Human-centered Design and Communities of Practice to Improve Home-based Tuberculosis Contact Investigation in Uganda: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Human-centered Design and Communities of Practice to Improve Home-based Tuberculosis Contact Investigation in Uganda (NCT05640648), a randomized na study. The primary outcome is Completion of TB Evaluation.

2. Clinical Question

The study focuses on how an implementation approach affects the delivery of routine contact investigation, rather than comparing TB medicines. The registry also lists secondary outcomes concerning TB yield, treatment initiation, and initiation of TB preventative therapy. The primary endpoint is specifically restricted to people who screened positive for symptoms. It therefore addresses evaluation completion in that subgroup, not completion among all contacts or the overall prevalence of TB.

The registry describes the work as a stepped-wedge, cluster-randomized implementation trial in 12 study sites. Its record also labels the allocation randomized, masking double, and model crossover. How those design labels map onto the stepped-wedge implementation and the reported comparison is not reported in the registry record. The comparison's direction is also not identified in the reported analysis details, so the numerical estimate should be read alongside that limitation.

3. Trial Design

The trial evaluates an implementation strategy in routine TB contact investigation at study sites in Uganda. The registry describes a stepped-wedge, cluster-randomized design. In general, a stepped-wedge design introduces an intervention to clusters in a sequence, so comparisons may draw on observations made during different implementation periods as well as across clusters. Here, the record describes 12 study sites and six sequences, with two health facilities identified for each sequence in the adverse-event group labels.

During the standard implementation period, community health workers receive standard TB program training on contact investigation and supportive supervision from the on-site National TB Program focal person. The enhanced period adds a user-centered strategy developed through serial prototyping guided by human-centered design and uses communities of practice as an adjunctive adaptation and sustainment strategy. The registry also describes implementation facilitation tools and continuous quality improvement techniques. Further detail about how each component was delivered is not reported in the registry record.

The reported enrollment is 10477. The registry lists the trial as completed, with a start date of 2022-03-07 and a primary completion date of 2023-10-26. The record classifies the allocation as randomized, the masking as double, and the model as crossover, while its study summary calls the design stepped-wedge and cluster-randomized. The practical meaning of the masking and crossover labels, and how they relate to the sequence-based implementation, are not reported in the registry record.

Clustering matters because contacts treated at a shared site may have outcomes that are more alike than outcomes from unrelated sites. The stepped-wedge structure also means that intervention status can be linked to calendar or implementation period. In general, analyses need to account for clustering and the design's timing. A simple individual-level comparison could understate uncertainty or otherwise misrepresent the design; the registry explicitly cautions against a simple individual-level re-analysis.

The registry lists six outcome measures and four analyses. The primary outcome is completion of TB evaluation within 60 days among contacts who screened positive for symptoms. It also lists secondary outcomes on TB yield and initiation of TB preventative therapy. Details of the sequence schedule, allocation procedures, period effects, and the protocol's statistical plan are not reported in the registry record.

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: In this stepped-wedge, cluster-randomized trial, 12 health facilities progress from Standard Contact Investigation \[Control\] to User-centered Contact Investigation \[Intervention\] in 6 clusters of randomly assigned pairs that cross over one cluster at a time \[Sequences\] at randomly assigned 8-week intervals \[Periods\]. All clusters start in Standard for 8 weeks, then 1 cluster crosses over to User-centered every 8 weeks for 48 weeks, and finally all clusters are in User-centered for 8 weeks.

Only the first 6 of 12 groups are shown.

Period 1 (Weeks 1-8)

Period 2 (Weeks 9-16)

Period 3 (Weeks 17-24)

Period 4 (Weeks 25-32)

Analysis population for the primary outcome, as defined in the registry: Only those who screened positive for symptoms of TB are included in this analysis

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.

Population: Provided are baseline characteristics for all that participated- both index and household contact participants.

6. Primary Endpoint

The registry lists the primary outcome as Completion of TB Evaluation, measured over 60 days.

The primary endpoint is completion of TB evaluation among close contacts who screened positive for TB symptoms. The registry defines the measure as the percentage of those participants who complete evaluation within 60 days of the TB patient's treatment initiation date. The population restriction is important: contacts who did not screen positive for symptoms are not included in this primary analysis.

This endpoint reflects a step in the contact-investigation care pathway. It asks whether symptomatic contacts complete evaluation within the stated time window, rather than whether all contacts are evaluated, whether TB is ultimately diagnosed, or whether treatment is initiated. Those latter questions are represented separately among the listed secondary outcomes.

In general, a timely evaluation measure can help distinguish whether an implementation approach changes follow-through after symptom screening. It is also a process outcome: its interpretation concerns delivery of evaluation, not by itself the accuracy of diagnosis or a person's health outcome. The registry reports the primary measure as a percentage but gives the analysis result as an incidence rate ratio. The exact model specification, denominator construction, and explanation for reporting this parameter for the stated percentage are not reported in the registry record. That makes it important not to treat the estimate as a percentage-point difference.

The 60-day window anchors the endpoint to the TB patient's treatment initiation date. The registry does not report how incomplete follow-up, contacts without a recorded evaluation status, or evaluations after the window were handled. Those details affect how closely the recorded endpoint represents completion in the underlying population.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a mixed models analysis and an incidence rate ratio for the primary outcome. The design is described as stepped-wedge and cluster-randomized. These features make the analysis more involved than a comparison that treats each contact as an independent observation.

Why account for clusters?

General statistical explanation: Participants receiving services through the same site may share staff, procedures, local conditions, and implementation practices. Their outcomes can therefore be correlated. A model that accounts for clustering allows for this dependence when estimating uncertainty. If dependence is ignored, the apparent amount of independent information can be overstated and confidence intervals may be too narrow. The registry itself cautions that a simple individual-level re-analysis would be misleading, and reports only the sponsor's own analysis.

What does a mixed model do?

General statistical explanation: A mixed model combines an estimate of the average contrast of interest with terms that represent variation associated with clusters or other design features. Depending on its specification, it can account for the fact that outcomes from the same site are related and can incorporate relevant design structure. The registry names a mixed models analysis, but the particular fixed effects, random effects, link function, and covariance assumptions are not reported in the registry record. It is therefore not possible to assess those choices from the reported result alone.

How should the ratio be read?

General statistical explanation: A rate ratio expresses one modeled rate relative to another. A value of one represents equality under the stated contrast; a value above one means the numerator rate is greater, and a value below one means it is lower. The reported incidence rate ratio is 1.84, with a 95% confidence interval from 1.62 to 2.09. The registry does not identify which group is the numerator, and it defines the endpoint as a percentage. Readers should therefore retain the reported parameter label and avoid translating it into a percentage-point change or assigning a direction that the contrast description does not establish.

What does the confidence interval convey?

General statistical explanation: A confidence interval conveys uncertainty around an estimate under the model and its assumptions. A narrower interval indicates greater precision on the model's scale; a wider interval indicates more uncertainty. The interval does not represent a range containing the true value with certainty, nor does it describe the likely range of outcomes for individual participants. Here, the reported interval is on the incidence-rate-ratio scale. It does not provide the absolute number or percentage of contacts completing evaluation in either group.

What does the p-value add?

General statistical explanation: A p-value describes how incompatible the observed data are with a specified null model, assuming that model and its assumptions hold. It is not the probability that the null hypothesis is true, and it does not quantify the clinical or operational importance of an effect. The registry reports p < 0.001. That result is evidence against the model's null comparison, but interpretation still depends on the contrast definition, outcome construction, and analysis assumptions.

Why does the stepped-wedge structure matter?

General statistical explanation: In a stepped-wedge design, clusters enter the intervention at different points in an implementation sequence. The comparison can be affected by both intervention status and time, because outcome conditions may change over periods for reasons unrelated to the intervention. An analysis may need to represent period and cluster structure to distinguish those influences. The registry describes a stepped-wedge, cluster-randomized trial but does not report how the primary model handled periods, sequence, or possible time trends.

The registry metadata also label the study randomized, double-masked, and crossover. Those labels do not fully explain how masking or crossover operated in the implementation setting, or how they relate to the stepped-wedge description. The detailed schedule and the mapping of coded groups to intervention contrasts are not reported in the registry record.

The registry lists six outcome measures and four analyses. Testing multiple outcomes or contrasts can raise multiplicity concerns: when several hypotheses are examined, the chance of at least one apparently unusual result can increase. Whether any adjustment or prespecified hierarchy was used is not reported in the registry record. The reported p-value should consequently be understood as the value attached to the stated analysis, not as evidence that multiplicity was addressed.

Finally, model-based inference depends on the quality and completeness of outcome data. The record does not report missing-data counts, exclusions from the primary analysis, or sensitivity analyses. It also does not provide enough detail to reproduce the mixed model. Those omissions limit independent assessment, even though the registry reports the model estimate and its confidence interval.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Standard Implementation Strategy Period", "Enhanced Contact Investigation Intervention Period")
n      <- c(560, 1329)
events <- c(266, 1144)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Standard Implementation Strategy Period  560  266  47.5  43.3  51.7  43.4  51.6  43.4  51.6
ARM  Enhanced Contact Investigation Intervention Period  1329  1144  86.1  84.1  87.9  84.1  87.8  84.1  87.8

11. Analyses Reported by the Sponsor

The registry reports the sponsor's mixed models analysis, with an incidence rate ratio of 1.84, a 95% confidence interval from 1.62 to 2.09, and p < 0.001. The study is cluster-randomized, and the registry cautions that a simple individual-level re-analysis would be misleading. For that reason, no simple unadjusted re-calculation is presented; the sponsor's own analysis is the relevant reported estimate.

This does not mean the estimate can be fully reconstructed from the registry entry. The contrast orientation, model details, and group-specific primary-outcome counts or percentages are not reported in the registry record. The endpoint is defined as a percentage, while the result is labeled an incidence rate ratio. Without the model specification and the underlying outcome data, it is not possible to independently verify the estimate or translate it into an absolute difference.

The reported confidence interval and p-value describe the sponsor's fitted model under its assumptions. They should not be read as a substitute for missing design and analysis details. In particular, readers need the comparison's numerator and denominator definition to assign a direction to the ratio.

12. Multiplicity, Missing Data and Analysis Populations

The registry lists six outcome measures and four analyses. It identifies completion of TB evaluation as the primary outcome and lists secondary outcomes concerning TB yield, diagnosis and treatment initiation, and initiation of TB preventative therapy. The registry record does not report whether the primary outcome was selected before analysis, whether secondary outcomes were organized in a hierarchy, or whether any adjustment for multiple comparisons was used.

General statistical explanation: When investigators test several outcomes or contrasts, the chance of obtaining at least one small p-value can increase, even if all null hypotheses are true. Methods such as a prespecified testing hierarchy or multiplicity adjustment can address that issue, but whether either was used in this trial is not reported in the registry record. The reported p < 0.001 should therefore be treated as the value associated with the reported primary analysis, without assuming a multiplicity procedure.

The registry record does not report the number of participants included in the primary analysis, the number excluded, or the extent and pattern of missing outcome data. It also does not describe how missing or unobserved evaluation status was handled, whether follow-up was complete through the 60-day window, or whether any sensitivity analyses examined alternative assumptions. Those details matter because the primary outcome requires a determination of whether evaluation was completed within a defined period.

The target population for the primary outcome is restricted to contacts who screened positive for TB symptoms. The record does not report the number screened positive or how symptom-screening information was handled if it was unavailable. It also does not explain the analysis population beyond the endpoint description. For example, the criteria for inclusion in the fitted model and the handling of changes in site or implementation status are not reported in the registry record.

The record's cluster-randomized stepped-wedge description makes analysis populations and timing particularly relevant. Details of period-specific denominators, protocol deviations, and the analysis set are not reported. These gaps prevent readers from determining how completely the reported estimate represents the intended population and design.

13. 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 registry lists serious adverse events, other adverse events, and deaths separately for the standard and user-centered contact-investigation groups across the six sequences. In every listed group, the reported count is zero for serious adverse events, zero for other adverse events, and zero for deaths. The at-risk counts vary by group; these are group-specific figures and should not be added together to create a separate total.

These entries indicate that no events were recorded in the listed categories for the groups as reported. They do not establish that no adverse events occurred outside the registry's definitions, reporting procedures, observation period, or captured population. The registry record does not provide detailed event definitions, ascertainment procedures, follow-up duration for safety reporting, or an explanation of how safety information was collected.

The intervention is described as an implementation strategy for contact investigation, rather than a drug or procedure. The adverse-event figures should still be read as registry safety reporting, not as proof of absence of risk. Because the record reports zero events in the listed categories, it does not provide an event-rate comparison that can distinguish the groups. The figures also do not address benefits or effectiveness; those questions require interpretation of the outcome analyses.

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

The reported primary result is a mixed models incidence rate ratio of 1.84, with a 95% confidence interval from 1.62 to 2.09 and p < 0.001. It describes the sponsor's model-based comparison for completion of TB evaluation among contacts who screened positive for symptoms. The endpoint concerns evaluation within 60 days of the TB patient's treatment initiation date.

The result does not directly state the absolute percentage of contacts completing evaluation in either group, or the absolute difference between groups. Those values are not reported in the registry record. The ratio's direction also cannot be assigned from the analysis details because the numerator and denominator groups are not identified. The result therefore should not be translated into a claim that the enhanced or standard strategy had a higher completion measure without clarification of the model contrast.

The endpoint is a process measure, not a direct measure of diagnosis, treatment initiation, or health status. The registry lists those as separate secondary outcomes. A change in evaluation completion would not, by itself, establish a change in TB diagnosis, preventative treatment, or health outcomes. Nor does a statistically small p-value establish the size or practical importance of an effect.

The registry describes a stepped-wedge, cluster-randomized implementation trial. The sponsor's mixed model is the reported analysis, and the registry cautions against a simple individual-level re-analysis. Still, the model specification, time-period handling, missing-data approach, and analysis population are not reported. Those limitations constrain independent assessment of the estimate and its assumptions.

The adverse-event entries report zero serious adverse events, zero other adverse events, and zero deaths in each listed group. This is a description of reported events, not evidence that no unrecorded events occurred or a complete assessment of safety. Overall, the registry presents a model estimate with stated statistical uncertainty, but leaves important information needed to interpret its direction, absolute scale, and robustness unavailable.

15. Why This Trial Matters Statistically

This trial addresses an implementation question: whether a user-centered strategy can change delivery of home-based TB contact investigation. That focus is statistically important because the intervention acts through service organization and practice, rather than being assigned independently to each contact. The registry describes a cluster-randomized, stepped-wedge structure, which makes the site and timing of implementation central to analysis.

For clinical researchers, the case illustrates why the unit of randomization and the unit of analysis need to be distinguished. Contacts may be the units whose outcomes are measured, while sites are the clusters assigned to implementation sequences. General statistical reasoning says that treating correlated observations as independent can distort uncertainty. The registry explicitly notes that a simple individual-level re-analysis would be misleading and reports the sponsor's mixed models analysis instead.

The trial also highlights the value of specifying the estimand and contrast clearly. The primary outcome is defined as a percentage among symptomatic contacts, but the reported parameter is an incidence rate ratio. The registry does not identify which group is the numerator or report the underlying percentages. Without those details, a model estimate with a confidence interval and p-value cannot fully answer the practical question of how much evaluation completion changed or in which direction.

Implementation studies often include multiple outcomes, from process measures to diagnosis and treatment initiation. The registry lists six outcome measures and four analyses, but does not report a multiplicity strategy, analysis populations, or missing-data handling. Clear reporting of these elements would help readers distinguish evidence for the primary endpoint from exploratory findings and assess how robust the results are.

The record therefore offers a useful example of both the strengths and reporting challenges of complex cluster trials. It gives the study context, endpoint, model label, estimate, confidence interval, and p-value, while leaving the contrast definition and several analysis details unavailable. Those distinctions are essential when communicating results without overstating what the statistics establish.

16. 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.

Several trial-specific details needed to interpret the result are not reported in the registry record. The registry describes the study as stepped-wedge and cluster-randomized, while its metadata label the model crossover and masking double. How these labels relate to the implementation sequence is not explained. The allocation schedule, sequence timing, and operational meaning of masking are not reported.

The primary endpoint is described as a percentage, but the result is labeled an incidence rate ratio. The record does not explain the model's outcome distribution or link function, the denominator used for the rate, or the exact contrast. It also does not identify which group is the numerator. Accordingly, direction and absolute magnitude cannot be established from the reported result alone.

The mixed model is not described in enough detail for independent evaluation or reproduction. Fixed and random effects, adjustment for time periods, assumptions about within-site correlation, and any covariate adjustment are not reported. The registry cautions against a simple individual-level re-analysis, so the reported sponsor analysis cannot be checked with an ordinary unadjusted comparison from the information in the record.

The record does not report group-specific primary endpoint counts or percentages, the number of symptomatic contacts, or the primary analysis population. Missing-data amounts and handling, exclusions, protocol deviations, and sensitivity analyses are not reported. These omissions limit assessment of whether the analyzed contacts represent the endpoint population as defined.

Finally, the registry lists multiple outcomes and analyses without reporting a multiplicity plan. The results available for the primary endpoint do not by themselves establish effects on the listed secondary outcomes. The adverse-event entries are zero in the reported categories, but event definitions and ascertainment details are not reported. These limits should accompany interpretation of both effectiveness and safety reporting.

17. Related Tutorials

18. Related Statistical Calculators

19. Sources

• ClinicalTrials.gov: Human-centered Design and Communities of Practice to Improve Home-based Tuberculosis Contact Investigation in Uganda — NCT05640648.
• Registry record last updated 2025-03-30; status last verified 2025-03.
• Results first posted 2025-03-30.
• 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.

20. Record Summary

The registry record describes a completed randomized implementation trial in Uganda of a user-centered strategy for TB contact investigation. Its study summary identifies a stepped-wedge, cluster-randomized design across 12 sites, with an enrollment of 10477. The registry lists a start date of 2022-03-07 and a primary completion date of 2023-10-26.

The primary outcome is completion of TB evaluation within 60 days among close contacts who screened positive for TB symptoms. The reported mixed models result is an incidence rate ratio of 1.84, with a 95% confidence interval from 1.62 to 2.09 and p < 0.001. The registry does not identify the contrast orientation or provide group-specific outcome percentages.

The record lists six outcome measures and four analyses, and reports zero serious adverse events, zero other adverse events, and zero deaths in each listed group. Model details, missing-data handling, analysis populations, and multiplicity procedures are not reported in the registry record.