A Study to Evaluate the Effectiveness, Safety, and Tolerability of Canagliflozin in Combination With Metformin in the Treatment of Patients With Type 2 Diabetes Mellitus With Inadequate Glycemic Control With Diet and Exercise: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study to Evaluate the Effectiveness, Safety, and Tolerability of Canagliflozin in Combination With Metformin in the Treatment of Patients With Type 2 Diabetes Mellitus With Inadequate Glycemic Control With Diet and Exercise (NCT01809327), a randomized phase 3 study. The primary outcome is Change in Glycated Hemoglobin (HbA1c) From Baseline at Week 26.

2. Clinical Question

The study focused on initial combination therapy in people whose diabetes remained inadequately controlled with diet and exercise. Its primary endpoint was a continuous measure of glycemic change, rather than a binary threshold: the analysis compared mean change in HbA1c between groups. The registry reports results for the modified intent-to-treat analysis set and sponsor analyses using a repeated-measures model. Simple unadjusted comparisons can also be calculated from the posted arm means, but those calculations are not the same as the sponsor's model-based results. The registry's reported analyses should therefore be kept distinct from those re-calculations when interpreting the comparisons.

3. Trial Design

The registry describes a completed Phase 3 study with randomized allocation and a parallel-group design. The official title calls it a randomized, double-blind, five-arm study and gives a 26-week treatment period; the masking field records quadruple masking. The record does not specify which people or roles were masked, so further details of masking are not reported in the registry record. The study enrolled 1186 participants and began on 2013-06-04, with primary completion on 2014-12-02.

Participants were assigned to one of five groups: metformin XR, canagliflozin 100 mg, canagliflozin 300 mg, canagliflozin 100 mg plus metformin XR, or canagliflozin 300 mg plus metformin XR. In the monotherapy groups, matching placebo capsules or tablets were used to match the regimens in the other groups. Metformin XR was taken once daily with the evening meal, with doses titrated over 9 weeks. Canagliflozin was administered as a capsule; the combination-arm descriptions place it with the evening meal. Treatment lasted 26 weeks.

The primary endpoint was change in HbA1c from baseline at Week 26. The posted primary arm analyses include 230 participants for metformin XR, 230 for canagliflozin 100 mg, 234 for canagliflozin 300 mg, 235 for canagliflozin 100 mg plus metformin XR, and 236 for canagliflozin 300 mg plus metformin XR. These analyzed counts are not the same as total enrollment, and the registry does not provide a complete account of the reasons participants were not evaluable. Original design assumptions, including the trial's planned sample-size justification and any interim-analysis plan, are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: A total of 1,186 participants were randomly allocated to the 5 treatment arms. All participants received at least 1 dose of study drug and were included in the modified intent-to-treat (mITT) analysis set.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: This analysis was conducted using the modified intent-to-treat analysis set, which included all participants who were randomly assigned to a treatment group and received at least 1 dose of study drug. Here "N" (Number of Participants Analyzed) signifies number of Participants who were evaluable for this outcome measure.

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 in Glycated Hemoglobin (HbA1c) From Baseline at Week 26, measured over Day 1 (Baseline) and Week 26.

The primary endpoint is change in glycated hemoglobin (HbA1c) from baseline to Week 26, reported as a least-squares mean in percentage of hemoglobin. HbA1c reflects glucose bound to hemoglobin, expressed as a proportion of the amount that can be bound. Because it summarizes glycemic exposure over time, it is commonly used as a continuous outcome in diabetes trials. The registry describes comparing the change from baseline among the treatment groups.

A continuous endpoint retains information about the magnitude of change across participants, rather than classifying each participant only as above or below a threshold. The reported arm values are adjusted least-squares means, with standard errors. A negative change indicates that the endpoint value at Week 26 was lower than at baseline, according to the direction used in the registry results. The arm estimates range from -1.3 for metformin XR to -1.78 for canagliflozin 300 mg plus metformin XR. The simple re-calculations compare these posted arm means, while the sponsor's reported comparisons use a mixed model for repeated measures. Those are different statistical summaries and should not be treated as interchangeable.

The endpoint is a measure of laboratory change, not a direct measure of every aspect of diabetes or of longer-term outcomes. Follow-up for the primary endpoint was Week 26; effects beyond that timeframe are not reported in the registry record. The registry's primary analysis population was the modified intent-to-treat set: randomized participants who received at least one dose and were evaluable for the endpoint. The complete definition of evaluability and the handling of missing endpoint measurements are not reported in the registry record.

7. Results Posted in the Registry

• Canagliflozin 100 Milligram (mg): mean -1.37 versus -1.30 on control; difference -0.07 (95% CI -0.27 to 0.13), standardized effect d = -0.07; z-test p = 0.486.
• Canagliflozin 300 mg: mean -1.42 versus -1.30 on control; difference -0.12 (95% CI -0.32 to 0.08), standardized effect d = -0.11; z-test p = 0.229.
• Canagliflozin 100 mg + Metformin XR: mean -1.77 versus -1.30 on control; difference -0.47 (95% CI -0.66 to -0.28), standardized effect d = -0.44; z-test p = <0.001.
• Canagliflozin 300 mg + Metformin XR: mean -1.78 versus -1.30 on control; difference -0.48 (95% CI -0.68 to -0.28), standardized effect d = -0.45; z-test p = <0.001.

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.

Percent Change in Body Weight From Baseline to Week 26

• 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:
  • Canagliflozin 100 Milligram (mg): mean -3.00 versus -2.10 on control; difference -0.90 (95% CI -1.73 to -0.07), standardized effect d = -0.20; z-test p = 0.034.
  • Canagliflozin 300 mg: mean -3.90 versus -2.10 on control; difference -1.80 (95% CI -2.63 to -0.97), standardized effect d = -0.39; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Canagliflozin 100 Milligram (mg) 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 Canagliflozin 300 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Percentage of Participants With Glycated Hemoglobin (HbAIc) Less Than 7 Percent at Week 26

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Canagliflozin 100 Milligram (mg): 38.8% (80/206) versus 43.0% (89/207) on control; risk difference -4.2 percentage points (95% CI -13.6 to 5.3), risk ratio 0.90 (0.72–1.14); unadjusted two-proportion z-test p = 0.390.
  • Canagliflozin 300 mg: 42.8% (92/215) versus 43.0% (89/207) on control; risk difference -0.2 percentage points (95% CI -9.7 to 9.2), risk ratio 1.00 (0.80–1.24); unadjusted two-proportion z-test p = 0.966.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Canagliflozin 100 Milligram (mg) and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Canagliflozin 300 mg and the control. The interval gives the range of effects that remain compatible with the data.

Change in Systolic Blood Pressure From Baseline at Week 26

• 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:
  • Canagliflozin 100 Milligram (mg): mean -2.24 versus -0.33 on control; difference -1.91 (95% CI -3.66 to -0.16), standardized effect d = -0.20; z-test p = 0.032.
  • Canagliflozin 300 mg: mean -2.36 versus -0.33 on control; difference -2.03 (95% CI -3.77 to -0.29), standardized effect d = -0.21; z-test p = 0.022.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Canagliflozin 100 Milligram (mg) 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 Canagliflozin 300 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes broaden the picture beyond HbA1c change. They include body-weight change, the percentage meeting an HbA1c threshold, systolic blood pressure, HDL-C, triglycerides, and treatment-emergent adverse events. The registry posts results for several measures, but does not present a complete inferential analysis for every listed secondary outcome. The general pattern is that several secondary measures have reported group contrasts, while the amount and type of statistical detail vary by outcome.

These measures relate to the primary result by describing other dimensions of the 26-week treatment experience, but they do not replace the primary HbA1c endpoint. Body weight and blood pressure are continuous outcomes, while the HbA1c threshold result is binary. The binary measure asks how many participants crossed a specified threshold, rather than how much HbA1c changed on average. Different outcomes can therefore show different patterns without contradicting the primary result: they measure different features and use different scales.

The registry's selected model-based secondary comparisons report sponsor analyses for body weight, the HbA1c threshold, and systolic blood pressure. Other secondary outcomes are listed, but corresponding comparative results are not all reported in the record. Consequently, the available secondary evidence is incomplete as a family of tests. The registry lists 18 analyses and 7 outcome measures, while the multiplicity plan and hierarchy are not reported in the registry record.

Unadjusted secondary results carry less weight as confirmatory evidence because testing multiple endpoints and contrasts creates more opportunities for chance findings. General explanation: unless a prespecified testing strategy controls that risk, isolated p-values should not be interpreted as if they were the only comparison conducted. The registry does not establish whether all secondary comparisons were adjusted, prespecified, or intended as confirmatory. The secondary findings are therefore most appropriately read as supporting context and as a source of questions for further evaluation, rather than as a set of independent definitive conclusions.

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

• Percent Change in Body Weight From Baseline to Week 26
• Percentage of Participants With Glycated Hemoglobin (HbAIc) Less Than 7 Percent at Week 26
• Change in Systolic Blood Pressure From Baseline at Week 26
• Percent Change in Fasting High-Density Lipoprotein Cholesterol (HDL-C) From Baseline to Week 26
• Percent Change in Triglycerides From Baseline to Week 26
• Number of Participants With Treatment Emergent Adverse Events (AEs)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary arm summaries are least-squares means with standard errors. A mean is a summary of the observed outcome scale; a least-squares mean is a model-based group estimate. The registry identifies the primary method for the sponsor's pairwise comparisons as a mixed model for repeated measures (MMRM). Such a model analyzes measurements taken at multiple visits and estimates treatment-group contrasts while accounting for the repeated structure of the observations. General explanation: repeated measurements from the same participant are related, so treating each visit as independent would fail to represent that within-participant relationship. The registry does not give the complete model specification, covariance structure, adjustment variables, or missing-data assumptions.

The posted primary arm values are -1.3 for metformin XR, -1.37 for canagliflozin 100 mg, -1.42 for canagliflozin 300 mg, -1.77 for canagliflozin 100 mg plus metformin XR, and -1.78 for canagliflozin 300 mg plus metformin XR. Standard errors are 0.071, 0.071, 0.07, 0.069, and 0.07, respectively. A standard error describes uncertainty in an estimated mean; it is not the standard deviation of individual participant outcomes. General explanation: standard errors tend to be smaller when an estimate is based on more information, but they do not describe the spread of individual responses.

The simple comparisons are recalculated from the posted arm summaries. For continuous outcomes, the comparison is the difference between the two arm means. These calculations use an approximate pooled standard deviation derived from the posted standard errors and analyzed sample sizes. The calculation is unadjusted: it does not reproduce the sponsor's repeated-measures model, its handling of measurements over time, or any covariate adjustment. The record flags a discrepancy, so the sponsor's own analysis takes precedence for interpreting the trial's reported comparisons. The simple calculations are useful as a transparent check on the arm summaries, not as a replacement for the registry-reported model analysis.

For the binary secondary outcome, the registry reports the percentage of participants with HbA1c below 7 percent at Week 26 and sponsor analyses as odds ratios from a generalized linear mixed model. An odds ratio compares the odds of meeting the threshold between groups; it is not the same quantity as a difference in percentages. General explanation: odds ratios can be less intuitive than absolute proportions, so both the group percentages and the modeled contrast are relevant to understanding the result. The generalized linear mixed model is the method named in the record; further details about its specification are not reported in the registry record.

Some secondary outcomes are continuous. Their posted arm estimates are least-squares means with standard errors, and the registry lists MMRM analyses for selected comparisons. General explanation: an estimated contrast describes the direction and magnitude of a group difference, while its uncertainty is reflected in a confidence interval and p-value. A p-value measures compatibility of the data with a specified statistical null model under the analysis assumptions; it is not the probability that a treatment effect is true. A small p-value does not measure clinical importance, and a larger p-value does not prove that two treatments are equivalent.

The registry lists 18 analyses and 7 outcome measures. It does not report a complete statistical analysis plan or explain how the family of analyses was defined. When many outcomes or contrasts are examined, the chance of at least one apparently notable result can increase if no adjustment or hierarchy is used. General explanation: multiplicity procedures control or otherwise address this issue by specifying how conclusions are evaluated across multiple tests. Whether and how such procedures were applied to all reported outcomes and comparisons is not reported in the registry record. Accordingly, unadjusted secondary comparisons should be treated as exploratory evidence rather than as a collection of independent confirmatory findings.

The reported primary population is the modified intent-to-treat analysis set, described as participants randomly assigned to a group who received at least one dose and were evaluable for the outcome. The registry does not give the full operational definition of evaluability. It also does not report the number of missing Week 26 measurements by group, reasons for missingness, or the method used to handle missing values in the sponsor's model. General explanation: missing-data assumptions matter because the observed outcomes may not represent participants whose measurements are absent. An MMRM can use available repeated measurements under model assumptions, but the record does not establish the precise assumptions or any sensitivity analyses used for this study.

Finally, the re-calculated p-values and intervals are not identical to sponsor model results because they arise from different calculations and inputs. For instance, the sponsor's MMRM estimate for canagliflozin 100 mg plus metformin XR versus metformin XR is -0.46, whereas the unadjusted re-calculation is -0.47. The sponsor reports p = 0.001, while the simple re-calculation reports p < 0.001. Small differences of this kind can arise from the distinction between a model-based analysis and a summary-statistic approximation; the record does not provide enough detail to attribute a particular difference to a specific model feature. The sponsor's reported analysis remains the primary reference for the trial's own inferential results.

11. Confidence Intervals and Precision

A confidence interval describes a range of values reasonably compatible with the data and the analysis assumptions. It conveys the precision of an estimate: a narrower interval indicates less uncertainty on that scale, while a wider interval leaves more possible values compatible with the analysis. General explanation: a confidence interval is not a guarantee that the true value lies within the particular interval, and it should be interpreted alongside the point estimate and the method used to calculate it.

The sponsor's primary MMRM comparisons are reported with 95% confidence intervals. For canagliflozin 100 mg plus metformin XR versus metformin XR, the estimated least-squares mean difference is -0.46, with a 95% confidence interval from -0.657 to -0.269. For canagliflozin 300 mg plus metformin XR versus metformin XR, the difference is -0.48, with a 95% confidence interval from -0.670 to -0.280. The reported intervals lie below zero, consistent with the negative point estimates on the change scale. Comparisons of combination groups with the corresponding monotherapy groups are also reported with intervals below zero: -0.40, with a 95% confidence interval from -0.594 to -0.207, and -0.36, with a 95% confidence interval from -0.557 to -0.169.

The simple unadjusted re-calculations have their own intervals. For example, the re-calculated difference for canagliflozin 100 mg plus metformin XR versus metformin XR is -0.47, with an interval from -0.664 to -0.276; the corresponding estimate for canagliflozin 300 mg plus metformin XR is -0.48, with an interval from -0.675 to -0.285. These are summary-statistic re-calculations and not the sponsor's MMRM intervals. The discrepancy flag means the sponsor's own analysis takes precedence.

For the monotherapy comparisons against metformin XR, the re-calculated intervals span zero: -0.267 to 0.127 for canagliflozin 100 mg, and -0.315 to 0.075 for canagliflozin 300 mg. Such intervals indicate uncertainty about the direction of the underlying difference under the re-calculation; they do not demonstrate equality. Precision is specific to the endpoint, population, and method. The registry does not report confidence intervals for all arm-level estimates or every secondary result, and the primary findings do not describe uncertainty in longer-term effects.

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 sample-size estimates are planning illustrations based on observed effects, not the trial's original sample-size calculation. The stated assumptions are 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. For these outcomes, the figures are participants per arm, not total participants across the two arms. The sample-size description notes that time-to-event calculations, if present, would instead give total events; the primary and secondary outcomes are not time-to-event outcomes.

The 80% power column gives the estimated participants per arm for a design that would have an 80% probability of detecting the specified effect under the planning assumptions, if that effect and assumptions were correct. The 90% power column represents a higher target probability of detection and consequently generally requires more participants per arm for the same assumed effect. These are design calculations, not evidence about how much power the completed trial had.

Each planning row uses a fraction of the observed effect. Planning for 100% of the observed effect uses the full observed estimate; planning for 75% or 50% uses a smaller assumed effect. Smaller effects are harder to distinguish from random variation, so they require more participants per arm to reach the same target power. This pattern is especially visible when an observed difference is small: the estimated participant requirement can become very large as the assumed effect is reduced. These figures should be read as conditional illustrations, not recommendations or a reconstruction of the sponsor's original design. The original design assumptions and sample-size justification are not reported in the registry record. These planning values are not observed power.

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 = -0.070; pooled standard deviation σ = 1.077.
2. n = 2 × (1.96 + 0.842)² × 1.077² / -0.070² = 3,714.4, rounded up to 3,715 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 <- -0.0700000000000001                      # observed difference in means
sigma <- 1.07676831305532                     # 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(1809327)
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 = 3,715 (80% power), 4,973 (90% power)
 75% of observed effect: n per arm = 6,604 (80% power), 8,840 (90% power)
 50% of observed effect: n per arm = 14,858 (80% power), 19,890 (90% power)
power.t.test: 3715.3 
Simulated power at n = 3,715 per arm: 0.799 (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 reports sponsor analyses using MMRM for the primary endpoint. Those results are distinct from the simple unadjusted re-calculations made from the posted arm means and standard errors. The sponsor's estimate for canagliflozin 100 mg plus metformin XR versus metformin XR is -0.46, with a 95% confidence interval from -0.657 to -0.269 and p = 0.001. For canagliflozin 300 mg plus metformin XR versus metformin XR, the estimate is -0.48, with a 95% confidence interval from -0.670 to -0.280 and p = 0.001. The sponsor also reports comparisons of combination treatment against the matching canagliflozin monotherapy, with estimates of -0.40 and -0.36; both have p = 0.001.

The re-calculated comparisons against metformin XR are close in direction and magnitude to the sponsor estimates for the two combination groups, but their intervals and p-values are not identical. The registry flags a discrepancy. Accordingly, the sponsor's own analysis takes precedence for the trial's inferential results; the re-calculations are best understood as unadjusted checks based on summary statistics rather than alternate official results.

The sponsor also labels the comparisons of canagliflozin monotherapy with metformin XR as non-inferiority or equivalence analyses. The registry lists p = 0.001 for these model results, but does not report the non-inferiority margin, the formal decision rule, or the supporting protocol details. A p-value alone is insufficient to determine non-inferiority or equivalence. Those comparisons therefore cannot be fully interpreted from the registry results alone.

15. Multiplicity, Missing Data and Analysis Populations

The registry record lists 18 analyses and 7 outcome measures, but does not report a complete multiplicity strategy. In particular, it does not describe whether a hierarchy of endpoints and contrasts was prespecified, which comparisons were considered confirmatory, or whether p-values were adjusted across the full set of analyses. These details matter because the interpretation of a p-value depends partly on how many opportunities there were to observe a result. General explanation: testing multiple hypotheses without an appropriate plan can increase the chance of at least one apparently strong result arising by chance.

The primary results include multiple comparisons among five treatment groups. The registry reports selected sponsor MMRM comparisons, including combination groups against metformin XR and against corresponding canagliflozin monotherapy. The reported p-values should be read as results for those named analyses; they do not, by themselves, resolve the overall multiplicity question. The record categorizes the two monotherapy comparisons with metformin XR as non-inferiority or equivalence analyses, but the margin and decision criteria are not reported in the registry record. Without those specifications, the p-values cannot be used to establish non-inferiority or equivalence.

The primary analysis population is described as the modified intent-to-treat set: participants who were randomized, received at least one dose, and were evaluable for the outcome. The registry does not give the complete definition of evaluability, the number of participants with missing Week 26 outcomes by treatment group, or the reasons for missing measurements. Handling of missing measurements, the assumptions underlying that handling, and sensitivity analyses are not reported in the registry record. These gaps limit independent assessment of how missing outcomes could affect the estimates.

The registry names MMRM for primary sponsor comparisons and for selected secondary contrasts, but does not provide the full model specification. Adjustment variables, covariance structure, model diagnostics, and sensitivity analyses are not reported in the registry record. The record also does not provide the protocol's original sample-size assumptions or an interim-analysis plan. These omissions do not change the posted estimates, but they limit the ability to evaluate the analysis plan in full or determine how closely the reported analyses followed prespecified procedures.

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 2% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: angina pectoris (0.8%), syncope (0.8%), anaemia (0.4%), immune thrombocytopenic purpura (0.4%), angina unstable (0.4%).

The posted serious adverse event counts are 7 among 237 participants at risk in the metformin XR group, 4 among 237 with canagliflozin 100 mg, 7 among 238 with canagliflozin 300 mg, 7 among 237 with canagliflozin 100 mg plus metformin XR, and 4 among 237 with canagliflozin 300 mg plus metformin XR. These are counts of participants with serious adverse events, alongside the number at risk recorded for each group. They are not estimates of causal effects and should not be compared as though treatment groups were guaranteed to be identical in all respects.

The registry also lists serious-event terms and percentages: Angina Pectoris and Syncope at 0.8% each, and Anaemia, Immune Thrombocytopenic Purpura, Angina Unstable, and Arrhythmia at 0.4% each. These terms describe events recorded in the registry; the record does not provide enough detail to assess their timing, severity beyond the serious-event classification, or relationship to treatment. Death counts are not reported in the registry record. Counts of other adverse events are also posted, but the figures alone do not establish comparative safety. Interpretation is limited by the 26-week study period and the information reported in the registry.

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

The primary results describe average change in HbA1c from baseline to Week 26 among participants included in the modified intent-to-treat analysis set. The posted arm means show decreases in all five groups, with the two combination groups having the most negative arm-level means. The sponsor's MMRM analyses report negative differences for combination treatment compared with metformin XR and compared with the corresponding canagliflozin monotherapy. These statements describe the direction and precision of the reported statistical contrasts; they are not a recommendation about treatment.

The results do not mean that every participant had the same change, or that an arm mean predicts an individual participant's outcome. A group average can conceal variation among participants. The registry does not report the distribution of individual responses in the primary summary, so the extent of that variation cannot be characterized from these results alone. The findings also do not establish whether the reported differences matter to participants in practical terms; no threshold for interpreting a clinically meaningful difference is reported in the registry record.

The results are limited to the studied population and the 26-week timeframe. The population described in the registry had type 2 diabetes with inadequate glycemic control with diet and exercise. Whether findings apply to other populations or settings is not established by the reported record. Longer-term outcomes and outcomes not listed in the registry are not reported. The primary endpoint measures HbA1c change, and the secondary outcomes address selected additional measures; neither should be treated as a complete account of health outcomes.

The sponsor's model-based analyses and simple summary-based re-calculations are not interchangeable. The record flags a discrepancy, so the sponsor's own analysis takes precedence. The registry does not report the full multiplicity plan, complete missing-data methods, or all model details. Those limitations matter when assessing how firmly the individual comparisons can support broader claims. The registry results can describe the posted estimates and their statistical uncertainty, but they cannot settle every question about generalizability, long-term effects, or the interpretation of unreported outcomes.

18. Why This Trial Matters Statistically

Statistically, this study illustrates why arm-level means, unadjusted contrasts, and model-based treatment comparisons should be kept separate. The primary arm estimates are least-squares means with standard errors, while the sponsor's pairwise results use MMRM. Re-calculations from summary statistics provide a transparent comparison of posted group means, but do not reproduce the repeated-measures analysis. The discrepancy flag is an important reminder that the sponsor's reported analysis takes precedence over a simplified calculation.

The five-arm design also generates multiple scientifically relevant contrasts: combination treatment against metformin XR, and combination treatment against the corresponding canagliflozin monotherapy. These address different questions. A comparison against metformin XR evaluates the difference between combination treatment and that single therapy, while comparison against a matching monotherapy addresses the incremental contrast within the canagliflozin dose groups. Results from one contrast do not automatically answer the other.

The endpoint structure further shows why continuous and threshold outcomes should not be conflated. Change in mean HbA1c captures the magnitude of a continuous measure; the percentage below 7 percent captures whether participants met a threshold. The secondary outcomes provide additional perspectives, but their interpretation depends on analysis methods and on how multiple comparisons were handled. The registry does not report the full multiplicity strategy, original sample-size justification, or complete missing-data methods.

Finally, the record distinguishes statistical evidence from design planning. Planning illustrations based on observed effects can help readers understand how assumed effect size influences participant requirements, but they do not establish the trial's original design assumptions or its achieved power. Taken together, the study is a useful case for reading treatment contrasts in context: identify the endpoint, analysis population, model, uncertainty, and limits of the record before drawing conclusions.

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 primary follow-up is Week 26, so the registry results do not characterize persistence of the measured changes beyond that visit. The study population is described as having type 2 diabetes with inadequate control despite diet and exercise; eligibility details beyond that description are not reported in the registry record. Applicability to other populations is therefore uncertain from the posted information.

The primary analysis population required randomization, receipt of at least one dose, and evaluability for the outcome. The complete evaluability definition and a full accounting of participants excluded from each analysis are not reported in the registry record. Missing measurement counts, reasons for missingness, and sensitivity analyses are also not reported. This limits assessment of whether missing data could influence group estimates.

The sponsor's primary analysis uses MMRM, but full model details are not reported. The registry does not give the complete adjustment set, covariance specification, or diagnostic information. It also does not report the full multiplicity plan, including how the 18 analyses and 7 outcome measures were organized or whether adjustment was applied across all comparisons. Non-inferiority or equivalence margins for the monotherapy comparisons are not reported, so those analyses cannot be fully assessed from the posted results.

The simple re-calculations are unadjusted approximations based on posted summaries and differ from sponsor results in some details. The discrepancy flag means that the sponsor's own analysis takes precedence. Secondary results are incomplete across the listed outcomes, and safety summaries do not by themselves establish treatment causation or comparative safety. Original sample-size justification and interim-analysis procedures are not reported in the registry record. These limitations concern what can be evaluated from the registry record; they should not be read as proof that particular procedures were absent from the full study documentation.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Study to Evaluate the Effectiveness, Safety, and Tolerability of Canagliflozin in Combination With Metformin in the Treatment of Patients With Type 2 Diabetes Mellitus With Inadequate Glycemic Control With Diet and Exercise — NCT01809327.
• Registry record last updated 2017-07-11; status last verified 2017-06.
• Results first posted 2016-01-06.
• 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

NCT01809327 is a completed randomized, parallel-group Phase 3 study of canagliflozin, metformin XR, and their combinations in participants with type 2 diabetes and inadequate glycemic control with diet and exercise. The record describes five treatment groups, quadruple masking, enrollment of 1186 participants, and 26 weeks of treatment.

The primary endpoint was change in HbA1c from baseline to Week 26. Posted group means were more negative in the combination groups than in the single-treatment groups. Sponsor MMRM analyses report negative differences for the combination groups against metformin XR and against corresponding canagliflozin monotherapy. A discrepancy is flagged between sponsor analyses and simple re-calculations, so the sponsor's analysis takes precedence.

Secondary results cover body weight, HbA1c threshold attainment, systolic blood pressure, lipids, and adverse events. The registry does not report the full multiplicity plan, complete model specifications, missing-data details, or original sample-size justification. The results should therefore be interpreted as estimates for the reported population, outcomes, analysis methods, and study duration.