This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
SCOUT was a randomized, double-masked, parallel phase 3 cardiovascular outcomes trial in overweight and obese patients with obesity as the registered condition. The primary endpoint was a composite time-to-event outcome consisting of nonfatal myocardial infarction, nonfatal stroke, resuscitated cardiac arrest, or cardiovascular death.
| Feature | SCOUT |
|---|---|
| Phase | Phase 3 |
| Condition | Obesity |
| Therapeutic area | Cardiovascular; Endocrinology & Metabolism |
| Design | Randomized, parallel, double-masked |
| Primary purpose | Treatment |
| Enrollment | 10777 |
| Arms | 3 |
| Primary endpoint | Risk of experiencing a Primary Outcome Event (POE) |
| Primary endpoint type | Time-to-event |
| Primary statistical method | Log-rank test |
| Effect measure | Hazard ratio |
| Hypothesis type | Superiority |
| Lead sponsor | Abbott |
2. Clinical Question
The central statistical question was whether randomized sibutramine was associated with a different time-to-first-primary-outcome-event experience than randomized placebo among the randomized study population followed from randomization up to 6 years.
Population
Overweight and obese patients enrolled in the SCOUT phase 3 trial under the registered condition of obesity.
Intervention
Sibutramine hydrochloride. The registry also describes a Lead-in Period during which subjects received sibutramine.
Comparator
Randomized placebo in the primary randomized comparison.
Primary question
How does the time-to-first Primary Outcome Event compare between randomized sibutramine and randomized placebo?
3. Trial Design
Sibutramine group
- Randomized sibutramine hydrochloride
- ITT analysis according to randomized intervention
- 4904 subjects at risk for the reported serious-adverse-event summary
Placebo group
- Randomized placebo
- ITT analysis according to randomized intervention
- 4881 subjects at risk for the reported serious-adverse-event summary
4. Endpoints
| Endpoint | Definition | Time frame | Type |
|---|---|---|---|
| Primary Outcome Event (POE) | Risk of experiencing a Primary Outcome Event, defined as nonfatal myocardial infarction (MI), nonfatal stroke, resuscitated cardiac arrest, or cardiovascular (CV) death | From randomization up to 6 years | Time-to-event |
| All-cause mortality | Risk of death from any cause | From randomization up to 6 years | Time-to-event |
| POE or revascularization | Risk of experiencing a POE or a revascularization procedure | From randomization up to 6 years | Time-to-event |
| Nonfatal MI included in POE | Risk of experiencing a nonfatal MI included in the POE | From randomization up to 6 years | Time-to-event |
| Nonfatal stroke included in POE | Risk of experiencing a nonfatal stroke included in the POE | From randomization up to 6 years | Time-to-event |
| Resuscitated cardiac arrest included in POE | Risk of experiencing a resuscitated cardiac arrest included in the POE | From randomization up to 6 years | Time-to-event |
| Cardiovascular death included in POE | Risk of experiencing cardiovascular death included in the POE | From randomization up to 6 years | Time-to-event |
For the registered primary endpoint, the registry states that each subject's POE status and time to first occurrence were evaluated using time-to-event analysis. All POEs confirmed by an independent adjudication committee were included in the analysis.
5. Analysis Population and Prespecified Risk Groups
The primary analyses were based on the intent-to-treat (ITT) population. The registry defines this population as all randomized subjects dispensed randomized study drug, grouped according to the intervention to which they were randomized.
The primary analysis also categorized subjects into three prespecified cardiovascular-risk groups:
| Risk group | Primary comparison | Hazard ratio | 95% CI | P-value |
|---|---|---|---|---|
| Overall randomized population | Sibutramine vs placebo | 1.162 | 1.029–1.311 | 0.015 |
| DM only | Sibutramine vs placebo | 1.010 | 0.738–1.384 | 0.948 |
| CV only | Sibutramine vs placebo | 1.276 | 0.916–1.776 | 0.149 |
| CV + DM | Sibutramine vs placebo | 1.182 | 1.024–1.365 | 0.022 |
These are distinct prespecified risk-group analyses of the same primary endpoint. A subgroup-specific hazard ratio describes the randomized comparison within that subgroup; it does not by itself establish that the treatment effect differs between subgroups. A formal claim of treatment-effect heterogeneity requires an interaction or other direct comparison of the subgroup-specific effects.
6. Statistical Methodology
Kaplan-Meier estimation and time-to-event analysis
The primary endpoint is inherently a time-to-event outcome: the analysis considers not only whether a participant experienced a POE, but also the time until the first POE. This makes survival-analysis methods appropriate because participants can have different follow-up times and can contribute information until an event or censoring.
Here, di represents events at time ti, while ni represents the number at risk immediately before that time. The registry does not post Kaplan-Meier curve coordinates, so the important statistical point is the time-to-event framework rather than a reconstructed curve.
Log-rank test
The registry reports the Log Rank method for the primary and secondary time-to-event comparisons. The log-rank test compares the observed and expected numbers of events between treatment groups across follow-up, with the comparison incorporating the timing of events rather than simply counting how many participants eventually experienced an event.
Cox proportional-hazards model
The effect measure reported by the registry is a Cox Proportional Hazard, expressed as a hazard ratio. For the overall primary comparison, the Cox model included treatment, country, gender, and age as a continuous variable at Lead-in Period baseline.
A hazard ratio above 1 indicates a higher estimated instantaneous event rate in the numerator group under the fitted model. A hazard ratio of 1 would correspond to equal hazards. The hazard ratio is not an absolute risk difference and is not the probability that a participant will experience the endpoint.
Intention-to-treat analysis
Using ITT means that participants are retained in the group to which they were randomized for the primary treatment comparison. This preserves the treatment contrast generated by randomization and avoids redefining treatment groups according to later exposure or adherence.
Sample-size assumptions
The registry states that the sample-size calculation used a two-tailed alpha level of 0.05 and power of 90%. The assumed annual composite event rate in the placebo arm was 7.0%. A sample of 3983 subjects in each of the 2 groups, corrected for a 30% noncompliance rate, was expected to provide 90% power to detect a relative risk reduction of 15% with sibutramine relative to placebo after at least 3 years of follow-up.
The noncompliance adjustment was specified as 15% in Year 1 and 6.3% in each year of Years 2 to 4. These assumptions belong to the design-stage power calculation and should not be confused with the observed hazard ratio of the completed trial.
7. Primary Results: Overall Randomized Population
The primary statistical comparison was randomized sibutramine versus randomized placebo for the risk of experiencing a POE from randomization up to 6 years. The analysis used the ITT population, a log-rank test, and a Cox proportional-hazards effect measure.
Primary Outcome Event
95% CI: 1.029–1.311 · P = 0.015
Randomized sibutramine vs randomized placebo
| Analysis component | Reported value |
|---|---|
| Endpoint | Risk of experiencing a Primary Outcome Event (POE) |
| Time frame | From randomization up to 6 years |
| Population | ITT |
| Comparison | Randomized Sibutramine vs Randomized Placebo |
| Method | Log Rank |
| Effect measure | Cox Proportional Hazard / hazard ratio |
| Estimate | 1.162 |
| 95% CI | 1.029–1.311 |
| P-value | 0.015 |
| Hypothesis | Superiority |
What the estimate means: A hazard ratio of 1.162 means that, under the Cox model used for this comparison, the estimated instantaneous rate of a first POE was 1.162 times that in the randomized placebo group. Because the sibutramine arm is the numerator, an HR above 1 corresponds to a higher estimated hazard in the sibutramine group.
What it does not mean: It does not mean that 16.2% more participants experienced a POE, nor does it represent an absolute increase in event probability. The hazard ratio is a relative, model-based measure of event rate over time.
What the confidence interval says: The two-sided 95% CI of 1.029–1.311 quantifies uncertainty around the estimated hazard ratio under the analysis framework. It does not describe the range of individual patient outcomes.
Why the P-value is not an effect size: The P-value of 0.015 addresses the statistical evidence against the null hypothesis in the prespecified superiority framework. It does not tell us how large the clinical effect is; the hazard ratio and its confidence interval provide that information.
Important caution: Cox-model interpretation relies on the proportional-hazards framework. The registry reports the Cox proportional-hazard effect measure but does not provide a proportional-hazards diagnostic in the information presented here. The comparison also concerns the ITT population and uses time-to-first POE, so it should not be interpreted as a simple comparison of cumulative event percentages.
8. Primary Results by Prespecified Cardiovascular-Risk Group
Diabetes mellitus only
DM-only analysis
95% CI: 0.738–1.384 · P = 0.948
The point estimate of 1.010 is very close to 1, indicating little difference in the estimated hazard between randomized sibutramine and randomized placebo in this subgroup. The 95% CI of 0.738–1.384 is considerably wider than the overall analysis interval, reflecting greater uncertainty in a subgroup analysis. The P-value of 0.948 is not a measure of effect magnitude and should not be read as proof that the two treatments are identical.
Cardiovascular disease only
CV-only analysis
95% CI: 0.916–1.776 · P = 0.149
The estimated hazard ratio is above 1, but the 95% CI extends from below 1 to above 1. That interval indicates substantial statistical uncertainty around the point estimate. The P-value of 0.149 does not establish a statistically significant superiority difference under the reported framework, but it also does not establish equivalence or absence of an effect.
Cardiovascular disease plus diabetes mellitus
CV + DM analysis
95% CI: 1.024–1.365 · P = 0.022
The CV + DM analysis estimates a hazard ratio of 1.182, with the sibutramine group as the numerator. The 95% CI of 1.024–1.365 lies above 1, indicating that the estimated hazard was higher for randomized sibutramine than randomized placebo within this prespecified risk group under the reported analysis. The P-value of 0.022 provides the corresponding hypothesis-test result.
This subgroup result should be interpreted as a subgroup-specific randomized comparison, not as evidence that the sibutramine effect necessarily differs from the effects in the DM-only or CV-only groups. Establishing effect modification requires a direct statistical comparison of treatment effects across groups.
9. What the Primary Hazard Ratio Tells Us
Direction
The overall HR of 1.162 is above 1 because the sibutramine arm was defined as the numerator and the placebo arm as the denominator.
Magnitude
The point estimate corresponds to a 16.2% higher estimated hazard in the sibutramine group relative to placebo under the Cox model.
Precision
The 95% CI of 1.029–1.311 describes uncertainty around the estimated relative hazard.
Hypothesis test
The P-value of 0.015 evaluates the statistical evidence under the reported superiority framework; it does not quantify clinical importance.
10. Secondary Time-to-Event Results
The registry reports six secondary time-to-event analyses comparing randomized sibutramine with randomized placebo over the period from randomization up to 6 years. Each used the ITT population, a log-rank test, and a Cox proportional-hazard effect measure.
| Secondary endpoint | HR | 95% CI | P-value |
|---|---|---|---|
| Risk of death from any cause (all-cause mortality) | 1.043 | 0.910–1.196 | 0.543 |
| Risk of experiencing a POE or a revascularization procedure | 1.097 | 0.999–1.204 | 0.051 |
| Risk of experiencing a nonfatal MI included in the POE | 1.276 | 1.036–1.571 | 0.022 |
| Risk of experiencing a nonfatal stroke included in the POE | 1.355 | 1.038–1.767 | 0.025 |
| Risk of experiencing a resuscitated cardiac arrest included in the POE | 1.582 | 0.613–4.081 | 0.343 |
| Risk of experiencing cardiovascular death included in the POE | 0.988 | 0.822–1.188 | 0.899 |
How to read the secondary endpoint pattern
The secondary results illustrate why a composite endpoint should be examined alongside its components. The overall POE combines several clinically distinct event types. Their individual hazard ratios vary in both magnitude and precision, so the composite cannot be understood simply by looking at one component.
The resuscitated-cardiac-arrest estimate is particularly imprecise: the 95% CI extends from 0.613 to 4.081. A wide interval means that the point estimate provides relatively limited precision about the underlying relative hazard. The appropriate interpretation is therefore different from an endpoint whose confidence interval is substantially narrower.
A P-value near 0.05 should not be treated as a universal boundary separating important from unimportant effects. For example, the POE-or-revascularization analysis has P = 0.051 and a 95% CI of 0.999–1.204, while the nonfatal-MI analysis has P = 0.022 and a 95% CI of 1.036–1.571. These results should be interpreted through their effect estimates and uncertainty, endpoint definitions, and prespecified analysis roles rather than by P-value alone.
11. Safety Results
The registry reports serious adverse events by arm using affected participants divided by participants at risk.
| Period / arm | Serious adverse events affected | At risk |
|---|---|---|
| Lead-in Period Sibutramine | 425 | 10744 |
| Randomized Sibutramine | 2063 | 4904 |
| Randomized Placebo | 1977 | 4881 |
These summaries describe affected participants relative to the corresponding number at risk. They are not time-to-event hazard ratios and should not be interpreted as interchangeable with the efficacy analysis.
12. Statistical Methods Explained
Why was a time-to-event analysis used?
The primary endpoint is defined by both event status and time to the first event. Time-to-event methods allow participants to contribute information for different lengths of follow-up and account for the fact that an event occurring earlier carries different information about the event process than one occurring later.
Why use a log-rank test?
The log-rank test is designed for comparing survival or event-time distributions between randomized groups. Instead of reducing each participant to a simple yes/no event indicator, it uses the timing of events and the risk set over follow-up.
What does an HR of 1.162 mean?
Because the registry defines the sibutramine arm as the numerator and placebo as the denominator, an HR of 1.162 means that the estimated instantaneous event rate under the Cox model is 1.162 times the corresponding rate in the placebo group. It does not mean that 16.2% of patients experienced an event or that every participant had a 16.2% increase in individual risk.
Why is the confidence interval important?
The point estimate alone does not communicate precision. The 95% CI of 1.029–1.311 for the overall primary analysis shows the range of relative-hazard values compatible with the statistical uncertainty represented by the model and sample. Wider intervals indicate less precise estimation than narrower intervals.
Why doesn't P = 0.015 measure effect size?
A P-value measures the compatibility of the observed data with a specified null hypothesis under the statistical model. It depends on both the magnitude of the observed effect and the amount of information available. Effect magnitude is communicated by the hazard ratio, while uncertainty is communicated by its confidence interval.
Why should subgroup hazard ratios not be ranked against one another?
The four primary analyses include the overall population and three prespecified cardiovascular-risk groups. Differences between subgroup point estimates can arise from sampling variation. To conclude that the treatment effect truly differs by risk group, a direct interaction or heterogeneity analysis is required; comparing whether individual subgroup P-values are above or below 0.05 is not a valid interaction test.
13. Interpreting the Sample-Size Calculation
The registry's design calculation illustrates the distinction between planning assumptions and observed results. The calculation used a two-tailed alpha of 0.05, 90% power, an assumed annual composite event rate of 7.0% in the placebo arm, and a target of detecting a relative risk reduction of 15% with sibutramine relative to placebo.
| Design quantity | Registry value |
|---|---|
| Alpha | 0.05, two-tailed |
| Power | 90% |
| Assumed annual composite event rate in placebo | 7.0% |
| Planned sample per randomized group | 3983 |
| Noncompliance adjustment | 30% |
| Expected follow-up for calculation | At least 3 years |
| Target detectable effect | Relative risk reduction of 15% |
The registry further states that the Cox model included treatment, country, gender, and age at Lead-in Period baseline. This means the prespecified modeling strategy incorporated baseline covariates in addition to the randomized treatment assignment.
The design calculation should not be interpreted as evidence that the final trial effect was expected to equal 15%. Power is a planning property conditional on assumptions about event rates, sample size, follow-up, noncompliance, and the target effect. The completed-trial hazard ratio of 1.162 is an observed estimate and answers a different question.
14. Limitations and Interpretation Issues
- No true placebo throughout the study: the registry states that all subjects received sibutramine during the Lead-in Period. Consequently, the randomized placebo comparison occurred after an initial sibutramine exposure rather than from the beginning of the entire trial pathway.
- Population and cardiovascular disease: the registry notes that most subjects were not indicated for sibutramine because of cardiovascular disease. This is an important consideration when interpreting the enrolled population and the context of the randomized comparison.
- Long-term exposure: the registry states that long-term sibutramine use for 6 years regardless of weight loss was inconsistent with the label, which limited use to 1–2 years and 5% weight loss.
- Time-to-event interpretation: hazard ratios describe relative event rates under the Cox model rather than absolute event probabilities.
- Proportional-hazards assumption: a Cox hazard ratio is most straightforward to interpret as a single relative hazard when the proportional-hazards framework is reasonable over follow-up. The registry information summarized here does not provide a diagnostic assessment of that assumption.
- Subgroup uncertainty: the DM-only, CV-only, and CV + DM analyses have wider confidence intervals than the overall analysis, illustrating the additional uncertainty that accompanies subgroup analysis.
- Composite endpoint: the POE combines nonfatal MI, nonfatal stroke, resuscitated cardiac arrest, and cardiovascular death. Components can have different frequencies and treatment effects, so the composite should be interpreted together with its individual components.
- Multiple reported analyses: the registry reports four primary-endpoint analyses and six secondary analyses. Individual P-values should be interpreted in the context of the prespecified analysis structure rather than treated as isolated tests without regard to the overall analysis plan.
15. Why This Trial Matters Statistically
SCOUT is a useful statistical case study because it combines a randomized treatment comparison with a long-term composite cardiovascular endpoint and multiple prespecified risk-group analyses. The trial demonstrates several recurring issues in clinical-trial statistics: the distinction between randomized treatment and prior exposure, the use of ITT analysis, survival-analysis methods, covariate-adjusted Cox modeling, and the interpretation of subgroup-specific hazard ratios.
| Concept | How it appears in SCOUT |
|---|---|
| Randomization | Randomized parallel comparison of sibutramine and placebo after a sibutramine Lead-in Period |
| Blinding | Double-masked design |
| ITT analysis | Primary analyses based on randomized subjects dispensed randomized study drug |
| Time-to-event endpoint | POE assessed from randomization up to 6 years |
| Log-rank test | Reported method for primary and secondary time-to-event comparisons |
| Hazard ratio | Cox proportional-hazard effect measure |
| Confidence interval | 95% two-sided intervals around reported hazard ratios |
| Covariate adjustment | Cox model included treatment, country, gender, and age at Lead-in Period baseline for the overall primary analysis |
| Composite endpoint | POE combined nonfatal MI, nonfatal stroke, resuscitated cardiac arrest, and CV death |
| Subgroup analysis | DM only, CV only, and CV + DM prespecified risk groups |
| Safety population context | Serious adverse events reported separately for Lead-in and randomized arms |
16. A Deeper Look at the Composite Endpoint
The primary endpoint is a composite of four clinically distinct outcomes: nonfatal MI, nonfatal stroke, resuscitated cardiac arrest, and cardiovascular death. The statistical advantage of a composite is that several event types contribute to the event count and therefore to the time-to-event analysis.
But a composite endpoint also requires careful interpretation. An overall hazard ratio summarizes the combined endpoint; it does not imply that the same relative effect applies to every component. The reported secondary analyses make this especially clear: the hazard ratios for nonfatal MI, nonfatal stroke, resuscitated cardiac arrest, and cardiovascular death are 1.276, 1.355, 1.582, and 0.988, respectively, with substantially different confidence-interval widths.
The composite POE result should therefore be viewed as the treatment comparison for the prespecified combined endpoint. The component analyses provide additional information about the pattern of events, but they do not transform the composite into four independent primary endpoints.
17. Understanding the Prespecified Risk Groups
The registry identifies three prespecified cardiovascular-risk categories: DM only, CV only, and CV + DM. The same POE endpoint and randomized sibutramine-versus-placebo comparison were evaluated within each category.
The visual comparison is descriptive rather than a formal interaction analysis. Point estimates can differ across groups because of random variation and different amounts of information. The confidence intervals provide the appropriate context for precision, and a direct treatment-by-risk-group interaction would be needed to formally assess heterogeneity.
18. Confidence Intervals as an Interpretation Tool
| Analysis | HR | 95% CI | Interpretive feature |
|---|---|---|---|
| Overall primary POE | 1.162 | 1.029–1.311 | Interval entirely above 1 |
| DM only | 1.010 | 0.738–1.384 | Wide interval spanning 1 |
| CV only | 1.276 | 0.916–1.776 | Wide interval spanning 1 |
| CV + DM | 1.182 | 1.024–1.365 | Interval entirely above 1 |
A confidence interval should be read as part of the effect estimate rather than as a decorative statistic appended to a P-value. For the overall primary analysis, the interval is relatively narrower and remains above 1. For the DM-only and CV-only analyses, the intervals are wider and include 1, reflecting greater uncertainty around the subgroup estimates.
19. Primary Versus Secondary Analyses
The registry identifies one registered primary endpoint, but four primary-endpoint statistical analyses: the overall randomized population and three prespecified cardiovascular-risk groups. It also reports six secondary time-to-event analyses.
| Analysis class | Number reported | Core statistical approach |
|---|---|---|
| Registered primary endpoint | 1 | Time-to-event analysis |
| Primary statistical analyses | 4 | Log-rank test + Cox proportional-hazard estimate |
| Primary analyses with estimate + CI | 4 | Hazard ratio with 95% two-sided CI |
| Secondary statistical analyses | 6 | Log-rank test + Cox proportional-hazard estimate |
This structure matters because the same P-value has different interpretive context depending on whether it belongs to the primary randomized comparison, a prespecified subgroup analysis, or a secondary endpoint. Statistical interpretation should therefore retain the endpoint hierarchy and analysis role rather than treating every reported test as equivalent.
20. Statistical Concepts in This Trial
Learn more about the methods used in this trial:
21. Related Statistical Calculators
22. Sources
- ClinicalTrials.gov: SCOUT, NCT00234832.
- PubMed: PMID 24534073.
- PubMed: PMID 24406481.
- PubMed: PMID 20818901.
- PubMed: PMID 20184783.
- PubMed: PMID 19828038.
Continue through Clinical Biostats
Connect the endpoints and statistical methods in this trial to deeper biostatistics tutorials and statistical tools.
23. Record Summary
SCOUT provides a detailed example of cardiovascular time-to-event analysis in a randomized clinical trial. The primary endpoint was a composite POE evaluated from randomization up to 6 years, with the primary comparison based on the ITT population and analyzed using a log-rank test and a Cox proportional-hazard effect measure. The overall hazard ratio was 1.162 with a 95% CI of 1.029–1.311 and P = 0.015.
The statistical interpretation is broader than the primary P-value. The analysis requires understanding the direction of the hazard ratio, the precision conveyed by the confidence interval, the distinction between relative hazard and absolute event probability, the role of the ITT population, and the implications of the prespecified cardiovascular-risk analyses. The secondary component analyses further show why a composite endpoint should be interpreted together with its individual event types.
The trial also illustrates an important design consideration: the randomized placebo comparison followed a Lead-in Period in which all subjects received sibutramine. Combined with the registry's stated concerns about the enrolled cardiovascular-risk population and the duration of sibutramine exposure, this feature is central to understanding what the randomized comparison does and does not represent.