Hybrid Effectiveness-Implementation Study to Improve Clopidogrel Adherence: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Hybrid Effectiveness-Implementation Study to Improve Clopidogrel Adherence (NCT01609842), a randomized na study. The primary outcome is Percentage of Adherent Patients.

2. Clinical Question

This is an effectiveness and implementation question: it concerns adherence in routine care as well as the delivery of a multi-part support strategy. The registry record does not report a primary outcome estimate, an uncertainty interval, or a statistical test comparing the study groups. Therefore, the posted information does not establish the direction or size of any difference in adherence.

3. Trial Design

The registry describes a randomized, parallel study with no masking. A separate design note identifies the study as cluster-randomized. In a cluster-randomized design, groups such as clinical sites, rather than individual participants, are assigned to study conditions. The number and identity of randomized clusters, the allocation procedures, and any stratification are not reported in the registry record. The listed enrollment is 14573.

The intervention was described as a multifaceted strategy involving an inpatient pharmacist or designated study team member and IVR refill reminder calls. Usual-care sites were described as having no interaction with study personnel, with database information used for comparison. The registry record does not report whether implementation of the intervention differed across sites or how adherence to the intervention procedures was assessed.

The recorded start date is 2014-01-20, the primary completion date is 2018-03-01, and the study status is completed. The phase is recorded as NA. The sponsor is the VA Office of Research and Development. The primary outcome timeframe is 12 months post PCI discharge. Hospitalizations and mortality are listed as secondary outcomes.

The primary outcome description says mixed logistic regression models for clopidogrel adherence were planned. However, the registry record lists no analyses, and it does not report the primary outcome results, model estimates, or uncertainty measures. The details of the statistical plan beyond the stated model 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: The participants reported in the Control Arm/Group were all unique patients; they were assigned as follows: pre-intervention control (n=6,081), post-intervention control (n=2,917), and concurrent intervention controls (n=4,869); there was no overlap between the control groups.

Pre-Intervention Control

Post-Intervention Control

Concurrent Intervention Control

Reasons for not completing this period:

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 Percentage of Adherent Patients, measured over 12 months post PCI discharge.

The primary endpoint is described as the percentage of patients whose clopidogrel prescription was filled at hospital discharge after PCI with stent placement, together with the percentage classified as adherent according to pharmacy refill data in the year after discharge. The adherence classification was binary: PDC greater than 80% was the stated threshold. PDC expresses the proportion of a follow-up period for which medication is available based on refill records.

Refill-based adherence is used to assess medication availability over time in routine care. It can capture whether prescriptions are refilled, but it does not by itself establish that medication was taken as prescribed. The discharge-fill measure and the later refill-based measure address different points in the treatment pathway: filling the prescription at discharge and continued medication availability during follow-up.

The endpoint description refers to both percentages, but the registry record does not report their values, group-specific results, or the number of participants contributing data to either measure. It also does not report how incomplete pharmacy records or other missing adherence information were handled. Although mixed logistic regression is stated as the planned approach for the binary adherence outcome, the registry record does not provide the fitted model results.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

General statistical explanation: a binary outcome places each participant into one of two categories. In this study’s primary outcome description, the adherence category was based on whether PDC exceeded 80%. Converting refill data to a binary outcome can make the result straightforward to summarize as a proportion meeting the threshold, but it also discards detail about how far a participant’s PDC lies from that cutoff. The registry record does not report a separate continuous PDC analysis.

Why use logistic regression?

General statistical explanation: logistic regression is commonly used when the outcome has two categories. It models the relationship between explanatory variables and the odds of the outcome. A model coefficient can be expressed on a log-odds scale or transformed to an odds ratio, but an odds ratio is not the same as a difference in probabilities. Interpretation depends on the model specification and the underlying outcome frequency. The registry record says mixed logistic regression was planned but does not provide coefficients, odds ratios, adjusted probabilities, covariates, or details about how the model was fitted.

Why does clustering matter?

General statistical explanation: participants treated within the same cluster may have more similar outcomes than participants in different clusters. For example, shared clinical workflows or local implementation conditions can create within-cluster dependence. A model that accounts for clustering is important because an analysis that treats every participant as fully independent may understate uncertainty. The study is identified as cluster-randomized in the registry design note. The registry record does not report the number of clusters, cluster sizes, the method used to represent clustering in the planned mixed model, or any estimated within-cluster correlation.

Mixed models can represent both participant-level information and variation associated with grouped settings, depending on their specification. This general description does not identify which random effects or other model components were planned for this study. Those details, as well as any fixed-effect covariates, are not reported in the registry record. The term mixed logistic regression alone is not enough to reconstruct an analysis or its estimates.

What does the PDC threshold mean?

General statistical explanation: PDC is based on the portion of a follow-up period during which refill records indicate medication availability. Applying a cutoff turns that measure into an adherent or not-adherent classification. The stated cutoff for this study was greater than 80%. A threshold can aid interpretation, but participants close to either side of the cutoff may have similar refill patterns while being placed into different outcome categories. The registry record does not report sensitivity analyses using another threshold, or results based on the underlying PDC values.

How should estimates and uncertainty be read?

General statistical explanation: an effect estimate describes the direction and size of a contrast under the chosen model. A confidence interval communicates uncertainty around that estimate, conditional on the data and modeling assumptions. A wide interval indicates that a range of effect sizes remains compatible with the analysis; a narrow interval indicates greater precision. A p-value addresses compatibility with a specified null hypothesis under the model and is not a measure of clinical importance. No primary effect estimate, confidence interval, or p-value is reported in the registry record, so none of these interpretations can be applied to the adherence comparison.

What can planning calculations tell readers?

General statistical explanation: sample-size calculations can illustrate how many participants per arm might be needed under stated assumptions about an effect, significance level, allocation, and approximation. Such planning values are not observed trial results and are not observed power. The registry record describes a planning illustration based on a fraction of the observed effect, with a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It does not report the numerical planning outputs in the registry record for this summary. A planning illustration also does not replace the design’s cluster structure; assumptions about clustering are not described in the registry record.

For time-to-event outcomes, a general sample-size approach may focus on the total number of events, while other outcomes may be expressed as participants per arm. Hospitalizations and mortality are listed as secondary outcomes, but the record does not provide a time-to-event analysis plan or results. The primary endpoint is a binary adherence measure, and the registry record reports no analysis output for it.

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("Intervention", "Pre-intervention Control", "Post-intervention Control", "Concurrent Intervention Control")
n      <- c(706, 6081, 2917, 4869)
events <- c(590, 4484, 2209, 3633)
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  Intervention  706  590  83.6  80.6  86.2  80.7  86.1  80.7  86.1
ARM  Pre-intervention Control  6081  4484  73.7  72.6  74.8  72.6  74.8  72.6  74.8
ARM  Post-intervention Control  2917  2209  75.7  74.1  77.3  74.1  77.2  74.1  77.2
ARM  Concurrent Intervention Control  4869  3633  74.6  73.4  75.8  73.4  75.8  73.4  75.8

11. Multiplicity, Missing Data and Analysis Populations

The record lists five outcome measures and identifies hospitalizations and mortality as secondary outcomes. It does not report the full set of outcome definitions, a hierarchy among endpoints, or a plan for controlling the chance of false-positive findings when more than one outcome or comparison is examined. Any adjustment for multiple comparisons, or decision to make none, is not reported in the registry record.

The primary outcome description says mixed logistic regression was planned, but the registry record lists zero analyses. It does not give model estimates or indicate which analysis populations were used. Definitions such as intention-to-treat, per-protocol, or modified intention-to-treat are not reported in the registry record. The exact participants included in each outcome analysis are also not reported.

Missing-data methods are not described. The record does not state how missing pharmacy refill data, incomplete follow-up, withdrawal, or other unavailable outcome information were handled. It also does not report the amount or pattern of missing data. These details matter because an adherence measure based on refill records may be unavailable for some participants, and the way unavailable values are treated can affect both the estimated outcome and its uncertainty.

For a cluster-randomized trial, the number and sizes of clusters and the handling of cluster-level dependence are relevant to analysis and precision. Those details are not reported in the registry record. The stated mixed-model plan is not sufficiently detailed to establish how the cluster structure was represented. Without these statistical and missing-data details, the planned analysis cannot be fully evaluated or reproduced from the registry record.

12. Safety Results

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

Deaths

Other (non-serious) adverse events

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

Most frequent serious adverse event terms

Serious events with the highest arm-level rate: other hospitalizations (12.9%), reported hospitalizations of the participants (0.8%).

The registry record lists serious adverse event counts separately for the intervention arm and usual care. The intervention arm has 706 participants recorded as at risk and 99 recorded as affected; usual care has 13867 recorded as at risk and 1782 recorded as affected. These are counts, not adjusted estimates of a between-group effect. The record does not report the event definitions, the observation period for these safety entries, or a statistical comparison.

The top serious-event terms include Other hospitalizations at 12.9% and Reported hospitalizations of the participants at 0.8%. The registry record does not clarify the relationship between these terms or the basis of those percentages. It also lists 34 deaths in the intervention arm and 858 in usual care. These entries are descriptive counts; the registry record does not report whether deaths were adjudicated as related to the intervention or provide a comparative mortality analysis.

The counts should not be interpreted as showing that the intervention caused or prevented serious events. Differences in the numbers recorded at risk, follow-up, ascertainment, and analysis could affect how group counts compare. The other adverse-event fields are recorded as zero, but their definitions and data-collection context are not reported in the registry record.

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

The trial addresses a practical clinical research problem: supporting clopidogrel adherence after PCI through pharmacy involvement and IVR refill reminders. Its primary outcome description focuses on prescription filling at discharge and pharmacy refill-based adherence during the year after discharge. The design is cluster-randomized, which is relevant to how comparisons and uncertainty should be analyzed.

The registry record does not report the primary adherence results. It therefore does not show whether the intervention and usual-care groups differed, the direction of any difference, or the precision of an estimate. A completed study and a large enrollment count are not themselves evidence of an effect. The planned use of mixed logistic regression describes an intended method, not an observed result.

The record also does not provide enough statistical detail to evaluate the primary analysis fully. The number of clusters, model specification, covariate adjustment, missing-data procedures, and analysis population are not reported. These omissions limit reproducibility and make it impossible to assess whether the analysis appropriately reflected the cluster-randomized design.

The serious adverse event entries provide counts and selected event labels, but they do not establish causation or a comparative safety effect. No adjusted safety estimates or uncertainty intervals are reported. Interpretation of these entries is further limited by the lack of reported event definitions and ascertainment details.

General statistical principle: evidence about an intervention depends on the contrast between groups, the precision of that contrast, and the assumptions used in the analysis. Without the primary estimates and their uncertainty, the registry information supports a description of the study’s aims and methods but not a conclusion about the effect of the intervention on adherence. It also cannot establish that the intervention changed hospitalization or mortality outcomes.

14. Why This Trial Matters Statistically

This study illustrates why the unit of randomization matters. Although the record describes individual participant counts, a separate design note identifies the trial as cluster-randomized. Participants in the same cluster may share care processes and implementation conditions, so statistical methods should account for dependence within clusters. A simple individual-level re-analysis could give misleading uncertainty estimates; the sponsor’s own analysis takes precedence, but no sponsor analysis is reported in the registry record.

The endpoint also demonstrates a common measurement choice in adherence research: using pharmacy refill records to classify whether a threshold is met. That approach makes the outcome operational and reproducible when refill data are available, but a binary cutoff can obscure differences in the underlying refill patterns. The registry record does not provide continuous PDC results or sensitivity analyses around the stated threshold.

Another statistical lesson is the distinction between a planned analysis and a reported result. The registry describes mixed logistic regression as planned, yet lists no analyses and gives no primary outcome estimate or uncertainty interval. A model name alone cannot show whether the analysis accounted for clustering adequately, which variables were included, or how missing data were handled.

Finally, serious event counts and mortality counts are not substitutes for comparative analyses. Clear definitions, comparable observation, and appropriate uncertainty estimates are needed to interpret safety outcomes. This record therefore offers a useful design and endpoint description, while leaving the central effectiveness and precision questions unresolved in the reported information.

15. Limitations

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

A major limitation of the available registry record is that it does not report the primary outcome results, despite the study being marked completed. There are no group-specific adherence values, effect estimates, confidence intervals, or p-values. The record therefore cannot support an assessment of the direction, size, or precision of the primary comparison.

The cluster-randomized design raises additional reporting needs. The number of randomized clusters, their sizes, the method of allocation, and the way clustering was represented in the model are not reported in the registry record. The description of mixed logistic regression does not specify the model structure or adjustment variables. No analysis population or missing-data approach is reported either.

The primary outcome description combines discharge prescription filling with refill-based adherence over the year after discharge. The record does not give separate results for these measures or explain how each was analyzed. It does not report the completeness of pharmacy data, loss to follow-up, or sensitivity analyses using the underlying PDC measure rather than its binary threshold.

For safety, the registry provides serious event counts and selected event terms, but the definitions, ascertainment methods, and observation period are not reported. No comparative estimates or uncertainty measures are included. Consequently, the counts should be treated as descriptive entries rather than evidence of a causal safety effect.

The record’s planning assumptions describe a sample-size illustration, not the trial’s own design justification. The design justification and any protocol-level sample-size rationale are not reported in the registry record. Interim analyses and the broader statistical analysis plan are also not reported.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Hybrid Effectiveness-Implementation Study to Improve Clopidogrel Adherence — NCT01609842.
• Registry record last updated 2023-07-21; status last verified 2023-07.
• Results first posted 2020-01-27.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

19. Record Summary

This completed, cluster-randomized, parallel study compared a multifaceted pharmacist and IVR intervention with usual care to address clopidogrel adherence after PCI. The registry lists enrollment of 14573, no masking, and a primary outcome timeframe of 12 months post PCI discharge. The primary outcome was described using discharge prescription filling and pharmacy refill-based adherence, with adherence defined by PDC greater than 80%.

Mixed logistic regression was described as the planned method for the binary adherence outcome. The registry record lists no analyses and does not report the primary outcome estimate, group-specific adherence results, or uncertainty measures. It also does not report key details such as cluster counts, model specification, missing-data handling, or analysis populations. Serious adverse event counts and deaths are listed descriptively, without a comparative analysis. The available record describes the study question and design but does not establish the intervention’s effect on adherence or safety.