KEEPS Mammographic Density And Breast Health Ancillary Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for KEEPS Mammographic Density And Breast Health Ancillary Study (NCT02524561), a randomized na study. The primary outcome is BIRADS Breast Density.

2. Clinical Question

The distinction between the study’s broader scientific aim and its registered primary measure matters. The description proposes examining possible breast-related effects of low-dose hormone-therapy regimens. Yet the primary measure is explicitly anchored before randomization. Baseline category frequencies can describe the participants’ breast density and how those frequencies were distributed by randomization status. They do not, by themselves, show a change caused by treatment after randomization.

The ancillary study is connected to KEEPS, whose primary goal was to examine low-dose transdermal versus oral estrogen combined with cyclic micronized progesterone in relation to progression of subclinical atherosclerosis. The breast-health record identifies abnormal mammogram/biopsy as a secondary outcome. It does not report a numerical comparison of the active regimens with placebo for that outcome, nor does it report a treatment-effect estimate for the primary BIRADS measure.

3. Trial Design

The registry classifies the study as randomized, double masked, and parallel. It records enrollment of 517 and a completed status. The listed study dates are 2005-09 for start and 2013-12 for primary completion. Phase is recorded as NA. These descriptors give the broad structure and timeline, but do not establish how allocation was generated, how masking was maintained, or how participants were followed; those details are not reported in the registry record.

The record describes three groups: a CEE pill with active progesterone, an estradiol patch with active progesterone, and placebo. The CEE regimen is conjugated equine estrogens 0.45 mg/day, with placebo patch and micronized progesterone 200 mg daily for the first 12 days of each month. The patch regimen is transdermal estradiol 50 mcg/day, with placebo tablet and the same listed progesterone schedule. The placebo group received placebo tablet, placebo patch, and placebo progesterone.

The record places the ancillary work within the ongoing, already-funded Kronos Early Estrogen Prevention Study. Its description says that KEEPS examined low-dose oral versus transdermal estrogen with cyclic micronized progesterone, while the ancillary study proposed examining breast density, abnormal mammogram rates, and circulating estrogens. The registry record does not report the allocation ratio, recruitment details, adherence, duration of treatment for each participant, or the timing of follow-up breast assessments.

Although the trial is classified as randomized, the primary BIRADS measure is specified at baseline, prior to randomization. Randomized design is central to interpreting post-randomization comparisons, but it does not turn a pre-randomization measurement into a treatment effect. The record does not explain whether baseline category frequencies were intended as a descriptive measure, a balance check, or part of a later analysis. The statistical plan for those purposes is 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.

Baseline: Pre-randomization Mammogram

Year 1: Post Randomization Mammograms

Final: Year 3 or 4 Post-randomization

Analysis population for the primary outcome, as defined in the registry: Women with mammograms are included.

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: Population with mammographic density data from at least one time point during study.

6. Primary Endpoint

The registry lists the primary outcome as BIRADS Breast Density, measured over Baseline (Prior to Randomization).

The primary outcome is BIRADS breast density, recorded as the frequency of participants in each category at baseline, prior to randomization. The record describes BIRADS as a radiologist-assessed scale from 1 to 4: 1 is most fatty and least dense, while 4 is most dense. The stated population for this measure is women with mammograms.

Breast density is used as a structured imaging classification. In general, a prespecified category system can make a clinical observation easier to describe consistently than an informal narrative alone. Here, the record’s definition provides an ordered scale from lower to higher density, but it does not report category counts or a statistical comparison of categories. It also does not specify how the categories were handled analytically, such as whether they were compared individually or treated as an ordered outcome.

The timing is essential to interpretation. Since the measure is taken before randomization, it describes baseline density rather than a change after assigned treatment. It cannot establish whether a regimen altered density. The registry record does not report follow-up BIRADS results, an outcome timepoint after treatment, or an estimate comparing treatment groups on change in density.

The study description also names abnormal mammogram/biopsy as a secondary outcome and discusses interest in circulating estrogens. Those broader outcomes do not change the definition or timing of the registered primary measure. Results for the secondary outcome and circulating estrogen are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Randomization, masking, outcome definition, and analysis are separate parts of trial methodology. The record identifies the study as randomized, double masked, and parallel. In general, randomization aims to make assigned groups comparable at baseline on average, while masking can reduce the influence of treatment knowledge on care, assessment, or reporting. These design features do not by themselves establish that allocation was concealed, that masking remained intact, or that a particular statistical model was appropriate. The registry record does not report those operational details.

What does the baseline timing mean?

The registered BIRADS measure is taken before randomization. In general, a pre-randomization measure describes participant characteristics at entry and may help characterize groups. It is not an outcome caused by assigned treatment, because the treatment assignment has not yet occurred. A treatment-effect analysis would ordinarily require a post-assignment outcome or a defined change from baseline, with an analysis appropriate to the measurement and design. The record does not report such a density analysis or its statistical method.

BIRADS is a categorical measure with ordered labels. The record describes its endpoints but does not report how categories were summarized or modeled. In general, categorical outcomes may be summarized with category counts and proportions, while an ordered-category analysis may use the order of the labels. The appropriate method depends on the question and assumptions. No such method can be attributed to this study from the registry record, and no category counts are available to interpret.

What can a confidence interval tell a reader?

General explanation: a confidence interval expresses uncertainty around an estimated quantity under the model and procedure used. A narrow interval indicates greater precision than a wide interval, but precision is distinct from whether an effect is important. An interval also depends on the chosen outcome definition, population, and analysis. The record does not report a confidence interval for the primary BIRADS measure or the secondary abnormal mammogram/biopsy outcome, so uncertainty around a treatment contrast cannot be assessed from these entries.

How should significance testing be understood?

General explanation: a p-value measures how compatible the observed data are with a specified null hypothesis, assuming the statistical model and its conditions. It does not measure the probability that the null hypothesis is true, and it does not express the size of an effect. A small p-value alone would not establish importance, while a larger p-value would not demonstrate equivalence. No p-value or treatment-effect test for the primary outcome is reported in the registry record.

What do the planning assumptions describe?

The record describes a planning illustration based on an observed effect, a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. The effect to detect is described as a stated fraction of the observed effect. These are planning assumptions, not the trial’s own design justification. The record does not report the protocol’s sample-size justification or the effect-size assumptions used when the trial was designed.

General explanation: a sample-size calculation estimates the information needed to detect an effect under specified assumptions. Such calculations depend on the outcome, the anticipated effect, variability or event frequency, allocation, and chosen error rates. Changing the assumed effect changes the planning requirement. A planning illustration based on an observed effect is hypothetical; it is not observed power, and it does not establish that the trial was adequately powered for an outcome. The posted record does not support a conclusion about power for the registered breast-density measure.

The illustration’s two-arm allocation assumption also needs to be kept distinct from the listed treatment groups. The record describes the CEE pill regimen, estradiol patch regimen, and placebo. A two-arm planning contrast does not by itself identify which clinical comparison was used or establish that all treatment comparisons were planned on the same basis. The comparison-specific details and numerical planning results are not reported in the registry record.

Why do multiple categories matter?

The registry notes that the primary outcome has several categories or time points and was not reduced to one comparison. General explanation: when a study considers multiple outcomes, categories, timepoints, or group contrasts, the chance of at least one apparently notable finding can increase if many tests are conducted without an appropriate plan. Multiplicity procedures may address this issue, but their selection depends on the hypotheses and analytic plan. The record does not report a multiplicity strategy, a hierarchy of tests, or adjusted results. The note about the outcome not being reduced to one comparison does not establish which comparisons, if any, were analyzed.

For categorical outcomes, interpretation also depends on how missing observations are handled and which participants are included. General explanation: excluding participants with missing measurements can change the analyzed population and may introduce bias if missingness is related to the outcome or treatment. Other methods may be appropriate under stated assumptions. The registry record does not report missing-data counts, reasons for missingness, imputation methods, or analysis populations. Therefore, neither the completeness of the BIRADS data nor the assumptions underlying any analysis can be evaluated from this record.

10. Multiplicity, Missing Data and Analysis Populations

The registry identifies four outcome measures and zero analyses. It also states that the primary outcome has several categories or time points and was not reduced to one comparison. This leaves the reader without a reported primary hypothesis test or a set of outcome-specific estimates in the registry record. The record does not state whether results were analyzed but not entered, reported elsewhere, or not analyzed; those possibilities cannot be distinguished from the posted information.

Multiplicity concerns can arise when investigators examine multiple outcomes, categories, timepoints, or contrasts. General explanation: if many statistical tests are considered, interpreting each test as though it were the only one can increase the risk of a chance finding. Prespecified hierarchies or adjustment methods can help define how evidence should be assessed. The registry record does not report whether any such method was planned or used, so it is not possible to determine how multiplicity was handled.

Missing-data handling is also not reported. The record does not give the number of participants with baseline mammograms, the completeness of category assessments, reasons for missing measurements, or any imputation or sensitivity-analysis approach. It specifies women with mammograms as the population for the primary measure, but does not say whether that is the full enrolled population or how eligibility for this measure was determined.

The analysis population is not reported in the registry record. In general, an analysis may include all randomized participants, participants with available outcome data, or a defined subset, and each choice affects interpretation. The record does not state whether an intention-to-treat approach was used, how protocol deviations were treated, or whether analyses were adjusted for baseline characteristics. These omissions prevent assessment of the analytic population and its relationship to the randomized groups.

11. 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 0% reporting threshold.

The registry lists 0 participants with serious adverse events and 0 with other adverse events in each of the three reported groups. The corresponding at-risk counts are 168 for the CEE pill with active progesterone group, 162 for the estradiol patch with active progesterone group, and 187 for placebo. These are the posted registry figures; they should not be interpreted as comparative rates or as evidence that the regimens have equal safety.

The entries do not provide event terms, event definitions, observation periods, or information about how adverse events were collected. The list of serious-event terms is empty. Deaths are recorded as null rather than as a reported count, so the registry record does not report deaths. A zero affected count in these entries is not equivalent to proof that no event occurred or that all safety outcomes were captured.

In general, safety comparisons depend on consistent surveillance, clear definitions, and adequate follow-up, as well as on how participants who discontinue treatment are handled. The registry entries do not provide those details. They therefore describe the recorded counts only and cannot establish the presence or absence of a safety difference between groups.

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

The record establishes that this was a randomized, double-masked, parallel study and describes two active hormone regimens and a placebo group. It defines baseline BIRADS breast density as the primary measure and identifies abnormal mammogram/biopsy as a secondary outcome. It also documents the study’s broader interest in breast density, abnormal mammograms, and circulating estrogens in the KEEPS setting.

The record does not report BIRADS category counts or a post-randomization comparison of breast density. Because the primary measure is explicitly baseline and prior to randomization, it cannot show that any treatment caused a change in breast density. No numerical result for abnormal mammogram/biopsy or circulating estrogens is reported either. Accordingly, the posted information does not establish the direction, size, or precision of a treatment effect on these breast-health outcomes.

General explanation: a randomized design can support causal interpretation when outcomes are measured after assignment and the analysis appropriately reflects assignment, follow-up, and missing data. Design labels alone are not results. They do not replace estimates, uncertainty intervals, outcome definitions, or transparent reporting of the analysis population. In this record, those elements are absent for the primary treatment-effect question.

The absence of a reported analysis should not be read as evidence of no effect, no difference, or equivalence. Equivalence requires a defined margin and an analysis capable of assessing whether uncertainty lies within that margin; neither is reported. Conversely, the study description’s rationale and proposed aims do not demonstrate that a treatment effect occurred. The registry record supports a description of the study structure and its intended subject matter, but not a quantitative conclusion about efficacy or breast-health risk.

The adverse-event entries likewise have a limited meaning: they give zero recorded affected participants for the displayed categories and group entries. They do not establish that there were no adverse events over the full study, because definitions and follow-up details are not reported and the death entries are null. The appropriate interpretation is therefore descriptive and bounded by what the record actually reports.

13. Why This Trial Matters Statistically

This record is statistically informative in part because it makes the timing of its primary measure explicit. BIRADS density is recorded at baseline, prior to randomization. That detail prevents a common interpretive error: treating a baseline group description as evidence that assigned treatment changed an outcome. Study design and outcome timing must be considered together before drawing a causal interpretation.

The study also illustrates the difference between a research rationale and a reported result. The description explains why investigators proposed examining breast density, abnormal mammograms, and circulating estrogens alongside a low-dose hormone-therapy trial. The registry record does not provide numerical results for those treatment questions. A plausible question and a randomized design are not substitutes for a posted estimate and its uncertainty.

Another statistical issue is the outcome’s categorical structure. BIRADS is described as a scale from 1 to 4, and the registry notes that the primary outcome has several categories or time points rather than a single reduced comparison. This makes the analytic question important: category-specific description, an ordered-category contrast, and a treatment-related change are not interchangeable. The record does not report which analysis was chosen.

Finally, the record shows why reporting analysis populations, missing-data handling, and multiplicity methods matters. Without those details, readers cannot determine how complete the outcome data were, which participants contributed to an analysis, or how multiple comparisons were interpreted. The study is therefore a useful example of how registry design details can clarify the scope of a question while an absence of posted outcome analyses limits what can be concluded.

14. 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 main limitation for interpretation is the lack of reported outcome results. Although the registry identifies the primary BIRADS measure and secondary abnormal mammogram/biopsy outcome, it gives no category frequencies, treatment contrasts, effect estimates, confidence intervals, or p-values. The record therefore does not allow assessment of the magnitude or precision of any breast-health comparison.

The primary measure’s timing is another specific limitation: it is defined as baseline, prior to randomization. That measure can describe density at entry but cannot establish a treatment-related change. The registry does not report follow-up BIRADS measurements or an analysis comparing change from baseline.

The record does not explain how BIRADS categories were analyzed, how abnormal mammogram/biopsy was defined, or how circulating estrogens were measured. It also does not report the assessment schedule, duration of treatment or follow-up for individual participants, allocation ratio, adherence, or handling of treatment discontinuation. These details are not reported in the registry record.

Missing-data procedures, analysis populations, and multiplicity handling are also absent. The record does not say how many participants had evaluable mammograms, how missing observations were handled, or whether analyses followed an intention-to-treat principle. Since the primary outcome was not reduced to one comparison and no analyses are posted, the analytic strategy and its implications cannot be assessed.

Finally, the safety entries have limited context. The record lists zero affected participants in the displayed adverse-event categories, but does not report definitions, surveillance methods, event terms, or follow-up periods. Deaths are not reported. These limits mean the entries cannot serve as a comprehensive account of safety or as a reliable basis for comparing safety between groups.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: KEEPS Mammographic Density And Breast Health Ancillary Study — NCT02524561.
• Registry record last updated 2017-03-08; status last verified 2017-01.
• Results first posted 2017-03-08.
• 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.

18. Record Summary

The registry describes the KEEPS Mammographic Density And Breast Health Ancillary Study as randomized, double masked, and parallel, with enrollment of 517 and completed status. The recorded start is 2005-09 and primary completion is 2013-12. The listed groups are CEE pill with active progesterone, estradiol patch with active progesterone, and placebo.

The primary measure is baseline BIRADS breast density among women with mammograms, recorded by category according to randomization status. The record defines the scale from 1 to 4 and lists abnormal mammogram/biopsy as a secondary outcome. It reports four outcome measures and zero analyses, with no primary outcome comparison reported. The treatment-effect results, missing-data methods, analysis populations, and multiplicity procedures are not reported in the registry record.

The posted adverse-event entries show 0 serious and 0 other affected participants in each group. The record does not report deaths. These entries describe the registry figures but do not provide enough detail for a comprehensive safety assessment.