Effect of Insulin Sensitizer Metformin on AD Biomarkers: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Effect of Insulin Sensitizer Metformin on AD Biomarkers (NCT01965756), a randomized phase 2 study. The primary outcome is Word List Memory Total - ADAS-cog.

2. Clinical Question

This design asks how the memory measure changed while participants received each treatment, with each participant’s periods potentially offering a within-person comparison. The timing and analysis needed to interpret such a comparison matter: the record does not report a specific contrast, the statistical model, or how the baseline and repeated measurements were used. The study is identified as a Phase II pilot, and enrollment is recorded as 20. That information describes the trial’s scale and stage, not the size or precision of any treatment effect.

3. Trial Design

The registry describes a Phase II randomized, quadruple-masked, placebo-controlled crossover pilot study. The listed model is crossover, and enrollment is recorded as 20. The sponsor is the University of Pennsylvania. The record gives the study status as completed, with a primary completion date of 2015-12-22. These details describe the trial’s registered design and status; they do not supply its outcome estimates.

In a crossover design, participants receive interventions in more than one period rather than remaining in a single assigned group throughout. The registry lists two sequences: Metformin, Then Placebo and Placebo, Then Metformin. The descriptions state that the first treatment period lasted 8 weeks, followed by a switch. The primary outcome timeframe is 16 weeks total, with measurements at baseline, week 8 at crossover, and week 16. A washout period, if any, is not reported in the registry record.

Randomization assigns participants to a treatment sequence by chance. Quadruple masking indicates that the registry classified the study as masking four roles, but the identities of those roles and the practical masking procedures are not reported in the registry record. The description does not establish whether masking remained intact throughout treatment or how it was assessed.

The registry lists metformin and placebo as the interventions. It does not report the randomization method, sequence-generation details, allocation concealment, the planned handling of period or sequence effects, or a sample-size justification. The study is marked completed, but completion alone does not indicate that the registered outcome was fully analyzed or that results were posted. The registry records no analyses for this outcome. Consequently, the design can be described, while the conduct and inferential results cannot be reconstructed from the information recorded.

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

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Word List Memory Total - ADAS-cog, measured over 16 weeks (total) - measured at baseline, week 8 (crossover), and week 16.

The primary endpoint is Word List Memory Total on the Alzheimer’s Disease Assessment Scale–Cognitive Subscale (ADAS-cog), expressed in words recalled. The registry describes three trials of 10 words each, for 30 words total. The measurement was planned at baseline, week 8 at crossover, and week 16. The registry identifies the summary parameter as a mean and the dispersion measure as a standard deviation.

Word recall is a direct performance measure of memory: the recorded value represents how many words were recalled across the described trials. In general, researchers use standardized cognitive measures to quantify performance in a consistent way across participants and assessment times. Repeated assessments can describe change over time and, in a crossover design, may inform comparisons between treatment periods. That interpretation depends on the analysis specifying which measurements are compared and how repeated observations are handled.

The registry does not report the observed means or standard deviations at any measurement point, the scale’s score distribution in this trial, or an estimated treatment contrast. It also does not report whether the primary analysis used all three assessments or a subset, or how missing assessments were handled. The endpoint’s definition is therefore available, but its results and precision are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies the outcome as continuous and specifies a mean as its summary parameter, with standard deviation as the measure of dispersion. In general, a mean describes the average observed score in a group or measurement period. A standard deviation describes how spread out individual scores are around that average. These summaries answer different questions: the mean concerns the group’s central value, while the standard deviation concerns variability among observations.

What does an effect estimate describe?

For a continuous outcome, an effect estimate usually expresses a difference in average scores between defined conditions or time points. In this crossover study, that could involve measurements collected during metformin and placebo periods, but the registry does not report the actual comparison or estimate. The choice of contrast is important. A contrast comparing treatment periods is not automatically the same as a change-from-baseline comparison, and neither can be assumed without the analysis specification.

General statistical interpretation distinguishes the estimated size of an effect from the uncertainty around it. A confidence interval gives a range of values compatible with the data under the model and its assumptions. A narrow interval indicates greater precision than a wide interval, while an interval spanning values on both sides of no difference can be compatible with effects in either direction. No confidence interval or treatment-effect estimate is reported in the registry record, so these interpretations cannot be applied to this trial’s primary outcome.

Why does the crossover structure matter?

General crossover analyses account for the fact that participants contribute observations in different treatment periods. The observations from one participant are usually related, rather than independent. An analysis that treats them as independent can misrepresent the uncertainty. A statistical method can model within-participant correlation and distinguish treatment effects from period or sequence effects, depending on the design and available data. The registry does not report which model was used, whether such effects were modeled, or whether any carryover assessment was planned.

Carryover is a general concern when an intervention’s effects may persist into a later period. A washout interval can sometimes be used to reduce this concern, but its existence and duration are not reported in the registry record. The record also does not provide an analysis plan that would show whether the treatment periods were interpreted under assumptions about persistence, time trends, or treatment-by-period differences. These are methodological questions, not claims that carryover occurred in this trial.

What do p-values and confidence intervals tell readers?

In general, a p-value measures how incompatible the observed data are with a specified null hypothesis under the assumptions of a statistical test. It is not the probability that the null hypothesis is true, and it does not measure the clinical importance of an effect. A confidence interval is generally more informative about the range and precision of an estimate, while still depending on the model and data. Neither a p-value nor a confidence interval is reported for the primary outcome in the registry record.

The record’s sample-size assumptions describe planning illustrations using a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. In general, a power-based sample-size calculation asks how many participants per arm would be needed to detect a specified effect under stated assumptions. The described illustrations are planning quantities, not observed power and not evidence that the trial had a particular probability of finding an effect. The registry record does not report a sample-size justification for the trial itself. No numerical planning results or observed effect estimate are reported in the record, so the assumptions alone cannot establish what effect the study could reliably detect.

General statistical practice also distinguishes statistical significance from practical importance. A small estimated difference can be statistically compatible with a non-zero effect if measured precisely, while a potentially important estimate can remain uncertain when data are sparse. Interpretation therefore considers the effect estimate, its interval, the outcome’s meaning, and the study design together. Since the record does not report primary outcome estimates or uncertainty intervals, it does not permit that assessment.

Why can over-dispersion matter?

Over-dispersion is a general issue in count or categorical data when variability exceeds what a simple statistical model expects. The primary endpoint is recorded as a mean continuous measure, not as an event count, so over-dispersion is not identified as an outcome-analysis issue in the registry record. More broadly, choosing a model that matches the type and distribution of an outcome matters because a poor fit can distort standard errors and confidence intervals. The record does not report diagnostic checks or model assumptions for the primary endpoint.

Repeated cognitive assessments can also be affected by measurement variability and changes over time. In general, repeated observations may be correlated within individuals, and a model’s assumptions should reflect that structure. The record identifies the assessment times but does not report reliability estimates, a learning-effect adjustment, or a repeated-measures model. None should be presumed to have been used or omitted in the actual conduct; they are simply not reported in the registry record.

10. Multiplicity, Missing Data and Analysis Populations

The primary outcome is measured at baseline, week 8 at crossover, and week 16. Multiple time points can create more than one possible comparison. If several comparisons are tested, the chance of at least one apparently unusual result can increase unless the analysis plan addresses multiplicity. In general, investigators may prespecify a primary contrast, adjust for multiple testing, or treat additional analyses as exploratory. The registry record does not report a multiplicity strategy or identify a single primary time-point contrast.

The record lists Trails-B as a secondary outcome and states that there are five outcome measures, but it does not list the full set of measures in the registry record for this account. It records zero analyses. No results for the primary or secondary outcomes, or any adjustment across outcomes, are reported. It is therefore not possible to judge whether multiplicity was handled, whether analyses were planned in advance, or whether any reported finding would be confirmatory.

Missing-data handling is also not reported in the registry record. It does not state how many participants completed each treatment period, whether any assessments were missing, why data might have been missing, or whether an imputation or other method was used. In general, missing observations can affect estimates when the reasons for missingness are related to participants’ outcomes or treatment experience. A complete-case analysis and an imputation approach rely on different assumptions; neither can be attributed to this study from the available record.

The record does not define analysis populations, such as an intention-to-treat population or a per-protocol population. Nor does it specify exclusions, adherence requirements, or how participants who did not complete both periods were to be treated. These details are particularly relevant in a crossover design because the intended within-person comparison depends on observations from treatment periods. Their absence from the registry record limits appraisal of the analysis, but does not establish what investigators did outside the information recorded.

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 non-serious adverse event terms

The registry’s safety fields list zero serious adverse events and zero deaths for both Metformin and Placebo. For other, non-serious adverse events, the Metformin group is listed with 20 at risk and 2 affected; the Placebo group is listed with 20 at risk and 0 affected. The record does not list serious adverse-event terms.

These are descriptive registry counts, not a formal comparison of safety between treatments. The record does not describe the two other adverse events, their timing, duration, severity, or relationship to treatment. It also does not explain how events were collected or classified. The recorded “at risk” figures should be read as the group figures in the registry, not as proof that every participant had complete safety follow-up during both crossover periods.

Zero recorded serious events or deaths means none are listed in these fields; it does not establish that no other health events occurred or that a treatment has no risk. Conversely, the two non-serious events listed for metformin do not, by themselves, demonstrate a treatment-related effect. A comparison would need clear event definitions, exposure and follow-up details, and an appropriate analysis, none of which are reported in the registry record.

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

The registry record provides a description of a randomized, quadruple-masked, placebo-controlled crossover pilot study, the treatment sequences, and a primary cognitive outcome. It identifies total word-list memory on the ADAS-cog as the outcome, measured at baseline, week 8, and week 16. This information helps readers understand what the investigators set out to examine and how the measurement was scheduled.

The record does not report the primary outcome scores, an estimated difference between metformin and placebo, a confidence interval, or a p-value. It therefore does not show the direction or size of a primary outcome result, or how precisely any effect was estimated. The absence of a reported analysis is not a finding of no effect. It also is not evidence that the treatments produced similar scores.

The pilot designation and recorded enrollment of 20 describe the study’s scale, but do not establish whether the study was adequately powered for a particular effect. The registry does not report a trial-specific sample-size justification or analysis plan. General planning calculations should not be read as observed power, and no treatment effect can be recovered from the design description alone.

The safety record lists no serious adverse events or deaths in either treatment group and records other non-serious events in the Metformin group. Those fields do not contain enough information to characterize event types, attribution, or comparative risk. Together, the design and safety fields provide a limited account of what was registered, rather than a complete outcome report.

Accordingly, the record supports statements about the study’s design, its endpoint definition, its schedule, and the safety counts entered. It does not support a conclusion that metformin improves, worsens, or leaves unchanged memory performance or safety in the population of interest. Any such conclusion would require outcome and analysis details not reported in the registry record.

13. Why This Trial Matters Statistically

This trial is statistically notable for its crossover structure. Rather than assigning each participant to only one treatment period, the listed sequences switch between metformin and placebo after 8 weeks. In general, this design can make within-person comparisons useful by accounting for some stable differences between participants. It also creates analytical requirements: the relationship between repeated observations, treatment period, sequence, and possible carryover must be considered. The registry record does not report how those issues were addressed.

The primary outcome is a repeated cognitive measure rather than a one-time event count. Its scheduled assessments span baseline, the crossover point, and the end of the 16-week period. This makes the exact treatment contrast central to interpretation. A comparison between two treatment-period measurements may answer a different question from a comparison of change from baseline. Without an analysis specification or posted estimate, the registered timing cannot identify which contrast was evaluated.

The study is described as a pilot and records enrollment of 20. Pilot studies can be useful for examining feasibility and generating preliminary estimates, but their findings require attention to precision and uncertainty. The record does not report a sample-size justification or primary outcome results, so it cannot show how much information the study yielded about the treatment comparison.

The registry record is also a reminder that a completed study status and a defined endpoint are not substitutes for posted results. Here, the design and outcome are identifiable, while the primary analysis is not reported. That distinction matters for statistical readers: the study question can be described, but its treatment effect cannot be quantified or assessed for precision from the record.

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 principal limitation for interpretation is that the registry record contains no primary outcome values or analysis. It therefore does not provide an estimated effect, uncertainty interval, or test result. Although the primary endpoint and assessment schedule are described, the exact contrast used for inference is not reported. This prevents a statistical assessment of the primary question.

The crossover structure raises questions that cannot be resolved from the registered details. The record does not report a washout period, how period and sequence effects were handled, whether carryover was assessed, or what model was used for repeated measurements. It also does not state how incomplete treatment periods or missing assessments were handled. These details are important to understanding the validity and precision of a within-participant comparison.

The record does not report participant characteristics, detailed eligibility criteria, adherence, withdrawals, or the number of participants contributing data at each measurement. The registered enrollment is 20, but enrollment alone does not show the analyzed population or completeness of follow-up. The masking classification is recorded as quadruple, but the masked roles and procedures are not described.

Safety reporting is limited to group counts for serious events, other events, and deaths. The types and timing of the other adverse events are not reported, and there is no comparative safety analysis. The registry also does not report a trial-specific sample-size justification, multiplicity strategy, analysis populations, or statistical analysis plan. These omissions limit what can be concluded from the record; they do not establish that the corresponding procedures were absent from the trial itself.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Effect of Insulin Sensitizer Metformin on AD Biomarkers — NCT01965756.
• Registry record last updated 2017-09-21; status last verified 2017-08.
• Results first posted 2017-09-21.
• 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

The University of Pennsylvania sponsored NCT01965756, a completed Phase II randomized, quadruple-masked, placebo-controlled crossover pilot study. Enrollment is recorded as 20. The listed sequences began with metformin or placebo for 8 weeks and then switched treatment. The primary outcome was total word-list memory on the ADAS-cog, measured at baseline, week 8, and week 16.

The registry record does not report primary outcome results, an analysis, a treatment-effect estimate, or its precision. Trails-B is listed as a secondary outcome, and the record lists zero analyses. Safety fields record zero serious adverse events and deaths in both groups, with other non-serious adverse events recorded for the Metformin group only. Event details are not reported. The design and endpoint can be described, but the primary treatment comparison cannot be evaluated from the registry record.