Memory and Insulin in Early Alzheimer's Disease: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Memory and Insulin in Early Alzheimer's Disease (NCT00581867), a randomized phase 1, phase 2 study. The primary outcome is fMRI Measure of Hippocampal Activation.

2. Clinical Question

The study description also says it sought to determine which parts of the brain are involved in insulin-related memory improvement. The posted primary endpoint is a brain-imaging measure of hippocampal activation, rather than a direct memory score. Global cognition is listed as a secondary outcome. The registry record does not report the trial's full hypothesis, the planned direction of the primary comparison, or how the study related the imaging measure to memory performance.

3. Trial Design

This was a randomized, double-masked, phase 1/phase 2 crossover trial with an enrollment of 31. Each participant was described as completing both intervention arms: intranasal insulin aspart and intranasal saline placebo. The order of saline and insulin was counterbalanced. The arm descriptions specify a 48-hour washout period between the two interventions and administration approximately 30 minutes before functional MRI scanning.

A crossover design compares interventions within participants, which can reduce the influence of stable differences between people. It also requires attention to the order of treatments, possible carryover effects, and the analysis of paired observations. Counterbalancing the order distributes treatment sequences across participants; a washout period is intended to separate intervention periods. These are general design considerations, not evidence that carryover was assessed or that the washout was sufficient for this study. The registry record does not report the method used to analyze the crossover structure or any assessment of period or sequence effects.

The primary outcome is a mean fMRI measure, reported for 29 participants in each arm. The registry lists an overall enrollment of 31, but participant flow and the reason for the difference between enrollment and the number reported for each arm are not reported in the registry record. The registry identifies Global Cognition as a secondary outcome, measured at 90 mins. The primary endpoint timeframe is 30 minutes after intervention administration.

The trial status is completed. The registry record does not report the original sample-size justification, the planned statistical analysis, or whether any interim analyses were planned or conducted.

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.

First Intervention

Second Intervention

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: 31 total participants were enrolled in the study. 29 completed the study and had, in a counterbalanced order, one-dose of intranasal insulin followed by fMRI and over 48 hours later a dose of intranasal saline followed by fMRI.

6. Primary Endpoint

The registry lists the primary outcome as fMRI Measure of Hippocampal Activation, measured over 30 minutes After Intervention Administration.

The primary endpoint is the percentage of active voxels among the total hippocampal volume of interest, measured using functional MRI 30 minutes after intervention administration. The registry reports this as a continuous measure and presents a mean and standard deviation for each arm. In broad terms, active voxels are imaging locations meeting the study's criterion for activation; the exact imaging acquisition, processing, and activation-threshold procedures are not reported in the registry record.

The hippocampus is a brain region relevant to memory, so measuring its activation provides an imaging-based way to examine brain response during the study. This is a physiological or imaging outcome, not a direct measure of a participant's memory ability. The registry's brief description refers to investigating brain regions involved in insulin-related memory improvement, but the posted primary endpoint itself is hippocampal activation. Global Cognition appears separately as a secondary outcome.

Using a continuous measure allows the result to be expressed as an average and the spread of participant measurements. Those summaries describe the posted arm results, but they do not by themselves establish how the endpoint was defined for each participant or whether the measure was analyzed using the paired structure of the crossover. The detailed endpoint protocol and analysis method are not reported in the registry record.

7. Results Posted in the Registry

• Intranasal Insulin Aspart: mean 36.40 versus 41.30 on control; difference -4.90 (95% CI -21.61 to 11.81), standardized effect d = -0.15; z-test p = 0.565.

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.

Global Cognition

• 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:
  • Intranasal Insulin Aspart: mean -0.05 versus -0.02 on control; difference -0.03 (95% CI -0.61 to 0.55), standardized effect d = -0.03; z-test p = 0.919.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Intranasal Insulin Aspart and the control. The interval gives the range of effects that remain compatible with the data.

Global Cognition is the one secondary outcome with posted results. Its mean is reported in z-scores at 90 mins, with summaries for both intervention arms and an unadjusted comparison. Broadly, the arm means are close to one another, and the unadjusted comparison is uncertain. The primary endpoint is also an unadjusted comparison with a confidence interval spanning values on both sides of zero, so the two outcomes do not provide a clear, consistent separation between the interventions in the posted summaries.

The secondary outcome does not directly replicate the primary measure: Global Cognition is a cognition measure, while the primary endpoint is hippocampal activation on functional MRI. A similarity in the broad pattern of estimates does not establish that the two outcomes measure the same process or validate one another. The registry record does not report how the secondary outcome was defined, how it was linked to the primary question, or whether a formal relationship between the outcomes was evaluated.

Secondary results generally carry less confirmatory weight when the analysis plan, multiplicity strategy, and inferential methods are not reported. When multiple outcomes or comparisons are examined, the chance of obtaining at least one apparently notable result can increase. Adjustments or a prespecified hierarchy can address this issue, but whether either was planned or used is not reported in the registry record. The posted secondary comparison is unadjusted, and the registry reports no trial analyses. It should therefore be treated as descriptive and exploratory rather than as independent confirmation of the primary result.

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

• Global Cognition

9. Statistical Interpretation

10. Statistical Methods Explained

Several statistical ideas help interpret the posted results. The primary endpoint is continuous, so the registry reports an arithmetic mean and standard deviation for each intervention arm. The mean describes the center of the measurements; the standard deviation describes how spread out individual measurements are around that center. A group mean is not an individual participant's result, and the standard deviation is not a confidence interval for the mean.

What does a mean difference describe?

The primary comparison is expressed as insulin aspart minus placebo. The posted mean difference is -4.9 percentage points in the endpoint's units. A negative value means the posted mean was lower in the insulin aspart arm than in the placebo arm. This direction is a description of the reported means, not a statement about clinical benefit or harm. The secondary mean difference is likewise expressed as insulin aspart minus placebo.

The reported comparisons are unadjusted recalculations from posted summary statistics. They are not listed as analyses performed by the trial. In particular, the registry reports no analyses. The calculations do not substitute for the trial's original statistical analysis, which is not reported in the registry record.

Why does the crossover design matter?

In a crossover trial, a participant receives both interventions, so the two measurements from that participant are related. A paired analysis uses that within-participant relationship. A comparison that treats the two sets of arm summaries as if they came from unrelated groups may not use all the information in the design and may produce a different estimate of uncertainty. The registry's arm summaries do not provide the within-participant differences, and the registry record does not report a paired analysis or another model for the crossover.

Order and period can also matter. Participants received the interventions in counterbalanced order, with a 48-hour washout period described between arms. These design features address sequence and separation in general, but they do not establish that carryover or period effects were absent. Whether such effects were evaluated is not reported in the registry record.

What does a confidence interval convey?

A confidence interval expresses uncertainty around an estimated difference under the assumptions of the method used to calculate it. A wider interval indicates less precision: a broader range of effects is compatible with the observed summaries and calculation. When an interval includes zero, the data do not clearly distinguish a positive difference from a negative difference at the corresponding confidence level. This does not prove that the true difference is zero. The primary interval extends from -21.61 to 11.81, so it includes values on both sides of zero. The interval for Global Cognition also includes zero.

How should a p-value be read?

A p-value measures how unusual the observed data, or more extreme data, would be under a specified null hypothesis and statistical model. It is not a measure of effect size or precision, and it is not the probability that a result occurred by chance. The primary unadjusted p-value is 0.565; the secondary unadjusted p-value is 0.919. Since the registry does not report the trial's analysis model, these values should be understood as attached to the posted unadjusted recalculations, not as documented results from a reported trial analysis.

Statistical significance thresholds and analysis plans need to be specified in advance to support confirmatory interpretation. The registry does not report the planned significance threshold for the trial or its full analysis plan. A planning assumption of two-sided significance level 0.05 is stated for the sample-size illustration; that illustration is not evidence that this was the trial's original design assumption.

Why can variation affect precision?

When measurements vary substantially among participants, a mean difference can be estimated less precisely unless the sample provides enough information to distinguish the intervention contrast from that variability. This is a general statistical principle. The reported standard deviations for the primary endpoint are 31.4 for insulin aspart and 33.5 for placebo, while the reported means are 36.4 and 41.3. The broad primary confidence interval reflects substantial uncertainty in the summary-based comparison. In a crossover study, within-participant correlation can also affect precision; that correlation is not reported in the registry record.

Over-dispersion is a general concern for outcomes whose observed variability exceeds what a chosen model expects. It can make model-based standard errors too small if ignored. The posted primary endpoint is continuous and summarized with standard deviations, but the registry does not report an over-dispersion assessment or a model that would allow such an assessment. The central methodological limitation is not evidence of over-dispersion; it is the absence of a reported analysis method and participant-level crossover information.

The effect-size and sample-size illustrations use observed differences and stated assumptions. Such calculations are sensitive to the estimated effect: a small observed effect can imply very large planning totals. Those totals are illustrations, not a reconstruction of the original trial's design. The original design assumptions and sample-size justification are not reported in the registry record.

11. Confidence Intervals and Precision

The primary mean difference is -4.9, with a 95% confidence interval from -21.61 to 11.81. This interval is wide relative to the point estimate and spans values below and above zero. It therefore indicates limited precision in the unadjusted summary-based comparison: the posted data do not narrowly locate the difference or its direction.

The interval should be interpreted in the endpoint's units, percentage of active voxels, and as uncertainty around a mean difference. It is not a range containing 95% of individual participant measurements. The arm standard deviations, 31.4 for insulin aspart and 33.5 for placebo, describe participant-level spread, whereas the confidence interval describes uncertainty in the estimated contrast.

Global Cognition has a posted mean difference of -0.03 and an unadjusted confidence interval that includes zero. The registry reports the endpoint in z-scores. The interval's inclusion of zero indicates that the posted summary calculation does not precisely distinguish a difference in either direction. Its exact confidence limits are not repeated.

These intervals come from unadjusted recalculations based on posted summaries, and the registry reports no trial analyses. In a crossover design, precision can depend on the relationship between each participant's results under the two interventions. That within-participant information and the analysis method are not reported in the registry record. Consequently, the displayed intervals should not be treated as a substitute for a reported paired analysis. The estimates are compatible with a range of differences, and the results are not precise enough to establish equivalence or absence of an effect.

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 figures are planning illustrations based on the observed effect, not the trial's original design calculation. For these continuous outcomes, the figures are participants per arm, not total events. The stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Each row considers an effect to detect equal to the stated fraction of the observed effect. The trial's own sample-size justification and original planning assumptions are not reported in the registry record.

The 80% power column gives the illustrated number of participants per arm for a design with an 80% chance of detecting the specified effect under the stated assumptions, if that effect and model assumptions were correct. The 90% power column gives the corresponding planning number for a 90% chance. Power is a design property under assumptions, not a measure of whether a particular result is true.

The curve visualizes how the planned participant count changes with the effect targeted. Planning for 75% of the observed effect requires more participants per arm than planning for 100% of the observed effect; planning for 50% requires still more. Smaller effects are harder to distinguish from variability, so they require more participants for the same target power. This is not observed power. The figures are illustrations based on the posted effects and specified assumptions, not evidence that the completed trial had those design characteristics.

The primary and secondary outcomes have different observed effects and variability, so their planning figures differ substantially. Very large illustrated counts can result when the observed difference is small relative to the spread of measurements. They should not be read as recommended enrollment targets without a prospective design and assumptions appropriate to the intended study.

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 = -4.900; pooled standard deviation σ = 32.467.
2. n = 2 × (1.96 + 0.842)² × 32.467² / -4.900² = 689.2, rounded up to 690 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 <- -4.9                      # observed difference in means
sigma <- 32.4669832291206                     # 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(581867)
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 = 690 (80% power), 923 (90% power)
 75% of observed effect: n per arm = 1,226 (80% power), 1,641 (90% power)
 50% of observed effect: n per arm = 2,757 (80% power), 3,691 (90% power)
power.t.test: 690.1 
Simulated power at n = 690 per arm: 0.798 (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

The registry lists a primary fMRI outcome and Global Cognition as a secondary outcome. It reports two outcome measures and one posted secondary endpoint, but no analyses. The number of outcome measures alone does not reveal how many statistical tests were planned or performed, and the registry record does not specify a testing hierarchy or a multiplicity adjustment.

Multiplicity arises when investigators evaluate several outcomes, time points, treatment contrasts, or subgroups. Without a prespecified strategy, interpreting each unadjusted p-value as if it were the only test can overstate the strength of evidence. Whether multiplicity correction was planned or applied is not reported in the registry record. The posted p-values are unadjusted recalculations, so they should not be treated as adjusted confirmatory evidence.

Missing-data handling is also not reported in the registry record. The overall enrollment is 31, while the posted primary and secondary arm summaries each include 29 participants per arm. The record does not explain participant flow, whether all participants completed both periods, why the analysis summaries include those counts, or how any unavailable measurements were handled. It also does not report whether the analyses used all randomized participants or a different analysis population.

These omissions limit evaluation of bias and precision. In a crossover study, missing one period can affect whether a participant contributes to a paired comparison, and the approach to incomplete pairs can influence the analysis. Whether this occurred and how it was addressed are not reported. The registry record also does not report prespecified analysis populations, sensitivity analyses for missing data, or the trial's complete statistical analysis plan.

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

Serious events with the highest arm-level rate: death (3.4%).

The posted serious adverse event summaries list 29 participants at risk in each arm. The insulin aspart arm has 0 serious adverse events affected; the placebo arm has 1. The registry also lists Death as a serious-event term at 3.4%. The death counts themselves are not reported in the registry record, so the term listing and the arm-level serious-event count cannot be reconciled from the available record alone.

No other adverse events affected are listed in either arm in the posted summary. These figures are descriptive counts from the registry, not a comparative safety analysis. With the small reported at-risk groups, the summaries provide limited information about uncommon events or differences between interventions. The record does not report event-level detail, timing, relatedness assessments, or a statistical comparison of adverse events. These safety figures do not establish that an event was caused by an intervention, nor do zero counts establish that no risk exists.

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

The posted results describe a primary hippocampal fMRI mean that was lower in the insulin aspart arm than in the placebo arm, with an unadjusted mean difference of -4.9. The confidence interval is broad and includes zero, and the unadjusted p-value is 0.565. These summaries do not establish a clear difference between the arm means, but neither do they establish equivalence or prove that the true difference is zero.

The endpoint is an imaging measure, not a direct memory outcome. Although the study description concerns brain regions involved in insulin-related memory improvement, the primary result alone does not establish a change in memory or explain a mechanism. Global Cognition is listed as a secondary outcome, but its posted summary is also unadjusted and the registry reports no trial analyses.

Interpretation is further limited because this is a crossover study and the registry does not report whether the comparison accounted for within-participant pairing, treatment order, period, or carryover. Participant-level results and details of missing-data handling are also not reported. The comparisons are recalculations from posted summary statistics, not a documented trial analysis.

The appropriate conclusion is therefore descriptive: the posted arm means and unadjusted calculations leave substantial uncertainty about the size and direction of differences. They do not support a definitive claim about treatment effect, clinical benefit, or absence of effect. The registry record alone cannot resolve those questions.

17. Why This Trial Matters Statistically

This trial illustrates why design and analysis details matter alongside point estimates. A crossover design allows the same participants to receive both interventions, which can make within-person comparisons informative. But valid interpretation depends on accounting for the paired measurements and considering order, period, and possible carryover. The registry describes counterbalancing and a washout period, yet does not report the model used to analyze the crossover.

The posted summaries also show why confidence intervals are essential. A mean difference alone gives a direction and size for the observed contrast, but the primary interval spans values on both sides of zero. The evidence is therefore not precise enough to establish a narrow range of plausible differences. A non-significant unadjusted p-value is not proof of no effect.

Finally, the sample-size illustrations demonstrate how planning requirements depend on the effect a future study aims to detect. Smaller target effects require more participants for the same planning power under the stated assumptions. Those calculations are not observed power and do not reveal the trial's original planning rationale, which is not reported in the registry record.

Together, these features make the record useful for statistical reading: it distinguishes descriptive arm summaries from inferential conclusions and highlights the importance of prespecified analyses, transparent participant accounting, and design-appropriate methods.

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 main limitation is that the registry reports no trial analyses. The numerical comparisons are unadjusted recalculations from arm summaries, and the record does not establish whether they reflect a paired crossover analysis. Within-participant results, period effects, sequence effects, and carryover assessments are not reported.

The record lists enrollment of 31 but reports 29 participants in each arm for the primary and secondary outcomes. It does not describe participant flow, reasons for unavailable outcome data, or how incomplete data were handled. The clinical-group composition is also unclear: the study description mentions early Alzheimer's disease and cognitively normal older adults, but the number in each group and group-specific results are not reported in the registry record.

The endpoint description does not provide detailed fMRI acquisition, processing, or threshold methods. The registry also does not report a full statistical analysis plan, original sample-size justification, analysis populations, multiplicity strategy, or interim-analysis plan. These missing details restrict assessment of whether the posted outcomes and comparisons were prespecified and how robust the estimates are.

The primary confidence interval is broad, and the sample is limited in size as reported for each arm. The posted summaries therefore offer limited precision. The serious adverse event summary is similarly sparse, with event-level details and death counts not reported in the registry record. These constraints mean that the record supports cautious description of the posted results, not definitive inference about treatment effects or safety.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Memory and Insulin in Early Alzheimer's Disease — NCT00581867.
• Registry record last updated 2015-07-14; status last verified 2015-07.
• Results first posted 2015-07-14.
• 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

The registry record describes a completed randomized, double-masked, phase 1/phase 2 crossover study with enrollment of 31. Participants received intranasal insulin aspart and intranasal saline placebo in counterbalanced order, with a 48-hour washout period described between arms.

The primary outcome was the percentage of active voxels in the hippocampal volume of interest, measured 30 minutes after intervention administration. The posted mean difference was -4.9, with a 95% confidence interval from -21.61 to 11.81 and an unadjusted p-value of 0.565. Global Cognition was listed as a secondary outcome. The registry reports no trial analyses; the comparisons are unadjusted recalculations from posted summaries.

Details of the original analysis plan, sample-size justification, participant flow, missing-data handling, and crossover-specific analysis are not reported in the registry record.