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
DIRECT was a randomized, parallel-group, phase 3 trial evaluating candesartan versus placebo in type 2 diabetes. The trial used quadruple masking and enrolled 4,717 participants. Its registered primary endpoint was the number of participants with a 3-step or greater increase in EDTRS Severity Scale from baseline to the end of the study.
| Feature | DIRECT |
|---|---|
| Phase | Phase 3 |
| Condition | Type 2 Diabetes |
| Brief title | DIabetic Retinopathy Candesartan Trials |
| Design | Randomized, parallel |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 4,717 |
| Intervention | Candesartan |
| Comparator | Placebo |
| Lead sponsor | AstraZeneca |
| Sponsor type | Industry |
| ClinicalTrials.gov | NCT00252694 |
2. Clinical Question
The primary clinical question was whether participants randomized to candesartan differed from those randomized to placebo with respect to the time to a 3-step or greater increase in the EDTRS Severity Scale during the study.
Population
Participants with type 2 diabetes enrolled in a phase 3 randomized clinical trial.
Intervention
Candesartan.
Comparator
Placebo.
Primary question
Is the time-to-event experience for a 3-step or greater increase in EDTRS Severity Scale different between candesartan and placebo?
3. Trial Design
Study timeline
Trial start
The DIRECT trial began in August 2001.
Longitudinal assessment
The primary endpoint was assessed from baseline to the end of study, defined as 5 years, with visits after a half year, one year, and thereafter once per year.
Primary completion
The registry lists February 2008 as the primary completion date.
Study arms
Candesartan
- Intervention: candesartan
- Randomized comparison group
- Evaluated against placebo for the registered primary endpoint
Placebo
- Comparator: placebo
- Randomized comparison group
- Evaluated against candesartan for the registered primary endpoint
4. Endpoints
The registered primary endpoint concerns progression on the EDTRS Severity Scale. The registry defines the event using a 3-step or greater increase and specifies repeated follow-up through the 5-year end-of-study period.
| Endpoint | Registered definition | Time frame |
|---|---|---|
| Primary endpoint | Number of participants with a 3-step or greater increase in Early Treatment of Diabetic Retinopathy Study (EDTRS) Severity Scale. | From baseline to end of study, i.e. 5 years, with visits after a half year, one year and thereafter one visit per year. |
What counts as a 3-step increase?
The registry defines a 3-step change as either a 1-step change in one eye and a 2-step change in the other eye, or a 3-step change in one eye only. The EDTRS scale is described as an 11-step scale ranging from 1 to 11.
5. Statistical Methodology
Intention-to-treat analysis
The posted primary analysis used the intention-to-treat population, defined in the registry as all randomized patients with any post-randomization data. Participants therefore remained associated with their randomized treatment group for the efficacy comparison.
Generalized log-rank test
The registered endpoint definition states that a generalized log-rank test was used to test the difference between treatments. The posted statistical analysis identifies the method as a log-rank analysis and places it within the survival-analysis family.
The log-rank framework compares the observed timing of events across randomized groups while accounting for follow-up information contributed before an event or censoring.
Hazard ratio
The treatment effect was summarized using a hazard ratio. The reported estimate comparing candesartan with placebo was 0.870, with a two-sided 95% confidence interval of 0.704 to 1.076.
An HR of 0.870 corresponds to an estimated instantaneous event rate about 13.0% lower in the candesartan group relative to placebo, under the hazard-ratio interpretation. It is not a 13.0% absolute reduction in the proportion of participants experiencing the endpoint.
Superiority framework
The posted analysis identifies the hypothesis type as superiority. The inferential question is therefore whether the randomized groups differ, rather than whether candesartan satisfies a prespecified non-inferiority margin.
6. Primary Result
The registry reports one formal statistical analysis for the primary endpoint. The analysis used the intention-to-treat population and compared candesartan with placebo using a generalized log-rank approach.
Time to a 3-step or greater EDTRS increase
95% CI: 0.704–1.076 · P = 0.1994
Analysis population: intention-to-treat population with any post-randomization data.
| Primary endpoint | Candesartan vs placebo |
|---|---|
| Endpoint | 3-step or greater increase in EDTRS Severity Scale |
| Time frame | Baseline to end of study, i.e. 5 years |
| Analysis population | Intention-to-treat |
| Method | Generalized log-rank / log-rank |
| Effect measure | Hazard ratio |
| Hazard ratio | 0.870 |
| 95% confidence interval | 0.704–1.076 |
| P-value | 0.1994 |
| Hypothesis | Superiority |
The estimated hazard ratio of 0.870 indicates a lower estimated instantaneous rate of the qualifying EDTRS event in the candesartan group relative to placebo, corresponding to approximately a 13.0% lower estimated hazard under the hazard-ratio interpretation.
The estimate does not mean that 13.0% fewer participants experienced a 3-step increase, nor does it describe an absolute risk difference. A hazard ratio is a relative time-to-event measure.
The 95% confidence interval of 0.704 to 1.076 describes statistical uncertainty around the estimated hazard ratio. Because the interval includes 1, the data are compatible with a range of relative effects that includes no hazard difference as well as effects in either direction.
The p-value of 0.1994 quantifies the compatibility of the observed test statistic with the null hypothesis under the specified statistical framework. It does not measure the size, clinical importance, or probability of the treatment effect.
The analysis is based on time-to-event methodology, so interpretation also depends on the handling of censoring and on the appropriateness of summarizing the treatment contrast with a hazard ratio. The registry does not report detailed censoring rules or a separate assessment of the proportional-hazards assumption.
7. Reading the Confidence Interval
The confidence interval is particularly important because the point estimate alone gives an incomplete picture of the statistical evidence.
An HR below 1 favors the candesartan direction for this endpoint, whereas an HR above 1 favors the placebo direction. The reported interval extends from below 1 to above 1, so the interval includes both a lower event hazard and a higher event hazard relative to placebo.
The point estimate of 0.870 is one estimate from the observed trial data. The confidence interval communicates how precisely that effect was estimated. An interpretation based only on the point estimate would ignore this uncertainty.
8. What the P-value Means
The primary analysis reports P = 0.1994. Under the superiority framework and the specified log-rank analysis, this is the probability, under the null model, of obtaining a test statistic at least as extreme as the observed statistic in the two-sided testing framework.
What it measures
Compatibility of the observed test statistic with the null hypothesis under the statistical model and test procedure.
What it does not measure
It is not the probability that candesartan has no effect, and it is not a measure of the magnitude or clinical importance of the observed hazard ratio.
The p-value should therefore be interpreted together with the hazard ratio and its confidence interval. Here, the HR provides the estimated direction and magnitude of the relative time-to-event effect, while the confidence interval describes its uncertainty and the p-value summarizes the corresponding hypothesis test.
9. Statistical Methods Explained
Why was a log-rank test used?
The primary endpoint is analyzed as a time-to-event outcome. A log-rank test is designed to compare survival or event-time distributions between groups while incorporating the timing of observed events and follow-up information from participants who have not yet experienced the event.
What does a hazard ratio of 0.870 mean?
It means that the estimated instantaneous rate of the qualifying EDTRS event was 0.870 times that of placebo, or approximately 13.0% lower under the hazard-ratio interpretation. It does not mean that the cumulative event probability was exactly 13.0% lower.
Why is the confidence interval important?
The 95% CI of 0.704–1.076 shows that the estimate has uncertainty extending on both sides of the null value of 1. The interval therefore provides information that the point estimate alone cannot: the observed data do not pin down a single precise relative effect.
Why does the p-value not measure effect size?
A p-value is determined by the test statistic and its reference distribution. It addresses evidence against a null hypothesis, not the magnitude of the treatment effect. Effect size is described here by the hazard ratio, while precision is described by the confidence interval.
Why was an intention-to-treat population used?
The registry defines the analysis population as all randomized patients with any post-randomization data. Maintaining randomized treatment assignment in the efficacy analysis preserves the treatment comparison created by randomization and avoids redefining treatment groups based on events that occur after randomization.
How should the EDTRS event definition be understood?
The event is not simply any change in the EDTRS scale. The registry specifies a threshold of 3 steps or greater, achieved either through a 1-step change in one eye plus a 2-step change in the other, or through a 3-step change in one eye alone. This prespecified threshold converts longitudinal changes in the severity scale into the primary event definition.
Why does the five-year follow-up matter statistically?
The registered time frame extends from baseline to the end of study, defined as 5 years, with assessments after a half year, one year, and annually thereafter. This repeated follow-up allows the primary analysis to use the timing of qualifying events rather than only whether an event was ever observed.
10. Safety Results
The registry reports serious adverse events by randomized treatment group as affected participants divided by participants at risk.
| Safety measure | Candesartan | Placebo |
|---|---|---|
| Serious adverse events | 301 / 949 | 267 / 953 |
These figures show the reported number of participants affected by serious adverse events relative to the corresponding number at risk. They are safety counts rather than the primary efficacy endpoint and should therefore be interpreted separately from the hazard ratio for EDTRS progression.
The serious-adverse-event figures are presented as affected participants over participants at risk. The primary efficacy analysis, by contrast, is a time-to-event comparison using a generalized log-rank method. The two summaries answer different statistical questions and should not be substituted for one another.
11. What the Registry Reports — and What It Does Not
The ClinicalTrials.gov record reports one registered primary endpoint and one posted formal statistical analysis for that endpoint. It also reports four posted outcome measures overall.
| Information category | Registry reporting |
|---|---|
| Primary endpoint | Reported |
| Formal primary statistical analysis | Reported |
| Primary hazard ratio | Reported |
| 95% confidence interval | Reported |
| P-value | Reported |
| Intention-to-treat analysis population | Reported |
| Serious adverse events by arm | Reported |
| Detailed baseline characteristics | Not reported in the ClinicalTrials.gov record |
| Median event time | Not reported in the ClinicalTrials.gov record |
| Subgroup estimates | Not reported in the ClinicalTrials.gov record |
| Detailed censoring or imputation procedures | Not reported in the ClinicalTrials.gov record |
The absence of these additional statistical details matters when interpreting the scope of the reported result. The primary analysis can be understood from the reported endpoint, population, method, hazard ratio, confidence interval, and p-value, but a fuller reconstruction of the underlying event-time distribution would require additional information.
12. Important Statistical Limitations
- Event-time detail: the registry does not report the underlying individual event and censoring times, so the complete Kaplan-Meier experience cannot be reconstructed from the posted summary alone.
- Median event time: the registry does not report a median time to the primary event.
- Baseline characteristics: detailed baseline distributions are not reported in the registry information summarized here.
- Subgroups: subgroup-specific treatment-effect estimates are not reported in the registry record.
- Censoring details: the posted record identifies a time-to-event analysis but does not provide detailed censoring rules.
- Proportional-hazards interpretation: the hazard ratio is a relative time-to-event measure, and the registry does not report a separate assessment of whether the proportional-hazards assumption was satisfied.
- Safety denominators: serious adverse-event counts are reported with the corresponding numbers at risk, but the registry summary does not provide a broader detailed safety table in this record.
13. Why This Trial Matters Statistically
DIRECT is a useful statistical case because its primary endpoint combines a clinically defined change in a severity scale with longitudinal follow-up. The analysis illustrates how a binary event definition can still be analyzed as a time-to-event outcome when the timing of the event is part of the information.
| Concept | How it appears in DIRECT |
|---|---|
| Randomization | Participants were randomized to candesartan or placebo. |
| Parallel-group design | The registry identifies the design model as parallel. |
| Quadruple masking | The trial is registered as quadruple masked. |
| Intention-to-treat analysis | The primary analysis included all randomized patients with any post-randomization data. |
| Time-to-event endpoint | The posted analysis identifies the primary endpoint as time-to-event. |
| Log-rank test | The treatment comparison used a generalized log-rank approach. |
| Hazard ratio | The treatment effect was reported as HR 0.870. |
| Confidence interval | The 95% CI was 0.704–1.076. |
| P-value | The two-sided p-value was 0.1994. |
| Superiority testing | The posted hypothesis type was superiority. |
14. Interpreting the Primary Result in Context
The estimated hazard ratio of 0.870 points toward a lower rate of the qualifying EDTRS event with candesartan compared with placebo. The corresponding 95% CI extends from 0.704 to 1.076.
Because the confidence interval includes 1, the interval includes a null relative hazard as well as estimates in both directions. The p-value of 0.1994 is consistent with that uncertainty under the specified superiority test.
The endpoint is specifically a 3-step or greater increase in EDTRS Severity Scale, defined using changes across one or both eyes. The hazard ratio therefore pertains to time to this prespecified retinal-severity event, not to all measures of diabetic retinopathy or to a general measure of diabetes control.
The reported hazard ratio is a relative measure. Absolute event probabilities, risk differences, and numbers needed to treat require corresponding cumulative event information, which is not reported in the registry summary used here.
15. Primary Analysis at a Glance
| Component | DIRECT primary analysis |
|---|---|
| Endpoint | 3-step or greater increase in EDTRS Severity Scale |
| Time frame | Baseline to end of study, i.e. 5 years |
| Population | Intention-to-treat population with any post-randomization data |
| Comparison | Candesartan vs placebo |
| Statistical method | Log-rank; generalized log-rank |
| Effect measure | Hazard ratio |
| Estimate | 0.870 |
| 95% CI | 0.704–1.076 |
| P-value | 0.1994 |
| Hypothesis type | Superiority |
16. Related Tutorials
Learn more about the methods used in this trial:
17. Related Calculators
18. Sources
- ClinicalTrials.gov: NCT00252694.
- PubMed: PubMed record 38682786.
- PubMed: PubMed record 21602709.
- PubMed: PubMed record 21309844.
- PubMed: PubMed record 21225239.
- PubMed: PubMed record 19451554.
Continue through the Clinical Biostats statistical methods
Explore the survival-analysis, clinical-trial, and statistical-inference concepts that underlie randomized time-to-event analyses.
19. Record Summary
DIRECT was a randomized, parallel, quadruple-masked phase 3 trial of candesartan versus placebo in 4,717 participants with type 2 diabetes. Its primary endpoint was a 3-step or greater increase in the EDTRS Severity Scale from baseline to the end of a 5-year study period, with scheduled assessments after a half year, one year, and annually thereafter.
The posted primary analysis used an intention-to-treat population and a generalized log-rank approach. The reported hazard ratio was 0.870, with a two-sided 95% confidence interval of 0.704–1.076 and a p-value of 0.1994. Statistically, the estimate points toward a lower instantaneous rate of the qualifying event with candesartan, while the confidence interval indicates substantial uncertainty around that estimate and includes the null value of 1.
The most important statistical lesson is that a time-to-event endpoint should be interpreted through all three components of the reported result: the hazard ratio for relative effect, the confidence interval for precision, and the p-value for the specified hypothesis test. The endpoint definition, intention-to-treat population, follow-up schedule, and generalized log-rank method are equally important for understanding what the reported estimate actually represents.