Post ExpOsure Prophylaxis for LEprosy in the Comoros and Madagascar: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Post ExpOsure Prophylaxis for LEprosy in the Comoros and Madagascar (NCT03662022), a randomized phase 3 study. The primary outcome is Compare Effectiveness in Curbing Transmission of Leprosy of Three Different Approaches of Post Exposure Prophylaxis.

2. Clinical Question

This is a cluster-randomized question, rather than a comparison in which each person was independently randomized. The unit of assignment and the way people within clusters were handled matter for interpreting the estimates: outcomes among people in the same cluster may be related. The registry reports that the sponsor used mixed models for the three incidence rate ratio analyses. It does not report the cluster unit, the precise model specification, or how the intervention arms were allocated across clusters.

The registry identifies the study as a Phase 3, randomized, parallel trial with no masking. These design labels describe the broad framework, but do not by themselves establish how well the comparison controlled bias or how complete outcome follow-up was. Details on recruitment, exclusions, and analysis populations beyond “participants examined” are not reported in the registry record.

3. Trial Design

This was a completed, randomized, parallel, cluster-randomized trial with no masking. It studied approaches to rifampicin post-exposure prophylaxis for leprosy in the Comoros and Madagascar. The registry lists the study as Phase 3, with an enrollment of 144000. The recorded start date is 2019-01-02 and the primary completion date is 2023-01-17.

Four approaches were described: no prophylaxis; prophylaxis for all household contacts of an incident leprosy patient; prophylaxis for household contacts and other people living within a 100m radius; and prophylaxis for those contact groups who tested positive in a fingerstick blood test detecting anti-M. leprae PGL-I IgM antibody. Rifampicin is listed as the drug intervention. These are different strategies for selecting contacts, not simply different drug treatments.

The primary outcome was assessed over 45 months. The registry describes incident cases as cases newly diagnosed while under follow-up and gives the outcome as new cases per 1000 person years. The sponsor’s statistical analysis reports three incidence rate ratios: the comparator arm divided by each intervention arm, based on incidence rates measured between the first and fourth household survey in each intervention arm.

The registry notes that some of the same participants were revisited across years, which might explain discrepancies in the overall number of participants analyzed. This means enrollment should not be read as a count of unique participants contributing to every analysis, nor does it explain the precise analysis denominators. The cluster unit, cluster counts, allocation procedures, and exact distribution of participants across arms are not reported in the registry record. The record also does not report the protocol’s sample-size justification, details of masking beyond “none,” or the operational procedures for maintaining follow-up.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: Some participants will not have been present during the door to door screening of each year. It is possible that some pts will only have been screened in a limited number of periods. This explains the discrepancy between the enrollment nr in the protocol section and the pts assigned to the arms in period 1 (and other periods).Please note that the pts counts in the 'baseline characteristics' are the total of all started pts (in all years). This number reflects the total number analyzed.

Year 1 2019

Reasons for not completing this period:

Year 2 2020

Reasons for not completing this period:

Year 3 2021

Reasons for not completing this period:

Year 4 2022

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: participants examined

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Compare Effectiveness in Curbing Transmission of Leprosy of Three Different Approaches of Post Exposure Prophylaxis, measured over 45 months.

The primary endpoint is the incidence of newly diagnosed leprosy cases during follow-up, expressed as the number of new cases per 1000 person/year. The registry defines incident cases as cases newly diagnosed while participants were under follow-up. It describes the assessment period as 45 months and the outcome population as participants examined.

An incidence rate uses the time people contribute under observation, rather than treating every person as if observed for an identical period. In general, this can be useful when follow-up time varies or when people enter and leave observation at different times. The rate is therefore an exposure-based measure, with person-time forming its denominator. The registry’s label for the outcome is a rate, but the specific person-time calculation and handling of interruptions in follow-up are not reported in the registry record.

The endpoint suits the stated aim of comparing new diagnoses over follow-up across approaches to contact selection. The reported incidence rate ratio compares the rate in the comparator arm with the rate in an intervention arm. In general, a ratio below 1 means the estimated rate in the numerator arm is lower than the rate in the denominator arm, while a ratio above 1 means it is higher. The registry states that the comparator rate is divided by each intervention rate, so directional interpretation depends on that stated ordering.

A rate does not, on its own, describe the number of people who remained free of diagnosis, the severity of disease, or the timing of each diagnosis. Nor does the endpoint alone establish whether differences between arms are attributable to the prophylaxis-selection approach. That interpretation also depends on the design, analysis, uncertainty, and completeness of follow-up.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The sponsor reports mixed models analysis for three incidence rate ratio comparisons. This is the analysis reported in the registry record; a simple individual-level re-analysis would be misleading for this cluster-randomized trial. No recalculated unadjusted estimates are presented. The registry gives the ratio direction as the comparator arm’s incidence rate divided by the incidence rate in each intervention arm.

What an incidence rate ratio describes

In general, an incidence rate is the count of new events relative to the amount of time participants contribute under observation. An incidence rate ratio compares two such rates. A value of 1 represents equal estimated rates in the numerator and denominator; a value below 1 represents a lower numerator rate, and a value above 1 represents a higher numerator rate. Here, the registry states that the comparator arm is the numerator. The three reported ratios are 0.95, 0.80, and 0.58, but the comparison group codes are not mapped to named interventions in the reported analyses.

Using person-time can be useful when observation time differs among participants. However, the registry does not give the person-time totals, the detailed calculation of the rates, or how specific follow-up circumstances were handled. Those details are necessary to reproduce the endpoint from raw data and are not reported in the registry record.

Why clustering matters

In general, cluster randomization assigns groups rather than independently assigning each participant. People within a group may share characteristics, exposure, or care context, so their outcomes may be more similar than outcomes from unrelated groups. This dependence affects how much information the data contain and how uncertainty should be estimated. Treating all participants as independent can make standard errors too small and confidence intervals too narrow when within-cluster correlation is present. The registry explicitly identifies the study as cluster randomized and notes that a simple individual-level re-analysis would be misleading.

Mixed models are a broad family of methods that can account for grouped or repeated observations by representing some sources of variation through model terms. That is a general description, not a claim about the exact specification used in this trial. The registry names mixed models analysis but does not report the model formula, link function, random effects, covariates, cluster identifier, or the method used to calculate the confidence intervals and p-values. The precise way clustering and repeat observations were addressed is therefore not clear from the registry record.

Confidence intervals and p-values

Each estimate is accompanied by a 98.3% confidence interval: 0.40 to 2.23, 0.34 to 1.87, and 0.22 to 1.56. In general, a confidence interval expresses uncertainty about an estimated quantity under the statistical model and its assumptions. It is not a statement that the true value has a specified probability of lying within that particular calculated interval. Here, all three intervals include 1, the value representing equal rates. Their breadth indicates that the reported estimates are compatible with a range of possible rate ratios.

The reported p-values are <0.017, 0.017, and <0.017. In general, a p-value measures how unusual data at least as inconsistent with a null hypothesis would be if that hypothesis and the analysis assumptions were true. It is not the probability that the null hypothesis is true, nor does it quantify the practical importance of an effect. P-values do not replace the estimates and their confidence intervals. The registry does not report the null-testing details or how the p-values relate to the confidence level.

Over-dispersion and model fit

In general, over-dispersion occurs when outcome variability is greater than a model’s assumptions allow. For count and rate outcomes, ignoring over-dispersion can understate uncertainty. A model may address this in different ways, depending on its form and assumptions. The registry record does not report whether over-dispersion was assessed, whether it was present, or how it was handled. It also does not report diagnostics or model-fit checks. These are unknowns, not evidence that a particular problem occurred.

Repeated observations and analysis population

The registry notes that some participants were revisited across years, potentially explaining discrepancies in the overall number analyzed. Repeated observations from the same person can be related, just as observations from the same cluster can be related. The model details needed to determine precisely how those dependencies were handled are not reported. The primary outcome population is described as participants examined, but the analysis counts, exclusions, and approach to missing outcomes are not reported in the registry record.

There are three reported comparisons, each with a 98.3% confidence interval. Multiple comparisons can increase the chance of at least one misleading finding if each test is interpreted without accounting for the set of tests. In general, a multiplicity plan may adjust significance thresholds, confidence levels, or interpretation. The registry does not state whether the reported intervals or p-values were adjusted, or give a prespecified multiplicity plan. Their particular confidence level should not be used to infer a method that the record does not describe.

10. Analyses Reported by the Sponsor

The posted estimates are the sponsor’s mixed-model analyses. The registry’s design note says that a simple individual-level re-analysis would be misleading because the trial was cluster randomized, and the posted results therefore present only the sponsor’s own analysis. No unadjusted recalculated incidence rate ratios or confidence intervals are reported for comparison.

The sponsor reports three ratios: 0.95, 0.80, and 0.58, each with a 98.3% confidence interval. The confidence intervals are 0.40 to 2.23, 0.34 to 1.87, and 0.22 to 1.56, respectively. The corresponding p-values are <0.017, 0.017, and <0.017. The registry states that each comparison divides the comparator rate by the rate for one intervention arm. The group codes are not connected to the named intervention strategies in the analysis entries, so the individual estimates cannot be reliably labelled by strategy from the registry record alone.

Because no independent recalculation is reported, there is no recalculated result to compare with the sponsor’s model-based estimates. The record also does not provide enough model specification or analysis data to reproduce those estimates. The reported results should consequently be understood as the sponsor’s own analysis, with its stated mixed-model method and the limitations in the reporting detail noted above.

11. Multiplicity, Missing Data and Analysis Populations

Three incidence rate ratio comparisons are reported, each contrasting the comparator with an intervention arm. Each has a 98.3% confidence interval, and the p-values are <0.017, 0.017, and <0.017. Multiple comparisons raise a general interpretive issue: considering several tests together can make it more likely that at least one result appears notable by chance if each is judged using a separate threshold. Methods for addressing this include prespecified adjustments or a planned hierarchy of comparisons, but the registry does not report whether any such approach was used.

The reported confidence level and p-values do not, by themselves, reveal whether multiplicity was handled. The rationale for the 98.3% intervals, the planned significance threshold, and the relationship between the intervals and tests are not reported in the registry record. The protocol’s statistical analysis plan, including whether the three comparisons were treated as a family, is not reported. It is therefore not possible to determine from these entries whether the displayed inference was adjusted or how the sponsor intended to interpret the set of comparisons.

The primary outcome population is described as participants examined. The registry also notes that some participants were revisited in different years and that this might explain discrepancies in the overall number of participants analyzed. It does not give the number analyzed for each comparison, exclusions from analysis, or the extent and pattern of missing outcome information. The methods for handling participants without complete follow-up or other missing data are not reported in the registry record.

Details about the analysis population are similarly limited. The record does not say whether analyses followed an intention-to-treat principle, how cluster membership was treated after any changes, or whether the analysis included all randomized clusters. These omissions do not establish that data were missing or that participants were excluded; they mean the available record does not describe these matters sufficiently to assess their impact.

The number or timing of interim analyses and any associated decision rules are not reported in the registry record. Neither are the protocol’s sample-size justification and the assumptions used to plan enrollment. Those details should not be inferred from the enrollment figure or from the final estimates.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The adverse-event entries list zero serious adverse events and zero other adverse events in the three prophylaxis arms. The numbers listed at risk are 731 for Household PEP, 15203 for PEP 100m, and 2850 for PEP 100m + Positive for Anti-PGL-I IgM Ab. The corresponding affected counts are 0 for serious events and 0 for other events in each of those arms.

For the No PEP arm, the record lists 0 at risk and 0 affected for both serious and other adverse events. With no participants listed at risk, that entry does not provide an event comparison for the no-prophylaxis arm. The record lists 0 deaths in each arm. It does not provide serious adverse event terms.

These entries describe what the registry reports, not a guarantee that no adverse events occurred. The adverse-event definitions, collection procedures, observation period for safety, and completeness of reporting are not reported in the registry record. The listed at-risk counts also differ from the overall enrollment, and the record does not explain how those safety populations were determined. Consequently, these figures should be read as registry-reported counts, not as a complete assessment of the relative safety of the approaches or as proof of absence of harm.

13. What the Result Does, and Does Not, Mean

The posted primary results estimate incidence rate ratios for three comparisons between a no-prophylaxis comparator and intervention arms using different approaches to selecting contacts for rifampicin post-exposure prophylaxis. The reported ratios are 0.95, 0.80, and 0.58, with 98.3% confidence intervals of 0.40 to 2.23, 0.34 to 1.87, and 0.22 to 1.56. The registry states that the comparator rate is divided by the intervention rate. The ratio estimates are therefore comparisons of rates, not counts of cases or direct measures of individual benefit.

All three estimates are below 1, but every confidence interval includes 1 and spans values below and above it. As a general statistical interpretation, that pattern means the estimates are uncertain and do not isolate a single rate-ratio value. The intervals are compatible with equal rates as well as with differing rates in either direction under the reported comparison. The registry’s p-values should likewise be considered alongside the estimates, intervals, and unreported multiplicity plan.

These results do not establish that a particular intervention approach is effective or ineffective. The comparison labels in the analysis entries are not mapped to the named approaches, and the registry does not report the detailed model specification, cluster information, or analysis denominators. Those limits prevent a more detailed reading of which contact-selection strategy corresponds to each estimate and how robust the estimates are to modeling decisions.

The endpoint captures new diagnoses during follow-up, expressed as cases per 1000 person/year. It does not directly count each transmission event, and the posted estimates do not, by themselves, demonstrate interruption of transmission. The study description identifies transmission reduction as an aim, but the evidence presented in the primary outcome is an incidence-rate comparison.

Finally, the registry’s adverse-event entries list no affected participants in the prophylaxis arms, but no at-risk participants are listed for the comparator, and safety reporting details are limited. Those figures do not establish absence of harm. The results are best understood as model-based, imprecise comparisons of recorded incidence rates, with important design and reporting details not available in the registry record.

14. Why This Trial Matters Statistically

This trial is statistically notable for asking a cluster-randomized question about different ways to identify contacts for post-exposure prophylaxis. The strategies range from household contacts to wider geographic contact selection and test-based selection. Comparing these approaches requires more than counting outcomes as if every participant had been independently randomized, because people within clusters may have correlated outcomes. The registry itself cautions that simple individual-level re-analysis would be misleading.

The endpoint is an incidence rate, which relates new diagnoses to time under observation. This is a useful distinction when follow-up is not identical for everyone, but it makes clarity about person-time and follow-up essential. The record also notes that some participants were revisited across years. Repeated observations and clustering are separate sources of dependence that a statistical analysis may need to address; the registry identifies mixed models but does not give enough detail to evaluate the exact model structure.

The three point estimates are all below 1, yet their 98.3% confidence intervals are broad and include 1. That combination illustrates why an effect estimate should not be read without its uncertainty. A point estimate gives one summary of the observed comparison, while the interval shows that the data and model do not pin down the ratio precisely. The three comparisons also raise a multiplicity question, but the registry does not explain the confidence level or the planned approach to interpreting the family of tests.

The record’s limitations are instructive for reading registry results: reported enrollment is not necessarily the same as the number analyzed for every outcome, safety populations may differ from overall enrollment, and repeated visits may complicate counts. The registry notes some of these issues but does not supply enough detail to resolve them. Careful interpretation therefore separates what is reported—the endpoint, estimates, intervals, p-values, and design labels—from what remains undocumented.

15. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry provides limited detail for evaluating the primary analysis beyond identifying it as mixed models analysis. The model specification, covariates, cluster structure, assumptions, diagnostics, and method for deriving uncertainty estimates are not reported in the registry record. The record identifies the trial as cluster randomized but does not specify the unit or number of clusters, allocation procedures, or the distribution of clusters among arms. These omissions limit independent assessment of the analysis.

The comparison entries use group codes, but the record does not map those codes to the named prophylaxis approaches. Although the intervention descriptions identify three contact-selection strategies, it is not possible to match each reported estimate to a particular strategy from the analysis entries alone. This limits strategy-specific interpretation.

The registry notes repeated visits to some participants and suggests that this might explain discrepancies in overall participant counts. It does not provide the number analyzed in each comparison, follow-up completeness, missing-data patterns, exclusions, or the rules for including participants and clusters. The analysis population is described only as participants examined. These details are relevant to judging possible selection effects and the contribution of repeated observations.

The outcome is incident diagnosis during follow-up, but the registry does not detail case ascertainment procedures, person-time calculation, or handling of changes in observation. The study’s stated transmission objective is broader than the recorded incidence endpoint, which is not a direct count of transmission events.

There are three reported comparisons, but the multiplicity plan, rationale for 98.3% confidence intervals, and interpretation threshold are not reported. The registry also does not report interim analyses, the protocol’s sample-size justification, or the statistical analysis plan. Safety reporting is limited: no safety events are listed in the prophylaxis arms, while the no-PEP arm has zero participants listed at risk. Definitions and collection procedures are not reported, so the safety entries cannot be treated as a complete comparative safety assessment.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Post ExpOsure Prophylaxis for LEprosy in the Comoros and Madagascar — NCT03662022.
• Registry record last updated 2025-03-04; status last verified 2025-02.
• Results first posted 2025-03-04.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

NCT03662022 is a completed Phase 3, randomized, parallel, cluster-randomized trial with no masking. It studied rifampicin post-exposure prophylaxis approaches for leprosy in the Comoros and Madagascar, with an enrollment of 144000. The study start date was 2019-01-02 and the primary completion date was 2023-01-17.

The primary outcome was incident leprosy cases per 1000 person/year over 45 months. The registry reports three sponsor mixed-model incidence rate ratios comparing the no-PEP comparator with intervention arms: 0.95, 0.80, and 0.58. Their 98.3% confidence intervals are 0.40 to 2.23, 0.34 to 1.87, and 0.22 to 1.56. The registry does not map the analysis group codes to the named approaches.

The estimates are imprecise, and the registry record lacks key details on model specification, multiplicity, missing data, cluster structure, and analysis populations. Some participants were revisited across years, according to the record. Safety entries list zero affected participants in the prophylaxis arms, while the no-PEP arm lists zero at risk.