Virtual Patient Behavioral Response Training for Family Caregivers: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Virtual Patient Behavioral Response Training for Family Caregivers (NCT04489303), a randomized phase 2 study. The primary outcome is Revised Memory and Behavior Problems Checklist.

2. Clinical Question

3. Trial Design

The registry record describes a completed phase 2, randomized, parallel-group trial comparing two behavioral programs for family caregivers. One arm received Virtual Patient Behavioral Response Training, described as the Caring Response mobile app with behavioral training. The other received Educational Training, described as a traditional educational program. The masking field is none, so the registry does not report masking of participants or other trial roles. Further operational details about allocation procedures, concealment, training delivery, or adherence are not reported in the registry record.

Registry enrollment is listed as 156. For the primary outcome summaries, the reported group counts are 60 for Virtual Patient Behavioral Response Training and 55 for Educational Training. These are the counts attached to the posted outcome summaries; the record does not explain the relationship between these counts and registry enrollment or describe exclusions, missing observations, or the analysis population. They should not be treated as evidence about why some participants do not appear in the outcome summaries.

The primary outcome was the Revised Memory and Behavior Problems Checklist, assessed at 3 months. The measure is continuous, and the registry reports group means and standard deviations. The listed sponsor is Photozig, Inc. The record marks the study completed and gives its start and primary-completion dates, but those dates do not add detail about the statistical analysis. The registry record does not report the trial’s original sample-size assumptions, a detailed statistical analysis plan, or analysis entries. The one posted primary comparison can therefore be interpreted from the available summary statistics, while important design and analysis details remain unknown.

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.

Overall Study

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 Revised Memory and Behavior Problems Checklist, measured over 3 months.

The primary endpoint is caregiver-reported conditional bother on the Revised Memory and Behavior Problems Checklist at 3 months. The checklist contains 24 items describing possible troublesome behaviors during the past month. Caregivers first indicate whether each behavior occurred and then rate how much it bothered or upset them. The conditional bother score uses the bother ratings for behaviors that occurred, rather than treating a behavior that did not occur as a source of bother.

The scale runs from 0 to 4 units. A score of 0 is the minimum and best value, while 4 is the maximum and worst value; higher scores indicate worse outcomes. The endpoint is relevant to the stated caregiver focus because it captures the caregiver’s reported distress associated with behaviors, rather than only counting whether behaviors occurred. It is important to distinguish this conditional-bother score from a measure of the number of behaviors alone: the definition incorporates caregivers’ ratings of how upsetting the behaviors were.

The registry reports the outcome as a mean with a standard deviation for each arm. The primary outcome description does not provide a separate measure of change from baseline, and baseline outcome values are not reported in the posted primary summaries. The record also does not describe a minimally important difference or a threshold for interpreting a difference as meaningful to caregivers. Accordingly, the scale direction and the observed group summaries can be described, but the clinical importance of the estimated difference cannot be established from the record alone.

7. Results Posted in the Registry

• Educational Training: mean 0.89 versus 0.66 on control; difference 0.23 (95% CI -0.04 to 0.50), standardized effect d = 0.31; z-test p = 0.095.

8. Statistical Interpretation

9. Statistical Methods Explained

What does a mean difference represent?

A mean is the arithmetic average of the observed scores in a group. For a continuous measure such as the conditional-bother score, comparing group means gives a concise description of the difference in average scores. The registry’s contrast is educational training minus virtual training. A positive estimate therefore means the educational-training group’s reported mean is higher. Because higher values are worse on this scale, the direction of the point estimate corresponds to more average caregiver-reported bother in that group. The direction of a contrast must always be read together with its stated group order and the scale’s meaning.

A mean difference retains the original measurement units. Here, the estimate is 0.23 units on the 0-to-4 scale. It is not a percentage change, a ratio, or a count of caregivers. Nor does it say that each caregiver’s score differed by that amount: it compares group averages. Individual scores can vary around each group mean, as indicated by the reported standard deviations.

What do standard deviations describe?

A standard deviation describes how spread out observations are around their group mean. It is a feature of the observed scores, rather than a measure of how precisely the mean difference has been estimated. The reported standard deviations are 0.69 in virtual training and 0.78 in educational training. Similar-looking averages can coexist with substantial variation among caregivers. Inference about the mean difference also depends on the number of observed scores and the design and analysis choices, so the standard deviations should not be read as confidence limits.

When group variances differ, the method used to compare means can affect uncertainty estimates. The registry record does not report the test or variance assumption used by a trial analysis. The posted unadjusted comparison is based on summary statistics, and no model specification is included. More generally, a comparison method should reflect the outcome’s distribution, group structure, and missing-data situation. The record does not provide enough detail to assess all those choices.

How should the p-value be read?

The unadjusted p-value is 0.095. In general, a p-value describes how incompatible the observed data are with a specified null model of no difference, under the assumptions of the test used. It is not the probability that the null hypothesis is true, and it does not measure the size or importance of the effect. The registry record does not identify a trial-specific significance threshold or explain whether the p-value came from a prespecified analysis. The estimate and its confidence interval are therefore needed alongside the p-value to understand direction and precision.

Because the p-value is unadjusted, it does not account for testing multiple outcomes or comparisons. The record lists three outcome measures but no secondary results and no multiplicity procedure. The primary outcome is identifiable in the record, but the analysis plan and status of any other planned tests are not reported. This limits what can be inferred about selective reporting or family-wise error control.

What does the confidence interval add?

A confidence interval expresses the uncertainty around an estimate under a particular statistical procedure. In general, it shows a range of values compatible with the data and the assumptions used to construct it; it is not a guarantee that the true value lies in the interval. For this comparison, the reported interval crosses zero. That means the posted data do not exclude no mean difference at the interval’s corresponding confidence level. The interval also permits values on either side of zero, so the result does not precisely establish the direction of a population difference.

Confidence intervals are useful because they show how much uncertainty remains, while a p-value compresses information into a test statistic. Their interpretation still depends on the analysis method, data completeness, and validity of assumptions. The registry does not provide sufficient detail to verify the interval’s construction or its treatment of missing observations. Accordingly, it is appropriate to interpret the reported interval as the posted uncertainty summary, without attributing unreported model choices to it.

What is a recalculated comparison?

A recalculated comparison uses the posted arm-level summary statistics to form a simple unadjusted contrast. It is not a substitute for a trial’s prespecified analysis, which can include design features, covariates, and missing-data procedures not recoverable from group means and standard deviations. The record marks no discrepancy between the comparison and the arm summaries. No detailed sponsor analysis is posted in the registry record, so the available contrast should not be described as an adjusted or fully documented trial analysis.

General statistical interpretation also distinguishes uncertainty from importance. A confidence interval can be narrow or wide relative to a decision-relevant threshold, but no such threshold is reported for this scale. The point estimate is on the scale’s original units, and the record gives its direction. Without a defined meaningful-difference benchmark, the statistical estimate cannot be translated into a definitive statement about practical importance.

10. Confidence Intervals and Precision

The estimated mean difference is 0.23 units, comparing educational training with virtual training. The reported confidence interval extends from a value just below zero to a positive value around one-half unit. It crosses zero, so the data are compatible with no average difference under the interval’s statistical assumptions. They are also compatible with a positive difference in the stated direction. The interval therefore does not pinpoint the size or direction of a population-level difference with high certainty.

A confidence interval is a measure of estimation uncertainty, not a range containing most caregivers’ scores. The standard deviations describe variation in individual observations; the interval describes uncertainty around the estimated group contrast. It should not be interpreted as showing that individual caregivers’ outcomes fall within its bounds. Nor does inclusion of zero prove that the groups are identical. It indicates that zero remains among the values compatible with the data and the method used.

The unadjusted p-value of 0.095 is consistent with this uncertainty: the result does not provide strong evidence against a no-difference null under the unadjusted test. But the p-value cannot tell readers whether an effect is important, and the interval cannot answer that question without a meaningful-difference threshold. No such threshold is reported in the registry record.

The precision assessment is limited by missing methodological detail. The registry record does not describe the confidence-interval procedure, the handling of missing endpoint data, or the analysis population. The endpoint group counts differ from the overall enrollment figure, but the record does not explain why. These omissions make it difficult to judge whether the uncertainty calculation accounts for the trial’s full design and data availability. The posted interval remains the appropriate uncertainty summary to report, while its assumptions and relationship to the trial’s planned analysis remain undocumented.

11. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures are participants per arm for 80% power and participants per arm for 90% power. They are not totals across groups and are not event counts. The registry record’s planning illustration uses the observed effect as its reference and shows scenarios for detecting 100%, 75%, and 50% of that observed effect. Under the stated illustration, the corresponding per-arm figures are 161 and 215 for 100% of the observed effect; 285 and 381 for 75%; and 641 and 857 for 50%, respectively.

The planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is the stated fraction of the observed effect. These are assumptions for the planning illustration, not reported assumptions from the trial’s original design. The trial’s own original sample-size assumptions are not reported in the registry record.

In general, 80% power means that a study designed under the stated assumptions would have an 80% chance of detecting an effect of the specified size using the stated test threshold, if that effect were the true one. The 90% column represents a higher planned chance of detecting that specified effect under the same assumptions. Neither column is a probability that the observed result is correct. These values are planning illustrations based on the observed effect, not observed power.

Smaller effects require more participants to distinguish them from random variation with the same planned power and significance level. The figures therefore increase as the assumed fraction of the observed effect decreases. Such calculations are sensitive to the effect used as the target: an observed effect from one trial is uncertain and may not be the effect a future study should expect. These values should be read as conditional planning scenarios, not as a recommendation or a description of the trial’s original design.

12. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 0.230; pooled standard deviation σ = 0.734.
2. n = 2 × (1.96 + 0.842)² × 0.734² / 0.230² = 160.0, rounded up to 161 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
delta <- 0.23                      # observed difference in means
sigma <- 0.734386144238545                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(4489303)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 161 (80% power), 215 (90% power)
 75% of observed effect: n per arm = 285 (80% power), 381 (90% power)
 50% of observed effect: n per arm = 641 (80% power), 857 (90% power)
power.t.test: 161 
Simulated power at n = 161 per arm: 0.802 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

13. Multiplicity, Missing Data and Analysis Populations

The registry record identifies the Revised Memory and Behavior Problems Checklist at 3 months as the primary outcome and lists three outcome measures overall. It posts no secondary outcome results and no analysis entries. A count of listed outcomes does not reveal how many hypotheses were tested, which outcomes were prespecified for formal testing, or whether results were selectively reported. Those details are not reported in the registry record.

Multiplicity arises when several outcomes, time points, subgroups, or comparisons are tested. In general, performing many tests can increase the chance of at least one apparently notable result under a no-effect setting. Researchers may address this through a prespecified hierarchy, adjustment procedure, or another strategy, but no such approach is reported. The posted p-value is unadjusted, so it should not be treated as controlling error across a family of tests.

The record does not describe missing-data frequency for the primary endpoint, reasons for missingness, imputation, sensitivity analyses, or the analysis population. It gives outcome-summary counts of 60 and 55, while registry enrollment is 156, but does not explain the difference. It would be unwarranted to infer whether missingness resulted from withdrawal, incomplete questionnaires, eligibility decisions, or another cause. No intention-to-treat, per-protocol, or other population designation is reported.

These reporting gaps affect interpretation. Missing outcome data can reduce precision and can bias a group comparison if availability differs systematically with outcome or treatment assignment. This is a general concern, not evidence that bias occurred in this trial. A fully assessable analysis would need information about participant flow, outcome availability by group, the planned estimand, and the handling of missing data. Those details are not reported in the registry record, so the posted means and comparison should be read as results for the observations represented in the summaries.

14. 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 registry reports adverse-event figures separately for the two arms. For Virtual Patient Behavioral Response Training, 78 were listed as at risk for serious adverse events and 78 as at risk for other adverse events; the reported affected counts were 0 for serious events and 0 for other events, with 0 deaths. For Educational Training, the corresponding at-risk counts were 78 for serious events and 78 for other events; the reported affected counts were also 0 and 0, with 0 deaths.

These are the counts recorded in the adverse-event fields, not a measure of the primary caregiver-bother outcome. The record does not provide serious-event terms, and it does not describe event definitions, surveillance procedures, duration of follow-up for safety, or how the at-risk counts relate to the primary outcome counts. A reported count of zero means no affected participants were entered in those registry fields; it does not establish that no event of any kind occurred outside the scope or reporting process represented by those fields.

The safety figures should also be interpreted with the trial’s behavioral-program context in mind without assuming that all possible harms were measured in the same way. The record does not provide enough detail to compare completeness of safety monitoring or to characterize unreported event types. These registry entries are a limited account of reported adverse events, separate from conclusions about comparative effectiveness.

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

The posted primary summaries show a lower mean conditional-bother score in the Virtual Patient Behavioral Response Training group than in the Educational Training group at 3 months. The difference in means is 0.23 units, expressed as educational training minus virtual training. Because lower scores are better on this scale, the point estimate’s direction is consistent with less average caregiver-reported bother in the virtual training group. The confidence interval crosses zero, and the unadjusted p-value is 0.095, so the estimate remains uncertain.

This result does not establish that the programs are equivalent. Equivalence requires a prespecified margin defining how small a difference must be to count as equivalent, and the record reports no such margin. Nor does the result establish that either program is superior. The point estimate is only one component of interpretation; uncertainty, outcome meaning, missing data, and analysis methods all matter.

The scale measures caregivers’ conditional bother from behaviors that occurred, not all aspects of caregiving, caregiver health, or the dementia patient’s overall condition. It is caregiver-reported and refers to behaviors in the past month at the stated assessment. The record does not report a threshold for a meaningful score difference, baseline primary-outcome summaries, or long-term results. Therefore, the 3-month contrast cannot by itself show whether the difference would be important to caregivers or persist over time.

The trial’s findings also should not be generalized automatically to all caregivers, settings, or training programs. The registry describes two specific program types, but details of their content and delivery are limited. The primary outcome summaries include fewer participants than registry enrollment, and the reasons and analysis approach are not reported. This does not demonstrate that the result is biased, but it limits assessment of representativeness and robustness.

In short, the posted data describe a modest positive mean contrast in the stated direction, with uncertainty that includes no difference. They do not support a definitive claim about treatment benefit, equivalence, clinical importance, or broader outcomes. Further interpretation would require more complete reporting of the planned analysis, participant flow, missing data, and the meaning of score differences to caregivers.

16. Why This Trial Matters Statistically

This trial raises several statistical questions that are common in behavioral research. The outcome is a caregiver-reported continuous score, and its direction matters: lower values represent better outcomes, while the posted contrast is educational training minus virtual training. Reading the sign without checking both the contrast order and the scale direction could reverse the substantive interpretation.

It also illustrates the difference between an observed point estimate and the evidence available to support a precise conclusion. The estimated contrast is 0.23 units, but its confidence interval crosses zero. The p-value is 0.095. These measures do not show that the groups are identical; they indicate that the observed data leave meaningful uncertainty about the population contrast. Statistical significance and practical importance are distinct questions, and the registry record gives no meaningful-difference threshold for this measure.

Outcome availability and analysis transparency matter alongside the numerical estimate. Registry enrollment is 156, whereas the primary summaries list group counts of 60 and 55. Without participant-flow and missing-data information, readers cannot determine why the outcome counts differ or whether the comparison represents the intended analysis population. The record also contains no detailed statistical plan or analysis entries. This is not evidence of a flaw in the trial, but it limits independent assessment of the analysis choices.

Finally, the sample-size illustration shows how planning depends on the target effect. Smaller assumed effects require larger per-arm sample sizes under the stated assumptions. Because the illustration is based on the observed effect rather than the trial’s documented original design assumptions, it should be used to understand the relationship between effect size and planning, not to reconstruct the trial’s design or judge its observed result as adequately powered. The trial is informative as a case study in reading continuous outcomes, uncertainty, registry reporting, and the limits of summary-statistic comparisons.

17. 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.

Several limitations specific to this trial’s registry record constrain interpretation. The primary outcome summaries include 60 participants in the virtual training group and 55 in the educational training group, while the registry enrollment is 156. The record does not explain the relationship among these figures, identify the endpoint analysis population, or describe participant flow. Reasons for missing outcome observations are not reported in the registry record.

The study is described as randomized and parallel, with no masking. Further details about sequence generation, allocation concealment, recruitment, adherence, program exposure, or crossover are not reported. The registry describes the app and traditional education in broad terms, so it is not possible to assess whether differences in delivery or content could affect interpretation.

The primary endpoint is caregiver-reported and reflects conditional bother for behaviors occurring in the past month at the 3-month assessment. The record does not post baseline values for this endpoint, a change-from-baseline estimate, or longer-term results. It also does not identify a threshold for a meaningful difference on the scale. The available contrast therefore cannot answer whether the observed difference is important to caregivers or durable.

The registry record does not report a detailed analysis plan, the method used to construct the confidence interval, adjustment for covariates, missing-data procedures, or multiplicity control. It lists three outcome measures but provides no secondary outcome results. These omissions do not establish that the trial lacked such procedures; they mean that they are not documented in the registry record.

Finally, the planning illustration uses the observed effect and stated planning assumptions, but the trial’s original sample-size assumptions are not reported. It cannot establish the basis for the trial’s enrollment or provide observed power. The adverse-event entries also lack event terms and monitoring detail. Together, these limitations call for a focused reading of the primary posted summaries rather than broader claims about the intervention or the trial’s full evidence base.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Virtual Patient Behavioral Response Training for Family Caregivers — NCT04489303.
• Registry record last updated 2023-05-18; status last verified 2023-05.
• Results first posted 2023-05-18.
• 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.

21. Record Summary

The registry record describes a completed phase 2 randomized parallel-group trial of Virtual Patient Behavioral Response Training versus Educational Training for family caregivers of people with Alzheimer disease and related dementia. The primary endpoint was caregiver-reported conditional bother on the Revised Memory and Behavior Problems Checklist at 3 months, with higher scores indicating worse outcomes.

The reported means were 0.66 for virtual training and 0.89 for educational training. The unadjusted mean difference, educational training minus virtual training, was 0.23 units; its confidence interval crossed zero, and the unadjusted p-value was 0.095. The result therefore describes a lower mean score in the virtual training group but remains uncertain.

The record lists enrollment of 156, while the primary outcome summaries report group counts of 60 and 55. It does not explain the difference or document missing-data handling, analysis populations, a detailed analysis plan, or secondary outcome results. Adverse-event entries report zero affected participants and zero deaths in both groups, with 78 at risk listed for each adverse-event category in each group.