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
FOCUS was a randomized, parallel-group, single-masked phase 3 trial evaluating two blood transfusion strategies in surgical patients with cardiovascular disease. The trial enrolled 2,016 participants and compared a liberal transfusion strategy with a restrictive, symptomatic strategy.
| Feature | FOCUS |
|---|---|
| Trial name | FOCUS |
| ClinicalTrials.gov identifier | NCT00071032 |
| Phase | Phase 3 |
| Status | Completed |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Single |
| Primary purpose | Treatment |
| Enrollment | 2,016 |
| Arms | 2 |
| Lead sponsor | Rutgers, The State University of New Jersey |
| Sponsor type | Other |
| Start | July 2003 |
| Primary completion | May 2009 |
2. Clinical Question
The central question was whether a liberal blood transfusion strategy with a hemoglobin threshold of 10 g/dL differed from a restrictive, symptomatic transfusion strategy with respect to functional ability or death 60 days after randomization in surgical patients with cardiovascular disease.
Population
Surgical patients with cardiovascular disease. The registry lists anemia, hematologic diseases, cardiovascular diseases, heart diseases, myocardial infarction, thromboembolism, pneumonia, and cerebrovascular accident among the conditions associated with the study.
Intervention
Liberal (10 g/dL) Transfusion Strategy.
Comparator
Restrictive (Symptomatic) Transfusion Strategy.
Primary question
At 60 days after randomization, what was the relative difference between the two strategies in inability to walk 10 feet or across a room without human assistance or death?
3. Trial Design
Liberal transfusion strategy
- Liberal (10 g/dL) Transfusion Strategy
- Biological intervention
- 76 serious adverse events among 1,007 participants at risk
Restrictive transfusion strategy
- Restrictive (Symptomatic) Transfusion Strategy
- Biological intervention
- 66 serious adverse events among 1,009 participants at risk
The randomized parallel-group structure is important statistically because randomization is intended to create treatment groups that can be compared without requiring investigators to decide which patients receive which transfusion strategy. The primary analysis therefore concerns the difference between the randomized strategies rather than a comparison of patients selected for treatment according to their observed characteristics.
4. Endpoints
The registry identifies one registered primary endpoint and reports a formal statistical analysis for it. The endpoint is binary because each participant can be classified according to whether the specified outcome occurred by the stated time frame.
| Endpoint | Registry definition | Time frame | Endpoint type |
|---|---|---|---|
| Primary | Inability to Walk 10 Feet or Across a Room Without Human Assistance or Death | 60 days after randomization | Binary |
| Secondary | Myocardial Infarction, Unstable Angina, or Death for Any Reason | In-hospital | Binary |
Primary endpoint ascertainment
The registry states that the primary endpoint was ascertained via telephone follow-up. The endpoint combines a functional outcome with death, so a participant who died is counted as having the primary outcome regardless of walking ability.
Why the composite endpoint matters
A composite endpoint combines multiple clinically distinct events into one binary outcome. Here, inability to walk the specified distance without human assistance and death are treated as components of the same primary endpoint. Statistically, this produces one comparison, but clinically the components can represent very different experiences.
For interpretation, it is therefore useful to distinguish the statistical question—whether the probability of the combined endpoint differs between randomized strategies—from the separate clinical meanings of impaired walking and death.
5. Primary Result: Walking Inability or Death at 60 Days
The registry reports an odds ratio comparing the liberal strategy with the restrictive strategy. The liberal arm is the numerator group and the restrictive arm is the denominator group.
Odds ratio for inability to walk or death
95% CI: 0.84–1.22 · Two-sided confidence interval
Comparison: Liberal (10 g/dL) Transfusion Strategy vs Restrictive Strategy
| Primary endpoint | Effect measure | Estimate | 95% CI | Hypothesis |
|---|---|---|---|---|
| Inability to Walk 10 Feet or Across a Room Without Human Assistance or Death | Odds ratio | 1.01 | 0.84–1.22 | Superiority |
An odds ratio of 1.01 means that the estimated odds of the combined primary outcome in the liberal-strategy group were approximately 1.01 times the odds in the restrictive-strategy group. Because the liberal group is the numerator and the restrictive group is the denominator, values above 1 correspond to higher estimated odds in the liberal group, while values below 1 correspond to lower estimated odds.
The estimate itself is very close to the null value of 1.00. The 95% confidence interval extends from 0.84 to 1.22, so the interval includes 1.00. Under the statistical framework represented by the reported estimate and interval, the data are compatible with both a lower and a higher odds of the combined endpoint for the liberal strategy relative to the restrictive strategy.
The confidence interval describes uncertainty around the estimated odds ratio; it does not describe the range of outcomes that individual participants experienced. Its width also indicates that the point estimate should not be interpreted as an exact measurement of a treatment effect.
The registry does not report a p-value for this analysis. A p-value would address the compatibility of the observed data with a specified null hypothesis; it would not measure the magnitude or clinical importance of the effect. The odds ratio and its confidence interval provide the effect estimate and its statistical precision.
6. Secondary Result: In-Hospital Cardiovascular Events or Death
The registry also reports a risk difference for the composite of myocardial infarction, unstable angina, or death for any reason during hospitalization.
Risk difference for in-hospital myocardial infarction, unstable angina, or death
99% CI: -3.3 to 1.6 · Two-sided confidence interval
Comparison: Liberal (10 g/dL) Transfusion Strategy vs Restrictive Strategy
| Secondary endpoint | Effect measure | Estimate | 99% CI | Hypothesis |
|---|---|---|---|---|
| Myocardial Infarction, Unstable Angina, or Death for Any Reason | Risk difference | -0.9 | -3.3 to 1.6 | Superiority |
A risk difference of -0.9 means that the estimated risk of the composite in-hospital outcome was 0.9 percentage points lower in the liberal-strategy group than in the restrictive-strategy group, using the direction specified by the reported comparison.
The corresponding 99% confidence interval ranges from -3.3 to 1.6. Because the interval includes 0, the reported estimate is compatible with a lower risk, essentially no difference, or a higher risk in the liberal group under the statistical framework represented by the interval.
This endpoint uses a risk difference rather than an odds ratio. The two measures are not interchangeable: a risk difference is an absolute comparison of event probabilities, whereas an odds ratio compares odds. A risk difference of -0.9 therefore should not be interpreted as an odds ratio of any particular value.
The registry does not report a p-value for this secondary analysis. The 99% confidence interval is the principal measure of statistical uncertainty reported for the effect estimate.
7. Statistical Methodology
The registry identifies the two posted statistical analyses by their effect measures—odds ratio for the primary endpoint and risk difference for the secondary endpoint—but does not report the specific statistical method or model used to obtain either estimate.
| Analysis feature | Registry information |
|---|---|
| Primary endpoint type | Binary |
| Primary effect measure | Odds ratio |
| Primary estimate | 1.01 |
| Primary confidence interval | 95%, two-sided; 0.84–1.22 |
| Secondary endpoint type | Binary |
| Secondary effect measure | Risk difference |
| Secondary estimate | -0.9 |
| Secondary confidence interval | 99%, two-sided; -3.3 to 1.6 |
| Hypothesis type | Superiority |
| Posted analysis method | Not reported |
What a binary endpoint requires statistically
For a binary endpoint, each participant contributes an outcome classification—for example, occurrence or non-occurrence of the specified composite outcome. The analysis then compares those classifications between randomized treatment groups.
An odds ratio can be estimated from a two-by-two event table or from a regression model for a binary outcome. In a simple two-group comparison, the odds in each group are the number with the event divided by the number without the event, and the odds ratio compares those two odds. Regression approaches can additionally incorporate prespecified covariates or stratification factors when appropriate.
A risk difference is more directly expressed on the probability scale. It is calculated as the event probability in one group minus the event probability in the comparison group. Its interpretation is therefore in percentage-point units when risks are expressed as percentages.
The two measures answer related but different questions. A risk difference describes an absolute difference in event probability; an odds ratio describes a relative comparison of odds.
8. Statistical Methods Explained
Why is the primary endpoint analyzed as a binary outcome?
The registered primary endpoint asks whether a participant was unable to walk the specified distance without human assistance or died by 60 days after randomization. That produces a yes/no outcome for each participant. Binary methods are therefore appropriate for comparing the proportion or odds of the combined outcome between treatment strategies.
What does an odds ratio of 1.01 mean?
An odds ratio of 1.01 means that the estimated odds of the primary endpoint in the liberal-strategy group were 1.01 times those in the restrictive-strategy group. This is close to the null value of 1.00. It does not mean that 1.01% of patients experienced the endpoint, nor does it mean that the liberal strategy changed individual patients' risk by 1%.
Why does the confidence interval include the null value?
For an odds ratio, the null value is 1.00. The primary 95% confidence interval is 0.84–1.22, which includes 1.00. Consequently, the interval encompasses effect estimates corresponding to lower odds, no odds difference, and higher odds for the liberal strategy relative to the restrictive strategy.
What does a risk difference of -0.9 mean?
The secondary risk difference compares event probabilities directly. A value of -0.9 means the estimated probability of myocardial infarction, unstable angina, or death was 0.9 percentage points lower in the liberal-strategy group than in the restrictive-strategy group, using the reported direction of comparison.
Why is the secondary confidence interval 99% rather than 95%?
The registry reports a 99% two-sided confidence interval for the secondary risk difference. A 99% interval uses a higher confidence level than a 95% interval and therefore generally produces a wider interval for the same underlying estimate and data. The reported interval is -3.3 to 1.6.
Why does the missing p-value matter?
A p-value and a confidence interval provide related but distinct summaries of statistical uncertainty. The registry's posted analyses provide estimates and confidence intervals but do not provide p-values. The confidence intervals therefore provide the available quantitative information about the precision of the reported effects and whether the corresponding null value lies within the interval.
9. Odds Ratio Versus Risk Difference
The two posted analyses use different effect measures. This is statistically important because the measures operate on different scales and should be interpreted separately.
| Feature | Primary analysis | Secondary analysis |
|---|---|---|
| Outcome | Inability to walk 10 feet or across a room without human assistance or death | Myocardial infarction, unstable angina, or death for any reason |
| Time frame | 60 days after randomization | In-hospital |
| Effect measure | Odds ratio | Risk difference |
| Estimate | 1.01 | -0.9 |
| Confidence level | 95% | 99% |
| Confidence interval | 0.84–1.22 | -3.3 to 1.6 |
| Null value | 1.00 | 0 |
| Direction of comparison | Liberal vs restrictive | Liberal vs restrictive |
The null value differs because the mathematical scales differ. An odds ratio of 1 indicates equal odds, whereas a risk difference of 0 indicates equal risks. This distinction is fundamental when reading results across endpoints within the same trial.
10. Safety Results
The registry reports serious adverse events by randomized strategy using the number affected and the number at risk.
| Safety measure | Liberal (10 g/dL) Transfusion Strategy | Restrictive Strategy |
|---|---|---|
| Serious adverse events | 76 / 1,007 | 66 / 1,009 |
The serious-adverse-event counts are descriptive safety information. They should not be conflated with either of the efficacy endpoints because the safety outcome and the two efficacy outcomes have different definitions and time frames.
11. Understanding the Primary Confidence Interval
The primary odds ratio of 1.01 is a point estimate. The accompanying 95% confidence interval of 0.84–1.22 communicates how much statistical uncertainty surrounds that estimate.
Point estimate
The estimate is 1.01. On the odds-ratio scale, this is very close to 1.00, the null value.
Lower boundary
The lower confidence limit is 0.84, corresponding to an odds ratio below the null value.
Upper boundary
The upper confidence limit is 1.22, corresponding to an odds ratio above the null value.
Interpretive range
The interval therefore includes both directions of association as well as the null value.
This is why the point estimate should not be read in isolation. An estimate of 1.01 with a confidence interval of 0.84–1.22 conveys substantially more information than the number 1.01 alone.
12. Understanding the Secondary Confidence Interval
The secondary risk difference is -0.9 with a 99% confidence interval from -3.3 to 1.6. The interval is expressed on an absolute risk scale rather than an odds scale.
Secondary effect estimate
99% CI: -3.3 to 1.6
Risk difference for myocardial infarction, unstable angina, or death for any reason, in-hospital.
The lower limit of -3.3 represents a larger estimated reduction in absolute risk for the liberal strategy, whereas the upper limit of 1.6 represents an estimated increase. The null value of 0 lies between these limits.
Importantly, the confidence interval does not say that the true effect has a 99% probability of falling between -3.3 and 1.6. In the conventional frequentist framework, the interval is a procedure that would contain the fixed underlying parameter in 99% of repeated samples under the relevant assumptions.
13. Randomization and Causal Interpretation
Randomization is central to the statistical interpretation of FOCUS. Participants were assigned to one of two transfusion strategies rather than simply observed under strategies selected according to clinical characteristics.
The treatment comparison is defined by randomized assignment. This provides the basis for interpreting between-group differences as evidence about the strategies, subject to the trial's design, conduct, endpoint ascertainment, and statistical assumptions.
Randomization does not guarantee that every baseline characteristic will be numerically identical between groups. Its statistical purpose is to make treatment assignment independent of prognostic characteristics in expectation, thereby reducing systematic confounding in the comparison.
14. Superiority Hypothesis
The registry classifies the hypothesis type as superiority. A superiority framework asks whether the randomized strategies differ in the specified direction of the statistical comparison rather than asking whether one strategy is merely no worse than another by a prespecified margin.
This distinction matters because non-inferiority trials use a different decision framework and require a prespecified non-inferiority margin. No non-inferiority margin is reported for the FOCUS analyses presented here.
Primary null value
For the odds ratio, the null value is 1.00.
Secondary null value
For the risk difference, the null value is 0.
15. Trial Timeline
Trial start
The registry lists July 2003 as the study start.
Randomized treatment comparison
The study used randomized, parallel allocation with two biological interventions: a liberal 10 g/dL transfusion strategy and a restrictive symptomatic strategy.
Primary endpoint time frame
The primary endpoint was inability to walk 10 feet or across a room without human assistance or death, ascertained via telephone follow-up.
Secondary endpoint
The registry reports a secondary composite of myocardial infarction, unstable angina, or death for any reason.
Primary completion
The registry lists May 2009 as the primary completion date.
16. Important Limitations and Interpretation Issues
- Unreported statistical method: The registry reports the effect measures and confidence intervals but does not identify the specific statistical method used to generate the posted estimates.
- No reported p-values: The posted statistical analyses contain estimates and confidence intervals but no p-values.
- Composite primary endpoint: The primary outcome combines inability to walk without human assistance with death. These components have different clinical meanings and should not be treated as interchangeable events.
- Different effect scales: The primary endpoint is summarized using an odds ratio, whereas the secondary endpoint is summarized using a risk difference. Their numerical values cannot be compared directly.
- Different confidence levels: The primary analysis uses a 95% confidence interval, while the secondary analysis uses a 99% confidence interval. The intervals therefore have different confidence levels and should not be compared solely by their widths.
- Endpoint time frames differ: The primary endpoint is assessed 60 days after randomization, whereas the secondary endpoint is an in-hospital outcome.
- Telephone ascertainment: The registry states that the primary endpoint was ascertained via telephone follow-up. Functional outcome ascertainment therefore depends on the follow-up assessment process.
- Analysis population: The posted statistical analyses identify the treatment comparison but do not specify an analysis population in the information reported here.
- Serious adverse events: Safety counts use the reported affected/at-risk denominators and represent a separate dimension of the trial from the efficacy endpoints.
17. What the Odds Ratio Does — and Does Not — Mean
An odds ratio of 1.01 means that the estimated odds of the primary composite outcome in the liberal-strategy group were 1.01 times those in the restrictive-strategy group.
It does not mean that 1.01% of participants experienced the endpoint, that the absolute risk difference was 1.01 percentage points, or that every individual participant experienced a 1% change in risk.
The 95% confidence interval of 0.84–1.22 shows the uncertainty around the estimated odds ratio. Because it spans the null value of 1.00, the interval encompasses estimates consistent with lower odds, no difference in odds, and higher odds for the liberal strategy relative to the restrictive strategy.
An effect estimate describes the magnitude and direction of the observed comparison. A p-value, when reported, addresses compatibility with a null hypothesis under a specified testing framework. A p-value is therefore not a measure of effect size, and a confidence interval provides information about precision that a p-value alone does not.
18. Why This Trial Matters Statistically
FOCUS is a useful teaching example because the registry presents two binary clinical endpoints using two different effect measures. That makes it possible to see how the same randomized-trial framework can produce different statistical summaries depending on the question being asked.
| Concept | How it appears in FOCUS |
|---|---|
| Randomization | Participants were randomized to two transfusion strategies. |
| Parallel-group design | The two strategies were evaluated as parallel treatment groups. |
| Binary endpoint | The primary endpoint classifies participants according to inability to walk or death. |
| Odds ratio | The primary endpoint is summarized using an OR of 1.01. |
| Risk difference | The secondary composite is summarized using an RD of -0.9. |
| Confidence interval | The primary analysis uses a 95% CI and the secondary analysis a 99% CI. |
| Superiority hypothesis | The registry identifies superiority as the hypothesis type. |
| Composite endpoint | The primary endpoint combines functional inability to walk with death. |
| Different time frames | The primary endpoint is at 60 days; the secondary endpoint is in-hospital. |
| Safety analysis | Serious adverse events are reported as affected/at-risk counts by arm. |
19. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary reported odds ratio is 1.01 with a 95% CI of 0.84–1.22. The secondary reported risk difference is -0.9 with a 99% CI of -3.3 to 1.6. Both intervals include their respective null values.
Clinical interpretation
The primary endpoint combines functional walking ability and death, while the secondary endpoint combines myocardial infarction, unstable angina, and death. Their clinical meanings therefore extend beyond the numerical effect measures alone.
The distinction is important: a statistical estimate is a quantitative summary of a randomized comparison, whereas clinical interpretation requires attention to the endpoint definition, time frame, absolute scale, uncertainty, and the consequences represented by the outcome.
20. Related Statistical Tutorials
Learn more about the methods and concepts used in this trial:
21. Related Statistical Calculators
22. Sources
- ClinicalTrials.gov: FOCUS, NCT00071032.
- PubMed: PMID 22168590.
- PubMed: PMID 41114449.
- PubMed: PMID 39294606.
- PubMed: PMID 37633261.
- PubMed: PMID 34932836.
Continue through Clinical Biostats
Use the related statistical concepts and tools to explore how randomized comparisons, binary endpoints, effect measures, and confidence intervals are analyzed in clinical research.
23. Record Summary
FOCUS provides a clear example of statistical interpretation for a randomized phase 3 trial with binary clinical outcomes. The primary endpoint—inability to walk 10 feet or across a room without human assistance or death 60 days after randomization—was reported with an odds ratio of 1.01 and a two-sided 95% confidence interval of 0.84–1.22. The secondary endpoint—myocardial infarction, unstable angina, or death for any reason during hospitalization—was reported with a risk difference of -0.9 and a two-sided 99% confidence interval of -3.3 to 1.6.
The most important statistical lesson is that effect measures must be interpreted on their own scales. The odds ratio uses 1.00 as its null value, whereas the risk difference uses 0. The confidence intervals provide information about precision and encompass their respective null values. The registry does not report p-values or the specific statistical methods used to generate the posted estimates, so the reported effect measures and confidence intervals are the principal quantitative summaries available from the registry analyses.
The trial also illustrates why endpoint definition matters. A composite endpoint can efficiently summarize clinically important outcomes, but its interpretation requires attention to the individual components, their clinical meaning, and the time frame over which they are assessed.