This page separates reported trial results from statistical interpretation. Numerical trial facts and reported estimates on this page are restricted to the ClinicalTrials.gov record.
1. Trial at a Glance
The Ring Study was a randomized, parallel, phase 3 treatment trial evaluating granulocyte transfusions for resolving infection in people with neutropenia. The registry reports 114 enrolled participants across 3 arms and provides a formal primary analysis comparing the control and granulocyte-transfusion groups.
| Feature | The Ring Study |
|---|---|
| Trial name | The Ring Study |
| Brief title | Safety and Effectiveness of Granulocyte Transfusions in Resolving Infection in People With Neutropenia |
| Phase | Phase 3 |
| Conditions | Neutropenia; Infection |
| Design | Randomized, parallel, single-masked |
| Primary purpose | Treatment |
| Enrollment | 114 |
| Arms | 3 |
| Start | April 2008 |
| Primary completion | April 2013 |
| Lead sponsor | Carelon Research |
| ClinicalTrials.gov | NCT00627393 |
2. Clinical Question
The central question was whether granulocyte transfusions, given with standard antimicrobial therapy, improved the probability that a participant would be alive and have a microbial response at Day 42 compared with the control group.
Population
People with neutropenia and infection enrolled in the phase 3 Ring Study.
Intervention
Granulocyte transfusions, with standard antimicrobial therapy.
Comparator
The control arm receiving standard antimicrobial therapy.
Primary question
Does granulocyte transfusion increase the percentage of participants who are alive at Day 42 and have had microbial response?
3. Trial Design
Standard antimicrobial therapy
- Standard antimicrobial therapy
- Registry intervention type: drug
Granulocyte transfusions
- Granulocyte transfusions
- Registry intervention type: biological
- Apheresis machine was listed as a device intervention
- G-CSF/dexamethasone was listed as a drug intervention
The ClinicalTrials.gov record identifies three study arms overall, while the posted statistical analyses compare the Control Arm with the Granulocyte Arm. The analysis records therefore define the principal reported comparison as control versus granulocyte transfusion.
4. Endpoints
| Endpoint | Registry definition / time frame | Type |
|---|---|---|
| Percentage of Participants Who Are Alive at 42 Days After Treatment and Have Had Microbial Response | Measured at Day 42. Microbial response was defined as a negative blood culture test at 42 days after randomization for subjects with fungemia (candidemia or fusariosis) or bacteremia, or improvement of signs and symptoms of infectious disease (complete or partial response) at 42 days after randomization. | Binary |
| Graft Versus Host Disease Among Recipients of Allogeneic Stem Cell Transplantation | Measured at Day 42. | Time-to-event |
| Long-term Survival | Measured at Month 3. | Time-to-event |
The registry data identify 1 primary endpoint and report 13 outcome measures overall. The primary endpoint is binary because each participant contributes a Day 42 outcome reflecting the combined requirement of being alive and having a microbial response.
5. Statistical Methodology
Primary endpoint: logistic regression
The primary endpoint was analyzed using ordinary multiple logistic regression in the intention-to-treat population consisting of all adjudicated or deceased subjects. The reported comparison was the Control Arm versus the Granulocyte Arm, with the control group as the reference group.
The model included treatment arm, infection strata, underlying disease, zubrod score, respiratory symptoms, and age. Thus, the reported odds ratio is an adjusted treatment comparison rather than an unadjusted comparison of two percentages.
The treatment coefficient is transformed into an odds ratio. An odds ratio of 1 represents equal odds of the modeled outcome between the treatment and reference groups.
Covariate adjustment
The primary model adjusted for infection strata, underlying disease, zubrod score, respiratory symptoms, and age. Covariate adjustment can improve precision and account for prespecified prognostic factors, but the adjusted odds ratio remains a relative measure of odds rather than a direct difference in probabilities.
Intention-to-treat analysis
The primary analysis population was described as all adjudicated or deceased subjects in intention-to-treat analyses. The ITT principle anchors the treatment comparison to randomized assignment rather than allowing later treatment exposure or protocol deviations to redefine the randomized groups.
Fine-Gray / competing-risks analysis
The secondary endpoint of graft versus host disease among recipients of allogeneic stem cell transplantation was analyzed using a competing-risks approach. The registry normalizes the method as Fine-Gray / competing risks. This is appropriate when participants can experience a competing event that prevents or changes the probability of observing the event of interest.
Log-rank test
The secondary long-term survival endpoint was analyzed with a log-rank test comparing survival distributions between the Control Arm and Granulocyte Arm. The reported hazard ratio was obtained for the same comparison, with the control group as the reference group.
A hazard ratio below 1 would correspond to a lower estimated instantaneous event rate in the Granulocyte Arm; a hazard ratio above 1 corresponds to a higher estimated instantaneous event rate under the model.
6. Primary Result: Day 42 Survival With Microbial Response
The primary endpoint was the percentage of participants who were alive at 42 days after treatment and had experienced a microbial response. The formal registry analysis compared the Control Arm with the Granulocyte Arm using multiple logistic regression.
Adjusted odds ratio
95% CI: 0.44–3.20 · P = 0.73
Control group = reference; two-sided 95% confidence interval.
| Primary endpoint | Analysis | Effect estimate | 95% CI | P-value |
|---|---|---|---|---|
| Alive at Day 42 and microbial response | Multiple logistic regression | OR 1.19 | 0.44–3.20 | 0.73 |
The model was adjusted for infection strata, underlying disease, zubrod score, respiratory symptoms, and age. The analysis population was the intention-to-treat population described by the registry as all adjudicated or deceased subjects.
The adjusted odds ratio of 1.19 means that the estimated odds of the combined Day 42 outcome were 1.19 times as high in the Granulocyte Arm as in the Control Arm, after adjustment for the covariates specified in the registry analysis.
This does not mean that the probability of the outcome was 19% higher. Odds and probabilities are different quantities, particularly when an outcome is not rare. The odds ratio also does not say that an individual participant had a 19% greater chance of being alive with microbial response.
The 95% confidence interval of 0.44–3.20 is wide relative to the point estimate. It represents substantial statistical uncertainty around the estimated adjusted odds ratio. Because the interval includes 1, the data are compatible with both lower and higher odds under the model.
The P-value of 0.73 addresses the evidence against the null hypothesis represented by an odds ratio of 1 under the specified model. It is not a measure of the size or clinical importance of the treatment effect. A large P-value does not establish that the two treatments are identical.
Because the primary endpoint is binary, logistic regression is directly aligned with the endpoint structure. The interpretation also depends on the adequacy of the adjustment model and the validity of its covariate specification. The registry does not provide enough information in the ClinicalTrials.gov record to evaluate model diagnostics or calibration.
7. Secondary Result: Graft Versus Host Disease
The registry reports graft versus host disease among recipients of allogeneic stem cell transplantation as a secondary time-to-event endpoint measured at Day 42. The analysis population was restricted to subjects who had allogeneic stem cell transplantation.
Reported competing-risks analysis
Reported effect measure: Log Rank P-Value · P = 0.43
Control Arm vs Granulocyte Arm · two-sided
| Endpoint | Population | Method | Reported measure | P-value |
|---|---|---|---|---|
| Graft Versus Host Disease Among Recipients of Allogeneic Stem Cell Transplantation | Subjects who had allogeneic stem cell transplantation | Fine-Gray / competing risks | Log Rank P-Value; estimate 0.43 | 0.43 |
The registry's statistical-analysis record gives an estimate of 0.43 and a P-value of 0.43, while identifying the effect measure as a log-rank P-value. Because the ClinicalTrials.gov record does not provide a confidence interval or a conventional Fine-Gray subdistribution hazard ratio for this analysis, no such quantity is inferred here.
The important statistical point is that this is a competing-risks analysis, not simply an ordinary binary comparison. In a competing-risks setting, an event that occurs first can prevent the event of interest from subsequently being observed, so the event structure must be incorporated into the analysis.
The reported value of 0.43 is not presented in the ClinicalTrials.gov record as a standard effect estimate such as a hazard ratio, so it should not be interpreted as "43% lower risk." The registry explicitly labels the reported effect measure as a log-rank P-value.
The reported P-value of 0.43 does not provide evidence that the treatment groups have identical GVHD incidence. It indicates that the reported analysis did not produce a small P-value under its specified testing framework.
The analysis is also restricted to participants who underwent allogeneic stem cell transplantation, making the analysis population different from the all-randomized population used for the long-term survival analysis.
8. Secondary Result: Long-term Survival
Long-term survival was measured at Month 3 and analyzed among all randomized subjects. The registry reports a log-rank comparison and a hazard ratio with a two-sided 95% confidence interval.
Hazard ratio for long-term survival
95% CI: 0.778–2.147 · P = 0.35
Control group = reference group.
| Endpoint | Analysis population | Method | Effect estimate | 95% CI | P-value |
|---|---|---|---|---|---|
| Long-term Survival | All randomized subjects | Log-rank test | HR 1.293 | 0.778–2.147 | 0.35 |
The hazard ratio of 1.293 compares the estimated instantaneous event rates for the Granulocyte Arm with the Control Arm, with the control group as the reference. Numerically, the estimate is above 1, so the fitted relative hazard estimate is higher in the Granulocyte Arm than in the reference group.
This does not mean that mortality was 29.3% higher, because a hazard ratio is not a relative risk or a risk ratio. It also does not imply that every participant experienced a 29.3% increase in risk.
The 95% confidence interval of 0.778–2.147 crosses 1 and is sufficiently broad to encompass both a lower and a higher hazard in the Granulocyte Arm. This illustrates why the point estimate should not be interpreted without its interval.
The P-value of 0.35 is a test statistic summary, not an effect-size measure. It should not be read as a 35% probability that the null hypothesis is true, nor as a 65% probability that the treatment works.
The ClinicalTrials.gov record identifies the log-rank test and hazard ratio but does not provide additional information needed to independently assess proportional hazards. Accordingly, the hazard-ratio interpretation should remain tied to the model used in the reported analysis.
9. Results in Context
The three posted statistical analyses answer different questions and use different analysis populations and methods. Keeping these distinctions explicit is important when interpreting a multi-endpoint clinical trial.
| Endpoint | Population | Method | Reported estimate | P-value |
|---|---|---|---|---|
| Primary: alive at Day 42 with microbial response | All adjudicated or deceased subjects in ITT analyses | Multiple logistic regression | OR 1.19; 95% CI 0.44–3.20 | 0.73 |
| Secondary: GVHD at Day 42 | Subjects with allogeneic stem cell transplantation | Fine-Gray / competing risks | Reported estimate 0.43 | 0.43 |
| Secondary: long-term survival at Month 3 | All randomized subjects | Log-rank test | HR 1.293; 95% CI 0.778–2.147 | 0.35 |
These estimates should not be collapsed into a single treatment-effect statement. The primary endpoint is binary and uses an adjusted odds ratio; GVHD is a competing-risks endpoint with a restricted analysis population; and long-term survival is a time-to-event endpoint evaluated in all randomized subjects.
10. Safety Results
The ClinicalTrials.gov record provides serious adverse-event counts by arm for the Control Arm and Granulocyte Arm.
| Arm | Participants with serious adverse events | At risk | Reported proportion |
|---|---|---|---|
| Control Arm | 25 | 58 | 25/58 |
| Granulocyte Arm | 26 | 56 | 26/56 |
The registry data therefore report 25/58 affected participants in the Control Arm and 26/56 in the Granulocyte Arm. No confidence interval, P-value, odds ratio, risk ratio, or formal between-arm statistical analysis for these serious adverse events is reported in the ClinicalTrials.gov record.
It is important not to substitute an inferred significance test for the registered results. The counts describe the observed serious-adverse-event experience in the two reported arms, but the ClinicalTrials.gov record does not establish a formal comparative hypothesis test for this safety endpoint.
11. Statistical Methods Explained
Why was logistic regression used for the primary endpoint?
The primary endpoint is binary: a participant either meets the combined Day 42 criterion of being alive and having a microbial response or does not. Logistic regression is designed for this type of outcome and models the log odds of the event as a function of treatment and covariates. The Ring Study used multiple logistic regression so that the treatment comparison could be adjusted for infection strata, underlying disease, zubrod score, respiratory symptoms, and age.
What does an odds ratio of 1.19 mean?
An odds ratio of 1.19 means that the estimated odds of the primary outcome were 1.19 times the odds in the reference group, after the specified adjustment. It does not mean a 19-percentage-point increase, a 19% increase in probability, or a 19% reduction in risk. Converting an odds ratio into an absolute probability difference requires baseline probability information.
Why does the confidence interval matter more than the point estimate alone?
The point estimate is only one possible estimate under the study's sampling process. The 95% confidence interval of 0.44–3.20 shows that the primary estimate is imprecise. It includes values below and above 1, so the observed estimate alone should not be treated as establishing a particular direction or magnitude of treatment effect.
Why was a competing-risks method used for GVHD?
When a participant can experience a competing event that prevents the event of interest from being observed in the same way, ordinary survival methods can misrepresent the probability of the event of interest. A Fine-Gray framework models the subdistribution associated with the cumulative incidence of the event while retaining competing events in the risk-set structure.
What does a hazard ratio of 1.293 mean?
For the long-term survival analysis, the hazard ratio compares the estimated instantaneous event rate in the Granulocyte Arm with that in the Control Arm. The estimate of 1.293 is above 1, but the 95% confidence interval of 0.778–2.147 includes 1. The hazard ratio therefore needs to be interpreted together with its uncertainty rather than as a standalone statement of treatment effect.
Why does a P-value not measure effect size?
A P-value describes how compatible the observed data are with a specified null hypothesis under the analysis model. It depends on the effect estimate, variability, sample information, and model structure. It does not tell us how large or clinically important an effect is. Effect measures such as odds ratios and hazard ratios, together with their confidence intervals, provide the information about estimated magnitude and precision.
Why is the analysis population important?
The primary analysis uses an intention-to-treat framework, while the GVHD analysis is limited to subjects who had allogeneic stem cell transplantation. These populations answer different questions. Restricting an analysis after randomization can change the interpretation because the resulting subgroup may no longer retain the same randomization-based comparability as the full randomized population.
12. Intention-to-Treat Analysis
The primary statistical analysis was conducted in an intention-to-treat population described by the registry as all adjudicated or deceased subjects in intention-to-treat analyses. The central principle is that participants remain associated with their randomized treatment assignment for the primary comparison.
Why ITT is valuable
Randomization creates the basis for a causal comparison. Preserving assignment in the primary analysis helps maintain that randomized structure even when treatment exposure changes.
What ITT does not solve
ITT does not automatically eliminate missing outcome information, nonadherence, treatment switching, or other sources of uncertainty. Those issues still require appropriate design and analysis decisions.
For this trial, the ClinicalTrials.gov record does not describe a separate imputation strategy for missing primary-endpoint observations. Therefore, no specific missing-data method should be attributed to the trial beyond what is explicitly contained in the ClinicalTrials.gov record.
13. Covariate Adjustment
The primary logistic regression model adjusted for five categories of baseline or clinical information: infection strata, underlying disease, zubrod score, respiratory symptoms, and age.
| Covariate included in primary model | Role in the analysis |
|---|---|
| Infection strata | Adjustment variable specified in the reported logistic model. |
| Underlying disease | Adjustment variable specified in the reported logistic model. |
| Zubrod score | Adjustment variable reflecting the participant characteristic specified in the registry model. |
| Respiratory symptoms | Adjustment variable specified in the reported logistic model. |
| Age | Adjustment variable specified in the reported logistic model. |
Covariate adjustment does not change the fact that treatment assignment was randomized. Rather, it changes the statistical model used to estimate the treatment association by conditioning on the listed variables.
The reported OR of 1.19 is therefore not equivalent to an unadjusted odds ratio calculated directly from two observed event proportions.
14. Time-to-Event Analysis
Two of the posted analyses concern time-to-event outcomes: graft versus host disease and long-term survival. These require a different statistical framework from the binary primary endpoint.
Kaplan-Meier estimation
Kaplan-Meier estimation is commonly used to describe time-to-event distributions when some participants are censored. Rather than reducing each participant to a simple yes/no outcome at a fixed time, the method uses the timing of events and censoring throughout follow-up.
Here, di represents events at a given event time and ni represents the number at risk immediately before that time.
The ClinicalTrials.gov record specifically report a log-rank method for the long-term survival analysis, but they do not provide Kaplan-Meier estimates or a survival curve. Accordingly, no median survival or time-specific survival probability is inferred.
Log-rank testing
The log-rank test compares the survival experience of randomized groups over follow-up. Its null hypothesis is generally framed around equality of the survival distributions under the specified testing framework.
Hazard ratio
The long-term survival analysis reports a hazard ratio of 1.293. A hazard ratio summarizes relative instantaneous event rates under the fitted time-to-event model. It should not be translated into a single absolute risk difference without additional information.
15. Competing Risks and GVHD
The GVHD endpoint illustrates why event definitions matter. The registry identifies the endpoint as time-to-event, restricts the population to subjects who had allogeneic stem cell transplantation, and specifies competing-risks methodology.
Event of interest
Graft versus host disease among recipients of allogeneic stem cell transplantation.
Analysis framework
Fine-Gray / competing risks, with the registry also reporting a log-rank P-value as the effect measure.
A competing-risks analysis is particularly important when the occurrence of another event changes whether or how the event of interest can subsequently occur. Treating all other events simply as ordinary censoring can answer a different question from the cumulative incidence of the event of interest.
The ClinicalTrials.gov record does not provide enough detail to reconstruct the complete competing-event structure, cumulative-incidence curves, or a Fine-Gray subdistribution hazard ratio. Those quantities are therefore not added here.
16. Statistical Interpretation of the Primary Endpoint
The primary adjusted odds ratio was 1.19. Relative to the Control Arm, the Granulocyte Arm had an estimated odds of the Day 42 combined outcome that was 1.19 times as high after adjustment for the specified covariates.
The 95% confidence interval was 0.44–3.20. The breadth of this interval indicates substantial uncertainty around the point estimate. The interval includes values below 1 and above 1, so the estimate alone does not establish a particular treatment direction.
The two-sided P-value was 0.73. This is evidence summary under the specified null model, not a probability that the treatment is ineffective and not a measure of effect size.
The primary endpoint combines two conditions: survival at Day 42 and microbial response. Consequently, the endpoint does not isolate mortality, microbiological clearance, or symptomatic improvement individually. A participant must satisfy the registered combined criterion to count as a primary-endpoint success.
17. Multiplicity and Hypothesis Testing
The ClinicalTrials.gov record identifies a single registered primary endpoint and three posted statistical analyses: one primary analysis and two secondary analyses. The primary and secondary analyses should therefore not be treated as interchangeable tests.
| Analysis | Role | Hypothesis type | Interpretive issue |
|---|---|---|---|
| Day 42 survival with microbial response | Primary | Superiority | Primary confirmatory question specified by the registry. |
| GVHD at Day 42 | Secondary | Superiority | Restricted analysis population and competing-risks framework. |
| Long-term survival at Month 3 | Secondary | Superiority | Time-to-event comparison using the log-rank test and hazard ratio. |
The ClinicalTrials.gov record does not specify an alpha-allocation strategy, multiplicity adjustment procedure, or hierarchical testing sequence. Therefore, no particular familywise-error interpretation is assigned to the collection of secondary P-values.
18. Crossover, Factorial Design, and Bayesian Methods
Crossover
The ClinicalTrials.gov record does not report a crossover design or a crossover analysis.
Factorial design
The registry identifies the design model as parallel, not factorial.
Bayesian methods
No Bayesian statistical method is reported in the statistical analyses posted on ClinicalTrials.gov.
Interim analysis
The ClinicalTrials.gov record does not report an interim-analysis method or alpha-spending procedure.
These distinctions matter because methods such as crossover, factorial randomization, Bayesian updating, or group-sequential monitoring can materially change how treatment effects and error rates are interpreted. They should not be attributed to the trial without registry support.
19. What the Odds Ratio Does — and Does Not — Mean
An odds ratio of 1.19 means that the estimated odds of the primary binary outcome in the Granulocyte Arm were 1.19 times those in the Control Arm, conditional on the adjustment variables included in the logistic regression model.
It does not mean that the Granulocyte Arm increased the probability of success by 19%, nor does it imply a 19-percentage-point difference.
The 95% CI of 0.44–3.20 describes uncertainty around the estimated odds ratio. It is not a range containing 95% of individual participant treatment effects.
The P-value of 0.73 does not measure the magnitude of the treatment effect. It summarizes the compatibility of the observed data with the specified null hypothesis under the model.
An odds ratio is a relative measure. To understand absolute clinical impact, one would ideally also examine the underlying event probabilities or risk differences. The ClinicalTrials.gov record does not provide those primary-endpoint percentages, so they are not reconstructed here.
20. Why This Trial Matters Statistically
The Ring Study is a useful teaching example because its registry record connects several fundamental clinical-trial methods within a relatively compact set of analyses.
| Concept | How it appears in The Ring Study |
|---|---|
| Randomization | The trial is randomized with a parallel design and 3 arms. |
| Binary endpoint analysis | The primary endpoint is measured at Day 42 and is analyzed using logistic regression. |
| Odds ratio | The primary treatment effect is reported as an adjusted OR of 1.19. |
| Confidence interval | The primary OR has a two-sided 95% CI of 0.44–3.20. |
| Covariate adjustment | The primary model adjusts for infection strata, underlying disease, zubrod score, respiratory symptoms, and age. |
| Intention-to-treat | The primary analysis population is described as ITT. |
| Competing risks | GVHD is analyzed using a Fine-Gray / competing-risks framework. |
| Log-rank test | Long-term survival is compared using the log-rank method. |
| Hazard ratio | Long-term survival is summarized by HR 1.293 with a 95% CI of 0.778–2.147. |
| Safety analysis | Serious adverse-event counts are reported by arm. |
From a statistical-learning perspective, the trial is especially useful because it demonstrates that the correct method depends on the endpoint structure. A Day 42 binary outcome calls for a categorical-data framework, while survival and GVHD endpoints require methods that explicitly account for event timing and, for GVHD, competing events.
21. Important Limitations and Interpretation Issues
- Wide primary confidence interval: the 95% CI of 0.44–3.20 indicates considerable uncertainty around the adjusted odds ratio.
- Binary endpoint: the primary endpoint combines survival and microbial response, so the effect estimate does not separately identify the treatment association with each component.
- Odds ratio interpretation: the OR should not be interpreted as a risk ratio or percentage-point difference in probability.
- Restricted GVHD population: the GVHD analysis is limited to subjects who had allogeneic stem cell transplantation and therefore differs from the all-randomized analysis population.
- Competing risks: GVHD incidence requires attention to competing events; ordinary survival methods can answer a different estimand.
- Secondary endpoints: the ClinicalTrials.gov record does not specify a multiplicity-adjustment strategy for the secondary analyses.
- Missing-data methodology: the ClinicalTrials.gov record does not report a specific imputation method for the primary endpoint.
- Proportional hazards: the long-term survival analysis reports a hazard ratio, but the ClinicalTrials.gov record does not provide information for independently evaluating the proportional-hazards assumption.
- Model specification: the primary analysis is adjusted for several covariates, but the ClinicalTrials.gov record does not provide model diagnostics or sensitivity analyses.
- Third study arm: the registry identifies 3 arms overall, while the statistical analyses posted on ClinicalTrials.gov focus on the Control Arm versus the Granulocyte Arm. No additional arm-specific statistical comparison is inferred.
- Safety comparison: serious adverse-event counts are available, but no formal comparative safety statistic is reported in the ClinicalTrials.gov record.
22. Statistical Interpretation vs Clinical Interpretation
Statistical interpretation
The primary adjusted OR was 1.19 with a 95% CI of 0.44–3.20 and P = 0.73. The registry also reports secondary time-to-event analyses for GVHD and long-term survival.
Clinical interpretation
The registry data describe the treatment comparisons and their uncertainty, but the ClinicalTrials.gov record does not provide enough detail to translate the primary OR into an absolute treatment benefit or to construct additional clinical-effect measures.
This distinction is important. Statistical evidence describes an estimate, its uncertainty, and the compatibility of the data with a hypothesis. Clinical interpretation additionally depends on the magnitude of the effect, absolute event rates, harms, patient priorities, and the broader evidence base. The ClinicalTrials.gov record does not support filling in those missing quantities.
23. Trial Timeline
Study start
The Ring Study began enrollment in April 2008.
Primary completion
The registry lists April 2013 as the primary completion date.
Trial status
The registry classifies the study as completed, with 114 participants enrolled.
24. Learning Pathway From This Trial
The Ring Study connects naturally to several core statistical concepts. Learn more about the methods used in this trial:
25. Related Statistical Calculators
26. Sources
- ClinicalTrials.gov: The Ring Study — NCT00627393.
- Linked publication: PubMed record for PMID 26333778.
Continue with the statistical methods
Use the related tutorials and calculators to explore the regression, survival-analysis, competing-risks, effect-measure, and inference concepts illustrated by The Ring Study.
27. Record Summary
The Ring Study provides a compact example of how endpoint structure determines statistical methodology. Its primary endpoint is a binary Day 42 outcome combining survival and microbial response, analyzed with adjusted logistic regression and reported as an odds ratio of 1.19 with a two-sided 95% CI of 0.44–3.20 and P = 0.73. The trial also reports a competing-risks analysis of GVHD and a log-rank analysis of long-term survival with an HR of 1.293 (95% CI 0.778–2.147; P = 0.35).
The most important statistical lesson is that these estimates cannot be interpreted independently of their endpoint definitions and analysis populations. The primary odds ratio is not a risk ratio; the GVHD statistic is not a conventional hazard ratio; and the long-term survival hazard ratio is not an absolute survival difference. Confidence intervals, P-values, covariate adjustment, competing events, censoring, and analysis populations all contribute to the correct interpretation.