This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. View the ClinicalTrials.gov record for NCT05646706.
1. Trial at a Glance
STEP UP was a randomized, parallel-group, quadruple-masked phase 3 trial in obesity. The registry reports 1,407 enrolled participants and three study arms, with semaglutide 7.2 mg versus placebo forming the comparison for both registered primary endpoints.
| Feature | STEP UP |
|---|---|
| Trial name | STEP UP |
| ClinicalTrials.gov identifier | NCT05646706 |
| Phase | Phase 3 |
| Condition | Obesity |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 1,407 |
| Interventions | Semaglutide; semaglutide; placebo |
| Primary endpoints | Two |
| Results posted | Yes |
| Statistical analyses posted | Two primary-endpoint analyses |
| Lead sponsor | Novo Nordisk A/S |
| Sponsor type | Industry |
2. Clinical Question
The primary statistical question was whether semaglutide 7.2 mg differed from placebo with respect to body-weight change and the probability of achieving at least 5% body-weight reduction by week 72.
Population
Participants enrolled in the phase 3 trial for the condition of obesity.
Intervention of primary comparison
Semaglutide 7.2 mg.
Comparator
Placebo.
Primary question
How does semaglutide 7.2 mg compare with placebo for relative body-weight change and achievement of at least 5% weight reduction?
The trial also included a semaglutide 2.4 mg arm. The two posted primary analyses in the registry, however, are explicitly identified as comparisons of semaglutide 7.2 mg versus placebo. No separate statistical estimate for semaglutide 2.4 mg is reported in the posted primary analyses summarized here.
3. Trial Design
Participants were assigned to study treatment using a randomized design.
The registry identifies the study as a parallel-group trial.
The registry identifies the trial as quadruple-masked.
The registry classifies the primary purpose as treatment.
Semaglutide 7.2 mg
- Semaglutide drug intervention
- Primary efficacy comparison versus placebo
- Relative body-weight change assessed from week 0 to week 72
- Achievement of ≥5% body-weight reduction assessed at week 72
Semaglutide 2.4 mg
- Semaglutide drug intervention
- Part of the three-arm randomized design
- Included in the enrolled study population
- No separate posted primary-endpoint estimate is reported in the summarized registry analyses
Placebo
- Placebo drug intervention
- Comparator for the posted primary analyses
- Compared with semaglutide 7.2 mg for both primary endpoints
4. Trial Timing and Status
Study start
The registry lists January 4, 2023 as the study start date.
Primary completion
The registry lists October 30, 2024 as the primary completion date.
Completed
The ClinicalTrials.gov record identifies STEP UP as completed and reports posted results.
5. Analysis Populations
Both posted primary analyses use a full analysis set that included all randomized participants. Participants were analyzed according to their randomized treatment. The number analyzed for a particular outcome was defined as participants with available data for that outcome measure.
| Analysis population | Registry description |
|---|---|
| Full analysis set | Included all randomized participants. |
| Treatment assignment | Participants were analyzed according to the randomized treatment. |
| Outcome-specific analysis set | Overall number analyzed was participants with available data for the outcome measure. |
| Primary comparison | Semaglutide 7.2 mg versus placebo. |
This structure is closely aligned with an intention-to-treat principle: randomization remains the basis for treatment assignment in the efficacy comparison. At the same time, the registry's wording about participants with available data means that the analyzed number can depend on outcome availability.
6. Primary Endpoints
| Endpoint | Time frame | Definition | Posted analysis |
|---|---|---|---|
| Semaglutide 7.2 mg versus placebo: relative change in body weight | Baseline (week 0), end of treatment (week 72) | Relative change in body weight from baseline (week 0) to end of treatment (week 72) is presented. | ANCOVA |
| Semaglutide 7.2 mg versus placebo: participants achieving body-weight reduction ≥5% | At week 72 | Participants are categorized as yes or no according to whether they achieved body-weight reduction ≥5% from baseline (week 0). | Binary regression model |
The two endpoints represent different statistical questions. The first treats body-weight change as a quantitative outcome and summarizes the treatment difference in percentage change. The second converts weight change into a binary response definition and compares the odds of reaching a clinically defined threshold.
7. Primary Result: Relative Change in Body Weight
The first posted primary analysis compared semaglutide 7.2 mg with placebo for relative change in body weight from baseline (week 0) to end of treatment (week 72). The registry reports an ANCOVA with a treatment difference of −14.81.
Semaglutide 7.2 mg versus placebo
Treatment difference in percentage change in body weight
95% CI: −16.21 to −13.41 · P < 0.0001
The treatment difference of −14.81 percentage points means that the modeled relative change in body weight was estimated to be 14.81 percentage points lower with semaglutide 7.2 mg than with placebo over the specified baseline-to-week-72 comparison.
The estimate is a between-group treatment difference, not an individual patient's predicted weight change. It does not mean that every participant lost 14.81% of body weight, nor does it describe the distribution of individual responses.
The 95% confidence interval, −16.21 to −13.41, describes the statistical uncertainty around the estimated treatment difference under the analysis framework. Its relatively narrow width indicates greater precision than would be conveyed by the point estimate alone.
The reported P < 0.0001 addresses evidence against the null hypothesis represented by the statistical test. It does not measure the size of the treatment effect, clinical importance, or the probability that the null hypothesis is true.
Because the analysis is ANCOVA-based, interpretation depends on the model specification and the handling of baseline information. The registry does not report additional model details such as covariate specification, adjustment variables, or missing-data imputation in the statistical analysis summarized here.
8. Primary Result: Achievement of at Least 5% Weight Reduction
The second posted primary analysis evaluated whether participants achieved body-weight reduction greater than or equal to 5% from baseline by week 72. The registry describes the endpoint as a binary yes/no outcome and reports a binary regression model with an odds ratio as the effect measure.
Odds of achieving ≥5% body-weight reduction
Semaglutide 7.2 mg versus placebo
95% CI: 8.32 to 17.61 · P < 0.0001
An odds ratio of 12.10 means that the estimated odds of achieving at least 5% body-weight reduction were 12.10 times the corresponding odds in the placebo group under the reported binary regression analysis.
An odds ratio is not the same as a risk ratio or probability ratio. An OR of 12.10 therefore does not mean that 12.10 times as many participants necessarily achieved the endpoint, nor does it directly state the absolute percentage-point difference in response rates.
The 95% CI of 8.32 to 17.61 describes uncertainty around the estimated odds ratio. Because the entire interval is above 1, the reported estimate is separated from the null value of 1 under this analysis.
The P < 0.0001 quantifies the evidence against the null hypothesis in the reported test. It does not quantify the magnitude of the effect; the OR and its confidence interval provide the effect-size information.
The analysis is based on a binary threshold, so it discards information about how far each participant's weight change was above or below the 5% threshold. The relative-change endpoint and the ≥5% endpoint therefore provide complementary rather than interchangeable summaries.
9. Statistical Methodology
ANCOVA for relative body-weight change
The registry reports ANCOVA for the primary analysis of relative change in body weight. ANCOVA, or analysis of covariance, is a linear-model framework that compares treatment groups while allowing specified covariates to be incorporated into the model.
For this endpoint, the outcome was percentage change in body weight from baseline (week 0) to end of treatment (week 72). The registry does not specify the complete covariate set for the posted analysis.
The key quantity reported by the registry is the treatment difference, rather than an odds ratio or hazard ratio. This is appropriate to the scale of a quantitative outcome: the result describes how far apart the modeled treatment groups were on the outcome scale.
Binary regression for the ≥5% endpoint
The second primary endpoint converted weight change into a binary outcome. Participants were categorized as yes if they achieved body-weight reduction ≥5% from baseline and no otherwise. The registry reports a binary regression model and an odds ratio.
An odds ratio above 1 indicates higher estimated odds of the binary outcome in the first group relative to the comparator; an odds ratio of 1 is the conventional null value.
Intention-to-treat principle
The full analysis set included all randomized participants, and participants were analyzed according to randomized treatment. This preserves the treatment comparison established at randomization rather than redefining groups according to treatment received after randomization.
Superiority testing
The registry identifies both primary analyses as superiority analyses. The statistical question is therefore whether the semaglutide 7.2 mg group differs favorably from placebo under the prespecified endpoint and model, rather than whether two treatments are sufficiently similar to establish non-inferiority.
10. Why Was ANCOVA Used?
ANCOVA is particularly useful when the outcome is continuous and the analysis is intended to estimate a treatment difference on the outcome scale. Rather than reducing body-weight change to a yes/no threshold, the first primary endpoint preserves quantitative information about the magnitude of change.
Continuous endpoint
Relative change in body weight is expressed as a percentage rather than a binary response category.
Treatment difference
The reported effect measure is the difference between treatment groups on the modeled percentage-change scale.
Model-based estimate
The ANCOVA estimate depends on the specified statistical model and covariates.
Complementary endpoint
The continuous analysis retains information that the separate ≥5% binary endpoint does not.
The registry does not provide the full ANCOVA specification in the posted analysis summary. Consequently, the treatment difference should be interpreted at the level actually reported: a model-based between-group difference in relative body-weight change.
11. What Does an Odds Ratio of 12.10 Mean?
The odds ratio compares odds, not probabilities directly. If an event has probability p, its odds are defined as:
The odds ratio then compares those odds between two treatment groups.
For STEP UP, the reported OR of 12.10 indicates that the estimated odds of achieving at least 5% body-weight reduction were substantially higher in the semaglutide 7.2 mg group than in the placebo group.
Importantly, an odds ratio cannot be translated into an absolute response probability without knowing the underlying comparator probability. The registry summary provides the OR, its confidence interval, and its P-value, but does not provide the response counts or percentages needed for that additional conversion.
12. What Does the −14.81 Treatment Difference Mean?
The first primary analysis reports a treatment difference of −14.81 for percentage change in body weight. The negative sign reflects the direction of the comparison as reported: semaglutide 7.2 mg minus placebo on the modeled percentage-change outcome.
The estimate indicates a 14.81-percentage-point lower relative change in body weight for semaglutide 7.2 mg compared with placebo under the reported ANCOVA model.
The estimate should not be confused with a 14.81-unit change in body weight, a 14.81-pound difference, or a statement about the response of every individual participant.
The confidence interval from −16.21 to −13.41 gives a range of values compatible with the statistical uncertainty around the estimated treatment difference. It is more informative than the point estimate alone because it conveys precision.
13. Confidence Intervals and P-values
Both primary analyses report two-sided 95% confidence intervals and P-values of <0.0001. These quantities answer related but different statistical questions.
Confidence interval
Describes statistical uncertainty around the estimated treatment effect under the specified analysis framework.
P-value
Quantifies the compatibility of the observed data with the null hypothesis used for the statistical test.
Effect size
The treatment difference and odds ratio describe the magnitude and direction of the estimated treatment effects.
Precision
The width of a confidence interval provides information about how precisely an effect has been estimated.
A very small P-value does not imply that an effect is large, clinically important, or precisely estimated. Conversely, a confidence interval should not be interpreted as the range in which individual patient outcomes will fall. The effect measure and its confidence interval are the principal tools for describing the magnitude and uncertainty of the treatment comparison.
14. Why the Two Primary Endpoints Are Complementary
The two primary endpoints deliberately summarize treatment effects on different scales.
| Feature | Relative body-weight change | Achievement of ≥5% reduction |
|---|---|---|
| Outcome type | Quantitative | Binary |
| Time frame | Baseline (week 0) to end of treatment (week 72) | At week 72 |
| Posted method | ANCOVA | Binary regression model |
| Effect measure | Treatment difference | Odds ratio |
| Reported estimate | −14.81 | 12.10 |
| 95% CI | −16.21 to −13.41 | 8.32 to 17.61 |
| P-value | <0.0001 | <0.0001 |
The continuous endpoint asks, in effect, how the groups differ in the magnitude of relative weight change. The binary endpoint asks whether participants crossed a prespecified threshold. A participant who loses substantially more than 5% and one who loses just over 5% receive the same "yes" classification for the binary endpoint, whereas the continuous endpoint retains that distinction.
15. Missing Data and Outcome Availability
The registry defines the number analyzed as participants with available data for the outcome measure. This wording matters because the full analysis set begins with all randomized participants, while the outcome-specific analysis is based on participants for whom the relevant outcome is available.
For the body-weight endpoint, the distinction is especially relevant because the analysis concerns change from baseline to week 72. For the binary endpoint, outcome availability determines whether a participant can be classified as achieving or not achieving the ≥5% threshold.
16. Randomization and Blinding
Randomization is a central design feature because it creates the basis for comparing outcomes across treatment groups without assigning treatment according to observed participant characteristics. STEP UP is identified by the registry as randomized, parallel, and quadruple-masked.
The registry does not report a crossover design, and the study is classified as parallel. The registry summary also does not report specific stratification factors for the posted primary analyses.
17. Multiplicity and Interim Analysis
STEP UP has two registered primary endpoints, and the registry reports two formal primary-endpoint analyses. The existence of multiple primary endpoints makes multiplicity an important statistical-design topic because the overall interpretation of repeated hypothesis tests depends on how the trial's type I error was controlled.
| Design topic | What is reported |
|---|---|
| Primary endpoints | Two |
| Primary analyses posted | Two |
| Hypothesis type | Superiority |
| Interim analysis | The registry summary does not report an interim-analysis strategy. |
| Multiplicity adjustment | The registry summary does not report a multiplicity-adjustment procedure for the two primary analyses. |
| Alpha-spending method | The registry summary does not report an alpha-spending method. |
These omissions do not change the reported estimates, confidence intervals, or P-values. They do limit the amount of statistical-design detail that can be inferred from the posted analysis summaries alone.
18. Stratification and Covariate Adjustment
The registry identifies ANCOVA as the method for the continuous primary endpoint, but the statistical analysis summary does not provide the complete model specification. In particular, it does not list the covariates used in the ANCOVA analysis in the information summarized here.
For the binary endpoint, the registry identifies a binary regression model but does not provide a full model specification in the summarized analysis record.
19. Serious Adverse Events
The registry reports serious adverse events by treatment arm as affected participants divided by participants at risk.
| Treatment arm | Participants with serious adverse events | Participants at risk |
|---|---|---|
| Semaglutide 7.2 mg | 73 | 1,004 |
| Semaglutide 2.4 mg | 22 | 201 |
| Placebo | 11 | 201 |
These counts provide a descriptive safety summary. The ClinicalTrials.gov record does not report a formal comparative statistical analysis of serious adverse events, so the counts should not be converted into an unreported treatment-effect estimate.
Semaglutide 7.2 mg
73 participants with serious adverse events among 1,004 participants at risk.
Semaglutide 2.4 mg
22 participants with serious adverse events among 201 participants at risk.
Placebo
11 participants with serious adverse events among 201 participants at risk.
Statistical role
These are descriptive safety counts rather than a posted inferential comparison.
20. Statistical Methods Explained
Why was ANCOVA used?
ANCOVA is a linear-model approach for a quantitative outcome. In STEP UP, the posted method was used for relative change in body weight, and the reported effect measure was a treatment difference. The method therefore preserves the continuous nature of the endpoint rather than converting all participants into response categories.
What does the treatment difference of −14.81 mean?
It is the reported difference between the semaglutide 7.2 mg and placebo groups on the percentage-change body-weight outcome. The negative value indicates that the modeled percentage change was lower in the semaglutide 7.2 mg group by 14.81 percentage points relative to placebo.
What does an odds ratio of 12.10 mean?
It means that the estimated odds of achieving at least 5% body-weight reduction at week 72 were 12.10 times as high in the semaglutide 7.2 mg group as in the placebo group under the reported binary regression analysis. It is not a direct ratio of probabilities.
Why are the confidence intervals important?
The confidence intervals describe uncertainty around the treatment-effect estimates. For the continuous endpoint, the 95% CI is −16.21 to −13.41. For the binary endpoint, it is 8.32 to 17.61. Reporting the interval alongside the point estimate makes the precision of the estimates visible.
Why does the P-value not measure effect size?
A P-value describes the statistical evidence against a null hypothesis under the specified test. It does not tell the reader how large the treatment effect is. The treatment difference and odds ratio are the effect measures, while their confidence intervals describe uncertainty around those estimates.
Why analyze participants according to randomized treatment?
Analyzing randomized participants according to their assigned treatment preserves the comparison generated by randomization. This is the core idea behind an intention-to-treat analysis and helps prevent post-randomization treatment decisions from redefining the primary comparison.
21. Interpretation of the Two Primary Results Together
The ANCOVA analysis estimated a −14.81 treatment difference in relative body-weight change, with a 95% CI of −16.21 to −13.41 and P < 0.0001.
The binary regression analysis estimated an OR of 12.10 for achieving at least 5% body-weight reduction, with a 95% CI of 8.32 to 17.61 and P < 0.0001.
The two estimates are not competing versions of the same statistic. They answer different questions. One estimates a difference on a continuous percentage-change scale; the other compares odds of crossing a predefined threshold.
The consistency of the direction across these distinct endpoint formulations provides two statistical descriptions of the same broad treatment comparison. The first emphasizes magnitude of relative weight change, while the second emphasizes the probability structure of reaching a threshold after conversion to a binary outcome.
22. What the Results Do — and Do Not — Establish
What the estimates establish statistically
The posted analyses estimate a negative treatment difference for relative body-weight change and an odds ratio above 1 for achieving ≥5% body-weight reduction.
What the confidence intervals add
They quantify uncertainty around the respective treatment-effect estimates rather than uncertainty about individual participant outcomes.
What the P-values add
They quantify evidence against the corresponding null hypotheses under the reported analyses; they do not quantify effect magnitude.
What remains separate
Statistical efficacy results and serious adverse-event counts address different dimensions of the trial and should not be collapsed into a single numerical conclusion.
The posted results do not, by themselves, provide an individual-level prediction of weight loss or guarantee that a particular participant will experience the estimated treatment difference. Trial-level estimates summarize average or modeled group-level behavior under the specified analysis.
23. Limitations and Interpretation Issues
- Outcome availability: the full analysis set includes all randomized participants, while the number analyzed for an outcome is defined as participants with available data for that outcome.
- Model specification: the registry identifies ANCOVA and binary regression but does not provide the complete covariate specifications in the summarized statistical analyses.
- Missing-data methods: the posted analysis summary does not specify a missing-data imputation method or detailed sensitivity-analysis strategy.
- Multiplicity: two primary endpoints are reported, but the summarized registry analysis does not provide a multiplicity-adjustment procedure.
- Interim analyses: the summarized registry analysis does not report an interim-analysis or alpha-spending strategy.
- Binary thresholding: the ≥5% endpoint reduces a continuous weight-change outcome to a yes/no classification and therefore loses information about the magnitude of change beyond the threshold.
- Odds ratio interpretation: the OR of 12.10 should not be described as a 12.10-fold increase in probability without the underlying response probabilities.
- Safety interpretation: serious adverse-event counts are descriptive in the posted information and do not constitute a formal comparative safety analysis.
- Three-arm design: the trial included semaglutide 7.2 mg, semaglutide 2.4 mg, and placebo, while the two posted primary analyses summarized here concern semaglutide 7.2 mg versus placebo.
24. Why This Trial Matters Statistically
STEP UP provides a useful teaching example because the same randomized clinical trial expresses its primary efficacy question through two fundamentally different statistical estimands: a continuous treatment difference and a binary odds ratio.
| Concept | How it appears in STEP UP |
|---|---|
| Randomization | Participants were randomized in a three-arm phase 3 parallel design. |
| Blinding | The registry identifies the trial as quadruple-masked. |
| Intention-to-treat principle | The full analysis set included all randomized participants, analyzed according to randomized treatment. |
| ANCOVA | Used for relative change in body weight from baseline to week 72. |
| Treatment difference | The effect measure for the continuous primary endpoint was −14.81. |
| Binary regression | Used for achievement of body-weight reduction ≥5% at week 72. |
| Odds ratio | The binary endpoint produced an OR of 12.10. |
| Confidence intervals | 95% two-sided intervals were reported for both primary estimates. |
| P-values | Both primary analyses reported P < 0.0001. |
| Superiority | Both primary analyses were classified as superiority hypotheses. |
| Three-arm design | Two semaglutide interventions and placebo were included in the study. |
| Safety analysis | Serious adverse events were reported descriptively by treatment arm. |
25. Related Tutorials
Learn more about the methods used in this trial:
26. Related Statistical Calculators
27. Sources
- ClinicalTrials.gov: STEP UP, NCT05646706.
- PubMed: Publication indexed under PMID 42764399.
- PubMed: Publication indexed under PMID 40961952.
Continue through Clinical Biostats
Explore statistical tutorials, calculators, and additional clinical-trial analyses to connect trial results with the methods used to generate them.
28. Record Summary
STEP UP illustrates how a randomized phase 3 trial can evaluate the same clinical domain through complementary statistical endpoints. The continuous primary endpoint used ANCOVA to estimate a treatment difference of −14.81 in relative change in body weight, with a 95% confidence interval of −16.21 to −13.41 and P < 0.0001. The binary primary endpoint used a binary regression model to estimate an odds ratio of 12.10 for achieving at least 5% body-weight reduction, with a 95% confidence interval of 8.32 to 17.61 and P < 0.0001.
The statistical interpretation is strongest when these results are kept on their proper scales. The treatment difference describes a quantitative between-group contrast, while the odds ratio describes a relative contrast in the odds of crossing a predefined threshold. Confidence intervals convey uncertainty around both estimates, and P-values address the corresponding hypothesis tests rather than measuring effect size.