This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Predictive Value of Bone Turnover Markers During Discontinuation With Alendronate (NCT03051620), a registered phase not stated study. The primary outcome is If Carboxy-terminal Collagen Crosslinks (CTX) Three and Six Months After Stopping Alendronate Predicted TH BMD (Total Hip BMD) Loss Above the Least Significant Change at Month 12 at the Individual Level..
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03051620 |
| Phase | Phase not stated |
| Condition | Osteoporosis |
| Allocation | Not Stated |
| Design model | Not Stated |
| Masking | Not Stated |
| Enrollment | 142 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Aarhus University Hospital |
| Results posted | Yes |
2. Clinical Question
Population
People with osteoporosis who stopped alendronate. The registry lists enrollment of 142, while the brief summary describes a cohort comprising 136 patients; the relationship between those figures is not explained in the registry record.
Intervention
Discontinuation of alendronate. The record describes the study as a cohort study of people stopping treatment, rather than a randomized treatment assignment.
Comparator
No like-for-like control group is described. Allocation is not stated, and the registry record does not report a randomized comparison.
Primary question
Could CTX measured 3 and 6 months after stopping alendronate predict total hip BMD loss beyond the least significant change at month 12, at the individual level?
The study focuses on prediction after treatment discontinuation, not on a comparison of continuing alendronate with stopping it. Its central statistical question is whether a biochemical marker measured after discontinuation can distinguish people who later have a change in total hip BMD exceeding the least significant change from those who do not.
The registry describes receiver operating characteristic analysis for CTX at 3 and 6 months. It also lists secondary questions concerning changes in bone turnover markers, BMD at the lumbar spine and total hip, and whether baseline bone turnover markers predict later BMD changes. The registered summary does not establish that these markers can identify an individual’s future outcome; it reports that none of the described ROC curves had an AUC above 60%.
3. Trial Design
The study is registered as a cohort study of people with osteoporosis stopping alendronate. The listed intervention is “Discontinue alendronate.” The registry does not describe a randomized allocation or a control group for a like-for-like comparison; allocation is not stated. The design therefore addresses prediction within a discontinuation cohort, not the comparative effect of stopping versus continuing treatment.
The registry lists enrollment of 142. Its brief summary describes a cohort comprising 136 patients. The relationship between those entries is not explained in the registry record, so they should not be treated as interchangeable counts or used to infer the number included in a particular analysis. The sponsor is Aarhus University Hospital. The study start is listed as 2017-02-01, primary completion as 2019-01-31, and status as COMPLETED.
The primary question concerns CTX measured 3 and 6 months after stopping alendronate and total hip BMD loss beyond the least significant change at month 12. Other registered outcomes cover percent changes in bone turnover markers at 3 and 6 months and BMD after one and two years, participants whose CTX increased beyond the least significant change, participants with BMD loss beyond that threshold at the lumbar spine and total hip, and prediction using baseline markers.
The registry lists 5 outcome measures and no analysis entries. It does not report a detailed protocol, a sample-size justification, a planned interim analysis, or a complete statistical analysis plan. Details such as eligibility criteria, recruitment procedures, and the analysis population 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.
Overall Study
| Stage | Study Population |
|---|---|
| STARTED | 142 |
| COMPLETED | 124 |
| NOT COMPLETED | 18 |
Reasons for not completing this period:
| Reason | Study Population |
|---|---|
| Lost to Follow-up | 6 |
| n=124 enrolled in extension (all completed) | 12 |
Analysis population for the primary outcome, as defined in the registry: None of the ROC curves investigating the ability of change in CTX to predict BMD loss beyond LSC at any site at month 12 at the individual level had AUC above 60% meaning that it was not possible to identify a specific cut-off value that could predict significant bone loss at the individual level.
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.
| Characteristic | Unit | Study Population |
|---|---|---|
| Participants analysed | 142 | |
| Age, Continuous | years | 68 (SD 6) |
| Sex: Female, Male: Female | Participants | 122 |
| Sex: Female, Male: Male | Participants | 20 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 0 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 142 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 0 |
| Bone mineral density (BMD) lumbar spine | g/cm3 | 0.806 (SD 0.09) |
| Bone mineral density (BMD) total hip | g/cm3 | 0.780 (SD 0.09) |
| Baseline CTX | ug/L | 0.21 (SD 0.1) |
| Baseline PINP | ug/L | 28 (SD 9.5) |
6. Primary Endpoint
The registry lists the primary outcome as If Carboxy-terminal Collagen Crosslinks (CTX) Three and Six Months After Stopping Alendronate Predicted TH BMD (Total Hip BMD) Loss Above the Least Significant Change at Month 12 at the Individual Level., measured over Baseline and one year after baseline.
| Element | Registry entry |
|---|---|
| Result type | Mean |
| Measure of spread | Standard Deviation |
| Unit | percentage change |
| Endpoint classification | Continuous Endpoint |
The primary endpoint asks whether CTX measured 3 and 6 months after alendronate discontinuation predicts total hip BMD loss greater than the least significant change at month 12, for an individual. The outcome is therefore framed as a prediction or classification question: can a marker measurement separate people who later cross the BMD-loss threshold from those who do not?
The least significant change is used to distinguish a change that exceeds the specified measurement threshold from a smaller observed change. In general, such a threshold helps avoid interpreting every measured fluctuation as a meaningful change. The registry does not give the numerical least significant change used in this analysis, nor does it report the rule used to define the predicted positive and negative classifications in detail.
The record says receiver operating characteristic curves were constructed and describes the endpoint as a percentage change, with mean and standard deviation as the listed summary. That description sits alongside the ROC-based prediction question; the registry does not clarify how the mean-and-standard-deviation reporting relates to the classification endpoint. Its narrative conclusion is that none of the ROC curves investigating CTX change had an AUC above 60%, and that a specific cutoff value for predicting significant bone loss at the individual level could not be identified.
The endpoint is useful for assessing discrimination, but it does not by itself establish why a person experiences BMD loss or whether acting on a marker result changes outcomes.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry’s primary-outcome narrative reports that none of the ROC curves assessing CTX change at 3 or 6 months had an AUC above 60% for predicting total hip BMD loss beyond the least significant change at month 12. The record interprets this as meaning that the analysis could not identify a specific CTX cutoff value that predicted significant bone loss at the individual level. This is a descriptive statement about the posted result, not a treatment comparison.
An AUC describes how well a marker-based rule ranks people who meet an outcome definition relative to people who do not. In general, a larger AUC indicates better discrimination across possible thresholds. An AUC alone does not specify which cutoff should be used, how many people would be classified above or below that cutoff, or the practical consequences of classification. The registry narrative gives no exact AUC values, confidence intervals, sensitivity, specificity, or cutoff estimates. The precision of the reported predictive performance therefore cannot be assessed from the registry narrative alone.
The phrase “no AUC above 60%” should not be expanded into a claim that every marker value has no relationship with BMD change, or that no individual could be predicted accurately. The record does not provide the underlying estimates or participant-level distributions needed to make those more specific assessments. It reports a limit on the performance described by the ROC analysis and says that a specific cutoff could not be identified.
The result also does not answer whether continuing alendronate would prevent bone loss, because the record does not describe a randomized continuation group or a like-for-like control comparison. It does not establish whether stopping treatment caused any observed change. The study’s question is narrower: whether CTX after discontinuation can serve as an individual-level predictor of a later BMD-loss threshold within the cohort under study.
The registry lists 5 outcome measures but no analysis entries. The primary-outcome narrative is therefore the main available description of the result, while numerical estimates and uncertainty measures are not reported in the registry record. No p-value or confidence interval is reported for the predictive result. Without such measures, the degree of statistical uncertainty cannot be quantified from the record.
The enrollment entry and brief summary also give different cohort figures: enrollment is 142, while the summary describes 136 patients. The record does not explain the relationship between these figures or specify which participants contributed to the primary ROC analysis. This limits interpretation of the analysis population and makes it important not to infer its size from either registry entry.
9. Statistical Methods Explained
The primary method named in the registry is receiver operating characteristic analysis. It is used when an analysis evaluates how well a marker or model distinguishes between two outcome categories across possible decision thresholds. In this study, the marker is CTX after alendronate discontinuation, and the outcome category concerns whether total hip BMD loss exceeded the least significant change at month 12.
What does an ROC curve assess?
In general, an ROC curve traces the trade-off between sensitivity and specificity as a classification threshold changes. Sensitivity describes the ability to identify people who meet the outcome definition; specificity describes the ability to identify people who do not. A strict threshold may classify fewer people as positive, while a more permissive threshold may classify more people as positive. The curve summarizes these trade-offs rather than selecting a single threshold automatically.
The registry says ROC curves were constructed for CTX measured 3 and 6 months after stopping alendronate. It reports that none of the curves had an AUC above 60%, and states that a specific cutoff value predicting significant bone loss at the individual level could not be identified. The record does not report the curve coordinates, the sensitivity and specificity at candidate cutoffs, or the method used to choose a cutoff.
What does the AUC mean?
In general, the area under an ROC curve (AUC) summarizes discrimination over the range of possible thresholds. It does not show whether predictions are calibrated to the observed probability of an outcome, and it does not measure whether using a prediction improves decisions. AUC also does not communicate the consequences of false-positive or false-negative classifications. Those questions require additional information about outcome frequencies, decision thresholds, and the intended use of the prediction.
The registry’s wording is a bound on the reported AUCs, not a list of exact estimates. No confidence intervals or other uncertainty measures for the AUCs are reported in the registry record. The result therefore does not show how precisely the discrimination estimates were measured or how much uncertainty surrounds the comparison of CTX measurement times.
Why use a least significant change threshold?
In general, a least significant change threshold defines how large a measured change must be to count as exceeding expected measurement variability under the chosen method. Using a threshold turns a continuous BMD change into an outcome category for prediction. That can make a classification analysis possible, but it also means that results depend on the threshold and on the measurement procedures underlying it. The numerical threshold and its derivation are not reported in the registry record.
Once the outcome is divided into categories, the ROC analysis evaluates discrimination for that defined category. It does not retain all information about the size of BMD change. A person just beyond the threshold and a person with a much larger change may both be classified in the same outcome category. The registry does not report additional analyses that would show how CTX relates to the continuous amount of BMD change.
What details remain unclear?
The record identifies the marker timings, outcome threshold, assessment time, and ROC method, but it does not report the full analysis specification. The approach to missing marker or BMD measurements, the handling of repeated measurements, and any adjustment for other predictors are not reported in the registry record. Nor are the number of participants contributing to each ROC curve, the number meeting the BMD-loss definition, or the uncertainty estimates available in the registered narrative.
The record also labels the primary outcome as a percentage change and lists mean and standard deviation, while its narrative describes ROC curves for an individual-level threshold outcome. These descriptions may refer to different aspects of the outcome reporting, but the registry does not explain their relationship. No additional model-performance measures, such as calibration or validation in a separate cohort, are reported in the registry record.
General interpretation of ROC analysis should not be confused with a claim that all those methods were used in this study. The registry specifically names ROC curves; other analytic choices and validation procedures are not reported. Sample-size figures associated with planning are planning illustrations, not observed power. The stated illustration assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect defined as a stated fraction of the observed effect. Those assumptions describe the illustration, not the cohort design or a stated rationale for the study’s enrollment.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists 5 outcome measures, including CTX changes at 3 and 6 months, BMD outcomes after one and two years, changes at the lumbar spine and total hip, and prediction using baseline markers. The primary question also involves more than one CTX measurement time. Considering several markers, time points, and outcome locations can create multiple opportunities to observe an apparently useful association or prediction result.
Whether and how the study accounted for multiplicity is not reported in the registry record. No hierarchy of secondary analyses, multiplicity adjustment, or prespecified decision rule for interpreting the collection of outcomes is described. The primary-outcome narrative does not give inferential estimates that would allow readers to assess the effect of any such adjustment.
Missing-data methods are not reported in the registry record. The record does not state how missing CTX measurements or BMD measurements were handled, whether any imputation was used, or whether analysis required complete measurements at particular time points. It also does not report the number contributing to each analysis. The enrollment entry and brief-summary cohort figure differ, and the registry does not explain which figure corresponds to the primary analysis population.
Analysis populations are not reported in the registry record. In particular, the record does not specify inclusion rules for the ROC analyses or describe whether all enrolled participants, only participants with the required measurements, or another defined population was analyzed. It lists no analysis entries. These omissions make it difficult to judge selection effects, the role of incomplete follow-up, and the precision of the primary result.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Study Population | 7 | 142 | 4.9% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Study Population | 0 of 142 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 5% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Study Population | 68 of 142 |
Most frequent serious adverse event terms
| Event term | Organ system | Study Population |
|---|---|---|
| Cancer | Investigations | 7 of 142 |
Most frequent non-serious adverse event terms
| Event term | Organ system | Study Population |
|---|---|---|
| Fracture | Musculoskeletal and connective tissue disorders | 15 of 142 |
| Arthralgia, osteoarthritis, back pain | Musculoskeletal and connective tissue disorders | 15 of 142 |
| Arrhythmia, acute myocardial infarction, hypertension | Cardiac disorders | 10 of 142 |
| Upper and lower respiratory tract infection, pneumonia, bronchitis | Respiratory, thoracic and mediastinal disorders | 10 of 142 |
| Cataract, glaucoma | Eye disorders | 9 of 142 |
| Lower urinary symptoms, infection and kidney stones | Renal and urinary disorders | 9 of 142 |
Serious events with the highest arm-level rate: cancer (4.9%).
The registry reports adverse-event figures for a single group titled “Study Population.” It records 142 at risk for serious adverse events and 7 affected, alongside 142 at risk for other adverse events and 68 affected. The death count is 0. The most frequent serious term listed is Cancer, at 4.9%.
These are registry counts, not a comparison between randomized treatment groups. The study is described as a cohort after alendronate discontinuation, and no like-for-like control group is reported. The figures therefore do not establish whether an event was caused by stopping alendronate or by any other study-related exposure. The registry entry does not provide the detailed event definitions, individual event narratives, or an attribution analysis needed to answer those questions.
The number at risk is the registry’s stated denominator for each category. The serious-event and other-event counts should be read as separate adverse-event summaries; the record does not establish whether people could appear in both. The listed cancer percentage is reported as the top serious term, but the registry narrative does not provide further detail about the cases or their timing.
These summaries describe what was recorded in the registry. They do not establish that an event was or was not caused by the study condition or the discontinuation of treatment.
12. What the Result Does, and Does Not, Mean
The primary result addresses whether CTX measured after alendronate discontinuation could identify people with total hip BMD loss beyond the least significant change at month 12. The registry narrative reports that none of the ROC curves had an AUC above 60% and concludes that a specific individual-level cutoff could not be identified. This indicates a limitation in the predictive discrimination described in the record.
It does not mean that CTX and BMD change are unrelated in every sense. The record does not give exact AUC estimates, confidence intervals, or analyses of continuous BMD change, so it does not allow a fuller assessment of association or uncertainty. Nor does the reported AUC statement show that every individual prediction was wrong or that prediction is impossible under every model or setting.
The result is not evidence from a randomized comparison of stopping with continuing alendronate. The registry describes a cohort of people stopping treatment and does not report a like-for-like control group. It therefore cannot, on its own, establish the effect of discontinuation on BMD or compare treatment strategies.
The finding also does not show whether a CTX-based decision rule would improve care or outcomes. ROC discrimination is only one aspect of a prediction tool; calibration, practical thresholds, consequences of classification errors, and validation in other data would also matter. Those details are not reported in the registry record.
Finally, the two cohort figures in the record are not reconciled, and the number of participants contributing to the ROC analyses is not reported. The outcome narrative is useful as a statement of the registered result, but the absent numerical estimates and analysis details limit how precisely that result can be evaluated.
13. Why This Trial Matters Statistically
This study is statistically interesting because it asks a prediction question rather than only describing average changes after treatment discontinuation. A marker can be associated with an outcome at the group level without accurately identifying which individual will experience that outcome. ROC analysis addresses discrimination across possible thresholds, making the distinction between an overall association and individual-level prediction central to interpreting the result.
The use of a least significant change threshold also highlights a common analytic choice: converting a continuous measurement into a category. That makes a classification question possible, but it can discard information about the size of the measured change and makes interpretation dependent on the chosen threshold. The registry’s focus on BMD loss beyond that threshold illustrates why the endpoint definition matters when assessing a biomarker’s predictive value.
The record’s conclusion is appropriately specific: no described curve had an AUC above 60%, and no specific cutoff could be identified for predicting significant bone loss at the individual level. It does not establish that CTX has no relationship with BMD change, nor does it test whether stopping alendronate is preferable to continuing it. Keeping those claims separate is essential to sound statistical interpretation.
The record also shows why transparent reporting matters. Exact AUCs, uncertainty intervals, analysis denominators, missing-data procedures, and the relationship between the listed cohort figures would help readers assess the result’s precision and reproducibility. Without those details, the narrative communicates the broad conclusion but not the full strength or uncertainty of the evidence.
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 registry record leaves important details about the cohort and analysis unresolved. It lists enrollment of 142, while the brief summary describes 136 patients. It does not explain the relationship between the entries or identify the number included in the primary ROC analyses. The analysis denominators and number of outcome events are not reported in the registry record.
The study is described as a cohort of people stopping alendronate, without a reported randomized comparison or like-for-like control group. That design limits interpretation to prediction within the described cohort and does not support a direct comparison of discontinuation with continued treatment. Allocation is not stated.
The registry does not report exact AUC values, confidence intervals, sensitivity, specificity, selected cutoffs, or detailed outcome distributions. It also does not provide the numerical least significant change threshold or explain how it was established. The result’s precision and the performance at any particular cutoff cannot therefore be assessed from the narrative.
Missing-data procedures, eligibility criteria, analysis-population definitions, adjustment for other predictors, and any validation approach are not reported in the registry record. The record lists 5 outcome measures and no analysis entries, while multiplicity handling is also not described. The outcome description’s percentage-change and mean-and-standard-deviation labels are not reconciled with the ROC-based classification narrative.
These gaps do not negate the registry’s reported conclusion, but they restrict the degree to which readers can evaluate its robustness, generalizability, and reproducibility.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Predictive Value of Bone Turnover Markers During Discontinuation With Alendronate — NCT03051620.
- Registry record last updated 2021-03-18; status last verified 2021-03.
- Results first posted 2021-03-18.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
NCT03051620 is a completed cohort study sponsored by Aarhus University Hospital, concerning people with osteoporosis stopping alendronate. The registry lists enrollment of 142; the brief summary describes a cohort comprising 136 patients, and the relationship between these figures is not explained. The listed study start is 2017-02-01 and primary completion is 2019-01-31.
The primary question was whether CTX at 3 and 6 months after stopping alendronate predicted total hip BMD loss beyond the least significant change at month 12. The registry narrative reports that none of the ROC curves had an AUC above 60% and says a specific individual-level cutoff could not be identified. Exact AUC estimates and uncertainty measures are not reported in the registry record.
The record lists 5 outcome measures and no analysis entries. It does not report a randomized comparison, detailed missing-data methods, analysis populations, or a sample-size justification.