A Study Of PF-04447943 Compared To Placebo In Subjects With Mild To Moderate Alzheimer's Disease: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Study Of PF-04447943 Compared To Placebo In Subjects With Mild To Moderate Alzheimer's Disease (NCT00930059), a randomized phase 2 study. The primary outcome is Alzheimer's Disease Assessment Scale-Cognitive Subscale 70 (ADAS-Cog 70) Score at Baseline.

2. Clinical Question

3. Trial Design

This was a Phase 2, randomized, multicenter, double-blind, placebo-controlled, parallel-group study of PF-04447943 in subjects with mild to moderate Alzheimer's disease. The registry record describes masking as quadruple. The study status is completed. It began on 2009-09-10, and the primary completion date was 2010-09-22. Total enrollment was 191.

The two posted groups were placebo and PF-04447943. The baseline ADAS-Cog 70 results include 100 participants in the placebo group and 91 in the PF-04447943 group. The baseline NPI results use those same group counts. These posted group sizes describe the participants represented in those outcome summaries; the registry record does not provide arm descriptions that explain treatment details, adherence, or participant flow.

The study's stated purpose was broader than the primary outcome shown in the results. It aimed to evaluate cognitive, behavioral, and overall symptoms, assess safety and tolerability, and determine plasma levels of PF-04447943 over the course of the study. Listed secondary outcomes include ADAS-Cog 70 changes at Week 3, 6 and 9, CGI-I, NPI at baseline, and NPI changes at Week 3, 6, 9 and 12. The registry record does not report results for most of those listed outcomes.

The primary outcome entry is a baseline score, not a follow-up contrast or change from baseline. This distinction matters when interpreting a randomized trial: randomization is intended to support comparisons of outcomes between assigned groups, but a baseline measure records status before the study's follow-up period. A difference at baseline is descriptive and is not evidence of a difference caused by study treatment. The record does not report the trial's original sample-size justification, detailed analysis plan, or planned approach to interim analyses.

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

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: FAS included all participants who consumed at least 1 dose of randomized study medication.

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: Full analysis set (FAS) included all participants who consumed at least 1 dose of randomized study medication.

6. Primary Endpoint

The registry lists the primary outcome as Alzheimer's Disease Assessment Scale-Cognitive Subscale 70 (ADAS-Cog 70) Score at Baseline, measured over Baseline (Day 1).

The primary outcome reported is the Alzheimer's Disease Assessment Scale-Cognitive Subscale 70 (ADAS-Cog 70) score at Baseline (Day 1). The registry describes ADAS-cog as a structured scale for assessing cognitive impairment in Alzheimer's disease. It combines scores across domains that include memory, naming, following commands, praxis, orientation, language, word-finding, and comprehension. The total score ranges from 0 (least impairment) to 70 (most severe impairment), and higher scores indicate worse cognition.

A structured scale turns several aspects of cognition into a consistent numerical summary. Such measures can be useful in clinical research because they apply the same scoring framework across participants and allow group averages and variability to be described. This endpoint is therefore relevant to the study's stated interest in cognition.

However, the timing and wording of this posted primary outcome are important. It is a baseline value, not a change from baseline or a score measured after treatment. The group comparison therefore describes cognitive scores at the start of the study, rather than whether scores changed during treatment or whether the groups differed in change. The record lists later ADAS-Cog 70 change outcomes as secondary outcomes, but results for those outcomes are not reported in the registry record. The baseline endpoint alone cannot answer the study's treatment-effect question.

7. Results Posted in the Registry

• PF-04447943: mean 22.66 versus 21.87 on control; difference 0.79 (95% CI -2.16 to 3.74), standardized effect d = 0.08; z-test p = 0.600.

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.

Neuropsychiatric Inventory (NPI) Total Score at Baseline

• 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:
  • PF-04447943: mean 10.67 versus 12.16 on control; difference -1.49 (95% CI -4.91 to 1.93), standardized effect d = -0.12; z-test p = 0.394.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between PF-04447943 and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes listed for the study cover several domains and time points: cognitive change on ADAS-Cog 70, clinical global improvement, and NPI scores and changes. The posted numeric secondary comparison is NPI total score at baseline. Its contrast is descriptive of starting scores, not a follow-up effect. Results for the other listed secondary outcomes are not reported in the registry record.

That pattern means the posted secondary results do not provide an independent set of follow-up findings that can be compared with the primary result. In particular, the listing of an outcome in the registry is not the same as a reported result. Without posted estimates, intervals, and analysis details for those outcomes, it is not possible to assess their direction, magnitude, or precision.

Secondary outcomes can broaden the clinical picture by measuring different aspects of a condition, but their interpretation depends on prespecification, completeness of reporting, and the analysis plan. When many outcomes or time points are examined, the chance of obtaining a nominally notable result by chance can increase. Unadjusted secondary results therefore generally carry less weight than a well-defined primary analysis, especially when the multiplicity strategy is not reported.

The available baseline NPI comparison does not resolve the primary study question about changes during treatment. It instead describes baseline neuropsychiatric scores in the two groups. The record lists NPI change outcomes over follow-up, but their results are not reported in the registry record. Accordingly, the secondary endpoint list indicates what the study intended to assess, while the posted results support only a limited descriptive account.

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

• Change From Baseline in Alzheimer's Disease Assessment Scale-Cognitive Subscale 70 (ADAS-Cog 70) at Week 3, 6 and 9
• Clinical Global Impression - Improvement (CGI-I)
• Neuropsychiatric Inventory (NPI) Total Score at Baseline
• Change From Baseline in Neuropsychiatric Inventory (NPI) Total Score at Week 3, 6, 9 and 12

9. Statistical Interpretation

10. Statistical Methods Explained

The posted continuous outcomes are summarized with means and standard deviations, and the group contrasts are expressed as mean differences with confidence intervals. These familiar summaries are useful, but interpretation depends on what was measured and when. Here, the primary outcome is ADAS-Cog 70 at Baseline (Day 1), so its group contrast describes the baseline distribution rather than a post-treatment difference.

What does a mean describe?

A mean is the arithmetic average of the observed scores in a group. It gives a compact description of the group's central level on the scale. For ADAS-Cog 70, the registry record reports a mean of 21.87 for placebo and 22.66 for PF-04447943. Since higher scores indicate worse cognition, the higher displayed mean corresponds to a higher average score on that scale. A mean does not show how every participant scored, nor does it describe the full distribution by itself.

The standard deviation summarizes spread among individual scores around the group mean. The posted standard deviations are 10.42 for placebo and 10.39 for PF-04447943. A standard deviation is not the standard error and is not a confidence interval: it describes observed variability between participants rather than the precision of the estimated group mean. A broad spread can coexist with a precisely estimated mean when many observations contribute, while a smaller group can leave greater uncertainty.

What does the mean difference represent?

The mean difference compares the group means on their original scale. For the primary outcome, the posted contrast is 0.79 units, with the PF-04447943 group as the comparison against placebo. Because the outcome is at baseline, the contrast is a difference in starting score summaries. It should not be interpreted as a treatment effect or as a difference in change.

The NPI baseline contrast is also reported as a mean difference: -1.49 units. It describes the difference between baseline group means on that measure. The meaning of a positive or negative value depends on which group is subtracted from which and on the scale's direction. For the primary ADAS-Cog 70, higher values specifically indicate worse cognition, as stated in the registry record.

How should a confidence interval be read?

A confidence interval gives a range of values compatible with the data and the method used to estimate the contrast. The primary baseline contrast has a confidence interval from -2.164 to 3.744. It includes zero, the value representing no mean difference between groups. The NPI baseline interval also includes zero, running from -4.914 to 1.934. An interval that includes zero does not prove that the true difference is zero; it indicates that the data and method do not distinguish zero from the values covered by the interval at the interval's stated confidence level. The confidence level itself is not reported in the registry record.

Confidence intervals describe uncertainty in an estimate, not the range containing most individual participant scores. They also do not account automatically for every possible source of bias, measurement error, missing data, or analysis choice. Their interpretation relies on the analysis being appropriate for the data and study design. The registry record does not report enough detail to assess every such assumption.

What does the p-value add?

The unadjusted p-value for the baseline ADAS-Cog 70 contrast is 0.6; for baseline NPI it is 0.394. A p-value summarizes how unusual the data, or more extreme data, would be under a specified null model. It is not the probability that the null hypothesis is true, nor does it measure clinical importance. An unadjusted p-value also does not account for testing multiple outcomes. The registry record does not report a multiplicity adjustment for these results.

For these outcomes, the p-values concern baseline group comparisons. They cannot answer whether treatment affected change after baseline. The clinical question and the statistical test need to refer to the same time point and outcome: a baseline test addresses a starting-score contrast, while a treatment-effect analysis would need an appropriately defined follow-up comparison or change analysis.

Why report variability and standardized effects?

Standard deviations retain the units of the scale and show how much individual scores vary. A standardized effect can express a mean contrast relative to a measure of variation, which may help compare effects across different scales. The posted comparison summaries include a pooled standard deviation and a standardized effect statistic, but these summaries do not change the baseline nature of the endpoint. A standardized effect is not a substitute for a clinically interpretable outcome or a follow-up treatment contrast.

General statistical note: when data are substantially more variable than a simple model assumes, standard errors and tests based on that model can be unreliable. In count outcomes, over-dispersion means that variation exceeds what a basic count model expects and may require a model that allows extra variation. That issue is not the central one for the posted mean summaries, and the registry record does not report the analysis model or diagnostics used.

What analysis details are unavailable?

The registry record identifies the primary analysis population as the FAS, including participants who consumed at least 1 dose of randomized study medication. It does not report the full statistical analysis plan, the original sample-size justification, detailed missing-data methods, or planned interim analyses. It also does not report results for most of the listed follow-up outcomes. Without those details, readers cannot assess how the planned analyses handled repeated measurements, missing values, or multiple comparisons, or whether the posted baseline results are part of a broader analysis strategy.

11. Confidence Intervals and Precision

The primary baseline mean difference is 0.79 units, with a confidence interval from -2.164 to 3.744. This interval is the main posted summary of precision for the contrast. Its lower and upper limits lie on opposite sides of zero, so the data are compatible with a baseline contrast favoring either group in numerical score terms, as well as with no difference. Since higher ADAS-Cog 70 scores indicate worse cognition, the direction of any numerical contrast must also be interpreted in light of the scale's scoring direction.

The interval's width reflects uncertainty in the estimated mean difference. It is not a range for individual scores, and it does not mean that all plausible values are equally likely. The standard deviations, 10.42 for placebo and 10.39 for PF-04447943, describe variation among participants, whereas the confidence interval describes uncertainty around the group contrast.

The baseline NPI mean difference is -1.49, with a confidence interval from -4.914 to 1.934. That interval also includes zero. These intervals do not establish equivalence; an equivalence conclusion would require an appropriate design, prespecified equivalence bounds, and analysis, which are not reported in the registry record.

Precision should be considered alongside endpoint timing. A confidence interval can quantify uncertainty about the baseline contrast, but it cannot turn that contrast into an estimate of change during treatment. The record does not report confidence intervals for most listed follow-up outcomes, so their precision cannot be assessed from the posted results.

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 sample-size table and curve are planning illustrations based on the observed effect in the posted comparison. They do not describe the number of participants actually enrolled; actual enrollment was 191. The figures are participants per arm, not total participants or total events. This is a continuous-outcome illustration using equal allocation between two arms and a normal approximation.

The listed assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and normal approximation. The effect to detect is expressed as a stated fraction of the observed effect. The trial's own original design assumptions and sample-size justification are not reported in the registry record.

The 80% power column gives the number of participants per arm that the planning calculation would require for an 80% chance of detecting the specified effect under its assumptions, if that effect were present and the statistical model applied. The 90% power column gives the corresponding per-arm planning figure for a 90% chance. These are conditional planning probabilities, not guarantees about any particular study.

Each row uses a different portion of the observed effect: the full observed effect, 75% of the observed effect, or 50% of the observed effect. The table shows that smaller effects need more participants per arm to be detected with the same target power. The curve offers a visual way to see this relationship across the displayed effect assumptions; it does not add evidence about the trial's actual treatment effect.

This is not observed power. It is a planning illustration calculated from the posted effect estimate, not a retrospective measure of whether the completed study was capable of producing a particular result. Because the observed primary comparison is a baseline difference rather than a treatment-effect estimate, the illustration should not be read as a sample-size recommendation for a follow-up treatment question.

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 = 0.790; pooled standard deviation σ = 10.406.
2. n = 2 × (1.96 + 0.842)² × 10.406² / 0.790² = 2,723.5, rounded up to 2,724 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 <- 0.789999999999999                      # observed difference in means
sigma <- 10.4057250725605                     # 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(930059)
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 = 2,724 (80% power), 3,647 (90% power)
 75% of observed effect: n per arm = 4,842 (80% power), 6,482 (90% power)
 50% of observed effect: n per arm = 10,895 (80% power), 14,585 (90% power)
power.t.test: 2724.5 
Simulated power at n = 2,724 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.

14. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when a study examines several outcomes, time points, groups, or analyses. If each test is considered separately without adjustment, the overall chance of at least one apparently notable result can be greater than the chance attached to an individual test. Appropriate interpretation therefore depends on knowing which outcome was primary, how secondary outcomes were prioritized, and whether the analysis plan controlled the error rate or used another prespecified strategy.

This study lists multiple secondary outcomes, including cognitive scores at several follow-up times, clinical global improvement, and NPI scores and changes. The registry record does not report a multiplicity adjustment or a hierarchy for interpreting these outcomes. The available NPI baseline comparison has an unadjusted p-value, and the primary baseline comparison also has an unadjusted p-value. These values should not be treated as if they account for the broader set of outcomes and time points.

Missing-data handling is another important part of interpreting repeated clinical measurements. Participants may not have measurements available at every visit, and different analysis methods can make different assumptions about unavailable values. The registry record defines the FAS as participants who consumed at least 1 dose of randomized study medication, but does not report how missing outcome measurements were handled, how many participants contributed at each follow-up, or whether sensitivity analyses were conducted.

The record also does not report the full statistical analysis plan, the trial's original sample-size justification, or interim-analysis plans. These omissions do not show that a method was or was not used; they mean it is not reported in the registry record. For the same reason, readers cannot verify from this record alone whether the posted analyses followed a prespecified plan or how the study addressed the listed follow-up outcomes.

These gaps matter especially because the posted primary measure is baseline, while the stated study purpose concerns effects over time. The baseline summaries do not substitute for the missing follow-up analyses. A complete interpretation of multiplicity, missingness, and analysis populations would require details and results that are not reported in the registry record.

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: intestinal obstruction (1.1%), fall (1.1%), hip fracture (1.1%), cerebral haematoma (1.1%), cardiac arrest (1.0%).

The registry record reports serious adverse events affecting 4 of 100 participants in the placebo group and 3 of 91 participants in the PF-04447943 group. Other adverse events affected 56 of 100 participants receiving placebo and 57 of 91 receiving PF-04447943. These are descriptive participant counts for the reported categories; the record does not provide an adjusted comparison of safety outcomes.

The displayed serious-event terms include intestinal obstruction, fall, hip fracture, cerebral haematoma, cardiac arrest, and chronic hepatic failure, each with the percentage shown in the posted results. A listed term is not, by itself, evidence that the study drug caused the event. Nor do the group summaries alone establish that risks were the same or different. Interpretation can depend on event definitions, exposure, timing, severity, and how events were collected, details not reported.

Deaths are not reported in the registry record. The adverse-event figures should therefore be read as the safety information posted in this record, not as a complete account of every safety question or a basis for medical decisions.

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

The posted data establish that the registry record contains baseline ADAS-Cog 70 and NPI summaries for the placebo and PF-04447943 groups, along with descriptive adverse-event counts. The baseline cognitive means are 21.87 for placebo and 22.66 for PF-04447943, and the posted mean difference is 0.79. The baseline NPI means are 12.16 and 10.67, with a posted mean difference of -1.49. These are descriptions of the recorded groups at baseline.

The results do not establish that PF-04447943 changed cognition or neuropsychiatric symptoms during follow-up. The primary outcome shown is an ADAS-Cog 70 score at Baseline (Day 1), not a change score or a post-treatment comparison. The registry record lists follow-up outcomes, but most of their results are not reported in the registry record. A treatment-effect conclusion would require relevant follow-up estimates and an analysis that addresses the study design and data.

The confidence intervals around the posted baseline differences include zero, indicating uncertainty about the size and direction of the baseline contrasts. They do not demonstrate that the groups are equivalent, and they do not provide evidence about a treatment effect. A non-significant p-value is also not proof of no effect; it summarizes the data under a particular statistical model and null hypothesis.

The safety summaries show reported serious and other adverse events in both groups, but they do not establish causality or settle comparative safety. The record does not report deaths or full details needed to interpret event timing, exposure, and attribution.

Overall, this record supports a limited statistical reading of baseline characteristics and posted safety summaries. It does not provide enough reported follow-up results to answer whether the intervention altered the outcomes the study set out to evaluate. That limitation is about the contents of the registry record, not a claim about unreported results elsewhere.

17. Why This Trial Matters Statistically

This trial illustrates why identifying the exact endpoint and its timing is essential before interpreting a statistical comparison. The study's stated purpose concerns effects on symptoms, yet the primary outcome displayed in the results is ADAS-Cog 70 at baseline. A precise comparison of starting scores answers a different question from an analysis of change during treatment. Statistical significance, confidence intervals, and sample-size calculations cannot compensate for a mismatch between the reported endpoint and the question of interest.

The record also shows why a registry outcome list and a set of posted results should be distinguished. Several follow-up cognitive and behavioral outcomes are listed, but results for most are not reported in the registry record. Without estimates and uncertainty intervals for those measures, readers cannot evaluate the full pattern across the planned outcome domains.

Finally, the sample-size illustration demonstrates how planning requirements depend on the effect size one wants to detect: smaller effects require more participants under the stated assumptions. Yet the illustration uses an observed baseline contrast, not an observed treatment effect, so it cannot be used to claim that the completed study had a particular probability of finding a treatment benefit. The trial is therefore a useful example of reading statistical summaries in context: endpoint timing, uncertainty, reporting completeness, and design assumptions all shape what a result can mean.

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 most important limitation for interpreting efficacy is that the primary outcome posted is a baseline score. The record does not report a primary follow-up contrast or change-from-baseline result. This prevents the displayed primary comparison from addressing whether cognitive scores changed differently between groups during the study.

Although the registry record lists several follow-up secondary outcomes, results for most are not reported in the registry record. This limits assessment of symptom trajectories and consistency across cognitive, behavioral, and global measures. It also limits the ability to judge the precision of those outcomes or relate them to the stated study purpose.

The record does not report the original sample-size justification, detailed statistical analysis plan, missing-data methods, or interim-analysis plans. It also does not report whether a multiplicity procedure was used. These are limitations of the information available in the registry record; they do not establish that the study lacked such procedures.

The safety information is also limited to posted participant counts and listed serious-event terms. Death counts are not reported in the registry record, and the available summaries do not provide enough detail to determine causality or fully compare safety across groups. Finally, the baseline group summaries and sample-size planning illustration should not be taken as substitutes for the unavailable follow-up treatment-effect results.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: A Study Of PF-04447943 Compared To Placebo In Subjects With Mild To Moderate Alzheimer's Disease — NCT00930059.
• Registry record last updated 2020-11-19; status last verified 2020-10.
• Results first posted 2020-11-19.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

22. Record Summary

This completed Phase 2 randomized, quadruple-masked, parallel-group study compared PF-04447943 with placebo in subjects with mild to moderate Alzheimer's disease. Enrollment was 191. The primary outcome posted is ADAS-Cog 70 score at Baseline (Day 1), with baseline means of 21.87 for placebo and 22.66 for PF-04447943. The mean difference is 0.79, with a confidence interval from -2.164 to 3.744 and an unadjusted p-value of 0.6.

The posted results also include baseline NPI summaries and adverse-event counts. Most listed follow-up outcome results are not reported in the registry record. Consequently, the available primary comparison describes baseline scores and does not estimate change during treatment.