Investigating Risk Factors for Racial Disparity in E-Cigarette Use with PATH Study
, Michael Celestin, Jr. 2 and Qingzhao Yu 3,*Round 1
Reviewer 1 Report (Previous Reviewer 1)
Comments and Suggestions for AuthorsI have no further comment.
Author Response
Thank you so much for the review.
Reviewer 2 Report (New Reviewer)
Comments and Suggestions for AuthorsThis paper examines racial disparities in e-cigarette use and explores the associated risk factors using data from the Adult Population Assessment of Tobacco and Health (PATH) study. The authors employed chi-square tests, multiple logistic regression, and mediation analysis to identify these risk factors and assess their impact on the association between race and e-cigarette use. Overall, the paper is well organized. My comments are listed below:
1. I recommend that the authors review the font consistency throughout the manuscript. For example, in the abstract, the text font before the “Results” section differs from the following.
2. To enhance readability, I suggest the authors use commas to separate large numbers in Tables 2, 3, and 4. For example, the number “8028170” may be presented as ‘8,028,170’.
3. Could the authors please verify section 3.1 - descriptive analysis results? It appears that the results discussed in lines 172 – 192 pertain to Table 3, not Table 2 as currently stated.
4. To maintain consistency and aid reader comprehension, I recommend expressing values in Table 4 as percentages (%). For example, the value “0.068” can be expressed as “6.8%.
5. To improve the tables' readability, I suggest using horizontal lines or color shading to better differentiate between risk factor categories. For example, in Table 3, adding a horizontal line below the income category "$100,000 or more" could clearly indicate that the following categories pertain to education levels.
6. Could the authors clarify the reason for bolding "Non-Hispanic White" in Table 4?
7. Please fully spell the abbreviations when they first appear, e.g., C.I., aOR.
8. To enhance the readability, I suggest the authors place the reference category as the first category for each risk factor in Table 5.
9. Figure 2 is too blurry. I kindly request that the authors replace it with a high-resolution image.
Author Response
This paper examines racial disparities in e-cigarette use and explores the associated risk factors using data from the Adult Population Assessment of Tobacco and Health (PATH) study. The authors employed chi-square tests, multiple logistic regression, and mediation analysis to identify these risk factors and assess their impact on the association between race and e-cigarette use. Overall, the paper is well organized. My comments are listed below:
- I recommend that the authors review the font consistency throughout the manuscript. For example, in the abstract, the text font before the “Results” section differs from the following.
- Authors: We thank the reviewer for pointing this out. We have reviewed all the fonts to make them consistent.
- To enhance readability, I suggest the authors use commas to separate large numbers in Tables 2, 3, and 4. For example, the number “8028170” may be presented as ‘8,028,170’.
- Authors: We agree with the reviewer. We have reviewed the tables and added the commas as recommended.
- Could the authors please verify section 3.1 - descriptive analysis results? It appears that the results discussed in lines 172 – 192 pertain to Table 3, not Table 2 as currently stated.
- Authors: We agree with the reviewer. Thanks for pointing this out. We have checked the labels of table and made necessary revisions.
- To maintain consistency and aid reader comprehension, I recommend expressing values in Table 4 as percentages (%). For example, the value “0.068” can be expressed as “6.8%.
- Authors: We agree with the reviewer. Thanks for pointing this out. We have made the revisions according to the recommendation.
- To improve the tables' readability, I suggest using horizontal lines or color shading to better differentiate between risk factor categories. For example, in Table 3, adding a horizontal line below the income category "$100,000 or more" could clearly indicate that the following categories pertain to education levels
- Authors: We have made the revisions as recommended.
- Could the authors clarify the reason for bolding "Non-Hispanic White" in Table 4?
- Authors: Thanks for pointing that out. We have corrected the typo.
- Please fully spell the abbreviations when they first appear, e.g., C.I., aOR.
- Authors: Thanks for pointing this out. We have spelled the abbreviations at their first appear.
- To enhance the readability, I suggest the authors place the reference category as the first category for each risk factor in Table 5.
- Authors: We appreciate the reviewer for pointing this out and have made the changes.
- Figure 2 is too blurry. I kindly request that the authors replace it with a high-resolution image.
- Authors: Figure 2 is now Figure 3. We have moved the subscript out of the figure to make it clearer.
Reviewer 3 Report (New Reviewer)
Comments and Suggestions for AuthorsThis paper provides a comprehensive analysis of racial disparities in e-cigarette use, shedding light on key risk factors contributing to these differences. The findings underscore that non-Hispanic White populations exhibit higher e-cigarette use compared to non-Hispanic Black and Hispanic populations. Importantly, factors such as previous smoking habits, exposure to e-cigarette advertising, and perception of e-cigarette harm play significant roles in explaining these racial disparities. This paper suggests that efforts to increase awareness of the dangers associated with e-cigarette use and to reduce exposure to e-cigarette advertisements could help mitigate these differences.
The topic of this paper is interesting, however, the manuscript should be revised before it can be accepted for publication.
--The abstract is too long, the authors should reduce it
--In the abstract, when you mention “Mediation analysis found risk factors that significantly (p-value < 0.05) help explain differences in the e-cig use between non-Hispanic Black and non-Hispanic White populations…”; the authors should mention how to obtain the p-value, which test has been performed?
--The authors should give more details on the established multiple logistic regression model, including how to estimate parameters in the model.
--A Conclusion part should be given.
Author Response
The topic of this paper is interesting, however, the manuscript should be revised before it can be accepted for publication.
- The abstract is too long, the authors should reduce it
- Authors: We agree with the reviewer. We have shortened the abstract.
- In the abstract, when you mention “Mediation analysis found risk factors that significantly (p-value < 0.05) help explain differences in the e-cig use between non-Hispanic Black and non-Hispanic White populations…”; the authors should mention how to obtain the p-value, which test has been performed?
- Authors: Thank you for pointing that out. We used the R package mma. In the abstract, we added the method name. In the method section, we added a description of the method.
- The authors should give more details on the established multiple logistic regression model, including how to estimate parameters in the model.
- Authors: Thank you for pointing that out. We used the SAS to fit the model. In the method section, we added a description of the model and the software we used.
- A Conclusion part should be given.
- Authors: We agree with the reviewer. A conclusion section was added.
Round 2
Reviewer 3 Report (New Reviewer)
Comments and Suggestions for AuthorsThe manuscript has been revised accordingly, it can be accepted for publication.
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
Comments and Suggestions for AuthorsPlease find the attached comment file.
Comments for author File: 
Comments.pdf
Author Response
We appreciate the reviewer's constructive comments and have revised the manuscript accordingly. Please see the following point-by-point responses to the comments.
- The authors explained that tobacco users were oversampled in the PATH They utilized ”survey” package for descriptive statistics and logistic regressions. How about the mediation analysis ? Did the authors take it account the oversampling in the mediation analysis ? If not, isn’t it too dangerous to interpret the current mediation result ?
Authors: This is an excellent question! We used the sample weights provided by the PATH survey to adjust for the over-sampling of tobacco users using the “survey” package. We used the same weights in the mediation analysis using the “mma” package so that the samples were weighted to represent the whole population. This is described in the statistical analysis section (highlighted).
- The numbers should be re-arranged in Table ”e” and ”E” in the p-values are mixed in several Tables.
Authors: Thanks for pointing this out. We have checked all the numbers and present the numbers more consistently.
- In this paper, Section 3 has many problems. First, the authors explained the exposure and the outcome only. Please clarify the mediators. Second. how about the relationship among the mediators ? It seems that the authors did not consider the situation where one mediator can affect another mediator, but not clear. What is the assumption for the mediators in the current mediation analysis ? Third, what is the purpose of Table 6 ? Is it from the mediation analysis ? What is the relationship be- tween Table 6 and 7 ? A clear explanation is necessary for aOR in Table 6. Fourth, in Table 7, direct effects should be provided. Fifth, the authors interpreted the indirect effect as a proportion, which is wrong. Sixth, in Table 7, ”Standard Deviation” should be changed to ”standard error”. Seventh, the 95% CIs are not obtained from ”Standard Deviation”. Some calculations are wrong. Eighth, how did the authors calculate the 95% CI for the difference of the indirect effects ? Ninth, when mediators affect each other, the current interpretation cannot be accepted.
Authors: We appreciate all the comments. We carefully rewrote a large part of the Section to include but not limited to: 1. The definition of mediators. 2. Assumptions for the mediation analysis were added. 3. Explained that Table 6 showed the existence of racial/ethnic differences in e-cigarette use using logistic regression, and Table 7 showed the results of the mediation analysis. 4. Direct effects were added to Table 7. 5. Table 7 gives the estimated relative effects. We rectified the concept. 6. Changed the “standard deviation” to “standard error”. 7. We used the quantile CI from the bootstrap samples, but did not use the normal assumption to build the confidence interval. 8. As described in 7, we added the description of the CI calculation. 9. Correct. We added the request that “mediators may correlate with each other, but we assume there are no causal associations among the mediators.”
Reviewer 2 Report
Comments and Suggestions for AuthorsOverall, the paper has been well written. I would suggest the authors deliver more integrated findings in the discussion section and help the reader understand the results from the mediation analysis and how this stood out from the previous analysis in the literature. Currently, it is not clear to the general reader how to interpret the results listed in the third paragraph of the discussion section; also, such detailed results should be in the result section.
Author Response
Authors: We appreciate the reviewer's comments. In the Results and Discussion Section. We have provided more details about interpreting the results and added literature.
Reviewer 3 Report
Comments and Suggestions for Authors1. This manuscript uses the 2018–2019 Wave 5 data of PATH study in the United States to discuss the composition of the risk for ethnic differences in E-cigarette use by mediation analysis.
2. The total sample size 28,073 is large, but not huge. Therefore, the general rule of statistical presentation should be followed. P values ​​are shown to 3 decimal places or only as p < 0.001 if less than 0.001. Frequency displays raw counts and percentages (accurate to 1 decimal place). Odds ratios are shown to 2 decimal places and 95% CI. The detail suggestions are as following.
3. Tables 2 and 3 require raw counts. Table 4 requires p-values ​​for each variable.
4. Results Section 3.1 should present the data in sentences, Section 3.2 needs to report the O.R. with a 95% CI. It is recommended to change the reference group so that the OR is greater than 1 for better understanding.
5. The two categories: excellent and very good of Mental Health variable are suggested to be collapsed to highlight the difference with the other categories
6. Figure 4 and 5 are recommended to be deleted for lacking information. Figure 6 and 7 require p-values
7. The Discussion section should compare the authors’ findings with other results from previous studies, not just summarize the Results section. Line 305-307 are suggested to be deleted.
Comments on the Quality of English Languagenone
Author Response
Authors: We have revised all the comments by the reviewer: we edited the presentation formats; added the required numbers in Tables 2, 3, and 4; added data with confidence intervals in sections 3.1 and 3.2; collapsed the excellent and very good mental health groups to highlight differences; added information and interpretation for Figures 4 to 7; and compared our findings with previous studies in the Discussion Section. We also edited the English extensively. We are pleased to find the results and discussions were much better presented. Thanks to the reviewer for the constructive suggestions!
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsThe previous comment is not reflected in this revision. I would like to clarify what the problem is again.
See the confidence intervals and the standard errors in Table 7. For example, a direct effect has the standard error of 46.598. But, its confidence interval is (-0.742,3.104). These numbers are really strange values. In fact, it seems that all the CIs and standard errors seem wrong when Non-Hispanic White and Non-Hispanic Other are compared in Table 7. In the point-by-point response, the authors said that the CI was obtained from the bootstrap, not from the normal theory. In fact, we can calculate s.e. from the 95% confidence interval and compare it with the last column in Table 7. This comparison will show that the current table has a serious problem. I am not sure whether this is a bug of the package or some editing problem.
Author Response
We are very grateful to the academic editor and Reviewer 1 for pointing this out. We made a mistake in the analysis. We find no difference in e-cigarette use between Non-Hispanic White and Non-Hispanic Others (Table 6, p-value=0.65). We should perform mediation analysis only on the populations that found significant differences. A mediation analysis should not be performed to explore the difference between non-Hispanic whites and non-Hispanic others. We had big standard errors for relative effects because the total effect (TE) is close to 0, and when calculating the relative effects, TE as the denominator makes the number very variable. When the bootstrap sample is not big enough, the quantiles of the estimates will not cover the estimates from the whole data set. To fix the problem, we removed the mediation analysis results comparing non-Hispanic whites with non-Hispanic others. We also explain the reasons in the manuscript.
Reviewer 2 Report
Comments and Suggestions for AuthorsThe revised version looks much better and my comments have been solved.
Author Response
We appreciate reviewer 2 for pointing this out. The results section (highlighted) provides a detailed explanation of the mediation effects.
Reviewer 3 Report
Comments and Suggestions for AuthorsRevisions to version 2 of the manuscript are acceptable, with the following exceptions.
1. Are the data in Table 4 percentages or probabilities? Clarification is needed, and annotations should be added under the table.
2. In Tables 6 and 7, P values ​​should be shown to 3 decimal places or only as < 0.001 if less than 0.001.
Author Response
- Are the data in Table 4 percentages or probabilities? Clarification is needed, and annotations should be added under the table.
Authors: Table 4 shows the probabilities. We have revised the title and added a footnote to the table.
- In Tables 6 and 7, P values ​​should be shown to 3 decimal places or only as < 0.001 if less than 0.001.
Authors: Thanks for pointing this out. We have revised all p-values to 3 decimal places or <0.001.
