Nonin 4 Wavelength Cerebral Oximeter Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Nonin 4 Wavelength Cerebral Oximeter Study (NCT01762722), a na na study. The primary outcome is Accuracy of Sensor.

2. Clinical Question

3. Trial Design

The registry describes a completed, single-group study in healthy volunteers, sponsored by Duke University. Its title concerns calibration and validation of a four-wavelength, non-invasive cerebral oxygen saturation oximeter and cerebral sensor. The intervention is reduction of inspired oxygen; the stated purpose is to evaluate device accuracy across a range of oxygen saturation values.

The brief summary describes placing a sensor on the forehead to emit and detect light, then comparing its displayed value with blood samples taken simultaneously from an artery in the wrist and a vein in the neck. Oxygen levels were to be reduced in a controlled manner by lowering inspired oxygen concentration. The summary characterizes this as equivalent to ascending to an altitude of 16,000 feet. The arm description refers to oxygen desaturation over a clinical range of 70 - 100%.

The registry labels the two groups Calibration and Validation, with 12 at risk in Calibration and 11 at risk in Validation for the adverse-event reporting. Those labels do not establish a randomized comparison: allocation is listed as NA, masking as NONE, and the record explicitly cautions that the arms are not a like-for-like control comparison. The record also describes the overall model as single group, so the group labels should not be interpreted as evidence of a treatment-control design.

The start is recorded as 2009-09 and primary completion as 2010-06; status is COMPLETED. The registry reports one outcome measure and no analyses. It does not report a detailed protocol, the sequence of calibration and validation procedures, or whether either group label corresponds to a separate phase of measurement.

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: Healthy volunteers recruited with specific exclusion of anemia \& hemoglobinopathy.

Overall Study

Analysis population for the primary outcome, as defined in the registry: Multiple data points from each individual were pooled and presented as group data.

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 Accuracy of Sensor, measured over Data collected from individual participants over 1 hour timeframe. Data from cohort of subjects collected over 6 month period..

The primary outcome is titled Accuracy of Sensor and uses percentage saturation as its unit. The registry describes a scatterplot with forehead sensor saturation on one axis and measured blood saturation on the other. The line of identity represents equality between the sensor and blood measurements: points on that line would indicate matching values for those paired measurements.

The stated summary measure is Arithmetic Root Mean Square, or ARMS. As described in the record, ARMS measures the dispersion of observed points around the line of identity. A smaller ARMS indicates that the data points lie closer to that ideal line, which is the registry’s stated interpretation of greater measurement accuracy. This is a measurement-agreement concept, not a comparison of clinical outcomes between assigned treatments.

The timeframe is described as data collected from individual participants over a 1 hour timeframe, with cohort data collected over a 6 month period. The registry says that multiple data points from each individual were pooled and presented as group data. It does not provide the numerical ARMS estimate, a confidence interval, or other numerical endpoint results. The plotted approach and endpoint definition therefore explain what was intended to be assessed, but do not establish the observed accuracy in this study.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary endpoint is a method for assessing agreement between two measurements: forehead sensor saturation and measured blood saturation. It is not described as a test of whether one group has a better clinical outcome than another. The relevant ideal is equality between the two readings, represented by the line of identity in the registry’s scatterplot description.

What does the line of identity represent?

In general, a line of identity marks the locations where paired measurements have the same value. A point lying on the line represents equality for that pair; distance away from it represents disagreement. Looking at a scatterplot can show whether readings tend to follow the ideal relationship and whether the spread changes across the measurement range. A scatterplot may also reveal patterns that a single summary measure cannot fully describe, such as systematic departures from equality or differing spread in different parts of the range. These are general features of agreement assessment, not additional findings reported for this study.

The registry’s endpoint description places forehead sensor saturation on the vertical axis and measured blood saturation on the horizontal axis. It says that ARMS summarizes the dispersion of the actual points around the line of identity. As a general statistical concept, a root-mean-square summary combines the size of deviations into a measure that gives greater influence to larger deviations than a simple average of signed errors would. Because the registry does not provide the numerical ARMS result or the underlying plotted values, the actual size and pattern of disagreement cannot be evaluated from the record.

Why consider repeated measurements?

The record states that multiple data points from each individual were pooled and presented as group data. In general, repeated readings from the same person are often correlated: they may share individual characteristics or measurement conditions. Treating every reading as if it came from an unrelated person can make the available information appear more independent than it is. Appropriate analysis may need to account for within-person clustering, depending on the estimand and analysis plan. The registry does not report whether such adjustment was used, how the pooled data were weighted, or how uncertainty was calculated.

The measurement schedule is described as data collected from individual participants over a 1 hour timeframe, with cohort data collected over a 6 month period. Those descriptions identify the stated collection windows; they do not give the count or timing of paired observations, or show whether measurements were complete for every participant. Those details are not reported in the registry record.

What can a confidence interval add?

In general, a confidence interval expresses the statistical uncertainty around an estimate under the assumptions of the method used. A narrow interval indicates greater precision than a wide interval, while the interval’s location indicates which values remain compatible with the data and method. An interval around an agreement measure would help readers judge how precisely the sensor’s agreement was estimated. No confidence interval for the primary endpoint is reported in the registry record, so its precision cannot be assessed from a reported interval.

Why does over-dispersion matter?

In general, over-dispersion means that the variability in data exceeds what a particular statistical model expects. If a model assumes more uniform or independent observations than the data support, standard errors and related uncertainty measures may be too small. With repeated measurements, dependence within individuals is one reason a simple independent-observation model may be inadequate. This general concern is relevant to understanding pooled repeated measurements, but the registry does not report a fitted model, a dispersion assessment, or whether over-dispersion was evaluated.

The study record provides an outcome definition rather than a full statistical analysis plan. It reports one outcome measure and no analyses. Details such as the analysis population, treatment of missing measurements, any adjustment for repeated observations, confidence-interval method, and criteria for interpreting the ARMS result are not reported in the registry record. Accordingly, the endpoint description can explain the intended statistic, but does not establish which inferential methods were ultimately applied.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists one outcome measure, Accuracy of Sensor, and reports no analyses. This limits what can be said about multiplicity: the record does not describe a family of analyzed outcomes, a hierarchy of hypotheses, or an adjustment for multiple testing. It also does not report a prespecified statistical analysis plan or explain whether any additional analyses were planned or conducted. The absence of these details should not be interpreted as evidence that a particular multiplicity procedure was or was not used.

Missing-data handling is not reported in the registry record. The description that multiple data points from each individual were pooled and presented as group data does not indicate whether all planned readings were obtained, whether some readings were excluded, or how incomplete measurement pairs were handled. The number of missing readings and the reasons for any missingness are also not reported. Without those details, it is not possible to determine whether the analyzed measurements represent all intended observations or whether missingness could affect the accuracy summary.

The analysis population is not reported in the registry record. The stated population includes healthy volunteers, and the registry reports enrollment of 23, but it does not specify which participants or measurements were included in the primary analysis. Nor does it state whether the primary endpoint was calculated for each individual before aggregation or directly from pooled observations. These distinctions matter because pooled repeated readings and participant-level summaries answer different statistical questions.

The record’s description of calibration and validation groups does not supply an analysis population definition or establish a controlled comparison. Allocation is listed as NA, the model is single group, and the registry warns that the arms are not a like-for-like control comparison. No inferential comparison between the two labels should be assumed from their names alone.

11. 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.

The adverse-event record lists Calibration and Validation separately. For serious adverse events, the number affected is reported as 0 in each group; for other adverse events, the number affected is also reported as 0 in each group. The corresponding at-risk counts are 12 for Calibration and 11 for Validation. The record does not list serious-event terms.

These figures describe what was entered in the registry’s adverse-event fields; they do not establish that every possible event was actively sought or recorded in the same way. The record does not report event definitions, ascertainment procedures, follow-up completeness, or a narrative safety assessment. Death counts are listed as null, so a death count is not reported in the registry record; null should not be read as a reported count of 0.

The group labels and at-risk counts are not evidence of a randomized safety comparison. The registry lists allocation as NA and cautions that the arms are not a like-for-like control comparison. The counts therefore should be read descriptively, not as comparative evidence about event risk. They also do not provide enough information to infer the absence of unreported or unmeasured events.

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

The registry describes a study intended to assess whether a forehead cerebral oximeter’s readings agree with blood saturation measurements during controlled oxygen desaturation in healthy volunteers. Its endpoint description gives a clear statistical target: closeness of paired readings to the line of identity, summarized by ARMS. The record also describes the sampling approach as pooling multiple measurements from each individual into group data.

However, the registry record does not report a numerical primary accuracy estimate, plotted values, confidence interval, or analysis result. The endpoint definition alone cannot establish how close the readings were to the blood measurements, how agreement varied across the described saturation range, or how precisely accuracy was estimated. No claim about the measured accuracy can be drawn from the outcome definition in the absence of reported results.

The trial is described as single group, with Calibration and Validation labels appearing in the record. Allocation is listed as NA, and the registry specifically cautions that the arms are not a like-for-like control comparison. The labels therefore do not support treating the study as a randomized comparison between interventions. Any interpretation should stay focused on the intended device-measurement question rather than a treatment effect.

The population is described as healthy volunteers. The record does not establish whether findings would apply to other populations or clinical settings, and it does not report participant characteristics that would allow such comparisons. The adverse-event fields report 0 affected for serious and other events in each named group, but those entries do not provide a complete account of ascertainment or follow-up; death counts are not reported.

In sum, the registry provides design and endpoint definitions, but not the numerical results needed to quantify accuracy or uncertainty. It explains what the study sought to measure, not what value the measurement produced. The planning illustrations are calculations for hypothetical future designs under stated assumptions, not observed power and not a replacement for the trial’s unreported endpoint result.

13. Why This Trial Matters Statistically

This study is statistically interesting because device validation is fundamentally an agreement problem. A sensor may be evaluated by asking whether its readings track a reference measurement, rather than by comparing clinical outcomes between assigned treatments. The registry’s line-of-identity and ARMS description makes that distinction explicit: the target is the spread of paired measurements around equality.

The record also highlights why the unit of analysis matters. It says that multiple measurements from each individual were pooled as group data. In general, repeated values from the same person can contain less independent information than the same number of observations collected from unrelated people. Clear reporting of how repeated readings are handled is therefore important for interpreting both the estimate and its uncertainty. The registry does not report those analysis details.

Another important statistical point is the distinction between a measurement protocol and a reported result. The registry describes simultaneous blood sampling, controlled reduction of inspired oxygen, and the intended accuracy metric. Those features help define the question, but they do not give the observed numerical agreement. A result estimate and an uncertainty measure are needed to judge what the data indicate and how precisely they do so.

Finally, the record cautions against inferring a comparative design from group names. Calibration and Validation are labels, while allocation is NA and the model is single group. Treating these labels as randomized arms could lead to an inappropriate between-group interpretation. Accurate reading of the design is as important as understanding the endpoint itself.

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 statistical interpretation is that the registry does not report the numerical primary endpoint result. It defines ARMS and describes the intended scatterplot, but gives no estimate, confidence interval, or underlying paired measurements. Consequently, the magnitude and precision of agreement cannot be assessed.

The record reports pooled group data from multiple measurements per individual, but does not report the number of paired observations or how within-person dependence was handled. It does not explain whether readings were summarized by participant before group aggregation, how observations were weighted, or how uncertainty was estimated. These omissions limit assessment of the information contributed by the repeated measurements.

Further details about missing readings, exclusions, and the analysis population are not reported in the registry record. The analysis methods, including any handling of repeated observations, are also not reported. The record gives a one-hour individual data-collection timeframe and a six-month cohort collection period, but does not provide the detailed timing or completeness of measurements.

Generalizability is also difficult to assess because the population is described as healthy volunteers, without additional participant characteristics in the record. The registry’s Calibration and Validation labels do not establish a randomized comparison; allocation is NA, and the record explicitly warns that the arms are not a like-for-like control comparison. These features constrain any comparative interpretation.

Finally, the adverse-event entries do not include event terms or further information on ascertainment, follow-up, or death counts. Death counts are not reported. The registry record is therefore insufficient for a comprehensive safety interpretation as well as a quantitative assessment of sensor accuracy.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Nonin 4 Wavelength Cerebral Oximeter Study — NCT01762722.
• Registry record last updated 2013-04-16; status last verified 2013-03.
• Results first posted 2013-04-16.
• 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 registry describes a completed, single-group study of a four-wavelength non-invasive cerebral oximeter in healthy volunteers. The intervention was reduction of inspired oxygen, and the stated aim was to compare forehead sensor readings with simultaneous arterial and venous blood saturation measurements. The primary outcome, Accuracy of Sensor, uses ARMS to describe dispersion around the line of identity.

Enrollment is reported as 23; the record lists Calibration and Validation labels, but does not describe a randomized, like-for-like comparison. It reports one outcome measure and no analyses. A numerical primary accuracy result and its uncertainty are not reported in the registry record. Adverse-event fields report 0 serious and 0 other events in each named group; death counts are not reported.