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
SURMOUNT-MMO is a randomized, double-masked, parallel-group phase 3 study evaluating tirzepatide versus placebo in adults with obesity or overweight. The primary endpoint is a time-to-first-occurrence composite that includes all-cause death, nonfatal myocardial infarction, nonfatal stroke, coronary revascularization, or heart failure events resulting in hospitalization or urgent visits, with a time frame of up to 5 years.
| Feature | SURMOUNT-MMO |
|---|---|
| Phase | Phase 3 |
| Population | Adults with obesity or overweight |
| Design | Randomized, parallel-group, double-masked |
| Allocation | Randomized |
| Primary purpose | Treatment |
| Primary endpoint | Time to first occurrence of any component event of a composite consisting of all-cause death, nonfatal myocardial infarction, nonfatal stroke, coronary revascularization, or heart failure events resulting in hospitalization or urgent visits |
| Primary endpoint time frame | Up to 5 Years |
| Enrollment | 15,374 |
| Interventions | Tirzepatide; Placebo |
| Lead sponsor | Eli Lilly and Company |
| Sponsor type | Industry |
| Study start | 2022-10-11 |
| Primary completion date | 2027-10 |
| ClinicalTrials.gov | NCT05556512 |
2. Clinical Question
The central statistical question is whether assignment to tirzepatide, compared with assignment to placebo, changes the time until the first occurrence of any event in the prespecified composite endpoint among adults with obesity or overweight.
Population
Adults with obesity or overweight.
Intervention
Tirzepatide.
Comparator
Placebo.
Primary question
Does tirzepatide alter the time to first occurrence of any component event in the composite endpoint compared with placebo?
3. Trial Design
Tirzepatide
- Study intervention: tirzepatide
- Randomized assignment
- Double-masked study design
- Parallel-group comparison
Placebo
- Study intervention: placebo
- Randomized assignment
- Double-masked study design
- Parallel-group comparison
4. Registered Primary Endpoint
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Time to First Occurrence of Any Component Event of Composite | Time to first occurrence of any component event of composite, all-cause death, nonfatal MI, nonfatal stroke, coronary revascularization, or heart failure events that results in hospitalization or urgent visits. | Up to 5 Years |
The endpoint is a time-to-first-event composite. The patient is considered to have reached the endpoint when the first qualifying component event occurs. The design therefore incorporates both whether an event occurs and when it occurs.
Components of the composite
| Component | Role in the primary endpoint |
|---|---|
| All-cause death | Death from any cause is a qualifying component event. |
| Nonfatal myocardial infarction | A nonfatal MI is a qualifying component event. |
| Nonfatal stroke | A nonfatal stroke is a qualifying component event. |
| Coronary revascularization | Coronary revascularization is a qualifying component event. |
| Heart failure events | Heart failure events resulting in hospitalization or urgent visits are qualifying component events. |
The composite is designed to capture several clinically important event types within one time-to-event outcome. Statistically, the first qualifying component determines the patient's time-to-event for this endpoint.
5. Statistical Structure of the Primary Endpoint
A conventional binary endpoint would ask whether a patient experienced an event during a specified observation period. SURMOUNT-MMO instead defines the primary endpoint as time to first occurrence, which preserves information about the timing of the first event.
If no qualifying event is observed during the available follow-up, the patient's observation is typically treated as right-censored at the appropriate end of follow-up. The precise censoring rules and statistical analysis specification are not reported in the registry information summarized here.
This distinction matters because two patients can both remain event-free through a particular assessment point while contributing different amounts of follow-up information. Time-to-event methods use the observed follow-up duration rather than reducing every participant to a simple event/no-event indicator.
6. Statistical Methodology
Kaplan-Meier estimation
For a time-to-first-event endpoint, the Kaplan-Meier estimator is a standard method for describing the probability of remaining event-free over time. It accommodates right-censored observations and produces an estimated event-free survival curve.
Here, di represents the number of events at event time ti, while ni represents the number at risk immediately before that event time.
Cox proportional-hazards modeling
A Cox proportional-hazards model is another standard approach for comparing time-to-event outcomes. It estimates a hazard ratio comparing the instantaneous event rates between randomized groups, subject to the model's assumptions.
An HR below 1 would indicate a lower estimated instantaneous rate of the composite event in the tirzepatide group under the fitted model. An HR above 1 would indicate a higher estimated instantaneous rate.
Log-rank comparison
The log-rank test is commonly used to compare survival distributions between randomized groups. For a time-to-first-event endpoint, it evaluates whether the observed event-time patterns differ between groups across follow-up.
Intention-to-treat analysis
For a randomized treatment-effect question, the intention-to-treat principle analyzes participants according to their randomized assignment. This preserves the comparison created by randomization and estimates the effect associated with assignment rather than attempting to reproduce an observational comparison among participants who remain continuously exposed to treatment.
Confidence intervals
A confidence interval around a hazard ratio quantifies statistical uncertainty in the estimated treatment effect. A narrow interval indicates greater precision than a wide interval, although precision is distinct from clinical importance.
A hazard ratio is a relative time-to-event measure. It is not a percentage of patients who will experience the event, and it is not an absolute difference in event probability. Absolute event-free probabilities at specific time points can provide complementary information when they are reported.
7. Planned Analysis
The ClinicalTrials.gov record identifies a single primary time-to-first composite endpoint with a time frame of up to 5 years. The registry information summarized here does not report posted statistical analyses or numerical efficacy results for this endpoint.
For an endpoint of this type, a conventional statistical analysis would generally describe the time-to-event distribution with Kaplan-Meier methods and compare randomized groups using a time-to-event test such as the log-rank test, with a Cox proportional-hazards model commonly used to estimate a hazard ratio and its confidence interval. The exact analysis population, censoring rules, covariate adjustment, hypothesis-testing procedure, and multiplicity strategy should be determined from the prespecified statistical analysis plan when available.
The primary endpoint is a composite, so the analysis also needs to preserve the prespecified definition of a first qualifying event. A patient who experiences more than one component does not contribute multiple first events to the primary time-to-event analysis; the endpoint is defined by the first occurrence.
8. Statistical Methods Explained
Why use a time-to-first-event endpoint?
A time-to-event endpoint incorporates follow-up duration rather than simply classifying every participant as having or not having an event. This is particularly useful when participants have different lengths of observable follow-up or when events occur throughout an extended study period.
Why combine several event types into one composite?
A composite endpoint can capture several clinically relevant outcomes within one prespecified outcome measure. It can increase the number of observed first events relative to relying on a single component, which can make a time-to-event comparison more informative. The trade-off is interpretational: the overall composite can be influenced disproportionately by whichever component occurs most frequently.
Why does the first event matter?
The registry defines the endpoint as time to first occurrence. If a participant experiences a nonfatal MI followed later by another qualifying event, the first qualifying event determines the endpoint time. This makes the primary analysis a time-to-first-event analysis rather than a count of all cardiovascular or mortality events.
What would a hazard ratio mean here?
A hazard ratio compares the estimated instantaneous event rate between the randomized groups. For example, an HR of 0.80 would describe a model-based event rate in the treatment group that is 80% of the control-group rate under the proportional-hazards interpretation. It would not mean that exactly 20% fewer participants experience the composite endpoint.
Why is censoring important?
Some participants may reach the end of their observable follow-up without experiencing a qualifying event. Time-to-event methods can retain their information up to the censoring time rather than treating them as if they had been followed indefinitely. Correct censoring rules are therefore an important part of the prespecified statistical analysis.
What does a 5-year time frame imply statistically?
A time frame of up to 5 years defines a long observation window for the primary endpoint. The analysis must therefore distinguish events that occur earlier from participants who remain event-free for longer periods, while accounting appropriately for censoring and the changing number of participants remaining under observation.
9. Composite Endpoint Interpretation
Composite endpoints are statistically efficient only when their individual components are clinically and analytically coherent. SURMOUNT-MMO combines all-cause death, nonfatal MI, nonfatal stroke, coronary revascularization, and qualifying heart failure events.
Strength of the composite
The endpoint captures several clinically meaningful event types rather than relying on a single outcome category.
Key interpretive issue
An overall treatment effect for the composite does not automatically imply that every component is affected to the same degree.
First-event structure
The registry defines the outcome around the first occurrence of a qualifying component event.
Component interpretation
Component-specific event patterns are important for understanding what drives any eventual composite result.
For this reason, a complete eventual statistical interpretation would ideally examine the total composite together with the frequency and nature of its individual components. The registry information summarized here does not provide numerical component results.
10. Randomization and Masking
Participants are assigned to study arms through randomization.
The study has two parallel intervention groups.
The registry classifies the study as double-masked.
The primary purpose is classified as treatment.
Randomization is important because it provides the design basis for comparing outcomes between the tirzepatide and placebo groups. Double masking can reduce the influence of knowledge of treatment assignment on participant behavior, clinical management, outcome assessment, or other processes, depending on how masking is implemented.
11. Enrollment and Study Scale
The registry reports enrollment of 15,374 participants across 2 arms. A large randomized population can provide substantial information for a time-to-event endpoint, particularly when the study is designed to observe participants for an extended period.
| Study characteristic | Registry value |
|---|---|
| Enrollment | 15,374 |
| Number of arms | 2 |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Primary purpose | Treatment |
| Phase | 3 |
Enrollment alone does not determine statistical power. For a time-to-event endpoint, the number and timing of qualifying events, follow-up duration, censoring, treatment effect, allocation, and statistical design all contribute to the amount of information available for the primary comparison.
12. Study Timeline
Study start
The registry lists 2022-10-11 as the study start date.
Current registry status
The ClinicalTrials.gov record lists the study as active, not recruiting.
Primary completion
The registry lists 2027-10 as the primary completion date.
The timeline is relevant to interpretation because the primary endpoint has a time frame of up to 5 years. A time-to-event analysis is inherently dependent on the accumulation of follow-up and qualifying events over time.
13. Analysis Population Considerations
The registry information identifies the randomized design and enrollment but does not report detailed analysis-population definitions in the information summarized here.
| Population concept | Statistical role |
|---|---|
| Randomized population | Defines the treatment comparison created by randomization and is the natural basis for an intention-to-treat efficacy analysis. |
| Safety population | Typically defined according to treatment exposure and used for describing adverse events; the registry information summarized here does not provide a detailed safety-population definition. |
| Per-protocol population | Can be useful for selected design questions, but the registry information summarized here does not identify a prespecified per-protocol analysis. |
For a randomized superiority-style treatment question, the intention-to-treat principle is generally important because excluding participants after randomization can disrupt the comparability created by the randomization process.
14. Missing Data and Censoring
Time-to-event analyses do not treat every missing observation in the same way as a conventional continuous or binary endpoint. Instead, participants without a documented qualifying event may contribute follow-up until an appropriate censoring time.
Censoring
Censoring represents the point after which the event status or event time is no longer observed under the analysis framework.
Informative loss to follow-up
If censoring is related to the future event process, standard survival estimates can be affected. Prespecified follow-up and censoring rules are therefore important.
Event ascertainment
Composite time-to-event studies require consistent identification and adjudication of qualifying events according to the endpoint definition.
Analysis-plan detail
The registry information summarized here does not report the detailed censoring or missing-data rules.
For this reason, a numerical reconstruction of the primary analysis requires more than the enrollment number and endpoint definition. The event dates, censoring information, and prespecified analysis rules determine the actual statistical estimate.
15. Multiplicity and Component Endpoints
The registry identifies one registered primary composite endpoint. The composite contains five categories of qualifying events, but the existence of multiple components does not by itself mean that each component constitutes a separate primary hypothesis.
This distinction is important when interpreting future results. A statistically significant composite comparison would address the prespecified composite endpoint. It would not automatically establish statistically significant effects for each individual component unless those component analyses were separately specified and appropriately controlled.
Component-specific questions are analytically distinct. Their interpretation depends on the prespecified statistical hierarchy, multiplicity control, event counts, and confidence intervals used for those analyses.
16. Statistical Precision and Confidence Intervals
When results are eventually reported, the hazard ratio should be interpreted together with its confidence interval rather than as a standalone number.
The point estimate summarizes the relative treatment effect under the chosen time-to-event model. For a hazard ratio, the reference value of 1 represents equal estimated hazards between the randomized groups.
The confidence interval describes statistical uncertainty around the estimated treatment effect under the specified model and sampling framework. It does not describe the range of effects experienced by individual participants.
A p-value addresses compatibility of the observed data with a specified null hypothesis under the statistical testing framework. It does not measure the magnitude of the treatment effect, clinical importance, or the probability that a treatment works.
17. Proportional-Hazards Assumption
If a Cox proportional-hazards model is used for the primary endpoint, one important modeling consideration is the proportional-hazards assumption. The model interprets the hazard ratio as a relative comparison that is stable over time, at least in the sense required by the model.
If the relative event rates change substantially over follow-up, a single hazard ratio can compress a more complicated time-varying pattern into one summary number. Kaplan-Meier curves, time-specific event-free probabilities, and other descriptive measures can therefore be useful complements to the hazard ratio.
18. Limitations
- No posted numerical efficacy analysis: the registry information summarized here does not contain numerical results for the primary endpoint or a posted statistical analysis.
- Future completion date: the registry lists a primary completion date of 2027-10, so the planned time frame extends beyond the current registry status.
- Composite interpretation: a treatment effect on the overall composite would not necessarily imply identical effects on all five component categories.
- First-event limitation: the primary endpoint captures the first qualifying event and therefore does not summarize the total number of recurrent events experienced by participants.
- Censoring assumptions: survival estimates depend on the analysis's censoring rules and assumptions concerning follow-up.
- Model assumptions: a Cox hazard ratio, if used, depends on the assumptions of the fitted model, including the proportional-hazards framework.
- Analysis-plan detail: the registry information summarized here does not report detailed statistical-analysis specifications such as exact censoring rules, multiplicity procedures, or covariate-adjustment rules.
- Generalizability: the enrolled population is defined by the study's eligibility criteria for adults with obesity or overweight; the registry summary does not provide a detailed baseline-characteristic table here.
19. Why This Trial Matters Statistically
SURMOUNT-MMO is a useful teaching example for understanding how a large randomized study can be structured around a long-term composite time-to-event endpoint rather than a simple short-term outcome.
| Statistical concept | How it appears in SURMOUNT-MMO |
|---|---|
| Randomization | The study uses randomized allocation to compare tirzepatide with placebo. |
| Parallel-group design | The registry identifies a two-arm parallel design. |
| Double masking | The registry classifies the study as double-masked. |
| Time-to-event analysis | The primary endpoint measures time to first occurrence of a qualifying event. |
| Composite endpoint | The primary endpoint combines five categories of qualifying events. |
| Censoring | Time-to-event methods can incorporate participants who remain event-free through their available follow-up. |
| Hazard ratio | A Cox model is a standard framework for summarizing relative event rates if prespecified for the analysis. |
| Kaplan-Meier estimation | A standard method for describing event-free survival over time. |
| Confidence intervals | Quantify uncertainty around estimated treatment effects. |
| Component interpretation | The overall composite must be distinguished from effects on its individual components. |
20. Statistical Interpretation of the Study Design
What randomization contributes
Randomization establishes the treatment groups before outcome information accumulates, supporting a causal comparison of assignment under the trial design.
What masking contributes
Double masking is intended to reduce the influence of treatment knowledge on trial conduct and outcome-related processes.
What the composite contributes
The composite captures several predefined event categories within one primary time-to-first-event outcome.
What the 5-year window contributes
The long time frame allows the primary analysis to evaluate when qualifying events occur across extended follow-up rather than at only one fixed assessment point.
21. What a Future Primary Result Would Need to Show
A complete statistical presentation of the primary endpoint would ideally include the number of qualifying first events in each randomized group, the estimated time-to-event distributions, an effect estimate such as a hazard ratio if prespecified, its confidence interval, and the corresponding hypothesis-test result where applicable.
Because the endpoint is a composite, a complete presentation would also benefit from reporting the contribution of the individual components. This helps distinguish an overall composite effect from the pattern of events that produced it.
| Result element | Why it matters |
|---|---|
| Number of first composite events | Shows how much event information contributes to the comparison. |
| Time-to-event curves | Shows how event-free experience changes over follow-up. |
| Hazard ratio | Summarizes the relative event rate under a time-to-event model. |
| Confidence interval | Shows the statistical precision of the estimated effect. |
| P-value | Addresses the prespecified hypothesis test; it does not quantify effect size. |
| Component events | Shows which event categories contribute to the composite. |
22. Current Statistical Record
The ClinicalTrials.gov record identifies SURMOUNT-MMO as a phase 3, randomized, double-masked, parallel-group study with 15,374 participants and two interventions: tirzepatide and placebo. Its registered primary endpoint is time to first occurrence of a composite of all-cause death, nonfatal MI, nonfatal stroke, coronary revascularization, or heart failure events resulting in hospitalization or urgent visits, with a time frame of up to 5 years.
The registry currently lists the study as active, not recruiting, with a study start date of 2022-10-11 and a primary completion date of 2027-10. No posted statistical analyses are reported in the ClinicalTrials.gov information summarized here.
23. Sources
- ClinicalTrials.gov: SURMOUNT-MMO, NCT05556512.