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
MODIFY II was a randomized, triple-masked, parallel phase 3 treatment trial in participants receiving antibiotic therapy for Clostridium difficile infection. The trial enrolled 1,203 participants and compared MK-3415A plus standard of care, MK-6072 plus standard of care, and placebo plus standard of care.
| Feature | MODIFY II |
|---|---|
| Trial phase | Phase 3 |
| Therapeutic area | Infectious Disease |
| Condition | Clostridium Difficile Infection |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Triple |
| Primary purpose | Treatment |
| Enrollment | 1,203 |
| Primary endpoint type | Binary |
| Registered primary endpoints | 6 |
| Outcome measures posted | 14 |
| Statistical analyses posted | 19 |
| Primary analyses with estimate + CI | 13 |
| Hypothesis type | Superiority |
| Lead sponsor | Merck Sharp & Dohme LLC |
| Sponsor type | Industry |
| Trial start | February 1, 2012 |
| Primary completion | May 22, 2015 |
| ClinicalTrials.gov | NCT01513239 |
2. Clinical Question
The primary clinical question was whether treatment with MK-3415A or MK-6072, when administered with standard of care, reduced the percentage of participants experiencing CDI recurrence over 12 weeks compared with placebo plus standard of care. The registry also designated several binary safety outcomes as primary endpoints.
Population
Participants receiving antibiotic therapy for Clostridium difficile infection.
Interventions
MK-3415A plus standard of care and MK-6072 plus standard of care.
Comparator
Placebo plus standard of care.
Primary question
Do the investigational biologics change CDI recurrence and the prespecified short-term safety outcomes relative to placebo?
3. Trial Design
MK-3415A + SOC
- MK-3415A biological intervention
- Standard of care therapy
- Primary recurrence assessment at 12 weeks
- Short-term safety assessment following infusion
MK-6072 + SOC
- MK-6072 biological intervention
- Standard of care therapy
- Primary recurrence assessment at 12 weeks
- Short-term safety assessment following infusion
Placebo + SOC
- Placebo biological intervention
- Standard of care therapy
- Reference group for primary efficacy comparisons
- Reference group for primary safety comparisons
Triple-masked parallel trial
- Randomized allocation
- Parallel design
- Triple masking
- Primary purpose: treatment
Trial structure
The parallel design creates three randomized treatment groups that can be compared directly. Because the principal efficacy endpoint is binary rather than time-to-event, the registry's primary efficacy analysis uses between-group differences in percentages rather than hazard ratios.
4. Primary Endpoints
The registry identifies six primary endpoint concepts. All are binary outcomes. CDI recurrence is assessed at 12 weeks, while the principal safety endpoints concern events occurring after infusion treatment.
| Endpoint | Time frame | Definition / assessment |
|---|---|---|
| Percentage of Participants With CDI Recurrence | 12 weeks | CDI recurrence is a new episode of diarrhea, defined as 3 or more loose stools in 24 or fewer hours, together with a positive local or central stool test for toxigenic C. difficile after clinical cure of the initial CDI episode. Clinical cure is no diarrhea, defined as 2 or fewer loose stools per 24 hours, for 2 consecutive days following completion of SOC therapy for the initial episode in participants who received a 14 day or less regimen. |
| Percentage of Participants With One or More Adverse Events During 4 Weeks Following Infusion Treatment | Up to 4 weeks | An AE is an unfavorable and unintended change in body structure, function, or chemistry temporally associated with use of the medicinal product, whether or not considered related to the product. Worsening of a preexisting condition can also constitute an AE. |
| Percentage of Participants With One or More Drug-related Adverse Events During 4 Weeks Following Infusion Treatment | Up to 4 weeks | A drug-related AE is an AE determined by the investigator to be related to the drug. |
| Percentage of Participants With One or More Serious Drug-related Adverse Events During 4 Weeks Following Infusion Treatment | Up to 4 weeks | A serious drug-related AE is an AE meeting the registry's serious-event criteria and determined by the investigator to be related to the drug. |
| Percentage of Participants Who Discontinued Study Medication Due to an Adverse Event During 4 Weeks Following Infusion Treatment | Up to 4 weeks | Participants discontinuing study medication because of an adverse event. |
| Percentage of Participants With One or More Infusion-specific Adverse Events on the Day of Infusion or the Day After Infusion | Up to 24 hours | Adverse events specifically identified as infusion-specific and occurring on the day of infusion or the following day. |
5. Statistical Methodology
Binary endpoint framework
Each primary endpoint is a binary outcome: a participant either experiences the specified event during the relevant time window or does not. The principal effect measure reported for the safety endpoints is the difference in percentages, while recurrence analyses are reported as an adjusted difference.
A negative difference means that the event percentage was lower in the first-listed treatment group than in the comparison group. A positive difference means it was higher.
Miettinen and Nurminen method
The registry reports the Miettinen and Nurminen method for the primary recurrence and most primary safety comparisons. This is a score-based approach for constructing confidence intervals and testing differences between two binomial proportions. It is particularly useful when the outcome is a proportion and the analysis is focused directly on the difference between groups.
The posted analyses identify the method as a score-based confidence interval approach and, in the statistical method, as Score-based CI for proportions (Miettinen-Nurminen, Newcombe, Wilson).
Two-sided 95% confidence intervals
The posted estimates use two-sided 95% confidence intervals. For a difference in percentages, an interval entirely below zero indicates that the first-listed group has a lower event percentage under the specified comparison and statistical framework. An interval that crosses zero is compatible with both a lower and a higher event percentage.
Superiority testing
The registry identifies the hypothesis type as superiority. In these comparisons, the null reference for a difference in percentages is zero. The P-value measures the compatibility of the observed data with the specified null hypothesis; it is not a measure of the size or clinical importance of the observed difference.
Analysis populations
The primary recurrence analyses use the Full Analysis Set (FAS). The registry defines this population as all randomized participants with exclusions for failure to receive infusion of study medication, lack of a positive local stool test for toxigenic C. difficile, or failure to receive protocol-defined standard-of-care therapy within a 1-day window of infusion.
The primary safety analyses use All Participants as Treated (APaT), based on the treatment actually received. The registry notes that one participant randomized to the MK-3415A + SOC arm who was treated with MK-3415 but not MK-6072 was not analyzed.
| Analysis population | Role in posted analyses |
|---|---|
| Full Analysis Set | Primary recurrence analyses and secondary efficacy analyses involving recurrence and global cure. |
| All Participants as Treated | Primary safety analyses, including adverse events, drug-related adverse events, serious drug-related adverse events, discontinuations, and infusion-specific adverse events. |
| Treated participants with clinical cure | Secondary analysis of CDI recurrence among participants who achieved clinical cure of the initial CDI episode. |
6. Results: CDI Recurrence at 12 Weeks
The primary efficacy endpoint was the percentage of participants with CDI recurrence at 12 weeks. Three pairwise comparisons were posted: each investigational treatment versus placebo plus SOC, and MK-3415A plus SOC versus MK-6072 plus SOC.
MK-3415A + SOC vs Placebo + SOC
Adjusted difference in CDI recurrence
95% CI: -16.4 to -5.1 · P < 0.0001
Analysis population: Full Analysis Set · Miettinen and Nurminen · Superiority
The reported adjusted difference was -10.7 percentage points, calculated as MK-3415A + SOC minus placebo + SOC. Thus, the observed recurrence percentage was lower in the MK-3415A + SOC group by 10.7 percentage points under the registry's analysis.
The 95% CI from -16.4 to -5.1 describes statistical uncertainty around that between-group difference. Because the interval is entirely below zero, it does not include a zero difference. The P-value of <0.0001 addresses the specified superiority test; it does not say that the effect is clinically important, nor does it describe the magnitude of the treatment effect.
This is a proportion-based analysis rather than a time-to-event analysis. It therefore should not be interpreted as a hazard ratio or as a statement that every participant had the same reduction in recurrence risk.
MK-6072 + SOC vs Placebo + SOC
Adjusted difference in CDI recurrence
95% CI: -15.5 to -4.3 · P = 0.0003
Analysis population: Full Analysis Set · Miettinen and Nurminen · Superiority
The reported adjusted difference was -9.9 percentage points, calculated as MK-6072 + SOC minus placebo + SOC. The negative value indicates a lower recurrence percentage in the MK-6072 + SOC group under this comparison.
The 95% CI of -15.5 to -4.3 quantifies uncertainty around the estimated difference and remains below zero. The reported P-value of 0.0003 provides evidence against the specified zero-difference null hypothesis under the registry's superiority analysis.
The estimate is an absolute difference in percentages, not a relative percentage reduction. The confidence interval also does not describe the range of individual participant responses.
MK-3415A + SOC vs MK-6072 + SOC
Adjusted difference in CDI recurrence
95% CI: -5.9 to 4.2 · P = 0.3718
Analysis population: Full Analysis Set · Miettinen and Nurminen · Superiority
The reported difference was -0.8 percentage points, calculated as MK-3415A + SOC minus MK-6072 + SOC. The point estimate therefore favors a lower recurrence percentage for MK-3415A + SOC in this particular contrast, but the estimate is close to zero.
The 95% CI of -5.9 to 4.2 crosses zero. Consequently, the interval is compatible with a modestly lower recurrence percentage, little difference, or a modestly higher recurrence percentage for MK-3415A relative to MK-6072 within the precision of this comparison.
The P-value of 0.3718 does not establish a difference between the two investigational groups. It also should not be interpreted as proving that the treatments are identical; the confidence interval is the more informative description of the range of differences compatible with the data.
| Primary recurrence comparison | Adjusted difference | 95% CI | P-value |
|---|---|---|---|
| MK-3415A + SOC vs Placebo + SOC | -10.7 | -16.4 to -5.1 | <0.0001 |
| MK-6072 + SOC vs Placebo + SOC | -9.9 | -15.5 to -4.3 | 0.0003 |
| MK-3415A + SOC vs MK-6072 + SOC | -0.8 | -5.9 to 4.2 | 0.3718 |
7. Results: Overall Adverse Events
The first primary safety endpoint was the percentage of participants with one or more adverse events during the 4 weeks following infusion treatment. These analyses used the APaT population.
| Comparison | Difference in percentages | 95% CI | P-value |
|---|---|---|---|
| MK-3415A + SOC vs Placebo + SOC | -2.9 | -9.8 to 4.0 | 0.408 |
| MK-6072 + SOC vs Placebo + SOC | -2.3 | -9.2 to 4.6 | 0.517 |
For MK-3415A + SOC versus placebo + SOC, the reported difference was -2.9 percentage points, with a 95% CI of -9.8 to 4.0. For MK-6072 + SOC versus placebo + SOC, the difference was -2.3 percentage points, with a 95% CI of -9.2 to 4.6.
Both confidence intervals cross zero, so the posted estimates are compatible with either a modestly lower or modestly higher percentage of participants experiencing one or more AEs. The P-values of 0.408 and 0.517 do not provide evidence against the specified zero-difference null hypotheses in these superiority comparisons.
These results should be read as comparisons of the percentage of participants with at least one AE during a defined period. They do not measure AE severity, number of events per participant, or whether an event was causally related to study medication.
8. Results: Drug-related Adverse Events
The second primary safety endpoint concerned one or more drug-related adverse events during the 4 weeks following infusion treatment.
| Comparison | Difference in percentages | 95% CI | P-value |
|---|---|---|---|
| MK-3415A + SOC vs Placebo + SOC | -0.2 | -3.8 to 3.5 | 0.931 |
| MK-6072 + SOC vs Placebo + SOC | 0 | -3.7 to 3.6 | 0.997 |
The estimated difference for MK-3415A + SOC versus placebo + SOC was -0.2 percentage points, with a 95% CI of -3.8 to 3.5. For MK-6072 + SOC versus placebo + SOC, the estimated difference was 0.0 percentage points, with a 95% CI of -3.7 to 3.6.
Both intervals span zero and are relatively concentrated around zero compared with the much larger uncertainty seen for some other binary comparisons. The corresponding P-values were 0.931 and 0.997. A P-value near 1 does not mean that a treatment is proven identical to placebo; it indicates that the observed difference is highly compatible with the zero-difference null hypothesis.
Because relatedness was determined by the investigator, drug-related AE results also depend on the classification of events as related or not related to study medication.
9. Results: Serious Drug-related Adverse Events
The third primary safety endpoint was the percentage of participants with one or more serious drug-related adverse events during the 4 weeks following infusion treatment.
| Comparison | Difference in percentages | 95% CI | P-value |
|---|---|---|---|
| MK-3415A + SOC vs Placebo + SOC | 0.5 | -0.8 to 2.0 | 0.328 |
| MK-6072 + SOC vs Placebo + SOC | -0.3 | -1.5 to 0.7 | 0.308 |
The MK-3415A + SOC comparison produced an estimated difference of 0.5 percentage points, with a 95% CI of -0.8 to 2.0. The MK-6072 + SOC comparison produced an estimated difference of -0.3 percentage points, with a 95% CI of -1.5 to 0.7.
Because serious drug-related events are less frequent outcomes, their estimates can be sensitive to small changes in event counts. Here, both intervals cross zero. The P-values of 0.328 and 0.308 therefore do not establish a superiority difference in the posted comparisons.
Importantly, absence of a statistically detectable difference is not equivalent to demonstrating identical safety. The confidence intervals describe the precision of the observed differences within the analysis population and time window.
10. Results: Discontinuation Due to Adverse Events
The fourth primary safety endpoint was discontinuation of study medication because of an adverse event during the 4 weeks following infusion treatment.
| Comparison | Difference in percentages | 95% CI | P-value |
|---|---|---|---|
| MK-3415A + SOC vs Placebo + SOC | 0 | -1.0 to 1.0 | >0.999 |
| MK-6072 + SOC vs Placebo + SOC | 0 | -1.0 to 1.0 | >0.999 |
The registry reports an estimated difference of 0.0 percentage points for both investigational comparisons, with identical 95% confidence intervals of -1.0 to 1.0 and P-values of >0.999.
The point estimate indicates no observed difference in the percentage of participants discontinuing study medication because of an AE for either comparison. The confidence interval indicates that the statistical uncertainty around the difference was limited to the reported range of -1.0 to 1.0 percentage points.
The endpoint is specifically discontinuation due to an AE during a defined 4-week period. It should not be interpreted as a measure of all treatment discontinuation for any reason or of longer-term treatment exposure.
11. Results: Infusion-specific Adverse Events
The final primary safety endpoint concerned one or more infusion-specific adverse events occurring on the day of infusion or the day after infusion.
| Comparison | Difference in percentages | 95% CI |
|---|---|---|
| MK-3415A + SOC vs Placebo + SOC | -0.4 | -4.2 to 3.3 |
| MK-6072 + SOC vs Placebo + SOC | 1.2 | -2.7 to 5.2 |
The estimated difference was -0.4 percentage points for MK-3415A + SOC versus placebo + SOC and 1.2 percentage points for MK-6072 + SOC versus placebo + SOC.
The corresponding 95% confidence intervals were -4.2 to 3.3 and -2.7 to 5.2. Both intervals cross zero, so the posted estimates are compatible with either a lower or higher percentage of infusion-specific adverse events relative to placebo within the precision of these comparisons.
Because the registry does not report the statistical method or P-value for these two analyses, interpretation should focus on the estimated differences and their confidence intervals rather than assigning an unreported hypothesis-test result.
12. Secondary Endpoint Results: Global Cure
Global cure at 12 weeks was a secondary binary endpoint. The registry reports three pairwise comparisons using the FAS population and the Miettinen and Nurminen method.
| Comparison | Adjusted difference | 95% CI | P-value |
|---|---|---|---|
| MK-3415A + SOC vs Placebo + SOC | 5.2 | -1.8 to 12.2 | 0.0722 |
| MK-6072 + SOC vs Placebo + SOC | 14.6 | 7.7 to 21.4 | <0.0001 |
| MK-3415A + SOC vs MK-6072 + SOC | -9.4 | -16.1 to -2.7 | 0.9969 |
For MK-3415A + SOC versus placebo + SOC, the estimated adjusted difference was 5.2 percentage points, with a 95% CI of -1.8 to 12.2. Because the interval crosses zero, the posted estimate is compatible with both a modestly lower and a higher global-cure percentage.
For MK-6072 + SOC versus placebo + SOC, the estimated adjusted difference was 14.6 percentage points, with a 95% CI of 7.7 to 21.4 and a P-value of <0.0001. The entire confidence interval is above zero, indicating a higher global-cure percentage for MK-6072 + SOC under the posted comparison.
The registry reports an adjusted difference of -9.4 for MK-3415A + SOC versus MK-6072 + SOC, with a 95% CI of -16.1 to -2.7 and a P-value of 0.9969. These reported quantities should be read exactly as posted; the P-value and interval should not be independently recomputed from the rounded estimate.
13. Secondary Endpoint Results: CDI Recurrence Among Participants With Clinical Cure
A secondary analysis evaluated CDI recurrence at 12 weeks specifically among treated participants who achieved clinical cure of the initial CDI episode. This is a conditional population: recurrence is evaluated only after the participant has first achieved clinical cure.
| Comparison | Adjusted difference | 95% CI | P-value |
|---|---|---|---|
| MK-3415A + SOC vs Placebo + SOC | -11.9 | -19.0 to -4.7 | 0.0006 |
| MK-6072 + SOC vs Placebo + SOC | -13.7 | -20.4 to -6.9 | <0.0001 |
| MK-3415A + SOC vs MK-6072 + SOC | 1.6 | -4.6 to 8.0 | 0.6962 |
Among treated participants who achieved clinical cure, the reported adjusted difference was -11.9 percentage points for MK-3415A + SOC versus placebo + SOC, with a 95% CI of -19.0 to -4.7 and P = 0.0006. For MK-6072 + SOC versus placebo + SOC, the difference was -13.7 percentage points, with a 95% CI of -20.4 to -6.9 and P < 0.0001.
The direct MK-3415A versus MK-6072 comparison produced a difference of 1.6 percentage points, with a 95% CI of -4.6 to 8.0 and P = 0.6962. The interval crosses zero and therefore does not establish a difference between these two investigational groups.
These results answer a narrower question than the overall recurrence endpoint because the analysis is restricted to participants who achieved clinical cure. Conditioning on an intermediate outcome can change the interpretation of a treatment comparison and means that these results should not be treated as interchangeable with the primary FAS recurrence analysis.
14. Serious Adverse Events by Arm
The registry reports serious adverse events by treatment group using affected participants over participants at risk. These counts provide an absolute safety summary that complements the primary percentage-difference analyses.
| Treatment group | Participants with serious AEs | Participants at risk |
|---|---|---|
| MK-3415A + SOC | 118 | 390 |
| MK-6072 + SOC | 111 | 396 |
| Placebo + SOC | 129 | 381 |
| MK-3415 + SOC | 1 | 1 |
| MK-3415A + SOC 9-ME | 6 | 112 |
| MK-6072 + SOC 9-ME | 7 | 100 |
| Placebo + SOC 9-ME | 3 | 83 |
The three principal treatment groups have 390, 396, and 381 participants at risk in the posted serious-AE summary. The registry also reports separate groups labeled with MK-3415 + SOC and 9-ME designations. These should be kept distinct from the principal three-arm randomized comparison rather than pooled into a single denominator.
15. Statistical Methods Explained
Why was a difference in percentages used?
All of the primary endpoints are binary outcomes measured over specified follow-up windows. A difference in percentages directly describes the absolute separation between groups. For example, an estimate of -10.7 means that the first-listed group's event percentage was 10.7 percentage points lower than the comparator under the reported analysis.
What does the Miettinen and Nurminen method contribute?
The Miettinen and Nurminen approach is a score-based method for inference on the difference between two binomial proportions. Rather than relying on a simple normal approximation to the difference in percentages, it constructs an interval and test using a score-based framework designed for two-group proportion comparisons.
How should a negative difference be interpreted?
The sign depends on the order of the comparison. For MK-3415A + SOC versus placebo + SOC, the analysis is explicitly MK-3415A + SOC minus placebo + SOC. A negative estimate therefore means a lower percentage of participants with the specified outcome in the MK-3415A group. The same logic applies to MK-6072 comparisons.
Why is the confidence interval important?
The point estimate is only one summary of the comparison. A 95% confidence interval communicates how precisely the treatment difference was estimated. An interval that is entirely below zero supports a lower event percentage for the first-listed group under the statistical framework; an interval that crosses zero leaves zero difference among the values compatible with the data.
Why doesn't a P-value measure effect size?
A P-value quantifies how compatible the observed data are with a specified null hypothesis under the statistical model. It depends on both the size of the observed difference and the amount of information available. Two studies can produce the same estimated difference but different P-values if their sample sizes or event information differ.
Why are the FAS and APaT populations different?
The FAS is used for the primary recurrence analyses and preserves a randomized-trial framework subject to the registry's prespecified exclusions. The APaT population is based on treatment actually received and is used for safety analyses. The distinction reflects the different objectives of efficacy and exposure-related safety assessment.
Why is the clinical-cure recurrence analysis different from the primary recurrence analysis?
The primary recurrence endpoint is evaluated in the FAS, while the secondary recurrence analysis is restricted to treated participants who achieved clinical cure. The latter therefore asks a conditional question: among participants who initially achieved clinical cure, how often did recurrence subsequently occur? It should not be interpreted as simply another analysis of the same population.
16. Understanding the Effect Measures
The reported effect is expressed in percentage points. Negative values correspond to a lower event percentage in the first-listed group; positive values correspond to a higher event percentage.
Estimate
The point estimate summarizes the observed between-group difference under the specified analysis.
Confidence interval
The 95% interval describes uncertainty around the estimated difference and identifies whether zero is within the reported interval.
P-value
The P-value addresses the specified hypothesis test. It does not quantify the probability that the treatment works or the clinical importance of the effect.
Time frame
A percentage is meaningful only together with its observation window, such as 12 weeks, 4 weeks, or 24 hours.
17. Interpreting the Trial's Primary Efficacy Pattern
The three primary recurrence comparisons show two investigational-versus-placebo estimates that are below zero: -10.7 for MK-3415A + SOC versus placebo + SOC and -9.9 for MK-6072 + SOC versus placebo + SOC. Their corresponding 95% confidence intervals are entirely below zero.
The direct comparison between the two investigational groups is different. Its estimate is -0.8 with a 95% CI of -5.9 to 4.2. This interval includes zero and is substantially narrower around zero than either investigational-versus-placebo estimate. Thus, the posted results support describing the two investigational treatments separately relative to placebo while not treating the direct comparison as evidence of a difference between them.
The visual is a representation of the posted point estimates, not a confidence-interval plot. It is important not to infer precision from bar length alone; the corresponding confidence intervals provide the appropriate measure of statistical uncertainty.
18. Multiplicity and Multiple Comparisons
The trial contains multiple primary endpoints and multiple pairwise comparisons. The registry identifies superiority as the hypothesis type and posts formal analyses for the primary endpoints, but the ClinicalTrials.gov record does not specify a multiplicity-adjustment strategy or an alpha-allocation hierarchy.
This distinction is particularly important because the recurrence endpoint has three posted pairwise comparisons. The two investigational-versus-placebo comparisons address the principal treatment-versus-control questions, while the direct MK-3415A-versus-MK-6072 comparison addresses a different question about relative effects between the two investigational groups.
19. Randomization and Masking
Randomization is central to the causal interpretation of the between-group comparisons. By assigning participants randomly, the design aims to balance measured and unmeasured baseline factors in expectation, allowing differences in outcomes to be attributed to randomized treatment assignment subject to adherence, missingness, post-randomization events, and other sources of uncertainty.
The registry describes the study as triple-masked. Masking can reduce the influence of treatment knowledge on participant behavior, clinical management, outcome assessment, and other trial processes. The extent to which masking is maintained for a particular endpoint can matter especially for outcomes involving subjective assessment, although the primary recurrence endpoint has a defined clinical and laboratory component.
20. Clinical Cure and Recurrence as Sequential Outcomes
The registry's definition of CDI recurrence requires that a participant first achieve clinical cure of the initial episode. This creates a clinically meaningful sequence: initial treatment is followed by assessment of cure, and recurrence is subsequently evaluated.
The secondary recurrence analysis explicitly conditions on clinical cure, whereas the primary recurrence analysis uses the FAS definition provided by the registry. These populations answer related but distinct questions, and their estimates should not be combined without preserving the population definition.
21. Safety Interpretation
The safety endpoints use several complementary definitions: any adverse event, drug-related adverse events, serious drug-related adverse events, discontinuation due to an adverse event, and infusion-specific adverse events. Each captures a different aspect of tolerability.
| Safety dimension | What it captures | Primary time frame |
|---|---|---|
| Any adverse event | Participants with one or more AEs | Up to 4 weeks |
| Drug-related adverse event | Participants with an AE judged related to the drug | Up to 4 weeks |
| Serious drug-related adverse event | Participants with a serious AE judged related to the drug | Up to 4 weeks |
| Discontinuation due to AE | Participants stopping study medication because of an AE | Up to 4 weeks |
| Infusion-specific AE | Participants with an infusion-specific AE near infusion | Up to 24 hours |
No single safety endpoint summarizes all aspects of tolerability. A treatment can have similar percentages of participants with any AE while differing in the seriousness, attribution, timing, or consequences of individual events. That is why the registry separates these safety dimensions.
22. Important Limitations and Interpretation Issues
- Multiple endpoints and comparisons: the registry contains several primary endpoints and pairwise comparisons. The registry-reported statistical information does not specify a multiplicity-adjustment strategy, so individual P-values should not automatically be interpreted as if they were isolated tests.
- Different analysis populations: recurrence analyses use the FAS, while safety analyses use APaT. The results therefore represent different analysis populations.
- Conditional secondary analysis: recurrence among participants with clinical cure is restricted to participants who achieved that intermediate outcome and therefore answers a different question from the primary FAS recurrence analysis.
- Binary endpoint interpretation: the reported differences are percentage-point differences, not hazard ratios or relative risk reductions.
- Time-window dependence: the safety endpoints are restricted to defined windows following infusion, while recurrence and global cure are evaluated at 12 weeks.
- Infusion-specific analysis method: the registry posts estimates and confidence intervals for the infusion-specific AE comparisons but does not report the statistical method or P-values for those analyses.
- Safety denominator structure: the registry reports serious-AE summaries for several labeled groups, including groups with 9-ME designations. These should not automatically be pooled with the principal three randomized treatment groups.
- Rounding and reported quantities: the statistical estimates and P-values are registry-reported quantities. Rounded estimates should not be used to reconstruct a different P-value or confidence interval.
23. Why This Trial Matters Statistically
MODIFY II is a useful teaching example because it illustrates a clinical-trial structure in which the principal efficacy outcome is binary rather than time-to-event. The analysis therefore centers on proportions, absolute percentage-point differences, score-based confidence intervals, and superiority testing.
| Concept | How it appears in MODIFY II |
|---|---|
| Randomization | Randomized three-arm parallel phase 3 design |
| Masking | Triple-masked trial |
| Binary endpoint | Recurrence, global cure, and safety endpoints are binary outcomes |
| Risk difference | Primary safety analyses use differences in percentages |
| Score-based confidence interval | Miettinen and Nurminen is reported for the primary recurrence and most safety comparisons |
| Full Analysis Set | Used for the primary recurrence and secondary efficacy analyses |
| As-treated analysis | Used for primary safety analyses |
| Superiority testing | Registry identifies superiority as the hypothesis type |
| Multiple comparisons | Several primary endpoints and pairwise treatment comparisons are posted |
| Conditional analysis | Secondary recurrence analysis is restricted to participants achieving clinical cure |
| Safety hierarchy | Any AE, drug-related AE, serious drug-related AE, discontinuation, and infusion-specific AE are separately defined |
24. Related Tutorials
Learn more about the methods used in this trial:
25. Related Calculators
26. Sources
- ClinicalTrials.gov: NCT01513239 — MODIFY II.
- PubMed: PMID 36906440.
- PubMed: PMID 36661328.
- PubMed: PMID 32504314.
- PubMed: PMID 32376702.
- PubMed: PMID 32101275.
Continue through Clinical Biostats
Connect this trial's binary endpoints, confidence intervals, effect measures, and clinical-trial design concepts to deeper statistical tutorials and calculation tools.
27. Record Summary
MODIFY II provides a clear example of statistical analysis for randomized clinical-trial outcomes expressed as binary proportions. The primary recurrence endpoint was analyzed using the Full Analysis Set with adjusted differences and Miettinen-Nurminen confidence intervals, while safety endpoints were analyzed in the as-treated population. The posted results include estimates, two-sided 95% confidence intervals, and P-values for most primary comparisons, allowing the magnitude, direction, and precision of each comparison to be examined separately.
The central statistical distinction is between the absolute treatment difference and the uncertainty around that difference. The recurrence estimates versus placebo were -10.7 for MK-3415A + SOC and -9.9 for MK-6072 + SOC, while the direct investigational-treatment comparison was -0.8. The corresponding confidence intervals show which comparisons are compatible with zero and which are not. Safety analyses add a separate perspective, with different definitions and a different analysis population.