This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for An Electronic Clinical Decision Support Tool to Reduce Low-value Antipsychotic Prescriptions (NCT04851691), a randomized na study. The primary outcome is Cumulative Total of New Antipsychotic Pill-days Prescribed.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT04851691 |
| Phase | NA |
| Condition | Alzheimer Disease, Dementia Alzheimers, Dementia of Alzheimer Type |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Single |
| Enrollment | 158 |
| Arms in the primary outcome | n/a |
| Lead sponsor | University of California, Los Angeles |
| Results posted | Yes |
2. Clinical Question
Population
Patients with Alzheimer’s disease and related dementias who received a new antipsychotic prescription from an eligible provider. The record describes older adults and identifies patients with an encounter in which a physician attempts to order a new prescription.
Intervention
A multi-pronged clinical decision support intervention embedded in the electronic health record, assigned to physicians.
Comparator
Control clinicians performed their duties as usual and did not receive the intervention.
Primary question
How did the cumulative total of new antipsychotic prescription days per eligible patient during the 12 months after rollout compare with the prior 12 months, in the randomized study of the intervention?
The intended question is whether an electronic decision support approach changes prescribing measured as days of new antipsychotic prescriptions. This is a prescribing-process outcome, not a direct measure of symptoms, quality of life, or whether a particular prescription was appropriate for an individual patient. The registry describes the study as targeting physicians newly ordering antipsychotic medication for older adults with Alzheimer’s disease and related dementias. It also states that patients were included when they had an encounter at which a physician attempted to place a new antipsychotic prescription and the physician had been assigned to one of the study groups.
The design makes the clinician’s assignment central to the comparison, while the primary outcome is described at the patient level. The record describes the outcome as prescription days per eligible patient across two 12-month periods, but does not explain the precise statistical contrast used to combine the periods and groups. It does not report a numerical estimate of the intervention’s effect, an uncertainty interval, or a test result. Consequently, the registry record identifies the question but does not establish the direction, size, or precision of any difference in prescribing.
3. Trial Design
This was a randomized, parallel-group study with an enrollment of 158. It is listed as completed. The study dates in the registry record are 2021-08-03 for the start and 2023-04-11 for primary completion. The intervention was an electronic health record clinical decision support tool; the control group received usual care. The registry lists the study phase as NA.
The assignment was made to physicians, according to the arm descriptions: physicians were assigned to the clinical decision support tool or to control. Patients were eligible in the described setting when a physician attempted to place an order for a new antipsychotic prescription. The intervention group’s patients were associated with a physician assigned to the tool, while control patients were associated with a physician assigned to control. This clinician-based assignment matters when interpreting patient-level outcomes because patients treated by the same clinician may have more similar prescribing patterns than patients treated by different clinicians.
The registry lists masking as single, but does not identify which people were masked or describe how masking was implemented. It does not report the number of physicians assigned to each group, the number of patients analyzed for the primary endpoint, or whether any participants or clinicians changed groups or were excluded after assignment. The detailed components of the multi-pronged tool are not reported in the registry record. The record also does not describe usual-care prescribing practices in further detail.
The planned primary observation period is described as 12 months after the intervention rollout date, compared with the prior 12 months. The record does not explain how the rollout date was defined for each clinician, how the pre- and post-rollout periods were linked to control clinicians, or whether the endpoint was analyzed as a between-group contrast, a within-group change, or a combination of these. Although the design is randomized and the allocation is described as clinician-based, the registry record does not supply a detailed protocol or a full account of the analysis implementation.
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: Patients were enrolled into the study when an eligible provider initiated a new antipsychotic prescription (e.g., Quetiapine, Olanzapine, Risperidone, Aripiprazole, Haloperidol, Clozapine) during a patient visit. Eligible providers were assigned an arm of the study based on their 2019 new antipsychotic orders for this population. Patients were enrolled into each arm based on the providers arm assignment. Healthcare Providers were not enrolled in the study.
Overall Study
| Stage | EHR CDS Tool | Control |
|---|---|---|
| STARTED | 67 | 91 |
| COMPLETED | 67 | 91 |
| NOT COMPLETED | 0 | 0 |
Analysis population for the primary outcome, as defined in the registry: Patients that received a new antipsychotic prescription (e.g., Quetiapine, Olanzapine, Risperidone, Aripiprazole, Haloperidol, Clozapine) from an eligible provider.
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: Providers were not assessed for baseline
| Characteristic | Unit | EHR CDS Tool | Control | Total |
|---|---|---|---|---|
| Participants analysed | 67 | 91 | 158 | |
| Age, Continuous | years | 81.19 (SD 15.14) | 84.78 (SD 11.77) | 83.26 (SD 11.73) |
| Sex: Female, Male: Female | Participants | 41 | 65 | 106 |
| Sex: Female, Male: Male | Participants | 26 | 26 | 52 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 52 | 70 | 122 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 6 | 10 | 16 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 9 | 11 | 20 |
| Race/Ethnicity, Customized: Race Category – Asian | Participants | 5 | 7 | 12 |
| Race/Ethnicity, Customized: Race Category – Other Race | Participants | 3 | 7 | 10 |
| Race/Ethnicity, Customized: Race Category – White | Participants | 45 | 63 | 108 |
| Race/Ethnicity, Customized: Race Category – Unknown or Not Reported | Participants | 8 | 11 | 19 |
| Race/Ethnicity, Customized: Race Category – Black or African American | Participants | 5 | 3 | 8 |
| Race/Ethnicity, Customized: Race Category – American Indian or Alaska Native | Participants | 1 | 0 | 1 |
6. Primary Endpoint
The registry lists the primary outcome as Cumulative Total of New Antipsychotic Pill-days Prescribed, measured over 12 month time point.
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | not applicable |
| Unit | New antipsychotic prescription days |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint is the cumulative total of new antipsychotic prescription days, expressed per eligible patient. It is described as prescription days from eligible clinicians during the 12 months after intervention rollout compared with the prior 12 months. The population for this endpoint consists of patients who received a new antipsychotic prescription from an eligible provider. Examples of medicines in the record include Quetiapine, Olanzapine, Risperidone, Aripiprazole, Haloperidol, and Clozapine.
Prescription days measure duration of prescribing rather than simply whether a prescription was initiated. A cumulative count can therefore reflect both how many new prescriptions were issued and how many days those prescriptions covered. The endpoint is relevant to the study’s stated quality improvement aim because the intervention was intended to influence prescribing behavior. However, the registry record does not report the numerical endpoint results, the exact formula for calculating prescription days, or the rules used to combine multiple prescriptions for a patient.
In general, an endpoint should be interpreted according to its definition and population. This measure describes prescribing, not medication use, adherence, clinical appropriateness, or patient benefit or harm. Those are distinct outcomes and cannot be inferred from prescription-day counts alone. The record lists five secondary outcomes concerning handouts, emergency visits, hospitalization, and death within 90 days after enrollment, but it contains no posted analysis results for those outcomes either.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record describes a randomized evaluation of a multi-pronged electronic health record decision support intervention, with usual care as the comparator. Its primary endpoint concerns cumulative new antipsychotic prescription days per eligible patient over a 12-month period after rollout, in comparison with the prior 12 months. This defines the intended outcome and time frame, but the record does not include the numerical primary outcome findings. It therefore does not show whether prescribing days were lower, higher, or similar between the study groups.
No effect estimate, confidence interval, p-value, or other primary statistical result is reported in the registry record. Without an effect estimate, the magnitude of any difference cannot be assessed. Without an uncertainty interval, its precision cannot be assessed. A significance test, if one had been reported, would address compatibility with a specified null model; it would not by itself convey the size or practical importance of an effect. Since no such result appears in the record, neither statistical significance nor a conclusion about the evidence against a null hypothesis can be drawn from this registry entry.
The design also affects how an estimate would need to be interpreted. Physicians, rather than individual patients, are described as assigned to intervention or control. Patient outcomes within a clinician’s practice may be correlated because clinicians share prescribing habits and care processes. An analysis that treats every patient as entirely independent can understate uncertainty when that correlation is present. The registry’s design note specifically cautions that a simple individual-level re-analysis would be misleading and indicates that the sponsor’s own analysis is the appropriate analysis to consult. However, the record lists no analyses, and no sponsor effect estimate or uncertainty measure is included.
The primary endpoint description refers to the period after rollout and the prior 12 months. A before-and-after comparison can be affected by changes over time as well as by an intervention. Randomization can support a comparative interpretation, but the record does not give enough detail about the endpoint calculation or statistical contrast to determine how time periods and randomized groups were handled. The results cannot be reconstructed from the enrollment count or the safety counts.
The appropriate conclusion is limited: this registry entry documents the study’s question, design features, and planned primary outcome, but does not report the primary efficacy result. The listed adverse event figures are not substitutes for the missing prescribing outcome, and they do not establish a treatment effect. No claim about whether the tool changed prescribing or improved clinical outcomes is supported by the results recorded in this entry.
9. Statistical Methods Explained
The registry describes a randomized parallel-group design in which physicians were assigned to the electronic health record clinical decision support tool or to control. The primary outcome is defined for eligible patients associated with those physicians. That structure creates a distinction between the unit of assignment and the unit of outcome: clinicians are assigned, while prescription days are counted for patients. This distinction is important for statistical analysis because patients cared for by the same clinician may share prescribing patterns. The registry design note cautions against a simple individual-level re-analysis and states that only the sponsor’s own analysis should be used. No sponsor analysis estimates are recorded in the registry entry.
What does the endpoint count?
The endpoint is a cumulative number of new antipsychotic prescription days per eligible patient. It is not described as a mean, a rate, or a percentage. A count outcome records how many units of an event or activity occur in a defined population and period. In general, count data are often nonnegative and can be unevenly distributed: many people may have few or no counted units while a smaller number have larger counts. Whether that pattern occurred in this study is not reported. The registry record does not state the distribution of individual patient counts or the rules for handling multiple prescriptions.
In general, a count model may use a Poisson distribution when its assumptions are reasonable. A common concern is over-dispersion, in which the observed variability is greater than the model expects. If over-dispersion is present and ignored, standard errors and confidence intervals can be too small, making results appear more precise than warranted. Other approaches can accommodate extra variability or clustering. These are general methodological considerations, not a description of the method used in this trial: the registry record does not report the model, link function, variance assumptions, or adjustment variables for the primary endpoint.
Why does clinician assignment matter?
When an intervention is assigned to clinicians, outcomes for patients within a clinician’s practice may be correlated. In general, this within-clinician correlation reduces the amount of independent information compared with a design in which every patient’s assignment is independent. Analyses of cluster-assigned studies can account for this structure using methods such as mixed-effects models, generalized estimating equations, or cluster-level summaries. Those are examples of general approaches only; the registry record does not identify which approach the sponsor used, nor does it give the number of clinician clusters or a measure of within-clinician correlation.
The registry explicitly warns that an individual-level recalculation would be misleading and says the sponsor’s own analysis takes precedence. This warning means that unadjusted computations from pooled patient-level summary data should not be treated as a substitute for a cluster-aware sponsor analysis. No discrepancy between an independent recalculation and sponsor result can be evaluated because no primary outcome estimate or sponsor analysis result is recorded.
What does the time comparison require?
The endpoint description compares prescription days in the 12 months after rollout with the prior 12 months. In general, repeated periods can be analyzed in several ways: investigators may compare change over time between randomized groups, model period-specific outcomes, or define another contrast. Each approach answers a somewhat different statistical question and depends on the available data and assumptions. The record does not specify which contrast was used, whether baseline prescribing was adjusted for, how dates were aligned across clinicians, or how any incomplete observation periods were handled.
A confidence interval, in general, gives a range of effect values compatible with the data and the statistical model at a stated confidence level. Its width reflects precision: wider intervals indicate greater uncertainty. A p-value, in general, describes how unusual data at least as extreme as those observed would be under a specified null model and its assumptions. Neither measure is present for the primary endpoint in the registry record. No conclusions about precision or hypothesis testing can therefore be drawn.
What do the planning assumptions mean?
The registry includes a planning illustration described as using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It says the effect to detect is a stated fraction of the observed effect. The record does not include the calculated planning values in the available results, so no participant requirement can be stated from it. Such calculations are illustrative planning scenarios, not observed trial results and not evidence that a study had a particular probability of detecting an effect after completion. In this cluster-assigned study, an illustration based on individual-level assumptions would also need to be interpreted cautiously if it does not account for clinician-level clustering.
The registry lists zero analysis entries. It does not report the primary endpoint’s statistical model, covariates, handling of repeated observations, missing-data approach, or analysis population. These omissions prevent a full appraisal of the analysis strategy and mean that general descriptions of count-data or cluster methods should not be mistaken for methods confirmed to have been used in this study.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists six outcome measures: one primary outcome and five secondary outcomes. The secondary outcomes concern the number of patients who receive a handout, emergency visits, hospitalization, hospitalizations including psychiatric hospitalizations, and death within 90 days after enrollment. Listing multiple outcomes raises a general statistical issue: when many hypotheses are tested, the chance of at least one apparently unusual result can increase unless the analysis plan accounts for multiple testing. The registry record does not report whether any multiplicity adjustment, hierarchy, or prespecified testing sequence was used.
The record lists zero analyses and does not give results for the primary or secondary outcomes. It does not report whether outcome data were missing, how missingness was defined, whether participants were excluded from particular analyses, or how missing observations were handled. Missing-data methods depend on why data are missing and on the analysis question; no particular mechanism or method can be inferred from the registry entry.
The analysis populations are also not described. The record does not state whether the primary analysis included all randomized clinicians or patients, only patients meeting the endpoint definition, or another population. It does not explain how patients were linked to assigned clinicians if their care involved more than one provider. These details matter for interpreting a randomized comparison and for assessing whether the analysis followed the assignment structure.
For a clinician-assigned study, multiplicity, missing data, and population definitions should be evaluated alongside clustering and the specified time contrast. Since these details are not reported in the registry record, uncertainty remains about the analysis procedures as well as the absent outcome estimates. No adjustment method or missing-data strategy should be assumed.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| EHR CDS Tool | 2 | 67 | 3.0% |
| Control | 12 | 91 | 13.2% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| EHR CDS Tool | 1 of 67 |
| Control | 2 of 91 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| EHR CDS Tool | 4 of 67 |
| Control | 6 of 91 |
Most frequent serious adverse event terms
| Event term | Organ system | EHR CDS Tool | Control |
|---|---|---|---|
| Inpatient Admission | General disorders | 2 of 67 | 12 of 91 |
Most frequent non-serious adverse event terms
| Event term | Organ system | EHR CDS Tool | Control |
|---|---|---|---|
| Emergency Department visits that did not require admissions | General disorders | 4 of 67 | 6 of 91 |
Serious events with the highest arm-level rate: inpatient admission (13.2%).
The registry reports serious adverse events and other adverse events separately for the EHR CDS Tool and Control groups. In the EHR CDS Tool group, 2 of 67 people at risk were affected by serious adverse events and 4 of 67 were affected by other adverse events; 1 death is listed. In Control, 12 of 91 people at risk were affected by serious adverse events and 6 of 91 by other adverse events; 2 deaths are listed. The record identifies Inpatient Admission as the top serious adverse event term, at 13.2%.
These counts describe events recorded in the registry, not a statistical comparison of safety between groups. The registry does not report uncertainty intervals, a comparative test, or details about the relationship between events and the intervention. It also does not clarify in the posted summary whether people could contribute more than one event, how overlapping event categories were handled, or the denominator used for the 13.2% term. The term’s percentage should therefore be read as a registry-reported figure, not independently interpreted as a group-specific risk.
Event counts can be affected by follow-up duration, ascertainment, and the population included. Those details are not reported sufficiently to make a comparative safety inference. The figures do not establish that the intervention caused or prevented any event, and they do not replace the absent primary prescribing analysis.
12. What the Result Does, and Does Not, Mean
The registry record gives a clear description of the study’s objective: evaluate an electronic health record clinical decision support tool intended to reduce low-value new antipsychotic prescriptions among older adults with Alzheimer’s disease and related dementias. It identifies the primary endpoint as cumulative new prescription days per eligible patient, comparing a 12-month period after rollout with the prior 12 months. It also records randomized parallel allocation, clinician-assigned study groups, and an enrollment of 158.
Those design details explain what the investigators set out to test, but they are not outcome evidence. The record contains no numerical primary endpoint result and no statistical analysis entries. It therefore cannot show whether prescription days changed, how large a change might have been, or how precisely an effect was estimated. The absence of a result in the registry entry is not evidence that there was no effect; it means the entry does not report an effect estimate that can be evaluated.
The study’s endpoint is a prescribing measure. In general, prescription days can describe the amount of prescribing over time, but do not by themselves establish whether patients took the medicine, whether the prescription was appropriate, or what happened to patients’ symptoms and health. The listed secondary outcomes include emergency visits, hospitalizations, handouts, and death within 90 days, but their results are also not reported in the registry entry.
Any future interpretation of a reported estimate should account for assignment by clinician and for the endpoint’s pre- and post-rollout periods. The record itself warns that a simple individual-level re-analysis would be misleading and identifies the sponsor’s own analysis as the relevant source. Since that analysis is not recorded here, the appropriate reading is descriptive and limited: the trial question and selected safety counts are documented, while the primary result and the analysis details needed to judge it are unavailable in the registry record.
No clinical recommendation follows from the information recorded. The study title and hypothesis state an intent to reduce prescribing, but an aim or hypothesis is not a finding. The available record cannot support a claim that the decision support tool did or did not achieve that aim.
13. Why This Trial Matters Statistically
This study raises a central issue in pragmatic clinical research: an intervention can be directed at clinicians while outcomes are counted for patients. That design can align with how a clinical decision support tool is implemented, but it also means that statistical analysis must respect the clinician-level assignment. Treating all patient outcomes as independent may overstate the information in the data when patients of the same clinician share care patterns. The registry’s warning against a simple individual-level re-analysis makes this design feature especially important to interpretation.
The endpoint also illustrates the difference between measuring a prescribing process and measuring a patient outcome. Prescription days can capture the duration of new prescribing across a defined period, which fits the stated quality improvement goal. But a process measure does not, by itself, resolve questions about medication use, appropriateness, or health outcomes. The registry lists some patient-level secondary outcomes, yet does not report their results.
Statistically, the contrast involves a period after rollout and a prior period, as well as randomized groups. Readers need to know how those elements were combined to understand the estimand—the precise effect the analysis aims to estimate. The registry record does not specify that contrast or the model. Without those details and without an estimate and uncertainty interval, a reader cannot judge the magnitude or precision of the primary result.
The entry is therefore useful for identifying the intended question and highlighting design features that affect analysis, but it is incomplete as a report of trial findings. It demonstrates why randomization alone does not make every re-analysis appropriate: assignment level, outcome level, time structure, and analysis method must be considered together. The absence of a posted primary analysis limits conclusions regardless of the trial’s completed status.
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 principal limitation for interpreting this registry entry is that it reports no primary outcome estimate or statistical analysis results. Although the study is listed as completed, the record provides no effect size, uncertainty interval, or hypothesis test for cumulative prescription days. No efficacy conclusion can be drawn from the recorded information.
Important design and analysis details are not reported in the registry record. The record does not identify which people were masked, describe the components of the multi-pronged decision support tool, state the number of randomized physicians, or explain how the intervention rollout dates and two 12-month periods were operationalized. It does not define the exact endpoint calculation or the statistical contrast used to compare groups and periods.
The registry also does not report the primary analysis model, any adjustment variables, how clustering by clinician was handled, the analysis population, missing-data methods, or multiplicity procedures. These are especially relevant because clinicians were assigned while outcomes were described for patients, and because the record explicitly cautions against a simple individual-level re-analysis. No sponsor result is available in the entry to resolve these issues.
The listed safety figures have limited context. The record does not describe event ascertainment, exposure time, potential overlap among events, or the denominator underlying the reported 13.2% for Inpatient Admission. They should not be treated as comparative safety estimates. The enrollment count alone does not establish how many patients contributed to the primary analysis or whether the study had adequate precision for that endpoint. The registry’s planning illustration assumptions do not fill these reporting gaps.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: An Electronic Clinical Decision Support Tool to Reduce Low-value Antipsychotic Prescriptions — NCT04851691.
- Registry record last updated 2024-09-19; status last verified 2024-08.
- Results first posted 2024-09-19.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
The registry record describes a completed randomized, parallel-group study of an electronic health record clinical decision support tool versus usual care. Physicians were assigned to the intervention or control, and the enrollment was 158. The study focused on older adults with Alzheimer’s disease and related dementias who received a new antipsychotic prescription from an eligible provider.
The primary endpoint was cumulative new antipsychotic prescription days per eligible patient, with a 12-month post-rollout period compared with the prior 12 months. The record lists five secondary outcomes concerning handouts, emergency visits, hospitalization, and death within 90 days after enrollment.
No primary or secondary outcome analyses are recorded. The registry cautions that a simple individual-level re-analysis would be misleading because of the study design and points to the sponsor’s own analysis. That analysis and its estimates are not included in the record. Safety counts are listed for both groups, but no comparative safety analysis is reported.