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
SPIRE-2 was a randomized, parallel-group, quadruple-masked phase 3 prevention trial comparing bococizumab with placebo in participants at high cardiovascular risk. Its primary endpoint was the first occurrence of an adjudicated major cardiovascular event, and the trial ended early, with status recorded as terminated.
| Feature | SPIRE-2 |
|---|---|
| Full title | The Evaluation of Bococizumab (PF-04950615; RN316) in Reducing the Occurrence of Major Cardiovascular Events in High Risk Subjects |
| Phase | Phase 3 |
| Condition | Cardiovascular Disease |
| Primary purpose | Prevention |
| Design | Randomized, parallel-group, quadruple-masked, placebo-controlled |
| Arms | 2: bococizumab (PF-04950615) and placebo |
| Enrollment | 10,564 |
| Primary endpoint | Event rate per 100 participant-years for first occurrence of major cardiovascular (CV) event |
| Primary effect measure | Hazard ratio from a stratified Cox proportional-hazards model; log-rank P-value |
| Status | Terminated; results posted |
| Dates | Start 2013-10-29; primary completion 2017-04-03 |
| ClinicalTrials.gov | NCT01975389 |
| Sponsor | Pfizer (industry) |
2. Clinical Question
The central question was whether adding bococizumab, compared with placebo, reduced the rate of first major cardiovascular events in high-risk participants. The trial also measured how strongly bococizumab changed LDL cholesterol and other lipid fractions, which allows the clinical-outcome comparison to be read alongside a biomarker comparison.
Population
High-risk participants in a cardiovascular disease prevention setting. Geographic region and complete statin intolerance were used as stratification factors in the analyses.
Intervention
Bococizumab (PF-04950615; RN316).
Comparator
Placebo, under quadruple masking.
Primary question
Does bococizumab lower the rate of first adjudicated major CV event (CV death, non-fatal MI, non-fatal stroke, or hospitalization for unstable angina needing urgent revascularization) relative to placebo?
3. Trial Design
Bococizumab
- Bococizumab (PF-04950615)
- Quadruple-masked administration
- Followed for adjudicated CV events
- Lipids and hs-CRP measured over time
Placebo
- Matching placebo
- Quadruple-masked administration
- Followed for adjudicated CV events
- Lipids and hs-CRP measured over time
Quadruple masking means that participants, care providers, investigators and outcomes assessors were all unaware of treatment assignment. For an endpoint built from clinical events such as hospitalization for unstable angina or revascularization, where clinical judgment enters into whether an event occurs and is recorded, masking and central adjudication are the main protections against differential ascertainment between arms.
4. Stratification and Analysis Populations
All time-to-event analyses used a Cox proportional-hazards model stratified by geographic region and complete statin intolerance, with treatment as a covariate. The lipid and biomarker models included the same two factors as fixed effects. Using the same factors in both places keeps the analyses aligned with the structure of the population and removes outcome variation that is explained by region or statin-intolerance status.
| Analysis population | Definition / role |
|---|---|
| Full analysis set (FAS) | All randomized participants, excluding those who attempted to be randomized more than once into a bococizumab CV outcomes trial (B1481022/B1481038) or into more than one CV outcomes trial, and all participants enrolled at study Site 3027, where a quality-related event was identified. Used for the cardiovascular outcome analyses. |
| FAS, evaluable participants | For lipid and hs-CRP endpoints, the analysis was performed on the FAS among participants evaluable for the specific outcome measure. |
| Safety (at risk) | 5276 participants in the bococizumab arm and 5279 in the placebo arm contributed to the serious adverse event summary. |
The FAS is close to, but not identical with, an intention-to-treat population. The exclusions are defined by events that are independent of the randomized treatment assignment itself: duplicate or multiple-trial randomization attempts, and a site-level quality problem. Exclusions of this kind are generally regarded as preserving the comparability that randomization creates, because they do not depend on how a participant responded to treatment. They still mean that the analyzed population is a defined subset of everyone randomized.
5. Endpoints
Primary endpoint
| Endpoint | Registry definition | Time frame |
|---|---|---|
| Event rate per 100 participant-years for first occurrence of major CV event | Event rate per 100 participant-years for first occurrence of major CV event (adjudicated by Adjudication Committee). Major CV event was defined as any of the following: CV death [defined as sudden cardiac death, fatal myocardial infarction (MI), death due to heart failure, death due to stroke (fatal ischemic stroke or fatal stroke of undetermined etiology), or death due to other CV causes], non-fatal MI, non-fatal stroke, and hospitalization for unstable angina needing urgent revascularization. Event rate was calculated as the number of events per 100 participant-years at risk. | From baseline until the date of first adjudicated and confirmed occurrence of major CV event (maximum duration: up to 3.4 years) |
Secondary endpoints with posted analyses
| Endpoint group | Endpoints | Time frame |
|---|---|---|
| Composite CV endpoints (event rate per 100 participant-years, first occurrence) | CV death, non-fatal MI or non-fatal stroke; all-cause death, non-fatal MI, non-fatal stroke or hospitalization for unstable angina needing urgent revascularization; all-cause death, non-fatal MI or non-fatal stroke; CV death, non-fatal MI, non-fatal stroke or hospitalization for unstable angina | From baseline until first adjudicated and confirmed occurrence (up to 3.4 years) |
| Component and related CV events | CV death; any MI (fatal or non-fatal); fatal MI; non-fatal MI; any stroke (fatal or non-fatal); any stroke of any etiology; non-fatal stroke; hospitalization for unstable angina (with and without the urgent-revascularization qualifier); hospitalization for congestive heart failure (CHF); coronary revascularization; CABG; PCI; any arterial revascularizations; all-cause death | From baseline until adjudicated and confirmed occurrence (up to 3.4 years) |
| LDL-C | Percent change and nominal change (mg/dL) from baseline at Week 14; percent change at last post-baseline measurement | Baseline, Week 14; baseline, last post-baseline measurement (any time up to Week 140) |
| Other lipids | Percent change from baseline in non-HDL-C, VLDL-C, RLP-C, Apo B, HDL-C, Apo A-I and total cholesterol; log-transformed triglycerides and Lp(a) | Baseline, Week 14 |
| Inflammation | Percent change from baseline in log-transformed high-sensitivity C-reactive protein (hs-CRP) | Baseline, Week 14 |
Every cardiovascular endpoint is labeled in the registry as an event rate per 100 participant-years, but the between-arm comparison for each is posted as a hazard ratio. These are related but distinct quantities. A crude event rate divides the number of first events by total participant-years at risk; a hazard ratio compares instantaneous event rates between arms under a proportional-hazards model and, here, within strata. Section 12 returns to why the two can differ.
6. Primary Endpoint Results
For the primary endpoint, the stratified Cox model produced a hazard ratio below 1, and the log-rank P-value was below the conventional two-sided 0.05 threshold. The registry lists the groups compared as placebo versus bococizumab; the hazard ratios are read here in the direction bococizumab relative to placebo, the same direction in which the lipid differences are negative for bococizumab.
Hazard ratio for first major cardiovascular event
95% CI: 0.65–0.97 (two-sided) · P = 0.021469 (log-rank)
Full analysis set · Cox model stratified by geographic region and complete statin intolerance · Superiority hypothesis
What the estimate means. A hazard ratio of 0.79 means that, under the fitted stratified Cox model and over the follow-up available before termination, the estimated instantaneous rate of a first major cardiovascular event in the bococizumab arm was 79% of the rate in the placebo arm, a 21% lower estimated hazard.
What it does not mean. It does not mean that 21% of participants avoided an event, that the absolute risk fell by 21 percentage points, or that every participant's own risk was reduced by the same proportion. A relative hazard says nothing on its own about how many events were prevented; that depends on how common events were in the placebo arm, which the hazard ratio does not display.
What the confidence interval says. The 95% CI of 0.65 to 0.97 excludes 1, so the data are not compatible, at the 5% two-sided level, with no difference in hazard. The interval is nonetheless fairly wide: its upper end (0.97) corresponds to only a 3% lower hazard, while its lower end (0.65) corresponds to a 35% lower hazard. The trial therefore establishes direction with more confidence than magnitude.
Why the P-value is not the effect size. P = 0.021469 measures how surprising the observed separation would be if bococizumab had no effect on the hazard. It does not measure how large the effect is. A smaller trial with the same hazard ratio would have produced a larger P-value; a larger one, a smaller P-value.
Cautions. The P-value comes from a log-rank test and the interval from a Cox model; the two are closely related but not guaranteed to agree exactly at the margin. Both rely on proportional hazards to summarize the effect with one number; if the effect grew or faded over time, 0.79 is an average over the observed follow-up. Participants without an event were censored at the end of their follow-up, and early termination made that follow-up shorter than planned; censoring is assumed to be unrelated to event risk within strata. Finally, the analysis population is the FAS rather than every participant randomized.
7. Secondary Cardiovascular Endpoints
The registry posts a stratified Cox hazard ratio and log-rank P-value for each of the following secondary time-to-event endpoints. All share the same analysis population (FAS), model and stratification factors as the primary analysis.
| Endpoint (first occurrence unless noted) | HR | 95% CI | P-value |
|---|---|---|---|
| CV death, non-fatal MI or non-fatal stroke (maximum duration: up to 3.4 years) | 0.75 | 0.60–0.93 | 0.007597 |
| All-cause death, non-fatal MI, non-fatal stroke or hospitalization for unstable angina needing urgent revascularization | 0.82 | 0.68–0.99 | 0.035958 |
| All-cause death, non-fatal MI or non-fatal stroke (maximum duration: up to 3.4 years) | 0.78 | 0.64–0.95 | 0.015694 |
| CV death, non-fatal MI, non-fatal stroke or hospitalization for unstable angina | 0.79 | 0.65–0.96 | 0.018053 |
| CV death | 0.82 | 0.50–1.36 | 0.446033 |
| All-cause death | 0.91 | 0.63–1.32 | 0.626157 |
| Any MI (fatal or non-fatal) | 0.75 | 0.57–0.97 | 0.029977 |
| Fatal MI | 0.44 | 0.14–1.44 | 0.162615 |
| Non-fatal MI | 0.77 | 0.59–1.00 | 0.051534 |
| Any stroke (fatal or non-fatal) | 0.67 | 0.41–1.09 | 0.104998 |
| Any stroke (fatal or non-fatal), of any etiology | 0.78 | 0.50–1.24 | 0.294331 |
| Non-fatal stroke (maximum duration: up to 3.4 years) | 0.67 | 0.41–1.09 | 0.104998 |
| Hospitalization for unstable angina needing urgent revascularization | 0.95 | 0.62–1.46 | 0.814224 |
| Hospitalization for unstable angina | 0.90 | 0.60–1.34 | 0.601200 |
| Hospitalization for CHF | 1.09 | 0.72–1.67 | 0.678061 |
| Coronary revascularization | 0.77 | 0.63–0.94 | 0.010457 |
| CABG | 1.19 | 0.71–2.01 | 0.509847 |
| PCI | 0.72 | 0.58–0.90 | 0.002981 |
| Any arterial revascularizations | 0.95 | 0.70–1.30 | 0.748975 |
Reading the pattern rather than the individual P-values
The four composite endpoints all have hazard ratios between 0.75 and 0.82 with intervals that exclude 1. Because these composites share most of their component events with the primary endpoint, they are not independent confirmations; they are largely the same events counted under slightly different definitions. Their agreement shows that the primary result is not driven by the particular choice of components, but it adds less new evidence than four separate trials would.
The component endpoints behave as one would expect from splitting a composite into smaller pieces. Myocardial infarction (any MI, HR 0.75) and PCI (HR 0.72) contribute estimates in the same direction as the composite, with intervals that exclude 1. Rarer components, such as fatal MI (HR 0.44, 95% CI 0.14–1.44) and CV death (HR 0.82, 95% CI 0.50–1.36), have much wider intervals. Wide intervals of this kind indicate little information, not evidence of no effect.
Non-fatal MI is a useful teaching case: HR 0.77 with a 95% CI of 0.59 to 1.00 and P = 0.051534. Its interval just reaches 1 and its P-value lies just above 0.05. Treating it as "negative" while treating any MI (HR 0.75, P = 0.029977) as "positive" would be an over-reading of a threshold; the two estimates are nearly identical.
A few estimates lie above 1, notably hospitalization for CHF (HR 1.09) and CABG (HR 1.19). Both intervals are wide and include values well below and above 1, so they neither demonstrate harm nor exclude it. As reported in the registry, the any-stroke and non-fatal-stroke analyses have identical estimates, intervals and P-values (HR 0.67, 95% CI 0.41–1.09, P = 0.104998).
8. Lipid and Biomarker Endpoints
The lipid endpoints show the pharmacological effect directly. All values below are least-squares (LS) mean differences, bococizumab minus placebo, from models including treatment, baseline value, geographic region and complete statin intolerance.
Percent change from baseline in LDL-C at Week 14
LS mean difference, percentage points · 95% CI: −57.91 to −55.89 · P < 0.001 (MMRM)
Nominal change at Week 14: −73.80 mg/dL (95% CI −75.11 to −72.50)
| Endpoint | Method | LS mean difference | 95% CI | P-value |
|---|---|---|---|---|
| LDL-C, percent change, Week 14 | MMRM (through Week 70) | −56.90 | −57.91 to −55.89 | <0.001 |
| LDL-C, nominal change (mg/dL), Week 14 | MMRM (through Week 70) | −73.80 | −75.11 to −72.50 | <0.001 |
| LDL-C, percent change, last post-baseline measurement (up to Week 140) | ANCOVA | −39.31 | −40.55 to −38.06 | <0.001 |
| Non-HDL-C, percent change, Week 14 | MMRM (through Week 70) | −51.87 | −52.81 to −50.94 | <0.001 |
| VLDL-C, percent change, Week 14 | MMRM (through Week 70) | −18.41 | −19.96 to −16.86 | <0.001 |
| RLP-C, percent change, Week 14 | MMRM (through Week 70) | −29.20 | −31.44 to −26.96 | <0.001 |
| Apo B, percent change, Week 14 | MMRM (through Week 52) | −51.40 | −52.37 to −50.42 | <0.001 |
| HDL-C, percent change, Week 14 | MMRM (through Week 70) | 6.91 | 6.33 to 7.50 | <0.001 |
| Apo A-I, percent change, Week 14 | MMRM (through Week 52) | 4.40 | 3.90 to 4.90 | <0.001 |
| Total cholesterol, percent change, Week 14 | MMRM (through Week 70) | −37.99 | −38.75 to −37.22 | <0.001 |
| Triglycerides (log-transformed), Week 14 | MMRM on log scale (through Week 70) | 0.82 | 0.81 to 0.83 | <0.001 |
| Lp(a) (log-transformed), Week 14 | MMRM on log scale | 0.68 | 0.67 to 0.69 | <0.001 |
| hs-CRP (log-transformed), Week 14 | MMRM on log scale | 1.06 | 1.02 to 1.09 | 0.002 |
Percentage points, not percent of a percent. The LDL-C estimate of −56.90 is a difference between two mean percent changes. If one arm's mean percent change were −X and the other's −Y, the reported value is the gap between them in percentage points. It is not the percent change in bococizumab participants alone.
Log-scale endpoints are ratios. For triglycerides and Lp(a), the model was fitted to log-transformed values and the 95% CI was derived by exponentiating the log-scale interval. A back-transformed value of 0.82 is a ratio: geometric-mean triglycerides relative to baseline were about 18% lower with bococizumab than with placebo, and 0.68 corresponds to about 32% lower Lp(a). The hs-CRP model was also fitted on the log scale; its value of 1.06 (95% CI 1.02–1.09) lies above 1, pointing to modestly higher hs-CRP relative to placebo at Week 14.
Narrow intervals from large samples. The lipid intervals are very narrow because continuous measurements in thousands of participants carry far more information than a count of relatively rare clinical events. That is why the LDL-C effect is estimated to within about one percentage point while the cardiovascular hazard ratio has an interval spanning 0.65 to 0.97.
Week 14 versus last measurement. The LDL-C difference at the last post-baseline measurement (−39.31) is smaller in magnitude than at Week 14 (−56.90). These are different estimands from different models: the last measurement could occur at any time up to Week 140, and an ANCOVA on each participant's last value mixes short and long exposures and absorbs whatever happened to participants over time. The registry does not attribute the difference to any specific cause, but it shows that the Week 14 effect was not the same as the effect observed at participants' final assessment.
9. Safety: Serious Adverse Events
The registry reports serious adverse events (SAEs) as the number of participants affected over the number at risk in each arm.
| Arm | Participants with SAEs | Participants at risk |
|---|---|---|
| Bococizumab (PF-04950615) | 934 | 5276 |
| Placebo | 994 | 5279 |
Fewer participants in the bococizumab arm than in the placebo arm had a serious adverse event, with nearly identical numbers at risk. No formal statistical comparison of SAEs is posted. Two cautions apply. First, serious adverse events in a cardiovascular outcomes trial include the cardiovascular events that are themselves efficacy endpoints, so an SAE count partly re-counts efficacy. Second, a count of participants with at least one SAE does not account for differences in time on study or for the type and severity of individual events; it is a coarse summary, not a full safety profile.
10. Statistical Methodology
Stratified log-rank test
The P-values for all cardiovascular endpoints come from a log-rank test. At each observed event time, the test compares the number of events in the bococizumab arm with the number expected if both arms shared the same hazard, then sums these observed-minus-expected differences across time. Stratification performs this comparison within each combination of geographic region and statin-intolerance status and pools the results, so that the treatment comparison is not distorted by differences in baseline risk between strata.
where Osi is the observed number of events in the bococizumab arm at event time i in stratum s, and Esi is the number expected under equal hazards given the numbers at risk.
Stratified Cox proportional-hazards model
Hazard ratios and 95% CIs came from a Cox proportional-hazards model stratified by geographic region and complete statin intolerance, with treatment as the covariate. Stratification lets each stratum have its own baseline hazard shape, while assuming the treatment effect (the hazard ratio) is common across strata.
h0s(t) is an unspecified baseline hazard for stratum s; x = 1 for bococizumab and 0 for placebo. An HR below 1 corresponds to a lower estimated hazard with bococizumab.
Event rates per 100 participant-years
The registry's outcome measures are named as event rates. An event rate of this kind is a crude incidence: the number of participants with a first event divided by total participant-years at risk, multiplied by 100. It is a useful absolute summary when follow-up differs between participants, as it inevitably does in a trial that stops early.
A crude rate assumes a roughly constant hazard over follow-up; the Cox hazard ratio does not require that, but does assume the ratio between arms is constant.
Mixed model for repeated measures (MMRM)
For Week 14 lipid endpoints, the registry reports an MMRM including observations through Week 70 (Week 52 for Apo B and Apo A-I), with fixed effects for treatment group, visit, treatment-by-visit interaction, baseline value, baseline-by-visit interaction, geographic region and complete statin intolerance. The model uses every available post-baseline measurement from each participant, models the correlation between repeated measurements within a participant, and reports the treatment difference at a specific visit (Week 14) from the treatment-by-visit interaction.
A key property of MMRM is how it handles missing visits. Participants with some missing measurements still contribute the measurements they have, and the estimate is valid under a missing-at-random assumption: whether a value is missing may depend on previously observed values, but not on the unobserved value itself.
Analysis of covariance (ANCOVA)
For LDL-C percent change at the last post-baseline measurement, the registry reports an ANCOVA with fixed effects for treatment group, baseline value, geographic region and complete statin intolerance. Unlike MMRM, this uses one value per participant. Adjusting for baseline LDL-C reduces residual variance and therefore narrows the confidence interval, and it corrects for any chance imbalance in baseline values between arms.
Log transformation of skewed biomarkers
Triglycerides, Lp(a) and hs-CRP were analyzed on the log scale. These markers are typically right-skewed, and a log transformation both stabilizes variance and turns multiplicative effects into additive ones. Exponentiating the log-scale difference yields a ratio, which is why their estimates sit near 1 instead of near 0.
11. Termination, Missing Follow-Up, and Multiplicity
Early termination
The trial was terminated, and the record ties the omission of health care resource utilization endpoints to discontinuation of the bococizumab development program. Stopping for reasons external to the observed efficacy data does not by itself bias the treatment comparison, but it limits information and precision.
Censoring
Participants without an event were censored at the end of their follow-up. The Cox and log-rank analyses assume censoring is non-informative within strata. Administrative censoring caused by stopping the whole trial at one time is generally the most benign form.
Missing lipid values
MMRM uses all observed visits and is valid under missing at random. The last-measurement ANCOVA instead takes whatever final value each participant has, which changes the estimand from "effect at a fixed time" to "effect at last assessment".
Multiplicity
One primary endpoint is supported by many secondary analyses. Without a described testing hierarchy, the primary comparison carries the confirmatory weight; secondary results are supportive.
12. Statistical Methods Explained
Why is the outcome called an "event rate" when the result is a hazard ratio?
The outcome measure describes what was counted in each arm: first major CV events per 100 participant-years. The statistical analysis compares arms with a stratified Cox model, which estimates a hazard ratio. The crude rate ratio and the hazard ratio will usually be close when hazards are roughly constant and proportional, but the hazard ratio also accounts for stratification and uses the exact timing of events, so it is the preferred summary for inference.
What does a hazard ratio of 0.79 with a 95% CI of 0.65 to 0.97 tell us?
It is the best single estimate of how much lower the hazard of a first major CV event was with bococizumab: about 21% lower. The interval indicates that effects ranging from about a 35% reduction down to about a 3% reduction are compatible with the data. The trial therefore supports a benefit in direction but leaves real uncertainty about its size.
Why stratify by geographic region and complete statin intolerance?
Baseline event risk can differ by region and by whether participants tolerate statins. Stratifying the Cox model and log-rank test lets each stratum have its own baseline hazard, so that the treatment comparison is made within comparable groups and then combined. This typically improves precision and matches the analysis to the structure of the randomization.
Why was an MMRM used for Week 14 LDL-C, and an ANCOVA for the last measurement?
The Week 14 endpoint targets the effect at a specific visit, and MMRM uses all repeated measurements to estimate it efficiently while handling missed visits under missing at random. The last-measurement endpoint has, by design, one value per participant taken at varying times, so a single-timepoint ANCOVA adjusted for baseline is the natural model.
What does 0.82 for triglycerides mean if it is labeled an LS mean difference?
The model was fitted on log-transformed values. A difference on the log scale becomes a ratio after exponentiation, so 0.82 means that, relative to baseline, geometric-mean triglycerides with bococizumab were about 82% of those with placebo at Week 14, an 18% relative reduction.
Why do fatal MI and CV death have such wide confidence intervals?
Precision in time-to-event analysis depends on the number of events, not the number of participants. Fatal MI and CV death are rarer than the composite endpoint, so their hazard ratios (0.44 and 0.82) are estimated with intervals that stretch well above 1. These results are uninformative about magnitude, and should not be read as showing no effect or as showing a large one.
Why does a P-value of 0.051534 for non-fatal MI not contradict the composite result?
The non-fatal MI hazard ratio (0.77) is in line with the composite hazard ratio (0.79); only the precision differs. A P-value just above 0.05 reflects fewer events in a component endpoint, not a different treatment effect. Dichotomizing at 0.05 hides that the two estimates tell a consistent story.
13. Limitations
- Early termination: the trial stopped when the development program was discontinued, shortening follow-up (maximum up to 3.4 years) and limiting the number of events, which widens every event-based interval.
- Precision of the primary estimate: the 95% CI of 0.65 to 0.97 excludes 1 but spans effects from modest to substantial.
- Analysis population: the FAS excludes participants with multiple randomization attempts and all participants at Site 3027; these exclusions are unrelated to assigned treatment but mean the analysis is not of every randomized participant.
- Proportional hazards: each hazard ratio is a single average over follow-up; the registry does not report checks of the proportional-hazards assumption or time-varying effects.
- Multiplicity: many secondary endpoints are reported with nominal P-values and no described adjustment; several overlap heavily with the primary composite.
- Absolute effects: the posted comparisons are relative; absolute risk differences or numbers needed to treat cannot be read from hazard ratios alone.
- Attenuation of LDL-C lowering: the LDL-C difference at the last post-baseline measurement (−39.31) is smaller than at Week 14 (−56.90), and the registry does not explain the change.
- hs-CRP: the estimate above 1 at Week 14 indicates that bococizumab did not lower this inflammatory marker relative to placebo.
- Unevaluated endpoints: health care resource utilization endpoints were not evaluated because of the program's discontinuation.
- Safety summary: SAE counts by arm are posted without a formal comparison or adjustment for exposure time.
14. Why This Trial Matters Statistically
SPIRE-2 is a useful teaching case because it combines a large event-driven outcomes comparison, a precisely estimated biomarker effect and an unplanned early stop. The contrast between a very narrow LDL-C interval and a comparatively wide hazard-ratio interval illustrates how information depends on the type of endpoint, and the trial shows how a result can be statistically significant while its magnitude remains uncertain.
| Concept | How it appears in SPIRE-2 |
|---|---|
| Time-to-event endpoints | First adjudicated major CV event and its components, followed up to 3.4 years |
| Hazard ratio | Primary HR 0.79 (95% CI 0.65–0.97) from a stratified Cox model |
| Log-rank test | Source of the posted P-values for every CV endpoint |
| Stratification | Geographic region and complete statin intolerance in all models |
| Confidence intervals | Narrow for lipids, wide for rare events such as fatal MI |
| P-values | Non-fatal MI at P = 0.051534 shows the cost of reading results by threshold |
| MMRM | Week 14 lipid differences using repeated measurements through Week 52 or 70 |
| ANCOVA | LDL-C at last post-baseline measurement, adjusted for baseline |
| Log transformation | Ratio-scale estimates for triglycerides, Lp(a) and hs-CRP |
| Composite endpoints | Several overlapping composites with closely similar HRs |
| Early termination | Reduced information and unevaluated resource-use endpoints |
15. Related Tutorials
Learn more about the methods used in this trial:
16. Related Calculators
17. Sources
- ClinicalTrials.gov: NCT01975389, The Evaluation of Bococizumab (PF-04950615; RN316) in Reducing the Occurrence of Major Cardiovascular Events in High Risk Subjects (SPIRE-2).
- PubMed: PMID 28304242.
- PubMed: PMID 29685591.
- PubMed: PMID 29716940.
- PubMed: PMID 35277540.
Continue learning
Connect the endpoints and methods in this trial to in-depth statistical tutorials and calculators.