The Study of Nasal Insulin in the Fight Against Forgetfulness (SNIFF): Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for The Study of Nasal Insulin in the Fight Against Forgetfulness (SNIFF) (NCT01767909), a randomized phase 2, phase 3 study. The primary outcome is Change in Global Measure of Cognition as Measured by the Alzheimer's Disease Assessment Scale-Cognitive 12 (ADAS-Cog12).

2. Clinical Question

3. Trial Design

SNIFF was a randomized, quadruple-masked, parallel-group study registered as a phase 2/phase 3 trial. The registry lists an enrollment of 240 and identifies the conditions as amnestic mild cognitive impairment and Alzheimer’s disease. The sponsor was the University of Southern California. The study is marked completed; the registry lists a start date of 2014-01-08 and a primary completion date of 2018-12-11.

Participants were assigned to intranasal insulin or placebo. The insulin regimen was 20 IU twice daily, totaling 40 IU per day, taken approximately 30 minutes after breakfast and dinner. The placebo group followed a corresponding twice-daily schedule. Both groups were assigned their blinded treatment for 12 months. A 6-month open-label period then followed, during which all participants were to receive insulin. Because treatment assignment no longer defined an insulin-versus-placebo comparison in that later period, results from the open-label phase would answer a different question from the blinded comparison.

The primary outcome was change in global cognition measured with ADAS-Cog12. Secondary outcomes included memory, daily functioning, cognitive deficit, hippocampal and entorhinal atrophy, and cerebrospinal fluid biomarkers. The registry lists six outcome measures. It does not report a detailed account of the randomization procedure, the masking roles, or the allocation ratio beyond the arm descriptions. It also does not report the protocol’s sample-size justification or a detailed statistical analysis plan. Those design details should not be inferred from the trial’s phase label or enrollment figure.

The registry record does not include a reported primary-outcome comparison or inferential analysis. Consequently, the design can be described, but the relative cognitive outcomes of the randomized groups cannot be characterized from the registry record alone.

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: Modified intent-To-Treat \[mITT\] population: All randomized participants in Impel Device with ADAS-Cog12 observed at baseline and at least one follow-up.

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: This represents the final baseline characteristics report for the blinded phase of the INI study. Population: Intent-To-Treat \[ITT\] population: All randomized participants in Impel Device.

6. Primary Endpoint

The registry lists the primary outcome as Change in Global Measure of Cognition as Measured by the Alzheimer's Disease Assessment Scale-Cognitive 12 (ADAS-Cog12), measured over 12 months (blinded phase) followed by 6 months (open label phase).

The primary endpoint was change in global cognition measured by the Alzheimer’s Disease Assessment Scale-Cognitive 12 (ADAS-Cog12). The instrument evaluates memory, attention, reasoning, language, orientation, and praxis. The ADAS-Cog12 version used in this study includes delayed word recall, which assesses episodic memory. The registry describes the outcome as a modeled change score and identifies the summary measure as a least-squares mean with a standard error.

The scale combines performance across several cognitive domains. The original portion ranges from 0, best, to 70, worse, with the number of items not recalled added for a maximum score of 80. A positive change indicates cognitive worsening. This direction matters when interpreting a treatment contrast: the sign must be read against the scale’s stated direction, rather than assuming that a larger score represents improvement.

A global cognitive measure is used to summarize change across multiple domains in one outcome. That can provide a broad view of measured cognition, while also making the result less specific about which domain may account for a change. The registry record identifies the endpoint and its scale, but does not report numerical change estimates or a between-group contrast. It therefore does not establish the size, direction, or precision of an observed difference.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does a least-squares mean represent?

In general, a least-squares mean is a model-based estimate of an average outcome. It can differ from a simple arithmetic mean when the analysis accounts for other variables or the structure of repeated observations. The registry identifies the primary measure as a least-squares mean for a modeled change score, but does not report the model specification, covariates, handling of repeated measurements, or exact contrast. Those details are not reported in the registry record, so the adjustment and interpretation of any model-based estimate cannot be reconstructed from this information.

For a randomized treatment question, the central quantity is generally a comparison between groups, not a change estimate considered in isolation. A treatment contrast expresses how the estimated outcome differs under the assigned strategies, subject to the chosen model and analysis population. For a change score, the direction must be interpreted using the outcome scale: the registry states that a positive ADAS-Cog12 change means cognitive worsening. The record does not give the contrast or its sign, so no direction can be stated.

What does a standard error tell the reader?

In general, a standard error describes the sampling uncertainty of an estimate under the statistical model. A smaller standard error indicates greater precision for that estimate, while a larger one indicates less precision. It is not the same as the spread of individual participant outcomes. The registry lists standard error as the dispersion measure for the least-squares mean, but no numerical standard errors are reported for the primary outcome.

A confidence interval is a range of values compatible with the data and analysis under a specified method. It conveys uncertainty about the estimated effect and is often more informative than a significance test alone. No primary-outcome confidence interval is reported in the registry record. General explanations of confidence intervals cannot substitute for the missing estimate or interval, and cannot show whether the observed effect was large, small, or uncertain.

Why does the outcome scale’s direction matter?

Statistical signs have no clinical meaning without knowing which direction is favorable on the measured scale. ADAS-Cog12 scores range from 0, best, to a maximum of 80, and the registry explicitly states that a positive change indicates worsening. This anchors interpretation of any change estimate. However, a within-group change does not by itself answer whether insulin and placebo differed. That question requires the between-group contrast at the relevant time point, as well as uncertainty around it.

The endpoint is described as a modeled change score. In general, a change score represents movement from a baseline measurement to a later measurement. Interpretation depends on which observations are included, how missing measurements are handled, and how the model accounts for the timing of assessments. The registry defines the primary analysis population by baseline measurement and at least one follow-up, but does not report further missing-data procedures or model details.

What can the planning illustrations establish?

In general, sample-size calculations estimate how much information a future study might need to detect a specified effect under stated assumptions. They are planning illustrations, not observed power and not evidence that the trial achieved a particular degree of precision. The registry describes an illustration based on the observed effect, using a two-sided significance level of 0.05, equal allocation, and a normal approximation. It describes effects as fractions of the observed effect. No numerical sample-size illustration or observed primary effect is reported in the registry information available for interpretation, so these assumptions do not yield a usable conclusion about the trial’s findings.

The registry does not report the trial’s own sample-size justification. The planning assumptions described for an illustration should not be treated as the original design rationale. Nor can a planning illustration be used to claim that the study was adequately or inadequately powered without the effect estimate, variability assumptions, and design details needed for that assessment.

Why is the blinded comparison distinct from the open-label period?

In general, masking can reduce the risk that knowledge of assigned treatment influences behavior, assessment, or reporting. The registry labels the study quadruple-masked, but does not identify the roles covered by that label or provide details of masking procedures. The initial phase compared assigned insulin with assigned placebo under masking. During the open-label period, both groups were to receive insulin, so a later comparison between original groups would not represent the same blinded insulin-versus-placebo contrast.

These explanations describe statistical concepts generally; they do not fill in methods that the registry does not report. In particular, the analysis model, covariate adjustment, confidence intervals, multiplicity strategy, and missing-data approach are not reported in the registry record. Without the primary estimates and those methodological details, the posted registry information supports a description of the endpoint and design, but not a full statistical evaluation of the treatment effect.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists six outcome measures, including the primary cognitive measure and secondary measures of memory, daily functioning, cognitive deficit, brain atrophy, and cerebrospinal fluid biomarkers. When a study assesses multiple outcomes, time points, or comparisons, the chance of obtaining at least one apparently notable result can increase. In general, a prespecified multiplicity strategy can address how to interpret such a family of tests. The strategy used in this trial is not reported in the registry record. The record also notes that the primary outcome has several categories or time points and was not reduced to one comparison. No inferential analyses are listed.

The primary analysis population is described as modified intention-to-treat: randomized participants in Impel Device with ADAS-Cog12 observed at baseline and at least one follow-up. This definition excludes randomized participants without the specified measurement pattern from that analysis population, but the registry does not report the number excluded or the reasons. The detailed handling of missing outcome data, including assumptions and sensitivity analyses, is not reported in the registry record.

In general, missing-data methods matter because estimates can depend on why measurements are absent and on how the analysis treats unobserved values. Requiring at least one follow-up defines who can contribute, but does not by itself explain how incomplete follow-up is handled. The registry does not report whether the primary analysis used all available repeated measurements, imputation, or another approach. These methods should not be inferred from the endpoint’s modeled-change description.

There are no reported primary-outcome estimates or tests with which to assess multiplicity-adjusted or unadjusted evidence. The record does not describe an interim-analysis plan, stopping rules, or the statistical analysis plan; those details are not reported in the registry record.

11. Safety Results

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

Deaths

Other (non-serious) adverse events

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

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: vascular disorders (3.3%), cardiac disorders (2.5%), gastrointestinal disorders (2.5%), general disorders and administration site conditions (1.7%), infections and infestations (1.7%).

The registry reports serious adverse events and other adverse events separately for the insulin and placebo groups. In the insulin group, 18 participants were reported as affected by serious adverse events, with 121 at risk; in the placebo group, 10 were reported as affected, with 119 at risk. The record lists 98 participants affected by other adverse events in the insulin group and 91 in the placebo group, with 121 and 119 at risk, respectively. These are the registry’s counts; no additional rates have been calculated.

The registry also records 1 death in the insulin group and 0 in the placebo group. It lists vascular disorders at 3.3%, cardiac disorders at 2.5%, gastrointestinal disorders at 2.5%, general disorders and administration site conditions at 1.7%, infections and infestations at 1.7%, and nervous system disorders at 1.7% among the top serious adverse event terms. The registry record does not explain the denominators for those percentages or provide enough detail to attribute individual events to treatment.

These figures describe reported safety counts in the registry, not a complete comparative safety analysis. They do not establish causation, and the counts should not be interpreted as proof of either harm or safety. Further details about event timing, severity, relatedness, and statistical comparisons are not reported in the registry record.

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

The registry record establishes that SNIFF was designed to compare intranasal insulin with placebo for cognitive and other outcomes in adults with amnestic mild cognitive impairment or mild Alzheimer’s disease. It identifies ADAS-Cog12 change as the primary endpoint and describes the scale’s direction. It also records the intended treatment schedules, the blinded and open-label periods, the analysis population definition, and selected adverse-event counts.

It does not report a numerical primary-outcome result. There is no reported group contrast, confidence interval, standard error, or p-value from which to judge the size or precision of a cognitive difference. The record therefore cannot support a conclusion that the randomized groups did or did not differ on the primary outcome. A completed study status is not itself evidence for either conclusion.

The treatment schedule also shapes what the phases can answer. The initial 12-month period was blinded and compared insulin with placebo. In the following 6-month open-label period, both original groups were assigned insulin. That later phase may describe outcomes during insulin exposure, but it does not retain a concurrent placebo group as described in the registry. Any interpretation of that period would require its outcome values and analysis details, which are not reported in the registry record.

The safety figures are descriptive counts. They are not adjusted comparisons and do not show whether any event was caused by treatment. The registry does not report the event-level context needed to assess causality or timing. Likewise, the secondary outcome list identifies domains of interest but does not supply results for those domains.

In short, the record documents a completed randomized trial and its stated questions, but does not report the efficacy estimates needed to answer the central cognitive comparison. This is an absence of a reported result in the registry record, not evidence that the treatment had no effect or that it had a beneficial effect.

13. Why This Trial Matters Statistically

This trial is statistically notable for asking a randomized treatment question across a defined blinded comparison and a later open-label period. Separating those phases is essential: the randomized insulin-versus-placebo contrast pertains to the blinded period, whereas both original groups were assigned insulin during open-label follow-up. Treating the later period as if it preserved the original comparison would misstate the design.

The primary endpoint also requires careful directional interpretation. ADAS-Cog12 measures several cognitive domains, and a positive change is defined as worsening. The scale definition makes it possible to interpret the sign of a reported change, but the registry does not report the estimates needed to apply that interpretation to the trial’s results. This distinction illustrates why endpoint definitions and outcome estimates must be read together.

The registry defines a modified intention-to-treat population using baseline and follow-up observations. That makes the analysis population part of the interpretation, particularly when follow-up is incomplete. Yet the record does not report the participant flow into that population or the missing-data methods. The absence of these details limits evaluation of how the reported population, if analyzed, would represent randomized participants.

Finally, a large enrollment figure, a completed status, and a list of secondary outcomes do not replace a treatment-effect estimate. Statistical interpretation requires the contrast and its uncertainty, alongside enough methodological detail to understand how it was obtained. Because the registry does not report primary-outcome results or analyses, this record illustrates an important distinction between a trial’s design and the evidence needed to evaluate its findings.

14. 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 this trial from the registry record is that the primary outcome is described without numerical results. There are no reported group estimates or treatment contrast, and no reported uncertainty interval or test. Consequently, the size, direction, and precision of the primary cognitive comparison cannot be evaluated. The study’s completed status does not resolve this information gap.

The registry also does not report a detailed statistical analysis plan. The model underlying the least-squares mean, the covariates included, the definition of each contrast, and the treatment of repeated measurements are not reported in the registry record. The record identifies the primary population but does not report its realized size or participant flow, and it does not describe missing-data assumptions or sensitivity analyses.

The primary timeframe includes both a blinded period and an open-label period. Because all participants were assigned insulin during open-label follow-up, interpretation of outcomes from that phase would require careful separation from the blinded randomized comparison. The registry does not provide the numerical results or details needed to assess the phases separately. It also notes that multiple categories or time points were not reduced to one primary comparison, while the approach to multiplicity is not reported.

The safety information is limited to summary counts and selected serious adverse event terms. The registry does not report the context needed to assess event timing, attribution, or comparative risk beyond those counts. These limitations apply to interpretation of the registry record and do not establish that the full trial report contains no additional information.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: The Study of Nasal Insulin in the Fight Against Forgetfulness (SNIFF) — NCT01767909.
• Registry record last updated 2024-11-20; status last verified 2024-11.
• Results first posted 2020-05-18.
• 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.

18. Record Summary

SNIFF was a completed randomized, quadruple-masked, parallel-group phase 2/phase 3 trial sponsored by the University of Southern California. It enrolled 240 participants with amnestic mild cognitive impairment or Alzheimer’s disease. The blinded comparison assigned intranasal insulin or placebo for 12 months, followed by a 6-month open-label period in which all participants were to receive insulin.

The primary endpoint was modeled change in ADAS-Cog12, with positive change indicating cognitive worsening. The registry identifies a modified intention-to-treat population and lists secondary outcomes covering memory, daily functioning, cognitive deficit, brain atrophy, and cerebrospinal fluid biomarkers. It reports selected safety counts but lists zero analyses and does not report a numerical primary-outcome result. The treatment effect and its precision therefore cannot be assessed from the registry record.