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
VIEW1 was a randomized, quadruple-masked, phase 3, parallel-group trial in ophthalmology that evaluated aflibercept injection (VEGF Trap-Eye) against ranibizumab in patients with macular degeneration. The registry reports 1,217 enrolled participants, four treatment groups, one registered binary primary endpoint, five posted outcome measures, and 15 posted statistical analyses.
| Feature | VIEW1 |
|---|---|
| Phase | Phase 3 |
| Condition | Macular Degeneration |
| Therapeutic area | Ophthalmology |
| Design | Randomized, parallel-group |
| Masking | Quadruple |
| Primary purpose | Treatment |
| Enrollment | 1,217 |
| Arms | 4 |
| Trial status | Completed |
| Start | August 2007 |
| Primary completion | September 2010 |
| Lead sponsor | Regeneron Pharmaceuticals |
2. Clinical Question
The central statistical question was whether the proportions of patients maintaining vision at Week 52 were sufficiently similar between ranibizumab 0.5 mg every 4 weeks and each of three aflibercept injection regimens to satisfy the prespecified non-inferiority framework.
Population
The registry identifies the condition as macular degeneration and describes the primary endpoint in terms of ETDRS letter score and maintenance of vision.
Reference treatment
Ranibizumab 0.5 mg Q4 was the reference group for each of the three posted primary comparisons.
Investigational regimens
The primary comparisons evaluated IAI 2.0 mg Q4, IAI 0.5 mg Q4, and IAI 2.0 mg Q8 against ranibizumab 0.5 mg Q4.
Primary question
Is the risk difference in the proportion maintaining vision compatible with non-inferiority under the prespecified 10% margin?
3. Trial Design
Ranibizumab 0.5 mg Q4
- Biological intervention
- Compared with each of the three IAI regimens
- Primary endpoint analyses used the PPS population
- Secondary endpoint analyses used the FAS population
Aflibercept injection
- IAI 2.0 mg Q4
- IAI 0.5 mg Q4
- IAI 2.0 mg Q8
- Each regimen was compared with ranibizumab 0.5 mg Q4 in the posted primary analyses
The registry does not report an allocation ratio among the four groups in the information summarized here. The statistical analyses instead identify the treatment groups explicitly for each pairwise comparison.
4. Primary Endpoint
| Endpoint | Registry definition | Time frame | Type |
|---|---|---|---|
| Percentage of Patients Who Maintained Vision at Week 52 - Last Observation Carried Forward (LOCF) | Maintenance of vision was defined as patients who lost fewer than 15 letters in Early Treatment Diabetic Retinopathy Study (ETDRS) letter score compared to baseline. | Baseline and at week 52 | Binary |
The primary endpoint is therefore a dichotomized visual-acuity outcome. Each patient is classified according to whether the Week 52 ETDRS letter score, using LOCF, represents a loss of fewer than 15 letters compared with baseline.
The analysis compares the proportion meeting this binary criterion between ranibizumab 0.5 mg Q4 and each IAI regimen.
5. Statistical Methodology
Risk difference for the binary primary endpoint
The primary endpoint was analyzed as a risk difference. For each pairwise comparison, the registry reports the difference between the proportion of patients maintaining vision in the ranibizumab group and the corresponding proportion in the IAI group.
The direction of the reported difference is important. A positive value favors IAI 2.0Q4 in the first comparison, whereas the registry's analysis notes describe the direction explicitly for each IAI regimen.
Wald / z-test and normal approximation
The registry identifies a Wald / z-test for all three primary analyses. The analysis notes describe a conditional sequence of calculations for the 95.1% confidence interval using a normal approximation to the difference between the proportions.
Per-protocol analysis population
The primary endpoint analyses used the PPS population. This matters because non-inferiority trials commonly place particular emphasis on adherence to the protocol: substantial deviations can make treatment groups appear more similar than they would under a strict randomized comparison. For this reason, the analysis population is part of the interpretation of the non-inferiority result rather than a minor technical detail.
Non-inferiority margin
The registry reports a non-inferiority margin of 10%. The trial protocol's sample-size estimation was reported to have 90% power. The registry also records a separate FDA communication stating that the 10% margin would be used for non-inferiority, while a 5% margin would be used to assess clinical equivalence.
Conditional sequence and multiplicity
The primary analysis notes describe adjustment for multiple comparisons through a conditional sequence of statistical hypotheses, with alpha = 0.049. The 95.1% confidence intervals correspond to this stated alpha level and are therefore not interchangeable with an unqualified 95% interval for the same inferential procedure.
ANCOVA for continuous secondary endpoints
Several secondary endpoints were continuous outcomes measured at Week 52. These were analyzed with ANCOVA using treatment group as a fixed factor across all four treatment groups and the baseline measurement as a covariate. Pairwise treatment contrasts were then used to estimate differences in least-squares means.
The baseline covariate adjusts the comparison for the corresponding baseline measurement and generally improves the precision of treatment-group comparisons when baseline values explain part of the outcome variation.
Chi-squared analysis for a secondary binary endpoint
The percentage of patients who gained at least 15 letters was analyzed with a chi-squared test. The reported effect measure was again a risk difference, with two-sided 95.1% confidence intervals and superiority as the hypothesis type.
6. Primary Results
The registry reports three formal analyses of the primary Week 52 maintained-vision endpoint. All three used the PPS population, a Wald / z-test, a risk difference, and a two-sided 95.1% confidence interval. The comparisons differ only in the IAI regimen evaluated against ranibizumab 0.5 mg Q4.
6.1 Ranibizumab 0.5 mg Q4 vs IAI 2.0 mg Q4
Risk difference in maintained vision
95.1% CI: −4.4 to 3.1
PPS population · Wald / z-test · Non-inferiority / equivalence framework
The reported risk difference was −0.7 percentage points, calculated as ranibizumab minus IAI 2.0 mg Q4. The confidence interval extends from −4.4 to 3.1 percentage points.
The point estimate of −0.7 means that the estimated difference in the proportion maintaining vision was 0.7 percentage points lower for ranibizumab than for IAI 2.0 mg Q4 under the registry's stated direction of calculation. Equivalently, the point estimate is close to zero, so the estimated difference between groups was small on this scale.
The 95.1% CI of −4.4 to 3.1 is the more important quantity for the non-inferiority question. It describes the range of treatment differences compatible with the stated statistical framework and shows that the interval remains well within the reported 10% non-inferiority margin on the relevant side of the comparison.
This does not mean that the two regimens are identical for every patient, nor does it establish that their effects are exactly equal. The result concerns the prespecified binary Week 52 endpoint and the PPS analysis population.
No p-value is reported for this primary analysis in the registry summary. A p-value, even when available, would not measure the magnitude of the treatment difference; the risk difference and its confidence interval describe the effect and its statistical precision.
6.2 Ranibizumab 0.5 mg Q4 vs IAI 0.5 mg Q4
Risk difference in maintained vision
95.1% CI: −5.1 to 2.1
PPS population · Wald / z-test · Non-inferiority / equivalence framework
The reported risk difference was −1.5 percentage points, calculated as ranibizumab minus IAI 0.5 mg Q4. The confidence interval ranges from −5.1 to 2.1 percentage points.
The point estimate of −1.5 indicates a small estimated difference in the direction specified by the registry, with the ranibizumab proportion estimated 1.5 percentage points below the IAI 0.5 mg Q4 proportion.
The 95.1% CI of −5.1 to 2.1 expresses uncertainty around that point estimate. Importantly for a non-inferiority assessment, the lower boundary is still inside the prespecified 10% margin described in the registry.
The confidence interval does not mean that all individual patients have treatment effects lying between −5.1 and 2.1 percentage points. It describes uncertainty around the population-level risk difference estimated from this trial.
The absence of a reported p-value should not be interpreted as an absence of statistical evidence. Non-inferiority is evaluated against the prespecified margin and inferential procedure, rather than by treating a conventional superiority p-value as the primary criterion.
6.3 Ranibizumab 0.5 mg Q4 vs IAI 2.0 mg Q8
Risk difference in maintained vision
95.1% CI: −4.5 to 3.1
PPS population · Wald / z-test · Non-inferiority / equivalence framework
The reported risk difference was −0.7 percentage points, calculated as ranibizumab minus IAI 2.0 mg Q8. The confidence interval ranges from −4.5 to 3.1 percentage points.
The point estimate is −0.7 percentage points, indicating a small estimated difference in the direction defined by the registry's calculation.
The 95.1% CI of −4.5 to 3.1 is relatively narrow compared with the 10% non-inferiority margin. The interval therefore provides substantially more information than the point estimate alone: it limits the range of population-level differences compatible with the analysis.
This result does not establish equality in every clinical outcome. It addresses the registered Week 52 maintained-vision endpoint, using LOCF and the PPS population, under the stated non-inferiority framework.
Again, the p-value is not the measure of treatment magnitude. The risk difference gives the estimated absolute difference in proportions, while the confidence interval communicates its statistical precision and relevance to the non-inferiority margin.
6.4 Primary endpoint comparison
| Comparison | Risk difference | 95.1% CI | Population | Method |
|---|---|---|---|---|
| Ranibizumab 0.5 mg Q4 vs IAI 2.0 mg Q4 | −0.7 | −4.4 to 3.1 | PPS | Wald / z-test |
| Ranibizumab 0.5 mg Q4 vs IAI 0.5 mg Q4 | −1.5 | −5.1 to 2.1 | PPS | Wald / z-test |
| Ranibizumab 0.5 mg Q4 vs IAI 2.0 mg Q8 | −0.7 | −4.5 to 3.1 | PPS | Wald / z-test |
The three confidence intervals all lie within the reported 10% non-inferiority margin. The registry describes the primary inferential procedure as a conditional sequence of hypotheses with alpha = 0.049, rather than three unrelated hypothesis tests considered independently.
7. Secondary Results: Best Corrected Visual Acuity
Mean change from baseline in Best Corrected Visual Acuity (BCVA), measured by ETDRS letter score at Week 52 using LOCF, was analyzed in the FAS population using ANCOVA. Treatment group was included as a fixed factor across all four groups and baseline measurement was included as a covariate.
| Comparison | Difference in LS means | 95.1% CI | P-value | Direction |
|---|---|---|---|---|
| IAI 2.0 mg Q4 vs ranibizumab 0.5 mg Q4 | 3.15 | 0.92 to 5.37 | 0.0054 | IAI 2.0 mg Q4 |
| IAI 0.5 mg Q4 vs ranibizumab 0.5 mg Q4 | 0.80 | −3.03 to 1.43 | 0.4793 | IAI 0.5 mg Q4 |
| IAI 2.0 mg Q8 vs ranibizumab 0.5 mg Q4 | 0.26 | −1.97 to 2.49 | 0.8179 | IAI 2.0 mg Q8 |
The registry defines these differences as IAI minus ranibizumab. Thus, positive estimates favor the corresponding IAI regimen for the mean change in ETDRS letter score.
The 3.15-letter least-squares mean difference for IAI 2.0 mg Q4 versus ranibizumab 0.5 mg Q4 was accompanied by a 95.1% CI of 0.92 to 5.37 and a reported p-value of 0.0054. The interval lies above zero, so the estimated difference was positive throughout the reported confidence interval.
For IAI 0.5 mg Q4, the estimate was 0.80, with a 95.1% CI of −3.03 to 1.43 and p = 0.4793. For IAI 2.0 mg Q8, the estimate was 0.26, with a 95.1% CI of −1.97 to 2.49 and p = 0.8179.
The p-values address evidence against the null hypothesis that the two adjusted mean changes are equal. They do not tell us how large a treatment difference is. The estimated least-squares mean difference and its confidence interval provide that information.
8. Secondary Results: Patients Gaining at Least 15 Letters
The registry also reports the percentage of patients who gained at least 15 letters of vision in the ETDRS letter score in the study eye at Week 52 using LOCF. These analyses used the FAS population and chi-squared testing, with risk difference as the effect measure.
| Comparison | Risk difference | 95.1% CI | P-value |
|---|---|---|---|
| IAI 2.0 mg Q4 vs ranibizumab 0.5 mg Q4 | 6.6 | −1.0 to 14.1 | 0.1042 |
| IAI 0.5 mg Q4 vs ranibizumab 0.5 mg Q4 | −6.0 | −13.2 to 1.2 | 0.1037 |
| IAI 2.0 mg Q8 vs ranibizumab 0.5 mg Q4 | −0.4 | −7.7 to 7.0 | 0.93 |
The direction is IAI minus ranibizumab for these comparisons. Accordingly, a positive risk difference indicates a higher estimated percentage gaining at least 15 letters in the IAI group.
For IAI 2.0 mg Q4, the estimated risk difference was 6.6 percentage points, but the 95.1% CI extended from −1.0 to 14.1 and the reported p-value was 0.1042. The interval therefore includes zero, so the analysis does not provide a precise separation from the null value for this secondary endpoint.
The IAI 0.5 mg Q4 comparison had an estimate of −6.0 with a 95.1% CI of −13.2 to 1.2 and p = 0.1037. The IAI 2.0 mg Q8 comparison had an estimate of −0.4 with a 95.1% CI of −7.7 to 7.0 and p = 0.93.
These are secondary superiority analyses. They should not be substituted for the primary maintained-vision endpoint, which used a different binary definition, a different analysis population, and a non-inferiority hypothesis.
9. Secondary Results: NEI VFQ-25 Total Score
Mean change from baseline in National Eye Institute Visual Functioning Questionnaire (NEI VFQ-25) Total Score at Week 52 using LOCF was analyzed with ANCOVA in the FAS population.
| Comparison | Difference in LS means | 95.1% CI | P-value |
|---|---|---|---|
| IAI 2.0 mg Q4 vs ranibizumab 0.5 mg Q4 | 1.28 | −0.73 to 3.28 | 0.2090 |
| IAI 0.5 mg Q4 vs ranibizumab 0.5 mg Q4 | −0.67 | −2.69 to 1.35 | 0.5128 |
| IAI 2.0 mg Q8 vs ranibizumab 0.5 mg Q4 | −0.60 | −2.61 to 1.42 | 0.5579 |
The reported differences are IAI minus ranibizumab. Positive values therefore represent a higher adjusted mean change in the corresponding IAI group.
The estimates ranged from −0.67 to 1.28, and all three reported confidence intervals crossed zero. The corresponding p-values were 0.2090, 0.5128, and 0.5579.
These findings are most appropriately read as estimates with uncertainty rather than as proof that the treatment groups have exactly the same patient-reported outcome. The confidence intervals describe the precision of the estimated adjusted mean differences, while the p-values address the corresponding equality hypotheses.
10. Secondary Results: Choroidal Neovascularization Area
Mean change from baseline in Choroidal Neovascularization (CNV) Area at Week 52 using LOCF was analyzed with ANCOVA in the FAS population. The registry reports the outcome unit as mm2.
| Comparison | Difference in LS means | 95.1% CI | P-value |
|---|---|---|---|
| IAI 2.0 mg Q4 vs ranibizumab 0.5 mg Q4 | −0.33 | −1.04 to 0.38 | 0.3575 |
| IAI 0.5 mg Q4 vs ranibizumab 0.5 mg Q4 | 0.71 | −0.01 to 1.42 | 0.0507 |
| IAI 2.0 mg Q8 vs ranibizumab 0.5 mg Q4 | 0.86 | 0.15 to 1.58 | 0.0173 |
The registry's analysis notes describe these differences as IAI minus ranibizumab. The individual notes state that a negative value favors IAI for each of the three comparisons.
The IAI 2.0 mg Q4 comparison produced an estimated difference of −0.33 mm2, with a 95.1% CI of −1.04 to 0.38 and p = 0.3575. The interval crosses zero.
The IAI 0.5 mg Q4 comparison produced an estimate of 0.71 mm2, with a 95.1% CI of −0.01 to 1.42 and p = 0.0507. The lower confidence bound is very close to zero, illustrating why the estimated effect should be interpreted together with the interval rather than from the point estimate alone.
The IAI 2.0 mg Q8 comparison produced an estimate of 0.86 mm2, with a 95.1% CI of 0.15 to 1.58 and p = 0.0173. This is a secondary superiority analysis; its statistical interpretation should remain distinct from the trial's primary non-inferiority endpoint.
11. Safety Results
The registry reports serious adverse events by treatment arm as the number affected among those at risk. These data provide an arm-level count and denominator rather than a formal comparative hypothesis test.
| Treatment group | Serious adverse events affected / at risk |
|---|---|
| Ranibizumab 0.5 mg Q4 | 94 / 304 |
| IAI 2.0 mg Q4 | 70 / 304 |
| IAI 0.5 mg Q4 | 88 / 304 |
| IAI 2.0 mg Q8 | 90 / 303 |
Ranibizumab
Serious adverse events were reported in 94 of 304 patients at risk.
IAI 2.0 mg Q4
Serious adverse events were reported in 70 of 304 patients at risk.
IAI 0.5 mg Q4
Serious adverse events were reported in 88 of 304 patients at risk.
IAI 2.0 mg Q8
Serious adverse events were reported in 90 of 303 patients at risk.
These counts should be kept separate from the efficacy analyses. The registry does not provide a comparative effect estimate or p-value for these serious-adverse-event counts in the statistical analyses summarized here.
12. Statistical Methods Explained
Why was a risk difference used for the primary endpoint?
The primary endpoint is binary: each patient either maintained vision according to the registered definition or did not. A risk difference expresses the absolute difference between the two proportions. This makes the effect directly interpretable in percentage-point terms and, importantly for a non-inferiority trial, allows the confidence interval to be compared directly with a prespecified absolute margin.
Why is the non-inferiority margin more important than a conventional superiority p-value?
A superiority test asks whether the data provide evidence that two groups differ. A non-inferiority analysis asks a different question: whether the experimental treatment is not worse than the reference treatment by more than a prespecified amount. Here, the registry reports a 10% non-inferiority margin. The relevant confidence interval must therefore be evaluated in relation to that margin.
What does a risk difference of −0.7 mean?
For the first and third primary comparisons, the reported risk difference was −0.7. Because the registry defines the difference as ranibizumab minus IAI for these comparisons, the point estimate corresponds to a 0.7-percentage-point lower estimated maintained-vision proportion for ranibizumab relative to the respective IAI regimen. It is an estimated population-level difference, not a statement about an individual patient's response.
Why was ANCOVA used for BCVA and other continuous outcomes?
ANCOVA compares treatment groups while including the baseline value of the outcome as a covariate. In VIEW1, the registry specifies treatment group as a fixed factor across all four treatment groups and the baseline measurement as a covariate. This produces adjusted least-squares mean comparisons that account for baseline outcome differences in the model.
What does the 3.15-letter ANCOVA estimate mean?
For IAI 2.0 mg Q4 versus ranibizumab 0.5 mg Q4, the reported difference in least-squares means was 3.15 letters, calculated as IAI minus ranibizumab. Thus, the fitted adjusted mean change in ETDRS letter score was estimated to be 3.15 letters higher in the IAI 2.0 mg Q4 group for this secondary analysis. The 95.1% CI of 0.92 to 5.37 describes uncertainty around that adjusted mean difference.
Why does the analysis population matter in a non-inferiority trial?
The primary endpoint was analyzed in the PPS population, whereas the reported secondary efficacy analyses used the FAS population. In non-inferiority settings, protocol deviations and treatment non-adherence can be particularly important because they may reduce observed differences between groups. Consequently, the population used for the primary analysis is part of the substantive interpretation of the non-inferiority evidence.
What does a 95.1% confidence interval tell us?
The 95.1% confidence interval describes statistical uncertainty around the reported effect estimate under the specified inferential procedure. It does not describe the range of individual patient responses. In the primary endpoint, its most important role is determining whether the plausible range of the risk difference remains compatible with the prespecified non-inferiority margin.
13. Confidence Intervals and the Non-Inferiority Decision Framework
The three primary analyses illustrate why confidence intervals are central to non-inferiority trials. The registry reports a 10% non-inferiority margin and uses two-sided 95.1% confidence intervals under a conditional sequence with alpha = 0.049.
| Comparison | Point estimate | Lower CI | Upper CI | Relationship to 10% margin |
|---|---|---|---|---|
| IAI 2.0 mg Q4 | −0.7 | −4.4 | 3.1 | Entire reported interval is within the 10% margin |
| IAI 0.5 mg Q4 | −1.5 | −5.1 | 2.1 | Entire reported interval is within the 10% margin |
| IAI 2.0 mg Q8 | −0.7 | −4.5 | 3.1 | Entire reported interval is within the 10% margin |
For an absolute-margin non-inferiority analysis, the key question is whether the confidence interval excludes differences worse than the prespecified margin. That is fundamentally different from asking whether the confidence interval excludes zero.
14. Multiplicity and the Conditional Sequence
Three primary comparisons were posted for the same binary endpoint. The registry explicitly states that adjustment for multiple comparisons was performed through a conditional sequence of statistical hypotheses with alpha = 0.049.
Why sequence matters
Three comparisons cannot automatically be interpreted as three independent confirmatory tests. The stated conditional sequence defines how the inferential evidence is carried across the comparisons.
Why alpha = 0.049 matters
The primary analyses report 95.1% two-sided confidence intervals, matching the stated alpha level used in the conditional statistical sequence.
The secondary analyses use 95.1% confidence intervals and report p-values, but the registry classifies these as superiority analyses. Their p-values should therefore be interpreted in the context of their endpoint role and analysis framework rather than treated as interchangeable with the primary non-inferiority testing procedure.
15. LOCF and Endpoint Construction
The primary endpoint and the reported secondary endpoints use Last Observation Carried Forward (LOCF). LOCF is a missing-data approach in which an earlier available observation is carried forward to represent the later assessment when the Week 52 value is unavailable.
Because LOCF replaces missing later observations with an earlier measurement, the resulting analysis depends on assumptions about how well that carried-forward observation represents the patient's later status. The method is therefore part of the endpoint definition, not merely a reporting detail.
For the primary binary endpoint, LOCF is applied before classifying whether the patient maintained vision according to the fewer-than-15-letter-loss criterion. The registry does not provide a separate missing-data sensitivity analysis in the statistical information summarized here.
16. Analysis Populations: PPS Versus FAS
| Population | Endpoints using it | Statistical role |
|---|---|---|
| PPS | Primary maintained-vision endpoint | Primary non-inferiority analysis |
| FAS | BCVA, gain of at least 15 letters, NEI VFQ-25, CNV area | Secondary efficacy analyses |
This distinction is important because the primary and secondary analyses answer related but not identical statistical questions. The primary endpoint uses a binary classification and a non-inferiority framework in the PPS population. The secondary endpoints include continuous and binary outcomes and are analyzed in the FAS population using ANCOVA or chi-squared methods.
It would be inappropriate to treat the PPS primary result and the FAS secondary results as if they were interchangeable estimates of one underlying endpoint. They differ in endpoint definition, population, effect measure, and hypothesis type. A statistically careful interpretation keeps those distinctions visible.
17. Comparing the Primary and Secondary Questions
| Question | Endpoint | Population | Method | Hypothesis type |
|---|---|---|---|---|
| Was vision maintained? | Percentage maintaining vision at Week 52 | PPS | Wald / z-test | Non-inferiority / equivalence |
| Did mean BCVA change differ? | Mean change in ETDRS letter score | FAS | ANCOVA | Superiority |
| Did more patients gain ≥15 letters? | Percentage gaining at least 15 letters | FAS | Chi-squared | Superiority |
| Did patient-reported visual function change? | Mean change in NEI VFQ-25 total score | FAS | ANCOVA | Superiority |
| Did CNV area change differently? | Mean change in CNV area | FAS | ANCOVA | Superiority |
This table illustrates a central principle of clinical-trial interpretation: statistical evidence is always attached to a specific estimand, endpoint definition, population, comparison, and hypothesis. A result from one row cannot simply be substituted for a result from another.
18. Overall Statistical Interpretation
The three primary risk differences were −0.7, −1.5, and −0.7, with corresponding 95.1% confidence intervals of −4.4 to 3.1, −5.1 to 2.1, and −4.5 to 3.1. All three reported intervals remain within the prespecified 10% non-inferiority margin.
The ANCOVA analysis reported a difference of 3.15 letters for IAI 2.0 mg Q4 versus ranibizumab 0.5 mg Q4, with a 95.1% CI of 0.92 to 5.37 and p = 0.0054. The corresponding estimates for IAI 0.5 mg Q4 and IAI 2.0 mg Q8 were 0.80 and 0.26, with p-values of 0.4793 and 0.8179, respectively.
For the percentage gaining at least 15 letters, the reported risk differences were 6.6, −6.0, and −0.4 percentage points for the three IAI comparisons, with p-values of 0.1042, 0.1037, and 0.93.
The NEI VFQ-25 adjusted mean differences were 1.28, −0.67, and −0.60, with all three confidence intervals crossing zero.
For CNV area, the reported adjusted mean differences were −0.33, 0.71, and 0.86 mm2. The corresponding p-values were 0.3575, 0.0507, and 0.0173.
The statistical story is therefore multidimensional. The primary endpoint was framed as a non-inferiority question using an absolute risk difference and a prespecified margin. Secondary endpoints addressed continuous visual acuity, a clinically defined gain in letters, patient-reported visual function, and CNV area using different estimands and methods. These results should be interpreted according to their respective statistical roles rather than collapsed into a single overall test.
19. Important Limitations and Interpretation Issues
- Non-inferiority interpretation: the primary endpoint uses a prespecified 10% margin, so interpretation depends on the confidence interval's relationship to that margin rather than on a conventional superiority p-value.
- Analysis population: the primary endpoint uses the PPS population, whereas the secondary analyses use the FAS population. These populations answer different analytic questions.
- LOCF: the primary and secondary endpoint definitions use Last Observation Carried Forward, making the missing-data method part of the reported analysis.
- Multiplicity: the three primary comparisons were analyzed using a conditional sequence with alpha = 0.049. The comparisons should not be treated as three unrelated tests.
- Secondary endpoints: the secondary analyses use superiority hypotheses and should not be interpreted as additional versions of the primary non-inferiority question.
- Confidence intervals: the intervals quantify uncertainty around population-level estimates; they do not describe the distribution of individual patient effects.
- Safety: serious adverse-event counts are reported by arm, but the summarized registry data do not provide a comparative effect estimate or p-value for those counts.
- Generalizability: the registry identifies the condition as macular degeneration and the therapeutic area as ophthalmology; the summarized record does not provide a detailed baseline-characteristic table for assessing representativeness across specific patient subgroups.
20. Why This Trial Matters Statistically
VIEW1 is a useful teaching case because its primary endpoint illustrates a core clinical-trial distinction: a non-inferiority question is not simply a conventional superiority test with a different label. The study combines an absolute risk difference, a prespecified non-inferiority margin, a conditional sequence for multiple comparisons, and a confidence interval constructed with a specified alpha level.
| Concept | How it appears in VIEW1 |
|---|---|
| Randomization | Randomized, parallel-group phase 3 design |
| Masking | Quadruple masking |
| Binary endpoint | Maintenance of vision defined by an ETDRS letter-score threshold |
| Risk difference | Primary effect measure for maintained vision and secondary gain-of-15-letters analyses |
| Non-inferiority | 10% prespecified margin for the primary endpoint |
| Confidence interval | Two-sided 95.1% intervals for the primary comparisons |
| Wald / z-test | Primary binary endpoint analyses |
| Multiplicity | Conditional sequence with alpha = 0.049 |
| ANCOVA | BCVA, NEI VFQ-25, and CNV-area secondary analyses |
| Covariate adjustment | Baseline measurement included as a covariate in ANCOVA |
| Chi-squared test | Secondary analysis of patients gaining at least 15 letters |
| LOCF | Used in the registered Week 52 endpoint definitions |
From a statistical perspective, the most important lesson is that the effect measure and hypothesis must be interpreted together. A risk difference near zero can support non-inferiority when its confidence interval excludes clinically unacceptable deterioration, while the same numerical estimate would have a different meaning under a conventional superiority framework.
21. Related Tutorials
Learn more about the methods used in this trial:
22. Related Calculators
23. Sources
- ClinicalTrials.gov: NCT00509795 — VIEW1.
- PubMed record: PMID 32374423.
- PubMed record: PMID 31826853.
- PubMed record: PMID 25439429.
Continue through the Clinical Biostats knowledge graph
Connect clinical-trial endpoints with deeper statistical tutorials and practical analysis tools.
24. Record Summary
VIEW1 provides a detailed example of non-inferiority methodology applied to a binary visual-acuity endpoint. The primary outcome was maintenance of vision at Week 52, defined as losing fewer than 15 ETDRS letters from baseline and analyzed using LOCF in the PPS population. The three reported risk differences were −0.7, −1.5, and −0.7, with 95.1% confidence intervals of −4.4 to 3.1, −5.1 to 2.1, and −4.5 to 3.1, respectively.
The statistical framework is particularly instructive because the interpretation depends on the prespecified 10% non-inferiority margin, the conditional sequence of hypotheses with alpha = 0.049, and the use of two-sided 95.1% confidence intervals. Secondary analyses then address different clinical dimensions using ANCOVA and chi-squared methods, with baseline-adjusted mean differences and risk differences respectively.
The most useful statistical reading therefore keeps several distinctions visible: non-inferiority versus superiority, risk difference versus adjusted mean difference, PPS versus FAS, and point estimates versus confidence intervals. The trial's reported serious-adverse-event counts provide an additional safety perspective but are not accompanied in the summarized registry analyses by a comparative effect estimate or hypothesis test.