SARS-CoV-2 Infection Increases the Risk of Muscle Injury in Professional Male Soccer Players—A Retrospective Analysis of the Italian and Spanish Major Leagues

Round 1

Reviewer 1 Report

The manuscript tried to retrospectively verify, on a larger sample, the correlation between SARS-CoV-2 infection and muscle injuries among professional footballers of two different championship Serie A and LaLiga. The method and analysis were scientific and the results were valuable and important. Some minor errors were found as follows.

 p.3 line 115-120 : two sentences are same : "A multiple linear regression analysis was carried out ...., and matches played. A multiple linear regression analysis was carried out...., and matches played.“

 p.4 line 127 : “The study included 634 players from Serie A and 649 player...” => But in Fig 1., the number of Serie A n is written as 643.

 p.8 line 240 : 4.2. Clinical implications : the title is same with “4.1 Clinical implications”. It is suggested to use “4.2 Study limitations”.

Author Response

Reviewer 1

The manuscript tried to retrospectively verify, on a larger sample, the correlation between SARS-CoV-2 infection and muscle injuries among professional footballers of two different championship Serie A and LaLiga. The method and analysis were scientific and the results were valuable and important. Some minor errors were found as follows.

Dear reviewer, we thank you for the comments, we are glade you appreciated the paper.

 p.3 line 115-120 : two sentences are same : "A multiple linear regression analysis was carried out ...., and matches played. A multiple linear regression analysis was carried out...., and matches played.“

Thanks for the comment, we corrected.

 p.4 line 127 : “The study included 634 players from Serie A and 649 player...” => But in Fig 1., the number of Serie A n is written as 643.

Thanks for the comment, we corrected.

 p.8 line 240 : 4.2. Clinical implications : the title is same with “4.1 Clinical implications”. It is suggested to use “4.2 Study limitations”.

Thanks for the comment, we corrected.

Reviewer 2 Report

Reviewers' comments:

Reviewer #1: In the submitted manuscript, the authors conducted a retrospective cohort study to evaluate the effect of COVID positive on muscle injuries. Football players in premier Italian and Spanish national football leagues were analyzed based on their COVID infection case history. The manuscript revealed that after the pandemic and league suspension, in 20/21 season, the number of muscle injury cases increased for players who were infected by COVID, compared with those who were not infected.

The manuscript was easy to read and well-organized based on valuable analyses. However, there are still limitations regarding statistical methods and the corresponding discussion that should be addressed prior to publication.

1.     The current title is not informative and does not briefly include the main findings of this study. The reviewer strongly suggests revising the title; including increased risk of muscle injuries.

2.     For the factorial design, the authors might consider using 2*2 analysis of variance because two factors of the season (19/20 and 20/21) and COVID infection were used in this study. Otherwise, the authors can consider testing the increment of average muscular injuries between seasons (change between seasons) using parametric t-tests.

3.     For the results of multiple linear regression, please specify R-squared, the coefficient of determination. Although some of the independent variables had significant effects, their proportion of total variance should be indicated for a better understanding of data and goodness of model fit.

3-1. The authors should discuss more about why the age effect was significant in the Italian championship in 20/21 season, while this was not significant in the Spanish championship in the same season.

3-2. Similarly, more explanation is needed for why the independent variable of minutes played, which seems associated with the number of muscular injuries, did not have a significant effect.

4.     Although the authors discussed the possible explanations for the correlation by quarantine and biological effect, the influence of COVID is very confounded and the authors should consider including more detailed factors (e.g., total period of quarantine, injured body part, etc.) to complement the current results.

5.      (Minor comment) For figure 3, please use uniformed scale for y-axis (i.e., survival probability).

English language level of the authors is generally good; but some sentences should be checked again (e.g., l. 97).

Author Response

Reviewer 2

Reviewer #1: In the submitted manuscript, the authors conducted a retrospective cohort study to evaluate the effect of COVID positive on muscle injuries. Football players in premier Italian and Spanish national football leagues were analyzed based on their COVID infection case history. The manuscript revealed that after the pandemic and league suspension, in 20/21 season, the number of muscle injury cases increased for players who were infected by COVID, compared with those who were not infected.

The manuscript was easy to read and well-organized based on valuable analyses. However, there are still limitations regarding statistical methods and the corresponding discussion that should be addressed prior to publication.

1. The current title is not informative and does not briefly include the main findings of this study. The reviewer strongly suggests revising the title; including increased risk of muscle injuries.

Thank for the comment, we modified the title as requested. The new title is:

“SARS-COV-2 infection increases risk of muscle injury in professional male soccer players. A retrospective analysis of the Italian and Spanish major leagues”

2. For the factorial design, the authors might consider using 2*2 analysis of variance because two factors of the season (19/20 and 20/21) and COVID infection were used in this study. Otherwise, the authors can consider testing the increment of average muscular injuries between seasons (change between seasons) using parametric t-tests.

Thanks for the comment. We implemented the results with parametric t-test analysis. However, as implemented in the discussion section, with such a large sample it is highly probable that even minimal variations between seasons are significant (especially in the C- group which is substantially more numerous than the C+ group), while the 95% confidence intervals shows that the average number of injuries between the two seasons is higher in group C+ only.

3. For the results of multiple linear regression, please specify R-squared, the coefficient of determination. Although some of the independent variables had significant effects, their proportion of total variance should be indicated for a better understanding of data and goodness of model fit.

Thanks for the comment. We implemented as requested in the tables.

3-1. The authors should discuss more about why the age effect was significant in the Italian championship in 20/21 season, while this was not significant in the Spanish championship in the same season.

Thanks for the comments. We implemented the results and the discussion section.

3-2. Similarly, more explanation is needed for why the independent variable of minutes played, which seems associated with the number of muscular injuries, did not have a significant effect.

Thanks for the comments. We implemented the results and the discussion section.

4. Although the authors discussed the possible explanations for the correlation by quarantine and biological effect, the influence of COVID is very confounded and the authors should consider including more detailed factors (e.g., total period of quarantine, injured body part, etc.) to complement the current results.

Thanks for the comment, we have implemented the limitation section of the study, as unfortunately from the database used it is not possible to trace further details such as total period of quarantine or injured body part.

5. (Minor comment) For figure 3, please use uniformed scale for y-axis (i.e., survival probability).

Thanks for the comment, we corrected as requested.

Round 2

Reviewer 2 Report

The reviewer is satisfied with the improved manuscript; however, it is still doubtful that the R-squared for each model is too small. The authors should elaborate on the reason for the low coefficient of determination, especially for the season 19-20.

Furthermore, presenting p-values in the results section should be improved (unified) before publication (ex. p = 0.xxx for non-significant results, p<0.05 for significant results).

The manuscript was easy to read.

Author Response

Dear Reviewer, thanks again for the comments, for the p-values we unified as requested.  

For the comment regarding the R-squared please consider that to better understand if the coefficient of determination is statistically significant we need to look at the p-value of the F statistic. In fact, the regression model we have built only makes sense if there is at least one independent variable that can explain the values of y variable. Another problem with the R squared could be that it increases each time an independent variable is added to the model, even if this variable is not explanatory at all. Indeed, it is not possible to explain less than the observed change for the dependent variable by adding explanatory variables to the model. To avoid this situation, in regression models with many independent variables, it is preferred to interpret the adjusted and predicted R-squared value. Also, if you construct a curve that fits "too much" to the data (e.g. using polynomial terms) you will probably get a model with a very high coefficient of determination. However, a model that adapts too much to a specific data set, following every slightest variation, turns out to be poorly generalized and with low predictive power. In statistics in these cases we speak of over-fitting problems. But as the reviewer can see it is not our case. Finally, based on these results, we are collecting data from championships around the world.