Effect of SARS-CoV-2 Infection on Renal and Hepatic Function after NSAID and Paracetamol Therapy
Round 1
Reviewer 1 Report
This paper investigated if COVID-19 is a risk factor for the development of renal or hepatic function impairment after NSAIDs. Some points should be addressed. First of all, explain the novelty of the study in comparison to the literature available. Information about dose and duration are missing
Authors should analyze the dose and the duration of treatment in association with COVID.
Also authors should add a table reporting cardiovascular risk factors that may impact the risk of kidney disease, such as hypertension, diabetes, etc
Some recent references are missing regarding NSAIDs use and safety in COXID patients (https://doi.org/10.1016/
S2665-9913(21)00104-1; https://doi.org/10.1038/s41533-022-00300-z)
This paper investigated if COVID-19 is a risk factor for the development of renal or hepatic function impairment after NSAIDs. Some points should be addressed. First of all, explain the novelty of the study in comparison to the literature available. Information about dose and duration are missing
Authors should analyze the dose and the duration of treatment in association with COVID.
Also authors should add a table reporting cardiovascular risk factors that may impact the risk of kidney disease, such as hypertension, diabetes, etc
Some recent references are missing regarding NSAIDs use and safety in COXID patients (https://doi.org/10.1016/
S2665-9913(21)00104-1; https://doi.org/10.1038/s41533-022-00300-z)
Author Response
We thank you for your review and helpful comments.
Comment 1:
This paper investigated if COVID-19 is a risk factor for the development of renal or hepatic function impairment after NSAIDs. Some points should be addressed. First of all, explain the novelty of the study in comparison to the literature available. Information about dose and duration are missing.
Response 1: We realized that we did not clearly communicate the aim of our study, i.e. the influence of a SARS-CoV-2 infection on the occurrence of side effects of specific drugs. We reorganized and reformulated the introduction to make our point clearer.
Comment 2: Authors should analyze the dose and the duration of treatment in association with COVID
Response 2: Furthermore, according to your suggestion, we added doses as predictors in our models. Overall, the general assertion of our study remains the same, however we did note a slight improvement of the predictive power of the trained ML models.
Unfortunately, we are not able to add treatment duration as in some cases, we could not determine the start of treatment. In some cases, NSAIDs and paracetamol were started prior to admission, as these are registered as OTC in Switzerland. This is a problem with the cohort dataset that cannot be addressed after the fact as this is a retrospective study.
Comment 3: Also authors should add a table reporting cardiovascular risk factors that may impact the risk of kidney disease, such as hypertension, diabetes, etc
Response 3: We also added comorbidities to the predictors. Unfortunately, these data are not complete, as the share of patients is very low. Again, this is a problem with the cohort dataset that cannot be addressed after the fact as this is a retrospective study.
Comment 4: Some recent references are missing regarding NSAIDs use and safety in COXID patients (https://doi.org/10.1016/S2665-9913(21)00104-1; https://doi.org/10.1038/s41533-022-00300-z)
Response 4: We added the missing references.
Reviewer 2 Report
In general, the text is well-written and direct. And the conclusion makes sense given the results found. However, some points need to be corrected:
- A point that raises doubts when interpreting the data in Table 1 is the absence of p-values ​​for the analyzed parameters. Even though the correction values ​​are expressed, the degree of significance expressed by p can help readers reach the same conclusions as the authors.
- Furthermore, supplementary tables 1 and 2 need to be placed in the body of the text, since they contain the main results of the work.
- Another important point in the current Coronavirus Disease (COVID-19) situation, it would be worthwhile for the authors to associate their results with the literature on long-term Covid or post-Covid syndrome.
- Finally, the data regarding renal and hepatic parameters need to be rewritten more clearly, as it is currently difficult to interpret whether the data refers to treatments or COVID-19.
- I suggest that the authors reformulate the work's title, in its current form as it leads to a different interpretation of the data presented.
- In all citations in the table, the following phrase appears: "Error! Reference source not found". This error needs to be corrected.
Author Response
Thank you for your thorough review and comments.
Comments 1: A point that raises doubts when interpreting the data in Table 1 is the absence of p-values ​​for the analyzed parameters. Even though the correction values ​​are expressed, the degree of significance expressed by p can help readers reach the same conclusions as the authors.
Responses 1: We did not add p-values to Table 4 (formerly Table 1) for the following reason: Using p-values to compare machine learning (ML) metrics such as accuracy, specificity, and other performance measures is generally not appropriate because p-values are designed to test hypotheses about population parameters, not to compare predictive performance metrics. In ML, practical significance (i.e., the actual performance difference between models) is often more relevant than statistical significance. A model with a statistically significant but practically small improvement in accuracy might not be meaningful in a real-world context. Instead of p-values, other approaches are typically used to compare ML models. The confidence intervals for the metrics (e.g. specificity, accuracy) assess the range of uncertainty around the estimates. Overlap** confidence intervals can indicate that the differences between models are not statistically significant. Additionally, bootstrap** to estimate the distribution of the metric differences can be used. This non-parametric approach does not rely on normality assumptions and can provide more robust estimates of uncertainty. However, we did not use any of these methods (or other approaches) as our model did not show good predictive power. This lack of predictive power led us to conclude that the chosen predictors, including the SARS-CoV-2 infection status, are not sufficient to predict the outcome. Therefore, SARS-CoV-2 infections status appears not to be an independent risk factor for the development of NSAIDs or paracetamol induced renal or hepatic impairment.
Comments 2: Furthermore, supplementary tables 1 and 2 need to be placed in the body of the text, since they contain the main results of the work.
Responses 2: As per your suggestion, we included the table from the supplements within the main body of the text.
Comments 3: Another important point in the current Coronavirus Disease (COVID-19) situation, it would be worthwhile for the authors to associate their results with the literature on long-term Covid or post-Covid syndrome.
Responses 3: We agree with you on the importance of investigation into long-term Covid and post-Covid syndrome. However, the relatively short temporal scope of our data set does not allow us to make any statement regarding these conditions. Our database does not contain any information on long-term patient follow-up.
Comments 4: Finally, the data regarding renal and hepatic parameters need to be rewritten more clearly, as it is currently difficult to interpret whether the data refers to treatments or COVID-19.
Responses 4: We did reorganize the introduction. In the results, we moved the COVID-test status from the outcome definition to the description of the features. Furthermore, we added the ranges for the variables responsible for the outcome definition (eGFR and ALAT, AP) to the study population tables. We hope that make it clearer that the outcome is the decline in organ function and that the COVID-19 testing status is a feature considered during ML training.
Comments 5: I suggest that the authors reformulate the work's title, in its current form as it leads to a different interpretation of the data presented.
Responses 5: We reformulated the title of our study, as well as restructured the introduction to make the scope and aim of our study clearer. We hope that you agree with our revision.
Comments 6: In all citations in the table, the following phrase appears: "Error! Reference source not found". This error needs to be corrected.
Responses 6: We apologize for our mistake with the missing hyperlinks. We tried to remove all hyperlinks from the original document per style guide from the journal, but we seem to have missed some in the body of text. This mistake has been corrected.
Round 2
Reviewer 1 Report
Authors have addressed all my queries in the novel version of the paper
Authors have addressed all my queries in the novel version of the paper
Reviewer 2 Report
The authors satisfactorily addressed the reviewers concerns.
The authors satisfactorily addressed the reviewers concerns.
