Study of Multiple Doses of Saxagliptin (BMS-477118): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study of Multiple Doses of Saxagliptin (BMS-477118) (NCT00950599), a randomized phase 2 study. The primary outcome is Analysis of Test for Positive Efficacy Trend in Change From Baseline in Hemoglobin A1c (A1C) at Week 12 in the 0-40 mg Cohort.

2. Clinical Question

The registry describes a multicenter, randomized, double-blind, placebo-controlled phase 2 trial of saxagliptin as monotherapy. The primary outcome was change from baseline in hemoglobin A1c (A1C), reported as a percentage of glycosylated hemoglobins. It focused on a dose trend rather than only one prespecified dose-versus-placebo contrast. The registry also lists a separate 100 mg cohort, but that cohort is not part of the primary 0-40 mg analysis.

The question is about the pattern in A1C change across dose groups over 12 weeks. The posted arm summaries describe mean changes, while the registry's reported analysis is a Kruskal-Wallis test across the six groups in the 0-40 mg cohort. Unadjusted recalculations also compare each saxagliptin dose with placebo. Those comparisons address a different question from a dose-trend test and should not be treated as a replacement for the sponsor's reported analysis.

3. Trial Design

This was a randomized, parallel-group phase 2 trial, with quadruple masking recorded in the registry. Its official title describes a multicenter, double-blind, placebo-controlled evaluation of saxagliptin as monotherapy in people with type 2 diabetes mellitus who had inadequate glycemic control. The registry lists total enrollment of 423 and trial status as completed.

The primary outcome concerned the 0-40 mg cohort, which included five saxagliptin dose groups and a placebo group. The 100 mg saxagliptin and corresponding placebo groups appear in a separate 0 & 100 mg cohort for some secondary outcomes. They are not included in the primary efficacy-trend analysis. The intervention list contains six saxagliptin entries and one placebo entry, consistent with the listed dose groups, but the record does not provide additional treatment administration details.

The primary timeframe was baseline to Week 12. The primary analysis population is described as randomized participants who had a baseline measurement and at least 1 post-baseline measurement; change to Week 12 was analyzed using Last Observation Carried Forward (LOCF). The registry lists 84 outcome measures and 2 reported analyses. It names additional outcomes involving A1C, fasting serum glucose, fructosamine, insulin, and C-peptide, with several measured at Week 6 and some at Week 12.

The registry record does not report the trial's original sample-size justification, detailed allocation procedure, or full protocol. The dates recorded for the study are not necessary to interpret the reported outcome results, and no further design assumptions should be inferred from the posted arm summaries.

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.

Only the first 6 of 10 groups are shown.

Double-Blind Treatment Period

Reasons for not completing this period:

Follow-Up Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Randomized participants. To be included in analysis of positive efficacy trend in change from baseline to Week 12 Last Observation Carried Forward (LOCF), participants must have had a baseline and at least 1 post-baseline measurement.

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.

Only the first 6 of 9 groups are shown.

6. Primary Endpoint

The registry lists the primary outcome as Analysis of Test for Positive Efficacy Trend in Change From Baseline in Hemoglobin A1c (A1C) at Week 12 in the 0-40 mg Cohort, measured over Baseline, Week 12.

The primary endpoint is the change from baseline in hemoglobin A1c (A1C) at Week 12 in the 0-40 mg cohort. A1C is reported as a percentage of glycosylated hemoglobins. The registry specifies an adjusted mean change and identifies the measurement timeframe as baseline and Week 12. The posted arm results are means with standard deviations.

Change from baseline expresses how the measured value at follow-up differs from the participant's starting value. In general, A1C is used in diabetes research as a measure of glycemic control over time; its use makes it possible to compare the average change across randomized dose groups and placebo. The study's stated primary aim was to assess whether there was a positive efficacy trend among doses, not simply to report a single dose's result.

The registry says that participants needed both a baseline and at least 1 post-baseline measurement to enter the primary analysis, and that Week 12 change was evaluated using LOCF. In general, LOCF carries a participant's last available post-baseline value forward to the specified endpoint when a later measurement is unavailable. The detailed handling of missing measurements beyond this description is not reported in the registry record.

Interpretation therefore depends on distinguishing the mean change in each arm, the sponsor's reported overall test, and the unadjusted dose-versus-placebo recalculations. They describe related but not identical statistical questions.

7. Results Posted in the Registry

• Saxagliptin 2.5 mg (0-40 mg Cohort): mean -0.67 versus -0.29 on control; difference -0.38 (95% CI -0.68 to -0.08), standardized effect d = -0.47; z-test p = 0.013.
• Saxagliptin 5 mg (0-40 mg Cohort): mean -0.95 versus -0.29 on control; difference -0.66 (95% CI -1.02 to -0.30), standardized effect d = -0.75; z-test p = <0.001.
• Saxagliptin 10 mg (0-40 mg Cohort): mean -0.81 versus -0.29 on control; difference -0.52 (95% CI -0.83 to -0.21), standardized effect d = -0.60; z-test p = <0.001.
• Saxagliptin 20 mg (0-40 mg Cohort): mean -0.74 versus -0.29 on control; difference -0.45 (95% CI -0.78 to -0.12), standardized effect d = -0.52; z-test p = 0.007.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Change From Baseline in A1C at Week 6 in the 0-40 and 0 & 100 mg Cohorts

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Saxagliptin 2.5 mg (0-40 mg Cohort): mean -0.71 versus -0.33 on control; difference -0.38 (95% CI -0.62 to -0.14), standardized effect d = -0.63; z-test p = 0.002.
  • Saxagliptin 5 mg (0-40 mg Cohort): mean -0.72 versus -0.33 on control; difference -0.39 (95% CI -0.64 to -0.14), standardized effect d = -0.63; z-test p = 0.002.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Saxagliptin 2.5 mg (0-40 mg Cohort) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Saxagliptin 5 mg (0-40 mg Cohort) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change From Baseline in Fasting Serum Glucose (FSG) at Week 6 in the 0-40 and 0 & 100 mg Cohorts

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Saxagliptin 2.5 mg (0-40 mg Cohort): mean -10.71 versus -6.20 on control; difference -4.51 (95% CI -14.15 to 5.13), standardized effect d = -0.17; z-test p = 0.359.
  • Saxagliptin 5 mg (0-40 mg Cohort): mean -18.63 versus -6.20 on control; difference -12.43 (95% CI -22.27 to -2.59), standardized effect d = -0.48; z-test p = 0.013.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Saxagliptin 2.5 mg (0-40 mg Cohort) and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Saxagliptin 5 mg (0-40 mg Cohort) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change From Baseline in FSG at Week 12 in the 0-40 mg Cohort

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard errors.
• Result:
  • Saxagliptin 2.5 mg (0-40 mg Cohort): mean -10.85 versus 2.81 on control; difference -13.66 (95% CI -25.75 to -1.57), standardized effect d = -0.42; z-test p = 0.027.
  • Saxagliptin 5 mg (0-40 mg Cohort): mean -21.68 versus 2.81 on control; difference -24.49 (95% CI -36.86 to -12.12), standardized effect d = -0.75; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Saxagliptin 2.5 mg (0-40 mg Cohort) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Saxagliptin 5 mg (0-40 mg Cohort) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes broaden the measurement beyond Week 12 A1C, including A1C at Week 6 and measures of fasting serum glucose, fructosamine, insulin, and C-peptide. Some outcomes are listed for both the 0-40 mg and 0 & 100 mg cohorts, while others concern only the 0-40 mg cohort. This gives context on which additional measures were assessed, but it does not make them equivalent to the primary endpoint.

The numerical secondary summaries included in the registry record show that arm patterns can differ by outcome and timepoint. That variation is important statistically: related measures may not provide identical estimates, and changes at Week 6 do not answer the same question as the primary Week 12 A1C analysis. The secondary results can provide descriptive context, but they do not overturn or replace the sponsor's reported primary analysis.

The registry lists 82 secondary outcomes as posted, but the detailed results for fructosamine, insulin, and C-peptide are not reported in the registry record. The available secondary summaries do not establish a single consistent pattern across every secondary measure, dose, cohort, and assessment time. Conclusions should therefore remain tied to the individual measures and analyses actually reported, rather than treating the secondary outcomes as one combined result.

In general, unadjusted secondary findings carry less weight when many outcomes, doses, and timepoints are examined. Each additional comparison creates more opportunity for chance findings, and the registry does not report a multiplicity strategy for the secondary analyses. These outcomes are most appropriately read as supportive or exploratory context, with the primary endpoint and its sponsor-reported analysis remaining distinct.

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

• Change From Baseline in A1C at Week 6 in the 0-40 and 0 & 100 mg Cohorts
• Change From Baseline in Fasting Serum Glucose (FSG) at Week 6 in the 0-40 and 0 & 100 mg Cohorts
• Change From Baseline in FSG at Week 12 in the 0-40 mg Cohort
• Change From Baseline in Fructosamine at Week 6 in the 0-40 and 0 & 100 mg Cohorts
• Change From Baseline in Fructosamine at Week 12 in the 0-40 mg Cohort
• Change From Baseline in Insulin at Week 6 in the 0-40 and 0 & 100 mg Cohorts
• Change From Baseline in Insulin at Week 12 in the 0-40 mg Cohort
• Change From Baseline in C-peptide at Week 6 in the 0-40 and 0 & 100 mg Cohorts

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint is continuous: each participant has an A1C change value, and the posted arm summaries report means and standard deviations. A mean describes the center of a set of measured values. A standard deviation describes how spread out individual values are around that mean; it is not the same as a standard error, which describes uncertainty in an estimated mean. In general, the standard deviation is useful for understanding variation among participants, while a confidence interval is useful for understanding the precision of an estimated contrast.

What does the reported test assess?

The registry lists a Kruskal-Wallis test across the six groups in the 0-40 mg cohort, with p = 0.989. In general, the Kruskal-Wallis test is a rank-based method for assessing whether outcomes differ across multiple independent groups. It is not specifically a test for a linear dose trend, nor does it directly provide a dose-versus-placebo mean difference. The registry supplies no effect estimate or confidence interval for this analysis, so the p-value alone does not show the size or precision of a particular contrast.

A p-value describes how compatible the observed data are with a specified statistical null model, under the assumptions of the test. In general, a small p-value is evidence against that null model, but it is not the probability that the null hypothesis is true and does not measure the clinical importance of an estimate. The registry's reported p-value belongs to the six-group Kruskal-Wallis analysis. It should not be interpreted as the p-value for each dose-versus-placebo comparison.

What do the recalculated comparisons add?

The recalculations use posted summary statistics to estimate unadjusted mean differences between each saxagliptin dose and placebo. The estimate is expressed as saxagliptin mean minus placebo mean, so negative estimates indicate a more negative mean change in the saxagliptin group. For example, the 5 mg contrast is -0.66 percentage points, with a 95% confidence interval from -1.017 to -0.303. The comparison is unadjusted: it does not include a correction for evaluating multiple doses.

These are simple comparisons of arm summaries rather than the sponsor's reported adjusted trend analysis. The registry describes the primary outcome as adjusted mean change, but the recalculations are unadjusted. The record does not report enough detail to establish whether the sponsor used the same model, covariate adjustment, or estimation procedure represented by these recalculated contrasts. Because a discrepancy is flagged, the sponsor's own analysis takes precedence for the trial's reported result.

Why is multiplicity relevant?

Multiplicity arises when several statistical questions are assessed using the same study data. Here, the recalculations compare five saxagliptin doses with one placebo group. In general, running multiple tests without adjustment increases the chance that at least one result appears noteworthy by chance when all corresponding null hypotheses hold. No adjustment for multiplicity is reflected in the listed pairwise p-values. The registry record does not report whether or how the sponsor addressed multiplicity in the primary analysis or across all outcomes.

The primary trend question and the set of pairwise contrasts are not interchangeable. A trend test evaluates a pattern across doses; a pairwise comparison evaluates a difference for one dose against placebo. A non-monotone sequence of arm means can yield different impressions under those two approaches. In general, an analysis should be interpreted according to its stated estimand and statistical method, rather than substituting a collection of other comparisons for the planned question.

How should missing measurements be understood?

The primary analysis population is defined as randomized participants with a baseline and at least 1 post-baseline measurement, and Week 12 change is described as using LOCF. In general, LOCF fills a missing later value with the last observed post-baseline value. This makes an analysis possible for participants without a Week 12 measurement, but it relies on assumptions about how the outcome behaves after the last observation. LOCF can misrepresent uncertainty when observations are carried forward as though they were actual endpoint measurements. The registry record does not report the amount or pattern of missing data, alternative missing-data analyses, or the assumptions behind the sponsor's approach.

All these methods answer statistical questions about the recorded trial outcomes. They do not, on their own, establish clinical importance or resolve uncertainty about design and analysis details that the registry does not report.

11. Confidence Intervals and Precision

The recalculated dose-versus-placebo estimates are accompanied by 95% confidence intervals. For example, the 2.5 mg comparison is -0.38 percentage points, with an interval from -0.680 to -0.080; the 5 mg comparison is -0.66, with an interval from -1.017 to -0.303. The other three dose contrasts also have intervals entirely below zero. These are unadjusted intervals based on posted summary statistics, not intervals for the sponsor's overall trend analysis.

In general, a confidence interval gives a range of estimates compatible with the data and the method used, while reflecting sampling uncertainty. A narrower interval indicates greater precision than a wider interval for the same kind of estimate. The interval does not capture every source of uncertainty: it does not account for potential bias, unreported analysis choices, or multiplicity unless the method explicitly addresses those issues. The registry record does not report a confidence interval for the sponsor's Kruskal-Wallis result.

The reported arm standard deviations show participant-to-participant variability, while the recalculated intervals express uncertainty in the differences between arm means. These quantities should not be confused. Likewise, a confidence interval for a mean difference does not establish that a difference is clinically important. No threshold for a clinically meaningful A1C difference is reported in the registry record.

Because the recalculated intervals are unadjusted, their coverage should not be read as simultaneous coverage across all dose comparisons. The multiplicity of the five contrasts adds another reason to treat them as exploratory. The sponsor's overall analysis remains the primary reported result, and its uncertainty is difficult to characterize from the registry because no effect estimate or interval accompanies the listed p-value.

12. Planning a Similar Trial: Sample Size

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

The planning figures are participants per arm for the two-arm comparisons; they are not total event counts. The planning illustration assumes a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It uses the observed effect as a reference and considers detecting 100%, 75%, or 50% of that observed effect. These are planning assumptions for the illustration, not the trial's original design assumptions, which are not reported in the registry record.

The 80% power column gives the estimated participants per arm for a future comparison to have an 80% chance of detecting the stated effect, if the planning assumptions and effect size are correct. The 90% power column gives the corresponding estimate for a 90% chance. In general, higher target power requires more participants for a fixed effect and variability. The figures are based on a two-arm planning model and should not be interpreted as a reconstruction of the trial's full multi-arm design.

The entries for 75% and 50% of the observed effect illustrate that smaller effects need more participants to detect under the same assumptions. The observed effect may be imprecisely estimated, so planning from it can yield optimistic or unstable figures, particularly when the original study is small. These numbers are not observed power; they are forward-looking planning illustrations. They do not change the interpretation of the sponsor's reported analysis or account for any design features not included in the listed assumptions.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = -0.380; pooled standard deviation σ = 0.811.
2. n = 2 × (1.96 + 0.842)² × 0.811² / -0.380² = 71.5, rounded up to 72 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- -0.38                      # observed difference in means
sigma <- 0.811052036188434                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(950599)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 72 (80% power), 96 (90% power)
 75% of observed effect: n per arm = 128 (80% power), 171 (90% power)
 50% of observed effect: n per arm = 287 (80% power), 383 (90% power)
power.t.test: 72.5 
Simulated power at n = 72 per arm: 0.800 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The registry's reported primary analysis is a Kruskal-Wallis test across the six groups in the 0-40 mg cohort, with p = 0.989. It does not list a corresponding effect estimate or confidence interval. This is the sponsor's reported result and takes precedence over recalculations from arm-level summaries.

The recalculated results compare each dose with placebo using unadjusted mean differences and confidence intervals. Those comparisons target individual dose contrasts, whereas the reported test evaluates the groups together using a rank-based method. They also do not reproduce the stated adjusted mean-change trend analysis. Consequently, the two results are not directly interchangeable. The discrepancy between the reported analysis and the recalculations is noted, but the registry record does not explain its source.

The recalculated contrasts can help readers understand the scale and precision of the posted arm-summary differences. They should not be presented as the sponsor's primary test or as a substitute for its result. The registry record does not report enough methodological detail to reconcile the analysis approaches or determine whether the recalculation matches any prespecified sponsor model.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists 84 outcome measures and 2 reported analyses. It also includes several doses, multiple cohorts, and multiple assessment times. This creates a setting in which multiple statistical questions may arise. In general, the more tests conducted, the greater the chance of at least one apparently unusual result by chance if no adjustment is made. The registry record does not report a multiplicity plan for the primary dose comparisons or the secondary outcomes.

The recalculated dose-versus-placebo tests and intervals are explicitly unadjusted. Their p-values and confidence intervals describe individual comparisons, not a familywise analysis across all doses. The reported Kruskal-Wallis test is a separate overall analysis, but the registry does not provide enough detail to determine how it relates to any planned trend test or the wider set of outcomes. The sponsor's statistical analysis plan is not reported in the registry record.

Missing-data reporting is also limited. The primary analysis required a baseline and at least 1 post-baseline measurement and used LOCF for the Week 12 outcome. The number of participants without Week 12 values, reasons for missingness, and sensitivity analyses are not reported in the registry record. In general, assumptions about missing outcomes can affect estimates and precision, and carrying the last observed value forward does not remove that uncertainty.

The registry identifies the primary analysis population as randomized participants meeting the measurement requirement. Further details about analysis populations, exclusions after randomization, handling of protocol deviations, or alternative analysis sets are not reported in the registry record. These omissions limit assessment of whether the posted results are robust to different assumptions or analysis choices.

16. 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 serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: rectocele (1.9%), gastroenteritis (1.9%), cystocele (1.9%), pneumonia (1.8%), appendicitis (1.6%).

The registry reports serious adverse events and other adverse events by group, with the number at risk and number affected. In the 0-40 mg cohort, serious adverse events affected 1 participant in the 10 mg group, 1 in the 2.5 mg group, 2 in the 20 mg group, none in the 5 mg or 40 mg groups, and 2 in the placebo group. The corresponding numbers at risk were 63, 55, 54, 47, 52, and 67, respectively.

For other adverse events, the affected counts were 36 for 10 mg, 35 for 2.5 mg, 33 for 20 mg, 23 for 5 mg, 27 for 40 mg, and 38 for placebo. The registry also lists the 100 mg and placebo groups from the separate 0 & 100 mg cohort. These figures are counts, not adjusted comparisons, and the record does not provide a statistical test for differences between groups.

Deaths are not reported in the registry record. Serious-event terms listed include Rectocele, Gastroenteritis, Cystocele, Pneumonia, Appendicitis, and Umbilical Hernia, with the percentages recorded for those terms. These summaries do not establish comparative safety or explain the clinical context, timing, or relatedness of individual events.

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

The registry records a randomized trial of multiple saxagliptin doses and a placebo group, with Week 12 change in A1C as the primary endpoint in the 0-40 mg cohort. The posted means describe the average changes observed in those groups. The sponsor's reported primary analysis is a Kruskal-Wallis test with p = 0.989; no effect estimate or confidence interval is reported for that analysis.

Unadjusted recalculations show dose-specific mean differences relative to placebo, but they answer a narrower question and do not replace the sponsor's overall analysis. Their results also do not account for testing multiple dose comparisons. The overall pattern of arm means is not steadily ordered across doses, so the descriptive values alone do not demonstrate a dose-response trend.

In general, statistical evidence and clinical importance are different considerations. A mean difference and its confidence interval describe an estimated contrast and its uncertainty; they do not by themselves say whether that contrast matters in practice. A clinically meaningful threshold is not reported in the registry record. Nor do the reported group summaries establish how results would generalize beyond the participants, outcome timeframe, and analysis population described.

The results do not establish a conclusion about treatment beyond the scope of the trial's posted data. Missing-data details, the original sample-size justification, the full statistical plan, and a reconciliation of the reported analysis with the recalculations are not reported in the registry record. Those gaps limit how confidently readers can assess the statistical methods and the broader implications of the findings.

18. Why This Trial Matters Statistically

This trial illustrates why the statistical question matters as much as the outcome measure. The stated primary objective concerns a positive efficacy trend across multiple doses, while the recalculations compare each dose separately with placebo. A trend analysis and a set of pairwise contrasts summarize different features of the data. They should not be conflated simply because both involve the same endpoint and groups.

It also illustrates how a collection of mean changes can appear suggestive while the sponsor-reported overall test gives a very different summary. The registry lists a Kruskal-Wallis result, not a trend estimate or interval. The dose-specific recalculations are unadjusted and should be viewed in light of the multiple comparisons. Identifying the analysis actually reported and its scope is essential for a balanced reading.

Finally, the record makes visible important boundaries of registry-based interpretation. The endpoint definition and analysis population are stated, but detailed information about multiplicity, missing data, and the sponsor's statistical plan is not reported. The sample-size figures are planning illustrations based on observed effects, not the trial's original assumptions. Recognizing these distinctions helps separate observed summaries, statistical tests, and future planning from one another.

19. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry's primary analysis result is a Kruskal-Wallis p-value without a corresponding effect estimate or confidence interval. That limits interpretation of the size and precision of the sponsor-reported result. The recalculated mean differences are not the same analysis: they are unadjusted pairwise comparisons derived from arm summaries, while the primary outcome description specifies adjusted mean change and an efficacy-trend question.

The record does not explain the discrepancy between the reported overall test and the recalculated contrasts. It also does not report the original sample-size justification, full statistical analysis plan, multiplicity procedure, number or pattern of missing Week 12 measurements, or sensitivity analyses for LOCF. These missing details limit evaluation of the robustness of the primary and secondary findings.

The arm summaries and unadjusted intervals do not account for potential multiplicity across several doses, many outcome measures, and assessment times. The record lists 84 outcome measures, but detailed results for some named secondary measures are not reported in the registry record. A full pattern across every secondary endpoint therefore cannot be assessed from the stated results.

The safety summaries give counts and terms but do not provide statistical comparisons or detail about event timing and relatedness. Deaths are not reported in the registry record. Finally, the trial's primary endpoint is limited to A1C change through Week 12 in the specified cohort and analysis population. Conclusions beyond that measure, timeframe, or group are not established by these registry results.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Study of Multiple Doses of Saxagliptin (BMS-477118) — NCT00950599.
• Registry record last updated 2015-04-03; status last verified 2015-03.
• Results first posted 2010-02-23.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

23. Record Summary

NCT00950599 is a completed, randomized, parallel-group phase 2 trial of saxagliptin as monotherapy in participants with type 2 diabetes mellitus and inadequate glycemic control. The registry lists enrollment of 423 and a primary assessment of A1C change from baseline to Week 12 in the 0-40 mg cohort.

The sponsor-reported primary analysis is a Kruskal-Wallis test across the five saxagliptin dose groups and placebo, with p = 0.989. Unadjusted recalculations provide dose-versus-placebo mean differences, but they do not replace the sponsor's analysis and do not account for multiplicity. Detailed information about the original design assumptions, missing-data sensitivity analyses, and multiplicity strategy is not reported in the registry record.