A Study To Learn About The Effects Of Pneumococcal Vaccine In People With HIV: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study To Learn About The Effects Of Pneumococcal Vaccine In People With HIV (NCT05794191), a registered phase not stated study. The primary outcome is PCV13 Vaccine Effectiveness (VE) for First Event of Invasive Pneumococcal Disease (IPD): Overall Follow-up.

2. Clinical Question

The study focuses on an outcome associated with pneumococcal disease that has spread into the bloodstream. The brief summary also describes pneumonia as a disease of interest, and the registry lists additional pneumonia outcomes by follow-up period. Those secondary questions are distinct from the primary question, which concerns invasive pneumococcal disease over overall follow-up.

The clinical question is therefore about an association estimated from health care databases, rather than a direct randomized comparison. The analysis aims to account for measured differences in vaccination status and for informative censoring through weighting. Those methods can address specified sources of imbalance, but they do not turn an observational comparison into a randomized one. The registry record does not report the numerical primary result or enough information to judge the balance of all relevant prognostic factors.

3. Trial Design

This completed study assessed PCV13 effectiveness among people with HIV in the United States using health care databases. The registry reports an enrollment of 350399, a start date of 2023-03-22, and a primary completion date of 2024-08-15. The brief summary describes identifying adults with HIV from medical diagnoses in databases, identifying vaccination from vaccine administration or PCV13 records, and following participants for disease outcomes in those databases.

The comparison is observational. Random allocation is not stated, and the registry flags that the arms are not a like-for-like control comparison. The intervention is described as vaccine administration, and the arm descriptions distinguish PCV13-vaccinated participants from unvaccinated participants. However, vaccination status is explicitly time varying: participants could move from unvaccinated to vaccinated status during follow-up and could be counted in both arms according to their status over time. The unvaccinated arm's overall number analyzed is defined as participants unvaccinated at the start of the study.

For the primary outcome, follow-up runs from the index date until the earliest of death, the end of health plan enrollment, or the end of the study period, with a maximum of up to 8.25 years. The primary outcome is the first event of invasive pneumococcal disease over overall follow-up. The registry describes weighted marginal structural Cox models for estimating vaccine effectiveness. It also lists secondary effectiveness outcomes for first events of pneumococcal pneumonia or pneumonia with unspecified causes, and for pneumonia with unspecified causes, across stated follow-up intervals.

The record labels the study completed, but a full protocol, detailed eligibility criteria, exact database sources, and a description of how the index date was defined beyond the primary outcome timeframe are not reported in the registry record. Allocation procedures are not stated. These omissions limit how fully the observational design and cohort construction can be reconstructed from the registry alone.

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: A total of 350399 participants were enrolled in the study. The six months of continuous enrollment in medical and pharmacy plans between 01-Jan-2014 and 31-Dec-2021 was considered the baseline period. The end of the study occurred on the last date that administrative claims data were available on September 30, 2022. Participants were followed up from index date to the earliest of death, end of health plan enrollment or end of study period (maximum up to 8.25 years).

Overall Study

Analysis population for the primary outcome, as defined in the registry: PCV13 vaccination status was a time varying exposure and participants could be counted in both PCV13 vaccinated or unvaccinated arms, depending on vaccination status during follow-up. Here, 'Overall Number of Participants Analyzed (N)' for Unvaccinated arm=participants unvaccinated at start of study evaluable for this outcome measure (OM); N for vaccinated arm=participants vaccinated by end of follow up for this OM; hence, N is different from those reported under participant flow and baseline.

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: Analysis population included all eligible participants whose data were retrieved and observed in the study. Baseline characteristics were analyzed in participants with vaccinated (PCV13 vaccinated participants) and unvaccinated (Unvaccinated participants) status at the index date, which was defined as the last day of baseline period.

6. Primary Endpoint

The registry lists the primary outcome as PCV13 Vaccine Effectiveness (VE) for First Event of Invasive Pneumococcal Disease (IPD): Overall Follow-up, measured over From index date to the earliest of death, end of health plan enrollment or end of study period (maximum up to 8.25 years).

The primary endpoint is the first event of invasive pneumococcal disease during overall follow-up. The registry describes invasive pneumococcal disease as being identified using International Classification of Diseases, 9th revision, Clinical Modification, or ICD-10-CM codes. It specifies that the total number of participants with invasive pneumococcal disease is reported descriptively, while vaccine effectiveness is reported in the statistical analysis.

This endpoint is used to assess a clinically consequential form of pneumococcal disease in which infection involves the bloodstream. As a first-event endpoint, it focuses on time until the initial recorded event rather than counting every recurrence. That makes a time-to-event model relevant to the stated question, because the analysis can incorporate the timing of the first event and the duration participants remain under observation.

The registry defines vaccine effectiveness as one minus the hazard ratio, multiplied by 100. The hazard ratio comes from marginal structural Cox models after treatment and censoring weights are applied. In general, a hazard ratio compares event rates over follow-up between exposure groups under the model; it is not the same as a simple comparison of event proportions. The registry's description is important because vaccination status can change during follow-up, and observation can end for reasons such as death or loss of health plan enrollment.

The numerical count of participants with the primary endpoint and the numerical vaccine-effectiveness estimate are not reported in the registry record. The endpoint definition and model therefore explain what was intended to be measured, but do not establish the magnitude or precision of the estimated association.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary analysis is described as a marginal structural Cox model with inverse probability of treatment weighting and inverse probability of censoring weighting. These methods are intended to handle features of the data that matter for an observational study with changing vaccination status and potentially informative loss of follow-up. The registry says the effectiveness estimate was derived from this weighted model and defines effectiveness as one minus the hazard ratio, multiplied by 100.

Why use a time-to-event model?

The outcome is the first recorded event of invasive pneumococcal disease, and follow-up begins at an index date and ends at the earliest of death, the end of health plan enrollment, or the end of the study period. A Cox model is a time-to-event method: it uses information about when an event occurs and how long participants are observed, rather than treating every participant as having the same observation period. In general, the hazard is the instantaneous event rate among those still at risk. A hazard ratio compares those rates between exposure conditions as represented by the model.

A hazard ratio is not a direct risk difference, and the registry's effectiveness transformation should not be interpreted as a simple subtraction of event counts or proportions. The registry entries do not include the numerical hazard ratio, the numerical transformed effectiveness estimate, or confidence intervals. Their precision and compatibility with a range of effect sizes therefore cannot be assessed from this record.

Why can vaccination status change?

The registry explicitly treats PCV13 status as a time-varying exposure. Participants were not necessarily assigned a permanent vaccinated or unvaccinated label at the start of follow-up; someone could receive PCV13 during follow-up. The record notes that participants could be counted in both the vaccinated and unvaccinated arms depending on status during follow-up. This approach reflects changing exposure, but also makes the arm labels different from fixed groups in a randomized trial.

In general, time-varying exposures can create analytic challenges when factors that influence treatment also change over time and are related to the outcome. A marginal structural model is designed to estimate an exposure-outcome association while accounting for measured treatment-related differences through weighting. The registry describes the aim as reducing imbalances in potential confounders by vaccination status. It does not list the specific confounders, their measurement timing, or the diagnostics used to assess balance.

What do the weights do?

The inverse probability of treatment weights were generated from predicted probabilities of vaccination status, described in the record as propensity scores. In general, weighting gives more influence to observations that represent people with similar measured characteristics who had a different observed treatment probability. The intended result is a pseudo-population with reduced measured imbalance between exposure conditions. This is a statistical construction, not a new set of observed participants, and it does not guarantee balance on unmeasured factors.

The inverse probability of censoring weights address the possibility that ending observation is related to factors associated with the outcome. The registry says these weights were used to adjust for informative censoring. In general, censoring can be informative when the people who stop being observed differ in their outcome prospects from those who remain. Weighting can adjust for measured predictors of censoring when the model and available data support that adjustment. The record does not report the censoring model, the predictors used, or diagnostics such as the distribution of weights.

How should the model output be read?

The registry says that the total number of participants with invasive pneumococcal disease is reported descriptively, while effectiveness is reported in the statistical section. It also says the effectiveness analysis used IPTW, IPCW, and imbalanced variables. The numerical results are not reported in the registry record available for this explanation. As a result, it is not possible to assess whether the estimated association was precise, whether weighting materially changed the estimate, or whether any model checks identified problems.

In general, a confidence interval describes the statistical uncertainty around an estimate under the model and its assumptions. It does not capture every source of uncertainty, such as errors in coding or residual confounding. No primary confidence interval is reported in the registry record. The treatment and censoring weights likewise depend on correctly specified probability models and adequate measurement of relevant predictors; those assumptions and diagnostics are not fully described in the record.

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("PCV13-vaccinated Participants", "Unvaccinated Participants")
n      <- c(85584, 320959)
events <- c(167, 747)
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  PCV13-vaccinated Participants  85584  167  0.2  0.2  0.2  0.2  0.2  0.2  0.2
ARM  Unvaccinated Participants  320959  747  0.2  0.2  0.3  0.2  0.3  0.2  0.3

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 14 outcome measures and 20 analyses. The primary outcome is first-event invasive pneumococcal disease over overall follow-up. Secondary outcomes include PCV13 effectiveness for first events of pneumococcal pneumonia or pneumonia with unspecified causes, and for first events of pneumonia with unspecified causes, at the stated follow-up intervals. A procedure for controlling multiplicity, such as adjustment for making several statistical comparisons, is not reported in the registry record. The record therefore does not establish whether the analyses were treated as confirmatory, exploratory, or handled under a prespecified hierarchy.

The primary analysis accounts for censoring using inverse probability of censoring weights, which the registry says were intended to adjust for informative censoring. This addresses a particular analysis issue; it should not be taken as a description of how missing covariate values or incomplete diagnosis records were handled. Imputation, complete-case analysis, and other missing-data procedures are not reported in the registry record. The sources and completeness of the health care database variables are also not described in the registry entries.

The record describes participants as potentially contributing to both vaccination categories as status changes over follow-up. It defines the unvaccinated arm's overall number analyzed as participants unvaccinated at the start of the study. Beyond this description, the analysis population rules, exclusions, and handling of participants with incomplete exposure or outcome information are not reported in the registry record. The primary event count and numerical effectiveness result are also not reported in the registry record, limiting assessment of the analysis populations and the precision of findings.

Any sample-size figures in the planning material are illustrations based on an observed effect, not observed power and not the trial's own design justification. The stated assumptions describe a two-sided significance level of 0.05, equal allocation, and a normal approximation; time-to-event illustrations use a total event count, while other illustrations use participants per arm. The underlying numerical outputs and the trial's own sample-size justification are not reported in the registry record.

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

The study addresses whether PCV13 vaccination is associated with the timing of a first recorded invasive pneumococcal disease event among adults with HIV identified in health care databases. Its primary measure is a weighted hazard-ratio-based effectiveness estimate. The registry's methods describe adjustment for measured differences in vaccination and for informative censoring, and the exposure is allowed to change during follow-up.

The available the registry entries do not report the numerical event count or primary effectiveness estimate. They therefore do not support a statement about the magnitude, direction, or precision of the primary result. A methods description alone cannot show whether the estimated association was large or small, or how uncertain it was. The secondary outcome descriptions likewise do not provide numerical outcome results in the registry record for this analysis.

The design also matters for interpretation. Vaccination was not assigned through a stated random allocation, and the registry notes that the arms are not a like-for-like control comparison. Participants could be classified in both arms as vaccination status changed. Consequently, an adjusted association from this analysis should not be presented as equivalent to a result from a randomized comparison. Statistical weighting can address measured differences under assumptions; it cannot remove all possible confounding or errors in database measurement.

The endpoint concerns first events recorded through diagnostic codes, so its interpretation depends on how those codes identify invasive pneumococcal disease in the databases. The registry names the coding systems but does not report validation details in the record. The follow-up definition also depends on death, health plan enrollment, and the end of the study period. The censoring weights were intended to address informative censoring, but the details and diagnostics are not reported.

These points describe what the study's design and methods can support, not medical guidance. The registry record is informative about the planned analysis framework, but the missing numerical primary result and unreported details constrain conclusions about the observed association.

13. Why This Trial Matters Statistically

This study is statistically notable because it evaluates a vaccine in routine health care databases while vaccination status can change during follow-up. A fixed vaccinated-versus-unvaccinated grouping could misrepresent when exposure begins. Treating vaccination as time varying allows the analysis to represent changes in status, though it also requires careful handling of factors related to both vaccination and disease risk.

The use of marginal structural Cox models with inverse probability of treatment and censoring weights reflects an effort to address measured confounding and potentially informative follow-up loss. These methods are relevant to observational data where exposure and observation processes are not necessarily independent of participant characteristics. Their value depends on data quality, model specification, and adequate measurement of the factors used to calculate weights. The registry description does not report the specific predictors or diagnostics, so those features cannot be evaluated from the available record.

The study also illustrates why a registry's design and results need to be read together. The record states that the comparison is not randomized and that participants may contribute to both exposure arms as vaccination status changes. That distinction is essential when interpreting an effectiveness estimate: adjustment can improve comparability with respect to measured factors, but does not make the groups equivalent by design.

Finally, the record defines a time-to-event primary endpoint and a follow-up period that can end for several reasons, but it does not report the numerical primary estimate or its uncertainty in the registry record for this analysis. The design can be described precisely, yet the result cannot be interpreted quantitatively without those values. This is a useful reminder that model sophistication and a clearly defined endpoint do not substitute for reporting event counts, effect estimates, and uncertainty.

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 the primary finding is that its numerical event count and vaccine-effectiveness estimate are not reported in the registry record. Without those values, the direction, magnitude, and precision of the association cannot be assessed. The record also does not report a primary confidence interval or enough numerical results to compare the descriptive count with the adjusted estimate.

The observational comparison has limitations beyond missing numerical results. Random allocation is not stated, and the record explicitly cautions that the arms are not a like-for-like control comparison. Although weighting was used to reduce imbalances in potential confounders and adjust for informative censoring, the specific covariates, model specifications, weight diagnostics, and balance assessments are not reported in the registry record. Residual confounding therefore remains a concern that cannot be quantified from the available description.

Exposure status was time varying, and participants could be counted in both arms during follow-up. This is an important design feature, but the record does not provide detailed rules for exposure timing, latency, or how vaccination dates were established beyond database records of vaccine administration or PCV13. It also does not report validation of the diagnostic codes used to identify invasive pneumococcal disease.

Details about database sources, missing covariate data, exclusion criteria, and the analysis population beyond the stated arm definition are not reported in the registry record. Nor is a multiplicity procedure reported despite the listed outcome measures and analyses. These gaps limit assessment of data completeness, model robustness, and the role of secondary analyses.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: A Study To Learn About The Effects Of Pneumococcal Vaccine In People With HIV — NCT05794191.
• Registry record last updated 2025-10-20; status last verified 2025-09.
• Results first posted 2025-10-20.
• 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

The registry describes a completed, database-based assessment of PCV13 effectiveness among adults with HIV in the United States. It reports an enrollment of 350399, with a start date of 2023-03-22 and a primary completion date of 2024-08-15. The primary outcome is a first event of invasive pneumococcal disease during overall follow-up, identified using ICD-9-CM or ICD-10-CM codes.

Vaccination status was time varying. The analysis used marginal structural Cox models with inverse probability of treatment and censoring weights, and defined effectiveness as one minus the hazard ratio, multiplied by 100. Follow-up ended at the earliest of death, the end of health plan enrollment, or the end of the study period, with a maximum up to 8.25 years.

The record also lists pneumonia-related secondary outcomes. The primary event count, numerical effectiveness estimate, and confidence interval are not reported in the registry record. Allocation is not stated, and the registry cautions that the comparison is not a like-for-like control comparison.