Image quality of abdominal ultrasonography after esophagogastroduodenoscopy is preserved by using carbon dioxide insufflation: A non-inferiority test in the same subject.

Because bowel gas deteriorates the image quality of abdominal ultrasonography (AUS), it is common to perform AUS prior to esophagogastroduodenoscopy (EGD). This one-way order limits the availability of examination appointments. To evaluate whether EGD using insufflation of carbon dioxide (CO2), which is rapidly absorbed by the gastrointestinal mucosa, preserves the image quality of AUS performed subsequently, we designed a non-inferiority test in which each subject underwent AUS, EGD with CO2 insufflation, and a second AUS, in that order. All saved AUS moving images were randomized and imaging quality was evaluated at 16 organs using a four-point Likert-like scale that divides the depiction rate by 25%. Sample size was calculated to be 26 using the following: non-inferiority margin of -0.40 corresponding to depiction rate of -10%, difference of means of 0.40, common standard deviation of 1.25, power of 90%, and 1-sided α-level of 0.025. We enrolled 30 subjects. The mean and 95% confidence interval (CI) of the image quality score of all 16 organs at pre- and post-EGD AUS in the 30 subjects were 3.54 [3.48-3.60] and 3.46 [3.39-3.52], respectively. The difference in the means was 0.08 of the scores, corresponding to a 2% depiction rate. The effect size was 0.172. The image quality of post-EGD AUS was not inferior, as demonstrated by the 97.5% CI of the difference, which did not cross the non-inferiority margin of -0.40. In conclusion, the use of CO2 for insufflation in EGD does not cause much deterioration in the image quality of AUS performed subsequently. Therefore, it is permissible to perform EGD prior to AUS, which is expected to improve the efficiency of examination setup.

Introduction

As it is the recommended first-line imaging study for specific clinical situations [1], abdominal ultrasonography (AUS) is commonly used as a diagnostic imaging test for patients with abdominal pain, a screening test for various types of basal disease, and for medical check-ups. However, ultrasonography does not always provide sufficient diagnostic information due to inadequate visualization of intra-abdominal structures, especially if bowel gas is present.

In endoscopy, gas is introduced to achieve adequate distension of the GI lumen for safe advancement of endoscopes and for careful visualization of the mucosa. Therefore, when AUS and esophagogastroduodenoscopy (EGD) are planned at the same time, it is common to perform AUS prior to EGD. This one-way order limits the availability of examination appointments and reduces the efficiency of examination. Room air, which is widely used for GI luminal distension, has the advantages of universal availability and low cost. However, room air is poorly absorbed by the GI tract and causes post-procedure pain related to distension. In contrast, carbon dioxide (CO2) is rapidly absorbed by the GI mucosa, driving increased interest in its use as an insufflation agent for endoscopic procedures [2].

We hypothesized that the use of CO2 insufflation in EGD would not deteriorate the image quality of AUS performed subsequently. Two trials associated with this hypothesis have been reported. As only a limited number of organs, such as the pancreas alone [3] or the pancreas, extrahepatic bile duct, and inferior pole of the right kidney [4], were available for observation by AUS, and the statistical analysis was inadequate in these trials in which no statistical analysis was done [3] or a significant difference test was done instead of a non-inferiority test [4], the results of the data analyses are insufficient to test our hypothesis. In addition, they reported partially negative results showing that, in about 27% of patients, the image quality of AUS was worse [3]. However, they also reported results based on which they concluded that the image quality of AUS was improved in some patients [4]. To prove this hypothesis, we designed a non-inferiority test [5] that compared image quality of AUS performed following EGD with CO2 insufflation (post-EGD AUS) with that of AUS prior to EGD (pre-EGD AUS) in the same subject.

Materials and methods

This prospective and observational study was conducted in compliance with Declaration of Helsinki at a single institute and was approved by the Institutional Review Board of Saiseikai Kanazawa Hospital (No. H28-19). All experiments were carried out in accordance with the approved study plan and relevant guidelines. The subjects were examinees aged ≥20 years who visited our institute for medical check-ups including both AUS and EGD between January and March 2017. Those with an underlying previous history of operations such as gastrectomy, pancreatectomy, hepatectomy, and cholecystectomy were excluded to avoid the possible effect of intestinal peristalsis on the results of AUS. Those contraindicated for antispasmodics were also excluded. Written informed consent was obtained from all subjects. Each subject underwent AUS, EGD with CO2 insufflation instead of air under planned moderate sedation, and a second AUS, in this order, within the same day. After EGD, subjects were transferred to a recovery bed and rested until their level of consciousness had recovered to the same level as before endoscopy.

Ultrasonography

Ultrasonography was performed by three sonologists using a Hitachi Preirus with a 1–5 MHz curvilinear transducer (EUP-C715; Hitachi, Ltd., Tokyo, Japan). One of the three sonologists was board-certified by the Japan Society of Ultrasonics in Medicine, and the other two sonologists had six or eight years of clinical experience and had performed a minimum of 500 clinical AUS examinations per year. The same technologist performed both pre- and post-EGD AUS for a given subject to reduce systematic error. All AUS moving images were saved. The 16 evaluation organs were the pancreas (head, body, and tail), liver (right, left, and caudate lobe), gallbladder, common biliary duct, spleen, kidney (right and left), aorta, celiac artery (CA), superior mesenteric artery (SMA), portal vein (PV), and splenic vein (SV), according to the AUS Cancer Screening Criteria of the Japanese Society of Gastrointestinal Cancer Screening [6].

Review of ultrasonograms

A method for objectively estimating the quality of AUS images has not been established. Previous studies [3, 4, 7–9] have used three- to five-point Likert scales such as 1—not interpretable, 2—barely interpretable, 3—adequate for interpretation but of poor quality, 4—interpretable and of average quality, 5—interpretable and of superior quality. In this study, we developed a more objective scale by replacing the abstract evaluation with the depiction rate as an interval scale suitable for statistical analysis [10]. All saved AUS moving images, which were marked only with a study code, were randomized and then evaluated. To reduce systematic error, four expert reviewers board-certified by the Japanese Society of Gastroenterology rated image quality in all moving images using a four-point Likert-like scale as an interval scale: 1—depiction rate of 0%–24%, 2–25%–49%, 3–50%–74%, 4–75%–100%. The mean of the evaluation score by the four reviewers was used as the final score for each organ in each subject. The reviewers were blinded to the time points of the moving images, including whether the images they were reviewing were obtained before or after EGD. The reviewers were also blinded to the identity of the sonologist who performed the scan, to the study subject being imaged, and to the scores of the other reviewers.

Endoscopic procedure

All EGD procedures were performed by the same endoscopist who was board-certified by the Japan Gastroenterological Endoscopy Society. EGD was performed using a GIF-H290 scope (Olympus Medical Systems, Tokyo, Japan) with a CO2 Regulation Unit (UCR) as the CO2 insufflation system (Olympus Medical Systems). Prior to endoscopy, 10 mg of butyl scopolamine bromide, an antispasmodic, was injected intramuscularly and diazepam was used for planned moderated sedation in all subjects.

Statistical analysis

The primary endpoint of the study was the difference in AUS image quality before and after EGD with CO2 insufflation using a non-inferiority test [5]. Secondary end points included BMI, abdominal circumference, gender difference, and age. Sample size calculation was based on a pre-defined margin of non-inferiority for image quality score set at –0.40, which is the same as a depiction rate of –10%. In the absence of published data, this margin was selected because we consider a larger difference as clinically relevant. The difference of means between pre- and post-EGD AUS was assumed to be 0.40 in the absence of previous data. The common SD was also assumed to be 1.25 in the absence of previous data. With a power of 90% and a 1-sided α-level of 0.025, we estimated that 26 subjects were needed to show non-inferiority in one-sample mean of post-EGD AUS. To compensate for unforeseeable problems, we aimed to enroll a total of 30 subjects. Sample size calculation was performed with the trial-size package of “R Project for Statistical Computing”.

Results

Thirty subjects were enrolled in the study and their characteristics are shown in Table 1. The mean body mass index (BMI) was 23.0 kg/m2 and was >30 kg/m2 in only 1/30 subjects. The mean procedure time of EGD was 4.7 min and the mean duration of the period after EGD to post-EGD AUS was 73.8 min (range, 53–110 min).

Table 1

Examinees’ characteristics.

Examinee	30
Male	20
Female	10
Age (years)	47.4 ± 7.8 (28–62)
BMI (kg/m2)	23.0 ± 3.2 (17.8–32.8)
BMI > 30	1
BMI < 18.5	0
Abdominal circumference (cm)	80.5 ± 9.1 (60–104)
EGD procedure time (min)	4.7 ± 1.2 (3–9)
Duration of period after EGD to post-EGD AUS (min)	73.8 ± 12.0 (53–110)

Data are number or mean ± SD (Range).

SD, standard deviation; BMI, body mass index; EGD, esophagogastroduodenoscopy; AUS, abdominal ultrasonography.

All 30 examinees underwent pre-EDG AUS and post-EGD AUS, and a total of 60 AUS moving images were obtained. The means and 95% confidence interval (CI) for the total image quality score for all 16 organs for pre- and post-EGD AUS were 3.54 [3.48–3.60] and 3.46 [3.39–3.52], respectively (Fig 1). The difference in the means was 0.08 of the score, which corresponds to 2% of the depiction rate. Calculated based on these data, the effect size (Cohen’s d) was 0.172, which is less than the value of 0.2, which is considered to be negligible.

Fig 1

Means and 95% CIs for total image quality score.

The means and 95% CI for the total image quality score for all 16 organs for pre- and post-EGD AUS are 3.54 [3.48–3.60] and 3.46 [3.39–3.52], respectively.

Fig 2 shows the result of the non-inferiority test of the post-EGD AUS in all 16 organs depicted as the difference in image quality score between the post- and pre-EGD AUS. Because the two-sided 95% CI of the post-EGD AUS did not cross the 0-outcome difference, it was statistical significantly different to the pre-EGD AUS. However, because the entire CI of the post-EGD AUS was above the non-inferiority margin, it was non-inferior to the pre-EGD AUS.

Fig 2

Non-inferiority test of post-EGD AUS.

The result of the non-inferiority test of the post-EGD AUS in all 16 organs depicted as the difference in image quality score between the post- and pre-EGD AUS.

Table 2 shows the means and 95% CIs of the image quality score for each organ at pre- and post-EGD AUS. Image quality scores for the pancreas tail, spleen, and celiac artery were <3, but those for other organs were >3. In all organs, the mean image quality for post-EGD AUS was less than that for pre-EGD AUS.

Table 2

Means and 95% CIs of image quality score for each organ at pre- and post-EGD AUS.

	pre-EGD AUS	post-EGD AUS
Pancreas Head	3.38 [3.18–3.57]	3.29 [3.01–3.57]
Pancreas Body	3.63 [3.43–3.83]	3.57 [3.35–3.78]
Pancreas Tail	2.41 [2.14–2.67]	2.35 [2.07–2.63]
Liver Right Lobe	3.97 [3.93–4.01]	3.89 [3.79–3.99]
Liver Left Lobe	3.98 [3.96–4.01]	3.94 [3.89–3.99]
Liver Caudate Lobe	3.78 [3.64–3.91]	3.64 [3.46–3.83]
GB	3.97 [3.93–4.00]	3.92 [3.84–3.99]
CBD	3.37 [3.18–3.56]	3.33 [3.07–3.58]
Spleen	2.93 [2.81–3.06]	2.93 [2.81–3.04]
Kidney Right	3.93 [3.89–3.98]	3.93 [3.88–3.97]
Kidney Left	3.84 [3.78–3.90]	3.82 [3.75–3.89]
Aorta	3.73 [3.56–3.89]	3.61 [3.38–3.84]
CA	3.13 [2.75–3.52]	2.98 [2.58–3.37]
SMA	3.30 [2.98–3.62]	3.05 [2.66–3.44]
PV	3.81 [3.68–3.93]	3.68 [3.50–3.87]
SV	3.46 [3.26–3.65]	3.38 [3.16–3.59]

CI, confidence interval; GB, Gallbladder; CBD, common biliary duct; CA, celiac artery; SMA, superior mesenteric artery; PV, portal vein; SV, splenic vein.

Fig 3 shows the results of the non-inferiority test of the post-EGD AUS in each organ, depicted as the difference in image quality score between the post- and pre-EGD AUS. In the celiac artery (CA) and superior mesenteric artery (SMA), non-inferiority of image quality of the post-EGD AUS could not be demonstrated when the margin was –0.40. In the other 14 organs, non-inferiority of image quality was demonstrated.

Fig 3

Non-inferiority test of post-EGD AUS in each organ.

The results of the non-inferiority test of the post-EGD AUS in each organ, depicted as the difference in image quality score between the post- and pre-EGD AUS.

Discussion

As well as its application as a diagnostic imaging test for patients with abdominal pain, AUS is also used for screening for metastasis in patients who have undergone resection for cancer [11-13], hepatocellular carcinoma in patients with viral hepatitis [14], pancreatic cancer in patients at high risk due to family history, obesity, and type 2 diabetes [15], among others. AUS is simple to perform, involves no ionizing radiation, and is less costly than CT and MRI. EGD is used as a diagnostic tool for patients with symptoms of acid-peptic disease and also in screening for Helicobacter pylori-related gastric cancer [16, 17] and for follow-up after endoscopic resection of esophageal cancer [18] and gastric cancer [11, 19]. Both AUS and EGD are used in medical check-ups for asymptomatic patients with no basal disease. For patients who require both AUS and EGD, it is preferred that these procedures be scheduled on the same day because they both need preparation of nil per os. Because bowel gas deteriorates the image quality of AUS, it is common to perform AUS prior to EGD, and both must be completed within a limited period in the morning with efficient setup. This one-way order of AUS prior to EGD limits the availability of examination appointments and reduces the efficiency of examination. In Japan, the government introduced endoscopic screening for gastric cancer as a national program in 2016. It has been reported that the program was difficult to introduce immediately because of insufficient medical resources [20]. Increasing examination efficiency will become even more important in the future. In addition to endoscopy, there is concern that materials such as oral radiocontrast can deteriorate the image quality of AUS performed subsequently, and previous studies have investigated whether such materials affect the image quality of AUS from the perspective of a time-saving strategy in the emergent situation [7, 8].

Room air, which is widely used for GI luminal distension in EGD, has the advantages of universal availability and low cost. However, room air is poorly absorbed by the GI tract, resulting in post-procedure pain related to distension. As CO2 is rapidly absorbed by the GI mucosa, there is increased interest in its use as an insufflation agent for endoscopic procedures. Many studies have shown the usefulness of CO2 in endoscopic procedures that are relatively lengthy, including colonoscopy, EGD with colonoscopy, endoscopic retrograde cholangiopancreatography (ERCP), balloon endoscopy, and endoscopic procedures such as endoscopic submucosal dissection (ESD) [2].

We hypothesized that the use of CO2 for insufflation in EGD would not deteriorate the image quality of subsequent AUS. To prove this hypothesis, we designed a non-inferiority test and demonstrated a negligible difference in the means of image quality of all 16 organs between pre- and post-EGD AUS, as well as a non-inferiority of post-EGD AUS. However, in the non-inferiority test, the two-sided 95% CI of post-EGD AUS did not cross the 0-outcome difference, which indicates a statistically significant difference from the pre-EGD AUS. This puzzling phenomenon might have resulted from having a sample size that was too large, or/and having too generous a non-inferiority margin [5]. An adequate sample size for this study had been calculated as around 26, but for a statistical study of all 16 organs in 30 cases, a significantly larger number of 480 samples was used. Regarding the non-inferiority margin, we set this at the point of –0.40, which is the same as a depiction rate of –10%, because we had considered a larger difference as clinically relevant. Furthermore, in the AUS studies after administration of oral radiocontrast, a one-point score difference in the five-point Likert scale was selected [7, 8]. Because the present –0.40-point margin of a four-point Likert-like scale is quite small compared with these reports, it is hard to say that our margin is too generous. Furthermore, because there is a possibility that the sample size was not adequate, we re-calculated the sample size using actual data obtained in the present study. Using a standard deviation (SD) of 0.69, which was calculated from the square root of the means of the sample variances of pre- and post-EGD AUS, the difference of means of 0.08, a power of 90%, and a 1-sided α-level of 0.025, resulted in a sample size of 18 subjects. The number of 26 subjects that we had estimated before the study is not much different from this re-calculated sample size. Ultimately, the fact that the effect size was 0.172 also supports our hypothesis.

Regarding each organ, even if all of the means of image quality score for post-EGD AUS for each organ were slightly less than those for pre-EGD AUS, 14/16 organs were demonstrated to be non-inferior. The two organs that were not demonstrated to be non-inferior were the celiac artery and superior mesenteric artery. It is not clear why the AUS image quality deteriorated after EGD in these two arteries. Because they are located close to each other, observation of the site at which these arteries branch seems to be poor even if CO2 insufflation is used. If the non-inferiority margin of –0.60, which is the same as a depiction rate of –15%, was acceptable rather than –0.40, all organs would be non-inferior.

We had expected that as CO2 would extrude the bowel gas downstream and be absorbed rapidly, the AUS image quality after EGD with CO2 insufflation would improve for some organs, such as the pancreas body and tail, where the quality is easily deteriorated by the presence of bowel gas in front of them. However, the mean image quality was reduced for all organs. Nakagawara et al. [4] reported that using CO2 insufflation instead of air improved the image quality of AUS at 60 min after EGD in some patients. They used a three-point Likert scale of better, unchanged, and worse for evaluation of image quality of AUS after EGD, compared with that of AUS images before EGD as control. They reported that image quality for the pancreas head and body, pancreas tail, and extrahepatic bile duct were judged better in 26.1% (6/23 cases), 26.1% (6/23 cases), and 43.5% (10/23 cases), respectively, for AUS after EGD. When the method for evaluating the depiction rate in our study was changed to that of difference in image quality, of better/unchanged/worse, we found 11/6/13 (better by 36.7%), 8/9/13 (26.7%), 12/7/11 (40%), and 12/8/10 (40%) cases in the pancreas head, body, tail, and common bile duct, respectively. In the present study, the image quality also improved in some patients post-EGD AUS (Fig 4), but for other patients the quality worsened, so it is misleading to state that CO2 insufflation improves AUS image quality even if this does occur in some patients. We concluded that CO2 insufflation in EGD does not improve image quality in AUS that is performed following EGD.

Fig 4

Image quality of pancreas body improved in some patients post-EGD AUS.

Nakagawara et al. [4] also reported that image quality post-EGD AUS depended on the duration of the period after EGD. They showed that the deterioration in quality at <15 min after EGD had recovered by 30 min or later. In the present study, because EGD was performed under planned moderate sedation, post-EGD AUS was performed 53–110 min after recovery from sedation. In the case of EGD performed without sedation, the shortest possible time between the end of EGD and AUS would be approximately 30 min.

Several limitations of the present study must be acknowledged. First, as the subjects were examinees for medical check-ups and enrolled in this study at our single institution, selection bias may have affected the results, particularly with regard to the subject characteristics. Second, in the absence of published data, the margin of non-inferiority was selected because we considered a larger difference to be clinically relevant. This margin size may have affected interpretation of the results. Third, because we demonstrated only the non-inferiority of the depiction rate, there is a possibility that EGD using CO2 insufflation may affect the image characteristics of certain diseases in the subsequent AUS examination.

Conclusions

The use of CO2 for insufflation in EGD does not cause much deterioration in the image quality of AUS performed subsequently. Therefore, it is permissible to perform EGD prior to AUS, which is expected to improve the efficiency of examination setup.

Dataset of examinees’ characteristics and Likert-like scales evaluated by the four reviewers for 16 organs in thirty subjects.

(XLSX)

Click here for additional data file.

12 Jul 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Gopal Krishna Dhali, MBBS, MD, DM

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf  and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

Additional Editor Comments:

Dear authors,

I regret to inform you that the statistical reviewer has raised major concerns regarding your manuscript. Kindly find the detailed comments below and make necessary comments if you want us to reconsider the manuscript.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: No

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: No

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Yes, I feel that the authors have written the manuscript in an intelligible fashion and in standard English. However, there are a few typographical and grammatical errors which I have requested for corrections in a separate word file attachment. Regarding the statistical tools and setting up of the non-inferiority margin for the study, I do not consider myself to be competent enough and request the editor to take an opinion from a statistician.

Reviewer #2: PONE-D-22-10775: Statistical review

SUMMARY. This is a study to test whether the use of CO2 insufflation in esophagogastroduodenoscopy (EGD) does not deteriorate the image quality of abdominal ultrasonography (AUS) if this is performed subsequently. Although the research question is clearly stated, I have serious concerns about the statistical analysis which, in my opinion, should be fully revised: see major points 1-3 below. I also append a list of specific points that should be addressed.

MAJOR POINTS

1. Linkert scale. Although the outcome of interest is a continuous variable, the depiction rate, this value is reduced to a (ordinal) categorical variable through a Linkert scale. It doesn't seem a good idea, for at least three reasons. First, transforming a continuous variable to a categorical one is a waste of information and it is in general not recommended. Second, mean differences between Linkert-based scores are not mathematically sound (ordinal variables are better analyzed by quantiles) and, as a result, they are difficult to interpret (see also specific point 3 below). Third, Linkert-based scores are obviously not normally distributed and, as a result, conventional confidence intervals are inappropriate. If the authors are willing to use a Likert scale, this choice should be well motivated and the statistical analysis should be based on multinomial tests. Otherwise, I'd suggest to exploit the depiction rate as the outcome of interest: if it is normally distributed, it can be examined under traditional inference methods.

2. One side test. Although an analysis based on confidence intervals is not incorrect, a more natural approach to test non-inferiority would rely on one-side statistical tests where the null hypothesis is H_0: mean < a_0 and the alternative is H_1: mean > a_0 (a_0 being the desired threshold). Under this setting, p-values and powers (and sample sizes) are easier to compute and the results are easier to interpret.

3. Organ-specific differences. In a second step of the analysis, the sample is stratified according to the type of organ and the analysis is separately repeated withing each stratum (Table 2). Again, this approach is not totally incorrect, but it is based on an unnecessary waste of information due to the reduction of sample size. A more natural approach would rely on a regression (multinomial regression in the case of a Linkert-based categorical outcome or a more traditional regression in the case of a continuous dependent variable) where differences across organs is examined simultaneously.

4. Covariates. The available covariates are summarized in Table 1, but they are afterwards totally ignored in the subsequent analysis. Is it possible that image quality is somehow affected by these covariates? If it is, image quality differences should be adjusted by covariate value, under a traditional regression setting (see also major issue 3 for an another reason that would motivate a regression approach).

SPECIFIC POINTS

1. Figures 2 and 3 are totally black!

2. Depiction rate is the main outcome and should be precisely defined. How is it computed?

3. The Likert score is interpreted as the depiction rate divided by 4 (see lines 31 and 36). I don’t understand: the Linkert score clusters depictions rates into 4 intervals, hence the Linkert score is not necessarily the depiction rate divided by 4. Please clarify.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

Submitted filename: PLOS.docx

Click here for additional data file.

30 Jul 2022

Point-by-point responses to the Editor and Reviewers

Editor

Comment: I regret to inform you that the statistical reviewer has raised major concerns regarding your manuscript. Kindly find the detailed comments below and make necessary comments if you want us to reconsider the manuscript.

Response: Thank you for your assessment. We have addressed the concerns of the reviewers, with special attention to the major concerns of the statistical reviewer (Reviewer #2), as noted below.

Reviewer #1

Comment: I feel that the authors have written the manuscript in an intelligible fashion and in standard English. However, there are a few typographical and grammatical errors which I have requested for corrections in a separate word file attachment. Regarding the statistical tools and setting up of the non-inferiority margin for the study, I do not consider myself to be competent enough and request the editor to take an opinion from a statistician.

Response: Thank you for your assessment. We have addressed the points you raised.

Reviewer comments on the manuscript titled,” Image quality of abdominal ultrasonography after esophagogastroduodenoscopy is preserved by using carbon dioxide insufflation: A non-inferiority test in the same subject”.

Comment: It is always considered better if a patient can undergo USG abdomen (AUSG) and Esophagogastroduodenoscopy (EGD) on the same day one after another. This would reduce the problems associated with the delay in the allocation of dates for the two procedures on the same subject. However due to the air inside the stomach and the duodenum, following EGD, the USG abdomen cannot be done on account of poor visibility. It is hypothesized that this can be overcome if CO2 is used as the insufflating agent in place of air. In order to prove this hypothesis, the authors have pursued the study. The strengths of the study are 1) the adequate sample size of the study subjects and 2) the blinding of the interpreters of the USG images done by three experienced USG operators. I would however like to obtain a few responses from the authors regarding the following issues:

Response: Thank you for your assessment.

Comment 1: Figure 1 does not have any title, legend or label. The X-axis and the Y-axis should be labelled.

Response: Thank you. Your comment made us aware that we had made a mistake in creating the figure, and we have revised it accordingly.

Comment 2: The authors have mentioned the study type as a prospective one. However, there hasn’t been any follow-up of the subjects. Therefore it is a cross-sectional study and not a prospective one.

Response: A cross-sectional study is defined as a research design in which researchers collect data from many different individuals at a single point in time. In cross-sectional research, researchers observe variables without intervening. In a longitudinal study, researchers repeatedly examine the same individuals to detect any changes that might occur over a period of time. Longitudinal studies are a type of correlational research in which researchers observe and collect data on a number of variables without trying to affect the variables. Because we collected USG data at two different time points, and we intervened by performing EGD between them, our study is neither a cross-sectional nor a simple longitudinal study. Even though it was only a few hours, the subjects were followed up, so that our study can be considered a prospective study.

Comment 3: USG of an organ is an operator-dependent procedure. That some of the images following EGD appeared better to the interpreters as compared to the images done before EGD emphasizes this fact only. I would request the authors to comment on this fact with reference to the results of the study.

Response: As you point out, USG of an organ is an operator-dependent procedure. To reduce this systematic error, the same technologist performed both pre- and post-EGD USG in our study. In fact, we did not conclude that the image following EGD improved. Nakagawara et al. [4] reported that using CO2 insufflation instead of air improved the image quality of USG at 60 min after EGD in some patients. Contrary to their report, we concluded that CO2 insufflation in EGD does not improve image quality in USG that is performed following EGD.

Comment 4: P values with confidence intervals were not demonstrated while comparing the USG image quality scores of the different organs before and after EGD. I would request the authors to provide that.

Response: We designed this as a non-inferiority study. As explained in reference 5 (Ahn S, Park SH, Lee KH. How to demonstrate similarity by using non-inferiority and equivalence statistical testing in radiology research. Radiology. 2013;267(2):328-38. doi: 10.1148/radiol.12120725.), neither a statistically significant difference between groups (P < .05) nor a lack of significant difference (P ≥ .05) from conventional statistical tests provides answers about equivalence/non-inferiority. Furthermore, because sample size was calculated for the non-inferiority test, not for a significant difference test, it is inappropriate to report p values in our study.

Comment 5: There are a few typographical and grammatical errors which the authors need to correct. These are as mentioned below.

Line 89: Insert “and” between AUS and EGD.

Response: The sentence is “Each subject underwent AUS, EGD with CO2 insufflation instead of air under planned moderate sedation, and a second AUS, in this order, within the same day”. This can be summarized as “Each subject underwent AUS, EGD, and a second AUS, in this order”. What you suggested is that “Each subject underwent AUS and EGD, and a second AUS, in this order”, in which the order of the first AUS and EGD are not explicitly stated and are, therefore, slightly unclear. Therefore, we left this sentence as it is.

Line 97: “sonologists was” should be” sonologists were”.

Response: The sentence was “One of the three sonologists was board-certified by the Japan Society of Ultrasonics in Medicine”. In other words, “One sonologist was board-certified by the Japan Society of Ultrasonics in Medicine”. Therefore, we left this sentence as it is.

Line 120: “was” should be “were”

Response: The sentence was “The mean of the evaluation score by the four reviewers was used as the final score for each organ in each subject”. This can be summarized as “The mean was used as the final score”. What you suggested is “The mean were used as the final score”, which is grammatically incorrect. Therefore, we left this sentence as it is.

Line 121: “as to” should be removed.

Response: We have revised the paper accordingly.

Changes: The following has been added to the revised manuscript on lines 120-122, and the revisions are shown in red.

The reviewers were blinded to the time points of the moving images, including as to whether the images they were reviewing were obtained before or after EGD.

→

The reviewers were blinded to the time points of the moving images, including whether the images they were reviewing were obtained before or after EGD.

Line 153: “mean” should be “the mean”

Response: We have revised the paper accordingly.

Changes: The following has been added to the revised manuscript on lines 152-153, and the revisions are shown in red.

Mean body mass index (BMI) was 23.0 kg/m2 and was >30 kg/m2 in only 1/30 subjects. Mean procedure time of EGD was 4.7 min and the mean duration of the period after EGD to post-EGD AUS was 73.8 min (range, 53–110 min).

→

The mean body mass index (BMI) was 23.0 kg/m2 and was >30 kg/m2 in only 1/30 subjects. The mean procedure time of EGD was 4.7 min and the mean duration of the period after EGD to post-EGD AUS was 73.8 min (range, 53–110 min).

Line 166: “considered to be” should be” is considered to be”.

Response: We have revised the paper accordingly.

Changes: The following has been added to the revised manuscript on lines 165-166, and the revisions are shown in red.

Calculated based on these data, the effect size (Cohen’s d) was 0.172, which is less than the value of 0.2 considered to be negligible.

→

Calculated based on these data, the effect size (Cohen’s d) was 0.172, which is less than the value of 0.2, which is considered to be negligible.

Line 214: “require” should be” requires.”

Response: For clarity, we have revised this sentence.

Changes: The following has been added to the revised manuscript on lines 213-215, and the revisions are shown in red.

Patients who require both AUS and EGD prefer to schedule these procedures for the same day because they both need preparation of nil per os.

→

For patients who require both AUS and EGD, it is preferred that these procedures be scheduled on the same day because they both need preparation of nil per os.

Line 247: for “a” statistical study.

Response: We have revised it accordingly.

Changes: The following has been added to the revised manuscript on line 247, and the revisions are shown in red.

An adequate sample size for this study had been calculated as around 26, but for statistical study of all 16 organs in 30 cases, a significantly larger number of 480 samples was used.

→

An adequate sample size for this study had been calculated as around 26, but for a statistical study of all 16 organs in 30 cases, a significantly larger number of 480 samples was used.

Line 248: samples “was” should be “were”

Response: The sentence is “An adequate sample size for this study had been calculated as around 26, but for statistical study of all 16 organs in 30 cases, a significantly larger number of 480 samples was used”. This can be summarized as, “An adequate sample size had been calculated as around 26, but for statistical study, a significantly larger number was used”. What you suggested is “a significantly larger number were used”, which is grammatically incorrect. Therefore, we left this sentence as it is.

Line 270: is same should be is “the” same.

Response: We have revised the paper accordingly.

Changes: The following has been added to the revised manuscript on line 270, and the revisions are shown in red.

If the non-inferiority margin of –0.60, which is same as a depiction rate of –15%, was acceptable rather than –0.40, all organs would be non-inferior.

→

If the non-inferiority margin of –0.60, which is the same as a depiction rate of –15%, was acceptable rather than –0.40, all organs would be non-inferior.

Line 287 image quality should be “the” image quality.

Response: We have revised the paper accordingly.

Changes: The following has been added to the revised manuscript on line 287, and the revisions are shown in red.

In the present study, image quality also improved in some patients post-EGD AUS (Fig 4), but for other patients the quality worsened, so it is misleading to state that CO2 insufflation improves AUS image quality even if this does occur in some patients.

→

In the present study, the image quality also improved in some patients post-EGD AUS (Fig 4), but for other patients, the quality worsened, so it is misleading to state that CO2 insufflation improves AUS image quality, even if it does occur in some patients.

Reviewer #2: PONE-D-22-10775: Statistical review

Comment: SUMMARY. This is a study to test whether the use of CO2 insufflation in esophagogastroduodenoscopy (EGD) does not deteriorate the image quality of abdominal ultrasonography (AUS) if this is performed subsequently. Although the research question is clearly stated, I have serious concerns about the statistical analysis which, in my opinion, should be fully revised: see major points 1-3 below. I also append a list of specific points that should be addressed.

Response: We recognize that there are some issues in the statistical analysis in which we differ. The concerns you have raised seem to be common, but we believe the two cited reports, references 5 and 10, have already resolved them. On the other hand, we do not believe it is possible to explain all statistical concepts in our paper, and we must assume that readers refer to these two references contained in our paper. Fortunately, PLOS ONE now offers accepted authors the opportunity to publish the peer review history, in which we would like to address readers who may have similar concerns here. We have addressed the points you have raised.

MAJOR POINTS

Comment:1. Linkert scale. Although the outcome of interest is a continuous variable, the depiction rate, this value is reduced to a (ordinal) categorical variable through a Linkert scale. It doesn't seem a good idea, for at least three reasons. First, transforming a continuous variable to a categorical one is a waste of information and it is in general not recommended. Second, mean differences between Linkert-based scores are not mathematically sound (ordinal variables are better analyzed by quantiles) and, as a result, they are difficult to interpret (see also specific point 3 below). Third, Linkert-based scores are obviously not normally distributed and, as a result, conventional confidence intervals are inappropriate. If the authors are willing to use a Likert scale, this choice should be well motivated and the statistical analysis should be based on multinomial tests. Otherwise, I'd suggest to exploit the depiction rate as the outcome of interest: if it is normally distributed, it can be examined under traditional inference methods.

Response: We agree that transforming a continuous variable to a categorical one is a waste of information. However, a method for evaluating the depiction rate of organs in AUS as continuous variable has not been established, and previous studies have used a Likert scale. To resolve this problem, we developed a more objective scale by replacing the abstract evaluation with the depiction rate as an interval scale more suitable for statistical analysis. We also considered that the minimum interval to distinguish the difference in the depiction rate subjectively by gastroenterologists who are familiar with AUS image evaluation is 25%. We mentioned this point in the Review of ultrasonograms in the Material and Methods section, as follows: A method for objectively estimating the quality of AUS images has not been established. Previous studies [3, 4, 7-9] have used three- to five-point Likert scales such as: 1 - not interpretable; 2 - barely interpretable; 3 - adequate for interpretation but of poor quality; 4 - interpretable and of average quality; and 5 - interpretable and of superior quality. In this study, we developed a more objective scale by replacing the abstract evaluation with the depiction rate as an interval scale suitable for statistical analysis [10]. All saved AUS moving images, which were marked only with a study code, were randomized and then evaluated. To reduce systematic error, four expert reviewers, board-certified by the Japanese Society of Gastroenterology, rated image quality in all moving images using a four-point Likert scale: 1 - depiction rate of 0%–24%; 2 - 25%–49%; 3 - 50%–74%; and 4 - 75%–100%. The mean of the evaluation score by the four reviewers was used as the final score for each organ in each subject.

The second point, that mean differences between Likert-based scores are not mathematically sound, and the third point, that Likert-based scores are obviously not normally distributed and, as a result, conventional confidence intervals are inappropriate, are true, strictly speaking. However, what is left unsaid is how much the chance of an erroneous conclusion is increased. This is what statisticians call “robustness”, the extent to which the test will give the right answer even when assumptions are violated. This point is described in reference 10 (Norman G. Likert scales, levels of measurement and the "laws" of statistics. Adv Health Sci Educ Theory Pract. 2010;15(5):625-32. doi: 10.1007/s10459-010-9222-y.). Norman also said that “Parametric statistics can be used with Likert data, with unequal variances, and with non-normal distributions, with no fear of “coming to the wrong conclusion”. These findings are consistent with empirical literature dating back nearly 80 years. The controversy can cease (but likely won’t).” We designed our study based on this “robustness”.

Comment: 2. One side test. Although an analysis based on confidence intervals is not incorrect, a more natural approach to test non-inferiority would rely on one-side statistical tests where the null hypothesis is H_0: mean < a_0 and the alternative is H_1: mean > a_0 (a_0 being the desired threshold). Under this setting, p-values and powers (and sample sizes) are easier to compute and the results are easier to interpret.

Response: For non-inferiority testing in which one is interested to know if the sensitivity (an index in which a larger value represents a better outcome) of the new test (PT) is not worse than the sensitivity of the active control test (PAC) without regard to its superiority to the active control test (H0: PT - PAC ≤ -Δ versus H1: PT - PAC > -Δ), only the relationship between the lower bound of the CI of PT - PAC and the non-inferiority margin (i.e., the lower bound of the equivalence range in this example) matters. Therefore, both one-sided and two-sided CIs of PT - PAC can be used for the analysis. This is described in reference 5 (Ahn S, Park SH, Lee KH. How to demonstrate similarity by using non-inferiority and equivalence statistical testing in radiology research. Radiology. 2013;267(2):328-38. doi: 10.1148/radiol.12120725.). They also state that neither a statistically significant difference between groups (P < .05) nor a lack of significant difference (P ≥ .05) from conventional statistical tests provides answers about equivalence/non-inferiority.

Comment: 3. Organ-specific differences. In a second step of the analysis, the sample is stratified according to the type of organ and the analysis is separately repeated withing each stratum (Table 2). Again, this approach is not totally incorrect, but it is based on an unnecessary waste of information due to the reduction of sample size. A more natural approach would rely on a regression (multinomial regression in the case of a Linkert-based categorical outcome or a more traditional regression in the case of a continuous dependent variable) where differences across organs is examined simultaneously.

Response: An adequate sample size for our non-inferiority test had been calculated as around 26. We enrolled 30 cases, and evaluated each organ in each case. This point was repeatedly described in our paper. Thus, the evaluation of organ-specific differences does not reduce the sample size, as you point out. On the contrary, in the statistical study for the primary endpoint, all 16 organs in 30 cases, a much larger number of 480 samples was used, which resulted in a puzzling phenomenon that both non-inferiority and a significant difference were demonstrated (Fig. 2). We mentioned this point in the Discussion section. Because gastroenterologists interested in our study realize that the depiction rate in daily practice may differ for each organ, it is important to report the mean value and confidence interval for each organ. On the other hand, this study is not aimed at statistically examining the differences across organs, and the primary endpoint is to prove non-inferiority of post-EGD AUS. Therefore, this study was designed as a non-inferiority study, not as a regression study to describe relationships between variables by fitting a line to the observed data.

Comment: 4. Covariates. The available covariates are summarized in Table 1, but they are afterwards totally ignored in the subsequent analysis. Is it possible that image quality is somehow affected by these covariates? If it is, image quality differences should be adjusted by covariate value, under a traditional regression setting (see also major issue 3 for an another reason that would motivate a regression approach).

Response: Our study design was a non-inferiority test in the same subjects. Therefore, the subjects’ characteristics summarized in Table 1 do not affect the results for non-inferiority of post-EGD AUS; for that reason, the variates are not covariates.

SPECIFIC POINTS

Comment: 1. Figures 2 and 3 are totally black!

Response: Thank you. Your comment made us aware that we had made a mistake in creating the figures, and we have revised them accordingly.

Comment: 2. Depiction rate is the main outcome and should be precisely defined. How is it computed?

Response: The depiction rate was not calculated. We mentioned this point in the Review of ultrasonograms in the Material and Methods section, as follows: A method for objectively estimating the quality of AUS images has not been established. Previous studies [3, 4, 7-9] have used three- to five-point Likert scales such as 1 - not interpretable, 2 - barely interpretable, 3 - adequate for interpretation but of poor quality, 4 - interpretable and of average quality, 5 - interpretable and of superior quality. In this study, we developed a more objective scale by replacing the abstract evaluation with the depiction rate as an interval scale suitable for statistical analysis [10]. All saved AUS moving images, which were marked only with a study code, were randomized and then evaluated. To reduce systematic error, four expert reviewers board-certified by the Japanese Society of Gastroenterology rated image quality in all moving images using a four-point Likert scale: 1 - depiction rate of 0%–24%; 2 - 25%–49%; 3 - 50%–74%; and 4 - 75%–100%. The mean of the evaluation score by the four reviewers was used as the final score for each organ in each subject. The reviewers were blinded to the time points of the moving images, including as to whether the images they were reviewing were obtained before or after EGD. The reviewers were also blinded to the identity of the sonologist who performed the scan, to the study subject being imaged, and to the scores of the other reviewers.

Comment: 3. The Likert score is interpreted as the depiction rate divided by 4 (see lines 31 and 36). I don’t understand: the Linkert score clusters depictions rates into 4 intervals, hence the Linkert score is not necessarily the depiction rate divided by 4. Please clarify.

Response: We used the Likert scale as an interval scale. We mentioned this point in the Review of ultrasonograms in the Material and Methods section as follows: A method for objectively estimating the quality of AUS images has not been established. Previous studies [3, 4, 7-9] have used three- to five-point Likert scales such as 1 - not interpretable, 2 - barely interpretable, 3 - adequate for interpretation but of poor quality, 4 - interpretable and of average quality, 5 - interpretable and of superior quality. In this study, we developed a more objective scale by replacing the abstract evaluation with the depiction rate as an interval scale suitable for statistical analysis [10]. All saved AUS moving images, which were marked only with a study code, were randomized and then evaluated. To reduce systematic error, four expert reviewers board-certified by the Japanese Society of Gastroenterology rated image quality in all moving images using a four-point Likert scale: 1 - depiction rate of 0%–24%; 2 - 25%–49%; 3 - 50%–74%; and 4 - 75%–100%. The mean of the evaluation score by the four reviewers was used as the final score for each organ in each subject. So, it is possible that the image quality score evaluated using our 4-point Likert scale corresponds to the depiction rate.

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

15 Aug 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Gopal Krishna Dhali, MBBS, MD, DM

Academic Editor

PLOS ONE

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: (No Response)

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: No

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: I Don't Know

Reviewer #2: No

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors have responded to my questions regarding the manuscript titled, “Image quality of abdominal ultrasonography after esophagogastroduodenoscopy is preserved by using carbon dioxide insufflation: A non-inferiority test in the same subject” satisfactorily.

There was however one comment to which I felt the authors did not explain clearly. This was the response to comment no 3. The authors said they did not conclude that the USG images improved after EGD using Co2 insufflation. Indeed, they did not conclude in that manner but this is also true that they wrote about their observation that the image quality improved in some patients after EGD using Co2 insufflation (page 14, lines 287,288) which I feel has no scientific explanation rather than the inherently subjective nature (operator dependence) of the investigation under question. This factor was bound to affect the assessment of the USG images in the study which was assessed based on the four-point Likert scale. However, they have rightly concluded that it is misleading to state that CO2 insufflation improves AUS image quality even if this does occur in some patients.

I feel that in the absence of a validated pre-existing tool for assessment of the differences in the quality of the two USG images done on the same subject in a short time gap, this paper has merit to be published, provided the statistical part (assessed by reviewer 2) is sound enough.

Reviewer #2: PONE-D-22-10775: Statistical review

In my previous review, I raised 4 major issues and 3 specific points. All specific points have been addressed and the authors have provided reasonable answers to major issues 2-4 (thanks). However, I can't accept their answer to major issue 1.

From the answer to specific point no 2, I understand the depiction rate was not computed but it was summarized by the levels of a Linkert scale. Under this setting, two important points raised in my previous review are still unsolved, namely

1) mean differences between Linkert-based scores are not mathematically sound (ordinal variables are better analyzed by quantiles) and, as a result, they are difficult to interpret

2) Linkert-based scores are obviously not normally distributed and, as a result, conventional confidence intervals are inappropriate.

The authors agree with me on these two points, but they claim that their methods are robust with respect to the violations of these assumptions, citing reference 10. It is possible that some methods of the paper are robust, but robustness must be quantified. Unfortunately, I don’t see a simple way to justify robustness under point 1 above, because, as the authors say, we don’t have the continuous value of the depiction rate. Point 2 would instead be addressed by an appropriate simulation study, which is the standard approach to evaluate robustness; citing reference 10 is not enough because the degree of robustness varies from one study to another.

I still believe that trying to prove robustness is not the easiest approach in this study. Acknowledging the qualitative nature of the Linkert scale and using multinomial-based methods (as I suggested in my previous review) is much simpler. Furthermore, robustness is usually invoked when there are not standard methods to avoid assumptions violation. However, multinomial methods are standard and available in many statistical packages.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

**********

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

2 Sep 2022

Point-by-point responses to the Editor and Reviewers

Reviewer #1

Comment: The authors have responded to my questions regarding the manuscript titled, “Image quality of abdominal ultrasonography after esophagogastroduodenoscopy is preserved by using carbon dioxide insufflation: A non-inferiority test in the same subject” satisfactorily.

There was however one comment to which I felt the authors did not explain clearly. This was the response to comment no 3. The authors said they did not conclude that the USG images improved after EGD using CO2 insufflation. Indeed, they did not conclude in that manner but this is also true that they wrote about their observation that the image quality improved in some patients after EGD using CO2 insufflation (page 14, lines 287,288) which I feel has no scientific explanation rather than the inherently subjective nature (operator dependence) of the investigation under question. This factor was bound to affect the assessment of the USG images in the study which was assessed based on the four-point Likert scale. However, they have rightly concluded that it is misleading to state that CO2 insufflation improves AUS image quality even if this does occur in some patients. I feel that in the absence of a validated pre-existing tool for assessment of the differences in the quality of the two USG images done on the same subject in a short time gap, this paper has merit to be published, provided the statistical part (assessed by reviewer 2) is sound enough.

Response: We believe that the sentence “In the present study, image quality also improved in some patients post-EGD AUS (Fig 4), but for other patients the quality worsened, so it is misleading to state that CO2 insufflation improves AUS image quality even if this does occur in some patients” is necessary to disprove the claims of previous papers (reference 4) and to negate the hypothesis that CO2 insufflation could improve image quality, which we had expected at the time of planning our study and is of interest to the readers. Thank you for your valuable comments.

Reviewer #2

Comment: In my previous review, I raised 4 major issues and 3 specific points. All specific points have been addressed and the authors have provided reasonable answers to major issues 2-4 (thanks). However, I can't accept their answer to major issue 1.

From the answer to specific point no 2, I understand the depiction rate was not computed but it was summarized by the levels of a Linkert scale. Under this setting, two important points raised in my previous review are still unsolved, namely

1) mean differences between Linkert-based scores are not mathematically sound (ordinal variables are better analyzed by quantiles) and, as a result, they are difficult to interpret

2) Linkert-based scores are obviously not normally distributed and, as a result, conventional confidence intervals are inappropriate.

The authors agree with me on these two points, but they claim that their methods are robust with respect to the violations of these assumptions, citing reference 10. It is possible that some methods of the paper are robust, but robustness must be quantified. Unfortunately, I don’t see a simple way to justify robustness under point 1 above, because, as the authors say, we don’t have the continuous value of the depiction rate. Point 2 would instead be addressed by an appropriate simulation study, which is the standard approach to evaluate robustness; citing reference 10 is not enough because the degree of robustness varies from one study to another.

I still believe that trying to prove robustness is not the easiest approach in this study. Acknowledging the qualitative nature of the Linkert scale and using multinomial-based methods (as I suggested in my previous review) is much simpler. Furthermore, robustness is usually invoked when there are not standard methods to avoid assumptions violation. However, multinomial methods are standard and available in many statistical packages.

Response: Our present study was designed and conducted based on statistical methods described in “Reviews and Commentary” of Radiology (reference 5), the most authoritative journal in diagnostic imaging. In clinical research, some flexibility is required to find a compromise between limitations in sample collection (in the evaluation of image quality, the limitation is that there is no method other than the Likert scale) and statistical correctness. In this review, the multinomial regression method that you are recommending is not mentioned, whereas a non-inferiority test using the mean difference is recommended to demonstrate non-inferiority of the outcome of one group to the outcome of another group, because non-inferiority cannot be proven correctly by multinomial regression methods. Furthermore, use of the mean difference between ordinal variables, such as Likert-based scores or grading scores is accepted. In fact, a study that is mentioned in this review used a non-inferiority test using the mean difference between grading scores. Please see the contents in the Abstract of that study below.

Hausleiter J, Martinoff S, Hadamitzky M, Martuscelli E, Pschierer I, Feuchtner GM, et al. Image quality and radiation exposure with a low tube voltage protocol for coronary CT angiography results of the PROTECTION II Trial. JACC Cardiovasc Imaging. 2010;3(11):1113-23.

Abstract

Objectives: The purpose of this study was to evaluate image quality and radiation dose using a 100 kVp tube voltage scan protocol compared with standard 120 kVp for coronary computed tomography angiography (CTA).

Background: Concerns have been raised about radiation exposure during coronary CTA. The use of a 100 kVp tube voltage scan protocol effectively lowers coronary CTA radiation dose compared with standard 120 kVp, but it is unknown whether image quality is maintained.

Methods: We enrolled 400 nonobese patients who underwent coronary CTA: 202 patients were randomly assigned to a 100 kVp protocol and 198 patients to a 120 kVp protocol. The primary end point was to demonstrate noninferiority in image quality with the 100 kVp protocol, which was assessed by a 4-point grading score (1 = nondiagnostic, 4 = excellent image quality). For the noninferiority analysis, a margin of -0.2 image quality score points for the difference between both scan protocols was pre-defined. Secondary end points included radiation dose and need for additional diagnostic tests during follow-up.

Results: The mean image quality scores in patients scanned with 100 kVp and 120 kVp were 3.30 ± 0.67 and 3.28 ± 0.68, respectively (p = 0.742); image quality of the 100 kVp protocol was not inferior, as demonstrated by the 97.5% confidence interval of the difference, which did not cross the pre-defined noninferiority margin of -0.2. The 100 kVp protocol was associated with a 31% relative reduction in radiation exposure (dose-length product: 868 ± 317 mGy × cm with 120 kVp vs. 599 ± 255 mGy × cm with 100 kVp; p < 0.0001). At 30-day follow-up, the need for additional diagnostic studies did not differ (13.4% vs. 19.2% for 100 kVp vs. 120 kVp, respectively; p = 0.114).

Conclusions: A coronary CTA protocol using 100 kVp tube voltage maintained image quality, but reduced radiation exposure by 31% as compared with the standard 120 kVp protocol. Thus, 100 kVp scan protocols should be considered for nonobese patients to keep radiation exposure as low as reasonably achievable. (Prospective Randomized Trial on Radiation Dose Estimates of Cardiac CT Angiography in Patients Scanned With a 100 kVp Protocol [PROTECTION II]; NCT00611780).

The sentence in your comment “as the authors say, we don’t have the continuous value of the depiction rate” appears to potentially be the result of a misunderstanding. We stated that “A method for evaluating the depiction rate of organs in AUS as continuous variables has not been established, and previous studies have used a Likert scale. To resolve this problem, we developed a more objective scale by replacing the abstract evaluation with the depiction rate as an interval scale more suitable for statistical analysis”. Again, the Likert scale we used is not an ordinal scale like the conventional Likert scale. It is an interval scale, even if it cannot be measured finely. Variables measured by the interval scale are continuous variables. Because it seems that the use of the word “Likert scale” is misleading, we have changed it to “Likert-like scale” in the revised manuscript.

There also appears to be a potential misunderstanding regarding reference 10. The paper (reference 10) is not just a single study. This paper shows that many studies, dating back to the 1930s, consistently show that parametric statistics are robust with respect to violations of many assumptions. Please see the contents in the Abstract below.

Reference 10. Norman G. Likert scales, levels of measurement and the "laws" of statistics. Adv Health Sci Educ Theory Pract. 2010;15(5):625-32. doi: 10.1007/s10459-010-9222-y.

Abstract

Reviewers of research reports frequently criticize the choice of statistical methods. While some of these criticisms are well-founded, frequently the use of various parametric methods such as analysis of variance, regression, correlation are faulted because: (a) the sample size is too small, (b) the data may not be normally distributed, or (c) The data are from Likert scales, which are ordinal, so parametric statistics cannot be used. In this paper, I dissect these arguments, and show that many studies, dating back to the 1930s consistently show that parametric statistics are robust with respect to violations of these assumptions. Hence, challenges like those above are unfounded, and parametric methods can be utilized without concern for ‘‘getting the wrong answer’’.

Although our present study is a clinical one with limited collection of samples with fine measurements, we believe that our analysis is based on recommended statistical methods by the authoritative journal, and that we have made every effort to be statistically accurate.

Changes 1: The following has been added to the revised manuscript on lines 113-119, and the revisions are shown in red.

In this study, we developed a more objective scale by replacing the abstract evaluation with the depiction rate as an interval scale suitable for statistical analysis [10]. All saved AUS moving images, which were marked only with a study code, were randomized and then evaluated. To reduce systematic error, four expert reviewers board-certified by the Japanese Society of Gastroenterology rated image quality in all moving images using a four-point Likert scale: 1 - depiction rate of 0%–24%, 2 - 25%–49%, 3 - 50%–74%, 4 - 75%–100%.

→

In this study, we developed a more objective scale by replacing the abstract evaluation with the depiction rate as an interval scale suitable for statistical analysis [10]. All saved AUS moving images, which were marked only with a study code, were randomized and then evaluated. To reduce systematic error, four expert reviewers board-certified by the Japanese Society of Gastroenterology rated image quality in all moving images using a four-point Likert-like scale as an interval scale: 1 - depiction rate of 0%–24%, 2 - 25%–49%, 3 - 50%–74%, 4 - 75%–100%.

Changes 2: The following has been added to the revised manuscript on lines 28-30, and the revisions are shown in red.

All saved AUS moving images were randomized and imaging quality was evaluated at 16 organs using a four-point Likert scale that divides the depiction rate by 25%.

→

All saved AUS moving images were randomized and imaging quality was evaluated at 16 organs using a four-point Likert-like scale that divides the depiction rate by 25%.

Changes 3: The following has been added to the revised manuscript on lines 252-254, and the revisions are shown in red.

Because the present –0.40-point margin of a four-point Likert scale is quite small compared with these reports, it is hard to say that our margin is too generous.

→

Because the present –0.40-point margin of a four-point Likert-like scale is quite small compared with these reports, it is hard to say that our margin is too generous.

Changes 4: The following has been added to the revised manuscript on lines 393-394, and the revisions are shown in red.

S1 Table. Dataset of examinees' characteristics and Likert scales evaluated by the four reviewers for 16 organs in thirty subjects.

→

S1 Table. Dataset of examinees' characteristics and Likert-like scales evaluated by the four reviewers for 16 organs in thirty subjects.

Submitted filename: Response to Reviewers 2nd. revise.docx

Click here for additional data file.

13 Sep 2022

Image quality of abdominal ultrasonography after esophagogastroduodenoscopy is preserved by using carbon dioxide insufflation: A non-inferiority test in the same subject

PONE-D-22-10775R2

Dear Dr. Shirota,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Gopal Krishna Dhali, MBBS, MD, DM

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: (No Response)

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: (No Response)

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: (No Response)

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: (No Response)

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: The authors have clarified a critical misunderstanding. In this revision, the authors remark that the Likert scale used in the paper is not the conventional Likert scale, but it is an interval scale. Under this setting, differences can be computed although they are approximated.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: No

**********

20 Sep 2022

PONE-D-22-10775R2

Image quality of abdominal ultrasonography after esophagogastroduodenoscopy is preserved by using carbon dioxide insufflation: A non-inferiority test in the same subject

Dear Dr. Shirota:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Gopal Krishna Dhali

Academic Editor

PLOS ONE