Government Restriction Efficiency on Curbing COVID-19 Pandemic Transmission in Western Europe
Round 1
Reviewer 1 Report (Previous Reviewer 3)
The manuscript is a revised version of an original work sent several months ago.
However, authors improved the original work by addressing some suggestions. In my opinion, the current version could be accepted.
Minor editing of English language required. Checking for typos.
Author Response
Thanks a lot for your positive feedback. We checked and correct typos in the manuscript
Reviewer 2 Report (Previous Reviewer 2)
The authors presented a significant improvement to the manuscript compare to the previous version. There are still some minor issues.
1. Could you add the Normalized time series of SI for each country, a similar plot Figure 1?
2. Please add some literature which criticised the lock downs and restrictions, then explain how the work presented here can support the implemented measures.
3. It is interesting to see that there is a relationship between "skewness" and SI. However, the skewness may not necessarily be an indicator for "pandemic transmission", as the main argument was about patients admitted to ICU.
Finally, remember that Figures 1 and 6 are not "density" or "distribution" curves. So my main question is that, would distributional properties such kurtosis, skewness, mean, sd, be valid? It is nice that you are using these properties, you just need to justify them to avoid misleading conclusions.
Author Response
Thanks for your positive feedbacks
1) We added the weekly normalized SI. It shows which nations implemented the restrictions earlier.
2) We added a paragraph about that
3) We agree that skewness alone may not directly indicate pandemic transmission as the main focus of our analysis was on patients admitted to the Intensive Care Unit (ICU). While our study primarily aim is to assess the effectiveness of government-imposed restrictions based on ICU admissions, we explored if and how some statistical parameters of the ICU admissions time series are affected by the implemented restrictions. We found a correlation between Skewness and SI index, that implies that the restrictions make the time series asymmetric. To corroborate this finding, we calculated the skewness for normal flu outbreaks. The highlighted correlation between skewness and the SI underscores a significant association between the implemented restrictions and the shape of the ICU admissions distribution, which is indeed a positive result.
4) That is correct. However, while the curves reported in those figures are NOT true distributions, they are normalized bell-like curves for which we seek to determine geometrical parameters such as skewness and kurtosis. Hence, instead of computing the moments directly through an integral, we use a resampling algorithm to draw random deviates within their support (the integer interval [1,M] in weeks) distributed as if the temporal trend was indeed a PDF. This step is accomplished through the corresponding cumulatives (computed simply as the cumulative sums as explained in the text.
In summary, in order to characterize the geometrical features of the temporal ICU trends (especially their asymmetry around the maximum), we treat them as (artificial) PDF and compute their moments as described above. This is now specified in the text.
Reviewer 3 Report (Previous Reviewer 1)
I am unable to find novelty or significant contribution of this manuscript. The analysis technique used is very basic and the results are too limited. As per my view, it is still below par to be published.
Need refining.
Author Response
We recognize the need to highlight more explicitly the novelty and significant contributions of our study. Specifically, our work introduces ICU admissions as a unique measure of COVID-19 severity, a metric that has not been used before in this capacity. This novel approach provides a more direct indicator of healthcare system strain and a precise understanding of the disease's impact, which is a clear departure from conventional methods. This aspect has been highlighted in the manuscript.
Regarding the analysis technique, we chose to use an arguably basic technique to ensure transparency, accessibility, and reproducibility of our findings. Our approach, predicated on simplicity, is consistent with that of co-author Piazza et al., whose renowned publication got over a thousand citations.
While we agree that more complex methodologies could potentially reveal additional insights, we believe our approach provides a valuable perspective, especially in a rapidly evolving public health crisis where clear and interpretable results are paramount. Regarding novelty, our aim in this research is not just to explore a new approach but to refine and potentially improve upon existing methodologies to assess the efficacy of non-pharmaceutical interventions in the context of COVID-19. Indeed, our application and execution may offer insights not previously considered.
As for the analysis technique, we understand that it may appear basic, but it is free of conceptual errors and it was intentionally chosen for its interpretability and reproducibility, given the global scale of the pandemic. We believe that complex methods may hinder understandability and application in real-world scenarios, especially in regions with limited resources. English has been checked by a native speaker.
Round 2
Reviewer 3 Report (Previous Reviewer 1)
Not applicable
Not applicable
Author Response
We have incorporated sections that emphasize the novelty of our publication and offer some future perspectives. We trust that these enhancements will address your concerns.
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
Thank you for giving me the chance to review this manuscript. The manuscript has many flaws and as per my view it doesn't make any significant contribution. Authors need to redesign their study and apply some significant results rather than mean, SDs etc. My detailed comments are as follows:
1- intro needs refinement. Current intro does not comprehensively cover the context of the study. The authors abruptly described the purpose of their study without explaining and mentioning the latest literature regarding covid 19.
2- it is suggested to create sub headings in intro part to clearly mention the already published lit and add a table summary for this purpose.
3- the methodology has been explained but it is suggested to add methodology flowchart.
4- why this methodology is chosen, which other options are available in lit regarding other similar tools.
5- how the indicators are chosen, what was sample size and population size.
6- it is suggested to mention biasedness in sampling. There are many tests, apply one of them and show the results.
7- discussion part is poorly written. Improve it. Compare your results with other published articles.
8- many equations have been shown in paper but none has been explained in results. Means the equations should be solved for atleast one indicator to show how the authors used these equations and attained the mentioned results.
9- conclusion is very weak. Add theoretical and practical implications as well.
Author Response
Thank you for your thorough evaluation of our manuscript and for the suggestions to improve it. We appreciate your effort and time invested in this task. Here are our responses to each point:
1- We have revised the introduction, adding a more comprehensive coverage of the context of the study, with a particular focus on the current literature regarding COVID-19. We have made efforts to ensure the purpose of our study is outlined more fluidly, with proper and updated references.
2- Following your suggestion, we have included subheadings in the introduction to clearly indicate the topics being discussed. We have also added a table summarizing the already published literature, which we believe improves the overall clarity of the introduction.
3- We added a flowchart that helps to visualize the different steps of analysis.
4- We have expanded our discussion about the choice of methodology, comparing it to other available tools described in the literature and giving reasons for our chosen approach.
5- In the methodology section, we have now clearly explained the selection of indicators, as well as provided a detailed description of our sample size and population size.
6- As suggested, we have addressed potential bias in sampling. We have applied the bootstrap test to demonstrate the validity of our sampling process and presented the results.
7- The discussion section has been significantly revised to provide a more nuanced and detailed analysis. We now compare our results with those from other relevant published studies, which we believe strengthens our arguments and conclusions.
8- We have addressed your concern about the equations. We have provided a detailed explanation of each equation in the results section and shown how they were applied to our indicators, using specific examples to demonstrate their use in achieving our results.
9- The conclusion section has been rewritten to clearly articulate the theoretical and practical implications of our study. We believe this provides a compelling summary of our findings and their implications.
We believe that the revisions made in response to your comments have greatly improved the quality and relevance of our manuscript. We hope that you will find it to be significantly improved and a valuable contribution to the field.
Reviewer 2 Report
The idea embedded in this paper was interesting but there were major issues related to the analysis and design study. Please find the main issues that I have identified below:
The design of the study
Knowing that the data may not follow a normal distribution, it is wrong to carry on the analysis. Therefore, appropriate measures had to be considered to assure the reader the conclusion is valid despite the violation of normality assumption.
Comparison between countries
1. Each district had to be separated not as the overall data point. Then at each time point the distribution properties had to be obtained. Do not look at the overall ICU cases per week.
2. Age range, Gender and other demographic variables had to be reported for each country.
Comparison between years
1. Similarly, each district should be shown separately. Then every week has a distribution, therefore has skewness and kurtosis. Looking at the overall distribution can be misleading.
2. Skewness cannot be the only measurement to show the effectiveness of the intervention. When comparing years from 2014-2019, we can see that 201718 has a fatter and longer tail from any other years.
3. Visually inspecting, 1718 has a very similar shape to the COVD year. Did you use any statistical test for the comparison between these years? In addition, it is not clear if similar interventions were used in 1718 same amount of change would have happened on the skewness.
4. The left tail of the distribution during COVID19 year, is shorter than previous years. Indicating a 10 week lag, in comparison to previous years.
Appropriate Statistical Test
Except a simple linear regression no other statistical test was performed.
Author Response
Thank you for your precious comments. We appreciate your input. Below the comments to each point.
While it is true that many statistical methods assume a normal distribution for the data, it is important to note that not all analyses are strictly dependent on this assumption. Skewness, as a measure of the asymmetry of a distribution, can provide valuable insights even when dealing with non-normal data. Without restrictions, there is not a difference among the distribution tails. Introducing the restrictions that are summarized by the Stringency index, the symmetry of the distribution is broken and skewness is able to capture this asymmetry. Comparison between countries: For some countries, data are available on daily basis, for some others are only weekly and for this reason we weekly aggregate the data for comparison. Moreover, age, gender and other demographic variables are not easily available for all the countries. We will take into consideration for future development. However, considering a large population number , especially for European countries, those parameters don't play a relevant role. Comparison between years For the first point, please see the answer above (comparison between countries) We agree with the reviewer that the skewness is not the unique available metric and indeed we showed even other metrics as kurtosis and standard deviation to assess the tail flatness. We didn't use statistical tests. We can consider this opportunity in future studies. About the second point, during prepandemic outbreaks as 1718 flu, no restrictions were applied. Thanks for the comment. We added the 10 week lag note into discussion/result Regarding the use of linear regression as the primary statistical test in our study, we opted for this approach for the following reasons:
-
Relationship Assessment: Linear regression allowed us to assess the relationship between variables and examine the impact of interventions over time.
-
Continuous Variables: Linear regression is suitable for analyzing continuous variables, which was the case for our dependent variable (ICU cases) and independent variables (such as time and interventions).
-
Controlling for Confounding Factors: Linear regression provided a framework to control for confounding factors and obtain more accurate estimates of the relationship of interest.
-
Interpretability: Linear regression offers interpretable coefficients, allowing for straightforward interpretation of the results.
Reviewer 3 Report
This paper assesses the effectiveness of the restrictions implemented by the government during COVID-19 outbreak, in order to limit the peak of Intensive Cure Units (ICU) and, generally, positive cases.
Authors analyzed the data related to COVID-19 of several countries (i.e., Belgium Finland, France, Germany, Italy, Sweden, Netherlands) by using a scale between 0 (no restrictions) and 100 (full lockdown). Furthermore, eight closure policies and containment indicators are evaluated.
The introduction could be extended by reporting more information on the models that studied COVID-19 trend, as well as it should mention that several countries adopted interventions to combat the spread (e.g., limitations, vaccination). Therefore, I suggest reporting some line of introduction related to COVID-19 trend, for instance authors could cite a study (e.g., PMID:35885152) related to epidemiology of Covid-19 in several countries (USA, Italy, France, Sweden, UK), the reference is reported as follows: https://pubmed.ncbi.nlm.nih.gov/35885152
In my opinion, the equations 2 and 3 could be removed because these referred to well-known arguments in statistics. The formulas could be replaced by citations for the topis.
However, Methodology does not explain how to the analysis has been performed. Did you develop a tool? Did you develop a new method? Did you use an existing tool (e.g., spss, R)? Did you implement or invoke by a tool the mentioned equations? In this latter case all equation should be removed because not directly applied, but these are used though a tool that implement these for you.
Authors performed a set of statistical analysis, however, these ones need to
“Discussion” is poor, and it does not report exhaustive information related to Results.
In my opinion Conclusion could be an isolate section to better structure the manuscript, instead it is joined with Discussion.
I think that this manuscript is very interesting; however, the mentioned issues should be addressed.
In my opinion the manuscript is not exhaustive and it needs a major revision, or a reject.
Minors: The manuscript contains typos and grammar mistakes.
Author Response
We greatly appreciate your time and efforts in evaluating our manuscript. Your valuable comments and suggestions will undoubtedly improve the quality and readability of our manuscript. We aim to address all your concerns in this response, with the understanding that a revised manuscript will be submitted for further review.
Regarding your suggestion to expand the introduction, we agree and have supplemented this section with a broader discussion on models that studied the COVID-19 trend. Additionally, we have included mention of various interventions that different countries have adopted to combat the spread of the virus. We appreciate your suggested study and have cited it to provide context on the epidemiology of COVID-19 across several countries.
We understand your concerns about Equations 2 and 3 being well-known in the field of statistics. We included them for the sake of completeness and clarity for a diverse readership. Those are also standard MATLAB functions. We added also a flowchart in the text to make the algorithm steps more clear. We developed a code in MATLAB that will be available if the manuscript will be accepted to ensure reproducibility.
We acknowledge your feedback about the need for a more comprehensive discussion section. We have expanded the discussion in our revised manuscript to provide more context and interpretation of our results. We have also separated the conclusions into its own section to improve the structure of the manuscript, as you suggested.
Finally, we appreciate your attentiveness to the typographical and grammatical errors in our original submission. We have thoroughly proofread the revised manuscript to correct these mistakes and improve the overall clarity of the paper.
