← Clinical Trials
Cardiovascular Diseases Phase 3 Completed NCT01492361

REDUCE-IT: Complete Statistical Analysis of AMR101 in Cardiovascular Disease

An independent statistical analysis of the randomized phase 3 REDUCE-IT trial evaluating AMR101 versus placebo in high-risk patients with cardiovascular disease and hypertriglyceridemia who were on statin therapy.

Trial start: 2011-11  ·  Primary completion: 2018-05  ·  Enrollment: 8179
Scope of this record

This page separates reported trial results from statistical interpretation. Numerical trial results, endpoint definitions, design characteristics, and statistical methods presented here are restricted to the ClinicalTrials.gov record data and its posted analyses.

Registry context: This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. View the REDUCE-IT ClinicalTrials.gov record.

1. Trial at a Glance

REDUCE-IT was a randomized, triple-masked, parallel phase 3 trial with a prevention objective. The registry reports 8179 enrolled participants and two study arms comparing AMR101 with placebo in the setting of statin therapy.

8179
Enrollment
Participants
2
Arms
AMR101 vs placebo
3
Phase
Phase 3
0.75
Primary HR
95% CI 0.68–0.83
FeatureREDUCE-IT
Trial nameREDUCE-IT
ClinicalTrials.gov identifierNCT01492361
Therapeutic areaCardiology
ConditionCardiovascular Diseases
PhasePhase 3
AllocationRandomized
Design modelParallel
MaskingTriple
Primary purposePrevention
Enrollment8179
StatusCompleted
Lead sponsorAmarin Pharma Inc.
Sponsor typeIndustry
Start2011-11
Primary completion2018-05

2. Clinical Question

The registered trial question was whether AMR101, compared with placebo, was associated with a difference in the time to first occurrence of the specified composite cardiovascular outcome in high-risk patients with cardiovascular disease and hypertriglyceridemia who were on statin therapy.

Population

High-risk patients with hypertriglyceridemia and cardiovascular disease who were on statin therapy, as described by the registered study title and condition.

Intervention

AMR101, with statin therapy listed among the study interventions.

Comparator

Placebo, with statin therapy listed among the study interventions.

Primary question

Does AMR101 change the time to first occurrence of the registered composite cardiovascular endpoint compared with placebo?

3. Trial Design

01
Randomize8179 participants
02
Two armsAMR101 vs placebo
03
ParallelParallel design
04
Follow-upUp to approximately 6 years
05
CompareTime-to-event outcomes
Allocation
Randomized allocation.
Design model
Parallel-group design.
Masking
Triple masking.
Primary purpose
Prevention.
ARM 1

AMR101

  • AMR101 listed as a drug intervention.
  • Statin therapy listed among the study interventions.
  • Compared with placebo for the registered time-to-event outcomes.
ARM 2

Placebo

  • Placebo listed as a drug intervention.
  • Statin therapy listed among the study interventions.
  • Compared with AMR101 for the registered time-to-event outcomes.

4. Randomization, Stratification, and Analysis Structure

The posted statistical analyses identify a stratified analysis. For the primary endpoint and each posted secondary analysis, the analysis notes state that the comparison was stratified by CV risk category, use of ezetimibe, and geographical region.

FeatureReported approach
AllocationRandomized
Analysis typeSurvival analysis
ComparisonAMR101 vs Placebo
Primary hypothesis typeSuperiority
Primary statistical testLog-rank test
Effect measureHazard ratio
Stratification factors in analysis notesCV risk category; use of ezetimibe; geographical region

The ClinicalTrials.gov record does not specify a separate intention-to-treat definition, an as-treated safety population definition, a crossover procedure, a factorial structure, an interim-analysis strategy, an alpha-spending procedure, or a missing-data/imputation method. Those topics therefore are not treated as features of the REDUCE-IT statistical design on this page.

5. Endpoints

The registry lists one primary endpoint and reports nine secondary analyses. All ten posted analyses are time-to-event comparisons of AMR101 versus placebo.

Endpoint roleRegistered / analyzed endpointTime frame
Primary Composite of CV Death, Nonfatal MI (Including Silent MI), Nonfatal Stroke, Coronary Revascularization, or Unstable Angina Determined to be Caused by Myocardial Ischemia by Invasive / Non-invasive Testing and Requiring Emergent Hospitalization. Total follow-up time of up to approximately 6 years.
Secondary Composite of CV Death, Nonfatal MI (Including Silent MI), or Nonfatal Stroke. Total follow-up time of up to approximately 6 years.
Secondary Composite of CV Death or Nonfatal MI (Including Silent MI). Total follow-up time of up to approximately 6 years.
Secondary Fatal or Nonfatal MI (Including Silent MI). Total follow-up time of up to approximately 6 years.
Secondary Non-elective Coronary Revascularization Represented as the Composite of Emergent or Urgent Classifications. Total follow-up time of up to approximately 6 years.
Secondary CV Death. Total follow-up time of up to approximately 6 years.
Secondary Unstable Angina Determined to be Caused by Myocardial Ischemia by Invasive / Non-invasive Testing and Requiring Emergent Hospitalization. Total follow-up time of up to approximately 6 years.
Secondary Fatal or Nonfatal Stroke. Total follow-up time of up to approximately 6 years.
Secondary Total Mortality, Nonfatal MI (Including Silent MI), or Nonfatal Stroke. Total follow-up time of up to approximately 6 years.
Secondary Total Mortality. Total follow-up time of up to approximately 6 years.
Endpoint construction matters: the primary endpoint is a composite. The registry definition specifies that the outcome is the first occurrence of any component during the follow-up period. A composite time-to-event endpoint therefore answers a different question from analyzing each component separately.

6. Primary Result

The posted primary analysis compares AMR101 with placebo using a stratified log-rank test and reports a hazard ratio as the effect measure. The hypothesis type is superiority.

Primary composite cardiovascular endpoint

Hazard ratio for the primary composite endpoint

0.75

95% CI: 0.68–0.83   ·   P = 0.00000001

Stratified by CV risk category, use of ezetimibe, and geographical region.

FeaturePrimary analysis
Groups comparedAMR101 vs Placebo
Endpoint typeTime-to-event
Statistical methodLog-rank test
Effect measureHazard ratio
Estimate0.75
Confidence interval95% CI 0.68–0.83
CI sidesTwo-sided
P-value0.00000001
HypothesisSuperiority
Clinical Biostats interpretation

The hazard ratio of 0.75 means that the estimated hazard in the AMR101 group was 75% of the estimated hazard in the placebo group under the reported time-to-event analysis. Expressed as a simple relative interpretation, an HR of 0.75 corresponds to a 25% lower estimated hazard for the primary composite endpoint.

This does not mean that 25% of patients avoided an event, that the absolute event probability was reduced by 25 percentage points, or that every individual patient experienced the same relative reduction. A hazard ratio is a relative time-to-event measure, not an absolute risk difference.

The 95% confidence interval of 0.68–0.83 describes the statistical precision of the estimated hazard ratio under the analysis framework. It is an interval for the estimated treatment effect, not a range containing the individual effects experienced by patients.

The very small P = 0.00000001 indicates strong evidence against the null hypothesis represented by the statistical test. It does not measure the size of the treatment effect. Effect size is described by the hazard ratio, while precision is described by the confidence interval.

The analysis is a time-to-event comparison and therefore incorporates the timing of events and censoring. The ClinicalTrials.gov record does not report Kaplan-Meier estimates, median event times, a proportional-hazards diagnostic, or a separate absolute-risk analysis. The hazard ratio should consequently be interpreted as the reported relative effect measure rather than as an absolute measure of benefit.

7. Secondary Endpoint Results

The registry reports eight secondary analyses in addition to the primary analysis. Each uses a log-rank test, a hazard ratio, a two-sided 95% confidence interval, and the same stratification factors noted for the primary analysis.

Secondary endpointHR95% CIP-value
Composite of CV Death, Nonfatal MI (Including Silent MI), or Nonfatal Stroke 0.74 0.65–0.83 0.0000006
Composite of CV Death or Nonfatal MI (Including Silent MI) 0.75 0.66–0.86 <0.0001
Fatal or Nonfatal MI (Including Silent MI) 0.69 0.58–0.81 <0.0001
Non-elective Coronary Revascularization Represented as the Composite of Emergent or Urgent Classifications 0.65 0.55–0.78 <0.0001
CV Death 0.80 0.66–0.98 0.0315
Unstable Angina Determined to be Caused by Myocardial Ischemia by Invasive / Non-invasive Testing and Requiring Emergent Hospitalization 0.68 0.53–0.87 0.0018
Fatal or Nonfatal Stroke 0.72 0.55–0.93 0.0129
Total Mortality, Nonfatal MI (Including Silent MI), or Nonfatal Stroke 0.77 0.69–0.86 <0.0001
Total Mortality 0.87 0.74–1.02 0.0915

How the secondary results should be read

The point estimates range from 0.65 to 0.87. Because these are hazard ratios below 1, each point estimate represents a lower estimated hazard in the AMR101 group relative to placebo for its respective endpoint. The confidence intervals provide the corresponding uncertainty around each estimate.

The registry reports different endpoint definitions, so these estimates should not be treated as interchangeable. For example, the analysis of fatal or nonfatal MI addresses a narrower event definition than the primary composite, while total mortality is a distinct endpoint from a composite containing mortality and nonfatal events.

Multiplicity matters: the ClinicalTrials.gov record identifies a superiority framework and report multiple secondary endpoint analyses, but they do not provide an alpha-allocation or multiplicity-adjustment procedure for these secondary tests. The individual P-values should therefore be read as the values reported for their respective analyses, without assuming an unreported multiplicity adjustment.

8. Comparing the Hazard-Ratio Estimates

Reported hazard ratios · lower values indicate lower estimated hazard in the AMR101 group
Primary composite
0.75
CV death / MI / stroke
0.74
CV death / MI
0.75
Fatal / nonfatal MI
0.69
Revascularization
0.65
CV death
0.80
Unstable angina
0.68
Stroke
0.72
Total mortality / MI / stroke
0.77
Total mortality
0.87

This visual is a direct display of the reported hazard-ratio point estimates rather than a reconstruction of event curves. It should not be interpreted as showing absolute risks, event counts, or Kaplan-Meier probabilities.

9. Safety Results

The ClinicalTrials.gov record reports serious adverse events by treatment arm as affected participants divided by participants at risk.

Safety measureAMR101Placebo
Serious adverse events1252 / 40891254 / 4090

The ClinicalTrials.gov record does not provide additional safety measures such as specific adverse-event categories, severity grades, discontinuations, or adverse-event rates beyond the serious-adverse-event affected/at-risk counts shown above.

Safety and efficacy are different statistical questions. The primary efficacy analysis is a time-to-event comparison using a log-rank test and hazard ratio. The serious-adverse-event ClinicalTrials.gov record is an affected/at-risk summary and is not itself a time-to-event efficacy analysis.

10. Statistical Methodology

Time-to-event analysis

All ten posted statistical analyses are classified as time-to-event analyses. This is important because the outcome is not simply whether an event ever occurred. The analysis concerns the timing of the first qualifying event during follow-up.

Time-to-event data commonly contain right-censored observations. A participant who has not experienced the endpoint by the end of available follow-up can still contribute information for the period during which the participant was observed. The ClinicalTrials.gov record does not specify the exact censoring rules, so this page does not add a trial-specific censoring convention.

Log-rank test

The registry reports Log Rank as the statistical method for the primary analysis and every registry-reported secondary analysis. The log-rank test is designed for comparing survival or event-time distributions between groups while accounting for the timing of observed events.

Core statistical question
H0: no difference in the event-time distributions between AMR101 and placebo

The reported P-value quantifies evidence against the null hypothesis represented by the test. It does not quantify how large or clinically important the observed treatment effect is.

Hazard ratio

The effect measure reported for every posted statistical analysis is the hazard ratio. A hazard ratio compares the instantaneous event rates between groups within a time-to-event framework.

Interpretation
HR < 1  →  lower estimated hazard in the AMR101 group

For example, the primary HR of 0.75 corresponds to a 25% lower estimated hazard, expressed as 1 − 0.75. This is a relative time-to-event interpretation and is not an absolute risk reduction.

Stratified analysis

The analysis notes specify stratification by CV risk category, use of ezetimibe, and geographical region. Stratification allows the treatment comparison to account for these prespecified categories within the reported analysis rather than treating every participant as if these characteristics were ignored.

Confidence intervals

The primary analysis reports a two-sided 95% confidence interval of 0.68–0.83. A confidence interval communicates the statistical precision of the estimated hazard ratio. A narrower interval generally indicates greater precision than a wider interval, although precision and clinical importance are separate concepts.

Superiority testing

The posted analyses identify the hypothesis type as superiority. That is different from a non-inferiority framework: the objective is to evaluate evidence for a difference favoring one treatment rather than to demonstrate that an experimental treatment is not worse than a comparator within a prespecified margin.

11. Statistical Methods Explained

Why use a log-rank test for REDUCE-IT?

The endpoints are defined as time-to-event outcomes over a follow-up period of up to approximately 6 years. A log-rank test is specifically suited to comparing event-time distributions between randomized groups while incorporating the timing of events and censoring.

What does an HR of 0.75 mean?

An HR of 0.75 means that the estimated hazard for the AMR101 group was 75% of the estimated hazard for the placebo group under the reported analysis. The simple relative interpretation is a 25% lower estimated hazard. It does not mean a 25% absolute reduction in the probability of an event.

Why is the confidence interval important?

The 95% CI of 0.68–0.83 shows the uncertainty surrounding the primary HR estimate. The point estimate alone gives one estimated effect; the interval gives additional information about the statistical precision of that estimate.

Why does the P-value not measure effect size?

The P-value addresses evidence against the null hypothesis under the specified statistical test. It is affected by both the observed data and the amount of information available. Effect size is instead summarized here by the hazard ratio, while its statistical precision is summarized by the confidence interval.

What does stratification change?

The registry identifies three stratification factors: CV risk category, use of ezetimibe, and geographical region. A stratified time-to-event analysis incorporates these categories into the comparison rather than conducting a completely unstratified treatment comparison.

Why is the primary endpoint a composite?

The registered primary endpoint combines CV death, nonfatal MI including silent MI, nonfatal stroke, coronary revascularization, and specified unstable angina requiring emergent hospitalization. A composite can capture the first occurrence of any qualifying component, but its interpretation depends on the collection of components rather than on any single component alone.

Does a hazard ratio below 1 prove that every component improved?

No. The primary HR describes the registered composite endpoint. A composite result should not automatically be interpreted as demonstrating the same magnitude of effect for every component. Separate component analyses require their own estimates and uncertainty, and the ClinicalTrials.gov record provides those separately only for the listed secondary endpoints.

12. Secondary Results in Statistical Context

MI-related endpoints

The hazard ratio for fatal or nonfatal MI including silent MI was 0.69, with a 95% CI of 0.58–0.81 and P < 0.0001.

Revascularization

The hazard ratio for non-elective coronary revascularization was 0.65, with a 95% CI of 0.55–0.78 and P < 0.0001.

CV death

The hazard ratio for CV death was 0.80, with a 95% CI of 0.66–0.98 and P = 0.0315.

Total mortality

The hazard ratio for total mortality was 0.87, with a 95% CI of 0.74–1.02 and P = 0.0915.

These analyses illustrate why individual endpoint estimates should be read independently. The reported hazard ratios differ across endpoints, and the confidence intervals differ in width. A treatment effect for a composite endpoint should not be substituted for an estimate of a specific component.

Statistical interpretation

The total-mortality estimate of 0.87 has a 95% CI of 0.74–1.02 and a reported P-value of 0.0915. This illustrates an important statistical distinction: a hazard ratio point estimate below 1 is not equivalent to statistical evidence meeting a conventional two-sided significance threshold. The point estimate, confidence interval, and P-value provide complementary information and should be considered together.

13. The Role of Censoring in Time-to-Event Analysis

The registered endpoints are followed for up to approximately 6 years, making time-to-event methodology appropriate for observations that may have different amounts of follow-up. Participants can contribute information before an event occurs or before follow-up ends.

Conceptual representation
Observed time = min(event time, censoring time)

When a participant is censored, the analysis uses the observed follow-up period without treating the participant as if the event necessarily occurred afterward.

The ClinicalTrials.gov record does not specify the detailed censoring rules or missing-data procedures used in REDUCE-IT. Those details are therefore not inferred here. The important statistical point is that time-to-event methods differ fundamentally from simply comparing proportions of participants who experienced an event.

14. Composite Endpoints and First Events

The primary endpoint is explicitly defined as the first occurrence of any component of the composite during the follow-up period. This creates a single time-to-first-event outcome from several clinically distinct event types.

Primary composite componentIncluded in registered primary endpoint?
CV DeathYes
Nonfatal MI, including silent MIYes
Nonfatal strokeYes
Coronary revascularizationYes
Specified unstable angina requiring emergent hospitalizationYes

A composite endpoint can increase the number of qualifying events available for analysis, but the resulting treatment effect represents the combined endpoint as defined. The statistical meaning of HR 0.75 is therefore specifically tied to the first occurrence of any component of this registered composite.

15. Blinding and Randomization

The trial is registered as randomized and triple-masked. These are design features rather than statistical tests, but they are important to the interpretation of the treatment comparison.

Randomization

Random allocation is intended to create treatment groups whose assignment is not determined by participant or investigator choice. This provides the foundation for comparing outcomes by randomized treatment assignment.

Triple masking

The registry identifies the study as triple-masked. Masking can reduce opportunities for knowledge of treatment assignment to influence trial conduct or assessment.

The ClinicalTrials.gov record does not specify which three groups of individuals were masked. Accordingly, the analysis identifies the study as triple-masked without assigning specific masking roles that are not present in the ClinicalTrials.gov record.

16. What the Primary Result Does — and Does Not — Establish

What it establishes statistically

The posted primary analysis reports an HR of 0.75 with a two-sided 95% CI of 0.68–0.83 and P = 0.00000001 for AMR101 versus placebo on the registered composite time-to-event endpoint, using a stratified log-rank test.

What it does not establish by itself

The primary HR does not provide an absolute risk reduction, number needed to treat, median time to event, survival probability at a particular time point, or component-specific treatment effect. None of those quantities is reported in the ClinicalTrials.gov record.

Why the analysis population matters

The ClinicalTrials.gov record reports enrollment of 8179 but do not provide a separate efficacy analysis-population definition. The analysis should therefore be described using the information actually registry-reported rather than assuming an unreported analysis population.

17. Multiplicity and Multiple Endpoints

REDUCE-IT has one registered primary endpoint and multiple posted secondary analyses. The ClinicalTrials.gov record identifies all of the analyses as superiority comparisons, but do not state an overall multiplicity-control strategy.

Analysis groupNumber reportedStatistical methodEffect measure
Primary endpoint1Log-rank testHazard ratio
Secondary endpoints9Log-rank testHazard ratio
Total statistical analyses posted10Survival analysisHazard ratio

Multiple statistical tests create an interpretive issue because the probability of observing at least one small P-value can increase when many hypotheses are tested. The ClinicalTrials.gov record does not report whether the secondary analyses were adjusted for multiplicity, so this page does not impose an adjustment that is not documented in the ClinicalTrials.gov record.

18. Missing Data, Interim Analysis, and Crossover

Several design topics commonly affect the interpretation of clinical-trial time-to-event analyses. For REDUCE-IT, the ClinicalTrials.gov record does not document an interim-analysis procedure, alpha-spending approach, crossover design, or missing-data/imputation strategy.

Design topicInformation posted on ClinicalTrials.gov for REDUCE-IT
Interim analysisNot specified in the ClinicalTrials.gov record.
Alpha spendingNot specified in the ClinicalTrials.gov record.
CrossoverNot specified in the ClinicalTrials.gov record.
Missing-data/imputation methodNot specified in the ClinicalTrials.gov record.
Bayesian methodsNot reported; the analyses posted on ClinicalTrials.gov use log-rank testing.
Factorial designNot reported; the registered design model is parallel.

This distinction is important: absence of a registry-reported detail is not evidence that the procedure was absent from the underlying protocol. It means only that the detail is not available in the trial data used for this page.

19. Longitudinal Trial History

2011-11

Trial start

The registered REDUCE-IT trial began in November 2011.

2018-05

Primary completion

The registry lists May 2018 as the primary completion date.

Completed

Registry status

The trial is listed as completed, with 8179 participants enrolled.

Results posted

Statistical results available

The registry data indicate that results were posted, including 10 outcome measures and 10 statistical analyses.

20. Statistical Interpretation of the Full Result Set

The primary analysis and the registry-reported secondary analyses form a coherent statistical structure: a randomized, triple-masked, parallel phase 3 comparison evaluated through time-to-event methods, with log-rank testing and hazard ratios as the principal inferential tools.

EndpointHR95% CIP-valueSimple relative interpretation of HR
Primary composite0.750.68–0.830.0000000125% lower estimated hazard
CV death / MI / stroke0.740.65–0.830.000000626% lower estimated hazard
CV death / MI0.750.66–0.86<0.000125% lower estimated hazard
Fatal / nonfatal MI0.690.58–0.81<0.000131% lower estimated hazard
Non-elective revascularization0.650.55–0.78<0.000135% lower estimated hazard
CV death0.800.66–0.980.031520% lower estimated hazard
Specified unstable angina0.680.53–0.870.001832% lower estimated hazard
Fatal / nonfatal stroke0.720.55–0.930.012928% lower estimated hazard
Total mortality / MI / stroke0.770.69–0.86<0.000123% lower estimated hazard
Total mortality0.870.74–1.020.091513% lower estimated hazard

The percentage interpretations in the final column are simple transformations of the reported hazard ratios using 1 − HR. They should not be confused with absolute percentage-point reductions in event probability.

21. Limitations and Interpretation Issues

22. Why This Trial Matters Statistically

REDUCE-IT is a useful teaching case because it demonstrates how a large randomized clinical trial can be analyzed when the principal outcomes are time-to-event composites rather than simple binary outcomes.

ConceptHow it appears in REDUCE-IT
RandomizationThe trial is registered as randomized.
BlindingThe trial is registered as triple-masked.
Parallel designThe design model is parallel.
Time-to-event endpointsThe primary and all registry-reported secondary analyses are classified as time-to-event.
Log-rank testReported as the statistical method for all 10 posted analyses.
Hazard ratioReported as the effect measure for every posted analysis.
Confidence intervalsEvery posted analysis reports a two-sided 95% CI.
Stratified analysisAnalyses are stratified by CV risk category, use of ezetimibe, and geographical region.
SuperiorityThe analyses posted on ClinicalTrials.gov identify superiority as the hypothesis type.
Composite endpointThe primary endpoint combines five specified cardiovascular event categories.
MultiplicityOne primary analysis and nine secondary analyses are posted.
Safety comparisonSerious adverse events are reported as affected/at-risk counts by arm.

23. Clinical Interpretation vs Statistical Interpretation

Statistical interpretation

The reported primary comparison produced a hazard ratio of 0.75 with a two-sided 95% CI of 0.68–0.83 and P = 0.00000001 using a stratified log-rank test.

Clinical interpretation

The statistical result concerns the registered composite cardiovascular time-to-event endpoint. Clinical interpretation requires keeping that composite definition, the relative nature of the hazard ratio, and the separate safety information in view.

A statistically precise relative treatment effect is not the same thing as an absolute measure of patient benefit. The ClinicalTrials.gov record does not provide the absolute event probabilities needed to calculate quantities such as an absolute risk difference or number needed to treat.

24. Statistical Concepts in This Trial

Learn more about the methods used in this trial:

25. Related Statistical Calculators

26. Sources

Continue through the Clinical Biostats statistical pathway

Explore the statistical concepts behind randomized trials, confidence intervals, hazard ratios, log-rank testing, and time-to-event endpoints.

27. Record Summary

REDUCE-IT provides a detailed example of randomized clinical-trial analysis centered on a composite time-to-event endpoint. The trial enrolled 8179 participants in a randomized, triple-masked, parallel phase 3 design. The primary analysis compared AMR101 with placebo using a stratified log-rank test and reported a hazard ratio of 0.75 with a 95% CI of 0.68–0.83 and P = 0.00000001. The analysis was stratified by CV risk category, use of ezetimibe, and geographical region.

The secondary analyses extend the same statistical framework to cardiovascular death, myocardial infarction, stroke, coronary revascularization, unstable angina, and mortality-related endpoints. Their hazard ratios range from 0.65 to 0.87, with confidence intervals and P-values reported for each analysis. The ClinicalTrials.gov record reports serious adverse events affecting 1252/4089 participants in the AMR101 group and 1254/4090 participants in the placebo group.

Clinical Biostats methodology: The statistical story of a trial is more than its smallest P-value. A rigorous interpretation connects the randomized design, endpoint definition, time-to-event structure, statistical test, effect measure, confidence interval, stratification, multiplicity considerations, and safety information while distinguishing reported evidence from educational interpretation.