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 A Randomized Phase 3 Study to Evaluate Two Formulations of Romosozumab in Postmenopausal Women With Osteoporosis (NCT02016716), a randomized phase 3 study. The primary outcome is Percent Change From Baseline in Bone Mineral Density (BMD) at the Lumbar Spine.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT02016716 |
| Phase | Phase 3 |
| Condition | Postmenopausal Osteoporosis |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Enrollment | 294 |
| Arms in the primary outcome | 3 |
| Lead sponsor | Amgen |
| Results posted | Yes |
2. Clinical Question
Population
Postmenopausal women with osteoporosis. The registry record reports enrollment of 294 participants. The primary analysis population was narrower: randomized participants with both a baseline and a month 6 lumbar spine DXA BMD measurement.
Intervention
Romosozumab at a concentration of 90 mg/mL, given as 210 mg monthly for 6 months using 2 subcutaneous injections of 1.17 mL each.
Comparator
Romosozumab at a concentration of 70 mg/mL, also given as 210 mg monthly for 6 months, using 3 subcutaneous injections of 1.0 mL each. Matching placebo arms were also included.
Primary question
How did the percent change in lumbar spine BMD from baseline to month 6 compare between the 90 mg/mL and 70 mg/mL romosozumab formulations, and did the registry’s reported analysis assess noninferiority?
The study focused on formulation concentration rather than a comparison of different monthly romosozumab amounts: both active regimens are described as 210 mg monthly. The injection number and volume differed between the active formulations. The record identifies a noninferiority analysis of the active groups, but it does not report the noninferiority margin or a formal conclusion about whether the criterion was met.
Placebo participants were assigned matching injections corresponding to the 90 mg/mL and 70 mg/mL formulations. The posted primary comparison, however, is between the active 70 mg/mL and 90 mg/mL groups. The primary endpoint was a change in a continuous measurement of lumbar spine BMD, not a count of clinical events. Secondary outcomes listed in the record included total hip and femoral neck BMD, P1NP, and CTX.
3. Trial Design
This was a completed, randomized, double-masked, parallel-group phase 3 study. The official title describes it as multicenter and states that its purpose was to evaluate noninferiority of romosozumab at a 90 mg/mL concentration compared with a 70 mg/mL concentration. The sponsor was Amgen. The record gives a start date of 2013-12-03 and a primary completion date of 2014-09-08.
Four intervention groups are described: romosozumab 90 mg/mL, matching placebo 90 mg/mL, romosozumab 70 mg/mL, and matching placebo 70 mg/mL. The active groups each received 210 mg monthly for 6 months. The 90 mg/mL regimen used 2 subcutaneous injections of 1.17 mL, while the 70 mg/mL regimen used 3 subcutaneous injections of 1.0 mL. The placebo regimens matched the injection schedule and volume for their corresponding formulation groups.
The registry reports enrollment of 294. For the primary outcome, the analysis population is described as all randomized participants who had both a baseline and a month 6 lumbar spine DXA BMD measurement. The active-group outcome estimates are reported for 110 participants in the 70 mg/mL group and 117 in the 90 mg/mL group; the placebo estimate is reported for 46 participants. These outcome counts should not be treated as interchangeable with enrollment or the safety denominators, which differ by group and analysis.
The randomization procedure, allocation ratio, concealment details, and reasons for participants not appearing in each outcome population are not reported in the registry record. The record also does not report how the masking was implemented or whether masking was assessed.
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: Eligible participants were randomized in a 22:5:22:5 ratio to receive * romosozumab 90 mg/mL * placebo 90 mg/mL * romosozumab 70 mg/mL * placebo 70 mg/mL For efficacy analyses the 2 placebo groups were combined, for safety analyses the data were reported separately.
Overall Study
| Stage | Placebo | Romosozumab 70 mg/mL | Romosozumab 90 mg/mL |
|---|---|---|---|
| STARTED | 53 | 118 | 123 |
| Received Study Treatment | 53 | 118 | 123 |
| COMPLETED | 47 | 110 | 117 |
| NOT COMPLETED | 6 | 8 | 6 |
Reasons for not completing this period:
| Reason | Placebo | Romosozumab 70 mg/mL | Romosozumab 90 mg/mL |
|---|---|---|---|
| Withdrawal by Subject | 5 | 5 | 5 |
| Protocol-specified Criteria | 0 | 1 | 0 |
| Lost to Follow-up | 1 | 2 | 1 |
Analysis population for the primary outcome, as defined in the registry: All randomized participants who had a baseline and a month 6 lumbar spine DXA BMD measurement
5. Baseline Characteristics
The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.
Population: All randomized participants (full analysis set)
| Characteristic | Unit | Placebo | Romosozumab 70 mg/mL | Romosozumab 90 mg/mL | Total |
|---|---|---|---|---|---|
| Participants analysed | 53 | 118 | 123 | 294 | |
| Age, Continuous | years | 68.4 (SD 8.0) | 67.4 (SD 7.1) | 67.7 (SD 7.6) | 67.7 (SD 7.5) |
| Age, Customized: < 65 years | Participants | 23 | 46 | 46 | 115 |
| Age, Customized: ≥ 65 years | Participants | 30 | 72 | 77 | 179 |
| Sex: Female, Male: Female | Participants | 53 | 118 | 123 | 294 |
| Sex: Female, Male: Male | Participants | 0 | 0 | 0 | 0 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 2 | 6 | 9 | 17 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 51 | 112 | 114 | 277 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 0 | 0 | 0 | 0 |
| Race/Ethnicity, Customized: Asian | Participants | 0 | 2 | 0 | 2 |
| Race/Ethnicity, Customized: Black (or African American) | Participants | 0 | 1 | 0 | 1 |
| Race/Ethnicity, Customized: White | Participants | 53 | 115 | 123 | 291 |
| Lumbar Spine Bone Mineral Density (BMD) T-score | T-score | -2.842 (SD 1.033) | -2.965 (SD 0.960) | -3.029 (SD 0.687) | -2.970 (SD 0.871) |
| Total Hip BMD T-score | T-score | -2.057 (SD 0.634) | -1.819 (SD 0.673) | -1.828 (SD 0.649) | -1.866 (SD 0.660) |
| Femoral Neck BMD T-score | T-score | -2.304 (SD 0.517) | -2.139 (SD 0.611) | -2.018 (SD 0.593) | -2.118 (SD 0.595) |
| Serum Procollagen Type 1 N-telopeptide (P1NP) | μg/L | 49.0 (38.0 to 61.0) | 53.0 (40.0 to 61.0) | 50.0 (40.0 to 64.0) | 51.0 (40.0 to 63.0) |
| Serum Type 1 Collagen C-telopeptide (CTX) | ng/L | 426.5 (296.5 to 560.5) | 440.0 (288.0 to 613.0) | 402.0 (299.0 to 530.0) | 426.0 (295.0 to 590.0) |
6. Primary Endpoint
The registry lists the primary outcome as Percent Change From Baseline in Bone Mineral Density (BMD) at the Lumbar Spine, measured over Baseline and month 6.
| Element | Registry entry |
|---|---|
| Result type | Least Squares Mean |
| Measure of spread | 95% Confidence Interval |
| Unit | percent change |
| Endpoint classification | Continuous Endpoint |
The primary endpoint was percent change from baseline in lumbar spine bone mineral density at month 6. BMD was measured by dual-energy X-ray absorptiometry (DXA). The record describes the reported group values as least-squares means and gives 95% confidence intervals. The endpoint is a continuous measure: it expresses change in a measured quantity rather than whether a participant experienced a particular event.
In general, BMD provides a quantitative way to assess bone mineral density at a specified skeletal site. Measuring change from baseline allows the analysis to describe how the measured value changed during the study period. The lumbar spine was the site specified for the primary outcome; total hip and femoral neck BMD were listed as secondary outcomes. The registry record does not explain why the lumbar spine was selected over those other sites, or state what magnitude of change would be clinically important.
The analysis population required a baseline and month 6 DXA measurement. Therefore, the posted primary estimate pertains to participants meeting that measurement requirement, rather than necessarily to every randomized participant. The record does not state how missing measurements were handled beyond defining this population. Nor does it report measurement reliability, procedures for reading DXA scans, or whether scans were centrally assessed. Those details matter when interpreting changes in a continuous imaging measure, but they are not reported in the registry record.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry reports least-squares mean percent change in lumbar spine BMD of 9.6 for romosozumab 70 mg/mL, with a 95% confidence interval from 8.8 to 10.4. For romosozumab 90 mg/mL, the corresponding estimate is 9.2, with a 95% confidence interval from 8.4 to 10.0. The placebo estimate is 0.8, with a 95% confidence interval from -0.4 to 2.1. Each estimate is based on the primary analysis population defined by having baseline and month 6 lumbar spine DXA measurements.
The registry also reports a least-squares mean difference of -0.4 for the comparison between the two active formulations, with a 95% confidence interval from -1.5 to 0.7. The record labels this as a noninferiority analysis. It does not specify the sign convention for the difference, so the direction represented by the negative sign should not be inferred from the record alone. The interval describes the range of values compatible with the analysis under its statistical assumptions; it is not a range containing a fixed proportion of individual participants’ changes.
In general, a confidence interval conveys uncertainty around an estimated quantity. A relatively narrow interval indicates greater precision for the estimate than a wider interval would, while the interval’s position indicates which values are compatible with the data and analysis. Here the reported interval for the active-formulation difference includes values on either side of zero. That fact alone does not establish noninferiority or equivalence. A noninferiority conclusion depends on a prespecified margin and the analysis rule used to compare the interval with that margin. Neither the margin nor a textual conclusion is reported in the registry record.
The placebo result gives context for the observed endpoint values, but it is not the stated noninferiority comparison. The posted analysis compares the active formulations. These estimates concern percent change in lumbar spine BMD over the specified period; they do not directly describe fractures, symptoms, long-term outcomes, or effects beyond the study period. No p-value is reported for the noninferiority comparison, and the record does not provide enough information to reconstruct a complete inferential decision rule.
The estimates are precise only to the extent represented by their reported confidence intervals and the analysis population. The record’s missing details about missing data, the noninferiority margin, and analysis procedures limit how fully the result can be independently assessed. The numerical estimates should therefore be read as posted, with the record’s stated population and endpoint, rather than as a complete account of every protocol decision.
9. Statistical Methods Explained
The primary outcome is continuous because it is measured on a numerical scale and reported as percent change. The registry gives least-squares means for the treatment groups and a least-squares mean difference for the active-formulation comparison. These are the reported statistical summaries; the record does not describe the fitted model or the covariates used to obtain them.
What does a least-squares mean represent?
In general, a least-squares mean is a model-based estimate of a group mean. Unlike a simple arithmetic average, it can reflect adjustment for variables included in a statistical model. The term alone does not reveal which variables were included, how the model was specified, or whether any adjustment was made in this study. Those details are not reported in the registry record. The least-squares mean difference is a contrast between estimated group means, but the record does not specify the subtraction order or sign convention.
How should the confidence intervals be read?
A confidence interval expresses uncertainty about an estimated parameter under the analysis assumptions. The reported 95% intervals accompany the group means and the active-group difference. In general, a confidence interval is not a range of individual participant outcomes, and it does not account for every possible source of uncertainty, such as unreported design or measurement details. The intervals should be interpreted alongside the endpoint definition and the population used to calculate the estimates.
For the active-formulation difference, the registry labels the analysis as noninferiority. In general, a noninferiority analysis asks whether the result rules out a loss greater than a prespecified acceptable margin. That margin is essential to interpretation: the estimated difference and its confidence interval must be judged against it under the analysis plan. The margin, the confidence-bound rule, and the protocol’s formal decision are not reported in the registry record. A confidence interval that includes zero cannot on its own determine the outcome of a noninferiority test.
Why does the analysis population matter?
The primary outcome population includes randomized participants with both baseline and month 6 lumbar spine DXA measurements. This definition makes clear that the posted estimate depends on available measurements at both time points. In general, restricting an analysis to participants with observed measurements can affect interpretation if availability is related to outcomes or treatment. The record does not report the number or reasons for missing measurements by randomized group, or whether the analysis used imputation, a sensitivity analysis, or another approach to missing data.
The reported outcome group sizes are 46 for placebo, 110 for romosozumab 70 mg/mL, and 117 for romosozumab 90 mg/mL. These are the numbers attached to the endpoint estimates, not the full enrollment figure. In general, randomization supports comparisons by balancing measured and unmeasured characteristics on average, but the details of the randomization procedure and the analysis population can affect how that principle applies to a particular estimate. The registry does not provide enough information to assess those details.
What can planning calculations tell a reader?
In general, a sample-size planning illustration describes how many participants per arm might be needed under stated assumptions to detect a chosen effect with a chosen power. The assumptions recorded for the planning illustration are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the target effect is expressed as a stated fraction of the observed effect. Such illustrations are not the trial’s original design justification unless the record establishes that they were used in the design. They are not observed power and do not describe the probability that the trial’s result is correct. The trial’s own sample-size justification and design assumptions are not reported in the registry record.
In general, the normal approximation is a method for approximating the sampling distribution of an estimator. Whether it is suitable for a particular analysis depends on the data and model. The registry does not describe whether this approximation was used in the sponsor’s primary analysis; the stated normal approximation pertains to the planning assumptions only. The record also does not report any interim analyses, stopping rules, or detailed statistical analysis plan.
Why can over-dispersion matter?
In general, over-dispersion means that observed variability is greater than a model expects. It is particularly relevant to some count-data models, where underestimating variability can make uncertainty look too small. The primary endpoint in this study is a continuous BMD measure, and the registry does not report an over-dispersion assessment or a model for count outcomes. The concept therefore does not supply a reason to reinterpret the posted continuous-outcome estimates; it illustrates why model assumptions must match the type and distribution of the outcome.
The registry records one analysis and lists five outcome measures, including the primary endpoint and four secondary outcomes. This accounting does not establish that only one statistical test was conducted. The analysis plan, multiplicity procedure, and results for the secondary outcomes are not reported in the registry record. A reader should distinguish the posted summary from a complete protocol-level account of modeling, missing-data handling, and inferential decisions.
10. Analyses Reported by the Sponsor
| Method | Estimate | 95% CI | P-value |
|---|---|---|---|
| not stated | Least Squares Mean Difference -0.4 | -1.5 to 0.7 | not reported |
The registry’s reported active-formulation comparison is a least-squares mean difference of -0.4, with a 95% confidence interval from -1.5 to 0.7, and is labeled as noninferiority. This is the comparison to use when describing the sponsor’s posted analysis. The registry does not report a separate unadjusted recalculation for this continuous two-arm comparison, so there is no recalculated estimate to set against the sponsor’s result.
The difference and interval do not, on their own, show whether the study met its noninferiority criterion. That interpretation requires the prespecified margin and the analysis rule. Neither is reported in the registry record. The record also does not state the direction used to calculate the difference, so the negative estimate should be quoted without assigning it to a particular formulation’s advantage.
No discrepancy between a sponsor analysis and a recalculation is identified in the registry material. The sponsor’s posted result therefore remains the available direct comparison, while important elements needed to evaluate its noninferiority conclusion remain unreported. In particular, the record does not include the model specification, adjustment variables, or a narrative statement of the decision against a margin.
11. Multiplicity, Missing Data and Analysis Populations
The registry lists five outcome measures and one analysis. The primary outcome is percent change in lumbar spine BMD; the four listed secondary outcomes are percent change in total hip BMD, femoral neck BMD, P1NP, and serum CTX. The record does not report results for those secondary outcomes in the posted outcome information, nor does it state whether they were confirmatory, exploratory, or part of a testing hierarchy.
When several outcomes or comparisons are tested, multiplicity can increase the chance of at least one apparently notable result arising by chance. In general, a statistical plan may address this through an adjustment, a hierarchy, or another prespecified strategy. The multiplicity procedure, family of tests, and allocation of significance levels are not reported in the registry record. The count of listed outcomes does not establish how many tests were actually conducted, and the single analysis entry does not establish that no other analyses occurred.
The primary analysis population required both baseline and month 6 lumbar spine DXA BMD measurements. The record does not report how many randomized participants lacked either measurement, reasons for missingness, or missingness by group. It does not describe imputation, sensitivity analyses, or a missing-data assumption. These omissions make it difficult to assess whether the observed-measurement population represents all randomized participants for the endpoint.
The registry does not report whether the primary analysis followed an intention-to-treat principle, a per-protocol principle, or another analysis-population rule beyond its measurement requirement. It also does not describe exclusions after randomization or protocol deviations. The exact consequences cannot be inferred from the endpoint counts alone. Thus, the posted estimate is best interpreted for the explicitly described population, while the relationship of that population to all randomized participants remains incompletely documented.
12. 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 |
|---|---|---|---|
| Placebo 70 mg | 4 | 26 | 15.4% |
| Placebo 90 mg/mL | 2 | 26 | 7.7% |
| Romosozumab 70 mg/mL | 7 | 119 | 5.9% |
| Romosozumab 90 mg/mL | 3 | 123 | 2.4% |
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 |
|---|---|
| Placebo 70 mg | 9 of 26 |
| Placebo 90 mg/mL | 5 of 26 |
| Romosozumab 70 mg/mL | 24 of 119 |
| Romosozumab 90 mg/mL | 38 of 123 |
Most frequent serious adverse event terms
| Event term | Organ system | Placebo 70 mg | Placebo 90 mg/mL | Romosozumab 70 mg/mL | Romosozumab 90 mg/mL |
|---|---|---|---|---|---|
| Coagulopathy | Blood and lymphatic system disorders | 0 of 26 | 0 of 26 | 1 of 119 | 0 of 123 |
| Atrial fibrillation | Cardiac disorders | 0 of 26 | 0 of 26 | 1 of 119 | 0 of 123 |
| Cardiac failure | Cardiac disorders | 0 of 26 | 0 of 26 | 1 of 119 | 0 of 123 |
| Hyperthyroidism | Endocrine disorders | 0 of 26 | 0 of 26 | 0 of 119 | 1 of 123 |
| Abdominal pain | Gastrointestinal disorders | 0 of 26 | 0 of 26 | 1 of 119 | 0 of 123 |
| Appendicitis noninfective | Gastrointestinal disorders | 0 of 26 | 0 of 26 | 1 of 119 | 0 of 123 |
| Gastritis | Gastrointestinal disorders | 0 of 26 | 0 of 26 | 1 of 119 | 0 of 123 |
| Mesenteric artery embolism | Gastrointestinal disorders | 1 of 26 | 0 of 26 | 0 of 119 | 0 of 123 |
| Death | General disorders | 1 of 26 | 0 of 26 | 0 of 119 | 0 of 123 |
| Appendicitis perforated | Infections and infestations | 0 of 26 | 0 of 26 | 1 of 119 | 0 of 123 |
Most frequent non-serious adverse event terms
| Event term | Organ system | Placebo 70 mg | Placebo 90 mg/mL | Romosozumab 70 mg/mL | Romosozumab 90 mg/mL |
|---|---|---|---|---|---|
| Nasopharyngitis | Infections and infestations | 3 of 26 | 0 of 26 | 9 of 119 | 19 of 123 |
| Asthenia | General disorders | 0 of 26 | 2 of 26 | 4 of 119 | 4 of 123 |
| Injection site pain | General disorders | 0 of 26 | 0 of 26 | 7 of 119 | 3 of 123 |
| Injection site erythema | General disorders | 0 of 26 | 0 of 26 | 2 of 119 | 7 of 123 |
| Urinary tract infection | Infections and infestations | 0 of 26 | 1 of 26 | 1 of 119 | 7 of 123 |
| Headache | Nervous system disorders | 3 of 26 | 0 of 26 | 2 of 119 | 2 of 123 |
| Upper respiratory tract infection | Infections and infestations | 1 of 26 | 2 of 26 | 1 of 119 | 2 of 123 |
| Fall | Injury, poisoning and procedural complications | 2 of 26 | 0 of 26 | 2 of 119 | 2 of 123 |
| Bone pain | Musculoskeletal and connective tissue disorders | 0 of 26 | 2 of 26 | 2 of 119 | 2 of 123 |
Serious events with the highest arm-level rate: mesenteric artery embolism (3.8%), death (3.8%), humerus fracture (3.8%), patella fracture (3.8%), spinal deformity (3.8%).
The safety summaries report serious adverse events and other adverse events separately, with an at-risk count and an affected count for each group. For serious adverse events, the recorded at-risk and affected counts are 26 and 4 for placebo 70 mg, 26 and 2 for placebo 90 mg/mL, 119 and 7 for romosozumab 70 mg/mL, and 123 and 3 for romosozumab 90 mg/mL. For other adverse events, the corresponding counts are 26 and 9, 26 and 5, 119 and 24, and 123 and 38.
These are descriptive counts within the listed groups. They do not by themselves establish that a difference between groups is statistically reliable, and the record does not give a comparative safety analysis or confidence intervals for these counts. The safety group denominators also differ from the primary endpoint group sizes, so the outcome-analysis counts should not be substituted for the safety at-risk counts.
The registry lists Mesenteric artery embolism, Death, Humerus fracture, Patella fracture, Spinal deformity, and Spinal osteoarthritis among the top serious-event terms, each with a reported 3.8%. The context and denominator for those term percentages are not reported in the registry record. Death is included as a term in that list, while deaths by group are not reported. These summaries do not establish causation or provide enough detail to compare the severity, timing, or clinical context of events across groups.
13. What the Result Does, and Does Not, Mean
The posted data describe month 6 lumbar spine BMD changes in the defined analysis population and a reported noninferiority comparison between the active formulations. The group estimates and confidence intervals quantify the observed endpoint and its statistical uncertainty under the sponsor’s analysis. They do not establish what the result would be in participants without the required DXA measurements or over a period beyond the endpoint timeframe.
The record labels the comparison as noninferiority, but does not give the margin or the formal decision. Consequently, the numerical difference and its interval cannot be used alone to declare that the noninferiority criterion was satisfied or not satisfied. The record also does not provide a p-value for the comparison or enough detail to reproduce the sponsor’s model. The distinction matters: a label describing an analysis is not the same as a fully documented inferential conclusion.
The endpoint is BMD, not a clinical event such as fracture. Secondary outcomes include other BMD sites and bone-turnover markers, but their numerical results are not reported in the outcome information described in the registry record. No inference about fractures, symptoms, long-term outcomes, or broader clinical benefit follows from the posted lumbar spine BMD estimates alone. The record also does not establish whether the observed changes are clinically meaningful.
The trial included matching placebo groups, but the primary comparison reported for noninferiority is between the two active formulations. The placebo estimate gives a separate reference point for the endpoint; it should not be confused with the active-formulation difference. Safety counts provide descriptive information about events during the study, but do not prove that an event was caused by an intervention or that one group is safer.
In general, statistical estimates describe data under a particular design, outcome definition, and analysis. They do not replace missing protocol details. Here, interpretation is constrained by the unreported noninferiority margin, statistical model, missing-data procedures, multiplicity approach, and secondary-outcome results. Those limitations do not erase the posted estimates, but they define what can responsibly be concluded from them.
14. Why This Trial Matters Statistically
This study is statistically notable because its stated central comparison concerns two concentrations of the same monthly romosozumab amount, with different injection schedules. That makes the estimand and sign convention important: readers need to know exactly which group is subtracted from which, and what loss the noninferiority design was intended to rule out. The registry reports the difference but not the subtraction order or noninferiority margin.
The primary outcome also illustrates the distinction between a group estimate and a treatment comparison. The least-squares means describe the endpoint in each active group, while the noninferiority analysis is represented by a difference and its confidence interval. The two forms of reporting answer related but not identical questions. A formal noninferiority interpretation depends on the prespecified margin, not merely on whether group estimates look similar or whether an interval includes zero.
The analysis population is another important statistical feature. Requiring both baseline and month 6 measurements defines who contributes to the primary outcome estimates. Without group-specific missingness information or sensitivity analyses, readers cannot fully assess the potential effect of participants lacking one of those measurements.
Finally, the registry’s outcome summaries are more complete for the primary continuous endpoint than for multiplicity and secondary outcomes. It lists additional measures but does not report their results or how multiple testing was handled. The study therefore offers a useful example of why estimates, intervals, population definitions, and protocol-level decisions all matter when interpreting a noninferiority analysis. A single difference estimate cannot convey all of those elements.
15. 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 provides the primary endpoint definition, group estimates, confidence intervals, and a labeled noninferiority comparison, but leaves several details needed for independent assessment unreported. The noninferiority margin is not given, and the record does not state a formal conclusion against that margin. The sign convention for the reported difference is also not specified. Without these details, the comparison cannot be fully interpreted as a noninferiority decision.
The statistical model is not described. The record does not identify adjustment variables, diagnostic checks, or the precise method used to obtain the least-squares means and difference. It also does not state whether the analysis was prespecified in a protocol or statistical analysis plan. The planning assumptions recorded for sample-size illustrations do not establish the trial’s original sample-size justification.
Missing-data information is limited to the definition of the endpoint population. Counts and reasons for missing baseline or month 6 DXA measurements, group-specific missingness, and methods for handling missing data are not reported in the registry record. The relationship between the primary outcome population and other analysis populations is therefore not fully described.
The record lists secondary outcomes but does not report their numerical results or a multiplicity strategy. It does not report interim analyses or stopping rules. Safety summaries are counts, without a reported comparative analysis, and the context for the listed serious-event term percentages is not specified. Deaths by group are not reported. These limitations restrict detailed assessment of statistical robustness and safety patterns; they should not be replaced with assumptions about the protocol or conduct.
16. Related Tutorials
17. Related Statistical Calculators
18. Sources
- ClinicalTrials.gov: A Randomized Phase 3 Study to Evaluate Two Formulations of Romosozumab in Postmenopausal Women With Osteoporosis — NCT02016716.
- Registry record last updated 2018-11-08; status last verified 2018-09.
- Results first posted 2018-11-08.
- 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.
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.
19. Record Summary
NCT02016716 is a completed randomized, double-masked, parallel-group phase 3 study sponsored by Amgen. It enrolled 294 postmenopausal women with osteoporosis and compared romosozumab 90 mg/mL with romosozumab 70 mg/mL, alongside matching placebo groups. Both active regimens are described as 210 mg monthly for 6 months, with different injection numbers and volumes.
The primary outcome was percent change from baseline in lumbar spine BMD at month 6, measured by DXA. The registry reports least-squares means of 9.6 for the 70 mg/mL group and 9.2 for the 90 mg/mL group, and a noninferiority comparison with a reported difference of -0.4 and 95% confidence interval from -1.5 to 0.7. The noninferiority margin and formal decision are not reported in the registry record.
Secondary measures included total hip and femoral neck BMD, P1NP, and CTX. Key details about the statistical model, multiplicity, missing data, and secondary-outcome results are not reported. Safety counts are descriptive and do not establish causation or comparative risk.