100 mg and -0.57 percent for metformin. The registry reports a net mean difference of 0.14, with a 95% confidence interval from 0.06 to 0.21, and categorizes the analysis as non-inferiority or equivalence.

The registry states that no missing data were imputed. It does not report the non-inferiority or equivalence margin, the statistical model, or a separate analysis for all randomized participants. Four secondary outcomes concern selected adverse experiences. The posted results should be interpreted within the stated per-protocol population and the limits of the statistical details recorded.