Combination Treatment Study for Memory Impairment and Depression: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Combination Treatment Study for Memory Impairment and Depression (NCT01658228), a randomized phase 4 study. The primary outcome is Selective Reminding Test (SRT) Total Recall.

2. Clinical Question

3. Trial Design

NCT01658228 is a completed Phase 4 trial with randomized allocation, quadruple masking, and a parallel-group model. The registry lists enrollment of 86. The trial is titled “Combination Treatment Study for Memory Impairment and Depression,” and its official title describes a pilot combination treatment trial for mild cognitive impairment with depression. The sponsor is New York State Psychiatric Institute. The registry lists a start date of 2011-09 and primary completion of 2016-01.

The treatment pathway began with open antidepressant treatment. The brief summary describes citalopram for 8 weeks, followed by reassessment. It says participants whose depression responded to citalopram would be randomized to add-on donepezil or placebo, while nonresponders would receive open venlafaxine treatment and be randomized later. The arm descriptions instead say that the initial open-label antidepressant period lasted 16 weeks and that all patients were then randomized, double-blind, to add-on donepezil or placebo for 62 weeks. These accounts do not align completely, so the exact sequence and timing for every participant are not fully clear from the registry record.

The registry lists citalopram, venlafaxine, donepezil, and placebo as interventions. Because antidepressant treatment preceded or accompanied the blinded add-on comparison, the posted comparison concerns donepezil versus placebo within that treatment context. The primary outcome is identified at Week 16, while the arm descriptions also describe blinded treatment after 16 weeks. The relationship between that timing and the outcome assessment is not explained in detail. The registry record does not report the randomization procedure, concealment method, or the rationale for the listed masking designation.

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: 86 subjects were enrolled in the open treatment phase of the study and then 61 subjects were randomized at 16 weeks to the add-on donepezil/placebo treatment groups.

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 Selective Reminding Test (SRT) Total Recall, measured over Week 16.

The primary endpoint is Selective Reminding Test (SRT) Total Recall at Week 16, reported as a mean number of words. The registry describes the SRT as a 12-item, 6-trial memory measure of verbal list learning and memory. Total immediate recall is the total number of words learned over the six trials. The measure is intended to capture memory performance through recall of a word list, rather than mood symptoms or a general clinical status measure.

A continuous endpoint retains information about each participant's score and allows the group result to be expressed as a mean and a difference between means. In general, a mean summarizes the central level of a continuous measure, while the standard deviation describes how spread out individual scores are around that mean. Here, the posted arm summaries give means and standard deviations for the two treatment groups, and the comparison estimates the difference in their means.

The SRT is relevant to the trial's stated focus on memory impairment in participants with depression and mild cognitive impairment. The endpoint definition does not specify a minimally important difference or a threshold that would classify a change as clinically meaningful. The registry record also does not identify whether the Week 16 result is an endpoint score, a change from baseline, or an adjusted estimate; the posted summaries are presented as group means. Thus the interpretation should remain focused on the reported scores and their statistical uncertainty, without assigning clinical importance that the record does not define.

7. Results Posted in the Registry

• Donepezil Treatment Group: mean 45.60 versus 46.60 on control; difference -1.00 (95% CI -6.65 to 4.65), standardized effect d = -0.09; z-test p = 0.729.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Alzheimer's Disease Assessment Scale - Cognitive (ADAS-Cog)

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Donepezil Treatment Group: mean 13.20 versus 13.90 on control; difference -0.70 (95% CI -4.24 to 2.84), standardized effect d = -0.10; z-test p = 0.699.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Donepezil Treatment Group and the control. The interval gives the range of effects that remain compatible with the data.

The registry lists the Alzheimer's Disease Assessment Scale - Cognitive (ADAS-Cog) as a secondary outcome and posts results for that measure at Week 16. Its reported comparison is another continuous mean difference between the donepezil and placebo groups. The direction of that estimate is negative, as is the primary mean difference, while its confidence interval also includes zero. In broad terms, the secondary result does not provide a clear contrasting pattern that resolves the uncertainty in the primary memory result.

The ADAS-Cog measures a different aspect of cognition from the SRT Total Recall measure. A result on one outcome does not automatically confirm or invalidate a result on another. The registry lists 15 outcome measures, but only one secondary outcome is posted in the posted registry content, and there are no reported analyses listed for that outcome. Consequently, the visible secondary evidence is limited and cannot establish a consistent pattern across all outcomes named or collected in the trial.

Secondary outcomes are often useful for exploring whether findings extend beyond the primary measure, but they generally require cautious interpretation when many outcomes may have been assessed. Testing multiple outcomes increases the chance that at least one result appears notable by chance unless the analysis accounts for multiplicity. No multiplicity adjustment or prespecified hierarchy is reported in the registry record. The posted secondary result is therefore unadjusted, and it carries less weight than a clearly prespecified primary analysis, especially when the analysis plan and full outcome reporting are unavailable.

The appropriate high-level reading is that the posted secondary cognitive measure does not remove the primary result's uncertainty. It should be regarded as additional, limited context rather than independent confirmation. Other secondary outcome results and their analysis details are not reported in the registry record, so no broader claim about the full set of secondary outcomes can be made.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Alzheimer's Disease Assessment Scale - Cognitive (ADAS-Cog)

9. Statistical Interpretation

10. Statistical Methods Explained

What does the mean difference represent?

A mean difference compares the average score in one group with the average score in the other. The posted comparison uses the Placebo Treatment Group as the reference and reports donepezil minus placebo. The estimate of -1.0 words corresponds to the two posted group means: 45.6 words in the donepezil group and 46.6 words in the placebo group. A negative estimate means the donepezil group average was lower on this measure at this timepoint. The sign alone does not indicate whether the difference matters clinically; that requires context about the measure and a meaningful difference threshold, which is not reported in the registry record.

How should the standard deviations be read?

The standard deviation describes variability among individual observations within a group. It is not the uncertainty in the group mean and is not a range containing every participant's score. The posted standard deviations, 11.2 words for donepezil and 11.3 words for placebo, indicate that individual scores varied around their respective group averages. When the question concerns the difference between group means, uncertainty also depends on the number of observations contributing to each mean. Standard deviations and sample sizes are therefore both relevant, but neither alone provides the full interpretation.

What does a confidence interval add?

A confidence interval gives a range of estimates compatible with the data under the statistical method used. In repeated applications of a suitable procedure, intervals constructed in this way would cover the underlying parameter at the stated rate. It is not a guarantee that the particular interval contains the true effect, nor does it assign a probability to each value within its limits. The reported interval, -6.647 to 4.647 words, includes zero and values on both sides. This makes the direction and size of the underlying difference uncertain. The registry does not state whether the interval was adjusted for covariates or multiplicity.

What does the p-value say?

The unadjusted p-value of 0.729 measures how unusual data at least as inconsistent with a null difference would be if the null model and the test assumptions were correct. It is not the probability that the null hypothesis is true, and it does not measure the probability that the result is due to chance. A p-value is also not a measure of clinical importance. The estimate and confidence interval are needed to understand the direction and range of effects compatible with the observed data. Because the record does not identify a multiplicity procedure, the posted value should be treated as unadjusted.

What is a standardized mean difference?

A standardized mean difference expresses the difference between group means using a pooled standard deviation as the scale. The reported value is -0.089. This can help compare the magnitude of continuous outcomes measured on different scales, but it is less directly interpretable than the raw difference in words. Standardization does not remove uncertainty, establish clinical importance, or resolve limitations in the design and analysis. The raw score scale remains the clearest basis for explaining the primary endpoint to readers.

Why does over-dispersion matter in some analyses?

General statistical context: over-dispersion occurs when data vary more than a model assumes. If a model underestimates variation, its standard errors may be too small and its confidence intervals too narrow; inference can then appear more certain than the data warrant. This is especially relevant to count outcomes and certain generalized models. The primary endpoint is reported as a continuous mean with standard deviation, and the registry record does not describe a model that assesses over-dispersion. It should not be presumed to have occurred in this analysis.

What can be inferred from randomization and masking?

In general, random allocation aims to make treatment groups comparable in expectation, while masking can reduce the influence of treatment knowledge on care, measurement, and reporting. These design features do not guarantee that groups are balanced in the realized trial or that missing data and analysis choices cannot bias a result. The record lists randomized allocation and quadruple masking but does not report the sequence-generation method, allocation concealment, the identities of masked parties, or whether masking was assessed. Those details are not reported in the registry record, so the design labels cannot answer every question about potential bias.

What is known about the analysis model?

The outcome is identified as a mean of a continuous measure, and the posted comparison is a mean difference. The registry does not report a detailed statistical analysis plan, model specification, covariate adjustment, handling of repeated measurements, missing-data method, or analysis population. It also reports no analyses in the analysis listing. The estimated comparison is therefore best described using the posted group summaries and comparison, without attributing unreported methods to it. A recalculated result, when labelled as such, would be a simple unadjusted calculation from posted summary statistics; it would not replace a sponsor's own analysis if the two differed. No discrepancy is indicated for this record.

11. Confidence Intervals and Precision

The primary estimate is a mean difference of -1.0 words, with a 95% confidence interval from -6.647 to 4.647. The interval is wide enough to include a range of plausible differences on either side of zero. It therefore leaves uncertainty about both the direction and the magnitude of the underlying group difference. The posted data do not precisely distinguish a small difference from a larger difference in either direction.

Precision is affected by the amount of information in the comparison, including the number of participants contributing outcome data and the variability of their scores. The primary outcome summaries include 30 participants in the donepezil group and 31 in the placebo group, with standard deviations of 11.2 and 11.3 words. These sample sizes are modest, and the observed variability is substantial relative to the estimated difference. That combination is consistent with the broad interval, although the registry does not report a formal precision analysis or the reasons these participants contributed to the posted outcome.

The interval's inclusion of zero should not be interpreted as proof of no effect. Likewise, the point estimate near zero should not be treated as a precise estimate of equivalence. Demonstrating equivalence or noninferiority would require a prespecified margin and an appropriate analysis; such a margin and analysis are not reported in the registry record. The interval instead shows the range of effects compatible with the data under the calculation used.

The unadjusted p-value of 0.729 is consistent with the interval's failure to rule out zero, but it does not make the interval narrower or resolve uncertainty. The estimate, interval, and sample size should be considered together. The registry record does not give a minimum clinically important difference, so it is not possible to determine whether particular values in the interval would be clinically meaningful.

12. 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 a continuous-outcome comparison. They are not totals of events. The planning illustration uses a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. It sets the effect to detect as a stated fraction of the observed effect. The registry record does not report the trial's original sample-size assumptions or justification, so these planning calculations should not be mistaken for the trial's design target.

The 80% power column gives the number of participants per arm that the planning calculation estimates would be needed for an 80% chance of detecting the specified effect, if that effect were the true one and the stated assumptions held. The 90% power column gives the corresponding planning number for a 90% chance. These are prospective planning probabilities under assumptions, not guarantees about any future study.

The entries are evaluated at 100%, 75%, and 50% of the observed effect. Smaller effects require more participants to distinguish them from random variation under the same assumptions. In this case, the observed primary estimate is close to zero, so planning figures derived from fractions of that estimate become very large. They should be read cautiously: the observed effect is uncertain, and using it as the basis for planning can produce unstable or unhelpful targets. The calculation does not establish what effect a future study should aim to detect; that choice requires a scientifically justified target.

The curve expresses the same planning relationship visually: as the effect to detect becomes smaller, the required number of participants per arm increases. It is not an additional trial result. This is not observed power. Observed power is not an informative substitute for reading the estimate and its confidence interval, and these planning values do not describe the probability that the completed trial detected an effect.

13. 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 = -1.000; pooled standard deviation σ = 11.251.
2. n = 2 × (1.96 + 0.842)² × 11.251² / -1.000² = 1,987.1, rounded up to 1,988 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                      # observed difference in means
sigma <- 11.2509585279038                     # 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(1658228)
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 = 1,988 (80% power), 2,661 (90% power)
 75% of observed effect: n per arm = 3,533 (80% power), 4,730 (90% power)
 50% of observed effect: n per arm = 7,949 (80% power), 10,641 (90% power)
power.t.test: 1988 
Simulated power at n = 1,988 per arm: 0.800 (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.

14. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study evaluates several outcomes, timepoints, subgroups, or statistical contrasts. If each is tested separately without adjustment or a planned hierarchy, the chance of at least one apparently unusual result can increase. The registry lists 15 outcome measures, while the posted results identify one primary outcome and one secondary outcome. The registry record does not report whether the remaining measures were analyzed, how outcomes were prioritized, or whether any adjustment for multiple comparisons was planned or applied.

The primary comparison is accompanied by an unadjusted p-value. The secondary comparison is also presented without a reported multiplicity adjustment. No prespecified testing sequence, family of hypotheses, or error-control method is reported in the registry record. The results should therefore be interpreted as outcome-specific estimates rather than as a multiplicity-controlled set of confirmatory tests. The absence of reported adjustment does not prove that none was used outside the registry, but its use is not documented in the record.

Missing-data handling is also not reported. The registry lists enrollment of 86, while the primary outcome summaries include 30 participants in the donepezil group and 31 in the placebo group. The reason that the posted outcome counts differ from overall enrollment is not reported in the registry record. The record does not state whether participants were excluded, lacked a Week 16 assessment, withdrew, or were handled through imputation or another method. It also does not report missingness by treatment group or reasons for missing observations.

The analysis population is not defined for the primary endpoint. It is not reported whether the posted comparison follows an intention-to-treat, per-protocol, or other population, nor how participants who changed treatment or did not follow the planned pathway were handled. The record also does not provide a statistical analysis plan or the model details needed to determine whether the result was adjusted for baseline score, site, or other factors. These omissions limit assessment of robustness and bias; they should not be filled in by assumption.

15. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: fall (6.5%), cardiac stents (3.3%), allergic reaction (3.3%), atrial flutter (3.3%), chest pain (3.3%).

The serious adverse event summary lists 30 participants at risk in the Donepezil Treatment Group, with 8 affected, and 31 at risk in the Placebo Treatment Group, with 7 affected. The other adverse event summary lists 30 at risk and 23 affected in the donepezil group, and 31 at risk and 17 affected in the placebo group. These are counts from the registry record. They do not by themselves establish whether an event was caused by treatment or whether differences between groups are reliable.

The listed serious adverse event terms include Fall at 6.5%, and Cardiac Stents, Allergic Reaction, Atrial Flutter, Chest Pain, and Colorectal Cancer at 3.3% each. The record does not identify the treatment group for each listed term, and the terms should not be interpreted as comparative risks. The registry record lists deaths as null for both groups; death counts are therefore not reported in the registry record.

Safety comparisons require context about event definitions, observation time, exposure, severity, and how repeated events were counted. Those details are not reported in the registry record. The counts are descriptive and should not be used alone to infer a treatment-related safety difference.

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

The posted results describe mean cognitive scores at a specified timepoint among the participants represented in the outcome summaries. For SRT Total Recall at Week 16, the donepezil group mean was 45.6 words and the placebo group mean was 46.6 words. The estimated difference was -1.0 words, and the 95% confidence interval ranged from -6.647 to 4.647. This is the evidence relevant to the primary comparison in the registry record.

The result does not demonstrate a clear difference between the group means, but it also does not establish that the groups are equivalent. The interval allows values on either side of zero, and the record provides no equivalence margin or noninferiority framework. The p-value of 0.729 does not prove that there is no effect. Rather, the estimate is imprecise enough that more than one interpretation remains compatible with the posted data.

The finding applies to the SRT measure and the Week 16 comparison as reported. It does not establish effects on every aspect of memory, depression symptoms, progression of cognitive impairment, dementia, daily functioning, or outcomes at later timepoints. Although the trial title and summary describe a broader clinical context, the primary result alone cannot support conclusions about all of those outcomes.

The comparison also took place within a treatment pathway involving antidepressant medication. The result should not be generalized as a comparison of donepezil alone against no treatment. The registry provides an account of the pathway, but its timing descriptions do not fully align, and the exact analysis population and handling of missing observations are not reported. These limits matter when considering how closely the posted comparison represents all enrolled participants or other populations.

The result is best understood as an uncertain estimate from a completed pilot trial, rather than as a definitive answer about treatment effectiveness. It can inform how the observed outcome and its precision are understood, but the registry does not report the design assumptions needed to judge whether the trial was intended to detect a particular clinically meaningful difference. No medical recommendation follows from this statistical interpretation.

17. Why This Trial Matters Statistically

This trial raises a common challenge in clinical biostatistics: interpreting a small randomized comparison when the estimate is close to zero but its uncertainty is substantial. The primary result is not just a p-value. The mean difference, its confidence interval, the variability of scores, and the number of participants contributing to the outcome jointly show why the data do not support a precise conclusion about the direction or magnitude of the difference.

The registry describes a complex treatment pathway rather than a simple isolated drug comparison. Antidepressant response appears to influence subsequent treatment steps, and the brief summary and arm descriptions differ in their account of timing. That makes clear reporting of the population, treatment sequence, timepoint, and analysis method especially important. Without those details, readers may struggle to connect the posted Week 16 result with the full treatment protocol.

The record also illustrates why planning calculations based on an observed effect need careful interpretation. The observed estimate is uncertain, and planning from it can yield very large participant requirements when the estimate is near zero. Such calculations are not a substitute for an a priori design rationale or a clinically justified target effect. The trial's original sample-size justification is not reported in the registry record.

Finally, the incomplete reporting of additional outcomes, multiplicity procedures, missing data, and analysis populations limits the strength of broader conclusions. The statistical value of the record is in its transparent description of what the posted primary and secondary estimates show, alongside what remains uncertain. That distinction is useful for researchers designing and reporting pilot trials, and for readers assessing how much weight a result can bear.

18. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record leaves several trial-specific details unresolved. The brief summary describes citalopram response assessment at 8 weeks and later randomization for nonresponders after venlafaxine, while the arm descriptions say the initial antidepressant period lasted 16 weeks before all patients were randomized. The timing and treatment sequence are therefore not fully consistent across the record. The Week 16 primary endpoint is also not explained in relation to that sequence and the subsequent blinded treatment period.

The primary outcome population field is blank, and the record does not define the analysis population. The posted primary outcome includes 30 participants in the donepezil group and 31 in the placebo group, compared with overall enrollment of 86. Reasons for the difference, missing outcome data, exclusions, withdrawals, and participant flow are not reported in the registry record. This prevents a full assessment of how representative the posted analysis is of everyone enrolled.

The registry does not report the randomization method, concealment details, the roles covered by quadruple masking, adherence, treatment discontinuation, or whether masking remained intact. It also does not give an analysis plan, model specification, covariate-adjustment details, missing-data approach, or the original sample-size justification. These matters are important for assessing the risk of bias and the precision expected from the trial.

Although 15 outcome measures are listed, only one secondary outcome is posted in addition to the primary result. The record does not explain the status of other measures or whether multiplicity was addressed. Safety event details are also limited: the record does not provide event timing, exposure duration by participant, attribution, or group assignment for the listed serious event terms. These gaps restrict interpretation beyond the posted descriptive counts and estimates.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Combination Treatment Study for Memory Impairment and Depression — NCT01658228.
• Registry record last updated 2017-10-17; status last verified 2017-10.
• Results first posted 2017-10-17.
• 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.

22. Record Summary

NCT01658228 is a completed Phase 4 randomized, quadruple-masked, parallel-group pilot trial sponsored by New York State Psychiatric Institute. It concerns combination treatment for participants with depression and mild cognitive impairment, with an open antidepressant pathway followed by a blinded add-on comparison of donepezil and placebo.

The primary endpoint is SRT Total Recall at Week 16. The posted mean difference was -1.0 words, with a 95% confidence interval from -6.647 to 4.647 and an unadjusted p-value of 0.729. The interval includes zero and values in either direction, so the estimate remains uncertain. One secondary cognitive outcome is posted, but the registry does not report a full analysis plan, multiplicity handling, missing-data methods, or a defined analysis population. The record supports a cautious reading of the posted comparison, not a definitive conclusion about treatment effects.