A Program to Reduce Inappropriate Medications Among Older Adults With Alzheimer's Disease: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Program to Reduce Inappropriate Medications Among Older Adults With Alzheimer's Disease (NCT05147428), a randomized na study. The primary outcome is Time to Dispensing of Inappropriate Medication Prescription.

2. Clinical Question

The study’s overarching aim was to assess whether a patient- and caregiver-centered educational intervention could reduce potentially inappropriate prescribing in people with Alzheimer’s disease and related dementias. Its stated hypothesis connected education for patients or caregivers and providers with lower medication-related morbidity and improved medication safety. Those broader goals are not identical to the primary endpoint: the primary outcome measured dispensing of targeted medication classes, not medication-related morbidity, quality of life, function, or overall safety. Those outcomes should not be inferred from the primary result.

3. Trial Design

D-PRESCRIBE-AD (NCT05147428) was a randomized, parallel-group trial with no masking. The registry lists an enrollment of 14442 and three groups: Patient-Provider, Provider Only, and Usual Care. The reported primary outcome analysis was based on a modified intention-to-treat cohort. The record does not explain the modifications that define that cohort or report the allocation ratio.

The Patient-Provider group received educational materials about inappropriate prescribing and deprescribing, as did providers in the Provider Only group. Usual Care was the comparator, but its contents are not reported in the registry record. The study description says it used the NIH Collaboratory’s Distributed Research Network, which employs FDA Sentinel System infrastructure. Further details about participating health plans, sites, or implementation are not reported in the registry record.

The primary endpoint was assessed during a 6-month observation period after intervention receipt. Its event window was day 91 to day 270. The registry reports a primary analysis using a log-rank method and a Cox proportional hazards model. The study is listed as completed, and its phase is recorded as not applicable. The listed start date is 2022-06-01 and the primary completion date is 2023-10-01.

The registry record does not report the randomization procedure, concealment arrangements, sample-size justification, or whether any interim analyses were planned or conducted. It also does not give a detailed protocol for delivery or adherence to the educational materials. These omissions limit assessment of how the intervention was implemented and how closely the analysis followed a prespecified plan.

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: To address the potential for within-prescriber contamination, only 1 randomly selected eligible patient was randomized per prescribing clinician.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Modified intention to treat cohort.

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: Modified intention to treat cohort.

6. Primary Endpoint

The registry lists the primary outcome as Time to Dispensing of Inappropriate Medication Prescription, measured over 6 months.

The primary endpoint was time to dispensing of an inappropriate medication prescription. The registry defines an event as dispensing from a targeted inappropriate prescription class between day 91 and day 270 during the 6 months following receipt of the intervention. It reports the outcome as a percentage of cumulative incidence, with a 95% confidence interval, for the modified intention-to-treat cohort.

A time-to-event endpoint incorporates both whether an event is recorded and when it occurs within the defined observation window. This can be useful when the timing of dispensing is relevant and participants may experience the event at different times. A cumulative incidence percentage describes the proportion estimated to have experienced the event over the specified period; it is not the same measure as a hazard ratio, which compares event rates over time under the fitted model.

The endpoint focuses on dispensing in targeted medication classes. It does not directly measure whether a prescription was clinically appropriate for an individual, whether medication use changed outside the specified event window, or whether participants experienced medication-related harm. The registry record does not identify the targeted medication classes in the primary endpoint description. The rationale for choosing this particular window and endpoint is not reported in the registry record. As a result, the endpoint supports interpretation about recorded dispensing as defined, rather than every aspect of prescribing quality or medication safety.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary statistical analysis used a log-rank test and a Cox proportional hazards model. The group estimates were reported as cumulative incidence percentages with 95% confidence intervals. These methods address related but different aspects of the endpoint: cumulative incidence describes event occurrence over the observation period, while the Cox model estimates a relative event rate over time.

What does the hazard ratio describe?

In general, a hazard ratio compares the instantaneous event rates between groups among participants who have not yet experienced the event. A value near 1 indicates that the fitted model estimates similar event rates over time; values on either side of 1 indicate different estimated rates in opposite directions, depending on which group is the reference. The reported hazard ratio was 0.99, with a 95% confidence interval from 0.94 to 1.04. The registry record does not clarify the reference coding or provide separate pairwise hazard ratios, so the result should be read as the reported overall comparison rather than as a detailed estimate for each pair of groups.

In general, a 95% confidence interval communicates the precision of an estimate under the analysis model and its assumptions. It gives a range of values compatible with the observed data and method; it is not a guarantee that the true value lies within that particular interval. For the primary analysis, the interval includes 1, which is the value corresponding to equal modeled hazards. The reported p-value, 0.76, is also consistent with the absence of clear statistical evidence against the null comparison. Neither quantity proves that there is no difference.

Why use the log-rank test?

In general, the log-rank test compares time-to-event experience across groups while accounting for the ordering of event times and censoring. It provides a test of group differences over the observed period. The registry reports a log-rank method and a Cox proportional hazards model, but does not describe how censoring was handled, whether the test was stratified, or whether the proportional hazards assumption was assessed. These details are not reported in the registry record.

How should the cumulative incidence estimates be read?

The primary outcome values were 76.7% for Patient-Provider, 77.9% for Provider Only, and 77.5% for Usual Care. Their respective confidence intervals were 75.45 to 78.0, 76.5 to 79.1, and 76.2 to 78.8. In general, each interval expresses uncertainty around that group’s estimated cumulative incidence. Overlap or non-overlap of separate group intervals is not a substitute for the reported group-comparison test, particularly for time-to-event data. The registry’s overall test and model result are the direct reported evidence for the primary comparison.

What do the secondary-outcome calculations add?

For selected binary secondary outcomes, the posted results also permit simple, unadjusted comparisons of proportions. A risk difference expresses the difference between two proportions; its confidence interval describes uncertainty around that contrast. A chi-squared test can assess whether the distribution of a binary outcome differs across groups. These calculations are distinct from the sponsor’s reported overall chi-squared tests and do not adjust for other participant characteristics. General statistical interpretation should therefore distinguish the recalculated pairwise estimates from the sponsor’s reported analysis.

For Dose Reduction, the reported overall chi-squared p-value was 0.33. For Percentage of Participants With Polypharmacy, it was 0.55, and for Percentage of Participants With Ambulatory Encounters, it was 0.47. The recalculated pairwise comparisons are unadjusted; the registry record does not report that they were prespecified or adjusted for multiple testing. They are exploratory context, not replacements for the registered overall analyses.

What are the planning illustrations?

Sample-size illustrations based on observed effects are planning exercises, not observed power and not the trial’s original design justification. The stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. For binary outcomes, the illustrations give participants per arm for 80% power and participants per arm for 90% power, at 100%, 75%, and 50% of the observed effect. For example, the Patient-Provider versus Usual Care Dose Reduction illustration gives 30,818 participants per arm for 80% power at 100% of the observed effect. Such figures depend on the observed effect and stated assumptions; they do not establish what the study was designed to detect. The registry record does not report the trial’s own sample-size justification.

In general, over-dispersion matters when the variability in data is greater than a model assumes. Ignoring it can make uncertainty estimates too narrow and tests too optimistic. This primary endpoint is analyzed as time-to-event data, and the registry record does not report an over-dispersion assessment or a corresponding adjustment. That general issue should not be mistaken for evidence that over-dispersion occurred in this trial.

10. Analyses Reported by the Sponsor

The primary result reported by the sponsor is the log-rank and Cox model comparison: hazard ratio 0.99, 95% confidence interval 0.94 to 1.04, and p = 0.76. This is the relevant reported analysis for the primary time-to-dispensing endpoint. No competing recalculated primary estimate is reported in the registry record, and the record does not flag a discrepancy.

For the secondary outcomes with posted numerical data, recalculated pairwise comparisons are simple, unadjusted analyses of the reported summary statistics. They do not replace the sponsor’s overall chi-squared tests. For example, the sponsor-reported overall p-values are 0.33 for Dose Reduction, 0.55 for Polypharmacy, and 0.47 for Ambulatory Encounters. The recalculated pairwise results address narrower contrasts and do not account for any broader analysis plan or multiplicity adjustment.

These analyses can differ because they answer different statistical questions and may use different comparison structures. The registry record does not report adjusted sponsor estimates for the secondary pairwise contrasts. The recalculations are therefore best treated as supplementary, unadjusted descriptions rather than as corrections to the sponsor’s reported analysis. The primary interpretation remains anchored to the sponsor’s reported time-to-event analysis.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary outcome and multiple secondary outcomes. When several outcomes or group contrasts are tested, the chance of finding at least one apparently unusual result by chance can increase. In general, a multiplicity strategy may control that risk by setting out which tests are confirmatory, how comparisons are prioritized, or how significance thresholds are adjusted. The registry record does not report a multiplicity plan, an outcome hierarchy beyond identifying the primary endpoint, or an adjustment for the secondary analyses.

This matters especially for the recalculated pairwise comparisons, which are unadjusted and examine individual intervention groups against Usual Care. The sponsor-reported secondary tests are overall chi-squared tests, but the registry record does not say whether these formed part of a prespecified testing sequence or whether multiplicity was otherwise addressed. Their p-values should therefore be interpreted as reported, without assuming they carry confirmatory status.

The primary analysis population is described as a modified intention-to-treat cohort. The registry record does not define the modifications, explain how participants were assigned to that cohort, or describe handling of missing data. It also does not report whether participants were analyzed according to randomized group, how censoring was handled, or what methods were used for any missing outcome information. These omissions prevent a full assessment of whether exclusions or incomplete follow-up could have affected the estimates.

Denominators reported for some secondary measures differ from the enrollment counts listed for the groups. The record does not explain why those analysis denominators differ, so they should not be interpreted as a specific pattern of missingness or exclusion. The number and reasons for missing observations, and any sensitivity analyses, are not reported in the registry record.

The absence of these details is a reporting limitation, not evidence that a particular bias occurred. It means that the reader cannot fully evaluate the analysis populations, missing-data assumptions, or the degree to which the collection of secondary findings was protected against false-positive interpretation.

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.

The adverse-event record lists zero serious adverse events and zero other adverse events in each of the three groups. It also records deaths of 51 in Patient-Provider, 59 in Provider Only, and 51 in Usual Care. The listed numbers at risk were 4286, 4230, and 4271, respectively. No serious-event terms are listed.

These are registry counts, not a complete account of the nature, timing, severity, or attribution of events. The safety observation window and definitions used for these entries are not reported in the registry record. Zero recorded events should not be interpreted as proof that no events occurred, or that the intervention caused no harm; it describes what the registry reports in these categories. The death counts alone do not establish a treatment effect or explain causes of death. The record does not report a comparative safety analysis, so no adjusted safety conclusion can be drawn from these figures.

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

The primary result means that the reported time-to-dispensing analysis did not establish a clear difference among Patient-Provider, Provider Only, and Usual Care. The hazard ratio was 0.99, its 95% confidence interval ranged from 0.94 to 1.04, and the reported p-value was 0.76. The cumulative incidence estimates were also close in magnitude. Together, these results describe the data for the specified outcome and observation period, under the analysis reported in the registry.

This is not proof that the groups were identical. A non-significant test can occur when any true difference is small, when uncertainty remains, or when the available data do not distinguish among plausible effects. The confidence interval communicates that uncertainty for the modeled hazard ratio, but the registry record does not state an equivalence margin or provide an equivalence analysis. The result therefore does not demonstrate equivalence.

The result also does not mean that the educational materials had no effect on every aspect of prescribing, medication use, or participant health. The endpoint measures dispensing of targeted inappropriate prescription classes during a defined period. It does not directly measure medication-related morbidity, health-related quality of life, function, or whether an individual prescription was clinically appropriate. Those broader outcomes cannot be concluded from this endpoint alone.

Likewise, the primary result does not establish that the study intervention is beneficial or harmful in other populations or health-care settings. The registry record does not report detailed eligibility criteria, implementation fidelity, or full characteristics of the analyzed cohort. The interpretation should remain tied to the enrolled population and the trial’s reported outcome definition, while recognizing that those details are incomplete.

The most defensible reading is neither that an effect has been ruled out nor that a meaningful effect has been established. The reported analysis found no clear statistical separation in the primary outcome, with uncertainty that remains visible in the confidence interval. Decisions about clinical significance would require prespecified thresholds and additional evidence, neither of which is reported in the registry record.

14. Why This Trial Matters Statistically

This trial is statistically informative because its primary question concerns a time-to-event outcome rather than only a single end-of-period proportion. The timing of dispensing can matter: two groups could have similar cumulative incidence while differing in when events occur. The registry reports both group cumulative incidence estimates and a log-rank/Cox comparison, allowing the reader to distinguish event occurrence over the window from the modeled relative event rate over time.

The three-arm structure also separates two delivery approaches. One group received educational materials for both patients and providers, while another directed those materials to providers only; usual care served as the comparator. That design makes the contrast relevant to how educational interventions might be delivered, although the reported primary analysis is an overall three-group comparison and does not provide detailed pairwise primary estimates. The registry record does not report the allocation ratio or provide further design details needed to assess all planned contrasts.

The close primary cumulative incidence estimates and hazard ratio near 1 make precision and uncertainty central to interpretation. A non-significant p-value alone is not enough to claim that approaches are equivalent. The confidence interval and absence of a reported equivalence margin matter because they define what the analysis can and cannot establish.

The trial also illustrates the difference between a primary endpoint and broader aims such as medication safety and morbidity. A result about dispensing is meaningful for the defined prescribing measure, but it cannot stand in for every patient-centered outcome. Separating those questions prevents overinterpretation of statistical findings and keeps conclusions aligned with what was measured.

Finally, the registry record’s incomplete reporting of multiplicity, missing data, and analysis populations shows why statistical interpretation depends on both estimates and methodological transparency. Those omissions do not negate the reported analysis, but they limit how fully its robustness can be evaluated.

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.

The registry record leaves important details about the primary endpoint and analysis unspecified. It does not identify the targeted medication classes in the endpoint description, explain why the observation window begins at day 91, or report how dispensing data were captured and validated. These details would help readers understand the scope and potential measurement limits of the outcome.

The primary cohort is called modified intention to treat, but the modifications are not described. The record does not report the missing-data strategy, censoring details, or assessment of the Cox proportional hazards assumption. Nor does it describe randomization procedures, allocation concealment, or whether the analysis followed a prespecified statistical plan. These omissions restrict evaluation of potential selection, information, or modeling concerns.

The trial was unmasked. The registry does not describe whether this could affect intervention delivery, outcome ascertainment, or other aspects of study conduct. The primary outcome is dispensing, and the record does not give enough detail to assess the extent to which its recording depended on processes that might be affected by knowledge of group assignment.

The intervention and comparator are also incompletely described. The educational materials are characterized in broad terms, and usual care has no description in the arm record. Details on intervention fidelity, exposure, and adherence are not reported. Consequently, the result cannot show whether a particular implementation of the educational approach would have a different effect.

Finally, the registry does not report a multiplicity strategy or full numerical results for all listed secondary outcomes. The secondary recalculations are unadjusted, and their interpretation is more limited than that of the primary analysis. These reporting gaps should be treated as uncertainty about how comprehensively the findings can be assessed, rather than as evidence that a specific bias occurred.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: A Program to Reduce Inappropriate Medications Among Older Adults With Alzheimer's Disease — NCT05147428.
• Registry record last updated 2024-11-27; status last verified 2024-11.
• Results first posted 2024-11-27.
• 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

NCT05147428, D-PRESCRIBE-AD, was a completed randomized, unmasked, parallel-group trial with a registry enrollment of 14442. It compared educational materials for patients and providers, materials for providers only, and usual care. The primary endpoint was time to dispensing of a targeted inappropriate medication prescription from day 91 to day 270 during the 6 months after intervention receipt, analyzed in a modified intention-to-treat cohort.

The reported primary analysis used a log-rank test and Cox proportional hazards model. The hazard ratio was 0.99 (95% CI 0.94 to 1.04; p = 0.76), with cumulative incidence estimates of 76.7%, 77.9%, and 77.5% for Patient-Provider, Provider Only, and Usual Care. The registry results do not establish a clear difference among groups for this primary endpoint. Important details about missing data, multiplicity, cohort definition, and model assessment are not reported in the registry record.