Plan and Pledge, HIV Self-testing in South Africa: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Plan and Pledge, HIV Self-testing in South Africa (NCT03898557), a randomized na study. The primary outcome is Report of HIV Self-test Result.

2. Clinical Question

This is a behavioral implementation question: the intervention adds a prompt to plan and pledge to an existing self-test distribution setting. The registry describes the pilot objective as examining whether commitment strategies can increase uptake of HIV self-testing. Its listed primary outcome, however, is framed around recipients reporting a self-test result and being HIV positive. That distinction matters: uptake, test completion, reporting a result, and reporting a positive result are not interchangeable outcomes. The registry record does not provide an outcome estimate showing how the arms compared on any of these measures.

The trial concerns people receiving self-tests through the STAR fixed-site distribution program implemented by Wits Reproductive Health and HIV Institute at the University of Witwatersrand in Johannesburg, South Africa. The registry record does not report further details about participants' ages, eligibility, or baseline characteristics. The comparison therefore needs to be read in the context of the specific distribution program and the outcome definition recorded for this pilot, rather than assumed to represent every self-testing setting.

3. Trial Design

The registry describes Plan and Pledge as a randomized, parallel-design pilot with no masking. It lists 1479 participants enrolled and marks the study as completed. The recorded start date is 2019-05-28, and the primary completion date is 2019-06-18. The registry identifies the condition as Human Immunodeficiency Virus and the sponsor as the University of Pennsylvania.

The two arms are Usual Care and Plan and Pledge. Both descriptions concern provision of an HIV self-test and use of a dedicated WhatsApp number for result reporting. The Usual Care card was described as similar to the existing card used in the STAR program. The Plan and Pledge arm was described as receiving that card together with a brief template for making a plan and pledge for test completion and results reporting. These descriptions identify the intended behavioral contrast; the registry does not report whether participants completed each step as intended or how consistently the prompts were delivered.

The record's analysis note identifies a cluster-randomized design and cautions that an individual-level re-analysis would be misleading. The registry does not describe the number or definition of clusters, the unit used for allocation, or how clustering was handled in the sponsor's analysis. Those omissions prevent a detailed reconstruction of the design and its implications for statistical precision. Although the record labels the allocation as randomized and the model as parallel, those labels do not replace a description of cluster structure or the analysis method.

The registry lists no analysis results. It also does not report the randomization procedure, allocation concealment, sample-size justification, protocol deviations, or any interim analysis plan. Such information is not reported in the registry record, so the design description should not be expanded beyond the stated randomized, parallel comparison and the analysis note about clustering.

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: Field staff were not considered enrolled in this study.

Overall Study

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 Report of HIV Self-test Result, measured over approximately 3 months.

The primary endpoint is titled Report of HIV Self-test Result. The registry describes it as the percent of self-test recipients who report their self-test result on the dedicated WhatsApp number and are HIV positive, by study arm, compared with the standard-of-care arm of the STAR program. Its unit is participants, its measure type is a count of participants, and its timeframe is approximately 3 months.

This endpoint is intended to capture a reporting outcome connected to the self-testing program. A participant must report a result through the specified channel to be represented by the reporting component of the endpoint; the description also specifies HIV-positive status. It should therefore not be casually paraphrased as all test completion, all result reporting, or the overall proportion of people with HIV. The registry lists Report of Positive HIV Self-test Result as a secondary outcome, but does not give the definition or results for that measure.

General: an endpoint needs a clear definition because the interpretation of a treatment comparison depends on exactly what counts as an event and who belongs in its denominator. In this record, the intended outcome description is available, but no observed numerator, denominator, percentage, or between-arm estimate is reported. The registry record also does not explain the handling of participants who did not report a result or whose result status was unknown.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The methods information supports a limited explanation of how a comparison might be interpreted, but not a reconstruction of the trial's analysis. The registry identifies the study as randomized and parallel, states that masking was none, and flags a cluster-randomized design. It does not report the cluster unit, the allocation procedure, the statistical model, or an analysis plan. Those details are important for understanding whether the uncertainty estimates account for the design.

Why cluster structure matters

General: in a cluster-randomized study, groups rather than necessarily separate individuals are the units assigned to study arms. Outcomes within one group can resemble one another because members share a setting, delivery process, or other context. This dependence reduces the amount of independent information relative to treating every person as unrelated. A valid analysis normally needs to reflect the cluster structure, for example through a cluster-level approach or a model that accounts for within-cluster dependence. The suitable method depends on design details and outcome data. The registry's analysis note explicitly cautions that a simple individual-level re-analysis would be misleading, but it does not give enough detail to identify the sponsor's method.

What a binary outcome means

The primary outcome is identified as binary and as a count of participants with an event. General: a binary outcome classifies each eligible participant into one of two categories, such as whether the defined event occurred or did not occur. A count can be converted into a proportion only when the corresponding denominator and analysis population are known. The primary outcome description refers to participants who report a result and are HIV positive, but the registry does not give the event count, denominator, or missing-result handling. An effect estimate cannot be recovered from the enrollment total alone.

General: a comparison of binary outcomes may be expressed as a difference in proportions, a ratio of proportions, or another model-based contrast. Each measure answers a somewhat different question. A confidence interval is used to show the uncertainty around an estimated contrast, conditional on the statistical model and its assumptions. A p-value tests compatibility with a specified null hypothesis under that model. The registry record reports none of these quantities for the primary comparison, so no effect measure or precision statement can be given.

Over-dispersion and uncertainty

General: over-dispersion occurs when observed variation is greater than a model expects. In cluster studies, ignoring within-cluster similarity can understate uncertainty; other forms of extra variation can also make a simple model fit poorly. Methods that account for clustering or extra variation can widen uncertainty appropriately. Choosing such a method requires information about the cluster structure and outcome distribution, which is not reported in the registry record. It would be inappropriate to select a correction or quantify its effect without those details.

The registry's outcome measure is described as a count of participants, not as a time-to-event outcome. The primary timeframe is approximately 3 months. General: for a fixed follow-up outcome, interpretation depends on a clearly specified assessment window and rules for participants without a recorded outcome. A time-to-event method would answer a different question and would require event timing and censoring information. No such event-time analysis is described for the primary endpoint.

Planning calculations are not results

General: a sample-size calculation is a planning exercise based on assumptions about the effect to detect, outcome variability, allocation, and the desired error rates. It is not evidence that the planned effect occurred, and it is not a substitute for an estimate with an uncertainty interval. The registry's stated planning illustration uses the observed effect, two-sided significance level 0.05, equal allocation, and a normal approximation; it describes the target effect as a stated fraction of the observed effect. The record does not report the resulting planning values in the clinical information used for this interpretation. In any case, such illustrations are not the trial's own design justification and must not be read as observed power.

The record does not report whether the study used an adjusted or unadjusted primary analysis, how clustering was incorporated, or whether any covariates were included. It also does not report model diagnostics, sensitivity analyses, or a prespecified analysis plan. These are not minor details: they affect which population the estimate represents and how much uncertainty should be attached to it. Without the sponsor's actual analysis output and relevant design details, the sound statistical conclusion is that the methods cannot be independently assessed and the primary effect cannot be quantified.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists six outcome measures and identifies Report of HIV Self-test Result as primary. It also lists Report of Positive HIV Self-test Result as a secondary outcome. The record does not report a multiplicity strategy, such as whether the primary outcome was the sole confirmatory test, whether secondary outcomes were adjusted for multiple comparisons, or how outcomes were prioritized in interpretation. This information is not reported in the registry record.

General: when multiple outcomes or comparisons are examined, the chance of at least one apparently unusual result can increase if each is tested separately without accounting for the number of tests. Prespecifying one primary outcome and a testing hierarchy can help clarify which conclusions are confirmatory and which are exploratory. The presence of a primary label alone does not establish what adjustment rules were used, especially when the analysis plan is not reported.

The registry does not report missing primary-outcome data, the number of participants with known result status, or how people who did not report via WhatsApp were classified. It also does not state whether analyses followed the assigned arm regardless of intervention uptake, or whether a different analysis population was used. These details are necessary to understand the denominator and potential effects of missingness.

General: missing outcomes can affect an estimate when the probability of being observed is related to arm assignment or to the outcome itself. Complete-case analysis, imputation, and sensitivity analysis make different assumptions; none can be attributed to this trial without a report of what was done. The registry's enrollment count and the adverse-event at-risk counts do not resolve the primary-outcome missing-data question. Accordingly, the robustness of any outcome comparison to missingness and the intended analysis population are not reported in the registry record.

11. 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's serious adverse event listing records 747 participants at risk in Usual Care and 731 in Plan and Pledge. It lists zero serious adverse events in each arm, zero other adverse events in each arm, and zero deaths in each arm. No serious adverse event terms are listed.

These are the event counts recorded in the registry, not a complete account of how safety events were collected, defined, reviewed, or attributed. The record does not report event surveillance procedures, duration of safety follow-up, severity criteria, or whether events were assessed for relatedness to the intervention. It also does not explain the difference between the enrollment total and the numbers listed at risk in the safety entries. The safety counts should therefore be read within the limits of the registry's reporting.

General: observing zero recorded events does not prove that an event is impossible or that the risk is exactly zero. The ability to characterize uncommon events depends on the amount and duration of observation and the completeness of reporting. Since no uncertainty interval or detailed safety analysis is reported, the zero counts should not be converted into a claim about the absence of risk. They are a description of what the registry lists, not a stand-alone safety conclusion.

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

The registry documents a pilot that compared an HIV self-test distribution approach with commitment and planning prompts against a Usual Care approach. The intervention description adds a planning and pledge template to a WhatsApp reporting card. The trial was recorded as randomized, parallel, and unmasked, and the analysis note identifies a cluster-randomized design.

The record does not report the primary endpoint results. It gives no arm-specific event counts, percentages, treatment contrast, confidence interval, or p-value. Therefore, the posted information does not establish the direction, size, or precision of a difference in the primary outcome. It cannot support a conclusion that the prompts changed reporting, test uptake, or the likelihood of a reported positive self-test.

The outcome definition is narrower than a general measure of self-testing: it refers to recipients who report a result through the dedicated WhatsApp number and are HIV positive. Even if an estimate were available, interpretation would require attention to who was included in the denominator and how unreported results were handled. Those details are not reported in the registry record.

General: a randomized comparison can help assess differences between assigned approaches, but valid interpretation still depends on appropriate analysis, follow-up, outcome definition, and uncertainty estimates. Cluster randomization particularly requires accounting for the dependence among participants in the same cluster. Here, the record warns against an individual-level re-analysis, and it does not include enough information to independently verify the sponsor's analysis.

The appropriate takeaway is limited: the registry describes the intervention, comparator, intended outcome, and some design features, but does not present the outcome analysis needed to judge the observed comparison. Missing results should not be interpreted as a null finding, a benefit, or evidence of equivalence. They indicate that the registry information available for this trial is insufficient to establish what the primary outcome showed.

13. Why This Trial Matters Statistically

This trial is statistically noteworthy because it asks an implementation question through a behavioral intervention: whether adding prompts to an existing self-test distribution and reporting process changes a reported outcome. The comparison is not simply between receiving and not receiving a test. Both arms are described as receiving an HIV self-test, while the Plan and Pledge arm adds a planning and commitment template to the WhatsApp card.

The endpoint emphasizes a particular point in the testing pathway: reporting a result through the dedicated channel, with the primary description further specifying HIV-positive status. Distinguishing testing, completion, reporting, and result status is important because a behavioral intervention might affect these stages differently. The registry objective discusses uptake, but the primary outcome wording should guide interpretation of any eventual estimate.

The cluster-randomized design is also important. General: when assignment occurs by group, the analysis needs to respect the grouping because participants within a cluster may not provide fully independent information. An individual-level analysis that ignores this feature can overstate precision. The registry explicitly warns against such a re-analysis, while leaving the cluster structure and sponsor's statistical method undescribed.

Finally, the record illustrates a basic distinction between a trial description and a results report. The design and intended endpoint are documented, but no primary outcome estimate is available in the registry record. A statistical reader should not fill that gap with assumptions or infer a null result. Without estimates and uncertainty, the comparative evidence cannot be evaluated for magnitude, precision, or compatibility with different effects. The trial's statistical importance is as a reminder that design labels and outcome definitions do not substitute for reported analysis results.

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 most important limitation for interpreting this trial is that the registry record does not report the primary outcome results. There are no outcome counts by arm, percentages, effect estimates, confidence intervals, or p-values. The registry lists zero analyses, so the comparative result cannot be independently evaluated from the posted information.

The analysis note identifies a cluster-randomized design and warns that simple individual-level re-analysis would be misleading. Yet the record does not describe the clusters, allocation unit, number of clusters, or the sponsor's method for accounting for clustering. These gaps limit assessment of both the estimate and its precision. The statistical analysis plan and sample-size justification are also not reported in the registry record.

The registry describes the primary endpoint as a percentage involving recipients who report their self-test result and are HIV positive, but does not provide the denominator definition, missing-result rules, or analysis population. It does not report follow-up completeness or whether the intervention was delivered or taken up as intended. These omissions make it impossible to separate outcome ascertainment from the behavioral processes that the intervention was designed to influence.

The record describes a pilot in the STAR fixed-site distribution program in Johannesburg, South Africa. It does not report detailed participant characteristics or enough setting information to assess how closely the study population and delivery context match other programs. The absence of masking is recorded, but the registry does not state how outcome reporting or assessment was conducted beyond the dedicated WhatsApp channel. These limitations are specific to the available record and should not be filled with assumptions about the protocol or conduct.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Plan and Pledge, HIV Self-testing in South Africa — NCT03898557.
• Registry record last updated 2024-06-26; status last verified 2024-06.
• Results first posted 2024-06-26.
• 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

NCT03898557, titled Plan and Pledge, HIV Self-testing in South Africa, is recorded as a completed randomized, parallel pilot with no masking. The sponsor is the University of Pennsylvania. The recorded start date is 2019-05-28, the primary completion date is 2019-06-18, and enrollment is 1479.

The trial compared Usual Care provision of an HIV self-test with provision of an HIV self-test plus commitment and planning prompts. The primary outcome was Report of HIV Self-test Result, described as a measure of participants who reported their result through a dedicated WhatsApp number and were HIV positive, with a timeframe of approximately 3 months.

The analysis note identifies a cluster-randomized design and cautions against simple individual-level re-analysis. The registry record lists no analyses and does not report the primary outcome estimate. Safety entries list zero serious adverse events, zero other adverse events, and zero deaths in both arms. The outcome comparison and its statistical precision cannot be determined from the registry information.