In-Home Technology for Dementia Caregivers: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for In-Home Technology for Dementia Caregivers (NCT03828383), a randomized na study. The primary outcome is Zarit Burden Interview-Short Form.

2. Clinical Question

3. Trial Design

The registry record identifies this as a randomized, single-masked, parallel-group trial. It lists an enrollment of 78 and a completed status. The recorded start date is 2019-02-26, and the primary completion date is 2020-11-27. The study is listed as not applicable for phase, and the sponsor is the University of California, Berkeley.

Two device interventions were compared: the In-Home Technology System and the Limited In-Home Technology System. The system description includes entry, activity, temperature, and water-leak sensors, voice control, a digital display, and a local router. In the limited arm, the record says monitoring of the water-leak sensor and associated warnings would be active. The available arm descriptions do not fully specify every service or feature enabled in each group. The record says that a research team member who was blind to study arm installed the system; it also labels the study single-masked, but does not identify all masked parties or explain the masking procedure in further detail.

The primary outcome was assessed 9 months after baseline. The outcome population description reports that 9 participants dropped out before completing the baseline measure and another 9 dropped out before the 9-month assessment, leaving 60 people in the primary analysis. The posted arm sizes for that analysis were 33 for the full system and 27 for the limited system. The registry record does not report the allocation procedure in detail, the original sample-size justification, or the intended analysis population beyond the information in the outcome description. It also does not report the trial's original statistical analysis 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: Note that outcomes are only assessed regarding caregiver health and well-being. Patients did not provide any information; therefore, all information reported for demographics and measures is only regarding caregivers.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: 78 individuals enrolled into the study, 9 individuals dropped out before completing the baseline measure, and another 9 dropped out before completing the 9-month assessment for a total 60 individuals for this analysis.

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: Although 78 individuals originally consented to being in the study, 9 individuals dropped out between consenting/enrolling and completing the baseline questionnaire assessing baseline outcome measures; therefore, these baseline numbers only include 69 participants.

6. Primary Endpoint

The registry lists the primary outcome as Zarit Burden Interview-Short Form, measured over 9 months after baseline.

The primary endpoint was the Zarit Burden Interview-Short Form score, assessed 9 months after baseline. It is a questionnaire intended to measure caregiver burden. The record describes 12 items, each rated on a 0-4 scale, with a total range of 0-48 and no subscales. Higher scores represent worse outcomes.

Caregiver burden is relevant to the study's stated aim of evaluating in-home assistive technology intended to address anxiety, burden, and loneliness among spousal and familial caregivers. The endpoint gives a numerical summary of reported burden, allowing group means and their uncertainty to be compared. It captures the measure's stated construct, but it is not itself a direct measurement of every aim named in the study summary. The record lists four outcome measures overall, but the other outcomes and their results are not reported in the registry record.

For this continuous outcome, the posted results report a mean and standard deviation for each group. The mean describes the group's average score, while the standard deviation describes how spread out individual scores are around that average. These group summaries do not show every participant's response or explain the reasons for differences between individuals. Interpretation also depends on the number of people contributing data and on how missing assessments were handled; details of the missing-data method are not reported in the registry record.

7. Results Posted in the Registry

• Limited In-Home Technology System: mean 25.59 versus 24.58 on control; difference 1.01 (95% CI -3.20 to 5.22), standardized effect d = 0.12; z-test p = 0.638.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is continuous: it is a score on a scale, and the registry record reports group means and standard deviations. A mean difference compares the average scores in two groups. For the posted comparison, the subtraction is described as limited-system mean minus full-system mean. The resulting estimate was 1.01 points. A positive estimate therefore means the limited-system group's mean was higher. Because higher scores represent worse outcomes on this measure, the sign gives a direction for the observed score contrast; it does not establish why the scores differed.

A standard deviation describes the spread of participant scores around a group's mean. It is not the uncertainty in the mean itself. The posted standard deviations were 8.09 in the full-system group and 8.44 in the limited-system group. The numbers contributing to the endpoint were 33 and 27, respectively. These group sizes matter because estimates of means are generally less precise when based on fewer observations, all else being equal. The registry record describes 60 people in the analysis after 9 dropped out before baseline and another 9 before the 9-month assessment.

A confidence interval gives a range of values compatible with the observed data under the method's assumptions. The interval for the mean difference ranged from -3.203 to 5.223. It includes zero, the value representing no difference in group means. Its inclusion means the data do not rule out no mean difference at the conventional interpretation of this interval. The range also includes negative and positive values, leaving uncertainty about whether the population contrast might favor one group or the other. A confidence interval is not a guarantee that the true value lies within its endpoints, and it does not describe the range of individual participants' scores.

The unadjusted p-value was 0.638. In general, a p-value measures the compatibility of the data with a statistical model that assumes no group difference, conditional on the analysis assumptions. A larger p-value does not prove that groups are identical, and a smaller p-value would not by itself establish clinical importance or causation. The registry record does not report a threshold for interpreting this endpoint, a multiplicity adjustment, or a detailed inferential plan. The reported p-value is unadjusted, so it should not be described as accounting for testing across outcomes or analyses.

The posted comparison also includes a standardized effect estimate, but the arm-specific means, standard deviations, and mean difference communicate the main result directly. In general, standardized measures express a contrast in units of variability, which can aid comparisons across differently scaled measures. Such quantities do not make an uncertain estimate precise, and their interpretation depends on the measure and analysis. The standardized estimate is not necessary to understand the reported score difference.

Randomization is intended, in general, to balance participant characteristics across groups on average and support a fair comparison. The registry record labels the study randomized, but does not report the sequence-generation procedure, allocation concealment, or a detailed account of adherence to assignment. These details are therefore not available for assessing how randomization was implemented. The record labels the trial single-masked and says that a research team member installing the system was blind to arm. It does not identify all masked parties or describe whether participants could infer their assignment from the technology they received.

Missing outcome data can affect an estimate if participants with unobserved scores differ systematically from those with observed scores. In general, analysts may use complete-case comparisons, imputation, or other methods, each requiring assumptions. The registry record reports the number who did not complete the baseline and 9-month measures but does not report the missing-data method, reasons for dropout, or sensitivity analyses. Consequently, the posted means should be understood as results for the people included in the analysis, without assuming that they represent all enrolled participants equally.

In general, over-dispersion refers to variability greater than expected under a statistical model. It is particularly relevant for count outcomes and certain generalized models; it is not a special correction that can be inferred from these continuous-score summaries. The primary endpoint is reported as a mean with a standard deviation, and the registry record does not describe a variance adjustment or model checking. A reader should not assume a particular regression model, covariate adjustment, or robust standard error procedure. Those methods are not reported in the registry record.

Finally, statistical significance, precision, and practical importance are distinct. The p-value summarizes compatibility with a no-difference model, the confidence interval conveys uncertainty about the mean contrast, and the score scale describes the measured outcome. The record does not specify a threshold for a meaningful change in burden. Without that threshold, the observed estimate and its interval can be described statistically, but should not be labeled clinically important or unimportant on the basis of a cutoff that is not reported.

10. Confidence Intervals and Precision

The mean difference estimate was 1.01 points, with a confidence interval from -3.203 to 5.223. The interval spans zero and includes both negative and positive differences. Thus, the posted data leave uncertainty about the direction of a population-level mean contrast as well as its magnitude. The interval is the most direct guide to precision in this comparison: it indicates that values materially different from the observed estimate remain compatible with the data under the method's assumptions.

A confidence interval is not a range containing most individual scores. Individual scores vary around each group mean, as reflected by standard deviations of 8.09 and 8.44. Nor does the interval assign a probability to each possible true difference. In general, confidence intervals describe the performance of a method across repeated studies under its assumptions; in a particular study, they provide an uncertainty range for the estimate.

The analysis included 33 people in the full-system group and 27 in the limited-system group. The outcome description states that the 60-person analysis followed 9 dropouts before baseline and another 9 before the 9-month assessment. The unequal group sizes and the loss of participants from enrollment may limit precision. The registry record does not report a missing-data approach or sensitivity analysis, so the interval does not resolve uncertainty about the possible influence of unavailable scores.

The interval is not evidence of equivalence. Establishing equivalence would require a prespecified range of differences considered sufficiently small, together with an analysis designed to assess that range. Such a margin is not reported in the registry record. The interval also should not be used to claim that a change from baseline occurred, because the posted comparison is between groups at 9 months and no baseline-adjusted result is reported.

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 expressed as participants per arm, not as total enrollment or total events. They are illustrations based on the observed effect, rather than a reconstruction of the trial's original design assumptions. The registry record does not report the original sample-size justification.

The listed planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. For each scenario, the effect to detect is a stated fraction of the observed effect. The 80% power column gives the number of participants per arm in the illustration intended to yield an 80% chance of detecting the specified effect under those assumptions, if that effect and the model assumptions were correct. The 90% power column gives the corresponding per-arm planning number for a 90% chance.

For planning at 100% of the observed effect, the figures are 1,048 participants per arm for 80% power and 1,402 participants per arm for 90% power. For planning at 75% of the observed effect, they are 1,862 and 2,492 participants per arm. For planning at 50% of the observed effect, they are 4,189 and 5,607 participants per arm. Smaller effects require more participants in these illustrations because they are harder to distinguish from random variation under the same assumptions.

These are planning illustrations, not observed power. They do not tell whether the completed study had a particular chance of finding an effect, and they should not be read as evidence about efficacy. They use the observed effect as the starting point, which is uncertain, and the registry record does not report whether the trial was originally planned using these assumptions or targets.

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 = 1.010; pooled standard deviation σ = 8.249.
2. n = 2 × (1.96 + 0.842)² × 8.249² / 1.010² = 1,047.1, rounded up to 1,048 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 <- 1.01                      # observed difference in means
sigma <- 8.24873323607934                     # 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(3828383)
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 = 1,048 (80% power), 1,402 (90% power)
 75% of observed effect: n per arm = 1,862 (80% power), 2,492 (90% power)
 50% of observed effect: n per arm = 4,189 (80% power), 5,607 (90% power)
power.t.test: 1048 
Simulated power at n = 1,048 per arm: 0.804 (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 Zarit Burden Interview-Short Form as the primary outcome and lists four outcome measures overall. It does not report results for secondary outcomes. The study summary also names anxiety and loneliness among its aims, but their measures and results are not reported in the registry record. No outcome-specific multiplicity procedure or overall testing strategy is reported.

Multiplicity is the issue that arises when many hypotheses or outcomes are tested: the chance of at least one apparently notable result can increase as the number of tests grows. In general, prespecified priorities and adjustment methods can help readers interpret multiple results. Since the registry record does not give an adjustment method or a full analysis plan, the unadjusted p-value for the primary comparison should not be described as adjusted for other tests. The presence of multiple listed outcomes alone does not show that any particular adjustment was required or used.

Missing data are relevant to the primary endpoint. The record states that 78 people enrolled, 9 dropped out before completing the baseline measure, and another 9 dropped out before the 9-month assessment, leaving 60 in the analysis. It reports endpoint group sizes of 33 and 27. The reasons for dropout, the distribution of missing outcomes across study groups beyond the reported endpoint sizes, and the method used to handle missing scores are not reported in the registry record. No sensitivity analysis addressing missingness is reported.

The analysis population is described only through the number of people included in this endpoint analysis. The registry record does not say whether the comparison followed an intention-to-treat principle, whether participants were analyzed according to random assignment regardless of technology use, or whether any other population definition was used. It also does not report adherence, crossover, or protocol-deviation summaries. These omissions limit assessment of how closely the posted comparison corresponds to the randomized assignment and how robust it is to incomplete follow-up.

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 adverse-event figures describe participants at risk and reported events for each group. In the In-Home Technology System group, 37 participants were at risk for serious adverse events and 0 were affected; 37 were at risk for other adverse events and 0 were affected. In the Limited In-Home Technology System group, 32 were at risk for serious adverse events and 0 were affected; 32 were at risk for other adverse events and 0 were affected. The recorded number of deaths was 0 in both groups.

These figures indicate that no serious or other adverse events and no deaths were recorded in the posted safety results for the listed at-risk groups. They do not establish that no events occurred outside the registry's reporting definitions or observation period. The record does not provide descriptions of event-monitoring procedures, the time period covered, or a narrative explaining how adverse events were collected and assessed. It lists no serious adverse-event terms.

Zero reported events should be read as a feature of the posted safety data, not as proof that an intervention has no risk. The at-risk group sizes are limited, and the record does not give additional detail about exposure time or follow-up for safety. These safety counts are separate from the primary caregiver burden score and do not determine the interpretation of that endpoint.

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

The posted primary outcome results show that the average 9-month caregiver burden score was 24.58 in the In-Home Technology System group and 25.59 in the Limited In-Home Technology System group. The estimated mean difference was 1.01 points, with a confidence interval from -3.203 to 5.223 and an unadjusted p-value of 0.638. Given the score's direction, the limited-system group had the higher observed mean, but the interval includes differences in either direction and no difference.

These findings do not establish that the two systems are equivalent. Equivalence requires a defined margin and a suitable analysis, neither of which is reported in the registry record. They also do not establish that either system reduced or increased burden relative to baseline. The posted primary comparison is at 9 months, and a baseline-adjusted result or a reported within-group change is not available in the registry record.

The result applies to the participants included in this analysis, not automatically to every caregiver or to all people with dementia. The record reports 78 enrolled participants and 60 in the endpoint analysis after dropouts before baseline and before the 9-month assessment. It does not provide enough information about participant characteristics, missing-data handling, or use of the technology to determine how well the analyzed group represents other caregivers.

The study's stated aims include anxiety, burden, and loneliness, but the posted result concerns the burden questionnaire. The registry record does not report secondary outcome results, so no conclusion about anxiety or loneliness follows from this primary endpoint. The zero recorded adverse events likewise describe the registry's safety entries and do not prove absence of risk. The most defensible interpretation is a modest observed difference in group means accompanied by uncertainty that does not clearly establish its direction or size.

16. Why This Trial Matters Statistically

This trial illustrates why an estimate should be read together with its uncertainty rather than in isolation. The observed mean difference is 1.01 points, but its confidence interval ranges from -3.203 to 5.223. Reporting only the direction of the point estimate would conceal the range of values compatible with the data, while relying only on the p-value would omit information about magnitude and precision.

The outcome is a caregiver-reported measure with a defined score range and a clear direction: higher scores represent worse outcomes. That makes the sign of the group contrast interpretable, but the registry record does not report a threshold for a meaningful difference. Statistical evidence about a mean contrast therefore cannot, by itself, answer whether a particular difference matters to caregivers.

The participant flow also matters for interpretation. The registry record states that 78 people enrolled and 60 contributed to this endpoint analysis after dropouts before baseline and before the 9-month assessment. The endpoint analysis groups contain 33 and 27 participants. Reporting these denominators alongside the result helps readers understand the evidence base and recognize that missingness could matter, especially when the method for handling absent outcomes is not described.

Finally, the study compares a full and a limited home technology system rather than a device against no technology. The available descriptions identify system components and the active water-leak monitoring in the limited arm, but do not fully characterize every difference between arms. Statistical interpretation depends on knowing what contrast the groups represent; the incomplete intervention detail constrains interpretation of the precise feature or service associated with any observed difference.

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.

The registry record offers limited detail for judging the conduct and analysis of the trial. It labels the design randomized, single-masked, and parallel, but does not describe the randomization sequence, allocation concealment, or who besides the installing staff was masked. It does not provide the original sample-size justification, a detailed statistical analysis plan, or the trial's original design assumptions.

The primary analysis includes 60 participants, compared with 78 enrolled. The outcome description records 9 dropouts before baseline and another 9 before the 9-month assessment. Reasons for dropout, group-specific missingness beyond the analysis sizes, and the statistical handling of missing data are not reported in the registry record. No sensitivity analysis for missing outcomes is reported.

The arm descriptions identify technology components but do not fully explain every operational difference between the full and limited systems. The record does not report how much participants used the technology, whether use varied over time, or whether any participants crossed between conditions. These details would help clarify the treatment contrast represented by the group comparison.

The posted primary result is a 9-month between-group comparison. Baseline group scores, a baseline-adjusted estimate, and change-from-baseline results are not reported in the registry record. A threshold for a meaningful difference is also not reported, so the clinical importance of the estimated score contrast cannot be classified using a predefined criterion. Finally, although four outcome measures are listed, secondary outcome results are not reported in the registry record, limiting assessment of the study's broader stated aims.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: In-Home Technology for Dementia Caregivers — NCT03828383.
• Registry record last updated 2022-05-10; status last verified 2022-04.
• Results first posted 2022-05-10.
• 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, randomized, single-masked, parallel-group study of in-home technology for dementia caregivers, sponsored by the University of California, Berkeley. It lists 78 enrolled participants and compares the In-Home Technology System with a Limited In-Home Technology System.

The primary outcome was the Zarit Burden Interview-Short Form score at 9 months after baseline, with higher scores representing worse outcomes. The analysis included 60 participants: 33 in the full-system group and 27 in the limited-system group. Mean scores were 24.58 and 25.59, respectively; the mean difference was 1.01, with a confidence interval from -3.203 to 5.223 and an unadjusted p-value of 0.638.

The interval includes zero and leaves uncertainty about the direction and magnitude of the difference. Missing-data methods, the original sample-size justification, and secondary outcome results are not reported in the registry record.