Study In Postmenopausal Women With Type 2 Diabetes Looking At Approved Diabetes Drugs And How They Affect Bone Health: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Study In Postmenopausal Women With Type 2 Diabetes Looking At Approved Diabetes Drugs And How They Affect Bone Health (NCT00679939), a randomized phase 4 study. The primary outcome is Adjusted Percent Change From Baseline in Femoral Neck (FN) Bone Mineral Density (BMD) Via Dual-energy X-ray Absorptiometry (DXA) at Week 76+10 Days.

2. Clinical Question

3. Trial Design

This was a phase 4, randomized, double-blind, multicenter, parallel-group study. The official title describes a 52-week randomized period followed by a 24-week open-label follow-up. The registry lists the study as completed, with a start date of 2008-04-21 and a primary completion date of 2010-09-16. Enrollment was 226.

The two groups differed in their double-blind-period treatment: rosiglitazone up to 8mg/day or metformin up to 2000mg/day. In the open-label period, both arm descriptions specify metformin. The endpoint group labels likewise distinguish rosiglitazone or metformin during the double-blind period, followed by metformin during the open-label period. This sequence matters when interpreting the endpoint: the Week 76 measurement reflects the study's treatment sequence, not just a comparison of two drugs taken throughout the entire study.

The primary outcome was adjusted percent change from baseline in femoral-neck bone mineral density at Week 76+10 days. Other listed outcomes included bone mineral density at additional skeletal sites and measures related to bone formation, bone resorption, and vitamin D. The registry lists 72 outcome measures and 19 posted secondary outcome results. These measures concern different outcomes and time points; they should not be treated as interchangeable evidence for the primary question.

The registry record does not report the original sample-size justification, the intended statistical power, or the planned handling of multiplicity and missing endpoint values. Randomization and masking are stated, but details such as the randomization procedure and how masking was maintained 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.

52-Week Double-Blind (DB) Period

Reasons for not completing this period:

24-Week Open-Label (OL) Period

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Safety Population. Only evaluable participants with a value at baseline and at Week 52 for the parameter of interest were analyzed. Only participants with Baseline DXA and Week 52 DXA measurements performed on or prior to initiating open-label MET are included in this primary analysis.

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 Adjusted Percent Change From Baseline in Femoral Neck (FN) Bone Mineral Density (BMD) Via Dual-energy X-ray Absorptiometry (DXA) at Week 76+10 Days, measured over Baseline and Week 76+10 days.

The primary endpoint is adjusted percent change from baseline in femoral-neck bone mineral density, measured by dual-energy X-ray absorptiometry, at Week 76+10 days. Bone mineral density expresses the mineral content of bone relative to its area. DXA is described in the registry as the principal technique used to measure this quantity. The endpoint's percent-change scale describes the change relative to each participant's baseline value, rather than the follow-up value alone.

Femoral-neck bone density is one specific measure at one skeletal site. The registry also lists outcomes at the total hip, trochanter, and lumbar spine, but those are separate endpoints and cannot be substituted for the femoral-neck result. The posted primary endpoint is a continuous measure, summarized as a mean with standard error.

The analysis description says that the endpoint was assessed using analysis of covariance, with terms for treatment, baseline value, prior therapy, and region. In general, adjusting for baseline can account statistically for initial differences in the measured outcome and can improve the precision of a treatment comparison when the adjustment is appropriate. The record's stated analysis population included only evaluable participants with both a baseline and a Week 76 value collected up to 10 days after stopping open-label metformin. The endpoint therefore represents that evaluable group and time window, not automatically every enrolled participant.

7. Results Posted in the Registry

• Metformin in DB Period; Metformin in OL Period: mean 0.31 versus -1.91 on control; difference 2.22 (95% CI 0.47 to 3.97), standardized effect d = 0.43; z-test p = 0.013.

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes the primary outcome as a continuous measure and summarizes each arm with a mean and standard error. In general, a mean describes the average outcome in a group. It does not show whether individual participants had similar values or whether the group included widely varying responses. The standard error describes the estimated precision of the group mean; it is not the spread of individual outcomes. The registry does not report a standard deviation for each primary endpoint group.

What does percent change represent?

The outcome is calculated relative to baseline: the change in bone mineral density from baseline to the follow-up measurement is divided by the baseline value and expressed as a percentage. This puts the change on a scale relative to the starting measurement. A negative group mean indicates that the mean follow-up measurement was lower than baseline on this percent-change scale; a positive group mean indicates a mean increase. A group mean does not imply that every participant changed in the same direction.

What does the adjusted analysis do?

The registry states that analysis of covariance was used, with terms for treatment, baseline value, prior therapy, and region. In general, analysis of covariance estimates a treatment comparison while accounting for specified covariates. Including baseline can account for the starting level of the outcome; including other prespecified factors can adjust the comparison for those factors. Such adjustment depends on the model and the data. It does not automatically remove bias from unmeasured differences, and it does not make an analysis population representative of people who were not included in it.

The posted group means and standard errors allow a simple unadjusted comparison to be recalculated. That comparison does not include the adjustment terms in the registry's ANCOVA description. The posted results do not include the model coefficient, adjusted group estimates, or the sponsor's treatment-comparison confidence interval and p-value. The unadjusted recalculation should therefore be interpreted as descriptive statistical context rather than the sponsor's adjusted result.

Why distinguish standard error from variability?

In general, standard error concerns uncertainty in an estimated mean, while a standard deviation describes variation among individual observations. A relatively small standard error does not mean participants' measurements were tightly clustered; it may reflect the precision of the mean estimate. The registry reports standard errors for the endpoint means, so those should not be read as participant-level ranges.

What is the role of the analysis population?

The primary endpoint population is described as the safety population, restricted to evaluable participants with a baseline value and a Week 76 value obtained up to 10 days after stopping open-label metformin. This complete-measure requirement means that participants without a usable baseline or follow-up value are not represented in the endpoint summary. The registry does not report the number or reasons for missing endpoint values, the method used to handle missing data, or whether any sensitivity analyses were conducted.

In general, restricting an analysis to participants with observed measurements can affect interpretation if the availability of measurements is related to treatment or outcome. That possibility cannot be assessed from the reported summaries alone. The registry record does not report an imputation procedure or a comparison of participants with and without endpoint data.

What can the recalculated p-value tell us?

The unadjusted p-value of 0.013 is a recalculation based on the posted summary statistics. In general, a p-value measures how incompatible the observed data are with a specified null model, under the assumptions of the test. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of a difference. This recalculated value does not incorporate the ANCOVA covariates described for the primary endpoint.

The registry lists multiple outcome measures and secondary results. When many outcomes or comparisons are examined, the chance of finding at least one apparently unusual result can increase. Appropriate interpretation depends on whether the analyses were prespecified and whether any adjustment or hierarchy was used. The registry record does not report a multiplicity strategy, so no claim about multiplicity control can be made.

Why can over-dispersion matter in other settings?

General statistical note: for count outcomes, over-dispersion means that the observed variation exceeds what a simple count model assumes. If it is ignored, uncertainty can be understated and statistical evidence can appear stronger than it is. The primary endpoint in this study is continuous, not a count outcome, so over-dispersion is not the central issue for this particular analysis. This general concept does not alter the interpretation of the posted femoral-neck means.

Overall, the main methodological distinction is between the adjusted ANCOVA described in the registry and the simple unadjusted recalculation possible from the posted summaries. The latter gives a transparent comparison of the displayed group means, but it cannot reproduce the sponsor's adjusted analysis or resolve missing-data and multiplicity questions that are not reported.

10. Confidence Intervals and Precision

The posted endpoint means are accompanied by standard errors, and the unadjusted recalculation includes a confidence interval for the mean difference. A confidence interval expresses a range of values compatible with the data and the assumptions of the calculation. In general, a narrower interval indicates greater precision for the estimated contrast, while a wider interval indicates greater uncertainty. An interval is not a guarantee that the true value lies within it.

The recalculated unadjusted interval is entirely above zero, consistent with the positive unadjusted mean difference of 2.22 percentage points. The recalculated unadjusted p-value is 0.013. These statistics describe the summary-based, unadjusted comparison; they do not give the confidence interval or test result from the registry's described adjusted ANCOVA. That sponsor analysis output is not reported in the registry record.

The interval's precision should also be considered in light of the analysis population. It is based on evaluable endpoint participants with the required measurements, rather than all enrolled participants. The registry does not report how many participants lacked an endpoint value or whether missingness differed by group. It also does not report a sensitivity analysis to show how the estimate might change under other assumptions about missing outcomes.

Precision is not the same as importance. A confidence interval concerns uncertainty in an estimate; it does not determine whether a difference matters clinically. Nor does an interval for a group mean difference describe the range of individual participant outcomes. The posted summaries support a cautious statement about the direction and uncertainty of an unadjusted group-average comparison, but do not provide the fuller information needed to evaluate individual variation or the adjusted primary 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, not total event counts. They are illustrations calculated from the observed effect and are not the original trial's sample-size plan. The registry record does not report the trial's original sample-size assumptions or justification.

The calculation assumes a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It considers planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. For planning at 100% of the observed effect, the estimates are 86 participants per arm for 80% power and 115 participants per arm for 90% power. For planning at 75% of the observed effect, the estimates are 153 participants per arm for 80% power and 204 participants per arm for 90% power. For planning at 50% of the observed effect, the estimates are 343 participants per arm for 80% power and 459 participants per arm for 90% power.

In plain language, the 80% power column gives the estimated number of participants per arm under the stated assumptions for a design with an 80% chance of detecting the specified effect if that effect were the true one. The 90% power column gives the corresponding estimate for a design with a 90% chance. These are prospective planning quantities, not findings about the completed trial. The figures are not observed power.

As the effect to be detected becomes smaller relative to the observed effect, the required number of participants per arm increases in these illustrations. This is a general feature of study planning: smaller differences are harder to distinguish from random variation, all else being equal. The planning estimates use the observed effect as their reference, so they should not be treated as independent evidence that the observed effect will recur in another study.

12. How the Sample Size Is Derived

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

Formula

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

Worked calculation for this trial

1. Observed difference Δ = 2.220; pooled standard deviation σ = 5.183.
2. n = 2 × (1.96 + 0.842)² × 5.183² / 2.220² = 85.6, rounded up to 86 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 <- 2.22                      # observed difference in means
sigma <- 5.18349015872162                     # 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(679939)
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 = 86 (80% power), 115 (90% power)
 75% of observed effect: n per arm = 153 (80% power), 204 (90% power)
 50% of observed effect: n per arm = 343 (80% power), 459 (90% power)
power.t.test: 86.6 
Simulated power at n = 86 per arm: 0.803 (20000 replicates)

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

13. Multiplicity, Missing Data and Analysis Populations

The registry lists multiple outcome measures, including bone mineral density at several sites, bone-turnover markers, and vitamin D measures. It reports 19 secondary outcome results. The registry record does not report whether the primary endpoint was part of a prespecified hierarchy, whether secondary comparisons were exploratory, or whether any correction for multiple testing was used.

In general, testing many outcomes can increase the chance of observing one or more results that appear unusual even when there is no corresponding effect. A multiplicity strategy may define which outcomes are confirmatory, sequence tests, or adjust the threshold for interpretation. Without the trial's stated strategy, it is not possible to determine how the recalculated unadjusted p-value should be interpreted in relation to all analyses conducted. The registry record does not report an interim-analysis plan or whether interim analyses occurred.

The primary endpoint analysis population required an evaluable baseline and a Week 76 measurement taken within 10 days after stopping open-label metformin. The record does not report the number of participants with missing measurements, the reasons values were unavailable, or how missing values were handled. It also does not report sensitivity analyses addressing missing data. In general, complete-case summaries can be affected if measurement availability is related to treatment, outcome, or both; the posted information is not sufficient to assess that possibility.

The endpoint is labelled as using the safety population, but the additional evaluability criteria mean that the primary endpoint summary is not necessarily based on every participant in that population. The registry does not report a separate intention-to-treat analysis for this endpoint. It also does not report how protocol deviations, treatment discontinuation, or adherence were handled in the primary analysis. These details limit assessment of the extent to which the reported estimate follows the randomized comparison.

14. 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: urinary tract infection (0.9%), enteritis infectious (0.9%), cerebral infarction (0.9%), asthma (0.9%), coronary artery disease (0.9%).

The registry reports serious adverse event counts and numbers at risk separately for the double-blind and open-label groups. During the double-blind period, serious events affected 7 of 114 participants at risk in the rosiglitazone group and 5 of 111 at risk in the metformin group. During the open-label metformin period, serious events affected 0 of 76 at risk in the group previously assigned rosiglitazone and 0 of 84 at risk in the group previously assigned metformin. These are period-specific registry counts; they should not be combined across periods or interpreted as a single continuous exposure comparison.

The registry also lists serious-event terms, each reported as 0.9%: urinary tract infection, enteritis infectious, cerebral infarction, asthma, coronary artery disease, and wrist fracture. These listed terms do not replace the overall serious-event counts. The registry record does not report deaths for these groups, so mortality should not be inferred from the blank death entries. The open-label period had metformin in both arm descriptions, and its counts should be interpreted in that context. These summaries are descriptive and do not establish whether a treatment caused an event or whether the groups differed statistically.

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

The primary endpoint is a measure of average femoral-neck bone mineral density change at a specified follow-up time, among participants with the required evaluable measurements. The posted means show a negative mean percent change in the rosiglitazone-then-metformin group and a positive mean percent change in the metformin-throughout group. A simple unadjusted recalculation gives a mean difference of 2.22 percentage points in the direction of the metformin-throughout group. The recalculated unadjusted confidence interval lies above zero, and its p-value is 0.013.

Those findings describe the posted group summaries; they do not amount to the sponsor's adjusted treatment estimate. The registry describes an ANCOVA that includes treatment, baseline value, prior therapy, and region, but does not report the adjusted contrast or its inferential result. The endpoint was measured after the open-label period, during which both groups were described as receiving metformin. It therefore reflects the whole treatment sequence and the endpoint's timing, rather than a simple comparison of two concurrent drugs at the measurement visit.

The result does not show how bone density changed for each individual participant, whether other skeletal sites followed the same pattern, or whether any observed difference has clinical importance. The registry lists other bone-related outcomes, but those outcomes address different measures and time points. A group-average femoral-neck result cannot answer all questions about bone health.

The result also does not establish causation on its own. Randomization supports a comparative design, but interpretation still depends on analysis methods, completeness of follow-up, adherence, and missing-data handling. Several of these details are not reported in the registry record. The unadjusted recalculation should therefore be read as one statistical description of the available posted summaries, not as a substitute for the adjusted analysis or a broad conclusion about treatment effects.

16. Why This Trial Matters Statistically

This trial illustrates why the timing and sequence of treatment matter when interpreting an endpoint. The primary femoral-neck measurement was taken at Week 76+10 days, after a study described as having a double-blind period followed by open-label metformin. Both arm descriptions specify metformin during the open-label period. Consequently, the endpoint comparison reflects earlier double-blind assignment together with the later common treatment period; it is not simply a concurrent comparison of rosiglitazone and metformin at the final measurement.

It also demonstrates the difference between an adjusted analysis specified in the outcome description and an unadjusted comparison that can be reconstructed from group summaries. The registry names the ANCOVA covariates, but does not report the model-based treatment contrast. The unadjusted mean difference and its uncertainty offer a useful view of the posted group averages, yet they cannot reproduce the sponsor's adjusted analysis.

The endpoint population is another important statistical feature. The reported analysis required evaluable measurements at baseline and follow-up within the stated timing window. Without information about missing values or sensitivity analyses, readers cannot assess how fully the endpoint summaries represent all randomized participants. This is a general concern whenever analysis depends on having observed measurements.

Finally, multiple bone and biomarker outcomes appear in the record, while the multiplicity strategy is not reported. That gap matters because evidence from one endpoint should be interpreted in the context of the full set of planned and analyzed outcomes. The trial is therefore a useful example of how to distinguish the endpoint's scale, analysis population, treatment sequence, adjusted model, and summary-based recalculation instead of treating one p-value as a complete account of the evidence.

17. Limitations

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

The primary endpoint analysis includes only evaluable participants with both baseline and follow-up measurements in the stated window. The registry does not report the number of participants excluded from this analysis, the reasons for missing endpoint data, or the method used to address missingness. The extent to which the endpoint analysis retains the benefits of randomization cannot be assessed fully from the record.

The registry describes an adjusted ANCOVA but does not report its treatment contrast, confidence interval, or p-value. The available recalculation is unadjusted and cannot account for the listed covariates. The values therefore do not permit independent verification of the adjusted primary result.

The endpoint was measured after a period in which both groups were described as receiving metformin. This sequencing limits interpretation as a direct comparison of two treatments continuing through the final measurement. The registry does not report adherence, treatment discontinuation details, or further information about treatment changes.

Multiple outcomes are listed, but the multiplicity plan and any prespecified hierarchy are not reported. The record also does not report the original sample-size justification or interim-analysis details. Safety summaries are period-specific, and death counts are not reported in the registry record. These limitations do not negate the posted summaries, but they constrain how broadly and confidently they can be interpreted.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Study In Postmenopausal Women With Type 2 Diabetes Looking At Approved Diabetes Drugs And How They Affect Bone Health — NCT00679939.
• Registry record last updated 2018-04-18; status last verified 2018-03.
• Results first posted 2011-03-07.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

21. Record Summary

The registry record describes a completed phase 4, randomized, double-blind, multicenter, parallel-group study in postmenopausal women with type 2 diabetes. Enrollment was 226. The study compared rosiglitazone with metformin during the double-blind period, followed by open-label metformin in both arm descriptions.

The primary endpoint was adjusted percent change from baseline in femoral-neck bone mineral density at Week 76+10 days, assessed by DXA. The record describes an ANCOVA adjusted for treatment, baseline value, prior therapy, and region, and restricts the endpoint population to evaluable participants with the required measurements. The posted group means support an unadjusted recalculation, but the adjusted treatment contrast is not reported in the registry record.

The record lists additional bone and biomarker outcomes and period-specific serious adverse event counts. Details on missing-data handling, multiplicity, and the original sample-size justification are not reported in the registry record.