Review of article titled: Abdominal Imaging Findings in COVID-19: Preliminary Observations. /doi.org/10.1148/radiol.2020201908

Overall statement: This is a retrospective, cross sectional, single center study describing the abdominal imaging findings of novel SARS-CoV-2 positive patients. The results of this study show that Covid positive ICU patients have significantly higher number of abdominal studies and that they have significantly higher bowel wall thickening, pneumatosis and fluid filled colon when compared to COVID positive in-patients.

Overall strengths of the article: The novelty and the current situation of world amidst various stages of pandemic makes the study relevant and important. This is the first series describing abdominal imaging findings in COVID positive patients.

Specific comments on weaknesses of the article and what could be done to improve it:

1) The inclusion and exclusion criteria mentioned in the methods sections does not match with Table 1 of the results section. Although the authors mention that no patients were excluded, patients who underwent abdominal radiographs were indeed excluded and study was limited to those who had atleast one cross sectional imaging study. Thus, the methods section must be re-written accordingly.

2) Of the 72 patients included for final analysis, 66 were admitted for an initial diagnosis of COVID and rest (n=6) were not initially suspected to have COVID infection but were subsequently tested positive for COVID. The details of the initial clinical diagnosis of the later six patients and the reasons which led to COVID testing is not mentioned. But this would be important because COVID testing may be done as a part of preoperative work up. We now know that concurrent COVID infection in patients presenting with acute surgical emergency might negatively influence their post-operative prognosis and recovery. Including such patients might confound the study results, affect comparability and generalisability of the results.

3) The study design did not have a matched comparison group of non-COVID patients. Therefore, the findings seen in this study cannot be attributed to COVID-19 infection with certainty. This is one of the major limitations of the study.

4) The study methods do not incorporate a sample size calculation. The authors have used using logistic regression analysis to study the association between imaging findings and patient demographics. However, the study sample is too small and not adequately powered to reliably answer questions pertaining to such association. Also, the results pertaining to logistic regression analysis only had odds ratio (OR) and p-value. The B-value and 95% confidence interval of OR are important for determining the true significance of the findings but are not mentioned.

5) Table 1 compares the demographic details of study populations and their imaging studies. It is hard to follow the section of this table which contains the information on the number of patients with abdominal imaging. It would have been better to have the number of patients (%) and the number of imaging studies (%) in separate columns. Also, the rows such as all CT and all USG could have been omitted and simplified as CT with its subcategories such as non-contrast and contrast; Ultrasound with its subcategories.

6) Some patients included in the study have undergone more than one abdominal imaging studies either with same or different imaging modality. The reason for repeat imaging could have been for the same or different clinical indication, the details of which are not mentioned.

7) In table 2 the results show N (%) of CT or ultrasound studies that had an imaging abnormality. But what is relevant is the N (%) of patients who had the imaging finding and this is not mentioned.

8) Figure shown are examples cases illustrating imaging and pathology findings. Since these figures are not study results, they are best shown in the discussion section.

9) The results provided in the text of the results section are mostly a repeat of table 1 and 2. Table 1 could have been a flow chart. The association between imaging findings and patient demographics would be better summarised in a table.

10) In the results section, under paragraph 2 of the subheading called CT findings, the authors have equated fluid filled colon to diarrhoea, which is their interpretation pertaining to an imaging finding. This interpretation best be limited to the discussion section. There is little additional information on fluid filled colon like associated wall thickening, increased enhancement etc. This will help the readers understand the relevance of this finding.

11) The first paragraph of the discussion only has the descriptive results. Subsequent three (paragraph 2, 4 and 5) out of six paragraphs of the discussion section are extensions of results section. The authors instead could provide a concise summary of all the study findings in the first paragraph of the discussion.

12) The authors have provided patho-immunological explanation for the results in paragraph 3 of the discussion. However, they have not compared their results with the available literature on the rates of the abdominal imaging findings in ICU patients in a non-COVID setting. Adding this would have been of great value.

13) Authors may have over interpreted the relevance of fluid filled colon on CT by equating this finding to diarrhoea. But there are multiple other causes for fluid filled colon in ICU patients and these have not been discussed.

Overall, the study shows that COVID positive ICU patients have significantly higher number of abdominal cross-sectional studies and that they have significantly higher bowel wall thickening, pneumatosis intestinalis and fluid filled colon when compared to COVID positive non-ICU in-patients. In practice, we see the same findings in non-COVID patients. Since there is no comparison group of non-COVID patients, it is difficult to understand clinical significance of the study findings.