Canonical Correlation for the Analysis of Lifestyle Behaviors versus Cardiovascular Risk Factors and the Prediction of Cardiovascular Mortality: A Population Study
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
In this study, authors investigated the role of some lifestyle behaviours as predictor of cardiovascular risk factors in an historical Italian cohort. Authors considered smoking habits, physical activity, dietary scores, marital status and high socio-economic status. Interestingly they showed that, when tested as predictors of cardiovascular outcome lifestyle behaviors correlate well with cardiovascular risk factors associated to CHD mortality.
The paper is interesting and the analysis well conducted. The most important strong point of this study is the long follow up.
I have only minor comments:
Discussion.
This paragraph should include what new features does this study bring to our knowledge
There are some typos through the paper that should be corrected
Comments on the Quality of English Language
The paper needs minor language revision
Author Response
Canonical correlation analyzing lifestyle behaviors versus cardiovascular risk factors and predicting cardiovascular mortality. A population study. By Alessandro Menotti (1), Paolo Emilio Puddu (2)
REFEREE 1
Open Review
(x) I would not like to sign my review report
( ) I would like to sign my review report
Quality of English Language
( ) I am not qualified to assess the quality of English in this paper
( ) English very difficult to understand/incomprehensible
( ) Extensive editing of English language required
( ) Moderate editing of English language required
(x) Minor editing of English language required
( ) English language fine. No issues detected
Yes Can be improved Must be improved Not applicable
Does the introduction provide sufficient background and include all relevant references?
( ) (x) ( ) ( )
Are all the cited references relevant to the research?
(x) ( ) ( ) ( )
Is the research design appropriate?
( ) (x) ( ) ( )
Are the methods adequately described?
( ) (x) ( ) ( )
Are the results clearly presented?
(x) ( ) ( ) ( )
Are the conclusions supported by the results?
( ) ( ) ( ) ( )
Comments and Suggestions for Authors
In this study, authors investigated the role of some lifestyle behaviours as predictor of cardiovascular risk factors in an historical Italian cohort. Authors considered smoking habits, physical activity, dietary scores, marital status and high socio-economic status. Interestingly they showed that, when tested as predictors of cardiovascular outcome lifestyle behaviors correlate well with cardiovascular risk factors associated to CHD mortality.
The paper is interesting and the analysis well conducted. The most important strong point of this study is the long follow up.
I have only minor comments:
QUESTION 1.
Discussion. This paragraph should include what new features does this study bring to our knowledge.
REPLY 1
A long paragraph on this issue has been added in the mid of Discussion.
Question 2
There are some typos through the paper that should be corrected
REPLY
They were identified and corrected.
Comments on the Quality of English Language
The paper needs minor language revision
ADDITIONAL INFORMATION FROM AUTHORS
A number of small mistakes and typos have been corrected and other minor additions have been introduced.
Submission Date
08 November 2023
Date of this review
28 Nov 2023 22:11:28
REPLIES: 17 Dec 2023
Reviewer 2 Report
Comments and Suggestions for Authors
This manuscript deals with an interesting scientific and epidemiological question. The strengths is the very long follow-up with an almost complete set of time-to-event dates. The approach with canonical correlation is very interesting. However, I have some remarks that should be addressed.
1) As the authors say canonical correlation is not so common in that kind of research yet. Therefore I think more detailed explanation on the methods and also on the results should given to the reader. I also recommend to add e.g. the canonical loadings so that the reader can better understand, what are the most improtant variables. See https://stats.oarc.ucla.edu/stata/dae/canonical-correlation-analysis/
2) The variables in Table 1 should be ranked according to the "rank" column.
3) The authors are dealing as primary outcome with death due to CVD/CHD, Stroke or HDUE. However, they dont consider in their Cox models competing risks. Any death due to other reasons is a competing risk for the primary event types of interest preventing those events. This should be addressed by either presenting cause-specific Cox model results for ALL event types including the competing events. Or I would recommend to use Fine.Gray models with subdistrbution HR for that.
4) Regarding table 8 and evalution of a predictive addon benefit including canonical variates into the model I think other improvement metrics than a likelihood ratio test should be presented. A better metric would be e.g. Akaikes or Bayes information criterion incl. weights. Regarding time-to-event data Harrels c or IDI or cNRI should be used. Time-dependent AUCs would also be possible.
Comments on the Quality of English Language
All tables should be edited regarding spelling mistakes, or line breaks e.g. Minor reevaluation of the main text body would also help.
Author Response
Canonical correlation analyzing lifestyle behaviors versus cardiovascular risk factors and predicting cardiovascular mortality. A population study. By Alessandro Menotti (1), Paolo Emilio Puddu (2)
REFEREE 2
Open Review
(x) I would not like to sign my review report
( ) I would like to sign my review report
Quality of English Language
( ) I am not qualified to assess the quality of English in this paper
( ) English very difficult to understand/incomprehensible
( ) Extensive editing of English language required
(x) Moderate editing of English language required
( ) Minor editing of English language required
( ) English language fine. No issues detected
Yes Can be improved Must be improved Not applicable
Does the introduction provide sufficient background and include all relevant references?
(x) ( ) ( ) ( )
Are all the cited references relevant to the research?
(x) ( ) ( ) ( )
Is the research design appropriate?
( ) (x) ( ) ( )
Are the methods adequately described?
( ) (x) ( ) ( )
Are the results clearly presented?
( ) (x) ( ) ( )
Are the conclusions supported by the results?
(x) ( ) ( ) ( )
Comments and Suggestions for Authors
This manuscript deals with an interesting scientific and epidemiological question. The strengths is the very long follow-up with an almost complete set of time-to-event dates. The approach with canonical correlation is very interesting. However, I have some remarks that should be addressed.
QUESTION 1
As the authors say canonical correlation is not so common in that kind of research yet. Therefore I think more detailed explanation on the methods and also on the results should given to the reader. I also recommend to add e.g. the canonical loadings so that the reader can better understand, what are the most important variables. See https://stats.oarc.ucla.edu/stata/dae/canonical-correlation-analysis/
REPLY 1
More methodological details were added in the chapter of Material and Methods, sub-chapter Statistical analysis.
In Table 2 the canonical loadings were added and then described and commented in the chapter of Results.
QUESTION 2.
The variables in Table 1 should be ranked according to the "rank" column.
REPLY 2
The content of table has been re-ranked as requested.
QUESTION 3
The authors are dealing as primary outcome with death due to CVD/CHD, Stroke or HDUE. However, they dont consider in their Cox models competing risks. Any death due to other reasons is a competing risk for the primary event types of interest preventing those events. This should be addressed by either presenting cause-specific Cox model results for ALL event types including the competing events. Or I would recommend to use Fine-Gray models with sub-distribution HR for that.
REPLY 3
The authors are aware of the possible existence of competing risks.
However, the problem has been already tackled on the same material and published in Reference N 11.
Now, a sentence has been added in the Discussion.
QUESTION 4
Regarding table 8 and evaluation of a predictive addon benefit including canonical variates into the model I think other improvement metrics than a likelihood ratio test should be presented. A better metric would be e.g. Akaike or Bayes information criterion incl. weights. Regarding time-to-event data Harrels c or IDI or cNRI should be used. Time-dependent AUCs would also be possible.
REPLY 4
The authors feel that the test called Informativeness, suggested by Prof Peto, would be enough the show the improvement in the predictive power of the model with the additional data are forced in.
However, the Akaike Information Criterion has been added in the new Table 8 together with the AUCs.
Consequently, the description of this Table has been enlarged in the new version.
Comments on the Quality of English Language
All tables should be edited regarding spelling mistakes, or line breaks e.g. Minor reevaluation of the main text body would also help.
ADDITIONAL INFORMATION FROM AUTHORS
A number of small mistakes and typos have been corrected and other minor additions have been introduced.
Submission Date
08 November 2023
Date of this review
13 Dec 2023 12:18:20
REPLIES: 17 Dec 2023
Round 2
Reviewer 2 Report
Comments and Suggestions for Authors
Thanks for addressing all my comments. I have no further remarks.
Just a minor remark: in tables 5-7 I assume the last line should be "tertile 3 Y covariate score". This should be corrected.
Author Response
Thanks for helping us correcting the errors introduced in Tables 5-7 by miss-spelling X instead of Y for Tertiles 3.