This page separates reported trial results from statistical interpretation. The numerical results and trial-specific facts presented here are restricted to the ClinicalTrials.gov record. Where the registry does not report the formal statistical method, that limitation is stated rather than filled in from external publications.
1. Trial at a Glance
MONET was a randomized, open-label, parallel-group phase 3 treatment trial in HIV infection. It compared darunavir/ritonavir 800/100 mg once daily as a regimen without the two background NRTIs against darunavir/ritonavir 800/100 mg once daily plus two NRTIs.
| Feature | MONET |
|---|---|
| Trial name | MONET |
| Phase | Phase 3 |
| Status | Completed |
| Therapeutic area | Infectious Disease |
| Condition | HIV Infections; AIDS Virus; Human Immunodeficiency Virus; Acquired Immunodeficiency Syndrome Virus |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 256 |
| Primary endpoint type | Time-to-event |
| Results posted | Yes |
| Outcome measures posted | 13 |
| Statistical analyses posted | 3 |
| Lead sponsor | Janssen-Cilag International NV |
| Sponsor type | Industry |
2. Clinical Question
The central statistical question was whether darunavir/ritonavir 800/100 mg once daily without the two background NRTIs could maintain virological response at Week 48 without exceeding the prespecified non-inferiority margin relative to darunavir/ritonavir 800/100 mg once daily plus two NRTIs.
Population
Participants enrolled in a phase 3 randomized treatment trial involving HIV infection and related registered conditions.
Intervention
Darunavir/ritonavir 800/100 mg once daily, abbreviated in the registry analysis as DRV/r.
Comparator
Darunavir/ritonavir 800/100 mg once daily plus two nucleoside reverse transcriptase inhibitors, abbreviated as DRV/r+2NRTIs.
Primary question
Is DRV/r once-daily monotherapy non-inferior to DRV/r plus two NRTIs for Week 48 virological response?
3. Trial Design
Triple combination
- Darunavir/ritonavir 800/100 mg once daily
- Two nucleoside reverse transcriptase inhibitors
- Comparator regimen for the non-inferiority question
Darunavir/ritonavir
- Darunavir/ritonavir 800/100 mg once daily
- No two background NRTIs in the registered treatment description
- Intervention regimen for the non-inferiority question
4. Primary Endpoint
| Endpoint | Registered definition / time frame | Analysis population |
|---|---|---|
| Virological Response [Per Protocol (PP) - Time to Loss of Virologic Response (TLOVR), < 50 Copies/ml, Week 48] | Week 48. Virological response is defined as the number of patients in the PP population with a plasma viral load < 50 HIV RNA copies/ml at Week 48. Treatment failure was defined as two consecutive HIV RNA levels ≥ 50 copies/mL, or discontinuation of randomized treatment, known as TLOVR. In addition, any switch in background NRTIs equaled failure. | PP population: all randomized patients who took study drug and did not deviate from the protocol. This excludes 10 patients with major protocol deviations. |
The registry classifies the primary endpoint as time-to-event, even though the clinical endpoint is expressed as virological response at a specified time point using a TLOVR framework. The distinction matters: TLOVR incorporates treatment failure and loss of response rules into the longitudinal classification rather than simply asking whether the viral load happened to be below the threshold at one isolated measurement.
5. Secondary Endpoints With Reported Statistical Analyses
| Endpoint | Time frame | Population | Effect measure |
|---|---|---|---|
| Virological Response [Intent To Treat (ITT) - TLOVR, < 50 Copies/ml, Week 48] | Week 48 | ITT population | Difference in proportion of response; risk difference |
| Virological Response [Intent To Treat (ITT), TLOVR - All Switches Included, < 50 Copies/ml, Week 144] | Week 144 | ITT population | Difference in proportion of response; risk difference |
6. Statistical Methodology
Non-inferiority comparison
The primary comparison was designed as a non-inferiority assessment. The registered analysis states that if the lower limit of the 95% two-sided confidence interval for the difference between DRV/r and DRV/r+2NRTIs exceeded -12% at Week 48, non-inferiority of DRV/r 800/100 mg once daily monotherapy versus the triple combination was concluded.
The registry analysis reports a maximum allowable difference of 12%, with a one-sided significance level of p=0.025 and 80% power in the design assumptions.
This is an important distinction from a conventional superiority test. In a non-inferiority trial, the question is not simply whether the observed treatment difference is statistically different from zero. The question is whether the data are sufficiently compatible with a treatment difference that remains better than the prespecified unacceptable-loss boundary.
Analysis population
The primary endpoint was analyzed in the per-protocol population. The registry defines this population as all randomized patients who took study drug and did not deviate from the protocol, excluding 10 patients with major protocol deviations.
The secondary Week 48 analysis used the intention-to-treat population, defined by the registry as all randomized patients who took study drug regardless of compliance with the protocol. This creates a useful teaching contrast: the primary non-inferiority comparison was PP, while a secondary analysis used ITT.
Risk difference
The registry reports the effect as a difference in proportion of response, normalized here as a risk difference. For the reported comparisons, the analysis text specifies the difference as DRV/r minus DRV/r+2NRTIs.
A negative estimate therefore indicates a lower observed response proportion in the DRV/r group, while a positive estimate indicates a higher observed response proportion in the DRV/r group.
Confidence interval
The primary result uses a 95% two-sided confidence interval. Although the trial's non-inferiority design uses a one-sided significance level of p=0.025, the registered decision rule is expressed through the lower bound of the two-sided 95% confidence interval.
Formal method reporting
The ClinicalTrials.gov record does not report the specific statistical method used to calculate the Week 48 risk difference and confidence interval. It also does not identify a specific regression model, variance estimator, or test statistic for this analysis. Those details should therefore not be inferred as if they were reported trial methods.
Sample-size assumptions
The registry states that, assuming a virologic response rate of 90% at 48 weeks for both treatment arms, 111 patients per treatment arm were required to establish non-inferiority with a maximum allowable difference of 12%, a one-sided significance level of p=0.025, and 80% power. The design also allowed for a maximum of 10% major protocol violations that would be excluded from the on-protocol analysis.
7. Primary Result: Week 48 Virological Response
The primary endpoint was the per-protocol TLOVR virological response at Week 48. The registered effect measure is the difference in response proportions, with the analysis comparison expressed as DRV/r minus DRV/r+2NRTIs.
Risk difference in virological response
95% CI: -10.1% to 6.8% · Two-sided confidence interval
Primary analysis: Per-protocol population
| Primary endpoint | Comparison | Estimate | 95% CI | Hypothesis |
|---|---|---|---|---|
| Virological Response, PP TLOVR, < 50 Copies/ml, Week 48 | DRV/r minus DRV/r+2NRTIs | -1.6% | -10.1% to 6.8% | Non-inferiority |
The estimated risk difference was -1.6 percentage points. Under the registry's stated comparison convention, this means the estimated proportion achieving the defined Week 48 virological response was 1.6 percentage points lower in the DRV/r group than in the DRV/r+2NRTIs group.
The estimate does not mean that an individual patient had a 1.6% lower probability of response in a deterministic sense, nor does it establish that every patient would experience the same difference. It is a group-level estimate from the specified analysis population.
The 95% confidence interval extends from -10.1% to 6.8%. Thus, the data are compatible with a modest disadvantage for DRV/r as well as with a modest advantage, within the statistical uncertainty represented by the interval.
Most importantly for the trial's non-inferiority design, the lower confidence limit of -10.1% is above the prespecified -12% non-inferiority boundary. The registry therefore reports this comparison under a non-inferiority framework. The p-value should not be used as a substitute for that margin-based reasoning.
No formal p-value for this primary comparison is reported in the ClinicalTrials.gov record. The registry instead gives the confidence interval and the non-inferiority decision rule.
8. Why the Non-Inferiority Margin Matters
The central design feature of MONET is the -12% non-inferiority boundary. A non-inferiority trial needs a clinically or scientifically justified limit for how much worse the experimental strategy could be while still being considered acceptably similar for the trial's purpose.
The visual above is only a conceptual representation of the two registered quantities; it does not reconstruct the underlying patient-level data. The key statistical relationship is simply that -10.1% is greater than -12%.
9. Secondary Result: Week 48 ITT Virological Response
The registry also reports an intention-to-treat analysis of virological response using TLOVR at Week 48. The analysis included all randomized patients who took study drug, regardless of protocol compliance.
ITT risk difference in virological response
95% CI: -9.9% to 7.8% · Two-sided confidence interval
Secondary analysis: ITT population
| Endpoint | Comparison | Estimate | 95% CI | Hypothesis |
|---|---|---|---|---|
| Virological Response, ITT TLOVR, < 50 Copies/ml, Week 48 | DRV/r minus DRV/r+2NRTIs | -1.0% | -9.9% to 7.8% | Non-inferiority |
The ITT estimate of -1.0 percentage point is close to the primary PP estimate of -1.6 percentage points. Both estimates point in the same direction under the registry's comparison convention.
The ITT 95% confidence interval extends from -9.9% to 7.8%. Its lower bound remains above the registered -12% non-inferiority margin.
This analysis does not erase the distinction between ITT and PP populations. The ITT analysis includes randomized participants according to the registry's stated definition, while the primary analysis excludes the 10 patients with major protocol deviations from the PP population.
As with the primary analysis, the confidence interval describes uncertainty around the estimated group-level difference; it does not describe the range of responses that individual patients can experience. The ClinicalTrials.gov record does not report a formal p-value for this secondary analysis.
10. Longer-Term Result: Week 144 Virological Response
The registry also reports an ITT analysis of virological response at Week 144 using TLOVR with all switches included. This analysis is classified as a secondary endpoint and uses a superiority hypothesis type.
Week 144 ITT risk difference
95% CI: -7.99% to 10.58% · Two-sided confidence interval
Difference in proportion: DRV/r minus DRV/r+2NRTIs
| Endpoint | Time frame | Estimate | 95% CI | Hypothesis |
|---|---|---|---|---|
| Virological Response, ITT TLOVR — All Switches Included, < 50 Copies/ml | Week 144 | 1.29% | -7.99% to 10.58% | Superiority |
The Week 144 estimate is 1.29 percentage points, meaning that under the registered comparison convention the estimated response proportion was 1.29 percentage points higher in the DRV/r group than in the DRV/r+2NRTIs group.
The 95% confidence interval ranges from -7.99% to 10.58%. Because the interval includes zero, the registry-reported estimate does not establish a statistically distinguishable difference from zero under a conventional two-sided interpretation.
The result should also be kept conceptually separate from the Week 48 non-inferiority analyses. Week 144 uses a different time frame, includes all switches in its TLOVR definition, and is classified in the registry as a superiority analysis rather than the primary non-inferiority comparison.
No formal p-value is reported in the ClinicalTrials.gov record. The appropriate interpretation therefore rests on the reported estimate, confidence interval, endpoint definition, analysis population, and stated hypothesis type.
11. Intention-to-Treat Versus Per-Protocol Analysis
MONET provides a useful example of why analysis population matters, especially in non-inferiority trials. The primary endpoint was analyzed in the PP population, while the Week 48 secondary endpoint used ITT.
| Feature | Primary PP analysis | Week 48 ITT analysis |
|---|---|---|
| Endpoint | Virological Response, TLOVR, < 50 Copies/ml | Virological Response, TLOVR, < 50 Copies/ml |
| Time frame | Week 48 | Week 48 |
| Population | PP; excludes 10 major protocol deviations | ITT; all randomized patients who took study drug regardless of compliance |
| Estimate | -1.6% | -1.0% |
| 95% CI | -10.1% to 6.8% | -9.9% to 7.8% |
| Hypothesis | Non-inferiority | Non-inferiority |
In a superiority trial, ITT is often emphasized because preserving randomized assignment generally provides a conservative and causally interpretable comparison. In non-inferiority trials, the situation is more subtle: protocol deviations, treatment switching, and non-adherence can make groups look more similar and can therefore make a non-inferiority conclusion easier to obtain. For that reason, examining both PP and ITT populations is often particularly informative.
12. Statistical Methods Explained
Why was a risk difference used?
The registered effect measure is the difference in the proportion of patients achieving virological response. Risk difference is directly interpretable on the percentage-point scale: a value of -1.6% means the estimated response proportion differed by -1.6 percentage points under the stated comparison direction. It is particularly useful for non-inferiority because the prespecified margin can be stated on the same absolute scale.
Why does the confidence interval matter more than simply asking whether it crosses zero?
For the primary non-inferiority question, zero is not the decisive boundary. The registered boundary is -12%. The primary 95% confidence interval extends from -10.1% to 6.8%, so its lower limit remains above -12%. A confidence interval can therefore include zero while still satisfying the stated non-inferiority criterion.
Why was a one-sided significance level used with a two-sided confidence interval?
The registry specifies a one-sided significance level of p=0.025 for the non-inferiority design while expressing the decision rule through the lower bound of a 95% two-sided confidence interval. These are equivalent ways of representing the relevant one-sided non-inferiority evidence when the confidence interval and test are constructed consistently.
Why does the analysis population matter in a non-inferiority trial?
Non-inferiority trials are unusually sensitive to deviations from the assigned treatment strategy. If protocol deviations or treatment changes make the two groups more alike, the observed difference may move toward zero, potentially making non-inferiority easier to demonstrate. MONET therefore provides a concrete example of the distinction between its PP primary analysis and ITT secondary analysis.
What does the -1.6% estimate actually mean?
It is an estimated absolute difference in response proportions, with DRV/r minus DRV/r+2NRTIs as the comparison direction. It does not mean that an individual patient has exactly a 1.6 percentage-point lower chance of response, and it does not quantify the response probability for either arm by itself.
Why is the Week 144 result not interpreted the same way as the Week 48 result?
The Week 144 analysis has a different endpoint specification, includes all switches in the TLOVR definition, uses the ITT population, and is classified as a superiority analysis. Its 95% confidence interval runs from -7.99% to 10.58%, so it describes uncertainty around a small positive point estimate rather than answering the primary Week 48 non-inferiority question.
13. TLOVR and Time-to-Event Thinking
The primary endpoint is registered as a time-to-event endpoint and uses Time to Loss of Virologic Response (TLOVR). The registry definition combines virological measurements with explicit treatment-failure rules.
The endpoint therefore does not simply classify a patient from one Week 48 laboratory measurement. Treatment changes and longitudinal virological information contribute to the response classification.
This structure illustrates why time-to-event concepts can be useful in clinical trials. Participants may be followed for different amounts of time, and a treatment failure can occur before the nominal assessment time. A TLOVR framework attempts to incorporate these longitudinal events into the endpoint definition.
At the same time, the registry does not provide the patient-level event and censoring data needed to reconstruct a Kaplan-Meier curve. It also does not report the specific formal statistical method used for the primary analysis. A Kaplan-Meier curve should therefore not be fabricated from the summary estimates on this page.
14. Week 48 and Week 144 Results Side by Side
| Analysis | Population | Time frame | Estimate | 95% CI | Hypothesis |
|---|---|---|---|---|---|
| Primary TLOVR | PP | Week 48 | -1.6% | -10.1% to 6.8% | Non-inferiority |
| Secondary TLOVR | ITT | Week 48 | -1.0% | -9.9% to 7.8% | Non-inferiority |
| Secondary TLOVR, all switches included | ITT | Week 144 | 1.29% | -7.99% to 10.58% | Superiority |
The two Week 48 estimates are close in magnitude, while the Week 144 point estimate is positive. These values should not be interpreted as a simple time trend in treatment effect because the endpoint definition, analysis population, and treatment-switch handling differ across the analyses.
15. Safety Results
The ClinicalTrials.gov record reports serious adverse events by treatment arm. These data are presented separately from efficacy because safety and virological response answer different statistical questions.
| Safety measure | DRV/r+2NRTIs | DRV/r | Total |
|---|---|---|---|
| Serious adverse events | 14/129 | 14/127 | 28/256 |
The registry limitation is important here: the study was not designed to demonstrate a safety benefit to stopping nucleoside analogues. Consequently, the absence of a reported difference in serious adverse-event counts should not be turned into a claim that one strategy is safer.
16. Sample Size and Statistical Power
The registry's design assumptions provide a useful example of non-inferiority sample-size planning. The calculation assumed a 90% virologic response rate at Week 48 in both treatment arms.
The registry also states that the design allowed for a maximum of 10% major protocol violations that would be excluded from the on-protocol analysis. The final PP analysis excludes 10 patients with major protocol deviations according to the ClinicalTrials.gov record.
17. Randomization and Open-Label Design
MONET used randomized allocation and a parallel design, with no masking. Randomization is the key design mechanism for creating comparable treatment groups in expectation, while the open-label structure means treatment assignment was known rather than concealed through blinding.
What randomization contributes
Randomization provides the foundation for comparing outcomes between treatment groups without deliberately assigning participants according to prognostic characteristics.
What randomization does not guarantee
It does not make the groups numerically identical in every characteristic, and it cannot eliminate bias arising from post-randomization treatment changes or non-adherence.
What open-label means
The registry identifies the study as having no masking. Participants and investigators therefore knew the assigned treatment.
Why this matters
Knowledge of treatment assignment can affect behavior, treatment management, reporting, and other aspects of a clinical study, depending on the endpoint.
18. Trial Timeline
Study start
The MONET trial began in June 2007 according to the registry data.
Randomized parallel comparison
The completed study used randomized allocation, two treatment arms, and no masking.
Primary virological response assessment
The primary PP TLOVR analysis estimated the difference in response proportions as -1.6%, with a 95% CI of -10.1% to 6.8%.
ITT secondary analysis
The ITT TLOVR analysis estimated a risk difference of -1.0%, with a 95% CI of -9.9% to 7.8%.
Longer-term response analysis
The ITT TLOVR analysis with all switches included reported a risk difference of 1.29%, with a 95% CI of -7.99% to 10.58%.
Primary completion
The registry lists February 2009 as the primary completion date.
19. Multiplicity, Interim Analysis, and Other Design Features
The registry-reported MONET data do not report a multiplicity-adjustment strategy, a formal interim-analysis procedure, an alpha-spending function, or a Bayesian analysis. These topics therefore cannot be attributed to the trial from the registry information provided.
| Design topic | What the ClinicalTrials.gov record supports |
|---|---|
| Non-inferiority margin | Yes. The primary criterion uses a -12% lower-bound margin. |
| One-sided testing | Yes. The design specifies a one-sided significance level of p=0.025. |
| Confidence interval | Yes. Primary and secondary reported estimates use 95% two-sided CIs. |
| Randomization | Yes. Allocation is randomized. |
| ITT analysis | Yes. Week 48 and Week 144 secondary analyses use ITT. |
| Per-protocol analysis, 125 patients were recruited in each treatment are, so 250 patients in total. Difference in proportion of response DRV/r minus DRV/r+2NRTIs. | Yes. The primary Week 48 analysis uses PP. |
| Crossover | Not reported in the ClinicalTrials.gov record. |
| Multiplicity adjustment | Not reported in the ClinicalTrials.gov record. |
| Interim analysis | Not reported in the ClinicalTrials.gov record. |
| Missing-data/imputation method | Not reported beyond the TLOVR treatment-failure rules. |
| Bayesian methods | Not reported. |
| Stratification factors | Not reported in the ClinicalTrials.gov record. |
This distinction is important for statistical interpretation. The absence of a method in the ClinicalTrials.gov record is not evidence that a method was not used; it means only that the available record does not provide enough information to describe it accurately.
20. Limitations and Interpretation Issues
- Open-label design: The study was not blinded. Knowledge of treatment assignment can introduce sources of bias that are not eliminated by randomization.
- Non-inferiority interpretation: The primary conclusion depends on the prespecified -12% margin. The statistical meaning of the result cannot be reduced to whether the confidence interval crosses zero.
- Analysis population: The primary endpoint used a PP population excluding 10 patients with major protocol deviations, while secondary analyses used ITT populations.
- Endpoint complexity: TLOVR incorporates longitudinal virological information and specified treatment-failure rules, so it is not equivalent to a single laboratory measurement taken at Week 48.
- Open-label safety assessment: The registry specifically cautions that the study was not designed to demonstrate a safety benefit from stopping nucleoside analogues.
- Formal method reporting: The ClinicalTrials.gov record does not identify the exact method used to calculate the reported risk differences and confidence intervals.
- No patient-level reconstruction: The ClinicalTrials.gov record contains summary estimates rather than individual participant event histories, so Kaplan-Meier curves or participant-level response tables cannot be reconstructed reliably.
- Later endpoint differs from the primary endpoint: The Week 144 analysis includes all switches and uses a superiority hypothesis, so it should not be treated as a direct replication of the Week 48 primary analysis.
- Safety evidence is limited in the ClinicalTrials.gov record: Serious adverse-event counts are reported, but no formal comparative safety analysis is provided.
- Registry scope: The analysis on this page is restricted to the ClinicalTrials.gov record and does not incorporate additional numerical results from the linked publication.
21. Why This Trial Matters Statistically
MONET is a useful teaching case because it concentrates several important clinical-trial concepts in a relatively compact randomized design: non-inferiority reasoning, absolute risk differences, one-sided testing, two-sided confidence intervals, per-protocol analysis, intention-to-treat analysis, treatment-failure definitions, and the interpretation of longer-term endpoint changes.
| Concept | How it appears in MONET |
|---|---|
| Randomization | Randomized allocation in a parallel-group phase 3 trial |
| Non-inferiority design | Primary Week 48 comparison uses a -12% non-inferiority margin |
| One-sided testing | Design specifies a one-sided significance level of p=0.025 |
| Confidence interval | Primary and secondary effect estimates use 95% two-sided CIs |
| Risk difference | Reported difference in proportion of virological response |
| Per-protocol analysis, 125 patients were recruited in each treatment are, so 250 patients in total. Difference in proportion of response DRV/r minus DRV/r+2NRTIs. | Primary endpoint analysis excludes 10 patients with major protocol deviations |
| Intention-to-treat analysis | Secondary Week 48 and Week 144 analyses use ITT populations |
| Time-to-event endpoint | Primary endpoint is registered as time-to-event and uses TLOVR |
| Treatment-failure rules | Two consecutive HIV RNA levels ≥ 50 copies/mL, discontinuation, and NRTI switching are incorporated into failure definitions |
| Longer-term endpoint | Week 144 TLOVR analysis with all switches included |
| Safety interpretation | Serious adverse events reported by arm, with a registry caveat concerning the safety objective |
22. Statistical Interpretation of the Primary Confidence Interval
The primary risk-difference estimate was -1.6%. With the registry's comparison convention, DRV/r was estimated to have a 1.6-percentage-point lower virological-response proportion than DRV/r+2NRTIs at Week 48.
The 95% confidence interval was -10.1% to 6.8%. This interval communicates uncertainty around the estimated difference. It is not a prediction interval for individual patients and does not imply that all plausible patient-level effects fall inside this range.
The key comparison is between the lower confidence limit and the prespecified -12% margin. Because the lower limit is -10.1%, it remains above the boundary by 1.9 percentage points. That is the central statistical relationship underlying the registry's non-inferiority framework.
The primary result does not by itself establish superiority, equality of treatment effects, absence of all clinically meaningful differences, or a safety advantage. It addresses the specific non-inferiority question defined by the trial design.
23. Risk Difference Versus Relative Measures
The trial reports an absolute risk difference rather than a hazard ratio or relative risk. This is important because different effect measures answer different questions.
| Measure | What it describes | MONET primary result |
|---|---|---|
| Risk difference | Absolute difference in response proportions | -1.6% |
| Relative risk | Ratio of response probabilities | Not reported |
| Odds ratio | Ratio of response odds | Not reported |
| Hazard ratio | Relative instantaneous event rate under a time-to-event model | Not reported |
The absence of a hazard ratio is particularly relevant because the endpoint is labeled time-to-event in the registry. Endpoint classification and effect-measure reporting are not necessarily identical: the registry supplies a risk-difference estimate for the reported statistical analysis, and that is the measure that should be interpreted here.
24. Safety and Efficacy Should Be Kept Separate
A common statistical mistake is to compress an entire clinical trial into one overall "benefit versus risk" number. MONET illustrates why that is inappropriate. Virological response and serious adverse events have different definitions, different statistical roles, and different limitations.
Efficacy
The primary efficacy analysis estimates the Week 48 difference in virological response using the PP population and a non-inferiority framework.
Long-term efficacy
The Week 144 secondary analysis uses ITT, TLOVR with all switches included, and a superiority hypothesis.
Safety
Serious adverse events are reported as 14/129 in the DRV/r+2NRTIs arm and 14/127 in the DRV/r arm.
Safety objective
The registry specifically states that the study was not designed to demonstrate a safety benefit from stopping nucleoside analogues.
25. Related Tutorials
Learn more about the methods used in this trial:
26. Related Calculators
27. Sources
- ClinicalTrials.gov: MONET — NCT00458302.
- Linked publication: PubMed record for PMID 20010070: PubMed 20010070.
Continue through the Clinical Biostats statistical library
Explore the underlying statistical concepts through tutorials and calculators covering confidence intervals, non-inferiority design, risk differences, randomization, and time-to-event analysis.
28. Record Summary
MONET is a useful statistical case study because its primary question is framed as non-inferiority, its principal effect measure is an absolute risk difference, and its primary analysis is based on a per-protocol population. The Week 48 primary estimate was -1.6% with a 95% CI of -10.1% to 6.8%. The registry's non-inferiority criterion required the lower confidence limit to exceed -12%, placing the reported lower bound above the prespecified boundary. A secondary ITT Week 48 analysis gave an estimate of -1.0% with a 95% CI of -9.9% to 7.8%. At Week 144, the ITT TLOVR analysis with all switches included reported a risk difference of 1.29% with a 95% CI of -7.99% to 10.58% and was classified as a superiority analysis.
The most important statistical lesson is that these estimates cannot be interpreted without their analysis populations, endpoint definitions, confidence intervals, and, for the primary analysis, the non-inferiority margin. The reported serious-adverse-event counts of 14/129 versus 14/127 provide additional safety information, but the registry explicitly cautions that the study was not designed to demonstrate a safety benefit from stopping nucleoside analogues.