This page separates trial information reported in the ClinicalTrials.gov record from statistical interpretation. The registry indicates that results have been posted, but the ClinicalTrials.gov record contains no formal statistical-analysis entries.
1. Trial at a Glance
SAVOR-TIMI 53 was a randomized, parallel-group, quadruple-masked phase 4 trial in participants with type 2 diabetes mellitus. The trial evaluated saxagliptin versus placebo, with 18,206 participants enrolled and a primary endpoint defined as any event in a composite of cardiovascular death, non-fatal myocardial infarction, or non-fatal ischaemic stroke.
| Feature | SAVOR-TIMI 53 |
|---|---|
| Trial name | SAVOR-TIMI 53 |
| ClinicalTrials.gov identifier | NCT01107886 |
| Phase | Phase 4 |
| Status | COMPLETED |
| Therapeutic area | Endocrinology |
| Condition | Type 2 Diabetes Mellitus |
| Allocation | RANDOMIZED |
| Design model | PARALLEL |
| Masking | QUADRUPLE |
| Primary purpose | TREATMENT |
| Interventions | Saxagliptin (drug); Placebo (drug) |
| Enrollment | 18,206 |
| Lead sponsor | AstraZeneca |
| Sponsor type | INDUSTRY |
| Start | May 2010 |
| Primary completion | May 2013 |
2. Clinical Question
The registered clinical question asks whether saxagliptin, used alone or added to other diabetes medications, is associated with a different risk of major cardiovascular events than placebo in participants with type 2 diabetes mellitus.
Population
Participants with Type 2 Diabetes Mellitus.
Intervention
Saxagliptin, identified in the registry as a drug intervention.
Comparator
Placebo, identified in the registry as a drug intervention.
Primary question
Whether participants experience an event from the registered composite of cardiovascular death, non-fatal myocardial infarction, or non-fatal ischaemic stroke during the specified follow-up period.
3. Trial Design
Participants were assigned to parallel trial arms according to the randomized design.
The registry identifies the study as quadruple-masked.
The two interventions were evaluated in a parallel-group design.
The registered primary purpose of the trial was treatment.
Saxagliptin
- Drug intervention
- Evaluated in the randomized parallel-group design
- Included in the primary cardiovascular composite comparison
Placebo
- Drug intervention listed as placebo
- Evaluated in the randomized parallel-group design
- Served as the comparator for the primary endpoint
4. Endpoints
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Primary composite cardiovascular endpoint | Participants With Any Event From the Composite of Cardiovascular Death (CV Death), Non-fatal Myocardial Infarction (MI), or Non-fatal Ischaemic Stroke | Randomization (day 0) up to 2.9 years | Binary |
Primary endpoint definition
The registered primary endpoint counts a participant as having an event if the participant experiences cardiovascular death, non-fatal myocardial infarction, or non-fatal ischaemic stroke.
The registry specifies that, if no event occurs, censoring occurs at the patient's withdrawal of consent, last contact, or death, when applicable, whichever was later.
The registered endpoint is therefore a composite binary outcome assessed from randomization through 2.9 years.
5. Statistical Methodology
The ClinicalTrials.gov record classifies the primary endpoint as Binary. That classification is important because it describes the endpoint as an event indicator over the registered follow-up period rather than providing a numerical time-to-event analysis.
For a binary endpoint defined over a fixed follow-up interval, a statistical comparison would ordinarily begin with the number and proportion of participants experiencing the composite event in each randomized arm. Depending on the prespecified statistical analysis plan, possible comparative measures include an absolute risk difference, relative risk, odds ratio, or a model that accounts for time to event.
However, the ClinicalTrials.gov record contains no formal statistical analyses. Consequently, no formal comparative estimate is reported on this page.
| Statistical element | What the ClinicalTrials.gov record establishes | Interpretation |
|---|---|---|
| Endpoint type | Binary | The registered primary outcome is represented as whether an event occurred during the specified period. |
| Analysis entries | None reported | No formal statistical estimate can be reproduced from the registry-reported analysis field. |
| Results posted | Yes | The registry indicates that results are posted, but the ClinicalTrials.gov record does not include the formal statistical-analysis results. |
| Outcome measures posted | 3 | Three outcome measures are identified as posted in the ClinicalTrials.gov record. |
6. Planned Analysis
The registry identifies the primary endpoint as a binary composite measured from randomization through 2.9 years. A conventional analysis of this type of endpoint would compare the occurrence of the composite event between the randomized saxagliptin and placebo groups.
What would be measured?
The number and proportion of participants experiencing cardiovascular death, non-fatal myocardial infarction, or non-fatal ischaemic stroke during the registered time frame.
What would be compared?
The event experience in the randomized saxagliptin group would be compared with that in the randomized placebo group.
What could the effect measure be?
For a binary endpoint, an analysis may use measures such as risk difference, relative risk, or odds ratio, depending on the prespecified statistical analysis plan.
What does the ClinicalTrials.gov record reports?
No formal statistical-analysis entry is reported. Therefore, no numerical comparative result is reported here.
Censoring information
The registry definition introduces censoring for participants who do not experience the event: censoring occurs at withdrawal of consent, last contact, or death when applicable, whichever was later. This wording is important because it shows that participant follow-up is not necessarily identical for every randomized participant.
7. How to Think About the Primary Composite
A composite endpoint combines multiple clinically distinct events into one outcome. In SAVOR-TIMI 53, the registered composite consists of cardiovascular death, non-fatal myocardial infarction, and non-fatal ischaemic stroke.
The statistical attraction of a composite is that more participants may experience at least one component than would experience any single component alone. That can provide more events for analysis. But interpretation depends on understanding what actually drives the composite.
| Component | Role in the registered endpoint |
|---|---|
| Cardiovascular death | One of the three events that qualifies a participant for the composite endpoint |
| Non-fatal myocardial infarction | One of the three events that qualifies a participant for the composite endpoint |
| Non-fatal ischaemic stroke | One of the three events that qualifies a participant for the composite endpoint |
A composite result should therefore not automatically be interpreted as demonstrating the same effect on each component. The ClinicalTrials.gov record does not provide component-specific event counts or effect estimates, so no component-level conclusion can be drawn here.
8. Statistical Methods Explained
Why use a composite endpoint?
A composite endpoint counts the first qualifying occurrence among several predefined event types. Here, the qualifying events are cardiovascular death, non-fatal myocardial infarction, and non-fatal ischaemic stroke. Combining them creates a single primary endpoint for the registered analysis.
What does “binary” mean for this endpoint?
A binary endpoint records whether the specified outcome occurred during the defined observation period. In this trial, the registry classifies the primary endpoint as binary and defines the observation window as randomization (day 0) up to 2.9 years.
Why does randomization matter statistically?
Randomization creates the principal basis for comparing the two treatment strategies. If randomization is properly implemented, systematic differences between treatment groups are reduced at the point of assignment, allowing subsequent differences in outcomes to be interpreted within a randomized comparison.
What does quadruple masking contribute?
Quadruple masking means the registry identifies the trial as having four masked roles or groups. Masking can reduce the risk that knowledge of treatment assignment influences behavior, assessment, reporting, or other aspects of trial conduct. The ClinicalTrials.gov record does not specify which four roles were masked.
Why is the censoring rule important?
Not every participant necessarily contributes identical follow-up. The registry specifies that, if no event occurs, censoring occurs at withdrawal of consent, last contact, or death when applicable, whichever was later. Any analysis must therefore respect the registered rules governing when outcome observation ends.
Why should the composite not be treated as three separate outcomes?
The registered primary endpoint is one composite outcome. A participant who experiences any one qualifying component meets the endpoint definition. The composite result therefore answers a combined question. Without component-specific results, it is not possible to determine which event type contributed most to any observed treatment-group difference.
Why can’t a p-value or confidence interval be reported here?
No formal statistical analyses were posted. A confidence interval and p-value require an underlying estimate and its statistical uncertainty. Supplying numerical values not present in the trial data would require importing external results, which would violate the data rules for this analysis.
9. Safety
The ClinicalTrials.gov record includes serious adverse-event counts by arm. These figures are reported as affected participants over the corresponding at-risk population.
| Arm | Participants with serious adverse events | At risk |
|---|---|---|
| Placebo | 2,075 | 8,212 |
| Saxagliptin | 2,114 | 8,280 |
The affected-participant counts should be interpreted together with their corresponding at-risk populations. The ClinicalTrials.gov record does not provide a formal statistical comparison, confidence interval, p-value, seriousness definitions beyond the registry category, or event-level breakdown for these serious adverse events.
10. Statistical Interpretation of the Available Safety Data
The serious-adverse-event data illustrate an important distinction between an event count and an event rate or risk. The placebo arm has 2,075 affected participants among 8,212 at risk, while the saxagliptin arm has 2,114 affected participants among 8,280 at risk.
The affected counts are close in absolute magnitude, but a rigorous comparison would normally account for the different denominators and for the way follow-up and event ascertainment were handled. The ClinicalTrials.gov record does not provide enough information to reproduce a formal inferential analysis.
The counts tell us how many participants are reported as affected by serious adverse events in each arm and how many participants were at risk. They do not by themselves establish whether the treatment groups differ statistically.
A confidence interval around a prespecified comparative measure would quantify statistical uncertainty. No such interval is included in the ClinicalTrials.gov record, so none is reported here.
Two groups can have similar numbers of affected participants but different risks if their numbers at risk differ. Conversely, a difference in raw counts does not necessarily imply a meaningful difference in risk.
11. Trial Timeline
Trial start
The ClinicalTrials.gov record identifies May 2010 as the trial start.
Randomized trial period
The trial used randomized allocation, a parallel design, and quadruple masking to evaluate saxagliptin and placebo.
Primary completion
The ClinicalTrials.gov record identifies May 2013 as the primary completion date.
Registry status
The trial is listed as COMPLETED, and the registry indicates that results have been posted.
12. Design Topics Supported by the Registry
| Design topic | Evidence in the ClinicalTrials.gov record | Statistical significance |
|---|---|---|
| Randomization | Allocation is RANDOMIZED. | Supports an intended randomized comparison between saxagliptin and placebo. |
| Parallel design | Design model is PARALLEL. | Participants are evaluated within parallel intervention groups rather than a crossover structure. |
| Blinding | Masking is QUADRUPLE. | Indicates extensive masking; the ClinicalTrials.gov record does not identify the four masked roles. |
| Factorial design | Not identified in the ClinicalTrials.gov record. | No factorial analysis is inferred. |
| Crossover | Not identified in the ClinicalTrials.gov record. | No crossover effect is inferred. |
| Non-inferiority margin | Not identified in the ClinicalTrials.gov record. | No non-inferiority analysis is inferred. |
| Interim analysis | Not identified in the ClinicalTrials.gov record. | No interim-analysis procedure is inferred. |
| Missing-data imputation | Not identified in the ClinicalTrials.gov record. | No imputation method is inferred. |
| Bayesian methods | Not identified in the ClinicalTrials.gov record. | No Bayesian analysis is inferred. |
| Stratification | Not identified in the ClinicalTrials.gov record. | No stratification variables are inferred. |
This distinction is important in a trial-analysis page. A statistical method should be attributed to the trial only when the ClinicalTrials.gov record supports it. Familiarity with a trial or with common cardiovascular-outcomes-trial practice is not sufficient evidence to add a method that is absent from the ClinicalTrials.gov record.
13. Multiplicity and the Primary Endpoint
the ClinicalTrials.gov record identifies 1 primary endpoint. That endpoint is the binary composite of cardiovascular death, non-fatal myocardial infarction, or non-fatal ischaemic stroke.
The ClinicalTrials.gov record therefore presents a single registered primary endpoint rather than a set of multiple primary endpoints.
Multiplicity can still arise from secondary outcomes, subgroup analyses, repeated looks at accumulating data, or multiple statistical comparisons. The ClinicalTrials.gov recordset does not identify such procedures, so no specific multiplicity-adjustment method is attributed to SAVOR-TIMI 53 here.
14. Analysis Population and Randomization
The trial enrolled 18,206 participants and used randomized allocation. For a randomized primary-efficacy comparison, preserving treatment assignment in the analysis is generally important because it retains the comparison created by randomization.
However, the ClinicalTrials.gov record does not explicitly identify an intention-to-treat population, per-protocol population, as-treated population, or modified intention-to-treat population. These analysis-population labels are therefore not assigned to SAVOR-TIMI 53 on this page.
What is known
The trial was randomized, parallel, quadruple-masked, and enrolled 18,206 participants.
What is not reported
The ClinicalTrials.gov record does not define the exact population used for the formal primary statistical analysis.
Why this matters
Different analysis populations can answer subtly different questions, particularly when participants discontinue treatment or have incomplete follow-up.
How to avoid overinterpretation
The absence of a named analysis population in the ClinicalTrials.gov record should not be filled with an assumed convention.
15. Understanding the Follow-Up Window
The primary endpoint is evaluated from randomization (day 0) up to 2.9 years. This wording defines the registered time frame for the primary outcome.
It is useful to distinguish this fixed registered window from a generic statement that participants were simply “followed for 2.9 years.” The endpoint definition explicitly starts the observation period at randomization and specifies a maximum endpoint time frame of 2.9 years.
| Time element | Registered information |
|---|---|
| Time origin | Randomization (day 0) |
| Primary endpoint window | Up to 2.9 years |
| No-event censoring | Withdrawal of consent, last contact, or death when applicable — whichever was later |
16. Why This Trial Matters Statistically
SAVOR-TIMI 53 provides a useful teaching example because its registry record combines a large randomized parallel-group design with a clinically important composite cardiovascular endpoint and an explicit follow-up window.
| Concept | How it appears in SAVOR-TIMI 53 |
|---|---|
| Randomization | The allocation is registered as RANDOMIZED. |
| Parallel-group design | The design model is PARALLEL. |
| Masking | The trial is registered as QUADRUPLE-masked. |
| Large trial population | 18,206 participants were enrolled. |
| Binary endpoint | The primary endpoint is classified as Binary. |
| Composite endpoint | CV death, non-fatal MI, or non-fatal ischaemic stroke comprise the registered primary outcome. |
| Fixed follow-up | The registered primary endpoint is assessed from randomization through 2.9 years. |
| Censoring | The registry specifies how participants without an event are censored. |
| Safety denominators | Serious adverse events are reported as affected participants over participants at risk. |
| Results reporting | The registry indicates that results are posted, but the registry-reported statistical-analysis array is empty. |
The trial is also a useful reminder that availability of posted results and availability of reproducible statistical-analysis data are not the same thing. A rigorous statistical review should distinguish between what the registry says exists and what numerical evidence is actually present in the ClinicalTrials.gov record.
17. Limitations
- No formal statistical analyses reported: no formal statistical analyses were posted, so numerical effect estimates, confidence intervals, and p-values for the primary endpoint cannot be reproduced from the ClinicalTrials.gov record.
- No primary event counts reported: the ClinicalTrials.gov record does not provide the number of participants experiencing the primary composite endpoint in either arm.
- No component-level results reported: there are no separate event counts or treatment effects for cardiovascular death, non-fatal myocardial infarction, or non-fatal ischaemic stroke.
- Arm-specific enrollment is not reported: although total enrollment is 18,206 and there are two arms, the ClinicalTrials.gov record does not give the randomized number in each arm.
- No statistical-analysis population is reported: the record does not specify an ITT, per-protocol, as-treated, or modified ITT population for the primary analysis.
- No analysis model is reported: no specific risk ratio, odds ratio, risk difference, logistic model, survival model, or other formal analysis is identified in the registry-reported statistical-analysis field.
- No multiplicity procedure is reported: the record does not identify alpha allocation, multiplicity adjustment, or hierarchical testing.
- No interim-analysis information is reported: an interim monitoring plan cannot be inferred from the presence of a large cardiovascular trial.
- No missing-data method is reported: although the registry provides a censoring rule, it does not provide an imputation strategy in the ClinicalTrials.gov record.
- No subgroup estimates are reported: treatment effects across demographic or clinical subgroups cannot be evaluated from the ClinicalTrials.gov record.
18. What a Complete Primary-Endpoint Result Would Normally Contain
For a binary composite endpoint, a complete statistical results table would ordinarily allow a reader to see the number of participants with an event in each randomized group, the corresponding event proportions, a comparative effect measure, its confidence interval, and a hypothesis-test result where appropriate.
| Result element | Why it matters | registry here? |
|---|---|---|
| Events in saxagliptin arm | Describes the observed primary endpoint burden. | No |
| Events in placebo arm | Provides the comparator event burden. | No |
| Event proportion by arm | Places event counts in the context of the number at risk. | No |
| Effect estimate | Quantifies the between-arm difference or relative effect. | No |
| Confidence interval | Quantifies statistical uncertainty around the effect estimate. | No |
| P-value | Provides evidence against a specified null hypothesis under the chosen test. | No |
| Component results | Shows how the composite is constructed in practice. | No |
This is why the absence of a formal statistical-analysis entry materially limits what can responsibly be concluded from the ClinicalTrials.gov recordset. The trial's design and endpoint can be explained in detail without fabricating the missing inferential results.
19. Interpreting P-values and Confidence Intervals
A p-value would quantify how incompatible the observed data are with a specified null hypothesis under the chosen statistical model. It would not measure the size or clinical importance of the treatment effect. Because no primary p-value is reported, no p-value is reported here.
A confidence interval would describe statistical uncertainty around a specified effect estimate. A narrow interval would indicate greater statistical precision than a wide interval under the same inferential framework. No primary confidence interval is reported in the provided statistical-analysis data.
The effect size is the quantity that directly describes the magnitude of the between-group difference or relative association. A p-value cannot substitute for an effect estimate, and neither can be reconstructed when the underlying event counts and analysis results are absent.
20. Composite Endpoints and Clinical Interpretation
The primary endpoint combines three event types that are clinically meaningful but not necessarily identical in severity or frequency. This has several implications for interpretation.
Statistical efficiency
A composite can accumulate events across several predefined outcomes, potentially providing more endpoint events than any one component would provide alone.
Clinical meaning
The composite is a combined outcome. Its interpretation should not automatically be extended to each individual component.
Component balance
The relative contribution of cardiovascular death, myocardial infarction, and ischaemic stroke matters when interpreting a composite result.
What is missing here
The ClinicalTrials.gov record contains no component-specific event counts or estimates, so the contribution of each component cannot be assessed.
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Statistical Calculators
23. Sources
- ClinicalTrials.gov: SAVOR-TIMI 53 — NCT01107886.
- Linked publication: PubMed record — PMID 33832610.
- Linked publication: PubMed record — PMID 31474116.
- Linked publication: PubMed record — PMID 31362530.
- Linked publication: PubMed record — PMID 29214305.
- Linked publication: PubMed record — PMID 28417296.
Continue through the Clinical Biostats statistical pathway
Explore clinical-trial methods, endpoint definitions, and statistical calculators that build the concepts used in randomized trial analysis.
24. Record Summary
SAVOR-TIMI 53 is a completed randomized, parallel-group, quadruple-masked phase 4 trial enrolling 18,206 participants with type 2 diabetes mellitus. The registered primary endpoint is a binary composite of cardiovascular death, non-fatal myocardial infarction, or non-fatal ischaemic stroke, assessed from randomization (day 0) up to 2.9 years. The registry specifies censoring for participants without an event at withdrawal of consent, last contact, or death when applicable, whichever was later.
The ClinicalTrials.gov record also report serious adverse events for the two arms: 2,075 of 8,212 participants in the placebo group and 2,114 of 8,280 participants in the saxagliptin group. These counts provide descriptive safety information but do not constitute a formal comparative analysis.
Most importantly for statistical interpretation, the ClinicalTrials.gov record says that results have been posted but contains no formal statistical analyses. A rigorous analysis therefore distinguishes the trial's documented design and endpoint from numerical efficacy conclusions that are not present in the ClinicalTrials.gov record. This preserves the central principle of clinical-trial statistics: reported evidence should be separated from assumptions about what the analysis might have shown.