This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Hyperbaric Oxygen Therapy for Cognition in Diabetic Elderly at High Dementia Risk (NCT03036254), a randomized na study. The primary outcome is Change in Global Cognition Score.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03036254 |
| Phase | NA |
| Condition | Diabetes Mellitus |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Double |
| Enrollment | 155 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Icahn School of Medicine at Mount Sinai |
| Results posted | Yes |
2. Clinical Question
The trial asked whether hyperbaric oxygen therapy (HBOT), compared with a sham intervention, was associated with change in global cognition among older adults with diabetes and mild cognitive impairment. The registry describes the target group as being at high risk of dementia and cognitive decline. It also identifies cerebral blood flow and glucose uptake as areas of interest in the trial summary, but the primary outcome described in the results record is a cognitive composite score.
Population
Older adults with diabetes and mild cognitive impairment who were considered at high dementia risk. The registry lists Diabetes Mellitus as the condition. Follow-up numbers were affected by a 14-month pause caused by the Covid-19 pandemic.
Intervention
The HBOT intervention used 100% oxygen at 2 ATA, with 5 minute air breaks every 30 minutes. Sessions lasted 90 minutes, took place 5 times/week, and were planned for 60 sessions. The registry describes a multiplace chamber and a nurse present throughout the session.
Comparator
The sham intervention included the other described conditions of HBOT, including time in the chamber, a face mask, vital-sign checks, and a nurse present. Pressure was the stated exception between the interventions.
Primary question
How did the groups compare in change in global cognition, measured with a composite z-score at baseline, 12 weeks, 6 months and 12 months? A comparative primary outcome result is not reported in the registry record.
3. Trial Design
The registry describes a randomized, double-masked, parallel-group trial of HBOT and a sham intervention. The trial is marked completed, with enrollment of 155. The phase is listed as NA. Icahn School of Medicine at Mount Sinai is named as the sponsor. The registry gives a start date of 2017-08-10 and a primary completion date of 2024-04-15.
The HBOT arm description specifies treatment in a multiplace unit, with 100% oxygen at 2 ATA, 5 minute air breaks every 30 minutes, and 90 minutes in the chamber per session. The stated schedule was 5 times/week for 60 sessions. A nurse was described as remaining in the chamber throughout the session. The sham intervention matched the other described conditions, including mask use, time in the chamber, session frequency, and nursing procedures; pressure was the stated difference.
Randomization means assignment to the study groups is determined by a chance-based process, in general. Double masking means that the registry classifies the trial as double-masked, but who was masked and how masking was maintained are not reported in the registry record. The allocation method and details of the randomization procedure are also not reported in the registry record.
The trial summary describes its purpose as investigating short- and long-term effects on cognition, cerebral blood flow, and glucose uptake in diabetic older adults with mild cognitive impairment. It also describes the trial as a pilot intended to inform a larger study. The registry record does not report a sample-size justification or the planned statistical analysis. The follow-up schedule for the primary outcome is listed as baseline, 12 weeks, 6 months and 12 months. A 14-month pause caused by the Covid-19 pandemic affected follow-up numbers; further details about the extent of missing follow-up 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.
Overall Study
| Stage | HBOT Intervention | Sham Intervention |
|---|---|---|
| STARTED | 77 | 78 |
| COMPLETED | 60 | 67 |
| NOT COMPLETED | 17 | 11 |
Reasons for not completing this period:
| Reason | HBOT Intervention | Sham Intervention |
|---|---|---|
| Adverse Event | 6 | 4 |
| Death | 3 | 1 |
| Lack of Efficacy | 5 | 1 |
| Withdrawal by Subject | 3 | 5 |
Analysis population for the primary outcome, as defined in the registry: The Covid-19 pandemic caused a 14-month pause, affecting follow-up numbers.
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.
| Characteristic | Unit | HBOT Intervention | Sham Intervention | Total |
|---|---|---|---|---|
| Participants analysed | 77 | 78 | 155 | |
| Age, Continuous | years | 71.7 (SD 4.182) | 70.2 (SD 3.560) | 70.94 (SD 3.938) |
| Sex: Female, Male: Female | Participants | 27 | 21 | 48 |
| Sex: Female, Male: Male | Participants | 50 | 57 | 107 |
| Race/Ethnicity, Customized: MENA: Middle Eastern North African | Participants | 33 | 44 | 77 |
| Race/Ethnicity, Customized: Ashkenazi | Participants | 41 | 32 | 73 |
| Race/Ethnicity, Customized: Unknown | Participants | 3 | 2 | 5 |
| Clinical Dementia Rating - Sum of Boxes (CDR-SB) | units on a scale | 1.6 (SD 1.16) | 1.51 (SD 0.80) | 1.56 (SD 0.99) |
| Total Mini-Mental State Examination | units on a scale | 28.55 (SD 1.28) | 28.49 (SD 1.36) | 28.52 (SD 1.32) |
| Total Beck Depression Inventory | units on a scale | 5.52 (SD 3.89) | 5.34 (SD 4.02) | 5.43 (SD 3.95) |
6. Primary Endpoint
The registry lists the primary outcome as Change in Global Cognition Score, measured over baseline, 12, weeks, 6 months and 12 months.
| Element | Registry entry |
|---|---|
| Result type | Mean |
| Measure of spread | Standard Deviation |
| Unit | Composite z-score |
| Endpoint classification | Continuous Endpoint |
The primary outcome is change in Global Cognition Score, expressed as a composite z-score. It combines 11 cognitive performance variables that assess different aspects of cognition. The registry says the measures are standardized using z-scores, allowing values on different scales to be combined and compared. The standardized values are averaged to form the composite, whose stated full range is -1 to 1.
Most of the described measures are scored so that higher values indicate better performance. Trail Making Test Part A and Trail Making Test Part B are exceptions: they measure time, so higher values indicate worse performance on those individual measures. The registry describes the composite as oriented so that a higher composite z-score reflects better cognitive performance and a more favorable outcome.
In general, combining standardized measures can provide a broader summary than relying on a single task, while reducing the influence of different measurement scales. A composite also depends on the chosen tasks and scoring rules. The registry identifies the number and general direction of the components, but further details about the construction and validation of this particular composite are not reported in the registry record.
The outcome was assessed at baseline, 12 weeks, 6 months and 12 months. These repeated assessments are intended to describe change over follow-up, but the registry record does not report a group comparison, an estimated change, a confidence interval, or a p-value for the primary outcome.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The record identifies the primary endpoint and its assessment schedule, but does not provide a primary outcome comparison. It therefore does not establish the direction or size of any difference between HBOT and sham in global cognition. An effect estimate, confidence interval, and p-value for this endpoint are not reported in the registry record. The appropriate interpretation is that the posted record does not give the numerical outcome results needed to judge the observed cognitive changes or their precision.
The primary endpoint is a composite average of standardized cognitive measures, with higher composite values defined as better performance. This orientation matters because two component tests measure completion time, for which higher individual values indicate worse performance. The registry describes the construction and intended direction of the composite, but does not report component-level results or the contribution of each component to the composite. Readers therefore cannot determine from the registry record whether any pattern in particular tasks was consistent with the composite.
The trial is described as randomized and double-masked, which are design features intended in general to reduce certain sources of bias. Those labels do not substitute for results or detailed information about how assignment and masking were conducted. The procedure for randomization, the identities of masked groups, and checks on whether masking was maintained are not reported in the registry record. These missing details limit assessment of how well the design features were implemented.
The registry reports that the Covid-19 pandemic caused a 14-month pause that affected follow-up numbers. This makes the completeness and timing of follow-up important when interpreting changes over time. The number of participants with available primary outcome data at each assessment, reasons for missing observations, and methods for handling missing data are not reported in the registry record. Without those details, the extent to which incomplete follow-up could affect the comparison cannot be assessed.
Several secondary cognitive and functional measures are listed, including the NIH Computerized Cognitive Test, CDR-SB, BDI, ADL-PI, ADCS-ADL MCI, and MMSE. A list of outcomes alone does not show whether they were analyzed, what they found, or how they relate to the primary outcome. The record lists 9 outcome measures and 0 analyses. A multiplicity strategy for interpreting results across outcomes and assessment times is not reported in the registry record.
Any planning illustration based on an observed effect should be read as a planning aid, not as observed power or as an outcome of this trial. The record's stated planning assumptions include a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation, with the effect to detect defined as a stated fraction of the observed effect. Those assumptions do not supply the missing primary outcome estimate, and the registry record does not report a sample-size justification for the trial's own design.
9. Statistical Methods Explained
Statistical methods help distinguish a measured difference from the uncertainty around that difference. In this record, the primary outcome is continuous and is described using a mean and standard deviation. The record does not report an estimate of the group difference or its uncertainty, so the statistical interpretation is limited to describing the planned measure and identifying what is absent.
What does a mean describe?
In general, the mean is an average that summarizes the location of a set of measurements. For this endpoint, each participant's composite z-score combines standardized cognitive measures, and a group mean could summarize the scores for participants in an arm at a given assessment. A comparison of means could then describe the average difference between arms, or the difference in change, depending on the analysis. The registry record does not report which specific comparison was planned or present a group mean for the primary outcome.
A mean is not a description of every participant's experience. Participants can have values above and below the group mean, and a similar average can arise from different distributions of individual scores. Reporting a mean without an accompanying measure of variability and a clear denominator would not fully describe the observed data. The registry identifies the standard deviation as the dispersion measure for the primary outcome, but numeric primary outcome means and standard deviations are not reported in the registry record.
What does the standard deviation describe?
In general, the standard deviation describes how spread out measurements are around their mean. A smaller standard deviation indicates that measurements are more concentrated around the mean; a larger one indicates greater spread. It is a descriptive measure of participant-level variation, not a measure of how precisely the group mean or a between-group comparison has been estimated. Precision of an estimate is typically conveyed with a confidence interval, which is not reported for the primary outcome in the registry record.
Standard deviations can also help readers assess how variable a continuous outcome is within each group. However, a standard deviation alone does not show whether the treatment groups differ, because that requires comparing the groups' estimates. Nor does the standard deviation establish whether an observed difference is clinically important. The registry record provides no primary endpoint estimates with which to make those assessments.
Why use a composite z-score?
In general, standardizing measurements expresses them on a common scale, making measures with different original units more suitable for combination. The record says that 11 cognitive performance variables are standardized and averaged. This approach gives the outcome a broad cognitive interpretation rather than limiting it to a single task. The record states that higher composite scores reflect better cognitive performance and that the full range is -1 to 1.
Composite construction requires decisions about which measures to include, how to orient each measure, and how to combine them. Here, the registry describes the measures generally and identifies the two timed Trail Making tests as exceptions to the usual direction of scoring. It does not report further details such as how missing component scores were handled or whether all components contributed to every participant's composite. Those details are not reported in the registry record.
How should repeated assessments be understood?
The primary outcome timeframe includes baseline, 12 weeks, 6 months and 12 months. In general, repeated measurements allow investigators to describe outcomes over time and compare patterns of change between groups. They also create analytic choices: an analysis may compare change from baseline at individual follow-ups or use a model that incorporates multiple measurements. These are general possibilities, not a description of the trial's analysis. The statistical model, the primary time point for inference, and the planned contrast are not reported in the registry record.
Repeated assessments are not independent in the usual sense because measurements from the same participant are related. General approaches account for this within-participant relationship when estimating change over time. The registry does not identify an approach, report model assumptions, or give a test of the primary group comparison. It is therefore not possible to infer the method used from the outcome schedule alone.
What does a confidence interval add?
In general, a confidence interval gives a range of values compatible with the data and the statistical method used, conveying the precision of an estimate. A narrow interval indicates greater precision than a wide interval, all else equal. It does not guarantee that the true effect lies in the interval, and it does not by itself establish clinical importance. No primary outcome confidence interval is reported in the registry record.
A p-value, in general, describes how incompatible the observed data are with a specified null hypothesis under the model used. It does not measure the size or importance of an effect, and it does not establish that a treatment works or does not work. No primary outcome p-value is reported in the registry record. With neither an estimate nor its uncertainty reported, readers should not infer a result from the trial design, enrollment, or safety figures.
How do planning illustrations differ from trial results?
The registry describes planning illustrations based on the observed effect, using a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. The effect to detect is a stated fraction of the observed effect. In general, such calculations can illustrate the number of participants per arm for 80% power under specified assumptions. They are planning calculations, not evidence that the trial had that power, and they are never observed power. The registry's sample-size justification and original design calculations are not reported in the registry record.
These illustrations depend on assumptions about the target effect and the statistical model. If the observed effect is uncertain, a planning value based on it is uncertain as well. A sample-size calculation cannot replace reporting the actual estimate, confidence interval, analysis population, and missing-data handling. The primary outcome comparison and its numerical results remain not reported in the registry record.
10. Multiplicity, Missing Data and Analysis Populations
The record lists multiple secondary outcomes in addition to the primary cognitive composite. These include the NIH Computerized Cognitive Test, Clinical Dementia Rating-Sum of Boxes, Beck Depression Inventory, two activities-of-daily-living measures, and total MMSE. The registry records 9 outcome measures and 0 analyses. A procedure for controlling or interpreting multiplicity across outcomes, time points, or component measures is not reported in the registry record.
In general, testing many outcomes or time points can increase the chance of finding at least one apparently notable result by chance. Multiplicity methods can address this issue, but they involve choices about which outcomes form a family and how evidence should be adjusted or interpreted. No such choices should be inferred from the outcome list alone. Since comparative analyses are not reported, there are no posted estimates to evaluate in light of multiplicity.
Missing-data information is especially relevant because the registry says a 14-month pause caused by the Covid-19 pandemic affected follow-up numbers. The number with outcome data at each assessment, reasons for missing observations, and the statistical method for handling missingness are not reported in the registry record. It is also not reported whether the analysis followed assigned groups, used another analysis population, or included all randomized participants with available measurements.
In general, missing data can affect both precision and the validity of a comparison, depending on why observations are absent and how the analysis addresses them. The pandemic-related pause gives context for possible follow-up disruption, but does not establish how many observations were missing or whether the groups were affected similarly. Those details are not reported in the registry record, so the impact on the primary outcome comparison cannot be assessed.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| HBOT Intervention | 18 | 77 | 23.4% |
| Sham Intervention | 6 | 78 | 7.7% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| HBOT Intervention | 3 of 77 |
| Sham Intervention | 1 of 78 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| HBOT Intervention | 55 of 77 |
| Sham Intervention | 51 of 78 |
Most frequent serious adverse event terms
| Event term | Organ system | HBOT Intervention | Sham Intervention |
|---|---|---|---|
| Deceased | General disorders | 3 of 77 | 1 of 78 |
| Cardiac catheterization | Surgical and medical procedures | 3 of 77 | 0 of 78 |
| Stroke | Nervous system disorders | 2 of 77 | 1 of 78 |
| Fall | Injury, poisoning and procedural complications | 1 of 77 | 1 of 78 |
| Inguinal hernia | Gastrointestinal disorders | 1 of 77 | 1 of 78 |
| Myocardial infarction | Cardiac disorders | 1 of 77 | 1 of 78 |
| Pneumonia | Infections and infestations | 1 of 77 | 1 of 78 |
| Retinal detachment | Eye disorders | 2 of 77 | 0 of 78 |
| Acute cough | Respiratory, thoracic and mediastinal disorders | 1 of 77 | 0 of 78 |
| Atelectasis | Respiratory, thoracic and mediastinal disorders | 1 of 77 | 0 of 78 |
Most frequent non-serious adverse event terms
| Event term | Organ system | HBOT Intervention | Sham Intervention |
|---|---|---|---|
| Barotrauma | Injury, poisoning and procedural complications | 20 of 77 | 4 of 78 |
| Covid 19 | Infections and infestations | 10 of 77 | 9 of 78 |
| Fall | Injury, poisoning and procedural complications | 5 of 77 | 4 of 78 |
| Cataract | Eye disorders | 6 of 77 | 2 of 78 |
| Back pain | Musculoskeletal and connective tissue disorders | 5 of 77 | 1 of 78 |
| Cardiac catheterization | Surgical and medical procedures | 3 of 77 | 1 of 78 |
| Cold | Infections and infestations | 2 of 77 | 2 of 78 |
| Inguinal hernia | Gastrointestinal disorders | 0 of 77 | 4 of 78 |
| Prostate resection | Surgical and medical procedures | 1 of 77 | 3 of 78 |
| Hip injury | Injury, poisoning and procedural complications | 1 of 77 | 2 of 78 |
Serious events with the highest arm-level rate: cardiac catheterization (3.9%), deceased (3.9%), retinal detachment (2.6%), stroke (2.6%), acute cough (1.3%).
The registry reports serious adverse event figures separately for the HBOT and sham groups. For HBOT, the serious adverse event count is 18, with 77 listed at risk; for sham, the count is 6, with 78 listed at risk. The record also lists other adverse event counts of 55 for HBOT and 51 for sham, and deaths of 3 and 1, respectively. These are descriptive registry figures, not a comparative analysis.
Counts should be read with their listed group denominators and in the context of how events were collected and defined. The registry record does not report the event-level timing, duration, severity details, relationship to intervention, or statistical comparison needed to interpret the group figures as evidence of a treatment-related difference. It also does not report whether the counts represent participants with events or total event occurrences in a way that resolves repeated events for an individual.
The record lists cardiac catheterization and deceased at 3.9% each, retinal detachment and stroke at 2.6% each, and acute cough and atelectasis at 1.3% each among the top serious terms. The denominator and the exact basis for these percentages are not reported in the registry record. These terms do not replace the full adverse-event record or establish whether an event was caused by the intervention. The figures should not be used alone to compare safety between groups.
12. What the Result Does, and Does Not, Mean
The registry record describes a randomized, double-masked comparison of HBOT with a sham intervention in older adults with diabetes and mild cognitive impairment. It defines a composite cognition outcome and gives its assessment schedule. It also reports group-level serious and other adverse event counts. These registry entries describe the trial's design and posted safety figures, but they do not provide the primary cognitive outcome comparison.
Without a reported estimate of change between groups, its uncertainty, and the analysis population, the record cannot show the size, direction, or precision of the primary outcome result. It is therefore not possible to say from the registry record whether cognition differed between the groups, how any difference changed over follow-up, or whether a difference would be clinically meaningful. No treatment conclusion should be drawn from the outcome definition or study design alone.
Likewise, the serious adverse event counts are not by themselves a causal comparison. Interpreting them would require fuller information about event definitions, ascertainment, timing, participant-level exposure, and analysis. The listed terms and percentages add descriptive detail, but the basis for those percentages and a comparative safety analysis are not reported in the registry record.
In general, a trial result is best interpreted using the estimated effect together with its confidence interval, the completeness of follow-up, and the analysis methods. Here, the registry notes a 14-month pandemic-related pause that affected follow-up numbers, but does not give the amount or handling of missing data. The posted record can identify the question and some design features, but its reported information is insufficient to determine the cognitive comparison or its precision.
13. Why This Trial Matters Statistically
This trial is statistically interesting because it uses a composite cognition measure built from standardized performance variables. Combining measures can provide a broader summary of cognitive performance than relying on a single test. The scoring direction is also important: most described measures treat higher values as better, while the timed Trail Making tests treat higher values as worse. The registry states that the composite is oriented so that higher values reflect better performance.
The repeated assessment schedule makes the question one of change over time, not just a comparison at a single assessment. In general, that requires clear choices about the time point or pattern of change that is intended to support the main conclusion. The registry gives the assessment schedule but does not report the planned primary contrast or statistical model. Readers therefore cannot identify which comparison was intended to carry the primary inference.
The pandemic-related 14-month pause makes completeness and timing of follow-up especially relevant. A trial can remain randomized while the interpretation of repeated measurements depends on how many observations are available and why others are absent. The registry does not provide the required follow-up counts or missing-data methods, so the influence of the pause on statistical precision and potential bias cannot be assessed.
Finally, the registry includes several secondary outcomes and descriptive adverse-event counts, yet reports 0 analyses and no primary cognitive comparison. This distinction matters: the existence of outcome definitions and safety counts is not equivalent to a complete statistical results report. The record defines an important clinical research question, but the unavailable comparative estimates prevent readers from evaluating the primary result's magnitude and precision.
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.
A central limitation of the registry record is that it does not report a comparative result for the primary endpoint. There are no posted group estimates, between-group change estimate, confidence interval, or p-value for the composite cognition score. The component-level cognitive results are also not reported in the registry record. This prevents assessment of both the overall outcome and whether its component measures moved in a consistent direction.
The registry notes that the Covid-19 pandemic caused a 14-month pause affecting follow-up numbers, but does not provide the number of participants assessed at each time point, the reasons observations were missing, or the approach used to address missing data. Details of the analysis population are not reported in the registry record. These omissions leave uncertainty about the completeness of follow-up and the precision and validity of any comparison.
Although the trial is labeled randomized and double-masked, the registry does not report the randomization procedure, who was masked, or how masking was assessed. The statistical plan, primary contrast, and multiplicity approach are also not reported in the registry record. The record describes the composite's general construction, but not all operational details, including how missing component values were treated.
The safety information is limited to group counts and listed serious-event terms and percentages. Further information about event definitions, participant-level recurrence, timing, attribution, and the basis for the listed percentages is not reported in the registry record. The posted safety figures therefore support only a descriptive reading, not a causal or adjusted comparison.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Hyperbaric Oxygen Therapy for Cognition in Diabetic Elderly at High Dementia Risk — NCT03036254.
- Registry record last updated 2025-08-11; status last verified 2025-07.
- Results first posted 2025-08-11.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
The registry record describes NCT03036254, a completed, randomized, double-masked, parallel-group trial sponsored by Icahn School of Medicine at Mount Sinai. Enrollment was 155. The trial compared an HBOT intervention with a sham intervention in older adults with diabetes and mild cognitive impairment.
The primary outcome was change in a global cognition composite z-score, assessed at baseline, 12 weeks, 6 months and 12 months. The composite averaged standardized values from 11 cognitive performance variables; higher composite scores were described as better performance. The registry reports that a 14-month Covid-19-related pause affected follow-up numbers.
The record lists secondary outcomes and adverse-event counts, but reports 0 analyses. A comparative primary outcome result, missing-data methods, analysis population, and statistical plan are not reported in the registry record. The posted information therefore describes the trial question and some design and safety details, but does not permit evaluation of the primary cognitive effect.