TatchSleep as a Tool to Aid in Sleep Apnea Analysis: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for TatchSleep as a Tool to Aid in Sleep Apnea Analysis (NCT04419714), a na na study. The primary outcome is Correlation Between Polysomnography (PSG) and Tatch Apnea/Hypopnea Index (AHI).

2. Clinical Question

In practical statistical terms, the study asks whether two measurement methods produce related respiratory-index values when used during the same sleep study. Correlation addresses whether values tend to vary together; it does not by itself establish that the values are close enough to be interchangeable. Agreement is therefore a separate part of the stated question. The registry also lists agreement in AHI severity classification as secondary outcomes, including percent agreement and Cohen’s kappa. The record does not report numerical results for those outcomes, either.

3. Trial Design

The registry describes a completed, single-center, single-arm quantitative study of TatchSleep Pro as an aid to sleep apnea analysis. The listed enrollment is 53. The dates recorded for the study are 2021-06-10 for start and 2021-11-04 for primary completion. The listed phase and allocation are NA, and masking is listed as none. These fields do not describe a randomized, masked comparison between treatment arms.

There was one test arm, named TatchSleep Pro Test Arm. Its description says that adult male and female participants wore the device while simultaneously undergoing an in-lab PSG. PSG is the comparator measurement for the device assessments, but it is not a separate assigned control arm. Thus, the design is a paired measurement comparison: each participant’s device measurement is considered alongside a PSG measurement taken during the same sleep study. This arrangement can support evaluation of association and agreement between measurement methods, but it does not test a treatment effect against a control group.

The primary outcome is described as correlation between PSG and TatchSleep AHI, with a 12-hour timeframe. The outcome description specifies AHI using a 3% desaturation criterion, Bland–Altman analysis for agreement, and Pearson correlation coefficients for AHI and ODI using a 3% criterion. The registry lists two secondary outcome topics: percent agreement in AHI severity classification and Cohen’s kappa for severity agreement.

The record lists three outcome measures but reports zero analyses. It does not include numerical primary or secondary outcome estimates. Additional details about recruitment, measurement procedures, analysis populations, or protocol deviations are not reported in the registry record. The listed enrollment should therefore not be treated as the number with usable paired measurements or as the denominator for every outcome.

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

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Subjects that used TatchSleep Pro while simultaneously undergoing a Polysomnography (PSG) sleep study

5. Baseline Characteristics

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

Population: Males and females, 18 years of age or older. Measure Analysis Population Description: Overall tested population was 53 but protocol deviation, equipment errors, and one withdrawal lead to 45 being analyzed

6. Primary Endpoint

The registry lists the primary outcome as Correlation Between Polysomnography (PSG) and Tatch Apnea/Hypopnea Index (AHI), measured over 12 hours.

The primary endpoint concerns the relationship between the AHI measured by TatchSleep Pro and the AHI measured by in-lab PSG. AHI is the apnea/hypopnea index, and the stated assessment uses a 3% desaturation criterion. The outcome description also identifies ODI, the oxygen desaturation index, and specifies Pearson correlation coefficients for AHI and ODI. The timeframe recorded for this outcome is 12 hours.

Comparing a device with PSG is useful because the study’s question is whether device-derived respiratory indices correspond to measurements from the in-lab reference method. Pearson correlation describes the degree to which two measurements vary together in a linear pattern. A strong correlation can coexist with systematic differences between methods, so it is not sufficient by itself to show close agreement. The registry accordingly states that agreement will also be examined using Bland–Altman analysis.

The registry names correlation as the primary endpoint but does not report the correlation coefficient or its 95% confidence interval. It likewise does not report numerical Bland–Altman results, or numerical results for ODI. Consequently, the primary endpoint’s observed magnitude and precision cannot be assessed from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome names Pearson correlation and Bland–Altman analysis. These methods answer related but distinct questions about paired measurements. Understanding that distinction helps prevent over-reading a correlation result as proof that two measurement methods can be substituted for one another.

What does correlation describe?

Pearson correlation measures the direction and strength of a linear relationship between two quantitative variables. In this study’s stated plan, the variables are device-derived and PSG-derived respiratory indices. If participants with larger PSG values also tend to have larger device values, the measurements may be positively associated. Correlation is a measure of co-variation, not a direct measure of how close paired values are.

A constant offset between methods can leave the correlation high even though the two methods differ systematically. Likewise, a broad range of values can contribute to a high correlation while individual paired readings remain far apart. Conversely, a restricted range can make correlation appear modest even when paired values are relatively close. These are general statistical considerations; the registry record does not provide the data needed to assess whether any of these patterns occurred in this study.

Why examine agreement separately?

Bland–Altman analysis focuses on the difference between paired measurements and how those differences relate to the measurements’ level. In general, this approach can help identify a consistent bias and describe the spread of differences across observations. Its interpretation depends on the measurement context and on whether the observed differences would be acceptable for the intended use. The registry states that Bland–Altman analysis would be used, but numerical estimates and plots are not reported in the registry record. No conclusion about bias or the range of differences can therefore be drawn.

The two methods should be read together rather than treated as interchangeable evidence. Correlation asks whether measurements move together; agreement asks how similar the paired values are. A method-comparison study often needs both perspectives to characterize measurement performance. Neither statistic alone establishes clinical usefulness, and the record does not describe a clinical decision threshold or decision-impact analysis.

What do the secondary outcomes add?

The listed secondary outcomes focus on agreement in AHI severity categories. Percent agreement is the proportion of classifications that match. It is straightforward to understand, but its value can depend on how common each category is and how the classifications are distributed. Cohen’s kappa adjusts observed categorical agreement for agreement expected by chance under the method’s assumptions. Kappa is not simply another form of percent agreement: the two measures can differ, particularly when category prevalence or marginal distributions are uneven.

The registry names both analyses but gives no result estimates, category counts, or uncertainty intervals. Thus, there is no basis for evaluating either raw classification agreement or chance-adjusted agreement in this study. The severity categories and their operational thresholds are also not described in the available record.

How should uncertainty be read?

A confidence interval describes the statistical uncertainty around an estimate under the analysis assumptions. A narrower interval indicates greater precision, while a wider interval indicates less precision. It does not guarantee that the estimate is unbiased or that the method is clinically acceptable. The primary outcome specifies a 95% confidence interval, but no interval is reported. Without an estimate and interval, neither the apparent strength of correlation nor its precision can be assessed.

The registry contains planning illustrations based on the observed effect, with a stated two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. These are planning assumptions, not the design of this single-arm paired measurement study. Such illustrations are not observed power and do not substitute for results. The record does not report the study’s own sample-size justification or enough outcome information to evaluate those calculations as a basis for this study.

What does the analysis count indicate?

The registry lists zero analyses. This is consistent with the absence of reported numerical outcome estimates in the record, but it does not establish why results are absent or whether analyses were conducted outside the registry. The registry does not report the analysis population, missing-measurement handling, or the number of complete paired observations. Those omissions matter because measurement comparisons require paired values, and incomplete pairs cannot contribute to a paired comparison in the same way as complete pairs.

All of these explanations are general statistical guidance. The actual performance of TatchSleep Pro in this study cannot be characterized without the numerical outcome results and their uncertainty.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary outcome and two secondary outcome topics. The primary assessment concerns correlation between device and PSG indices, while the secondary outcomes concern AHI severity classification agreement. Considering multiple outcomes raises a general multiplicity issue: the more statistical questions examined, the greater the opportunity for at least one result to appear notable by chance if each is interpreted without regard to the others. The registry record does not report a multiplicity adjustment, a hierarchy among secondary analyses, or a statistical analysis plan describing how multiple results would be interpreted.

The record lists three outcome measures and zero analyses, but it does not provide numerical results for the primary or secondary outcomes. It is therefore not possible to assess whether multiplicity adjustments were used or how they might affect interpretation. No interim analyses or stopping rules are reported in the registry record.

Missing-data handling is also not reported in the registry record. The study question depends on paired device and PSG measurements, but the record does not state how incomplete pairs, unusable device recordings, or missing PSG measurements were handled. It does not identify an analysis population or state how many participants had evaluable paired data. The enrollment figure is not necessarily the denominator for each analysis, and the safety-group size should not be assumed to be an outcome-analysis denominator.

These omissions limit reproducibility and make it difficult to judge whether the reported analysis population, if results become available elsewhere, corresponds to all enrolled participants or a subset. No conclusions about missingness, exclusions, or analysis choices can be made from the registry record alone.

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 safety listing describes one group, Medical Device: TatchSleep Pro: Single Arm. It records 45 participants at risk for serious events and 0 affected by serious events. The listing also records 45 at risk for other events and 0 affected by other events, with 0 deaths. No serious event terms are listed.

These figures describe what is entered in the registry safety fields; they do not establish that every participant in the study was included in the safety group. The listed enrollment is 53, and the registry does not explain why the safety denominator differs from that enrollment figure. The record also does not give further detail on how adverse events were collected, the observation period for safety reporting, or whether event definitions or attribution procedures were used. These details should not be inferred from the zero counts.

A recorded zero means that no affected participants or deaths are entered for the corresponding field in this registry listing. It does not prove that no events occurred outside the scope or reporting process represented by these fields. The safety figures are descriptive and do not establish comparative safety, since the study has no separate control arm. They also do not answer the primary measurement-performance question.

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

The record describes a study designed to compare TatchSleep Pro respiratory-index measurements with in-lab PSG measurements taken during the same sleep study. It identifies the intended primary analyses—correlation and Bland–Altman assessment—and secondary analyses of AHI severity classification. This provides a clear outline of the measurement question the study was meant to address.

What the registry does not provide is the numerical evidence needed to answer that question. The primary correlation coefficient and its 95% confidence interval are not reported. Bland–Altman results, ODI results, percent agreement, and Cohen’s kappa are also not reported. With no numerical estimates, the strength of association, the degree of agreement, and the precision of those estimates cannot be evaluated.

Even if correlation results were available, correlation alone would not show that device and PSG values are sufficiently close for interchangeable use. Agreement analyses would be important for understanding systematic differences and the spread of paired differences. Classification agreement would address a related but distinct question: whether the methods place measurements in the same severity categories. The record reports plans for these analyses, not their findings.

The design is not a randomized comparison of interventions. It is a single-arm assessment of a device against a reference measurement. It therefore cannot establish the effect of device use on clinical decisions, care, or patient outcomes. Nor does the safety listing, with its recorded zero affected participants and zero deaths, establish comparative safety or exclude events beyond those recorded.

The appropriate conclusion is limited to the study’s design and intended measurements. The registry record does not report results sufficient to determine whether TatchSleep Pro measurements correlate with, agree with, or classify AHI similarly to PSG measurements. Those questions remain unanswered by the posted registry results.

13. Why This Trial Matters Statistically

This study is statistically relevant because evaluating a measurement device requires more than asking whether its values correlate with a reference method. Correlation can show that two measurements tend to change together, but method agreement is needed to understand whether the values are close. The registry’s stated plan to combine Pearson correlation with Bland–Altman analysis reflects these complementary questions.

The paired design is also appropriate to the stated measurement question in principle: device and PSG measurements are collected simultaneously for participants in the test arm. Pairing makes the comparison about two measurements made in the same sleep-study setting, rather than a contrast between separate groups. The registry, however, does not report the number of complete pairs or the results of the planned comparisons, so the statistical contribution of that design cannot be evaluated from the posted results.

The secondary focus on severity classification illustrates another important distinction. Numerical agreement between index values and agreement in categories are related but not identical. A method may have some value-level discrepancies that do not change categories, or small discrepancies that cross a classification boundary. Reporting both percent agreement and Cohen’s kappa could help describe classification concordance from different angles, but those results are absent.

Finally, this record shows why registry design information and outcome reporting serve different purposes. The design identifies the question, comparator, and intended methods. Numerical estimates and uncertainty are needed to judge what the study found. Because those results are not reported, the record is informative about the planned evaluation but does not establish the magnitude or precision of device-to-PSG correspondence.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The principal limitation for interpretation is that numerical results for the primary and secondary outcomes are not reported in the registry record. The planned correlation, confidence interval, Bland–Altman assessment, ODI analysis, percent agreement, and Cohen’s kappa therefore cannot be evaluated. The record lists zero analyses, and no outcome-specific denominators or uncertainty estimates are available.

The analysis population is unclear. Although the registry lists enrollment of 53, it does not report how many participants had usable measurements from both TatchSleep Pro and PSG. It also does not describe missing-data handling, exclusions, protocol deviations, or the completeness of recordings. These details are particularly relevant to a paired method-comparison study.

The record identifies a single-center, single-arm design and no masking. PSG serves as the reference measurement during simultaneous device use, rather than as a randomized control arm. The record does not provide information on the selection of participants beyond the arm description, nor does it report device or PSG measurement procedures in enough detail to assess implementation. Generalizability to other settings or populations cannot be determined from the registry record.

The listed enrollment and safety-group denominator differ, and the record does not explain that difference. Safety reporting is limited to registry counts; event definitions, collection procedures, and observation details are not reported. The severity classification thresholds are also not described.

Finally, the record does not report a statistical analysis plan, multiplicity approach, or sample-size justification for the study. The planning illustrations elsewhere are not the trial’s own design justification and should not be treated as observed power. These limitations leave the study’s actual measurement performance and result precision unresolved.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: TatchSleep as a Tool to Aid in Sleep Apnea Analysis — NCT04419714.
• Registry record last updated 2025-08-20; status last verified 2025-07.
• Results first posted 2025-08-20.
• 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-center, single-arm quantitative study evaluating TatchSleep Pro during simultaneous in-lab PSG. Enrollment is listed as 53, and the primary outcome is correlation between device and PSG AHI using a 3% desaturation criterion, with a 12-hour timeframe. The stated methods include Pearson correlation and Bland–Altman analysis; secondary outcomes concern AHI severity classification agreement.

The record lists zero analyses and does not report numerical results for the primary or secondary outcomes. The safety listing records 45 at risk for serious events, 0 affected, 45 at risk for other events, 0 affected, and 0 deaths. The relationship between the safety group and listed enrollment is not explained. The registry therefore describes the study’s design and intended measurement questions but does not establish the observed association, agreement, or precision.