Assessment of Continuous Positive Airway Pressure Therapy in OSA and ILD: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Assessment of Continuous Positive Airway Pressure Therapy in OSA and ILD (NCT03901534), a non-randomized na study. The primary outcome is Surfactant Protein D (SP-D) [Aim 1].

2. Clinical Question

The question is therefore an association and biomarker question rather than a straightforward test of a treatment against a concurrent randomized control. The record describes several aims, with cohorts assigned to different measurements. Aim 1 uses MESA biospecimens, and the primary outcome listed for that aim is SP-D. Aim 2 uses participants with newly diagnosed moderate-to-severe OSA from HeartBEAT and BestAir, while Aim 3a concerns CPAP treatment of OSA and Aim 3b concerns OSA screening in people with ILD.

These distinctions matter when interpreting any result: evidence from one cohort or aim would not automatically answer the question posed by another. The record does not report how biomarker levels were compared, whether associations were adjusted for other characteristics, or what specific contrast was planned for each aim. It also does not report primary outcome results. The study status is terminated, but the record does not state why it ended or whether all planned measurements were completed.

3. Trial Design

The registry record describes a non-randomized, unmasked, sequential study of CPAP therapy and biomarkers in OSA and ILD. Its stated purpose is to evaluate whether biomarkers of lung epithelial and endothelial injury are associated with these conditions. The record lists Columbia University as sponsor, an enrollment figure of 2021, a start date of 2023-01-01, a primary completion date of 2023-10-16, and a terminated status. It does not explain the enrollment figure or the reason for termination.

Four cohorts are described. The MESA cohort includes biospecimens collected from 1852 participants in a sleep study, with serum biomarkers measured by immunoassay. The HeartBEAT and BestAir cohort includes 82 people with newly diagnosed moderate-to-severe OSA and without known severe lung disease. A separate cohort of 87 participants with OSA is described as receiving CPAP therapy. The ILD cohort is described as undergoing OSA screening by polysomnography. Each cohort is linked to specific aims; the MESA cohort is for Aim 1, and the CPAP and ILD cohorts are designated for Aim 3a and Aim 3b respectively.

The design is not a randomized comparison in this registry record. The fact that some HeartBEAT and BestAir participants previously took part in randomized trials does not establish random allocation for the current study or make the cohorts interchangeable. The sequential model is recorded, but the sequence of measurements, recruitment, or treatment is not explained further. The record does not report the protocol rationale, eligibility criteria in full, allocation procedures, masking procedures beyond none, or whether the planned enrollment was achieved.

There are 11 outcome measures listed and zero analyses listed. The record gives no statistical analysis plan, sample-size justification, interim analysis description, or account of deviations from the planned design. As a result, the design can be described from its cohort structure and registry fields, but not used to reconstruct the intended analysis in detail.

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.

Aim 1

Aim 2

Aim 3a

Aim 3b

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: No participants were enrolled for Aim 3b.

6. Primary Endpoint

The registry lists the primary outcome as Surfactant Protein D (SP-D) [Aim 1], measured over 1 day.

The listed primary outcome is Surfactant Protein D (SP-D) for Aim 1. It is measured in pg/mL, with the outcome parameter described as a mean and dispersion described as a standard deviation. The timeframe is 1 day. The record identifies SP-D as a marker of alveolar epithelial cell injury and states that this outcome applies only to participants in the Aim 1 phase of the study.

In general, a continuous biomarker outcome retains information about measured concentrations rather than reducing observations to categories. A mean describes the central value of a set of measurements, while a standard deviation describes their spread. These summaries answer different questions: a mean alone does not show how variable individual measurements are, and a standard deviation does not identify the direction or size of a group difference.

SP-D is used as a biological marker relevant to the stated question about lung injury. A biomarker can help characterize biological processes, but a measurement by itself is not a clinical outcome and does not establish a cause-and-effect relationship. Interpretation would depend on the comparison being made, the participants included, measurement procedures, and uncertainty around any estimated association.

The registry record does not report the observed SP-D mean, standard deviation, sample analyzed, confidence interval, or statistical test. It also does not report results for the other outcome measures. The listed endpoint therefore identifies what was intended to be measured, but the record does not allow readers to assess its observed level or precision.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the stated outcome summary mean?

SP-D is a continuous measurement reported in pg/mL. In general, describing a continuous outcome with a mean and standard deviation separates its central tendency from its variability. The mean gives a summary of the measured values, while the standard deviation describes how dispersed those values are around the mean. Neither quantity alone expresses uncertainty about the estimated mean. A standard error or confidence interval is often used for that inferential purpose, but neither is reported for SP-D in this registry record.

The endpoint is assigned to Aim 1 and is measured over a timeframe of 1 day. That framing tells readers which aim the measure belongs to and the listed timeframe; it does not explain whether repeated measurements were taken, how a single value was selected, or how measurements were processed. Those details are not reported in the registry record. The record also does not report an observed mean, standard deviation, analyzed sample size, or comparison for this endpoint.

How should the study structure affect interpretation?

The design is described as non-randomized, unmasked, and sequential. In general, random allocation can help balance measured and unmeasured participant characteristics between treatment groups, while a non-randomized design does not provide that same design-based protection. Differences between groups may reflect treatment, participant characteristics, selection, or other factors. This is a general consideration, not a claim that any particular bias occurred in this study.

The record lists cohorts with different roles: MESA biospecimens for Aim 1; HeartBEAT and BestAir participants for Aim 2; a CPAP treatment cohort for Aim 3a; and an ILD cohort screened for OSA for Aim 3b. A sequential model is recorded, but the exact sequence and how cohort membership relates to analysis are not detailed. Results would need to be interpreted in the context of the relevant cohort and aim rather than combined automatically across all participants.

The HeartBEAT and BestAir cohort is described as participants from two randomized trials. That description concerns the participants' prior study context. The present record identifies its own allocation as non-randomized, so the prior trial design cannot be assumed to govern the current biomarker assessment. The record does not report whether analyses used original randomization, treatment assignment, or other trial information.

What can a confidence interval and a p-value contribute?

In general, a confidence interval communicates statistical uncertainty around an estimated quantity. Its width reflects precision under the model and assumptions used; its endpoints are not a guarantee that the true value lies within that particular interval. A p-value, when reported, describes how incompatible the observed data are with a specified null model, under the assumptions of the test. It does not measure the probability that a hypothesis is true, nor does it describe the size or clinical importance of an effect.

No confidence interval, p-value, test statistic, model estimate, or analysis method is reported for the primary endpoint. The record lists zero analyses, so there is no posted inferential result to interpret. It also does not state whether comparisons were adjusted, which covariates were considered, how distributions were assessed, or how missing biomarker measurements would be handled.

Why can variability and over-dispersion matter?

For continuous measurements, substantial variability can make a mean less precisely estimated, particularly when the number of observations is limited. In general, a distribution with unusual values or strong skew can make a mean and standard deviation less representative of a typical participant. Researchers may examine the distribution and select a suitable model or summary, but the registry record does not describe such checks or choices for SP-D.

Over-dispersion is most often discussed for count outcomes when observed variability exceeds what a basic count model expects. It matters because an overly restrictive variance assumption can make uncertainty estimates too small. SP-D is recorded as a continuous outcome, and the registry record does not identify a count model or report an over-dispersion assessment. This is a general statistical concept, not a reported feature of this analysis.

What do the planning assumptions say?

The stated planning framework describes a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It says that the effect to detect is a stated fraction of an observed effect. For time-to-event outcomes, the framework describes total events; for other outcomes, it describes participants per arm. These are planning assumptions, not observed trial findings and not a report of achieved power.

The registry record does not report the observed effect used for a planning illustration, the actual calculated sample-size values, or a design-stage sample-size justification. It also does not show that equal allocation between two arms matches the multi-cohort structure described for this study. Planning illustrations based on an observed effect are not evidence that the study was originally designed around that effect, and such calculations are not observed power.

What is not available for re-analysis?

There are no posted outcome estimates to recalculate from summary statistics. No unadjusted or adjusted comparisons, standard errors, confidence intervals, or group summaries appear in the registry record. Consequently, an independent recalculation cannot be made from the recorded results. The record's zero-analysis listing also means no sponsor analysis is described for comparison. The appropriate conclusion is that the analysis methods and numerical results are not reported in the registry record, rather than that any particular statistical method was used or any particular result was obtained.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("MESA")
n   <- c(1852)
m   <- c(8)
sp  <- c(6)
sd  <- sp      # posted standard deviations
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  MESA  1852  8.000  6.000  7.727  8.273

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists 11 outcome measures and zero analyses. It does not state how many hypotheses were tested, which outcomes were designated primary beyond the listed Aim 1 SP-D measure, whether outcomes were ordered hierarchically, or whether any multiplicity adjustment was planned. With several outcomes or aims, repeated testing can increase the chance of at least one apparently notable result by chance. In general, methods such as family-wise error control or false discovery rate procedures can address particular multiplicity goals, but the record does not say that any such method was used.

The record does not report a statistical analysis plan, model specifications, covariates, subgroup analyses, interim analyses, or rules for choosing among multiple comparisons. It is not possible to determine whether analyses were prespecified or how analytic decisions were handled. The planning framework mentions a two-sided significance level of 0.05, but that statement alone does not establish the testing strategy for each outcome or how multiplicity was considered.

Missing-data methods are also not reported in the registry record. There is no description of the amount or pattern of missing biomarker measurements, exclusions, imputation, or sensitivity analyses. Without these details, readers cannot assess whether the analyzed participants differed from those with missing measurements or how missingness may have affected any estimate.

The analysis populations are not described in sufficient detail to determine who was included in each outcome analysis. The cohort descriptions identify intended aim-specific groups, but do not establish the number actually analyzed for each outcome. The registry enrollment figure is 2021; the record does not explain how this figure relates to the separate cohort descriptions or the number with available measurements. The absence of these details prevents an assessment of analysis denominators, exclusions, and generalizability.

12. 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 record reports serious adverse events, other adverse events, and deaths by cohort. For the MESA cohort, 1852 participants are listed at risk, with 0 serious adverse events, 0 other adverse events, and 0 deaths. For HeartBEAT and BestAir, 82 are listed at risk, with 0 serious adverse events, 0 other adverse events, and 0 deaths. For the CPAP treatment of OSA cohort, 87 are listed at risk, with 0 serious adverse events, 0 other adverse events, and 0 deaths.

For the ILD screening for OSA cohort, the record lists 0 participants at risk and 0 events in the safety fields. Because no participants are listed at risk for that cohort, these zeros should not be interpreted as a safety observation among participants who underwent screening.

In general, a count of zero recorded events does not establish that an intervention is risk-free. Interpretation depends on how events were defined, collected, and monitored, the period of observation, and the number of participants with safety follow-up. These details are not reported in the registry record. The displayed counts are registry safety summaries and do not establish a comparative safety effect, particularly because the study is non-randomized and does not list a like-for-like control comparison.

The record does not list any top serious adverse event terms. The zero counts therefore describe what is entered in the safety fields, not a detailed account of event surveillance or adjudication.

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

The record establishes the study's stated purpose, its cohort structure, the primary Aim 1 biomarker, and its administrative status. It identifies SP-D as a marker of alveolar epithelial cell injury and specifies that the measure is a mean in pg/mL with standard deviation as its dispersion summary. It also describes cohorts related to OSA, ILD, CPAP treatment, and biospecimens from prior study settings.

It does not establish whether SP-D was associated with OSA or ILD, whether biomarker values differed between groups, or whether CPAP changed a biomarker. There are no numerical primary outcome results, estimates of uncertainty, or listed statistical analyses. The status of terminated does not answer those scientific questions and does not explain whether data collection or analysis was completed.

In general, an observed association is not necessarily causal. In a non-randomized study, group differences may have several explanations, including participant characteristics or selection into a cohort. A causal interpretation would require careful design and analysis, along with assumptions that cannot be assessed from the current registry description. The record does not report the adjustment strategy or comparison details needed to evaluate those assumptions.

The cohort-specific nature of the aims also matters. Findings from MESA biospecimens would concern that cohort and its measurements; findings from participants receiving CPAP or undergoing OSA screening would address different questions. The registry record does not provide results that could be used to compare these aims or generalize across them.

Thus, the record documents a biomarker study and its planned measurement framework, but not an interpretable estimate of association or treatment effect. It also does not support a conclusion about the presence or absence of a biological relationship. Any such conclusion would require results and methods that are not reported in the registry record.

14. Why This Trial Matters Statistically

This study is statistically notable for combining several cohorts and aims under a non-randomized, sequential design. That structure makes it especially important to distinguish the population and question associated with each measure. A single overall description of OSA, ILD, biomarkers, and CPAP could obscure that the cohorts serve different purposes and do not form a simple randomized treatment comparison.

The primary outcome is a continuous biomarker, SP-D, summarized by a mean and standard deviation. In general, continuous outcomes can retain information about measured levels, but their interpretation depends on the distribution, measurement quality, cohort definition, and analysis model. The registry record supplies the endpoint description but no numerical result or precision estimate, so statistical significance and practical relevance cannot be assessed.

The study also illustrates the importance of separating study registration from posted findings. A registry can identify intended outcomes, interventions, and design features without presenting the completed analyses. Here, the record lists 11 outcomes but zero analyses and marks the study terminated. Those details make clear why design information alone cannot answer the substantive biomarker question.

Finally, the study's cohort structure underscores the limits of treating previously randomized-trial participants as if the current study were randomized. The registry allocation is non-randomized. The causal protections associated with a randomized design cannot be assumed for a separate biomarker assessment merely because some participants came from prior randomized trials. The record does not describe analyses that would address confounding or selection.

Statistically, the key lesson is the need to align conclusions with the actual reported estimand, analysis population, and uncertainty. In this record, these components are not reported for the primary result. The study's clinical and biological questions are identifiable, but the evidence needed to quantify an answer is not present.

15. 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 registry record does not report results for the primary SP-D outcome or the other listed outcomes. It lists zero analyses, so there are no effect estimates, group summaries, confidence intervals, or p-values to evaluate. The reason for termination and whether all planned measurements were completed are not reported in the registry record.

The study has several distinct cohorts, and their relationship to the enrollment figure of 2021 is not explained. The record gives cohort descriptions, including 1852 MESA participants, 82 HeartBEAT and BestAir participants, and 87 participants in the CPAP treatment cohort, but does not report the numbers actually analyzed for each outcome. The ILD screening cohort has no participants listed at risk in the safety fields; this does not establish its enrollment or whether study measurements were completed.

The study is non-randomized and unmasked. The record does not describe a concurrent comparator, detailed allocation process, blinding of outcome assessment, or methods to address confounding. It does not say whether participants were selected for particular analyses in ways that could affect comparisons. The sequential model is recorded, but its practical details are not reported.

Measurement and analysis details are also limited. The record does not explain biomarker assay procedures beyond identifying immunoassay measurement for MESA serum biomarkers, nor does it describe quality control, transformation, distributional assessment, covariate adjustment, missing-data handling, or sensitivity analysis for the primary endpoint. The exact rationale for the sample-size assumptions and the actual planning calculations are not reported.

These limitations constrain both reproducibility and interpretation. They do not demonstrate that the study methods were inadequate; they indicate that the registry record does not contain enough information to assess those methods or the findings.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Assessment of Continuous Positive Airway Pressure Therapy in OSA and ILD — NCT03901534.
• Registry record last updated 2024-12-20; status last verified 2024-12.
• Results first posted 2024-12-20.
• 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.

19. Record Summary

The registry record describes a terminated, non-randomized, unmasked, sequential study sponsored by Columbia University. Its stated purpose is to evaluate associations between biomarkers of lung epithelial and endothelial injury and OSA and ILD. The record lists a start date of 2023-01-01, a primary completion date of 2023-10-16, and an enrollment figure of 2021.

Four cohorts are described across the study aims: MESA biospecimens, HeartBEAT and BestAir participants, a CPAP treatment cohort for OSA, and an ILD cohort screened for OSA. The listed primary outcome is SP-D for Aim 1, measured in pg/mL and summarized as a mean with standard deviation over a timeframe of 1 day. The record lists 11 outcome measures and zero analyses.

No primary outcome results, inferential estimates, analysis methods, or reason for termination are reported in the registry record. Safety fields list zero serious adverse events, zero other adverse events, and zero deaths for cohorts with participants at risk, while the ILD screening cohort has zero participants listed at risk in those fields. These registry entries describe the available record but do not establish a treatment effect or a complete account of study findings.