Saxagliptin + Metformin Compared to Saxagliptin or Metformin Monotherapy in PCOS Women With Impaired Glucose Homeostasis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Saxagliptin + Metformin Compared to Saxagliptin or Metformin Monotherapy in PCOS Women With Impaired Glucose Homeostasis (NCT02022007), a randomized phase 3 study. The primary outcome is Glucose Metabolism.

2. Clinical Question

The study was framed as a pilot investigation comparing clinical, endocrine, and metabolic effects of the combination with those of each medicine alone. Its primary outcome is a count of participants meeting a categorical status-change definition, rather than a direct measure of average glucose change. That distinction matters: the primary comparison concerns how many participants crossed the stated threshold for normal status, while the secondary outcomes include continuous glucose measures and other metabolic or endocrine measures.

3. Trial Design

The registry record describes NCT02022007 as a randomized, parallel-group, single-blinded study, classified as Phase 3. Its official title calls it a single-blinded randomized pilot study. Enrollment was 38, and the record lists the study as completed. The recorded start was 2014-03 and primary completion was 2016-08.

Participants were assigned to metformin XR, saxagliptin, or saxagliptin-Metformin XR. Each regimen was once daily for 16 weeks. The recorded doses were 2000 mg metformin XR, 5 mg saxagliptin, or the combination pill containing 5 mg saxagliptin and 2000 mg metformin XR. The registry record says the study was single-blinded but does not identify which people involved in the study were blinded.

The primary outcome was glucose metabolic status after treatment, expressed as the number of participants changing from impaired to normal. For this outcome, the posted arm counts are 3 of 12 for metformin XR, 6 of 11 for saxagliptin, and 10 of 11 for the combination. The denominators used for this outcome do not account for all 38 enrolled participants; reasons for the difference and the analysis population are not reported in the registry record.

The registry record also lists secondary measures spanning fasting glucose, glucose during an oral glucose tolerance test, insulin sensitivity, pancreatic beta-cell compensatory function, body mass index, waist circumference, menstrual cycle interval, and the triglyceride-to-HDL cholesterol ratio. Nine secondary outcomes are posted, while the listed outcome set includes 12 measures. The study's original sample-size assumptions, allocation details beyond randomization, and planned analysis populations 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: Exclusion criteria included diabetic subjects, smokers, suspected pregnancy, desiring pregnancy or injectable hormonal contraceptive use within 6 months; and use of oral contraceptives, other steroid hormones, drugs that affect gastrointestinal motility or carbohydrate metabolism and/or anti-obesity drugs within 3 months prior to study entry.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Change from impaired to normal was evaluated. No patients were diabetic at the start or completion of the study

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: All participants had PCOS and were glucose impaired

6. Primary Endpoint

The registry lists the primary outcome as Glucose Metabolism, measured over 16 weeks.

The primary endpoint counts participants whose glucose metabolic status changed from impaired to normal after treatment. The impairment definition in the registry record is fasting glucose greater than 100 mg/dL and/or 2-hour glucose greater than 140 mg/dL. The outcome description also states that no participants were diabetic at the beginning or end of the study. Thus, the outcome focuses on movement across the stated impaired/normal categories, rather than change in glucose concentration as a continuous measurement.

A categorical outcome can be useful when the clinical question is framed around whether participants meet a defined status threshold. It provides an easily interpreted count within each treatment group. However, a status count does not convey how close each participant was to the threshold, the size of any change in glucose, or the distribution of individual changes. Those are different questions, addressed in part by continuous secondary measures such as fasting glucose and mean blood glucose during the oral glucose tolerance test.

The primary outcome is listed at 16 weeks. The posted counts are 3 participants in the metformin XR group, 6 in the saxagliptin group, and 10 in the combination group. These counts describe the observed status changes in the recorded outcome groups; they do not by themselves establish the size or precision of a causal effect. The registry record does not report further details about how status was adjudicated, the handling of missing outcome measurements, or the statistical plan for this endpoint.

7. Results Posted in the Registry

• Saxagliptin: 54.5% (6/11) versus 25.0% (3/12) on control; risk difference 29.5 percentage points (95% CI -8.7 to 67.8), risk ratio 2.18 (0.71–6.68); unadjusted two-proportion z-test p = 0.147.
• Saxagliptin-Metformin XR: 90.9% (10/11) versus 25.0% (3/12) on control; risk difference 65.9 percentage points (95% CI 36.1 to 95.7), risk ratio 3.64 (1.34–9.86); unadjusted two-proportion 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.

Fasting Glucose

• 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 deviations.
• Result:
  • Saxagliptin: mean 5.30 versus 5.40 on control; difference -0.10 (95% CI -0.60 to 0.40), standardized effect d = -0.16; z-test p = 0.694.
  • Saxagliptin-Metformin XR: mean 5.00 versus 5.40 on control; difference -0.40 (95% CI -0.85 to 0.05), standardized effect d = -0.71; z-test p = 0.081.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Saxagliptin 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 Saxagliptin-Metformin XR and the control. The interval gives the range of effects that remain compatible with the data.

Mean Blood Glucose During the OGTT

• 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 deviations.
• Result:
  • Saxagliptin: mean 7.10 versus 7.90 on control; difference -0.80 (95% CI -2.05 to 0.45), standardized effect d = -0.53; z-test p = 0.210.
  • Saxagliptin-Metformin XR: mean 6.20 versus 7.90 on control; difference -1.70 (95% CI -2.76 to -0.64), standardized effect d = -1.30; z-test p = 0.002.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Saxagliptin 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-Metformin XR than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Matsuda Index of Insulin-Sensitivity (SI OGTT)

• 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 deviations.
• Result:
  • Saxagliptin: mean 3.50 versus 2.50 on control; difference 1.00 (95% CI -1.20 to 3.20), standardized effect d = 0.38; z-test p = 0.374.
  • Saxagliptin-Metformin XR: mean 4.10 versus 2.50 on control; difference 1.60 (95% CI -0.18 to 3.38), standardized effect d = 0.75; z-test p = 0.079.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Saxagliptin 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 Saxagliptin-Metformin XR and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes cover several related but distinct domains: fasting and oral-glucose-tolerance measurements, insulin sensitivity and beta-cell function, body-size measures, menstrual cycle interval, and a lipid ratio. The posted results therefore offer a broader metabolic and endocrine description than the primary categorical status-change outcome alone.

Across the posted continuous outcomes, the group means do not all move in the same way or show the same degree of separation. Some recalculated pairwise estimates are more pronounced than others, while several intervals include no difference. The registry record also lists overall ANOVA p-values for three outcomes. These patterns can provide context for the primary result, but they do not independently confirm it: the measures assess different features, and the record does not report a prespecified hierarchy linking secondary measures to the primary endpoint.

Unadjusted secondary results carry less weight because testing many outcomes increases the chance that at least one result will appear notable by chance. The record lists multiple outcomes and analyses, but it does not report whether the sponsor adjusted for multiplicity, designated particular secondary outcomes as confirmatory, or distinguished exploratory from confirmatory analyses. In addition, the posted secondary outcome set is incomplete relative to the total number of listed measures. Without a full accounting of planned and reported results, selective availability cannot be assessed.

The secondary findings are therefore best read as exploratory context for a pilot study. They may help identify measures and contrasts to examine in future research, but the available record does not establish that a pattern across these measures is robust or that it reflects a common underlying treatment effect.

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

• Fasting Glucose
• Mean Blood Glucose During the OGTT
• Matsuda Index of Insulin-Sensitivity (SI OGTT)
• Pancreatic ß-cell Compensatory Function
• Body Mass Index at 16 Weeks
• Waist Circumference at 16 Weeks
• Menstrual Cycle Interval at 16 Weeks
• Triglyceride (TRG) /HDL-cholesterol Ratio

9. Statistical Interpretation

10. Statistical Methods Explained

Randomization is intended to make treatment groups comparable, on average, with respect to measured and unmeasured baseline characteristics. It does not guarantee that small groups will be balanced. The record identifies randomized allocation but does not report the randomization sequence method, concealment procedures, or baseline characteristics needed to assess balance.

What does the primary outcome count?

The primary outcome is binary at the participant level: a person either changed from impaired to normal status or did not. Arm summaries report the number meeting the change definition and the denominator. A proportion is often used to describe such an outcome, and a risk difference expresses the difference between two proportions. Here, the recalculated risk differences compare each treatment group with metformin XR. A positive value indicates a higher observed proportion changing status in the named group, under the comparison's stated direction. This is a general explanation of the measure, not a claim about the study's original analysis plan.

The recalculated confidence intervals and p-values are unadjusted comparisons based on the posted summary counts. They are not adjusted for other outcomes or potential covariates. Because the outcome is binary and group sizes are limited, estimates can be sensitive to a small number of participants. Risk ratios and odds ratios are alternative ways of describing binary outcomes, but neither should be read without its uncertainty interval or the context of the study design.

How should a confidence interval be understood?

In general, a confidence interval represents a range of values compatible with the data and the assumptions of the method. A wide interval indicates limited precision: the data are consistent with a broad range of possible effect sizes. If an interval for a difference includes zero, it is compatible with no difference as well as with effects in either direction within the interval. An interval that excludes zero does not, on its own, resolve questions about bias, multiplicity, missing data, or generalizability.

What does an unadjusted p-value say?

In general, a p-value measures how unusual data at least as inconsistent with a specified null model would be if that model were true and the analysis assumptions held. It is not the probability that the null hypothesis is true, nor does it measure clinical importance. A small p-value does not convey the size or precision of an effect, and a larger p-value does not establish equivalence. The sponsor's reported overall p-value is distinct from the recalculated pairwise p-values, since the methods and comparison structures differ.

Why do methods matter for continuous outcomes?

Continuous outcomes are summarized using group means and standard deviations in the posted results. A mean difference describes the contrast between group averages. Standard deviations describe variability among observations, rather than uncertainty in the mean itself. The recalculated mean-difference intervals and p-values are unadjusted. The registry record lists ANOVA for the sponsor's overall analyses of fasting glucose, mean glucose during the oral glucose tolerance test, and the Matsuda index. ANOVA is a general method for comparing means across groups; the record does not provide test statistics, degrees of freedom, model details, or follow-up pairwise tests.

For general context, standard methods for mean comparisons can depend on assumptions such as the form of the outcome distribution and how variability is modeled. In count or binary data, sparse observations can also make conventional approximations less dependable. Over-dispersion is a general concern when observed variability exceeds what a model expects; it can make uncertainty estimates too small if not handled appropriately. The registry record does not report whether over-dispersion was assessed or whether alternative models were used.

What is and is not documented?

The record identifies McNemar for the primary reported analysis and ANOVA for three reported secondary analyses. It does not report a comprehensive statistical analysis plan, covariate adjustment, model diagnostics, treatment of missing measurements, or methods for controlling repeated testing. As a result, the posted analysis methods can be described, but the full analytic workflow and its assumptions cannot be reconstructed from the registry record.

11. Confidence Intervals and Precision

The primary outcome estimates are based on small denominators: 12 participants in the metformin XR group and 11 in each of the other groups. With few observations, each participant represents a substantial part of the group outcome, so a change in one or a few observed outcomes could alter the estimated proportions and contrasts.

For saxagliptin versus metformin XR, the recalculated risk difference is 0.295 and its interval extends from -0.087 to 0.678. The interval includes zero, so the recalculation is compatible with no difference as well as a range of positive differences. For the combination versus metformin XR, the recalculated risk difference is 0.659, with an interval from 0.361 to 0.957. This interval is above zero, but it remains broad, reflecting uncertainty in the magnitude of the contrast. These are unadjusted recalculations; the sponsor's own reported analysis takes precedence where it differs.

A confidence interval does not show where the true effect is with certainty, and it is not a guarantee that the effect lies within its limits. It describes sampling uncertainty under the method's assumptions. It does not incorporate uncertainty from missing data, outcome measurement, selective reporting, or a mismatch between the analysis and the original plan. Those issues cannot be assessed fully because relevant details are not reported in the registry record.

The small group sizes also limit how confidently the observed pattern can be generalized beyond the study participants. The overall reported p-value indicates a result from the sponsor's listed analysis, but without a corresponding estimate and interval for that analysis, its magnitude and precision cannot be assessed from the reported analysis alone.

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 illustrations use the observed effect as a starting point and consider planning for 100%, 75%, or 50% of that observed effect. For each scenario, the 80% power column gives the planned number of participants per arm for a design targeting an 80% chance of detecting that effect under the stated assumptions if the effect is present. The 90% power column gives the number of participants per arm for a higher target chance of detection. These are planning quantities, not outcomes observed in this study.

The stated planning 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 the stated fraction of the observed effect. For the outcomes presented, figures refer to participants per arm; the general note says time-to-event illustrations instead give the total number of events. No time-to-event outcome is described in the posted results. The trial's own original sample-size assumptions and justification are not reported in the registry record.

The illustrations are sensitive to the effect chosen: planning for a smaller effect requires more participants to achieve the same target power. This is why the 50% observed-effect scenarios generally call for larger groups than the full observed-effect scenarios. The calculations are based on effects from small observed groups, which may not provide stable estimates for planning a future study. The effect assumed for planning should therefore be selected with substantive justification rather than treated as a prediction of what a future trial will find.

These figures are not observed power. They are hypothetical planning illustrations under the stated assumptions, not a retrospective measure of whether this trial was adequately powered and not a reason to interpret a nonsignificant result as evidence of no effect.

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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.250; treatment proportion p₁ = 0.545; average p̄ = 0.398.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6922 = 1.3566.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6599 = 0.5554.
4. n = (1.3566 + 0.5554)² / (0.2955)² = 41.9, rounded up to 42 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
p0 <- 0.25                          # control proportion
p1 <- 0.545454545454545                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^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 (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(2022007)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 42 (80% power), 56 (90% power)
 75% of observed effect: n per arm = 73 (80% power), 97 (90% power)
 50% of observed effect: n per arm = 157 (80% power), 209 (90% power)
power.prop.test: 41.9 
Simulated power at n = 42 per arm: 0.822 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-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 record lists a sponsor-reported McNemar analysis across the three primary-outcome groups, with p = .007. For three secondary measures, it lists overall ANOVA analyses with p-values of < 0.001, .038, and 0.003. These are the analyses attributed to the sponsor in the record.

The displayed recalculations are simple, unadjusted pairwise comparisons derived from posted group summaries. For the primary outcome, those comparisons use metformin XR as the reference group and yield different p-values from the sponsor's overall analysis. The recalculations are not a correction to, or reproduction of, the sponsor's analysis: they ask pairwise questions using a different method and available summary data. The registry record does not describe enough detail to reproduce the sponsor's analyses or determine exactly how the reported tests were implemented.

Because the record flags a discrepancy, the sponsor's own analysis takes precedence when describing the trial's reported statistical result. Recalculated estimates can still help readers understand the direction and approximate size of pairwise contrasts, but they should be clearly distinguished from the sponsor's analysis. The reported overall p-value should not be treated as a pairwise comparison, and the pairwise recalculated p-values should not be presented as the sponsor's results.

15. Multiplicity, Missing Data and Analysis Populations

The registry record lists a primary outcome and multiple secondary outcomes, with analyses across three treatment groups and additional pairwise contrasts in the recalculations. Repeated testing raises a multiplicity issue: as more hypotheses are tested, the chance of at least one apparently unusual result can increase. This is a general statistical concern and does not show that any particular finding is false. The record does not report a multiplicity adjustment, a testing hierarchy, or which outcomes were intended to be confirmatory.

Missing-data handling is also not reported. The primary outcome denominators are 12, 11, and 11, while enrollment was 38. The registry record does not state why the outcome denominators differ from the enrolled total, whether all enrolled participants were randomized, whether participants withdrew, or how missing measurements were handled. It also does not identify an intention-to-treat, per-protocol, or other analysis population. Consequently, the effect of missing outcomes on the posted estimates cannot be evaluated from the record.

Other important analysis details are not reported, including baseline adjustment, the allocation concealment procedure, the identity of the blinded party, prespecified handling of protocol deviations, and the full statistical analysis plan. The primary reported method is named as McNemar, but its implementation and assumptions are not described. The secondary record names ANOVA for several outcomes without model details or diagnostic information.

These omissions limit reproducibility and make it difficult to determine whether all outcomes and analyses were reported as planned. They do not establish that the analyses were inappropriate; they mean the registry record does not provide enough information to assess those questions.

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 non-serious adverse event terms

The adverse-event record lists zero serious adverse events in each group: the serious-event risk counts are 13 for metformin XR, 12 for saxagliptin, and 13 for the combination, with zero serious events reported in all three groups. The record does not list serious adverse-event terms.

For other, non-serious adverse events, the record lists 1 event among 13 at risk in the metformin XR group, 0 among 12 at risk in the saxagliptin group, and 2 among 13 at risk in the combination group. These are counts in the registry record, not estimates of comparative risk with uncertainty intervals. The safety denominators differ from the primary-outcome denominators, and the record does not explain the difference.

Deaths are not reported in the registry record. The absence of reported serious events in these small groups does not establish that a treatment has no safety risks. The record does not provide event descriptions, severity details for other events, exposure-adjusted analyses, or enough information to compare uncommon events reliably.

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

The posted primary counts show that more participants in the combination group changed from impaired to normal glucose status than in either monotherapy group during the 16-week study period. The sponsor's reported overall analysis has p = .007. The recalculated pairwise comparison with metformin XR is consistent with a larger observed change-status proportion in the combination group, but it is unadjusted and differs from the sponsor's reported analysis. The sponsor's analysis takes precedence as the trial's own reported result.

This finding does not establish the magnitude of an effect with high precision. The groups are small, and the registry record does not provide the sponsor's estimate and confidence interval for the overall analysis. Nor does the result establish what would happen over a longer period, in a different population, or under a different treatment regimen. The record concerns women with PCOS and impaired glucose regulation and reports outcomes at 16 weeks.

The finding also does not show that a treatment is clinically beneficial or appropriate for an individual. It reports a study outcome, not medical advice. A categorical transition across a glucose threshold is not the same as a complete account of glucose control, metabolic health, or longer-term outcomes. Secondary measures provide related information but are subject to uncertainty and multiplicity concerns.

Finally, the registry record does not provide enough detail to assess missing data, the analysis population, the complete statistical plan, or the reasons for the difference between enrollment and outcome denominators. These limits mean the results are best interpreted as pilot evidence that describes observed group patterns and motivates careful evaluation in further research, rather than as a definitive estimate.

18. Why This Trial Matters Statistically

This study is statistically instructive because it places a categorical primary outcome alongside continuous metabolic measures. A count of participants changing from impaired to normal status is easy to communicate, but it depends on threshold definitions and does not show the size of the underlying glucose changes. The secondary continuous measures address different aspects of metabolism and can add context without being interchangeable with the primary endpoint.

The study also illustrates why results from small randomized groups need careful attention to precision. A small number of outcomes can produce noticeable changes in an estimated risk difference, and a broad confidence interval can remain compatible with a wide range of effects. Reporting counts, estimates, intervals, and the analysis method helps readers distinguish the observed pattern from the certainty of its magnitude.

A further statistical issue is the difference between an overall sponsor-reported test and recalculated pairwise comparisons. They answer different questions, so a p-value from one should not be treated as if it directly tested the other. Clear reporting of the analysis plan, model details, and treatment of multiple outcomes would help readers understand those distinctions and reproduce the findings.

Finally, the registry record's incomplete information about analysis populations, missing data, and multiplicity demonstrates how these details affect interpretation even when the outcome counts are available. For a pilot study, the results can inform questions and design choices for later work, but the planning illustrations should not be mistaken for proof that a future effect will match the observed estimate.

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.

Several limitations are specific to the information recorded for this trial. The outcome groups are small, with primary-outcome denominators of 12, 11, and 11. Enrollment was 38, but the registry record does not explain why the primary-outcome denominators account for fewer participants or how missing outcomes were handled. The analysis population is not reported.

The primary result is a count of participants crossing from impaired to normal glucose status. The record supplies the threshold definition but not details about the timing or procedure for assessing status, how repeat or borderline measurements were handled, or whether outcome assessors were blinded. Although the study is described as single-blinded, the blinded party is not identified.

The record labels the study Phase 3 while its official title describes a pilot study. It does not report the original sample-size justification, protocol, detailed statistical analysis plan, multiplicity strategy, or treatment of protocol deviations. The sponsor's primary analysis is named, but the method's implementation and its estimate with an interval are not reported. The discrepancy between the sponsor analysis and simple recalculations further limits direct comparison of the reported p-values.

Only nine secondary outcomes are posted despite 12 listed measures, and the record does not explain the difference. It also does not report whether unposted measures were collected or analyzed. Safety reporting is limited to event counts and does not include serious-event terms, while deaths are not reported in the registry record. These reporting gaps constrain assessment of completeness, precision, and generalizability.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Saxagliptin + Metformin Compared to Saxagliptin or Metformin Monotherapy in PCOS Women With Impaired Glucose Homeostasis — NCT02022007.
• Registry record last updated 2017-06-28; status last verified 2017-05.
• Results first posted 2017-06-28.
• 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

NCT02022007 is a completed randomized, parallel-group, single-blinded study classified as Phase 3 and described in its official title as a pilot study. It enrolled 38 women with PCOS and impaired glucose regulation to compare metformin XR, saxagliptin, and their combination over 16 weeks.

The primary outcome counted participants changing from impaired to normal glucose status. The posted counts were 3 of 12 with metformin XR, 6 of 11 with saxagliptin, and 10 of 11 with the combination. The sponsor reported a three-group McNemar analysis with p = .007. Recalculated unadjusted pairwise comparisons are supplementary, and the sponsor's analysis takes precedence where results differ.

Secondary measures covered glucose, insulin sensitivity, body-size measures, menstrual cycle interval, and a lipid ratio. The record does not report the original sample-size justification, a full analysis plan, missing-data methods, multiplicity handling, or the analysis population. Serious adverse events were recorded as zero in all groups; deaths are not reported in the registry record.