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
ATLANTIC was a randomized, parallel, triple-masked phase 4 study evaluating pre-hospital versus in-hospital initiation of ticagrelor in STEMI patients planned for PCI. The registry reports 1875 enrolled participants, two co-primary binary endpoints, logistic-regression analyses, and odds ratios as the principal effect measure.
| Feature | ATLANTIC |
|---|---|
| Trial name | ATLANTIC |
| ClinicalTrials.gov identifier | NCT01347580 |
| Phase | Phase 4 |
| Status | Completed |
| Start | 2011-09 |
| Primary completion | 2013-11 |
| Condition | Myocardial Infarction; Segment Elevation Myocardial Infarction (STEMI) |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Triple |
| Primary purpose | Treatment |
| Lead sponsor | AstraZeneca |
| Sponsor type | Industry |
| Enrollment | 1875 |
| Interventions | Ticagrelor; Placebo |
| Registered primary endpoints | 2 |
| Posted outcome measures | 12 |
| Posted statistical analyses | 8 |
2. Clinical Question
The clinical question was whether initiating ticagrelor before hospital arrival, rather than initiating it after arrival at the hospital, was associated with different early measures of coronary reperfusion and ST-segment resolution in STEMI patients planned for PCI.
Population
STEMI patients planned for percutaneous coronary intervention, as described by the registered trial title and condition.
Intervention
Pre-hospital initiation of ticagrelor.
Comparator
In-hospital initiation of ticagrelor.
Primary question
Does pre-hospital initiation differ from in-hospital initiation with respect to TIMI flow grade 3 at initial angiography and ST-segment elevation resolution of at least 70%?
The registry classifies the primary hypothesis type as superiority. This is important statistically: the analysis asks whether the two randomized strategies differ in the direction represented by the odds ratio, rather than testing whether one strategy remains within a prespecified non-inferiority margin.
3. Trial Design
Pre-hospital ticagrelor
- Ticagrelor initiated before hospital arrival.
- Patients were planned for PCI.
- Primary angiographic and electrocardiographic endpoints were assessed around PCI.
In-hospital ticagrelor
- Ticagrelor initiated in hospital.
- Patients were planned for PCI.
- The same registered primary endpoints were used for comparison.
The registry describes the study as triple-masked. Because the treatment strategies concern the timing of drug initiation in an urgent clinical setting, masking is a design feature worth identifying separately from randomization: randomization addresses allocation to strategy, while masking is intended to limit knowledge of assignment when outcomes are assessed or care is delivered.
4. Endpoints
ATLANTIC has two registered co-primary endpoints. Both are binary outcomes, which explains why logistic regression and odds ratios are central to the posted statistical analyses.
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Thrombolysis In Myocardial Infarction (TIMI) Flow Grade 3 of MI Culprit Vessel at Initial Angiography (Co-primary Endpoint) | (TIMI) flow grade classification is used to assess coronary blood flow in acute coronary syndromes. Grade 0: no reperfusion, grade 1: penetration without perfusion, grade 2: partial reperfusion, grade 3: complete perfusion. | At initial angiography, pre PCI | Other / unclear; analysed as binary |
| ST-segment Elevation Resolution Pre PCI ≥70% (Co-primary Endpoint) | ST segment elevation resolution is the mean ST elevation pre-hospital minus the mean ST elevation pre-PCI divided by the mean ST elevation pre-hospital. It is expressed as a percentage and split in 2 categories, complete (≥70%) versus incomplete (<70%) resolution. | Between baseline and PCI | Binary |
The two primary endpoints represent different aspects of early reperfusion. TIMI flow grade is an angiographic classification of coronary blood flow, whereas ST-segment resolution converts a continuous percentage measure into two categories: complete resolution at ≥70% versus incomplete resolution at <70%.
5. Statistical Methodology
Why logistic regression was used
All eight posted statistical analyses use logistic regression. This is appropriate for the binary structure of the reported outcomes: each patient contributes an outcome category rather than a continuous measurement or a time-to-event observation.
For a binary endpoint, the treatment indicator can be represented by X. The exponentiated treatment coefficient, exp(β1), corresponds to an odds ratio under the fitted model.
Odds ratio as the effect measure
The registry reports odds ratios for all eight posted analyses. An odds ratio compares the odds of the specified binary outcome between the groups represented in the analysis. An odds ratio of 1 corresponds to equal odds; values above 1 indicate higher estimated odds in the numerator group, while values below 1 indicate lower estimated odds.
An odds ratio is not itself a risk ratio or an absolute risk difference. The numerical distinction becomes especially important when the outcome is not rare.
Analysis populations
The two primary endpoint analyses were conducted in the modified intention-to-treat (mITT) population, specifically described as patients with non-missing values. The secondary analyses were generally reported in the mITT population, with some analyses additionally restricting to patients with non-missing TIMI flow or ECG values.
| Endpoint category | Analysis population reported |
|---|---|
| Primary TIMI flow grade 3 | mITT, on patients with non-missing values |
| Primary ST-segment resolution ≥70% | mITT, on patients with non-missing values |
| 30-day composite endpoints | mITT |
| Definite stent thrombosis | mITT |
| Post-PCI TIMI flow grade 3 | mITT, on patients with non-missing TIMI flow grade values |
| Post-PCI ST-segment resolution ≥70% | mITT on patients with non-missing ECG values |
| Thrombotic bail-out with GPIIb/IIIa inhibitors | mITT |
Missing data
The registry analysis records for both primary endpoints identify multiple imputation / missing data as an additional concept. However, the ClinicalTrials.gov record does not specify the imputation model, variables included in the imputation model, number of imputations, or the exact way the imputed analyses were combined. Those details therefore should not be inferred.
6. Primary Results: TIMI Flow Grade 3
The first co-primary endpoint was TIMI flow grade 3 of the MI culprit vessel at initial angiography, pre PCI. The posted analysis used logistic regression in the mITT population among patients with non-missing values.
Odds ratio for TIMI flow grade 3
95% CI: 0.799–1.327 · P = 0.8214
Pre-hospital ticagrelor vs in-hospital ticagrelor
| Feature | Reported analysis |
|---|---|
| Endpoint | Thrombolysis In Myocardial Infarction (TIMI) Flow Grade 3 of MI Culprit Vessel at Initial Angiography (Co-primary Endpoint) |
| Time frame | At initial angiography, pre PCI |
| Analysis population | mITT, on patients with non missing values |
| Method | Logistic regression |
| Effect measure | Odds Ratio (OR) |
| Estimate | 1.030 |
| 95% CI | 0.799–1.327 |
| P-value | 0.8214 |
| Hypothesis type | Superiority |
The estimated odds ratio of 1.030 is very close to 1. In the fitted logistic-regression comparison, the estimated odds of the binary TIMI flow grade 3 outcome were therefore close to those represented by the reference group.
The 95% confidence interval of 0.799–1.327 is substantially wider than a single point estimate and includes 1. This means the reported data are compatible with a range of odds ratios on both sides of the null value of 1. The interval describes uncertainty around the estimated odds ratio; it does not describe the range of outcomes that individual patients could experience.
The P-value of 0.8214 is a measure of evidence against the null hypothesis used for the statistical test. It is not a measure of the size or clinical importance of the effect. A P-value should therefore not be interpreted as the probability that the treatment strategies are equivalent, nor as the probability that the observed estimate occurred by chance.
Because this is a logistic-regression analysis of a binary endpoint, proportional-hazards assumptions are not relevant to this estimate. The more relevant considerations are the binary outcome definition, the analysis population, missing observations, model specification, and the interpretation of an odds ratio rather than a risk ratio.
7. Primary Results: ST-Segment Elevation Resolution ≥70%
The second co-primary endpoint was complete ST-segment elevation resolution, defined in the registry as resolution of at least 70% between baseline and PCI. The analysis again used logistic regression in the mITT population among patients with non-missing values.
Odds ratio for ST-segment resolution ≥70%
95% CI: 0.801–1.441 · P = 0.6322
Pre-hospital ticagrelor vs in-hospital ticagrelor
| Feature | Reported analysis |
|---|---|
| Endpoint | ST-segment Elevation Resolution Pre PCI ≥70% (Co-primary Endpoint) |
| Time frame | Between baseline and PCI |
| Analysis population | mITT, on patients with non missing values |
| Method | Logistic regression |
| Effect measure | Odds Ratio (OR) |
| Estimate | 1.074 |
| 95% CI | 0.801–1.441 |
| P-value | 0.6322 |
| Hypothesis type | Superiority |
The odds ratio of 1.074 is modestly above 1. On the scale of the fitted logistic model, this is an estimated increase in the odds of the defined ≥70% ST-segment resolution outcome for the group represented in the numerator of the reported comparison.
The 95% CI of 0.801–1.441 includes 1. Consequently, the point estimate should not be treated as establishing a difference between the randomized strategies. The interval also demonstrates why the point estimate alone is insufficient: plausible values extend below and above the null value.
The P-value of 0.6322 does not quantify the magnitude of the treatment effect. It summarizes the statistical evidence under the specified testing framework. It should not be converted into a probability that the null hypothesis is true or that one strategy has no clinical effect.
As with the TIMI endpoint, this is not a time-to-event analysis, so a proportional-hazards assumption does not apply. The main statistical issues instead concern the binary threshold, the analysis population, missing ECG information, and the interpretation of the odds ratio.
8. What the Two Co-Primary Results Mean Together
The two co-primary analyses provide complementary measurements of early reperfusion. The first concerns angiographic TIMI flow grade 3 at initial angiography; the second concerns the binary threshold of at least 70% ST-segment elevation resolution between baseline and PCI.
| Co-primary endpoint | OR | 95% CI | P-value | Statistical scale |
|---|---|---|---|---|
| TIMI Flow Grade 3 at initial angiography, pre PCI | 1.030 | 0.799–1.327 | 0.8214 | Odds ratio |
| ST-segment elevation resolution pre PCI ≥70% | 1.074 | 0.801–1.441 | 0.6322 | Odds ratio |
Both point estimates are close to 1, and both 95% confidence intervals include 1. The registry classifies both analyses as superiority tests. Taken together, the posted analyses do not provide a basis for describing a statistically demonstrated difference in either co-primary binary endpoint from the reported estimates alone.
9. Secondary Endpoint Results
The registry posts six secondary statistical analyses. Each uses logistic regression and an odds ratio. The secondary analyses cover 30-day clinical composite endpoints, definite stent thrombosis, post-PCI TIMI flow, post-PCI ST-segment resolution, and thrombotic bail-out with GPIIb/IIIa inhibitors.
| Secondary endpoint | Time frame | OR | 95% CI | P-value |
|---|---|---|---|---|
| 1st Composite Clinical Endpoint | during the 30 days of treatment | 1.027 | 0.661–1.595 | 0.9056 |
| 2nd Composite Clinical Endpoint | within 30 days of study | 1.215 | 0.760–1.942 | 0.4168 |
| Definite Stent Thrombosis | during 30 days of treatment | 0.189 | 0.042–0.856 | 0.0307 |
| TIMI Flow Grade 3 Post-PCI | at coroangiography post-PCI | 1.132 | 0.876–1.462 | 0.344 |
| ST Segment Elevation Resolution Post-PCI ≥70% | Between baseline and ECG 60 mn post-PCI | 1.225 | 0.996–1.506 | 0.0547 |
| Thrombotic Bail-out With GPIIb/IIIa Inhibitors at Initial PCI | during PCI | 0.803 | 0.588–1.096 | 0.1660 |
Definite stent thrombosis
Reported odds ratio
95% CI: 0.042–0.856 · P = 0.0307
Pre-hospital ticagrelor vs in-hospital ticagrelor
The posted secondary analysis estimates an odds ratio of 0.189 for definite stent thrombosis during 30 days of treatment. An odds ratio of 0.189 corresponds to substantially lower estimated odds in the numerator group relative to the reference group; expressed descriptively, 0.189 is 81.1% below an odds ratio of 1. The 95% confidence interval, 0.042–0.856, lies below 1, and the reported P-value is 0.0307.
This is a secondary endpoint, not one of the two co-primary endpoints. The estimate is therefore important to report but should not be interpreted in isolation from the trial's endpoint hierarchy and multiplicity considerations.
The odds ratio is not the same as an 81.1% reduction in absolute event probability. The percentage comparison above describes the location of the odds ratio relative to 1; it does not convert the result into a risk difference or relative risk.
The confidence interval indicates substantial uncertainty around the point estimate. Its lower bound is 0.042 and its upper bound is 0.856, so the estimated magnitude is not known precisely from the confidence interval alone.
The registry caveat states that no prespecified hypothesis and procedure for adjustment was made on secondary endpoints. That is especially relevant when interpreting a nominal P-value for one secondary outcome among multiple analyses.
Post-PCI ST-segment resolution
The registry analysis record lists the group compared as “Pre-hospital Ticagrelor” and uses logistic regression in the mITT population with non-missing ECG values.
The confidence interval comes very close to 1 at its lower boundary but includes 1. The P-value is 0.0547. This is a useful example of why statistical interpretation should rely on the full estimate and confidence interval rather than applying an automatic interpretation to a P-value close to a conventional threshold.
10. Secondary Endpoint Analysis in Context
The six secondary estimates span both directions of the odds-ratio scale. Some estimates are below 1 and some are above 1. This is not surprising when multiple clinical and procedural outcomes are examined, but it creates an important interpretive distinction: the secondary analyses are supportive and descriptive within the ClinicalTrials.gov record, rather than a collection of independently confirmatory tests.
| Secondary endpoint | Interpretive feature |
|---|---|
| 1st Composite Clinical Endpoint | OR 1.027; CI includes 1; P = 0.9056 |
| 2nd Composite Clinical Endpoint | OR 1.215; CI includes 1; P = 0.4168 |
| Definite Stent Thrombosis | OR 0.189; CI below 1; P = 0.0307 |
| TIMI Flow Grade 3 Post-PCI | OR 1.132; CI includes 1; P = 0.344 |
| ST Segment Elevation Resolution Post-PCI ≥70% | OR 1.225; CI includes 1; P = 0.0547 |
| Thrombotic Bail-out With GPIIb/IIIa Inhibitors | OR 0.803; CI includes 1; P = 0.1660 |
The registry explicitly notes that no prespecified hypothesis and procedure for adjustment was made on secondary endpoints. Consequently, the individual P-values should not be treated as though each secondary endpoint had its own protected confirmatory type-I-error budget.
11. Safety Results
The ClinicalTrials.gov record reports serious adverse events by randomized arm. These are presented as affected patients over patients at risk.
| Safety measure | Affected / at risk |
|---|---|
| Ticagrelor In-Hosp | 143 / 950 |
| Ticagrelor Pre-Hosp | 140 / 908 |
Ticagrelor In-Hosp
Serious adverse events were reported in 143 of 950 patients at risk.
Ticagrelor Pre-Hosp
Serious adverse events were reported in 140 of 908 patients at risk.
The ClinicalTrials.gov record does not provide a formal statistical comparison, confidence interval, or P-value for serious adverse events by arm. Accordingly, the safety result is reported descriptively rather than converted into an unreported comparative effect estimate.
12. Statistical Methods Explained
Why was logistic regression used?
The primary endpoints were binary outcomes: TIMI flow grade 3 versus other grades, and complete ST-segment elevation resolution ≥70% versus incomplete resolution <70%. Logistic regression is designed to model the probability of a binary outcome and expresses the treatment contrast through an odds ratio.
What does an odds ratio of 1.030 mean?
An odds ratio of 1.030 means that the estimated odds ratio is close to the null value of 1. It does not mean that the treatment effect is 3.0% better, because an odds ratio is a ratio of odds rather than a percentage change in probability. To translate an odds ratio into absolute probabilities, the underlying baseline probability and model specification are also needed.
Why does the confidence interval matter?
The point estimate is only one estimate from the data. The 95% confidence interval communicates statistical uncertainty around that estimate under the analysis framework. For the two co-primary endpoints, the intervals are 0.799–1.327 and 0.801–1.441, respectively, and both include 1.
Why doesn't the P-value measure effect size?
A P-value measures the compatibility of the observed result with the statistical null hypothesis under the specified testing framework. It does not tell us how large the treatment effect is. Effect size is represented here by the odds ratio, while precision is represented by the confidence interval.
Why does the mITT population matter?
Randomized trials derive much of their inferential strength from randomization. An mITT analysis can differ from a pure intention-to-treat analysis when some randomized patients are excluded from the analysis population. Here, the primary analyses are explicitly described as mITT among patients with non-missing values. That means the reader should not automatically describe these posted estimates as analyses of all 1875 enrolled participants.
How should the missing-data information be interpreted?
The registry associates the primary analyses with multiple imputation / missing data, but the ClinicalTrials.gov record does not specify the imputation model or implementation. At the same time, the stated analysis population is limited to patients with non-missing values. A careful statistical review therefore distinguishes the registry's identification of missing-data methodology from the details required to reproduce that methodology.
Why is multiplicity important for the secondary results?
The trial reports two co-primary endpoints and six secondary statistical analyses in the ClinicalTrials.gov record. The registry specifically states that no prespecified hypothesis and procedure for adjustment was made on secondary endpoints. A nominal P-value of 0.0307 for a secondary endpoint therefore should not automatically be interpreted as if it came from a separately protected confirmatory test.
13. Understanding Odds Ratios in This Trial
Odds ratios are especially important in ATLANTIC because they are the common effect measure across the posted statistical analyses. The odds scale is related to probability, but it is not the same scale.
The odds ratio compares these odds between groups. A risk ratio instead compares probabilities directly, so the two effect measures should not be substituted for one another.
OR = 1
The estimated odds are equal between the compared groups on the model's odds scale.
OR > 1
The estimated odds of the specified outcome are higher in the numerator group.
OR < 1
The estimated odds of the specified outcome are lower in the numerator group.
CI crossing 1
The confidence interval includes the null odds ratio, so the interval alone does not establish a difference under the corresponding two-sided framework.
For example, the definite stent thrombosis analysis reports an OR of 0.189, while the first composite clinical endpoint reports an OR of 1.027. These values cannot be compared as though they were percentages of patients because they describe different binary outcomes on an odds-ratio scale.
14. Randomization and Masking
Randomization is a foundational part of the trial design because treatment assignment is randomized rather than selected by investigators or participants. In principle, this supports comparison of outcome distributions between treatment strategies by reducing systematic differences in treatment assignment.
The registry also identifies the study as triple masked. Masking and randomization solve different problems: randomization concerns how treatment assignment is generated, while masking concerns who knows the assignment during the conduct and assessment of the study.
| Design feature | ATLANTIC | Statistical role |
|---|---|---|
| Allocation | Randomized | Creates the randomized treatment comparison. |
| Design model | Parallel | Patients remain assigned to one of the two parallel strategies. |
| Masking | Triple | Limits knowledge of treatment assignment during the trial. |
| Hypothesis type | Superiority | Tests for evidence of a difference rather than non-inferiority within a margin. |
15. What This Trial Does Not Establish
Several statistical conclusions would require information that is not contained in the ClinicalTrials.gov record. Keeping these distinctions explicit prevents an analysis page from turning a reported estimate into a stronger claim than the underlying record supports.
- No non-inferiority conclusion: the registry identifies superiority hypotheses and provides no non-inferiority margin.
- No proportional-hazards issue: the posted analyses are logistic regressions of binary outcomes, not Cox proportional-hazards models.
- No survival endpoint interpretation: the statistical analyses posted on ClinicalTrials.gov do not report a time-to-event treatment effect.
- No subgroup conclusions: the ClinicalTrials.gov record does not provide subgroup estimates.
- No formal safety comparison: serious adverse-event counts are reported, but no comparative safety estimate is reported.
- No reconstructed absolute risk estimates: the registry provides odds ratios but not the corresponding event probabilities needed to calculate absolute risk differences.
- No detailed imputation reconstruction: multiple imputation / missing data is identified as an analysis concept, but the ClinicalTrials.gov record does not provide the implementation details.
- No multiplicity-adjusted secondary inference: the registry explicitly states that no prespecified hypothesis and procedure for adjustment was made on secondary endpoints.
16. Limitations
Urgent clinical setting
The registry notes that 8.6% of patients with symptoms of MI in the ambulance ultimately did not have a STEMI diagnosis in the cathlab. This is a specific limitation of an urgent pre-hospital study population.
Secondary endpoint multiplicity
No prespecified hypothesis and procedure for adjustment was made on secondary endpoints, according to the registry caveat.
mITT analysis populations
The primary analyses were performed in mITT patients with non-missing values rather than simply treating all 1875 enrolled participants as the analysis population.
Missing-data detail
The registry associates the primary analyses with multiple imputation / missing data but does not provide enough detail in the ClinicalTrials.gov record to reconstruct the procedure.
The 8.6% figure is particularly relevant to interpretation because the trial operated in an urgent setting in which diagnosis was not necessarily definitive at the pre-hospital stage. The registry caveat therefore identifies a real distinction between the population intended for the intervention comparison and the population ultimately confirmed to have STEMI in the cathlab.
A second limitation is statistical rather than clinical: the trial has two co-primary endpoints plus multiple secondary outcomes. The registry's explicit statement that no prespecified adjustment procedure was made for secondary endpoints means the secondary P-values require more cautious interpretation than a single prespecified primary hypothesis test.
17. Why This Trial Matters Statistically
ATLANTIC is a useful teaching example because it demonstrates how the statistical method should follow the structure of the endpoint. Unlike a survival trial, its principal outcomes are binary and occur around a defined clinical procedure. The appropriate statistical language therefore centers on logistic regression, odds ratios, confidence intervals, missing-data handling, and multiplicity rather than Kaplan-Meier curves or hazard ratios.
| Concept | How it appears in ATLANTIC |
|---|---|
| Randomization | Randomized allocation to pre-hospital versus in-hospital ticagrelor initiation. |
| Parallel design | Two parallel treatment strategies. |
| Masking | Triple masking. |
| Binary endpoints | TIMI flow grade 3 and ST-segment resolution ≥70% are analysed as binary outcomes. |
| Logistic regression | Method used for all eight posted statistical analyses. |
| Odds ratio | Common effect measure for all posted statistical analyses. |
| Confidence intervals | 95% two-sided intervals are reported for all eight posted odds-ratio analyses. |
| P-values | Reported for both co-primary endpoints and the six posted secondary analyses. |
| Missing data | Multiple imputation / missing data is identified in the analysis text for the primary endpoints. |
| mITT analysis | Primary analyses use mITT patients with non-missing values. |
| Multiplicity | Two co-primary endpoints and multiple secondary analyses require separate interpretation. |
| Safety analysis | Serious adverse events are reported descriptively by arm. |
18. A Worked Reading of the Primary Statistical Results
A disciplined interpretation of a trial result can be performed in a fixed sequence. ATLANTIC provides a particularly clear example because both primary outcomes use the same statistical framework.
The first outcome is TIMI flow grade 3 at initial angiography, pre PCI. The second is ST-segment elevation resolution ≥70% between baseline and PCI.
Both are analysed using logistic regression in the mITT population among patients with non-missing values.
The treatment effect is reported as an odds ratio. The null value for an odds ratio is 1.
The TIMI flow confidence interval is 0.799–1.327, while the ST-segment resolution confidence interval is 0.801–1.441. Both include 1.
The corresponding P-values are 0.8214 and 0.6322. These are statistical-test measures, not effect-size measures.
These were superiority analyses of co-primary endpoints. Therefore, the results should be interpreted as evidence from the prespecified randomized comparison rather than as a non-inferiority or equivalence demonstration.
19. Related Tutorials
Learn more about the methods used in this trial:
20. Related Calculators
21. Sources
- ClinicalTrials.gov: ATLANTIC, NCT01347580. Official registry record containing the ClinicalTrials.gov record, registered endpoints, posted outcome measures, statistical analyses, and registry limitations.
- PubMed: PMID 25175921.
- PubMed: PMID 30353444.
- PubMed: PMID 30302940.
- PubMed: PMID 29421015.
- PubMed: PMID 29421001.
The numerical results presented on this page are restricted to the registry-reported ATLANTIC trial data. No additional numerical trial results have been added from external publications or from general knowledge of the study.
Continue through the Clinical Biostats learning pathway
Use the related tutorials and calculators to explore the statistical concepts represented by this randomized clinical trial.
22. Record Summary
ATLANTIC illustrates a randomized clinical-trial analysis in which the statistical method follows the endpoint structure. The two co-primary outcomes are binary measures of early reperfusion, and the registry reports logistic regression with odds ratios, two-sided 95% confidence intervals, and P-values. The primary TIMI flow analysis produced an OR of 1.030 with a 95% CI of 0.799–1.327 and P = 0.8214. The primary ST-segment resolution analysis produced an OR of 1.074 with a 95% CI of 0.801–1.441 and P = 0.6322.
The secondary analyses show why interpretation should extend beyond isolated P-values. The definite stent thrombosis analysis reported an OR of 0.189, 95% CI 0.042–0.856, and P = 0.0307, while the other posted secondary estimates had confidence intervals that included 1. The registry explicitly notes that no prespecified hypothesis and procedure for adjustment was made on secondary endpoints, making multiplicity an important part of the interpretation.
Finally, the registry identifies several design and analysis features that matter when reading the results: randomized parallel allocation, triple masking, mITT analysis populations, non-missing-value restrictions, multiple imputation / missing-data concepts, and descriptive serious-adverse-event counts by arm. The urgent clinical setting also produced a registry-documented diagnostic limitation, with 8.6% of patients with symptoms of MI in the ambulance ultimately not having a STEMI diagnosis in the cathlab.