TB Innovation Project: A Pre- and Post- Implementation Assessment (TIPPI): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for TB Innovation Project: A Pre- and Post- Implementation Assessment (TIPPI) (NCT03948698), a registered phase not stated study. The primary outcome is Children Screened for TB.

2. Clinical Question

3. Trial Design

TIPPI is an evaluation of the Catalyzing Pediatric TB Innovations project, also called CaP-TB. The registry characterizes it as a pre- and post-implementation assessment across ten countries. The project sought to use innovative methods and capacity building to strengthen health systems in relation to pediatric TB case detection, early accurate diagnosis and effective treatment. The evaluation focused on selected service-delivery outcomes.

The registry’s brief description says the evaluation would assess CaP-TB interventions in up to 450 sites and compare selected outcomes with routine TB programming in a subset of project sites. This description does not establish that every site contributed to every endpoint, or identify the sites used for the primary endpoint. The exact number of sites contributing data is not reported in the registry record.

The assessment period for the primary outcome is December 2018–September 2021. The registry lists the study start as 2018-12-01, primary completion as 2021-09-30 and status as completed. Those dates identify the record’s study milestones; the specific dates or lengths of the pre-implementation and post-implementation observation windows are not reported in the registry record.

This was not a randomized comparison. The allocation method is not stated, and the record explicitly cautions that the arms are not a like-for-like control comparison. The registry describes an intended comparison with routine TB program service delivery, but does not detail how sites or periods were selected, whether the same sites contributed before and after implementation, or how differences between intervention and routine-program settings were handled.

The registry lists an enrollment of 4,159,533, while the primary endpoint population description gives 7,198,865 clinic attendees and 4,749,072 children screened. These figures are reported as different record elements; their relationship is not explained in the registry record. A detailed participant flow or accounting of site-level data is not reported.

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: This study does not have a set number of participants. The number of participants is number of children screened.

Overall Study

Analysis population for the primary outcome, as defined in the registry: The number of participants analyzed, 7198865, is the clinic attendees and 4749072 is number of children screened. 4749072 is the number indicated in the participant flow because this data is generated by the study. Clinic attendees data is already available in health facilities.

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 Children Screened for TB, measured over December 2018-September 2021.

The primary endpoint is “Children Screened for TB.” It measures the number and proportion of children age 0–14 screened for TB among clinic attendees, over December 2018–September 2021. The registry identifies 4,749,072 as the number of children screened and 7,198,865 as the clinic-attendee population for this endpoint. The record explains that the screened count is indicated in participant flow because that information was generated by the study, while clinic-attendee data were already available in health facilities.

A count describes the volume of screening activity, while a proportion places that count in relation to clinic attendance. Using both measures can make service delivery easier to interpret: counts reflect the amount of activity recorded, and proportions describe screening relative to attendance. Neither measure, on its own, establishes why screening levels may differ across settings or periods. General statistical note: a proportion is interpretable only when its numerator and denominator refer to a clearly defined population and observation period.

For this endpoint, the registry gives the screened count and clinic-attendee denominator in its population description, but does not report the endpoint’s calculated proportion or an uncertainty interval. It also does not report a period-specific breakdown, an adjusted comparison, or a statistical test for change. The endpoint therefore describes recorded screening activity, rather than a causal effect attributable solely to the intervention.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a count of children screened for TB, with the number and proportion considered among clinic attendees. Count outcomes are different from continuous measurements such as weight or blood pressure: the observed value is a number of events or people meeting the endpoint definition. The registry’s description identifies the counted group as children age 0–14 who were screened, and the population as clinic attendees.

What does the count describe?

A count is useful for describing service volume. It answers how many screening events or people were recorded under the stated definition, but it does not automatically answer whether service coverage changed. Clinic attendance can vary, and the same count may have a different meaning when the number of attendees differs. The proportion among clinic attendees addresses that denominator context. General statistical note: counts and proportions are complementary summaries, and their interpretation depends on consistent definitions and complete recording.

The registry record gives 4,749,072 children screened and 7,198,865 clinic attendees. It does not report the calculated proportion, its uncertainty, or a breakdown by implementation period. It also does not specify whether the count represents unique children, screening encounters, or another counting convention. The record’s use of “number of children” describes the endpoint, but the detailed counting rules are not reported in the registry record. This distinction matters because repeated visits can affect counts if the unit is an encounter rather than a unique child.

How would a pre/post comparison be interpreted?

A pre/post assessment compares outcome measurements from periods before and after implementation. For a count endpoint, a useful analysis might account for the amount of observation or clinic attendance in each period and present the estimated change with uncertainty. The registry record does not report the specific pre- and post-implementation windows, the period-specific counts, a model, or an effect estimate. Therefore no such comparison can be reconstructed from the record’s description alone.

General statistical note: count data are often analyzed with a model that links the expected count to explanatory variables and, when appropriate, to the amount of exposure or observation. A Poisson model is a common starting point, but it assumes a particular relationship between the average count and its variance. If the observed variability is greater than that assumption allows, over-dispersion can make standard errors too small and confidence intervals too narrow. Alternative variance estimates or a model that allows extra variation may then be considered. This is general methodological context, not a description of the analysis used for TIPPI; the statistical model is not reported in the registry record.

Why does the comparison design matter?

The registry identifies the assessment as pre- and post-implementation and describes a comparison with routine TB program outcomes in a subset of sites. It also says the study was not randomized, that allocation was not stated, and that the arms are not a like-for-like control comparison. In a nonrandomized setting, implementation and comparison groups may differ before the intervention. Changes over time can also affect both service delivery and recording. General statistical note: adjustment for measured differences can improve comparability, but it cannot guarantee removal of bias from unmeasured differences.

Site-level data may also be correlated. For example, observations from the same facility can share staffing, procedures and local context. General statistical note: treating correlated observations as independent can understate uncertainty; analyses may need to account for clustering. Whether TIPPI used a site-level or clustered analysis is not reported in the registry record. The number of sites contributing to the primary endpoint is also not reported.

What do planning illustrations mean?

The registry’s planning assumptions describe illustrations based on an observed effect, using a two-sided significance level of 0.05, equal allocation between two arms and a normal approximation. The effect to detect is described as a stated fraction of the observed effect. For non-time-to-event outcomes, those illustrations are expressed as participants per arm; time-to-event illustrations are described as total events. These are planning assumptions, not evidence that the evaluation had a prespecified sample-size target or that it achieved a particular power.

General statistical note: a planning calculation depends on the assumed effect, outcome variability, significance level, allocation and analysis method. If it uses an effect estimated from the same data, it is an illustration of future planning needs rather than an independent validation of the observed result. The registry record does not report the trial’s own sample-size justification, numerical illustration results, or a prespecified power calculation. There is no basis in the record for describing any calculation as observed power.

The record lists zero analyses. It does not report a statistical analysis plan, covariate adjustment, handling of repeated observations, treatment of missing data, or an inferential result for the primary endpoint. The methods concepts above explain considerations relevant to count outcomes and pre/post comparisons in general; they should not be mistaken for methods documented as having been used in this evaluation.

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("CaP TB TIPPI Study")
n      <- c(7198865)
events <- c(4749072)
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  CaP TB TIPPI Study  7198865  4749072  66.0  65.9  66.0  65.9  66.0  65.9  66.0

11. Multiplicity, Missing Data and Analysis Populations

The registry lists twelve outcome measures and zero analyses. It does not report a multiplicity strategy for interpreting results across outcomes, such as a hierarchy of endpoints or an adjustment to account for multiple statistical tests. Because no set of inferential results is reported, the record does not permit assessment of how multiple comparisons may have affected conclusions.

Missing-data procedures are not reported in the registry record. The entry does not describe how absent clinic records, incomplete screening information, changes in site participation, or other unavailable observations were handled. It also does not report the extent or pattern of missing data. General statistical note: missingness can affect a count or proportion if missing records are related to whether screening occurred or was documented. The direction and size of any effect depend on the circumstances, which are not described.

The analysis population is not fully defined. The primary endpoint description identifies clinic attendees as the population and children age 0–14 as the screened group, but the record does not specify inclusion and exclusion rules, the precise unit counted, or which sites and periods contributed. It reports 4,749,072 children screened and 7,198,865 clinic attendees, while the registry enrollment is 4,159,533; the relationship among these quantities is not explained in the registry record.

The registry does not report whether all eligible records were analyzed, whether data were aggregated by site or period, or whether any exclusions were applied. These omissions restrict assessment of completeness and reproducibility. They do not establish that data were missing or that a particular analysis population was used; those details are simply not reported in the registry record.

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

The posted primary endpoint figures describe a large volume of recorded pediatric TB screening in the clinic-attendee population: the registry identifies 4,749,072 children screened and 7,198,865 clinic attendees. The outcome is relevant to service delivery because it tracks screening activity among children age 0–14. The registry frames the evaluation as assessing selected outcomes related to CaP-TB implementation in participating settings.

These figures do not, by themselves, establish that the CaP-TB intervention caused a change in screening. The record describes a pre/post assessment, not a randomized comparison, and warns that the arms are not a like-for-like control comparison. The registry does not report period-specific estimates, an effect estimate, a confidence interval or a statistical test for the primary endpoint. The calculated proportion is also not reported in the registry record.

The record therefore does not show how screening differed before and after implementation, how the intervention sites compared with routine TB program sites, or how precise any difference might be. It also does not establish whether the screening count represents unique children or another counting unit; detailed endpoint counting rules are not reported.

General statistical note: a descriptive count can document activity, while causal interpretation requires a comparison that addresses other explanations for differences. Changes in clinic attendance, documentation or site composition could affect observed counts or proportions. The record does not describe whether or how such factors were addressed.

The enrollment value of 4,159,533 should not be substituted for the clinic-attendee population stated for the primary endpoint. The registry reports both, but does not explain their relationship. The endpoint figures are best read according to their own labels, without combining or reconciling them.

Overall, the registry supports a description of screening volume in the stated population and period. It does not support a quantified causal claim, a conclusion about statistical significance, or a judgment about the precision of an implementation effect.

13. Why This Trial Matters Statistically

TIPPI addresses a service-delivery question: whether approaches intended to strengthen pediatric TB case detection, early accurate diagnosis and effective treatment were associated with changes in selected outcomes. The primary endpoint focuses on screening children age 0–14 among clinic attendees. From a biostatistical perspective, this makes the denominator and the definition of the counted unit important, not just the screening count itself.

The registry describes an evaluation across ten countries and up to 450 sites, with routine TB program service delivery as a comparator in a subset of project sites. That setting highlights a common analytical challenge: implementation evaluations may operate across sites with different attendance patterns, documentation practices and local contexts. General statistical note: comparisons across time and settings need to distinguish intervention-associated change from variation in who attended, how data were recorded and what other changes occurred.

The record also illustrates why registration details and outcome results need to be read separately. The registry gives a study enrollment of 4,159,533 and an endpoint population description of 7,198,865 clinic attendees, alongside 4,749,072 children screened. It does not explain the relationship among these figures. Maintaining those distinctions avoids treating an enrollment label as if it were automatically the denominator for a service-delivery outcome.

Statistically, the most consequential gap is that no inferential analysis is reported for the primary endpoint. Without period-specific estimates and uncertainty, the size and precision of any change cannot be evaluated. The nonrandomized design and the registry’s warning about non-comparable arms further limit causal interpretation. These are not reasons to discard descriptive data; they define what those data can support.

For readers assessing implementation evidence, the record underscores the value of clear outcome definitions, explicit analysis populations, transparent pre/post windows, and methods that reflect the structure of site-level data. Those details are not reported in the registry record, so the posted information supports a cautious descriptive reading rather than a firm comparative conclusion.

14. 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 main limitation for interpreting this evaluation is that the registry record does not report an inferential analysis of the primary endpoint. Although it identifies 4,749,072 children screened and 7,198,865 clinic attendees, it does not report the calculated proportion, a pre/post contrast, an effect estimate, a confidence interval or a test of statistical significance. Precision and comparative change therefore cannot be assessed from the registry entry.

The design is described as pre- and post-implementation rather than randomized. Allocation is not stated, and the record cautions that the arms are not a like-for-like control comparison. The approach to selecting sites or comparison periods, and any adjustment for differences between them, are not reported in the registry record. This limits causal interpretation of any apparent difference.

Several details needed to understand the endpoint are also absent. The exact pre- and post-implementation windows, the sites contributing data, the counting rules for children screened, and the handling of repeat attendance are not reported. The registry does not explain why its enrollment of 4,159,533 differs from the endpoint description’s 7,198,865 clinic attendees, or how these quantities relate.

Data completeness and analysis choices are unclear. Missing-data procedures, inclusion and exclusion rules, analysis populations, statistical models and any adjustment for site-level clustering are not reported in the registry record. The record lists twelve outcome measures and zero analyses, and does not describe a multiplicity approach.

These limitations concern what can be established from the registry record. They do not demonstrate that the evaluation lacked additional analyses or that data were incomplete; such details are not reported in the registry record.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: TB Innovation Project: A Pre- and Post- Implementation Assessment (TIPPI) — NCT03948698.
• Registry record last updated 2023-04-18; status last verified 2023-04.
• Results first posted 2021-12-20.
• 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.

18. Record Summary

TIPPI is a completed pre- and post-implementation assessment of the Catalyzing Pediatric TB Innovations project, conducted across ten countries. The sponsor is the Elizabeth Glaser Pediatric AIDS Foundation. The registry lists a start date of 2018-12-01 and a primary completion date of 2021-09-30.

The primary endpoint counts children age 0–14 screened for TB among clinic attendees during December 2018–September 2021. The registry identifies 4,749,072 children screened and 7,198,865 clinic attendees in the endpoint population description. The overall enrollment is listed as 4,159,533; its relationship to the endpoint figures is not explained.

The evaluation description refers to up to 450 sites and comparison with routine TB program outcomes in a subset. It was not randomized, allocation is not stated, and the registry warns that the arms are not a like-for-like control comparison. The record lists twelve outcome measures and zero analyses; an inferential result for the primary endpoint is not reported in the registry record.