Cognitive Stimulation Therapy for Dementia: A Two-Armed Pragmatic Trial: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Cognitive Stimulation Therapy for Dementia: A Two-Armed Pragmatic Trial (NCT05860127), a randomized na study. The primary outcome is Cognitive Impairment.

2. Clinical Question

3. Trial Design

This was a randomized, two-arm, parallel-group pragmatic trial, with no masking. The title identifies it as a two-armed pragmatic trial, and the study description calls it an embedded pragmatic clinical trial. The sponsor was the University of Massachusetts, Amherst. The registry reports enrollment of 133, a start date of 2023-04-10, primary completion on 2025-06-12, and a completed status.

Participants with mild to moderate dementia identified through standard care were allocated to referral for CST or standard of care. In the referral group, a physician referred participants for CST treatment; in the comparison group, care continued without a CST referral. The study description specifies that people who accepted the referral would participate in V-CST, an online delivery of CST. Thus, random assignment was to a referral strategy, not necessarily to completed therapy. Acceptance and completion were separately identified as secondary outcomes, as was non-completion.

The primary outcome was a cognitive score measured at 12 months. The record describes assessment using one of three instruments—MoCA, SLUMS, or MMSE—and conversion to a MoCA score using a crosswalk. The record lists five outcome measures and two analyses, but details of the additional measures and analyses are not reported in the registry record. The phase is listed as NA. Specific details about sites, allocation concealment, randomization procedures, and the intended analysis population 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: 133 participants were randomly assigned to the intervention or control group. Participants were enrolled at the time of randomization.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Participants were randomized to intervention or standard of care group.

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: This protocol utilized a randomization stratified by site by a random number sequence in a 1:2 ratio.

6. Primary Endpoint

The registry lists the primary outcome as Cognitive Impairment, measured over 12 months.

The primary endpoint, titled Cognitive Impairment, is a continuous score on a scale assessed at 12 months. The registry says that participants were assessed with one of three cognitive instruments: the Montreal Cognitive Assessment (MoCA), the Saint Louis University Mental Status (SLUMS) exam, or the Mini-Mental State Examination (MMSE). A crosswalk converted the measures to a MoCA score so that scores from different assessment tools could be represented on a common scale.

The registry describes MoCA scores as ranging from 0-30. It says scores of 26+ are generally considered normal, while scores below that suggest cognitive issues; 18-25 is often mild impairment, 10-17 moderate impairment, and under 10 severe impairment. Participants started with baseline MoCA scores from 10 to 26, which the record characterizes as mild-moderate cognitive impairment. These ranges provide context for the scale, but do not establish that a particular change is clinically important.

A continuous score retains gradations in performance rather than reducing the outcome to a yes-or-no category. Comparing scores can therefore describe differences in average measured cognition. However, the registry does not report a threshold for meaningful change, the details or validation of the crosswalk, or whether the same participants used the same instrument at follow-up. Those details are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary measure is continuous, so its arm summaries are means and confidence intervals rather than counts of participants with an event. A mean describes the central value of a set of scores. The posted means are 19 for the referral group and 19 for standard care. Because the values are reported as rounded summaries, they should not be treated as proof that every participant's score was the same or that the underlying unrounded group means were identical.

A confidence interval expresses uncertainty about an estimated quantity under the statistical procedure used to construct it. The arm-specific 95% confidence intervals are 5.64 to 34.64 and 6.51 to 31.49. These are intervals around the group means, not ranges containing 95% of individual participants' scores. Their width indicates that the group summaries are imprecise. The registry gives no standard deviations or other direct description of score variability, and does not explain why these intervals extend beyond the MoCA range described for the scale.

The reported group comparison comes from a hierarchical regression model, not simply from subtracting the rounded arm means. Regression models can estimate a comparison while accounting for modeled structure or covariates, depending on their specification. “Hierarchical” can refer generally to a model that represents levels or groupings in data, but the registry does not identify which levels, grouping variables, covariates, link or distribution assumptions, or estimation procedure were used. Those are not reported in the registry record, so it is not possible to explain exactly what the fitted model adjusted for or how the reported estimate was obtained.

The model result is an effect size of -2.78 with a 95% confidence interval from -4.61 to -0.96, and a reported p-value of <0.05. The confidence interval represents uncertainty in the model's estimate under the model and its assumptions. The p-value is a measure of how incompatible the observed result would be with a specified null model, assuming that model and the analysis assumptions. It is not the probability that the null hypothesis is true, nor a measure of clinical importance. The registry does not define the effect-size scale or contrast order. Although zero is a common reference for a difference, the exact interpretation of zero for this reported effect is not specified.

The record labels the result “EQUIVALENCE.” Equivalence testing generally asks whether an effect is sufficiently close to zero within a prespecified acceptable margin. That margin, the null and alternative hypotheses, and the decision procedure are not reported in the registry record. The confidence interval and p-value alone cannot establish that an equivalence criterion was met. In particular, the reported confidence interval is on one side of zero, so the equivalence label merits caution rather than an assumption that the groups were equivalent.

The intervention is a referral strategy. In pragmatic trials, outcomes can reflect both assignment and what happens after assignment, such as whether a referral is accepted. Here, the registry separately names acceptance, CST completion, and non-completion as secondary outcomes. That separation matters: a comparison by randomized assignment addresses the effect of offering or making the referral, while a comparison based on actual participation would answer a different question and may no longer preserve the original randomized groups. The analysis population and whether the primary analysis followed assignment are not reported in the registry record.

General statistical considerations also apply to a crosswalk between cognitive instruments. A common scale can facilitate combined summaries, but the result depends on how the measures are mapped and how well the mapping represents the underlying construct. The crosswalk's method, validation, and handling of instrument-specific measurement properties are not reported in the registry record. The record also does not say whether measure choice differed by participant or site, or how such differences were handled in the model.

The planning illustration associated with the posted results uses a two-sided significance level of 0.05, equal allocation, a normal approximation, and a target effect defined as a stated fraction of the observed effect. These are illustrative planning assumptions based on the observed effect, not evidence about the trial's original design or a measure of its observed power. The registry does not report the trial's own sample-size justification, assumptions, or any interim-analysis plan. Planning sample sizes are hypothetical design calculations and should not be interpreted as the probability that the completed trial detected an effect.

10. Analyses Reported by the Sponsor

The sponsor-reported primary comparison is the hierarchical regression result: effect size -2.78, 95% confidence interval -4.61 to -0.96, and p <0.05. The registry also reports the rounded arm means and their confidence intervals. These are different forms of summary: the group means describe each arm, while the regression result is a model-based comparison. Since the model details and effect definition are not reported, the model estimate cannot be reconstructed from the arm summaries alone.

No discrepancy is flagged between the sponsor-reported result and the recalculated summaries. A simple recalculation from posted summary statistics is unadjusted and cannot reproduce a hierarchical regression analysis without the model specification and underlying data. It should not replace the sponsor's own analysis. Conversely, the sponsor's result cannot be interpreted beyond its reported scale and group coding, which are not reported in the registry record.

The registry labels the analysis as equivalence, but does not report an equivalence margin or decision procedure. The label alone is not enough to conclude equivalence. The numerical estimate, interval, and p-value should be retained as reported, with their interpretation limited by the missing model and contrast details.

11. Multiplicity, Missing Data and Analysis Populations

The registry lists five outcome measures and two analyses. It identifies the 12-month cognitive impairment score as the primary outcome and names three secondary outcomes: the percentage who accepted the referral, the percentage who completed CST, and the percentage who did not complete CST. The other outcome measures and the details of the additional analysis are not reported in the registry record.

When several outcomes or analyses are examined, the chance of obtaining at least one apparently notable result can increase. Statistical approaches to multiplicity can control or otherwise address that issue, but the suitable approach depends on the questions and analysis plan. Whether the trial prespecified a multiplicity strategy, adjusted significance thresholds, or designated a testing hierarchy is not reported in the registry record. The reported p-value should therefore be understood as the value attached to the posted analysis, not as evidence that multiple testing was accounted for.

The primary result describes a continuous score, but the number of participants with follow-up data, reasons for missing scores, and handling of missing observations are not reported in the registry record. The record does not describe imputation, complete-case analysis, sensitivity analyses, or how missing outcome data were handled in the hierarchical regression. Missingness can affect estimates when the availability of a score is related to participant characteristics or outcomes; the direction and extent of any such effect cannot be assessed from the posted information.

The registry reports outcome-group counts of 50 for referral and 83 for standard care. It does not state whether the primary analysis was intention-to-treat, per-protocol, or based on another analysis population. Nor does it report whether participants were analyzed according to random assignment, whether crossovers occurred, or how acceptance and therapy completion were incorporated. These omissions matter because the intervention was referral, and taking up the referral was a separate event.

Details of the randomization procedure, allocation concealment, model covariates, clustering structure, and prespecified analysis plan are not reported in the registry record. Without those details, it is not possible to assess fully the protection against bias, the model's assumptions, or whether the posted analysis followed a prespecified strategy.

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 5% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: respiratory illness (4.0%), cardiovascular exacerbation (2.0%), failure to thrive (2.0%), febrile illness/sepsis (2.0%), pulmonary embolism (2.0%).

The registry reports serious and other adverse-event counts separately for the two groups. Among 50 participants at risk in the referral group, 6 were reported as affected by serious adverse events and 3 by other adverse events; 3 deaths were reported. Among 83 participants at risk in standard care, 9 were reported as affected by serious adverse events and 13 by other adverse events; 3 deaths were reported. These are registry counts, not evidence by themselves that an event was caused by referral, CST, or standard care.

The listed serious adverse-event terms include Respiratory illness (4.0%), Cardiovascular Exacerbation (2.0%), Failure to thrive (2.0%), Febrile Illness/Sepsis (2.0%), Pulmonary embolism (2.0%), and Perinephric hematoma (1.2%). The registry does not specify in this summary how the listed percentages relate to the groups or whether people could contribute more than one event. They should not be reassigned to an arm or interpreted as comparative risks without further detail.

The record does not provide event timing, severity beyond the serious/other categories, relatedness assessments, or the method used to collect adverse events. Group counts also do not establish whether the exposure periods were comparable. Accordingly, these figures describe what was posted, but do not support a causal safety comparison on their own.

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

The central result is that the registry reports a model-based effect size of -2.78 for the 12-month cognitive outcome, with a 95% confidence interval from -4.61 to -0.96 and p <0.05. The arm summaries report a mean score of 19 in both groups, with wide confidence intervals. These pieces of information answer different descriptive and model-based questions; the registry does not provide enough analytical detail to reconcile them or to identify the direction of the modeled group contrast.

The result does not establish that the referral group had better or worse cognitive scores. The contrast order and effect-size definition are not reported in the registry record. The trial's objective refers to cognitive decline, but the posted primary outcome is a continuous converted score, and the record does not define a threshold for decline or report its frequency by group. A conclusion about how often decline occurred cannot be drawn from the posted effect estimate.

The equivalence label also does not establish that the groups were equivalent. An equivalence margin and the applicable decision rule are not reported. Likewise, a p-value below 0.05 is not a measure of clinical importance and does not indicate the probability that a treatment effect exists. The interval communicates uncertainty around the reported model estimate, but its practical meaning is constrained by missing information about the scale, coding, and model.

Assignment was to referral, while accepting and completing CST were separate outcomes. The primary result therefore should not be treated as a direct estimate of the effect of completing V-CST. The registry does not report the relevant participation counts or the primary analysis population in enough detail to distinguish these questions.

Overall, the posted information documents a completed randomized pragmatic trial and a sponsor-reported statistical comparison, but not all elements needed for a full independent interpretation. The result should be read as an estimate with uncertainty and important unresolved interpretation details, not as a standalone statement about treatment benefit, equivalence, or individual outcomes.

14. Why This Trial Matters Statistically

This trial raises an important statistical distinction between assigning a referral and receiving an intervention. The randomized contrast concerns referral for CST versus standard care without a CST referral. Because acceptance and completion were separate secondary outcomes, the effect of the referral strategy is not automatically the same as the effect of participating in V-CST. Keeping assignment, uptake, and completion distinct is central to interpreting pragmatic trials.

The study also combines cognitive measurements from MoCA, SLUMS, and MMSE by converting them to a MoCA score. A crosswalk can make a common continuous outcome possible when different instruments are used, but it also introduces a measurement step that deserves clear documentation. The registry does not report how the crosswalk was constructed or assessed. That omission limits the ability to judge how much the common scale reflects comparable measurement across participants.

The posted result illustrates why the statistical model and estimand need to be described alongside an effect estimate. Equal rounded arm means coexist with a nonzero regression estimate, but the registry does not report the contrast coding or model specification needed to explain the relationship. Clear reporting of the outcome definition, model, analysis population, missing-data handling, and interpretation of the estimate would make the result more reproducible and accessible.

Finally, the equivalence label demonstrates the importance of reporting a prespecified margin and decision rule when equivalence is the stated analytic aim. A model estimate and confidence interval alone do not tell readers whether a result met an equivalence criterion. The record's combination of a pragmatic referral design, multiple outcomes, and a crosswalked continuous measure makes transparent statistical reporting especially valuable.

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 main limitation for statistical interpretation is that the hierarchical regression model is not described beyond its name. The registry does not report the effect-size definition, group-order coding, covariates, clustering structure, or estimation assumptions. The posted estimate therefore cannot be translated confidently into the direction or exact meaning of the comparison.

The registry labels the analysis as equivalence but does not report an equivalence margin or decision procedure. It also does not explain why the arm-specific 95% confidence intervals extend beyond the stated MoCA score range of 0-30. The displayed values may be presented as reported, but the registry does not explain their construction or their relation to bounded scores.

The endpoint combines three possible assessment instruments through a crosswalk. Details of the conversion, its validation, measure assignment, and handling of measurement differences are not reported in the registry record. A threshold for meaningful change and a definition of cognitive decline are also not reported, so the model result cannot be translated into a clinically interpretable change or the stated frequency of decline.

Information about follow-up completion, missing outcome data, analysis populations, adherence, crossovers, and the handling of acceptance or completion is not reported. The record also does not state whether multiplicity was addressed across its five outcome measures and two analyses, or provide the full analysis plan and randomization details. These limitations restrict assessment of bias, precision, and reproducibility. They do not establish that the analysis was inappropriate; they identify details needed to evaluate it.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Cognitive Stimulation Therapy for Dementia: A Two-Armed Pragmatic Trial — NCT05860127.
• Registry record last updated 2026-03-23; status last verified 2026-03.
• Results first posted 2026-03-23.
• 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

NCT05860127 is a completed randomized, unmasked, parallel-group pragmatic trial sponsored by the University of Massachusetts, Amherst. It enrolled 133 participants and compared physician referral for CST, described as virtual CST in the study summary, with standard care and no CST referral. Assignment to referral was distinct from accepting or completing therapy.

The primary outcome was a continuous cognitive score at 12 months, assessed with MoCA, SLUMS, or MMSE and converted to a MoCA score. The posted arm means were 19 in both groups. The reported hierarchical regression estimate was -2.78, with a 95% confidence interval from -4.61 to -0.96 and p <0.05; the record labels the analysis as equivalence. The contrast coding, model details, and equivalence margin are not reported in the registry record, limiting interpretation of direction and equivalence.

Acceptance, completion, and non-completion were listed as secondary outcomes. Missing-data methods, analysis populations, multiplicity procedures, and details of the crosswalk are not reported in the registry record.