Enhancement of Cerebral Vasoreactivity and Cognition by Intranasal Insulin in Type 2 Diabetes: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Enhancement of Cerebral Vasoreactivity and Cognition by Intranasal Insulin in Type 2 Diabetes (NCT01206322), a randomized phase 2 study. The primary outcome is Cognitive Outcome: Brief Visuospatial Spatial Memory Test -Total Recall (Unit T Score).

2. Clinical Question

The trial title and stated hypotheses concern possible acute effects on cerebral blood flow, vasoreactivity to a CO2 challenge, and cognitive functioning. The cognitive outcome summarized in the registry is visuospatial memory total recall; the broader hypotheses also mention attention, memory, and executive function. The record describes regional perfusion and vasoreactivity assessment using 3-D continuous arterial spin labeling MRI at 3 Tesla. These aims frame the research question, but they should not be confused with the specific numerical cognitive summaries available in the posted results.

3. Trial Design

The registry identifies this as a phase 2, randomized, double-blind, crossover study, with a completed status. The study is sponsored by Beth Israel Deaconess Medical Center. The registry lists a start date of 2010-05 and primary completion date of 2013-04.

In a crossover design, participants receive different study conditions at different times, so each participant can contribute observations under more than one condition. The outcome description states that a single dose of intranasal insulin or placebo was administered on day 2 and a single dose of insulin or placebo on day 3, with the order randomized. Randomization is intended to reduce systematic differences in treatment order, while blinding is intended to limit the influence of knowledge about the assigned condition on participants’ experience or outcome assessment. The registry does not report how blinding was maintained or assessed.

The brief summary describes a proposed pilot involving participants with type 2 diabetes and healthy controls. For the posted cognitive outcome, the registry lists separate placebo and insulin groups within both the diabetes and control categories. Those arm summaries are not a complete description of the within-person data structure. The record does not report the sequence allocation details, the handling of period or order effects, or whether analyses modeled the crossover pairing. These details matter because measurements from the same participant across conditions are related rather than independent.

The registry records enrollment as 30. The brief summary proposes 30 participants with type 2 diabetes and 30 healthy controls, while the posted outcome arm counts are 15 and 15 for the diabetes group and 14 and 14 for the control group. The record does not explain these differences, so the figures should be read in their stated contexts rather than reconciled by assumption. The original sample-size justification and detailed protocol are not reported in the registry record.

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: Thirty consented subjects, 15 type 2 diabetic and 14 controls completed the protocol. One control was excluded after starting the study. Screened participants that did not start study: 27 were excluded after screening visit, 7 withdrew consent. Enrollment Table list the characteristics of 29 participants who completed the protocol.

Overall Study

Reasons for not completing this period:

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: Twenty nine consented subjects, 15 type 2 diabetic and 14 controls completed the protocol.

6. Primary Endpoint

The registry lists the primary outcome as Cognitive Outcome: Brief Visuospatial Spatial Memory Test -Total Recall (Unit T Score), measured over Acute changes within 2 hours.

The primary cognitive endpoint is total recall on the Brief Visuospatial Spatial Memory Test, expressed as a T-score. It is intended to quantify performance on a visuospatial memory task, making it a specific measure relevant to the study’s question about acute cognitive effects. It does not, by itself, represent every aspect of cognition or establish effects on attention, executive function, or other memory domains.

The registry specifies that the outcome reflects acute changes within 2 hours after administration of insulin or placebo. The accompanying description places this measure within a broader assessment that also included regional perfusion and vasoreactivity to a CO2 challenge. Those physiological outcomes are part of the study’s stated aims, but the cognitive endpoint is the measure for which the listed group means and standard deviations are available.

A mean T-score gives the average observed result for a group, while the standard deviation describes how spread out individual scores are around that average. The posted summaries report a mean of 39.46 with standard deviation 8.86 for the diabetes placebo group, and 41.23 with standard deviation 9.85 for the diabetes insulin group. For controls, the placebo mean is 43.1 with standard deviation 17.29, and the insulin mean is 50.9 with standard deviation 9.66. The registry record does not describe the T-score reference scale or specify the interpretation of a particular score value. The measure is therefore best read as a standardized cognitive outcome, with the reported group summaries describing performance rather than a diagnosis or a complete account of cognitive functioning.

7. Results Posted in the Registry

• Diabetes Group: Insulin: mean 41.23 versus 39.46 on control; difference 1.77 (95% CI -4.93 to 8.47), standardized effect d = 0.19; z-test p = 0.605.
• Control Group: Placebo: mean 43.10 versus 39.46 on control; difference 3.64 (95% CI -6.47 to 13.75), standardized effect d = 0.27; z-test p = 0.480.
• Control Group: Insulin: mean 50.90 versus 39.46 on control; difference 11.44 (95% CI 4.68 to 18.20), standardized effect d = 1.24; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

What do the group summaries describe?

The posted results give a mean and standard deviation for each combination of group and treatment. A mean summarizes the average T-score; the standard deviation describes the dispersion of individual scores around that average. The mean does not show whether observations are clustered or whether individual participants changed between conditions. The standard deviation is not a confidence interval and does not describe the precision of the mean by itself.

The registry lists 15 participants for each diabetes treatment group and 14 for each control treatment group. Because the trial is described as a crossover, the treatment observations may come from the same people under different conditions. The summaries do not show the within-person correlation or the distribution of participant-level changes. Without that information, the mean and standard deviation alone cannot reconstruct a paired crossover analysis.

What is a mean difference?

A mean difference expresses the distance between two group means in the outcome’s units. The recalculated comparisons use the posted means and standard deviations to estimate an unadjusted difference and its confidence interval. For the diabetes insulin-versus-placebo comparison, the estimate is 1.77. For control placebo compared with diabetes placebo, the estimate is 3.64. For control insulin compared with diabetes placebo, the estimate is 11.44. The meaning of each estimate depends on which groups are being contrasted; they are not interchangeable treatment-effect estimates.

In general, a confidence interval gives a range of values compatible with the observed estimate under the statistical calculation and its assumptions. A wider interval indicates less precision. An interval crossing zero includes no mean difference among the values compatible with that calculation. These intervals are not guarantees that the true value lies within them, and they do not correct for design features that were not included in the calculation.

How should the p-values be understood?

In general, a p-value describes how unusual data at least as discrepant as those observed would be under a specified null hypothesis and statistical model. It is not the probability that the null hypothesis is true, nor does it directly measure the size or practical importance of a difference. The recalculated p-values are 0.605 for the diabetes insulin-versus-placebo comparison, 0.48 for control placebo compared with diabetes placebo, and <0.001 for control insulin compared with diabetes placebo.

These are unadjusted values calculated from posted summary statistics. The registry lists zero analyses, and it does not report the sponsor’s statistical analysis plan or a model that incorporates the crossover structure. The p-values therefore describe those unadjusted calculations, not a documented full trial analysis. A small p-value for a contrast does not make that contrast equivalent to the intended randomized comparison.

Why does the crossover structure matter?

In general, crossover trials compare observations from the same participant under different conditions. Accounting for that pairing can improve precision when measurements from a participant are more similar to each other than measurements from different participants. It also requires attention to treatment sequence, period effects, and possible carryover from one condition to another. Those considerations are specific to the design; the registry does not report how they were handled analytically.

The recalculated comparisons use posted group means and standard deviations, not the participant-level changes between insulin and placebo. They therefore cannot represent a paired treatment-effect analysis unless additional assumptions and data support that interpretation. The registry record does not report the within-person covariance, a sequence-specific analysis, or an assessment of period or carryover effects.

What does over-dispersion mean in general?

For continuous outcomes, substantial variation among participants can make a group mean less precise. More generally, over-dispersion refers to observed variability greater than a model assumes. A model that understates variability can produce intervals that are too narrow and p-values that appear more decisive than warranted. The posted standard deviations differ across the groups, including the control placebo standard deviation of 17.29 and the control insulin standard deviation of 9.66. These summaries indicate that the observed spread is not identical across all groups, but the registry does not report a variance model or a formal assessment of its assumptions.

What is and is not established by these calculations?

The calculations give useful descriptive contrasts and quantify their uncertainty under the stated simple comparison approach. They do not substitute for an analysis accounting for repeated observations, randomized treatment order, group status, and any prespecified adjustment strategy. The registry does not report the protocol’s primary analysis method, missing-data approach, analysis populations, or multiplicity plan. It also does not report how the two stated hypotheses or the several cognitive domains were handled statistically.

In general, a prespecified analysis plan helps clarify which outcome and contrast are primary, how the design is represented, and how multiple comparisons are interpreted. Because those details are not reported in the registry record, the recalculated estimates should be treated as exploratory descriptions of posted summaries, not as a definitive reconstruction of the study’s intended analysis.

10. Confidence Intervals and Precision

The confidence intervals show how uncertain the unadjusted mean-difference estimates are under the calculations based on the posted summaries. For diabetes insulin compared with diabetes placebo, the estimate is 1.77 and the 95% confidence interval extends from -4.935 to 8.475. For control placebo compared with diabetes placebo, the estimate is 3.64 and the interval extends from -6.466 to 13.75. Both intervals include zero and allow values on either side of zero, so the direction and size of these contrasts are not estimated precisely.

For control insulin compared with diabetes placebo, the estimate is 11.44 and the 95% confidence interval extends from 4.679 to 18.2. This interval does not include zero under the unadjusted calculation, although its width still reflects uncertainty about the magnitude of the difference. The contrast itself compares different group and treatment combinations and is not the within-control insulin-versus-placebo effect.

In general, confidence intervals are more informative than a p-value alone because they communicate both the estimated direction and the range of compatible magnitudes. A wide interval can include effects that would be interpreted differently, even when the point estimate appears clear. Here, the diabetes and placebo group contrasts are especially imprecise, while the control-insulin contrast is estimated more distinctly from zero by this calculation.

The intervals are recalculated from posted summary statistics and do not incorporate a reported crossover model. The registry does not provide the participant-level paired differences needed to assess how accounting for within-person correlation might alter precision. The intervals should therefore be read as uncertainty measures for the unadjusted calculations, not as a complete account of uncertainty under the trial’s design.

11. Planning a Similar Trial: Sample Size

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

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

The 80% power column gives the illustrative number of participants per arm under those assumptions for a design with an 80% chance of detecting the specified effect if that effect is present. The 90% power column gives the corresponding per-arm number for a design with a 90% chance. These are planning concepts; the figures are not observed power.

Rows describing 100%, 75%, and 50% of the observed effect show how the planning requirement changes when the target effect is smaller than the observed estimate. Smaller effects need more participants to detect, all else equal. For example, the diabetes comparison’s illustration rises from 440 participants per arm at 100% of the observed effect to 782 at 75% and 1,759 at 50% for 80% power. The very different planning figures across comparisons reflect their different observed estimates and variability; they should not be treated as reliable forecasts of the sample required for a future study without clinically justified assumptions about the target effect and design.

These calculations use a simple two-arm planning framework. The registry describes a crossover trial, but the listed planning assumptions do not describe a crossover-specific design. The figures therefore serve as illustrations rather than a reconstruction of the trial’s design or a direct recommendation for a future study.

12. How the Sample Size Is Derived

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

Formula

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

Worked calculation for this trial

1. Observed difference Δ = 1.770; pooled standard deviation σ = 9.368.
2. n = 2 × (1.96 + 0.842)² × 9.368² / 1.770² = 439.7, rounded up to 440 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 <- 1.77                      # observed difference in means
sigma <- 9.36808678439733                     # 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(1206322)
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 = 440 (80% power), 589 (90% power)
 75% of observed effect: n per arm = 782 (80% power), 1,047 (90% power)
 50% of observed effect: n per arm = 1,759 (80% power), 2,355 (90% power)
power.t.test: 440.7 
Simulated power at n = 440 per arm: 0.804 (20000 replicates)

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

13. Multiplicity, Missing Data and Analysis Populations

The trial description states hypotheses about both regional perfusion and vasoreactivity and cognitive functioning, with cognitive domains including attention, memory, and executive function. Testing multiple outcomes or contrasts can increase the chance of finding at least one apparently notable result by chance. In general, multiplicity is addressed through a prespecified hierarchy, adjustment method, or clearly defined emphasis on estimation. The registry record does not report a multiplicity plan or whether any adjustment was used.

The registry lists two outcome measures and zero analyses. The numerical summaries addressed in this statistical interpretation concern the primary cognitive outcome. The record does not report posted secondary outcome results, including numerical results for the stated perfusion and vasoreactivity measures. It is therefore not possible to assess how the full family of outcomes was analyzed or whether the reported cognitive result was selected from a broader set of comparisons.

The registry record does not report whether all randomized participants were included in each analysis, how the analysis population was defined, or how participants with incomplete observations were handled. Missing data may matter in a crossover study if a participant does not contribute measurements under both conditions. The record also does not report imputation, complete-case methods, sensitivity analyses, or reasons for any missing measurements.

Other analysis details are not reported in the registry record, including a statistical model for repeated observations, treatment sequence and period terms, carryover assessment, and covariate adjustment. The stated crossover design makes these details relevant to interpreting treatment comparisons. Without a reported analysis plan and participant-level data, the unadjusted recalculations cannot answer how the investigators handled these issues or whether the posted comparisons were prespecified.

14. 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 non-serious adverse event terms

The registry reports serious adverse event counts of zero for the diabetes placebo group, diabetes insulin group, control placebo group, and control insulin group. Each listed group has 15 participants at risk for the serious adverse event summary. No serious adverse event terms are listed. Death counts are not reported in the registry record.

For other adverse events, the registry reports zero affected participants in the diabetes placebo, diabetes insulin, and control placebo groups, and 1 affected participant in the control insulin group. These counts describe the adverse-event fields posted in the record; they do not establish that no other symptoms or events occurred. The registry does not provide event descriptions for the other adverse event in the control insulin group, nor does it report timing, severity, attribution, or whether the event was related to study treatment.

Safety counts are descriptive and should be interpreted in light of the small group counts and the record’s limited detail. A zero count in a posted category is not proof that an event cannot occur. The registry’s adverse event information does not by itself provide a comparative safety analysis or explain how events were collected and classified.

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

The registry describes a randomized, double-blind crossover study intended to examine acute cognitive and cerebral perfusion effects of intranasal insulin in participants with type 2 diabetes and healthy controls. The posted cognitive means are higher under insulin than placebo in both group categories. The unadjusted diabetes insulin-versus-placebo recalculation is 1.77, with a confidence interval that includes zero. The unadjusted comparison of control placebo with diabetes placebo also has an interval that includes zero. The control-insulin versus diabetes-placebo contrast is 11.44, with a confidence interval above zero, but it compares different treatment and group combinations.

These values do not establish that insulin caused a cognitive change. The registry provides no reported analysis that accounts for the crossover pairing, and the unadjusted recalculations do not show participant-level changes between conditions. A group difference between controls and participants with diabetes is also not a treatment effect. The contrast involving control insulin and diabetes placebo cannot determine whether the effect of insulin differs by diabetes status; that question requires a comparison structured to test a treatment-by-group difference.

The cognitive outcome measures visuospatial memory total recall and does not represent all cognitive functioning. Although the study description includes regional perfusion and vasoreactivity, numerical results for those measures are not included in the reported cognitive summaries. The record does not describe the analysis populations, missing-data handling, multiplicity approach, or the detailed statistical plan, limiting interpretation of the posted results as a complete account of the trial.

In general, statistical evidence should be interpreted through the estimate, its uncertainty, the design, and the question the contrast actually addresses. A p-value alone does not establish importance or causality. The record supports a description of the posted group summaries and simple unadjusted contrasts; it does not support a definitive conclusion about treatment efficacy, effects across cognitive domains, or the physiological outcomes.

16. Why This Trial Matters Statistically

This trial illustrates why statistical interpretation depends on design as well as arithmetic. A crossover design can use each participant’s experience under different conditions to make treatment comparisons, but the analysis needs to reflect that within-person structure. Group means and standard deviations are informative descriptions, yet they do not disclose how scores changed within individuals or how strongly observations from the same person are related.

The posted comparisons also demonstrate why the identity of the comparator matters. The insulin-versus-placebo contrast within the diabetes group addresses a different question from a placebo comparison between controls and participants with diabetes. A contrast between control insulin and diabetes placebo combines treatment and group differences, so it cannot isolate either component. The numerical estimate and p-value must be read alongside the exact groups being compared.

Precision matters as much as the point estimate. The diabetes treatment contrast has a confidence interval spanning negative and positive values, while the control-insulin versus diabetes-placebo comparison has an interval above zero. Neither finding answers every stated hypothesis, and the latter contrast is not itself a randomized within-group insulin effect. Distinguishing uncertainty from the research question prevents an apparently notable result from being overinterpreted.

The record also connects observed effects to planning illustrations. Those illustrations show how assumptions about the target effect affect the number of participants per arm, but they are not the trial’s original justification and do not account for every feature of the crossover design. Together, these issues make the study a useful example of reading registry outcomes carefully: distinguish descriptive summaries from modeled results, verify that a contrast matches the scientific question, and treat precision and analysis details as central to interpretation.

17. Limitations

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

The registry enrollment is 30, while the brief summary describes a proposed 30 participants with type 2 diabetes and 30 healthy controls. The posted cognitive outcome lists 15 participants in each diabetes treatment group and 14 in each control treatment group. The record does not explain the difference between the proposed counts, the enrollment figure, and the posted outcome counts, so the participant flow cannot be reconstructed from these entries alone.

The crossover design is central to the study, but the available summaries do not provide participant-level paired changes, within-person correlations, or sequence-specific results. The record does not report whether treatment order, period effects, or carryover were modeled. Consequently, the recalculated contrasts based on group summaries may not reflect the analysis needed to estimate a crossover treatment effect.

The reported primary cognitive outcome is specific to visuospatial memory total recall. It does not cover all cognitive domains named in the hypotheses. The study description also identifies cerebral perfusion and vasoreactivity outcomes, but the numerical summaries are cognitive results. The record lists two outcome measures and zero analyses, and does not provide a full account of secondary results or the analysis population.

Other limitations of the registry record include the absence of a reported original sample-size justification, multiplicity strategy, missing-data method, and detailed statistical plan. The adverse event entries provide counts but little detail about event descriptions, timing, severity, or attribution. These gaps limit evaluation of the study’s complete analysis and safety reporting; they do not establish that unreported procedures or outcomes were absent.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Enhancement of Cerebral Vasoreactivity and Cognition by Intranasal Insulin in Type 2 Diabetes — NCT01206322.
• Registry record last updated 2019-06-26; status last verified 2019-06.
• Results first posted 2014-10-28.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

21. Record Summary

The registry identifies NCT01206322 as a completed phase 2 randomized, double-blind crossover study sponsored by Beth Israel Deaconess Medical Center. Its title concerns intranasal insulin, cerebral vasoreactivity, and cognition in type 2 diabetes. The record gives a start date of 2010-05, primary completion of 2013-04, and enrollment of 30.

The stated study question concerns acute effects of intranasal insulin compared with placebo on cognition and cerebral perfusion or vasoreactivity in participants with type 2 diabetes and healthy controls. The posted primary cognitive outcome is Brief Visuospatial Spatial Memory Test total recall, measured as a T-score within 2 hours. Group means and standard deviations are reported for diabetes and control participants under placebo and insulin.

Unadjusted recalculations from those summaries show a mean difference of 1.77 for diabetes insulin compared with diabetes placebo, 3.64 for control placebo compared with diabetes placebo, and 11.44 for control insulin compared with diabetes placebo. The registry does not report a crossover-specific analysis, detailed statistical plan, or secondary outcome results. The comparisons therefore provide descriptive statistical context, not a complete analysis of the trial’s stated hypotheses.