Validation of Oximetry for Diagnosing Obstructive Sleep Apnea in a Clinical Setting

Round 1

Reviewer 1 Report

The authors compared ODI measured by oximetry with the AHI/REI measured by PSG and out-of-center sleep testing (OCST) in 119 participants in the Nagahama study. Their results suggest that oximetry ODI has only moderate correlation to AHI/REI, and it is difficult to set a threshold value of ODI to distinguish moderate and severe OSA. This is a very interesting study, and I have the following comments for the authors: 

1. In Bland-Altman plot, it is important to show the 95% CI of the mean difference and the 95% limits of agreement (LoA), and checking the normality of the differences. In your current results I can only see the 95% LoA in Figure 1 (b). Please add the 95% CI of the mean difference and check the normality of the differences in all of your Bland-Altman analyses.
2. How many moderate-to-severe and severe OSA patients are there in your Figure 2? It will also be interesting to see how many patients changed their original diagnoses that only based on oximetry ODI, after doing in-lab PSG and OCST diagnosis.  
3. The authors should provide details of the oximetry devices (e.g., transmission or reflectance pulse oximetry? where is the oximetry sensor worn? fingertip, arm or forehead?) used in their study, and also compare the correlations and agreements between the ODI measured from the oximetry devices and the ones measured by PSG and OCST. 
4. I do not understand why the authors could conclude that PSG and Type 3 OCST have similarity in agreement and increasing the number of channels in others could provide more accurate data similar to PSG. Their studies did not compare AHI and REI measured by PSG and OCST.
5. I suggest to add the supplementary results into the main text, as they include some interesting key results of this study.
6. One main limitation of your study is that the oximetry ODI was not measured simultaneously together with PSG/OCST, which may also explain why the correlations between ODI and AHI/REI are lower compared to previous studies. Please discuss this limitation in your manuscript.

Author Response

Reviewer 1:

Thank you very much for your very useful suggestions. According to your suggestions, I revised the manuscript as the followings.

1. In Bland-Altman plot, it is important to show the 95% CI of the mean difference and the 95% limits of agreement (LoA), and checking the normality of the differences. In your current results I can only see the 95% LoA in Figure 1 (b). Please add the 95% CI of the mean difference and check the normality of the differences in all of your Bland-Altman analyses.
→ Figures 2b and 3: Bland-Altman plots are Figures 2b, 3a,3b and 3c. I put 95%CI (±1.96 SD) lines to all the plots.

2. How many moderate-to-severe and severe OSA patients are there in your Figure 2? It will also be interesting to see how many patients changed their original diagnoses that only based on oximetry ODI, after doing in-lab PSG and OCST diagnosis.
→ Lines 182-186: We added table 2 to present the diagnostic agreement between oximetry in the previously performed epidemiological setting and PSG/OCST in our clinical setting.

3. The authors should provide details of the oximetry devices (e.g., transmission or reflectance pulse oximetry? where is the oximetry sensor worn? fingertip, arm or forehead?) used in their study, and also compare the correlations and agreements between the ODI measured from the oximetry devices and the ones measured by PSG and OCST.
→ Lines 69-77, 151-152, 182-186, 209-210 and 233-237: We revised the manuscript following your suggestion. Details of oximetry were provided in the introduction section. Agreement between oximetry and PSG/OCST was presented in the table 2. Correlation was presented in the results and discussion (figures 2a, 5a and 5b).

4. I do not understand why the authors could conclude that PSG and Type 3 OCST have similarity in agreement and increasing the number of channels in others could provide more accurate data similar to PSG. Their studies did not compare AHI and REI measured by PSG and OCST.
→ Lines 220-223: I added more discussion concerning this comment in the second paragraph of the discussion.

5. I suggest to add the supplementary results into the main text, as they include some interesting key results of this study.
→ Figures 3, 5 and 6: We added supplementary results into the main text as you suggested

6. One main limitation of your study is that the oximetry ODI was not measured simultaneously together with PSG/OCST, which may also explain why the correlations between ODI and AHI/REI are lower compared to previous studies. Please discuss this limitation in your manuscript.
→ Lines 250-258: I added this as the limitation. Thank you.

Reviewer 2 Report

The authors have conducted a study validating oximeter for the diagnosis of obstructive sleep apnea. I congratulate the authors for the interesting work they have conducted.  The manuscript is interesting but needs little modifications. 

Please see the attached file for comments/suggestions.

Comments for author File: Comments.pdf

Author Response

Thank you very much for all the comments you made. We revised the manuscript following your suggestions.

Line 15: which population?

→ We changed from “this population” to “Japanese population”

Line 23: not sure what this means.

→ We deleted “in large epidemiological settings”

Line 25: What population? please define.

→ We deleted “in this population”

Lines 46-47: Please reword for clarity.

→ We revised the sentence.

Lines 51-66: Need to provide more details on first two studies and their findings for all three studies.

→ We added more detailed of the three studies as you suggested.

Line 78: Which study?

→ We changed from “this epidemiological study” to “the Nagahama Study”

Lines 81-82: This is not clear. Was the Oximetry performed as a part of an epi study and then the data was compared with the data obtained from clinic in same patients? or Is this study designed to compare home-based vs clinical values?

Please clarify.

→ We added “we compared OSA diagnosis data obtained in our clinical setting to oximetry data obtained in the Nagahama Study.” to clarify.

Figure 1: Please provide a consort flow diagram for this information.

→ We added the diagram (figure 1) and verify some numbers.

Lines 242-243: This belongs in the discussion

→ We moved “suggesting that use of ODI to distinguish between moderate-to-severe and severe OSA is difficult.

Lines 117-120: This is methods not results.

→ We moved “For OCST, we usually ask patients to record for two nights, in case they have trouble recording. This multiple night strategy is useful in the reduction of the failure rate in OCST [1]. We have two OCST recordings for most participants with an interval of 1.40 ± 0.877 days and can compare ODI and REI from different nights for the same participants” from results to methods.

Lines 215-216 is a better description of the study and needs to be clarified in the earlier sections.

→ Thank you very much. We added “we compared OSA diagnosis data obtained in our clinical setting to oximetry data obtained in the Nagahama Study” in lines 81-82 in introduction.

Lines 234-237: report correlation coefficient value here and mention if it is similar or not.

→ We added correlation coefficient values. Thank you.

Lines 250-262: This is good but the discussion is not sufficient in terms of describing results. The Authors are recommended to discuss on how they interpret supplementary data and Figure 1b data.

→ We added discussion concerning figure 3 (previously supplementary figure S1) in the second paragraph of discussion (lines 217-223). I also added discussion in lines 251-259 about figures 5 and 6 (previously supplementary figures S2 and S3).

Lines 265: Again, epidemiological studies or settings? This is confusing.

→ We changed the word from “settings” to “studies”. Thank you.

Round 2

Reviewer 1 Report

My comments have been correctly answered by the authors. I have no further comments. I suggest to accept this manuscript.