An Efficacy Study of Canagliflozin or Sitagliptin to Determine Glucose Variability in Mexican Participants With Type 2 Diabetes Mellitus Inadequately Controlled on Metformin: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for An Efficacy Study of Canagliflozin or Sitagliptin to Determine Glucose Variability in Mexican Participants With Type 2 Diabetes Mellitus Inadequately Controlled on Metformin (NCT03267576), a randomized phase 4 study. The primary outcome is Change From Baseline in Glycemic Coefficient of Variation (CV) in Treatment Period 1.

2. Clinical Question

3. Trial Design

This was a Phase 4, randomized, double-blind, controlled crossover study. The registry lists an enrollment of 64 and describes Mexican participants with type 2 diabetes inadequately controlled on metformin. Participants received canagliflozin 300 mg or sitagliptin 100 mg in addition to metformin, then crossed over to the other active treatment. Treatment periods were 4 weeks each. The sequence descriptions identify the regimens as Treatment Sequence AB and Treatment Sequence BA; the description for BA specifies treatment B first and treatment A second.

The BA sequence description specifies a metformin-only washout lasting at least 16 days, from days 28 to 43, between treatment periods. The AB description is truncated in the registry record, so its full schedule is not reported in the registry record. The posted primary-outcome timeframe is baseline up to the end of Treatment Period 1, identified as days 22 to 27. CGM was conducted over a 6-day period to obtain a continuous 72-hour reading at baseline and after each active treatment.

The primary analysis population is described as an intent-to-treat analysis set, but its registry definition requires receipt of at least one dose and successful CGM recordings at baseline and after each active treatment. For the reported primary outcome, the sequence groups include 31 participants in Treatment Sequence AB and 32 in Treatment Sequence BA. These are the group sizes for this outcome, not a statement about how many randomized participants had complete data for every other outcome.

The registry lists the study as completed, with a start date of 2017-10-27 and primary completion date of 2018-10-01. Details of randomization generation, allocation concealment, adherence, and the complete protocol are not reported in the registry record. Although the record calls the study double-blind, it does not specify which trial roles were masked.

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: A total of 64 participants were enrolled in this study but 1 participant has withdrawn consent before assigning to any study treatment hence, 63 participants were randomized to study treatment arms.

Period 1: Day 0 to 27

Washout Period: 16 Days

Period 2: Day 44 to 71

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-treat (ITT) analysis set included all participants who received at least one dose of medication and in whom CGM recordings at baseline and after each active treatment were successful (greater than \[\>\] 70% of tracings available).

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: Intent-to-treat (ITT) analysis set included all participants who received at least one dose of medication and in whom CGM recordings at baseline and after each active treatment were successful (greater than \[\>\] 70 percent (%) of tracings available).

6. Primary Endpoint

The registry lists the primary outcome as Change From Baseline in Glycemic Coefficient of Variation (CV) in Treatment Period 1, measured over Baseline up to End of Treatment Period 1 (Days 22 to 27).

The primary endpoint is change from baseline in glycemic coefficient of variation during Treatment Period 1. Continuous glucose monitoring captured glucose readings, and the coefficient of variation was calculated by dividing the standard deviation of the readings by their mean. It is expressed as a percentage of the coefficient of variation. The endpoint timeframe runs from baseline to the end of Treatment Period 1, days 22 to 27.

As a general statistical concept, the coefficient of variation expresses variability relative to the mean rather than variability in the original measurement units alone. This can help describe how dispersed repeated glucose readings are in relation to their average level. It is distinct from mean glucose: two sets of readings can have different average levels and different degrees of relative variability. The registry also lists glycemic standard deviation and mean glucose profile among the secondary outcomes, alongside fasting, post-prandial, and time-in-range measures.

The registry describes the primary outcome as a mean change and reports standard deviations for the sequence groups. The record’s endpoint description says that participants were analyzed according to the treatment received in Treatment Period 2 as per the sequence reported, while the outcome title and timeframe specify Treatment Period 1. This wording is not fully reconciled in the registry record. The result should therefore be read with attention to the stated endpoint and sequence-group definitions, rather than treated as a clearly documented treatment-specific estimate.

7. Results Posted in the Registry

• Treatment Sequence BA: mean 0.24 versus -0.69 on control; difference 0.93 (95% CI -3.05 to 4.91), standardized effect d = 0.11; z-test p = 0.647.

8. Statistical Interpretation

9. Statistical Methods Explained

The result concerns a continuous outcome: change in a percentage measure of glycemic coefficient of variation. The registry reports means and standard deviations for each sequence group. The comparison is an unadjusted recalculation from those posted summary statistics; it is not described as the sponsor’s adjusted crossover analysis. The reported estimate, confidence interval, and p-value should be understood in that limited context.

What does a mean change represent?

A mean is a measure of the center of a group’s observed values. Here, the values are changes from baseline in the coefficient of variation, rather than raw glucose concentrations. A negative mean indicates an average change below baseline on the measure’s scale; a positive mean indicates an average change above baseline. The sequence-group means are -0.69 for AB and 0.24 for BA. The comparison is stated as BA versus AB, with a mean difference of 0.93. These figures describe group averages and do not imply that every participant changed in the same direction.

The standard deviation describes the dispersion of participant-level values around a group mean. The reported standard deviations are 5.17 for AB and 10.23 for BA. As a general statistical concept, a larger standard deviation indicates more spread in the observed values, not a larger average treatment effect. Variation in outcomes affects how precisely a group mean or mean difference can be estimated.

How should the confidence interval be read?

A confidence interval gives a range of values compatible with the observed data under the method used to calculate it. For the unadjusted mean difference of 0.93, the reported interval extends from -3.054 to 4.914. Because this interval crosses zero, the data do not distinguish the observed difference from no difference at conventional two-sided testing thresholds. The interval also contains values on either side of zero, so the direction of a possible underlying difference is uncertain. A confidence interval is not a guarantee that the true value lies within its endpoints, and its interpretation depends on the analysis assumptions.

The registry gives an unadjusted p-value of 0.647. In general, a p-value describes how unusual data at least as incompatible with a null hypothesis would be under that hypothesis and the specified statistical model. It is not the probability that the null hypothesis is true, nor a measure of clinical importance. Here it is an unadjusted result from the sequence-summary comparison; the registry record does not report a complete model specification for a treatment contrast that accounts for the crossover design.

Why does crossover structure matter?

As a general statistical concept, a crossover trial exposes participants to treatments in different orders. This can allow comparisons that account for person-to-person differences, but the analysis needs to reflect repeated observations within participants and the sequence and period structure. Treatment effects can also be difficult to separate from period effects or lingering effects from a preceding treatment if those possibilities are not addressed. Whether and how these features were modeled is not reported in the registry record.

The posted primary summaries are organized by Treatment Sequence AB and Treatment Sequence BA. The summary comparison is therefore not, by itself, a fully specified within-participant treatment contrast. The endpoint title and timeframe identify Treatment Period 1, while the endpoint description says participants were analyzed according to treatment received in Treatment Period 2 as per the sequence. That wording adds uncertainty about the exact analysis definition. A reader should not assume an unreported crossover model or adjustment.

What does the coefficient of variation add?

The registry defines coefficient of variation by dividing the standard deviation of glucose readings by the mean of the corresponding readings. In general, this expresses variability relative to the average level. It complements measures of mean glucose or absolute standard deviation, which describe different aspects of a glucose profile. A change in the coefficient of variation does not by itself establish a change in mean glucose, time in a target range, or any other secondary measure. Those outcomes require their own reported estimates and uncertainty.

CGM recordings were obtained at baseline and after each active treatment. The primary analysis set required successful recordings at baseline and after each active treatment, with more than 70% of tracings available, as well as receipt of at least one medication dose. This definition describes who qualified for the primary analysis, but the registry record does not explain how missing recordings or incomplete measurements were handled outside that eligibility rule. The detailed statistical plan, including model terms, estimation method, and sensitivity analyses, is not reported in the registry record.

10. Confidence Intervals and Precision

The confidence interval around the unadjusted mean difference is broad relative to the point estimate: it ranges from -3.054 to 4.914 around an estimate of 0.93. It includes zero and allows for a possible difference favoring either sequence in the numerical direction of the outcome. The result therefore does not precisely identify the direction or size of the underlying sequence-group difference.

As a general statistical concept, precision refers to how narrowly an estimate is bounded by its uncertainty. A confidence interval’s width depends on factors such as sample size and variability, as well as the analysis method. The posted standard deviations are 5.17 for AB and 10.23 for BA, so there is meaningful dispersion in the observed change values, particularly in the BA sequence group. The registry record does not provide participant-level data or an adjusted crossover-model interval that would clarify how within-participant correlation affects precision.

The interval should not be treated as an equivalence range. Equivalence requires a prespecified margin defining differences small enough to count as practically similar, along with an analysis designed to test that question. Such a margin and an equivalence analysis are not reported in the registry record. The absence of a statistically clear difference is not evidence that the treatments have identical effects.

The confidence interval and p-value describe uncertainty for the unadjusted sequence comparison. They do not account for any unreported multiplicity strategy, nor do they resolve the ambiguity in the endpoint description about treatment period. Precision for other outcomes cannot be inferred from this interval.

11. 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 sample-size figures are planning illustrations based on the observed effect, not the trial’s original design assumptions. They use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set to the stated fraction of the observed effect. The reported figures are participants per arm, not total participants or total events.

The column for 80% power gives the planned number of participants per arm under the stated assumptions to have an 80% chance of detecting the chosen effect if that effect is the true one. The 90% power column gives the corresponding planning number for a 90% chance. For 100% of the observed effect, the figures are 1,204 participants per arm for 80% power and 1,612 participants per arm for 90% power. For 75% of the observed effect, they are 2,141 and 2,866 participants per arm. For 50% of the observed effect, they are 4,816 and 6,447 participants per arm.

Smaller effects require more participants to detect under the same assumptions, because they are harder to distinguish from random variation. These figures are not observed power. They are hypothetical planning results derived from the observed effect and should not be read as evidence about the trial’s chance of detecting an effect before it was conducted. The registry record does not report the original sample-size justification or the assumptions used to design the trial.

The illustration uses an equal-allocation, two-arm planning framework. The study itself is a crossover design with treatment sequences, so the planning setup is not a substitute for a design-specific calculation based on the trial’s analysis model and within-participant information. The registry record does not report such a design-specific planning calculation.

12. 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 = 0.930; pooled standard deviation σ = 8.144.
2. n = 2 × (1.96 + 0.842)² × 8.144² / 0.930² = 1,203.9, rounded up to 1,204 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 <- 0.93                      # observed difference in means
sigma <- 8.14429992764727                     # 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(3267576)
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 = 1,204 (80% power), 1,612 (90% power)
 75% of observed effect: n per arm = 2,141 (80% power), 2,866 (90% power)
 50% of observed effect: n per arm = 4,816 (80% power), 6,447 (90% power)
power.t.test: 1204.8 
Simulated power at n = 1,204 per arm: 0.801 (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.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists 13 secondary outcomes, including changes in glycemic standard deviation and mean 24-hour glucose profile, fasting plasma glucose, post-prandial glucose, and time spent in specified glucose ranges. Considering multiple outcomes can increase the chance that at least one result appears statistically noteworthy by chance if each is tested separately. As a general statistical concept, multiplicity procedures can control or otherwise address that issue.

Whether the trial used a multiplicity adjustment, a prespecified hierarchy of outcomes, or another strategy is not reported in the registry record. The posted primary comparison is identified as unadjusted. The registry record does not provide secondary outcome estimates in the registry record, so their direction, precision, and statistical results cannot be assessed alongside the primary outcome.

The primary analysis population is described as an intent-to-treat analysis set, but the stated eligibility criteria require at least one dose and successful CGM recordings at baseline and after each active treatment, with more than 70% of tracings available. This is a registry definition of the analyzed population; the record does not explain how it relates to all randomized participants or how participants with incomplete recordings were handled. The numbers excluded, reasons for missing data, and any imputation or sensitivity analyses are not reported in the registry record.

The crossover structure also makes the analysis population and data completeness relevant: the primary outcome description requires successful CGM at baseline and after each active treatment. Whether partial-period observations were used, how missing values were handled, or whether analyses were repeated under alternative population definitions is not reported in the registry record. The full statistical analysis plan and detailed model specification are not reported either. These omissions limit assessment of multiplicity, missing-data assumptions, and robustness of the estimate.

14. 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 0% reporting threshold.

Most frequent non-serious adverse event terms

The registry lists serious adverse events for 63 participants at risk in each treatment group. No serious adverse events are reported for canagliflozin 300 mg or sitagliptin 100 mg, and deaths are reported as 0 in each group. Other, non-serious adverse events are reported for 7 participants at risk in the canagliflozin group and 2 participants at risk in the sitagliptin group.

These are registry counts, not rates calculated for exposure time, and they should not be converted into percentages without additional information. The figures are presented by treatment group and do not establish whether events occurred in the same or different participants across periods. The registry record does not report event descriptions for the non-serious events in the registry record. It also does not provide enough information to interpret event duration, severity beyond the serious/non-serious categories, relationship to study treatment, or exposure-adjusted comparisons.

The absence of reported serious events in these group summaries is not proof that serious events cannot occur. Safety counts from this study should be read as descriptive observations within the reported registry record, not as a comprehensive comparison of safety.

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

The primary result is a comparison of mean change in glycemic coefficient of variation between Treatment Sequence AB and Treatment Sequence BA. The means were -0.69 and 0.24, respectively, and the unadjusted recalculated difference for BA versus AB was 0.93. Its 95% confidence interval, from -3.054 to 4.914, includes zero, and the unadjusted p-value is 0.647. These findings describe the posted sequence summaries and show substantial uncertainty about the size and direction of the difference.

The result does not establish that canagliflozin is better, worse, or equivalent to sitagliptin for this outcome. The comparison is organized by treatment sequence, and the registry does not provide a clearly documented adjusted crossover analysis in the registry record. Its endpoint description also refers to treatment received in Treatment Period 2, while its title and timeframe refer to Treatment Period 1. This makes the precise treatment contrast difficult to interpret from the summary alone.

The endpoint is a measure of relative glucose variability derived from CGM. It is not the same as mean glucose, a clinical event, or a patient-reported outcome. No conclusion about those other outcomes follows from the primary endpoint. Although secondary measures are listed, their results are not included in the registry record.

The safety summaries report no serious adverse events and no deaths in either treatment group, alongside other adverse events in both groups. These are descriptive counts, not proof of comparative safety. The registry record does not provide the detail needed to evaluate event timing, exposure, or causal attribution.

Overall, the posted data support a cautious account of the observed sequence-group means and their uncertainty. They do not, by themselves, answer every treatment-effect question that might be asked about a crossover trial.

16. Why This Trial Matters Statistically

This trial is statistically informative as an example of the distinction between a treatment comparison and a sequence comparison. In crossover studies, participants receive treatments in an order, and analysis needs to use that structure. The registry’s primary result is presented by treatment sequence, while the endpoint description does not clearly align its reference to treatment period with the title and timeframe. That makes the model and estimand important: they determine whether the estimate represents an effect of treatment, a difference between sequences, or a mixture of treatment and period-related features.

The outcome also illustrates why the measure used matters. Coefficient of variation expresses glucose variability relative to the mean glucose level, whereas standard deviation and mean glucose describe other properties of the readings. A result for one measure should not be treated as a result for all measures of glucose control. The registry lists several secondary glucose outcomes, but the primary result cannot stand in for their estimates.

The wide confidence interval is important alongside the point estimate and p-value. A p-value alone does not show how large a compatible difference might be, while the interval makes clear that the data allow values on either side of zero. The planning illustrations likewise show how the number of participants required depends on the effect a study is intended to detect and on the planning assumptions. They are not evidence about the trial’s original design rationale.

Finally, the definition of the primary analysis set requires successful CGM recordings at baseline and after each active treatment. That makes completeness of measurement part of the interpretation, not merely a technical detail. The registry does not report how missing recordings were handled or the full crossover analysis plan. These are useful reminders to assess the endpoint definition, analysis population, uncertainty, and design-specific model together.

17. 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 primary comparison is reported as a simple unadjusted recalculation from the posted sequence-group summaries. The registry record does not report a complete treatment-effect model for the crossover design, including how participant-level pairing, period, sequence, or possible carryover was handled. The reported comparison therefore cannot be assumed to be an adjusted within-participant contrast.

The primary endpoint description has an ambiguity: the title and timeframe specify Treatment Period 1, but the description says participants were analyzed according to treatment received in Treatment Period 2 as per the sequence. The registry entries do not resolve that wording. The AB sequence description is also truncated, so its complete dosing schedule is not reported in the registry record.

The analysis population is described as intent-to-treat but is restricted to participants with at least one dose and successful CGM recordings at baseline and after each active treatment, with more than 70% of tracings available. The registry record does not report the number of randomized participants excluded from this analysis, reasons for exclusion, or how missing and incomplete CGM data were treated.

The registry does not report the full statistical analysis plan, multiplicity adjustment, sample-size justification, or sensitivity analyses. Secondary outcome titles are listed, but detailed secondary results are not part of these registry entries. Safety information is limited to counts by treatment group and does not include event descriptions, timing, or treatment attribution. These limitations constrain interpretation of both treatment effects and safety.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: An Efficacy Study of Canagliflozin or Sitagliptin to Determine Glucose Variability in Mexican Participants With Type 2 Diabetes Mellitus Inadequately Controlled on Metformin — NCT03267576.
• Registry record last updated 2019-11-29; status last verified 2019-11.
• Results first posted 2019-11-29.
• 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.

21. Record Summary

NCT03267576 is a completed Phase 4 randomized, double-blind crossover study of canagliflozin 300 mg versus sitagliptin 100 mg, each added to stable metformin, in Mexican participants with type 2 diabetes inadequately controlled on metformin. The registry lists enrollment of 64 and a study period from 2017-10-27 to primary completion on 2018-10-01.

The primary outcome was change from baseline in glycemic coefficient of variation during Treatment Period 1, measured using CGM. The posted sequence-group means were -0.69 for AB and 0.24 for BA. The unadjusted recalculated mean difference for BA versus AB was 0.93, with a 95% confidence interval from -3.054 to 4.914 and an unadjusted p-value of 0.647. The estimate is imprecise and does not establish a treatment-specific difference. The registry record does not report the complete crossover analysis plan or original sample-size assumptions.