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 Accuracy of ScanWatch Pulse Oximeter With Profound Hypoxia (NCT04380389), a na na study. The primary outcome is RMSE Compared to CO-oximetry.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT04380389 |
| Phase | NA |
| Condition | Oximetry |
| Allocation | Not applicable (single group) |
| Design model | Single Group |
| Masking | None (open label) |
| Enrollment | 14 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Withings |
| Results posted | Yes |
2. Clinical Question
Population
The protocol describes healthy volunteers undergoing brief, stable arterial oxygen desaturation. The registry records an enrollment of 14. The reported population for the primary endpoint is not specified in the registry record.
Intervention
The intervention is hypoxia, listed as a diagnostic test. During the protocol, the ScanWatch pulse oximeter is tested while a stable level of hypoxia is attained. The record describes sampling arterial blood at that point.
Comparator
The device’s functional blood oxygen saturation (SpO2) is compared with arterial blood oxygen measured by a bench CO-oximeter. The registry describes the CO-oximeter as the gold-standard measurement. This is a measurement reference, not a randomized control group.
Primary question
The primary endpoint is RMSE (%) between the device’s SpO2 and the arterial blood oxygen measurement. It asks how far the device readings, considered together with their errors, differ from the reference readings under the test conditions. The numerical RMSE result is not reported in the registry record.
3. Trial Design
This is a completed, single-group performance test rather than a randomized treatment comparison. The registry lists allocation as not applicable and masking as none. It records enrollment of 14 and describes a protocol in which healthy volunteers undergo brief, stable arterial oxygen desaturation. Arterial blood is sampled once a stable level of hypoxia has been attained, and the sample is analyzed for oxygen saturation using a bench CO-oximeter.
The intervention is listed as hypoxia, with a profound-hypoxia arm description. No separate randomized control arm is described. The left-versus-right-hand label appears in the adverse-event record, but it does not establish a randomized, like-for-like control comparison. The registry explicitly cautions that the arms are not a like-for-like control comparison. Results should therefore be read as a device-performance assessment against a measurement reference, not as an estimate of a treatment effect between independent trial arms.
The registry lists the start date as 2020-03-11 and the primary completion date as 2020-03-12; status is completed. The phase is recorded as not applicable. The brief summary says the protocol is aligned with the latest ISO and FDA guidance documents for pulse-oximeter testing. That description concerns the stated protocol; it does not, by itself, report a numerical accuracy result or establish what the test found.
The registry record does not report a detailed sampling schedule, the number of arterial samples per participant, or further procedural details. It also does not report a statistical analysis plan or a sample-size justification. The primary endpoint is identified, but its numerical result and an accompanying analysis are not listed.
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 | Left vs Right Hand |
|---|---|
| STARTED | 14 |
| COMPLETED | 14 |
| NOT COMPLETED | 0 |
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 | Left vs Right Hand |
|---|---|---|
| Participants analysed | 14 | |
| Age, Continuous | year | 27.1 (SD 3.9) |
| Sex: Female, Male: Female | Participants | 6 |
| Sex: Female, Male: Male | Participants | 8 |
| Race and Ethnicity Not Collected | Participants | — |
| Skin tone: Fair skin tone | Participants | 4 |
| Skin tone: Medium skin tone | Participants | 7 |
| Skin tone: Dark skin tone | Participants | 3 |
| BMI (kg/m²) | kg.m-² | 23.4 (SD 2.1) |
| Height | cm | 168.7 (SD 9.8) |
| Weight | kg | 66.6 (SD 8.6) |
6. Primary Endpoint
The registry lists the primary outcome as RMSE Compared to CO-oximetry, measured over 1 hour.
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | not applicable |
| Unit | RMSE (%) |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint is root mean squared error between functional blood oxygen saturation (SpO2) from the device under test and arterial blood oxygen measured with the CO-oximeter. Its unit is RMSE (%), and its timeframe is 1 hour. The endpoint is intended to describe the size of the device’s errors relative to a reference measurement under the test conditions.
Generally, RMSE combines the size of individual errors into one summary, with larger errors contributing more strongly than smaller ones. It is useful in measurement studies because it expresses discrepancy on the measurement scale and reflects both bias and variability in the readings. It does not, by itself, show whether errors are consistently high or low, how errors vary across oxygen-saturation levels, or how precisely the summary has been estimated. Those questions require additional results or analyses.
The registry record does not report the observed RMSE, its uncertainty, the number of paired measurements used in the calculation, or a breakdown of errors across measurement levels. Consequently, the endpoint’s intended role can be described, but the recorded information does not permit a quantitative assessment of the observed accuracy or its precision.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry identifies a focused measurement question: how closely ScanWatch SpO2 readings correspond to arterial blood oxygen measured by a bench CO-oximeter during profound hypoxia. The primary outcome is RMSE (%), with a timeframe of 1 hour. However, the numerical RMSE result is not reported in the registry record. Without that estimate, the direction and magnitude of the observed discrepancy cannot be described, and no conclusion about the measured accuracy can be drawn from the posted results.
Precision is also not reported. There is no confidence interval or other uncertainty estimate for the endpoint in the registry record. Generally, a confidence interval indicates a range of values compatible with the data under the assumptions of an analysis; it helps distinguish a precise estimate from one that remains uncertain. Since neither an estimate nor an interval is recorded, the degree of statistical uncertainty around the endpoint cannot be assessed.
The design informs how to interpret the question. This is a single-group performance test with no randomized control comparison. The CO-oximeter is a measurement reference, not a treatment arm. Thus, the central contrast concerns readings from two measurement approaches, not outcomes between allocated groups. The adverse-event record’s left-versus-right-hand label does not change the trial into a randomized comparison, and the registry cautions against treating its arms as like-for-like controls.
The registry describes testing in healthy volunteers during brief, stable desaturation. That setting defines the scope of the protocol as recorded. The record does not report enough detail to determine how the results would vary across other populations, settings, oxygen levels, or measurement conditions. More generally, the performance of a device under one protocol should not automatically be interpreted as performance in every context.
The absence of a reported result is not evidence that the device had a particular error level, and it is not evidence that error was absent. It means only that the primary numerical finding cannot be evaluated from the registry record. There are no listed analyses, and no analysis results are reported. The record also does not give details about the number of paired readings, missing measurements, or the method used to produce the endpoint summary.
Accordingly, the defensible statistical reading is limited: the trial was designed as a comparison of device readings with a reference measurement during profound hypoxia, and RMSE was the primary endpoint. The registry records completion and enrollment of 14, but does not report the RMSE value or its precision. No claim about the amount of agreement, the consistency of measurement errors, or performance beyond the stated test conditions can be supported from these posted results.
9. Statistical Methods Explained
What does RMSE describe?
RMSE is a summary of the differences between paired measurements. In this trial, the pair consists of a functional SpO2 reading from the device under test and arterial blood oxygen measured with a CO-oximeter. Generally, RMSE squares each difference, averages those squared differences, and takes the square root. The result is expressed on the scale of the measured quantity, here recorded as RMSE (%).
Because differences are squared before averaging, comparatively large errors have more influence on RMSE than small errors. The measure combines information about the overall size of discrepancies, but it does not reveal their direction. For example, the same general RMSE could arise from readings that tend to be above the reference, readings that tend to be below it, or a mix of both. This is why an RMSE summary is not a complete description of agreement.
The registry record identifies RMSE as the primary endpoint but does not report its numerical value. It also does not report the paired observations, a confidence interval, or supporting summaries of bias and variation. The endpoint definition therefore clarifies what the study intended to summarize, but cannot establish the magnitude or precision of the result.
Why use a reference measurement?
In a measurement-performance study, a reference method supplies the comparator against which a device reading is evaluated. Here, the registry describes arterial blood sampling and measurement by a bench CO-oximeter, which it calls the gold standard. The comparison is between measurements, not between assigned treatments. Generally, this distinction matters because study designs and interpretations for measurement agreement differ from those for estimating the effects of an intervention.
The record describes collecting arterial blood after a stable level of hypoxia has been attained. It does not state how many paired readings were obtained, how repeated readings were handled, or how the endpoint calculation accounted for multiple observations from the same volunteer. Those details are not reported in the registry record, so the independence and structure of the observations cannot be assessed from the record.
What does the study design imply?
The study is listed as single-group, with allocation not applicable and masking none. A single-group measurement test can evaluate device readings against a reference, but it does not provide a randomized treatment comparison. The registry explicitly cautions that the listed arms are not a like-for-like control comparison. In particular, the left-versus-right-hand adverse-event label should not be treated as evidence of random allocation or as a basis for a treatment-effect calculation.
Generally, randomization supports comparisons by balancing measured and unmeasured characteristics across assigned groups, while masking can reduce some forms of measurement or assessment bias. Neither feature is listed for this study. This does not by itself determine the validity of the measurement comparison, but it does define what kinds of causal conclusions the design can support. The stated question is device accuracy against a reference during a specified testing protocol.
How should statistical significance and precision be read?
The registry record lists no analyses and does not report a p-value, confidence interval, or other uncertainty estimate for the primary endpoint. Statistical significance cannot be assessed from an absent result. More generally, a point estimate describes the observed magnitude in the analyzed data, while an uncertainty interval describes how much the estimate might vary under the statistical model and sampling process. Both are needed for an informative account of precision.
The record does not report a prespecified threshold for acceptable RMSE or a hypothesis test for the primary measure. No such threshold or testing procedure should be inferred. The phrase “aligned with the latest ISO and FDA guidance documents” appears in the protocol summary, but the registry does not state a numerical acceptance criterion in the endpoint information.
What do the planning assumptions represent?
Planning illustrations associated with this record are described as calculations based on an observed effect, not as the trial’s own design. Their stated assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect to detect expressed as a stated fraction of the observed effect. For time-to-event illustrations, the stated quantity is the total number of events; other illustrations are described in participants per arm.
These assumptions do not describe the completed single-group measurement protocol, which has no randomized two-arm allocation. The observed RMSE is not reported in the registry record, so the record does not permit a reader to evaluate the effect on which those illustrations are said to be based. Planning values are not observed power, and they do not substitute for the trial’s measured endpoint or an uncertainty interval. The registry record does not report the trial’s own sample-size justification or statistical plan.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists one primary outcome measure: RMSE compared with CO-oximetry. It lists no analyses, and no numerical result for that endpoint is reported. The record does not report whether additional outcomes were prespecified outside the listed primary measure, whether multiple hypotheses were tested, or whether any adjustment for multiplicity was planned. Therefore, the multiplicity strategy is not reported in the registry record.
Missing-data handling is also not described. The record does not state whether any device readings or arterial samples were missing, how incomplete paired measurements would have been handled, or whether any imputation method was used. The number of paired observations contributing to the endpoint is not reported. Without that information, the completeness of the primary analysis cannot be assessed.
The analysis population is not specified in the registry record. The record gives enrollment of 14 and describes a healthy-volunteer protocol, but does not say which participants or measurements were included in the primary endpoint calculation. It does not report an exclusion rule, a per-protocol definition, or another analysis-population definition. These omissions limit assessment of how the endpoint would relate to the enrolled population.
Generally, multiplicity matters when several outcomes or tests are examined because the chance of a nominally notable result can increase as the number of analyses grows. Missing-data methods matter because conclusions can depend on which measurements are unavailable and how those absences are handled. Those concepts are relevant to evaluating a study, but the registry record does not provide the details needed to determine how they applied in this trial.
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 |
|---|---|---|---|
| Left vs Right Hand | 0 | 14 | 0.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Left vs Right Hand | 0 of 14 |
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 |
|---|---|
| Left vs Right Hand | 0 of 14 |
The registry’s adverse-event entry is labeled “Left vs Right Hand.” It records 14 at risk for the serious-event category and 0 affected, and 14 at risk for the other-event category and 0 affected. Deaths are recorded as 0. These are the counts listed in the adverse-event record; they do not provide a description of event definitions, monitoring procedures, or follow-up duration.
These figures should be read as recorded counts, not as proof that no adverse events could occur or that the study ruled out safety risks. A zero count means no affected participants are recorded for that category in this entry. It does not quantify how precisely an event rate was estimated, and the registry record does not report an uncertainty interval for these counts.
The left-versus-right-hand label is not evidence of a randomized control comparison. The registry cautions that the arms are not a like-for-like control comparison, and the design is listed as single-group. The safety figures should therefore not be interpreted as a comparative estimate of risk between randomized groups. Further safety details, including event definitions and observation duration, are not reported in the registry record.
12. What the Result Does, and Does Not, Mean
The record describes a completed performance test of a pulse oximeter during profound hypoxia. Its primary question concerns the agreement between device-reported functional SpO2 and arterial blood oxygen measured by a bench CO-oximeter. The chosen summary is RMSE (%), with a stated timeframe of 1 hour. This identifies the measurement question and the intended summary of error.
The record does not report the numerical RMSE result. As a result, it does not show the observed size of the discrepancy or the precision of that estimate. The registry also does not report a confidence interval, a numerical breakdown of errors, or the number of paired measurements used. These omissions prevent a quantitative interpretation of accuracy from the registry record.
The design does not compare randomized treatment groups. The CO-oximeter is the measurement reference, and the trial is listed as single-group with allocation not applicable. The adverse-event label involving left and right hands should not be read as a randomized control contrast. Thus, the study’s stated statistical purpose is measurement performance in the test protocol, not estimation of a treatment effect.
The protocol summary describes brief stable desaturation in healthy volunteers and says the protocol is aligned with the latest ISO and FDA guidance documents. That description does not supply an endpoint result, and it does not establish the device’s performance outside the described test conditions. The record does not report further details that would allow generalization to other populations or measurement settings.
The study’s completion status and enrollment are recorded, but neither replaces the missing primary result. The registry record supports describing what was tested and how the endpoint was defined. It does not support stating whether the observed error was small or large, whether the measurements met an acceptance criterion, or whether performance was consistent across the range of readings. Those conclusions require results and analysis details that are not reported in the registry record.
13. Why This Trial Matters Statistically
Pulse oximetry is a measurement problem: the device reading is compared with a reference measurement, and the statistical question concerns the size and behavior of measurement error. This trial makes that structure explicit by defining RMSE between device SpO2 and arterial blood oxygen measured with a bench CO-oximeter as its primary endpoint. That endpoint is meaningful as an error summary, but its interpretation depends on the reported value and its precision.
The study also illustrates why the comparator must be described accurately. A reference instrument is not the same as a randomized control arm. Here, the registry lists a single-group design and says allocation is not applicable. The measurement comparison can address correspondence with the reference under the test protocol, but it should not be recast as a causal comparison of treatments or as a randomized left-versus-right-hand contrast.
The protocol’s focus on stable profound hypoxia in healthy volunteers makes the testing context central to interpretation. Statistical conclusions are tied to the measurements and conditions observed; extending them beyond those conditions requires evidence not reported in this record. The brief summary’s statement that the protocol is aligned with guidance documents is contextual information, not a substitute for numerical results.
Finally, the record shows the difference between a defined endpoint and an interpretable result. Naming RMSE establishes what the primary measure was intended to capture, while the absence of the estimate and uncertainty leaves the performance question unanswered in the registry record. Reporting the endpoint value, its precision, the number of paired observations, and relevant analysis details would be necessary for readers to judge the result statistically.
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 main limitation for statistical interpretation is that the primary RMSE result is not reported in the registry record. There is no numerical estimate or uncertainty interval, so the magnitude and precision of the measured error cannot be evaluated. The record also does not report the number of paired device and arterial measurements used to calculate the endpoint.
Details of the primary analysis are not reported. The registry does not specify how repeated readings were handled, what analysis population was used, whether any observations were excluded, or how missing measurements were addressed. It does not state a threshold for acceptable RMSE or describe a hypothesis test for the endpoint. These gaps prevent assessment of the analytic choices behind the intended summary.
The study is described as a single-group performance test, with no allocation and no masking. The reference measurement is not a randomized control group. The registry’s caution that the arms are not a like-for-like control comparison is important when reading the left-versus-right-hand adverse-event label. That label does not support a randomized group comparison.
The protocol summary describes healthy volunteers undergoing brief, stable arterial oxygen desaturation. The registry does not provide further detail about the enrolled population, measurement conditions, or representativeness beyond that description. The extent to which findings would apply in other populations or settings is therefore not reported in the registry record. The stated timeframe for the primary measure is 1 hour, but further timing details are not reported.
The record lists no analyses, and does not report a statistical plan, multiplicity approach, or sample-size justification for the trial. Planning illustrations are described as being based on an observed effect, but the primary observed result itself is not reported. These limitations mean that the registry supports a description of the intended measurement comparison, not a quantitative conclusion about device performance.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Accuracy of ScanWatch Pulse Oximeter With Profound Hypoxia — NCT04380389.
- Registry record last updated 2026-05-04; status last verified 2026-04.
- Results first posted 2026-05-04.
- 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
This completed, single-group performance test examined ScanWatch pulse-oximeter readings during profound hypoxia in healthy volunteers. The registry records enrollment of 14 and describes arterial blood sampling at a stable level of hypoxia, with oxygen saturation measured using a bench CO-oximeter. The primary endpoint is RMSE (%) between device SpO2 and the reference measurement, with a timeframe of 1 hour.
Allocation is listed as not applicable, masking as none, and the phase as not applicable. The protocol summary describes alignment with the latest ISO and FDA guidance documents. The registry lists no analyses and does not report the numerical primary endpoint result, its precision, or details of missing-data handling and analysis populations. The adverse-event entry records zero affected for serious and other events and zero deaths, with the listed at-risk counts recorded under “Left vs Right Hand.”