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
TEMPRANO was a randomized phase 3 factorial trial evaluating early antiretroviral treatment and/or early isoniazid prophylaxis against tuberculosis in HIV-infected adults. The study enrolled 2073 participants and used four treatment arms.
| Feature | TEMPRANO |
|---|---|
| Phase | Phase 3 |
| Conditions | HIV Infections; Tuberculosis |
| Design | Randomized, factorial |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 2073 |
| Number of arms | 4 |
| Study start | 2008-03 |
| Primary completion | 2015-01 |
| Status | Completed |
| Lead sponsor | French National Agency for Research on AIDS and Viral Hepatitis |
| Sponsor type | OTHER_GOV |
| ClinicalTrials.gov | NCT00495651 |
2. Clinical Question
The TEMPRANO design addresses two related treatment questions in HIV-infected adults: the effect of early antiretroviral treatment and the effect of early isoniazid prophylaxis against tuberculosis. Because the trial used a factorial design, the statistical structure allows the effects of the intervention dimensions to be considered within the same randomized study.
Population
HIV-infected adults enrolled in the TEMPRANO phase 3 trial, with tuberculosis as a clinical condition of interest.
Interventions
Antiretroviral medications and isoniazid prophylaxis were evaluated in the factorial treatment structure.
Comparator structure
The four-arm factorial design included combinations involving antiretroviral medications and antiretroviral medications plus isoniazid prophylaxis.
Primary questions
How does early treatment affect severe HIV-related disease or death, and how does early isoniazid prophylaxis contribute to the trial's primary clinical outcome?
3. Trial Design
A factorial trial evaluates more than one intervention dimension within a single randomized framework. In TEMPRANO, the registry identifies early antiretroviral treatment and/or early isoniazid prophylaxis as the intervention questions and records four arms. The factorial structure is statistically important because the main effects of the intervention dimensions can be evaluated efficiently, while the possibility of an interaction between them must also be considered.
Antiretroviral treatment
- Antiretroviral medications
- Evaluated within the four-arm factorial design
Isoniazid prophylaxis
- Isoniazid prophylaxis
- Evaluated in combination with the antiretroviral treatment structure
4. Registered Primary Endpoints
| Endpoint | Time frame | Registry definition |
|---|---|---|
| Death (all-cause), or severe HIV-related disease | 30 months | Severe HIV-related disease is defined as AIDS-defining diseases, non-AIDS-defining malignancies, and non-AIDS-defining invasive bacterial diseases. Invasive bacterial diseases include bacteremia or bacterial infection of any solid organ or aseptic cavity, including pneumonia, pleurisy, meningitis, pyomyositis, pyelonephritis, prostatitis, orchitis, epididymitis, salpingitis, endometritis, endocarditis, cholecystitis, and visceral abscesses. |
| Prevalence of HIV resistance (ANRS12253 associated study) | 30 month after ARV initiation | Prevalence of HIV resistance associated with the ANRS12253 study. |
The first primary endpoint is a composite clinical outcome. It combines all-cause death with a defined group of severe HIV-related diseases. The second primary endpoint is a prevalence measure concerning HIV resistance 30 months after antiretroviral initiation.
5. Statistical Methodology
The registry identifies a randomized factorial design but does not post formal statistical analyses for the primary endpoints. For the first endpoint, an analysis of the occurrence of death or severe HIV-related disease would typically treat the endpoint as a composite clinical outcome and compare the randomized treatment groups or factorial intervention factors using an appropriate binary or time-to-event framework, depending on the prespecified analysis plan.
For the second endpoint, HIV resistance prevalence is a categorical outcome measured at a defined follow-up point. A typical analysis would compare the proportion with resistance across the randomized intervention groups or factorial factors, with an effect measure such as a risk ratio, odds ratio, or risk difference accompanied by a confidence interval.
The first primary endpoint counts the occurrence of either component as an endpoint event. The interpretation therefore concerns the combined clinical outcome rather than one component in isolation.
Because the study is factorial, the statistical analysis should also distinguish the main effects of the intervention dimensions from any interaction between them. If the effect of early antiretroviral treatment is materially different depending on isoniazid prophylaxis exposure, a simple main-effect summary may not fully describe the treatment relationship.
The interaction term asks whether the effect associated with one intervention depends on the level of the other intervention.
6. Planned Analysis
Death or severe HIV-related disease
The first primary endpoint is defined as death from any cause or severe HIV-related disease within 30 months. A conventional analysis would begin by defining the analysis population according to the randomized treatment assignment and prespecified endpoint rules. If the endpoint is analyzed as time to the first qualifying event, Kaplan-Meier methods can describe the event-free distribution, with a log-rank comparison and a Cox proportional-hazards model providing a relative treatment-effect estimate. If the prespecified analysis instead uses event status at 30 months, the corresponding comparison would be based on proportions experiencing the composite endpoint.
HIV resistance prevalence
The second primary endpoint is prevalence of HIV resistance 30 months after ARV initiation. This is naturally expressed as the proportion of assessed participants with resistance at that time point. A confidence interval around each prevalence estimate quantifies uncertainty, while a between-group comparison can quantify the association with randomized treatment assignment.
Factorial main effects and interaction
For a factorial analysis, the principal comparisons would generally estimate the effect of each intervention dimension while accounting for the other dimension. An interaction term tests whether the effect of one intervention differs according to exposure to the other. When there is little evidence of interaction, the main effects can provide a concise summary of the two intervention questions. When interaction is important, the four combinations may need to be interpreted separately.
Primary clinical endpoint
Death or severe HIV-related disease is a composite outcome with several clinically distinct components. Its analysis should preserve the prespecified definition rather than silently replacing it with one component.
Resistance endpoint
Resistance prevalence is a point-in-time outcome. Its interpretation differs from a time-to-event endpoint because the outcome is assessed at a specified follow-up point after ARV initiation.
7. Statistical Methods Explained
Why use a factorial design?
A factorial design permits two intervention questions to be studied within one randomized trial. Instead of conducting completely separate trials, the design can evaluate the effects of antiretroviral treatment and isoniazid prophylaxis within the same participant population.
What is a main effect?
A main effect is the average effect associated with one intervention dimension across the levels of the other intervention. In TEMPRANO, this means considering the effect associated with early antiretroviral treatment while accounting for the isoniazid dimension, and vice versa.
Why test an interaction?
An interaction asks whether the effect of one treatment depends on the other treatment. For example, if early antiretroviral treatment had one effect among participants receiving isoniazid prophylaxis and a different effect among participants not receiving it, the two intervention effects would not be adequately summarized by independent main effects alone.
How should the composite endpoint be interpreted?
The endpoint combines all-cause death with severe HIV-related disease. A participant experiencing either qualifying component meets the composite endpoint definition. The combined endpoint can improve statistical efficiency when its components are clinically relevant, but the individual components may differ in clinical importance and frequency.
How would a 30-month prevalence endpoint be analyzed?
A prevalence endpoint at a defined time point is generally summarized as the proportion of assessed participants with the characteristic of interest. Confidence intervals describe uncertainty around the proportion, while a comparison between randomized groups can quantify the difference or relative association.
What is the difference between a time-to-event endpoint and a 30-month binary endpoint?
A time-to-event analysis uses information about when an event occurs and can account for censoring. A binary 30-month analysis asks whether the participant has experienced the specified outcome by the fixed assessment time. The appropriate method depends on the endpoint definition and prespecified statistical analysis plan.
8. Understanding the Composite Clinical Endpoint
The first primary endpoint is deliberately broader than mortality alone. It includes all-cause death and severe HIV-related disease, with severe disease defined through specific clinical categories.
| Component | Included in primary endpoint |
|---|---|
| All-cause death | Yes |
| AIDS-defining diseases | Yes, as severe HIV-related disease |
| Non-AIDS-defining malignancies | Yes, as severe HIV-related disease |
| Non-AIDS-defining invasive bacterial diseases | Yes, as severe HIV-related disease |
The definition of invasive bacterial disease is extensive. The registry includes bacteremia and bacterial infection of solid organs or aseptic cavities, with examples including pneumonia, pleurisy, meningitis, pyomyositis, pyelonephritis, prostatitis, orchitis, epididymitis, salpingitis, endometritis, endocarditis, cholecystitis, and visceral abscesses.
A composite endpoint should be interpreted as a single prespecified outcome containing multiple event types. It does not imply that all component events have identical clinical meaning. An overall treatment comparison should therefore be accompanied, when available, by an understanding of which components contribute to the composite.
9. The 30-Month Time Frame
The primary clinical endpoint is evaluated over 30 months. A fixed follow-up horizon can be useful because it places all participants' outcome assessment on a common time scale, but the statistical interpretation depends on how follow-up, censoring, and incomplete observations are handled in the prespecified analysis.
Time-to-event interpretation
If analyzed as time to first qualifying event, participants contribute information according to their observed follow-up time, with censoring handled according to the analysis plan.
Fixed-time interpretation
If analyzed as outcome status at 30 months, the comparison concerns the proportion experiencing the composite endpoint by that defined time point.
The registry specifies the 30-month time frame but does not post the formal statistical analysis used for this endpoint. The distinction between these approaches is therefore important when interpreting any future reported numerical analysis.
10. Factorial Analysis in Depth
Factorial trials are especially useful statistically because they can evaluate more than one intervention question without requiring a separate trial for each question. TEMPRANO's four-arm structure reflects the factorial evaluation of early antiretroviral treatment and early isoniazid prophylaxis.
| Statistical question | Purpose |
|---|---|
| ARV main effect | Estimate the effect associated with early antiretroviral treatment across the isoniazid dimension. |
| Isoniazid main effect | Estimate the effect associated with early isoniazid prophylaxis across the antiretroviral-treatment dimension. |
| ARV × isoniazid interaction | Assess whether the effect of one intervention depends on the other intervention. |
| Four-arm comparison | Describe the outcomes of the individual treatment combinations when an interaction or other design consideration makes this necessary. |
The main-effect interpretation is most straightforward when the interventions do not materially interact. If there is an important interaction, the average main effects can obscure meaningful differences between individual treatment combinations.
A formal interaction analysis compares treatment effects directly. Different p-values within separate groups do not, by themselves, establish that the treatment effects differ.
11. Analysis of HIV Resistance Prevalence
The second primary endpoint is the prevalence of HIV resistance 30 month after ARV initiation. Unlike a survival endpoint, this outcome is defined at a particular follow-up point rather than by the time at which resistance first occurs.
The denominator and handling of participants without an evaluable resistance assessment are important parts of a formal analysis and should follow the prespecified analysis population.
A comparison can be expressed using an absolute prevalence difference, a risk ratio, or an odds ratio. The choice should be determined by the statistical analysis plan and the scientific interpretation intended for the endpoint.
A prevalence estimate describes the proportion among the population assessed for the endpoint. If some participants lack an evaluable assessment at the specified time point, the statistical analysis must define how those observations are handled rather than silently treating them as either resistant or non-resistant.
12. Statistical Uncertainty
Any treatment-effect estimate should be accompanied by a confidence interval when available. The estimate provides the observed magnitude of association in the analyzed population, while the confidence interval describes statistical uncertainty around that estimate.
A confidence interval is not a range containing the true treatment effect with a fixed probability. It is an interval produced by a statistical procedure that, under repeated sampling and its assumptions, has the stated coverage property.
A p-value addresses the compatibility of the observed data with a specified null hypothesis under the statistical model. It does not measure the size of a treatment effect, the probability that the treatment works, or the clinical importance of the result.
Relative measures and absolute differences answer different questions. For a binary endpoint such as resistance prevalence at 30 months, both the underlying proportions and an appropriate comparative effect measure are useful for interpretation.
13. Trial Timeline
Study start
The TEMPRANO study began in March 2008.
Randomized factorial trial
The study evaluated early antiretroviral treatment and/or early isoniazid prophylaxis in a randomized factorial design.
Primary clinical endpoint time frame
The primary clinical composite endpoint was assessed over a 30-month time frame.
Resistance endpoint
Prevalence of HIV resistance was specified as a primary endpoint at 30 month after ARV initiation.
Primary completion
The registry lists January 2015 as the primary completion date.
14. Safety and Adverse Events
The registry information available for this analysis does not report serious adverse-event counts by treatment arm. No arm-specific serious adverse-event table is therefore included here.
This distinction is important statistically. Safety analyses require a defined population, exposure period, event definition, and denominator. Without those quantities, a meaningful comparison of serious adverse-event rates cannot be constructed.
15. Missing Data and Follow-Up
The registry specifies the endpoint time frames but does not post a formal missing-data or imputation strategy for the primary endpoints in the statistical analyses available on ClinicalTrials.gov.
For the 30-month clinical endpoint, incomplete follow-up can matter differently depending on whether the analysis is based on time to first event or event status at the fixed 30-month point. Time-to-event methods can use partial follow-up through censoring, whereas a fixed-time binary analysis requires a rule for participants without an evaluable 30-month status.
For HIV resistance prevalence, missing resistance measurements at the specified follow-up can directly affect the denominator. A formal analysis should therefore prespecify the analysis population and handling of missing assessments.
16. Multiplicity and Multiple Statistical Questions
TEMPRANO has two registered primary endpoints within a factorial design. This creates several statistical questions: the clinical composite endpoint, the resistance endpoint, the antiretroviral-treatment effect, the isoniazid effect, and potentially their interaction.
| Question | Statistical consideration |
|---|---|
| Clinical composite endpoint | Requires a prespecified comparison appropriate to the endpoint's event structure. |
| HIV resistance prevalence | Requires a prespecified categorical-outcome comparison at the defined follow-up point. |
| ARV main effect | Factorial main-effect estimate. |
| Isoniazid main effect | Factorial main-effect estimate. |
| ARV × isoniazid interaction | Tests whether the intervention effects depend on one another. |
When multiple hypotheses are tested formally, the statistical analysis plan should specify how type I error is controlled or how the hierarchy of hypotheses is interpreted. The registry does not post formal multiplicity-adjustment details for these endpoints.
17. Why This Trial Matters Statistically
TEMPRANO is a useful statistical teaching example because its design combines randomization, a factorial treatment structure, a clinically meaningful composite endpoint, a point-in-time prevalence endpoint, and two intervention dimensions.
| Concept | How it appears in TEMPRANO |
|---|---|
| Randomization | Participants were randomized within a phase 3 treatment trial. |
| Factorial design | The study used a four-arm factorial structure to evaluate antiretroviral treatment and isoniazid prophylaxis. |
| Composite endpoint | Death or severe HIV-related disease formed the first registered primary endpoint. |
| Fixed follow-up | The primary clinical endpoint had a 30-month time frame. |
| Prevalence analysis | HIV resistance prevalence was specified at 30 month after ARV initiation. |
| Interaction | The factorial design makes assessment of interaction between intervention dimensions statistically relevant. |
| Multiplicity | Multiple primary and factorial questions require prespecified statistical interpretation. |
| Missing data | Incomplete follow-up and missing endpoint assessments can affect both clinical and resistance analyses. |
18. Limitations
- No posted numerical statistical analyses: ClinicalTrials.gov does not post formal statistical analyses for the registered primary endpoints.
- Composite endpoint: the first endpoint combines death with several categories of severe HIV-related disease, which may differ in clinical importance and frequency.
- Factorial interaction: main effects can be difficult to interpret if the intervention dimensions interact.
- Fixed-time resistance assessment: prevalence at 30 month after ARV initiation depends on the defined assessment population and handling of missing measurements.
- Analysis population: interpretation depends on the prespecified population used for each endpoint.
- Censoring and incomplete follow-up: if the clinical endpoint is analyzed as time to event, censoring assumptions become relevant; if analyzed at 30 months as a binary outcome, incomplete assessments require a prespecified rule.
- Multiplicity: the factorial structure creates several related statistical comparisons that should be interpreted according to a prespecified testing strategy.
- Generalizability: the trial population, eligibility criteria, geographic setting, treatment implementation, and follow-up procedures determine how broadly the findings can be applied.
19. Clinical Interpretation vs Statistical Interpretation
Statistical interpretation
The study is a randomized factorial phase 3 trial with 2073 participants and two registered primary endpoints. The appropriate analysis must account for the factorial structure and the different statistical properties of the clinical composite endpoint and HIV resistance prevalence.
Clinical interpretation
The clinical questions concern whether early antiretroviral treatment and early isoniazid prophylaxis affect severe HIV-related disease, death, and HIV resistance in the defined study population and follow-up periods.
20. A Statistical Reading Guide for TEMPRANO
When interpreting a future numerical analysis of TEMPRANO, several questions should be answered before focusing on the p-value.
| Question | Why it matters |
|---|---|
| What is the analysis population? | The denominator and treatment assignment determine what population the estimate describes. |
| Is the clinical endpoint time-to-event or fixed-time? | The distinction determines whether censoring and event timing enter the analysis. |
| How is the factorial design modeled? | Main effects and interaction can lead to different interpretations. |
| What are the component events? | A composite endpoint can be driven by different clinical outcomes. |
| How are missing assessments handled? | Missing data can affect prevalence and fixed-time endpoint estimates. |
| What effect measure is reported? | Risk differences, risk ratios, odds ratios, and hazard ratios have different interpretations. |
| How is multiplicity handled? | Multiple primary and factorial hypotheses affect confirmatory interpretation. |
21. Sources
- ClinicalTrials.gov: NCT00495651 — TEMPRANO.
- PubMed: PubMed record associated with the TEMPRANO study.
- PubMed: PubMed record associated with the TEMPRANO study.
- PubMed: PubMed record associated with the TEMPRANO study.
- PubMed: PubMed record associated with the TEMPRANO study.
- PubMed: PubMed record associated with the TEMPRANO study.
Continue through the Clinical Biostats knowledge graph
Explore the statistical concepts that connect factorial clinical-trial design with practical methods for planning and analysis.
22. Related Tutorials
Learn more about the methods used in this trial:
23. Related Calculators
24. Record Summary
TEMPRANO is a phase 3 randomized factorial trial that enrolled 2073 participants and evaluated early antiretroviral treatment and/or early isoniazid prophylaxis in HIV-infected adults. Its two registered primary endpoints address different statistical problems: a 30-month composite clinical outcome consisting of all-cause death or severe HIV-related disease, and prevalence of HIV resistance 30 month after ARV initiation.
The central statistical feature is the factorial design. A rigorous analysis must distinguish the main effects of the intervention dimensions from their potential interaction. The clinical composite also requires careful attention to its individual components, while the resistance endpoint requires a clearly defined assessment population and treatment of missing measurements. For either endpoint, confidence intervals, effect measures, analysis populations, and prespecified multiplicity procedures are important parts of interpretation rather than secondary details.