Accuracy of Pulse Oximeters With Profound Hypoxia During Motion: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Accuracy of Pulse Oximeters With Profound Hypoxia During Motion (NCT04814342), a registered phase not stated study. The primary outcome is Accuracy of Sensor Oxygen Saturation by Arms Calculation.

2. Clinical Question

The study concerns whether a pulse oximeter can measure oxygen saturation accurately during controlled motion in conditions involving hypoxia. The registry describes a comparison between the device’s noninvasive SpO2i measurement and oxygen saturation obtained from a blood sample, SRi. Accuracy is defined using an arithmetic root mean square error, or Arms, over the stated 30 seconds. The listed primary outcome is “Accuracy of Sensor Oxygen Saturation by Arms Calculation.”

This is a measurement-accuracy question, not a treatment-effect question. The registry record does not report a numerical Arms result, so the extent of agreement between device readings and the blood-sample reference cannot be quantified from the posted outcome information.

3. Trial Design

NCT04814342 is a completed device study sponsored by Nihon Kohden. The registry lists an enrollment of 5. Its recorded start date is 2021-03-24, and its primary completion date is 2021-05-24. The condition listed is Hypoxia. The brief summary describes evaluating pulse oximeters with patients during controlled motion, while the safety group is titled Adult Healthy Subjects. The record does not provide additional detail about the participants or explain the wording difference.

The intervention is a pulse oximeter. The study’s primary outcome is sensor oxygen-saturation accuracy by Arms calculation; its secondary outcome is sensor pulse-rate accuracy by arms calculation. The registry lists 2 outcome measures. It does not describe distinct randomized treatment groups, and allocation is not stated. This is therefore not a randomized comparison between like-for-like control arms. The blood-sample measurement serves as the reference for the primary device-accuracy calculation, rather than as a separate treatment arm.

The record identifies the primary outcome population as “Qualified subjects” and gives a timeframe of 30 seconds. It does not describe the detailed measurement schedule, how motion was applied, the procedures used to establish profound hypoxia, or how many readings contributed to an individual Arms calculation. Those protocol details are not reported in the registry record.

The study status is COMPLETED, but completion status alone does not establish that a numerical endpoint result is available. The registry record lists zero analyses and does not report the primary outcome’s numerical value or a numerical result for the secondary outcome. It also does not report the rationale used to select the enrollment size, so the enrollment should not be taken as evidence of a particular precision target or statistical design objective.

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

Analysis population for the primary outcome, as defined in the registry: Qualified subjects.

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: Qualified subjects.

6. Primary Endpoint

The registry lists the primary outcome as Accuracy of Sensor Oxygen Saturation by Arms Calculation, measured over 30 seconds.

The primary endpoint is “Accuracy of Sensor Oxygen Saturation by Arms Calculation.” It compares the pulse oximeter’s noninvasive oxygen-saturation reading, SpO2i, with the oxygen-saturation value from a blood sample, SRi. The unit is percentage of RMS error, and the listed timeframe is 30 seconds. The registry defines Arms as the square root of the sum of the squared differences across n samples, divided by n. In general, this kind of root mean square summary reflects the size of discrepancies while preventing positive and negative differences from cancelling each other out.

For an accuracy assessment, a reference measurement is needed to judge how close a device reading is to the target value. A blood-sample result plays that reference role in this endpoint. Arms is an error measure: smaller errors would indicate closer readings in the observations included in the calculation, but the record gives no numerical Arms result with which to assess the observed closeness. The statistic also does not, by itself, describe whether readings tend to be higher or lower than the reference, because squaring removes the sign of each difference.

The registry does not report a confidence interval, dispersion measure, or other uncertainty estimate for this endpoint. It therefore does not show how precisely the observed accuracy has been estimated. The endpoint description identifies the calculation, but does not specify how repeated measurements were grouped, how missing readings were handled, or how the result was aggregated across qualified subjects. These details are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary statistical quantity is an arithmetic root mean square error, abbreviated Arms in the registry’s endpoint description. In general, an error statistic compares measurements with a reference. Here, each pulse oximeter SpO2i reading is compared with a blood-sample SRi value. The signed difference is squared, the squared differences are averaged across n samples, and the square root is taken. The registry describes the result as a percentage of RMS error and gives a timeframe of 30 seconds.

What does root mean square error convey?

As a general statistical concept, squaring errors makes their magnitudes positive, so readings on opposite sides of a reference do not cancel out. Larger discrepancies have more influence on the average squared error than smaller discrepancies. Taking the square root returns the summary to the scale of the underlying measurement differences. Arms therefore summarizes the typical magnitude of discrepancy in the observations used, rather than a signed tendency to read high or low.

This distinction matters for interpretation. A mean signed difference can reveal an average direction of bias, whereas a root mean square error combines discrepancy magnitude without preserving the direction of each difference. The registry specifies the latter calculation, but does not report a separate signed-bias statistic. In general, two devices or datasets could have similar root mean square errors while differing in whether readings tend to fall above or below the reference. No such comparison can be made from the posted results in this record.

Why use a reference measurement?

In general, an accuracy study needs a reference against which device readings can be assessed. The endpoint identifies the blood-sample SRi value as that reference and the pulse oximeter SpO2i reading as the noninvasive measurement under evaluation. The calculation only summarizes the readings included in it; interpretation also depends on how well the reference is measured and how the observations were collected. The registry record does not provide further details about reference measurement procedures, calibration, timing alignment, or quality control.

What does a confidence interval add?

In general, a confidence interval expresses the uncertainty around an estimated quantity under the assumptions of the statistical method used. An error estimate without an uncertainty interval gives less information about its precision: it reports a summary value but does not show the range of values compatible with the data and analysis. The registry record does not report a confidence interval or another precision estimate for the primary Arms outcome. It also does not report a numerical Arms estimate, so neither the magnitude nor the precision of the endpoint can be assessed from the posted results.

Variation among repeated readings and variation among participants are distinct sources of uncertainty in general. How those sources should be handled depends on the study design and analysis. For example, readings taken repeatedly from the same participant are not necessarily independent observations. The registry record does not describe the measurement structure in enough detail to determine how repeated readings were treated, or whether the analysis accounted for clustering within participants.

Why can over-dispersion matter?

In general, over-dispersion occurs when observed variability is greater than a statistical model assumes. If a model treats correlated or highly variable observations as more independent information than they are, uncertainty can be understated. That can make an interval appear narrower or a test appear stronger than warranted. This is a general caution, not a statement that over-dispersion occurred in this study. The registry record does not identify a statistical model, report a dispersion parameter, or describe checks for model fit.

The endpoint is a calculated error summary rather than a reported treatment contrast. The registry does not report a test statistic, p-value, or inferential comparison for the primary outcome. The device-accuracy question is therefore not answerable by importing the logic of a randomized treatment comparison. The measurement reference and the device reading are paired conceptually in the endpoint definition, but the details of pairing, analysis, and any confidence interval method are not reported.

What can be said about planning?

In general, sample-size planning uses assumptions about the target effect, variability, significance level, allocation, and analysis method. The planning illustrations associated with the results are explicitly illustrations based on an observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. Those are planning assumptions, not evidence about the study’s original design. The sample-size rationale for this trial is not reported in the registry record. Any planning illustration should be read as a hypothetical planning aid, not as observed power or a description of the enrolled study’s design.

The registry record does not report the statistical analysis plan, the choice of any model, handling of multiple measurements, assumptions underlying the error calculation, or a plan for confidence intervals. It lists zero analyses. Accordingly, a detailed reconstruction of the statistical approach is not possible. General statistical concepts explain what Arms and uncertainty measures mean, but they cannot substitute for the missing numerical endpoint results and analysis details.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists 2 outcome measures: primary accuracy of sensor oxygen saturation by Arms calculation, and secondary accuracy of sensor pulse rate by arms calculation. It does not report numerical results or analyses for either measure; the recorded number of analyses is zero. The primary endpoint is described in more detail than the secondary outcome, but no analysis results are posted for the primary question.

When several outcomes are assessed, multiplicity can arise because examining multiple statistical tests increases the chance of at least one apparently notable result by chance. In general, a prespecified hierarchy or adjustment can address that issue, depending on the objectives and analysis plan. The registry record does not report whether any multiplicity procedure was planned or used, whether the secondary outcome was intended to support a confirmatory claim, or how outcomes were prioritized beyond the primary and secondary labels. No adjustment should be assumed.

Missing-data methods are also not described. The registry does not report whether any readings or participant data were unavailable, how missing measurements were handled, or whether any sensitivity analysis was planned. The absence of these details does not establish that data were missing; it means the approach cannot be assessed from the registry record.

The analysis populations are not reported in the registry record. The primary outcome description identifies “Qualified subjects,” and the safety grouping is Adult Healthy Subjects, but the record does not explain inclusion in an analysis set, exclusions, or whether all enrolled participants contributed to each endpoint. Allocation is not stated, and the record does not describe a randomized arm-based analysis. Without these details and without numerical outcome estimates, neither multiplicity control nor the handling of missing data and analysis populations can be evaluated.

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.

The registry’s adverse-event record lists Adult Healthy Subjects, with 5 at risk for serious events and 0 affected. For other events, it lists 5 at risk and 0 affected. The recorded number of deaths is 0, and no serious-event terms are listed. These are the registry’s reported counts for the listed group.

In general, an at-risk count describes the group included in the event reporting, while an affected count describes how many in that group were recorded as experiencing the event category. Zero recorded events is a description of the entries in this record; it is not proof that no event could occur or that every type of harm was captured. The record does not give additional information about event-collection procedures, observation duration, or how adverse events were defined.

These safety counts should be kept separate from the device-accuracy endpoint. They do not show how closely SpO2i readings matched SRi values, and they do not provide a statistical estimate of measurement accuracy. With an enrollment of 5, the safety listing describes a limited group and cannot establish the frequency of uncommon events. No broader safety conclusion can be drawn from the counts alone.

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

The registry record establishes that NCT04814342 was a completed pulse oximeter device study with an enrollment of 5 and a primary endpoint based on comparing SpO2i readings with blood-sample SRi values. The endpoint uses Arms error over 30 seconds. The study aim refers to profound hypoxia and controlled motion. This information clarifies the question the study was designed to address, but does not provide the measured answer.

The numerical primary outcome is not reported in the registry record. Therefore, the posted results do not show the magnitude of oxygen-saturation error, the direction of any measurement bias, or the precision of an estimate. A reader cannot determine from the endpoint definition alone whether readings were close to or far from the reference. The record also does not provide a numerical result for sensor pulse-rate accuracy, the listed secondary outcome.

The study’s non-randomized measurement design should not be interpreted as a treatment comparison. There is no reported allocation to arms, and the record notes that arms are not a like-for-like control comparison. The central question is device measurement accuracy against a reference, not whether one treatment changes a clinical outcome relative to another treatment.

The enrollment and safety listing describe a small recorded group. The registry does not report enough detail about participants, measurement conditions, repeated observations, or analysis populations to determine how representative the observations might be of other settings. Although the title and brief summary refer to profound hypoxia and motion, specific operating conditions and subgroup results are not reported in the registry record.

The safety record lists 0 serious events, 0 other events, and 0 deaths among the listed group, but those entries do not establish that all risks are absent. Likewise, completion status does not mean that a numerical accuracy result has been posted. The defensible interpretation is limited to the study objective, endpoint definition, enrollment, and recorded safety counts. Claims about device performance or clinical consequences would require results and context that are not reported in the registry record.

13. Why This Trial Matters Statistically

Measurement accuracy is a statistical question with direct methodological importance: a device reading is meaningful only in relation to a reference and a clear summary of discrepancy. This study’s endpoint makes that structure explicit by comparing noninvasive SpO2i measurements with blood-sample SRi values and summarizing error with Arms. The endpoint therefore illustrates the difference between evaluating a measurement instrument and comparing clinical interventions.

The motion and hypoxia context also highlights why measurement performance should be interpreted under the conditions in which it is assessed. General performance claims cannot be inferred solely from the name of an endpoint or from a study’s stated aim. The relevant conditions, measurement procedures, and analysis determine what the resulting estimate would describe. Detailed protocol conditions are not reported in the registry record, so the scope of any conclusion cannot be specified further.

The record also illustrates the role of uncertainty reporting. A point estimate of error would describe a measured summary, while a confidence interval would help convey its precision. Neither a numerical primary estimate nor a confidence interval is reported. With an enrollment of 5, the limited enrollment is an additional reason to avoid broad claims about how accurately the device might perform in other groups or settings.

Finally, the registry separates outcome measures from analyses: it lists a primary oxygen-saturation measure and a secondary pulse-rate measure, but zero analyses. Without numerical results, analysis methods, or a missing-data description, readers cannot evaluate the size, uncertainty, or robustness of the findings. This makes the record useful for identifying the intended statistical question while also showing why endpoint definitions alone cannot answer it.

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 interpretation is that the registry record does not report the numerical primary Arms result. It also does not report a numerical result for the secondary pulse-rate accuracy outcome. No uncertainty interval, participant-level result, or other precision estimate is available. Consequently, the magnitude and precision of device-measurement error cannot be judged from the posted results.

The enrollment was 5. The registry does not report a sample-size rationale or explain what precision or performance target informed enrollment. The planning assumptions listed for illustrations do not establish the original study design or provide observed power. The analysis population is also not described beyond the primary outcome’s reference to qualified subjects and the safety grouping titled Adult Healthy Subjects.

The record does not provide detailed procedures for controlled motion, the conditions used to establish profound hypoxia, or the timing and number of observations contributing to each calculation. It does not say how repeated observations were handled, whether all enrolled participants contributed to the endpoint, or how missing readings were addressed. These details matter for judging the representativeness and uncertainty of a measurement-accuracy estimate.

Allocation is not stated, and this is not a randomized comparison with like-for-like control arms. The blood-sample value serves as the reference in the primary calculation; it does not constitute a separate randomized group. The registry does not report the statistical analysis plan, multiplicity handling, confidence-interval method, or any model diagnostics.

The safety record lists zero affected participants for serious and other events and zero deaths in the recorded group. The record does not provide further information about event definitions or collection procedures. These entries are not a substitute for detailed safety reporting or for the missing accuracy outcomes. Together, these limitations mean that the registry supports description of the study question and design elements, but not a quantitative conclusion about accuracy or its generalizability.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Accuracy of Pulse Oximeters With Profound Hypoxia During Motion — NCT04814342.
• Registry record last updated 2025-09-12; status last verified 2025-09.
• Results first posted 2023-11-21.
• 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

NCT04814342, sponsored by Nihon Kohden, is listed as COMPLETED. The registry records an enrollment of 5, a start date of 2021-03-24, and a primary completion date of 2021-05-24. The study concerns pulse oximeter accuracy in hypoxia and motion. Its primary endpoint compares noninvasive SpO2i readings with blood-sample SRi measurements using an arithmetic root mean square error over 30 seconds. The secondary outcome concerns sensor pulse-rate accuracy.

The registry record does not report numerical results for either outcome and lists zero analyses. It does not report the statistical analysis plan, missing-data approach, or analysis populations. The safety listing records 0 serious events, 0 other events, and 0 deaths among the listed Adult Healthy Subjects group, with 5 at risk in each event category. The record defines the measurement question, but does not provide a numerical basis for assessing accuracy or its precision.