Treatment of Hyperandrogenism Versus Insulin Resistance in Infertile Polycystic Ovary Syndrome (PCOS) Women: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Treatment of Hyperandrogenism Versus Insulin Resistance in Infertile Polycystic Ovary Syndrome (PCOS) Women (NCT00704912), a randomized phase 2 study. The primary outcome is Live Birth Rate.

2. Clinical Question

3. Trial Design

This was a phase 2 randomized, parallel-group trial with no masking. The record lists three treatment groups: lifestyle intervention, oral contraceptives (OCP), and lifestyle intervention combined with OCP. The lifestyle intervention comprised orlistat, meal replacement, and lifestyle modification; the OCP treatment was Loestrin 1/20. The combined arm received the combination of treatments. The registry lists an enrollment of 217. The primary outcome results are reported for groups of 50, 49, and 50 participants, respectively; how those outcome counts relate to the overall enrollment is not explained in the registry record.

The stated study goal was to assess the relative roles of treating hyperandrogenism and addressing obesity, described in the summary as the largest modifiable factor contributing to insulin resistance, in infertility treatment and pregnancy outcomes among obese women with PCOS. The investigators’ stated hypothesis was that correcting hyperandrogenism with OCP would be key to restoring ovulation leading to live birth, while improving insulin sensitivity through weight loss would be key to avoiding later pregnancy complications. These are the investigators’ hypotheses, not conclusions established by the outcome counts alone.

The registry lists ovulation rate, change in weight, and prevalence of metabolic syndrome as secondary outcomes. It also records four outcome measures and 15 analyses. Further details about randomization procedures, allocation concealment, treatment duration before conception attempts, adherence, and the original statistical analysis plan are not reported in the registry record. The registry status is completed. The recorded study start is 2008-09 and primary completion is 2014-03; these dates do not, by themselves, describe the follow-up experience of each participant.

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: 217 subjects consented for the study, of which 149 were randomized to one of the 3 treatment groups. Sixty-eight subjects were not randomized because they withdrew prior to randomization or were determined ineligible during the screening process.

Overall 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.

6. Primary Endpoint

The registry lists the primary outcome as Live Birth Rate, measured over Participants were followed for 4 months of attempted conception and those who conceived were then followed for the duration of their pregnancy, approximately 9 months..

The primary endpoint is live birth rate, presented as the number of participants with a live birth in each treatment group. It is a binary participant-level outcome: each participant either had a recorded live birth or did not. The listed counts are 13 of 50 in the lifestyle intervention group, 5 of 49 in the OCP group, and 12 of 50 in the combined group. These counts describe the posted results; they do not explain the reasons a participant may not have had a recorded live birth.

Live birth is relevant to a trial focused on infertility because it captures an outcome beyond ovulation or conception. The registry describes participants as attempting conception for 4 months and says those who conceived were followed for the duration of pregnancy, approximately 9 months. The time frame therefore includes a conception period and, for pregnancies, subsequent pregnancy follow-up. Further endpoint details, including the precise operational definition of live birth, handling of multiple births, and treatment of missing outcome information, are not reported in the registry record.

A count of participants with an event is not a measure of how quickly the event occurred. The posted primary outcome is a group count, not a time-to-event estimate. The record does not report a separate time-to-live-birth analysis. For interpretation, the counts should be read alongside the reported comparisons and their confidence intervals, while keeping in mind that the registry also contains analyses that differ from simple unadjusted recalculations.

7. Results Posted in the Registry

• Oral Contraceptives (OCP): 10.2% (5/49) versus 26.0% (13/50) on control; risk difference -15.8 percentage points (95% CI -30.6 to -1.0), risk ratio 0.39 (0.15–1.02); unadjusted two-proportion z-test p = 0.042.
• Lifestyle/OCP Combined: 24.0% (12/50) versus 26.0% (13/50) on control; risk difference -2.0 percentage points (95% CI -19.0 to 15.0), risk ratio 0.92 (0.47–1.82); unadjusted two-proportion z-test p = 0.817.

8. Statistical Interpretation

9. Statistical Methods Explained

Statistical methods help translate observed event counts into comparisons while making assumptions explicit. The registry lists a log-binomial model for the sponsor-reported analyses and also permits simple unadjusted recalculations from the posted counts. These approaches answer related but not necessarily identical questions. The sponsor’s model-based results are the trial’s reported analyses, and the recalculations are descriptive checks based directly on the posted group summaries.

What does a risk ratio describe?

A risk ratio compares the probability of an event in one group with the probability in another. A value of 1 represents equal event probabilities in the two groups under the measure. Values above or below 1 indicate a relative difference, with direction depending on which group is placed in the numerator. The registry reports risk ratios for three pairwise group comparisons. The analysis method is a log-binomial model. Details about covariates, adjustment, model diagnostics, and the exact handling of any missing observations are not reported in the registry record.

A risk ratio is relative, not an absolute difference. It does not directly state how many more or fewer participants had a live birth. The group counts and the risk difference provide complementary descriptions, but the two measures should not be treated as interchangeable. Interpretation also depends on comparison order, which is why the groups named in each reported analysis matter.

What does a risk difference describe?

A risk difference subtracts one group’s event probability from another’s. It is expressed on the probability scale, so its sign depends on the order of the comparison. The recalculated risk difference for OCP versus lifestyle intervention is -0.158, with a 95% confidence interval from -0.306 to -0.010. The recalculated risk difference for the combined group versus lifestyle intervention is -0.02, with a 95% confidence interval from -0.190 to 0.150. These are unadjusted recalculations from posted summary statistics, not the sponsor’s model estimates.

The p-values attached to those recalculations are also unadjusted. A p-value describes how compatible the observed data are with a specified null hypothesis under the calculation’s assumptions. It does not measure the probability that the null hypothesis is true, the probability that a result is due to chance, or the size or importance of an effect. The registry’s sponsor-reported p-values differ from the recalculated p-values in the comparisons where both are available; the sponsor’s own analysis takes precedence for the reported trial inference.

Why use a log-binomial model?

General explanation: a log-binomial model is used for binary outcomes when the aim is to estimate risk ratios directly. It models the logarithm of event probability as a function of group and any modeled covariates. The registry identifies this model for the reported risk ratios, but does not report whether covariates were included or provide enough detail to reconstruct the fitted models. A model-based estimate may differ from a simple comparison of observed counts because the model’s specification and analysis population can affect the estimate.

How should confidence intervals and p-values be read together?

A confidence interval gives a range of estimates compatible with the observed data and the method’s assumptions at the stated confidence level. The 95% intervals reported for the sponsor analyses are broad: 1.0 to 6.6, 0.9 to 6.1, and 0.5 to 2.1 for the three comparisons. Such intervals signal uncertainty about the magnitude of the relative differences. A p-value is a separate summary of evidence against a null value, and neither a p-value nor an interval alone establishes a causal explanation.

General explanation: when a confidence interval includes the null value for a measure, the data do not rule out that value at the interval’s confidence level using that method. Conversely, an interval that does not include the null does not establish clinical importance or remove concerns about study design, selection, missing outcomes, or multiple analyses. The intervals should be read in conjunction with the reported model and the limitations of the record.

Why can the recalculation differ from the reported analysis?

The recalculations use the posted event counts and group sizes without adjustment. The registry’s analyses are described as log-binomial models. Different methods can produce different estimates, intervals, and p-values, especially when the model incorporates information or assumptions not represented by a simple unadjusted calculation. The registry does not report sufficient model details to identify the particular reason for each difference. The discrepancy flag indicates that the reported and recalculated results do not align; the sponsor’s own analysis takes precedence.

General explanation: a planned statistical analysis can also distinguish an intention-to-treat population from a per-protocol or other analysis population, or specify covariate adjustment and missing-data procedures. Those choices can affect results. The registry record does not report which analysis populations were used, whether adjustment was applied, or how missing outcome data were handled. Consequently, these methods cannot be assumed or reconstructed from the summary results.

What does multiplicity change?

General explanation: when researchers examine several outcomes or comparisons, the chance of finding at least one apparently notable result can increase if each is interpreted independently. A multiplicity strategy may specify which outcomes are primary, how comparisons are ordered, or whether significance thresholds are adjusted. The registry lists four outcome measures and 15 analyses, but the multiplicity approach is not reported in the registry record. The presented p-values should not be assumed to account for all outcomes and comparisons unless the record explicitly says so.

In short, the model-based risk ratios are the registry’s sponsor-reported inferential results, while the risk differences and their p-values are simple unadjusted recalculations. Both describe the posted data from different statistical perspectives. The limits of the record prevent a fuller assessment of model specification, analysis populations, missingness, or multiplicity.

10. Confidence Intervals and Precision

Precision concerns how narrowly an estimate is determined by the data and statistical method. The registry reports 95% confidence intervals for its log-binomial risk ratios. For lifestyle intervention versus OCP, the risk ratio is 2.5 with an interval from 1.0 to 6.6. For OCP versus the combined group, it is 2.3 with an interval from 0.9 to 6.1. For lifestyle intervention versus the combined group, it is 1.1 with an interval from 0.5 to 2.1. These intervals span substantially different possible relative effects, so the estimates are not precise measurements of a single effect size.

The simple unadjusted risk-difference recalculation for OCP versus lifestyle intervention is -0.158, with a 95% confidence interval from -0.306 to -0.010. For the combined group versus lifestyle intervention, it is -0.02, with an interval from -0.190 to 0.150. These intervals are also broad relative to their point estimates. The recalculated intervals describe uncertainty under the unadjusted calculation, not the sponsor’s model.

General explanation: a confidence interval is a method-based range that reflects sampling uncertainty under the assumptions used to calculate it. It is not a statement that every value inside the interval is equally likely, nor does it account automatically for possible bias, missing outcomes, or all analysis choices. Wider intervals generally indicate less precision. In this trial, the event counts are modest and the reported intervals allow a range of effect magnitudes, so fine distinctions between treatments are not well resolved by these estimates.

The reported p-values should not be read as precision measures. A p-value does not tell how large an effect is or how certain its magnitude is. The registry analyses and unadjusted recalculations also differ, so the interval associated with one method should not be substituted for another. The sponsor-reported model results take precedence as the trial’s reported analysis.

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 planning figures are participants per arm for 80% power and participants per arm for 90% power. They are planning illustrations based on the observed effect, not the trial’s original design assumptions. The assumptions listed for these illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. Each planning scenario uses the stated fraction of the observed effect. The original trial’s own sample-size assumptions are not reported in the registry record.

For the OCP-versus-lifestyle comparison, planning for 100% of the observed effect gives 93 participants per arm for 80% power and 123 participants per arm for 90% power. Planning for 75% of the observed effect gives 179 participants per arm for 80% power and 239 participants per arm for 90% power. Planning for 50% of the observed effect gives 432 participants per arm for 80% power and 577 participants per arm for 90% power.

For the combined-versus-lifestyle comparison, the corresponding figures are 7,358 and 9,849 participants per arm when planning for 100% of the observed effect; 13,168 and 17,627 when planning for 75%; and 29,821 and 39,921 when planning for 50%. These figures follow directly from the stated planning scenarios and should not be interpreted as enrollment targets validated by the original trial protocol.

In plain language, the 80% column gives the estimated per-arm sample size under the stated assumptions for a design with an 80% chance of detecting the specified effect if that effect were true. The 90% column gives the corresponding size for a design with a 90% chance. Smaller effects need more participants to distinguish them from random variation under the same assumptions. This is not observed power; it is a prospective planning illustration based on the observed effect. The posted primary endpoint is a participant event count, not a time-to-event endpoint, so the listed figures are participants per arm, not total events.

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

Worked calculation for this trial

1. Control proportion p₀ = 0.260; treatment proportion p₁ = 0.102; average p̄ = 0.181.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5445 = 1.0672.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5329 = 0.4485.
4. n = (1.0672 + 0.4485)² / (-0.1580)² = 92.1, rounded up to 93 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 <- 0.26                          # control proportion
p1 <- 0.102040816326531                          # 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(704912)
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 = 93 (80% power), 123 (90% power)
 75% of observed effect: n per arm = 179 (80% power), 239 (90% power)
 50% of observed effect: n per arm = 432 (80% power), 577 (90% power)
power.prop.test: 92.1 
Simulated power at n = 93 per arm: 0.811 (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.

13. Analyses Reported by the Sponsor

The registry’s sponsor-reported primary-outcome comparisons use log-binomial models and report risk ratios, confidence intervals, and p-values. These are the analyses that should be used when describing the trial’s own inferential results. The simple recalculations use only the posted group counts and sizes to estimate unadjusted risk differences and associated intervals and p-values. They are not model-based reproductions of the sponsor’s analyses.

The recalculated OCP-versus-lifestyle risk difference is -0.158, with a 95% confidence interval from -0.306 to -0.010 and p = 0.042. The sponsor reports a risk ratio of 2.5 for lifestyle intervention versus OCP, with a 95% confidence interval from 1.0 to 6.6 and p = 0.06. For the combined group versus lifestyle intervention, the recalculated risk difference is -0.02, with a 95% confidence interval from -0.190 to 0.150 and p = 0.817; the sponsor reports a risk ratio of 1.1 for lifestyle intervention versus the combined group, with a 95% confidence interval from 0.5 to 2.1 and p = 0.82.

Because the sponsor results and simple recalculations differ, they should not be combined as if they were the same analysis. The sponsor’s own analysis takes precedence. The registry does not report enough detail about model specification or analysis populations to explain fully why the estimates differ.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists four outcome measures and 15 analyses, including the primary outcome and secondary outcomes of ovulation rate, change in weight, and prevalence of metabolic syndrome. It does not report a multiplicity strategy. It is therefore not possible to determine from the record whether the analyses were treated as a planned family, whether one comparison was designated as the principal inferential test, or whether any adjustment was used for examining multiple outcomes and comparisons.

General explanation: testing multiple hypotheses can increase the chance that at least one result appears unusual when each test is considered on its own. Methods for addressing multiplicity may adjust the testing threshold, specify a hierarchy, or distinguish confirmatory from exploratory analyses. Without a reported plan, the p-values should be interpreted as results for their individual analyses rather than assumed to control error across all analyses.

The registry record does not report how missing primary-outcome data were handled, whether all randomized participants were included in the analyses, or which analysis populations were used. It also does not describe whether participants were analyzed according to randomized assignment or according to treatment received. These details matter because analysis populations and missing-outcome procedures can affect the estimates and their uncertainty.

Further details about allocation concealment, protocol deviations, adherence, and the statistical analysis plan are not reported in the registry record. The record lists randomized allocation but does not give enough information to evaluate all safeguards against bias or to reconstruct the sponsor’s models. The differences between the sponsor-reported analyses and simple unadjusted recalculations reinforce the need to distinguish the reported model results from recalculated summaries. The sponsor’s own analysis takes precedence as the trial’s reported inferential analysis, but the available record does not resolve every methodological question.

15. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: abnormal uterine bleeding (2.0%), ectopic pregnancy (2.0%), preterm delivery (2.0%), perforated appendix (2.0%).

The registry reports serious adverse events for 50 participants at risk in the lifestyle intervention group, 49 in the OCP group, and 50 in the combined group. Serious events affected 3, 0, and 1 participants, respectively. The record also lists other adverse events affecting 32 of 50, 37 of 49, and 43 of 50 participants, respectively. These are counts of participants affected, not counts of all event episodes.

The listed serious adverse event terms are abnormal uterine bleeding, ectopic pregnancy, preterm delivery, and perforated appendix, each reported as 2.0%. The record does not provide enough detail in these entries to establish how each term was distributed across treatment groups or how the percentages were calculated. Death counts are not reported in the registry record; a missing value should not be interpreted as zero deaths.

These figures describe the safety information posted in the record. They do not, by themselves, establish that an event was caused by a treatment or allow comparison of risk without considering follow-up, event definitions, and ascertainment. The registry record does not report further details about safety monitoring or the time period used for adverse-event collection.

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

The record describes live-birth counts across three randomized groups and includes model-based comparisons and unadjusted recalculations. The counts were 13 of 50 for lifestyle intervention, 5 of 49 for OCP, and 12 of 50 for the combined treatment. These observations show the outcomes recorded among participants in this study; they do not by themselves establish a general treatment effect.

The sponsor-reported analyses give risk ratios of 2.5 for lifestyle intervention versus OCP, 2.3 for OCP versus the combined group, and 1.1 for lifestyle intervention versus the combined group. Their 95% confidence intervals are broad, and the corresponding p-values are 0.06, 0.08, and 0.82. The simple unadjusted recalculations differ from the sponsor’s model-based results. The sponsor’s own analysis takes precedence, and the record does not include enough model detail to fully account for the discrepancy.

General explanation: a randomized comparison can help assess differences between assigned groups, but interpretation still depends on implementation, outcome ascertainment, missing data, analysis choices, and uncertainty. The available record does not describe all of these features. The primary outcome concerns live birth, while the study summary also emphasizes ovulation, weight loss, insulin sensitivity, and pregnancy complications. The posted primary outcome counts alone do not establish the proposed biological pathway or explain why outcomes occurred.

The results do not establish that one treatment is superior, equivalent, or ineffective. Nor do they determine the effects for people outside the study population or under different care settings. The registry summary identifies the intended population as obese women with PCOS and infertility, but detailed eligibility and participant characteristics are not reported. The results are best read as limited evidence from this trial, with imprecision and unresolved reporting details, rather than as a definitive answer for all clinical contexts.

17. Why This Trial Matters Statistically

This trial is statistically interesting because it compares three approaches that the registry frames around different potential targets: hyperandrogenism and obesity-related insulin resistance. The primary endpoint, live birth, connects the infertility focus to an outcome occurring after conception attempts and pregnancy follow-up. The record also lists ovulation, weight change, and metabolic syndrome as secondary outcomes, indicating that the study addressed more than one stage or aspect of the proposed treatment rationale.

Three-arm trials create several possible comparisons. Here, the registry reports comparisons between lifestyle intervention and OCP, OCP and the combined group, and lifestyle intervention and the combined group. Each comparison has its own estimate and uncertainty. The number of outcomes and analyses recorded makes it important to know how comparisons were prioritized and whether multiplicity was addressed; that approach is not reported in the registry record.

The comparison between sponsor-reported log-binomial results and simple unadjusted recalculations is also a useful statistical lesson. Analyses based on the same posted outcome counts can differ when they use different methods or analysis populations. A recalculation can help describe the raw summaries, but it does not replace the trial’s reported model. In this record, the sponsor’s own analysis takes precedence, while the missing model details limit independent assessment.

Finally, the wide intervals make precision central to interpretation. A point estimate alone can hide the range of effects compatible with the data and method. Reading estimates together with their intervals, the study design, and the record’s reporting limits helps prevent stronger conclusions than the evidence supports.

18. Limitations

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

Several trial-specific limitations arise from the available record. Although the study is described as randomized, parallel, and unmasked, details of the randomization procedure and allocation concealment are not reported in the registry record. The record does not explain treatment adherence, discontinuation, protocol deviations, or co-interventions. These details would help assess how closely the groups’ assigned treatments were implemented.

The overall enrollment is listed as 217, while primary outcome groups are reported with 50, 49, and 50 participants. The record does not explain the relationship between the overall enrollment and those outcome denominators. It also does not report the analysis population, the number of participants with missing outcome data, or how missing data were handled. The follow-up description specifies 4 months of attempted conception and pregnancy follow-up for those who conceived, approximately 9 months, but participant-level completion is not described.

The sponsor’s log-binomial results differ from simple unadjusted recalculations. The sponsor’s own analysis takes precedence, but model specification, covariate adjustment, and the basis for the differences are not reported in the registry record. The multiplicity approach for the 15 analyses and four outcome measures is also not reported. These omissions limit independent evaluation of the inferential results.

The record offers limited information about the primary endpoint definition and safety-event ascertainment. It does not report the detailed criteria used for live birth, the distribution of listed serious adverse-event terms by treatment group, or the time period used to collect adverse events. The record also does not report deaths; that should not be interpreted as evidence of no deaths. Finally, detailed baseline characteristics and eligibility criteria are not available in the stated information, limiting assessment of how broadly the results may apply.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Treatment of Hyperandrogenism Versus Insulin Resistance in Infertile Polycystic Ovary Syndrome (PCOS) Women — NCT00704912.
• Registry record last updated 2016-11-07; status last verified 2016-11.
• Results first posted 2015-12-17.
• 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.

22. Record Summary

NCT00704912 is a completed phase 2 randomized, unmasked, parallel-group trial of treatments for infertile women with PCOS. The registry lists an enrollment of 217 and three groups: lifestyle intervention, OCP, and lifestyle/OCP combined. The primary outcome is live birth rate, recorded as participants with an event. The posted counts are 13 of 50, 5 of 49, and 12 of 50, respectively.

The sponsor-reported comparisons use log-binomial models. The record also supports simple unadjusted recalculations that differ from those model results; the sponsor’s own analysis takes precedence. Confidence intervals are broad, and details about missing data, analysis populations, model specification, and multiplicity are not reported in the registry record. The study summary states hypotheses about hyperandrogenism, insulin sensitivity, ovulation, and pregnancy complications, but the posted primary outcome counts alone do not establish those proposed mechanisms.