Socioeconomic Disparities Along the Cancer Continuum for Hepatocellular Carcinoma: A Systematic Review

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors,

Thank you for having the opportunity to read this kind of article.

The abstract is well-organized, but the conclusion is slightly confusing.

The introduction is too short, but at the end of the introduction is an explanation of novelties. „ There is a gap in our knowledge on the role of prominent  SES-related factors such as income. Insurance, employment, and education.“ The aim is to conduct a clear review of the literature on the impact of SES on the clinical care continuum of HCC.

The Materials and Methods are accurately written according to PRISMA guidelines. Study selection and data extraction are properly conducted. The result is the PRISMA flow diagram of included studies and analysis. The review consists of 63 studies and 179 analyses on incidence, surveillance, diagnosis, treatment, survival, and end-of-life care for the HCC patients. The results are organized into chapters, and each section of the discussion explains the effect of SES on a different point in the HCC continuum. The most relevant final result is insurance, and it is most relevant. Is it acceptable to authors, and what is the possible way of explaining that?

Why does the insurance have the most impact on the different stadiums of the HCC continuum?

While previous studies have focused on the influence of race and ethnicity on HCC outcomes, the authors found among them studies connecting SES and points, being part of the HCC continuum, varied widely in their racial and ethnic categories. What induced the change in significance from unadjusted to adjusted models? Did you expect those results?

The study limitations are significant, but the data are limited until 2021.

The conclusion is not completely defined. The focus on insurance information is missing; insurance coverage is a major determinant of outcomes across the continuum of HCC care. On the other hand, the authors emphasized that future research should focus on HCC surveillance and end-of-life/survivorship, with greater emphasis on examination of modifiable individual-level social determinants.

This is a very interesting review linking socioeconomic factors and the HCC continuum in patients. Undoubtedly, insurance represents the main predictor in most of the analyses, and generally speaking, too. That kind of result is quite expecting and now proven. The authors very efficiently analysed a lot of studies in order to connect socioeconomic status and the HCC continuum. The most significant conclusion of the research is that should be investigated over the HCC continuum is surveillance and end-of-life.

Kind regards

Author Response

Authors would like to thank the reviewer for their time providing constructive criticism of our initial submission draft. Please find the attached comments for our point-by-point response. 

Comment 1:

The Materials and Methods are accurately written according to PRISMA guidelines. Study selection and data extraction are properly conducted. The result is the PRISMA flow diagram of included studies and analysis. The review consists of 63 studies and 179 analyses on incidence, surveillance, diagnosis, treatment, survival, and end-of-life care for the HCC patients. The results are organized into chapters, and each section of the discussion explains the effect of SES on a different point in the HCC continuum. The most relevant final result is insurance, and it is most relevant. Is it acceptable to authors, and what is the possible way of explaining that?Why does the insurance have the most impact on the different stadiums of the HCC continuum?

Response 1: 
Authors thank the reviewer for this important question. This is acceptable, and our interpretation of the significance of insurance is described in the Discussion section, paragraph 5 starting with “Not surprisingly, insurance coverage is a major determinant [...]” We have added the following to explain the possible reasons why insurance strongly impacts the HCC continuum: “The reasons for this are manifold, including and not limited to: (a) insurance dictates where a patient can receive care which can affect quality of healthcare delivery, (b) treatment access differs by insurance status (i.e., transplant not uniformly accessible for those with public or no insurance), (c) insurance barriers result in lack of or delays in healthcare utilization, and (d) insurance is a proxy for financial well-being and financial toxicity is well-known to impact receipt of care.”

Comment 2:
While previous studies have focused on the influence of race and ethnicity on HCC outcomes, the authors found among them studies connecting SES and points, being part of the HCC continuum, varied widely in their racial and ethnic categories. What induced the change in significance from unadjusted to adjusted models? Did you expect those results?

Response 2: 
Authors appreciate this comment by reviewers. Many racial and ethnic minorities are also of lower SES, underlining the intersectionality of these two factors. We postulate that SES is a mediator between race/ethnicity and HCC outcomes along the clinical continuum, therefore we did expect some change in the model results with SES adjustment. However, we cannot conclude whether SES alone induced the change in significance from unadjusted to adjusted models since these models were multivariate in nature (not bivariate) and other factors were included in the models such as clinical and tumor characteristics. This has been clarified in the Discussion, paragraph 6.
The study limitations are significant, but the data are limited until 2021.
 
Comment 3:
The conclusion is not completely defined. The focus on insurance information is missing; insurance coverage is a major determinant of outcomes across the continuum of HCC care. On the other hand, the authors emphasized that future research should focus on HCC surveillance and end-of-life/survivorship, with greater emphasis on examination of modifiable individual-level social determinants.

Response 3:  
This is a valid point by our reviewer. The concluding section of the manuscript has been edited in response to: “In conclusion, this systematic review demonstrates that lower SES – reflected by poverty, limited education, and inadequate insurance – consistently predicts worse HCC outcomes. Particularly, health insurance access and coverage is a significant and modifiable factor across the care continuum. Future research should target underrepresented stages of the care continuum such as surveillance and survivorship and emphasize modifiable individual-level social determinants of health.”
 
Comment 4:
This is a very interesting review linking socioeconomic factors and the HCC continuum in patients. Undoubtedly, insurance represents the main predictor in most of the analyses, and generally speaking, too. That kind of result is quite expecting and now proven. The authors very efficiently analysed a lot of studies in order to connect socioeconomic status and the HCC continuum. The most significant conclusion of the research is that should be investigated over the HCC continuum is surveillance and end-of-life.

Response 4: 
Authors thank the reviewer for their time in reading our work.

Reviewer 2 Report

Comments and Suggestions for Authors

Dear authors, 

This manuscript provides a comprehensive and timely systematic review addressing the impact of socioeconomic status on outcomes along the hepatocellular carcinoma care continuum in the United States. It offers a broad synthesis of 63 studies, with clear clinical and policy implications. The topic is important and relevant; however, several sections require improved clarity, consistency, and depth to enhance scientific rigor and readability.

1. Abstract

• Line 18–23: The sentence “Studies were heterogeneous…” is lengthy and overly dense. Consider breaking it into two sentences and briefly quantifying heterogeneity or providing an example of differing SES measures.
  Suggestion: “The included studies were heterogeneous regarding both SES definitions (e.g., individual vs. area-level measures) and outcome reporting.”
• Line 24–26: The phrase “particularly amongst analyses utilizing national cancer databases” could be clarified with examples (e.g., SEER, NCDB).
  Suggestion: Add: “—particularly in analyses using SEER and NCDB data.”
• Line 26–27: The claim that racial/ethnic disparities were “attenuated in about a quarter of analyses” should be supported with the total number (e.g., 23 analyses; 6 affected).

 

2. Introduction

• Line 35–37: The reference to ACS 2025 estimates appears predictive; ensure citation [2] corresponds to the most recent official ACS projection rather than a forward-looking estimate.
• Line 43–50: The introduction mentions previous reviews focusing on race but not SES. It would strengthen the rationale to briefly state what knowledge gap this study uniquely fills (e.g., area-level SES vs. individual-level; end-of-life outcomes underexplored).

 

3. Methods

• Line 87–103:
  The data extraction section should clarify how duplicate publications (e.g., abstracts later published as full manuscripts) were handled—only “most recent full-text used” is mentioned. Specify if earlier versions were excluded or compared.
• Line 105–121:
  The authors mention using the modified Newcastle-Ottawa Scale (NOS). Please indicate how modifications differed from the standard NOS and whether interrater reliability was assessed (e.g., kappa statistics).

 

4. Results

• Figure 1 (page 5): Well-constructed PRISMA diagram. However, clarify what is meant by “No SES factor (n=60)” in the exclusion box — does this include missing SES data or absence of SES analysis?
• Table 1 (pages 6–8):
  Comprehensive but extremely dense. Suggest splitting into two subtables: one for study characteristics and one summarizing SES predictors/outcomes. Consider abbreviating common databases (SEER, NCDB, etc.) and grouping by SES domain (income, insurance, etc.).

• Line 159–170 (Incidence): The authors summarize mixed findings but do not discuss possible reasons for inconsistency (e.g., area-level vs. individual-level data). A short explanatory note would enhance interpretation.
• Line 171–179 (Surveillance): The description of odds ratios is accurate, but please clarify “insured individuals had higher rates… but no difference in annual surveillance” — does this refer to adherence to 6-month intervals?
• Line 180–205 (Diagnosis):
  Excellent synthesis. However, the mix of ORs and HRs across studies can be confusing; consider summarizing directionality qualitatively and leaving specific statistics to Supplementary Table 3.
• Line 207–282 (Treatment subsections):
  This section is detailed but lengthy. To improve readability, summarize common trends per SES domain (e.g., “Income generally associated with higher likelihood of any treatment; insurance strongest predictor across all treatment modalities”).
• Line 283–337 (Survival):
  Well-analyzed, but the transition between subsections (income, education, insurance) is abrupt. Suggest using subheadings (e.g., 3.6.1 Income and Survival).
  Also, “OS” (Overall Survival) should be defined at first mention.

• Table 2 (page 13–16):
  Valuable table, but readability is poor due to cramped formatting and multiple abbreviations. Recommend splitting into two—one for SES-income and one for SES-insurance analyses.

 

5. Discussion

• Lines 370–384:
  The discussion effectively contextualizes the findings, but several assertions (e.g., “worse outcomes were most robust for clinical outcomes of stage at diagnosis, receipt of treatment, and survival”) could benefit from citation of specific supporting figures/tables (e.g., “see Supplemental Table 5”).
• Line 397–411:
  The paragraph on HCC surveillance is insightful; consider adding concrete examples of potential interventions (e.g., patient navigators, reminder systems) to move from observation to recommendation.
• Line 426–437:
  The paragraph on the Affordable Care Act is strong but slightly overstated—phrase such as “led to increased detection” should be qualified (e.g., “associated with increased detection”).
• Line 440–447:
  The intersectionality discussion is important. However, the interpretation that “race likely exerts an independent influence” would benefit from referencing specific results (Table 2 examples).
• Line 451–466 (Limitations):
  Well structured, but an additional limitation should be acknowledged: the potential publication bias toward studies from high-resource databases.

 

6. Conclusion

• Line 468–473:
  Strong closing. Consider tightening phrasing:
  “This systematic review demonstrates that lower SES—reflected by poverty, limited education, and inadequate insurance—consistently predicts worse HCC outcomes. Future research should target underrepresented phases such as surveillance and survivorship, emphasizing modifiable, individual-level SDOH"

Author Response

Authors would like to thank the reviewer for their time providing constructive criticism of our initial submission draft. Please find the attached comments for our point-by-point response. 

1. Abstract

Comment 1:
Line 18–23: The sentence “Studies were heterogeneous…” is lengthy and overly dense. Consider breaking it into two sentences and briefly quantifying heterogeneity or providing an example of differing SES measures.
Suggestion: “The included studies were heterogeneous regarding both SES definitions (e.g., individual vs. area-level measures) and outcome reporting.”

Response 1:
Authors agree with this feedback. This sentence has been modified to reflect the suggestions, in lines 23-24 of the updated manuscript document.

Comment 2:
Line 24–26: The phrase “particularly amongst analyses utilizing national cancer databases” could be clarified with examples (e.g., SEER, NCDB).
Suggestion: Add: “—particularly in analyses using SEER and NCDB data.”

Response 2:
Authors agree with this feedback. The sentence has been modified according to suggestions in lines 26-27 of the updated manuscript document.

Comment 3:
Line 26–27: The claim that racial/ethnic disparities were “attenuated in about a quarter of analyses” should be supported with the total number (e.g., 23 analyses; 6 affected).

Response 3:
Authors agree with this feedback. This sentence has been qualified with numerical values as suggested, now in lines 26-27 of the updated manuscript document.

2. Introduction

Comment 4:
Line 35–37: The reference to ACS 2025 estimates appears predictive; ensure citation [2] corresponds to the most recent official ACS projection rather than a forward-looking estimate.

Response 4:
Thank you for pointing this out. After further review of this reference and link, this figure still represents the ACS’s most updated ACS projection for 2025; it is also within close approximation of the ACS’s 2024 projected numbers of 41,630 new cases and 29,840 deaths. As such, authors have elected to keep this reference however with an updated reference access date of October 21, 2025, reflected in line 515 of the revised manuscript. 

Comment 5:
Line 43–50: The introduction mentions previous reviews focusing on race but not SES. It would strengthen the rationale to briefly state what knowledge gap this study uniquely fills (e.g., area-level SES vs. individual-level; end-of-life outcomes underexplored).

Response 5:
Authors agree with this point. As such, we have added some clarifying phrasing in the Introduction, paragraphs 2 and 3: “Previous systematic reviews on HCC disparities have focused on race and ethnicity as the primary predictor of interest3,4, and there remains a gap in our knowledge on the role of prominent SES-related factors such as income, insurance, and education, including both area-level and individual-level measures. Furthermore, prior reviews more narrowly focus on one aspect of the care continuum for HCC such as treatment or survival, which leaves other key stages of the continuum under-explored.”

3. Methods

Comment 6:
Line 87–103:
The data extraction section should clarify how duplicate publications (e.g., abstracts later published as full manuscripts) were handled—only “most recent full-text used” is mentioned. Specify if earlier versions were excluded or compared.

Response 6:
Earlier versions were excluded. This has been clarified in the Methods.

Comment 7:
Line 105–121:
The authors mention using the modified Newcastle-Ottawa Scale (NOS). Please indicate how modifications differed from the standard NOS and whether interrater reliability was assessed (e.g., kappa statistics).

Response 7:
Authors thank reviewers for noticing this error. We did not modify the NOS scale and the term “modified” has been removed.

4. Results

Comment 8:
Figure 1 (page 5): Well-constructed PRISMA diagram. However, clarify what is meant by “No SES factor (n=60)” in the exclusion box — does this include missing SES data or absence of SES analysis?

Response 8:
Thank you for this feedback, and authors agree that clarification is needed. This has been clarified to “missing SES data”. An updated Figure 1 (line 151) has been uploaded into the revised manuscript document.

Comment 9:
Table 1 (pages 6–8):
Comprehensive but extremely dense. Suggest splitting into two subtables: one for study characteristics and one summarizing SES predictors/outcomes. Consider abbreviating common databases (SEER, NCDB, etc.) and grouping by SES domain (income, insurance, etc.).

Response 9:
Authors thank reviewers for pointing this out and agree with this feedback. Table 1 has been split into two subtables, Table 1a (summary characteristics) and Table 1b (summary predictors and outcomes) for improved readability. These can be found on pages 6-8, lines 157-158 of the revised manuscript document. Line 139 in the result section has been modified to reflect the change in nomenclature to Tables 1a and 1b. Authors have elected to keep the order of the table by manuscript author, as multiple included papers have multiple SES predictor domains and outcomes represented in its analyses, and organizing by SES domains may lead to unnecessary redundance and worsened readability.

Comment 10:
Line 159–170 (Incidence): The authors summarize mixed findings but do not discuss possible reasons for inconsistency (e.g., area-level vs. individual-level data). A short explanatory note would enhance interpretation.

Response 10:
Authors agree with this comment. After further review of these citations, a mixture of area- and individual level data were found and line 175-176 in the revised manuscript was added highlighting this.

Comment 11:
Line 171–179 (Surveillance): The description of odds ratios is accurate, but please clarify “insured individuals had higher rates… but no difference in annual surveillance” — does this refer to adherence to 6-month intervals?

Response 11:
While the clinical standard is six month intervals, the studies in question were referring to a more liberal one-year surveillance time interval.

Comment 12:
Line 180–205 (Diagnosis):
Excellent synthesis. However, the mix of ORs and HRs across studies can be confusing ; consider summarizing directionality qualitatively and leaving specific statistics to Supplementary Table 3.

Response 12:
Authors agree with this comment. As such, quantitative OR and HRs were removed from this section for improved readability. For consistency and improved readability across other results sections, quantitative ratios were removed as well.

Comment 13:
Line 207–282 (Treatment subsections):
This section is detailed but lengthy. To improve readability, summarize common trends per SES domain (e.g., “Income generally associated with higher likelihood of any treatment; insurance strongest predictor across all treatment modalities”).

Response 13:
A summary statement was added to 3.5. Treatment header as recommended.

Comment 14:
Line 283–337 (Survival):
Well-analyzed, but the transition between subsections (income, education, insurance) is abrupt. Suggest using subheadings (e.g., 3.6.1 Income and Survival).
Also, “OS” (Overall Survival) should be defined at first mention.

Response 14:
Subheaders have been added to the revised manuscript to improve readability. Additionally, overall survival has been explicitly defined prior to being abbreviated thereafter. 

Comment 15:
Table 2 (page 13–16):
Valuable table, but readability is poor due to cramped formatting and multiple abbreviations. Recommend splitting into two—one for SES-income and one for SES-insurance analyses.

Response 15:
Table 2 has been divided into Table 2a (income as SDOH modifier) and Table 2b (insurance as SDOH modifier) for improved readability.

5. Discussion

Comment 16:
Lines 370–384:
The discussion effectively contextualizes the findings, but several assertions (e.g., “worse outcomes were most robust for clinical outcomes of stage at diagnosis, receipt of treatment, and survival”) could benefit from citation of specific supporting figures/tables (e.g., “see Supplemental Table 5”).

Response 16:
This citation has been added into the text as suggested.

Comment 17:
Line 397–411:
The paragraph on HCC surveillance is insightful; consider adding concrete examples of potential interventions (e.g., patient navigators, reminder systems) to move from observation to recommendation.

Response 17:
Authors appreciate this useful feedback and agree. As such, a concluding sentence has been added recommending these interventions to strengthen this section: “We propose investment in patient navigators and technology-aided patient and provider reminder systems as examples of potential low-cost interventions to improve surveillance rates.”

Comment 18:
Line 426–437:
The paragraph on the Affordable Care Act is strong but slightly overstated—phrase such as “led to increased detection” should be qualified (e.g., “associated with increased detection”).

Response 18:
We agree that this statement should be qualified and the sentence has been modified according to this suggestion.

Comment 19:
Line 440–447:
The intersectionality discussion is important. However, the interpretation that “race likely exerts an independent influence” would benefit from referencing specific results (Table 2 examples).

Response 19:
A reference to Tables 2a and 2b was added in addition to specific breakdowns of changes in income and insurance effect estimates.

Comment 20:
Line 451–466 (Limitations):
Well structured, but an additional limitation should be acknowledged: the potential publication bias toward studies from high-resource databases.

Response 20:
An additional sentence has been added acknowledging this publication bias in the limitations section.

6. Conclusion

Comment 21:
Line 468–473:
Strong closing. Consider tightening phrasing:
“This systematic review demonstrates that lower SES—reflected by poverty, limited education, and inadequate insurance—consistently predicts worse HCC outcomes. Future research should target underrepresented phases such as surveillance and survivorship, emphasizing modifiable, individual-level SDOH"

Response 21:
The conclusion has been tightened accordingly: “In conclusion, this systematic review demonstrates that lower SES – reflected by poverty, limited education, and inadequate insurance – consistently predicts worse HCC outcomes. Particularly, health insurance access and coverage is a significant and modifiable factor across the care continuum. Future research should target underrepresented stages of the care continuum such as surveillance and survivorship and emphasize modifiable individual-level social determinants of health.”

Reviewer 3 Report

Comments and Suggestions for Authors

Overall, it is a well-written literature review. Although a formal meta-analysis was not conducted, the authors did well in summarizing a large volume of studies for multiple SES exposures and HCC outcomes. Clarifications on the methods and criteria used to come to the conclusions will strengthen the manuscript.

 

The introduction started with racial/ethnicity disparities seen in the care continuum in HCC patients and then focused on how SES factors could potentially contribute to these disparities. However, it turns out that understanding the contribution of SES factors to racial/ethnicity disparities is only part of the analysis. It is confusing to start with such introduction then end up with a different focus in your main analysis.

 

Line 24-26 “Worse outcomes were observed with lower indicators across all SES domains and HCC outcomes, particularly amongst analyses utilizing national cancer databases.”

It is a bit too simple to put this SLR into this sentence of conclusion without considering the heterogeneity we saw in the results, especially when a formal meta-analysis was not conducted. Please clarify what criteria were used to come to this conclusion. Was it simply based on the number of studies which reported worse SES was associated with worse HCC outcomes?

 

Line 95-96 should be moved up to study selection criteria

 

Data extraction

Please clarify how data across multiple years were extracted and summarized, given that the publication years were between 2000 and 2021 and it is possible that one study reported relevant data for multiple years. From table 1, it seems that if a study reported relevant results for multiple years, all data years were combined in the final report. Please clarify it in the Methods.

 

Quality assessment

Please clarify that assessment items were modified for cohort and case-control studies. For example, in case-control studies, ascertainment of exposure should be assessed, not outcome as listed in supplemental table 6.

What scoring and classification criteria were used for the quality assessment? Per supplemental table 6, it seems that score above 7 is considered high quality. Please clarify it in the Methods.

 

Line 123-125 please remove. They seem to be irrelevant.

 

PRISMA

It says 19 studies were excluded due to ineligible study design, however, in the study selection section, the criteria based on study design is not clearly described.

Six studies were excluded due to duplicate cohort or study. If the same cohort or study resulted in a different publication, it might be due to an updated data or a different study question. If it is the latter case, it is fine not to include this publication as no new data was reported. Please provide explanation of this exclusion criteria and also add this criteria to the study selection section.

 

Line 366

Please clarify in the Methods what criteria are used to determine good quality studies. As the main study objective is to understand how each SES factor affects the HCC healthcare outcomes, bias assessment is very important. However, quality assessment is very brief and lacking essential details of how bias was assessed and taken into account in the narrative summary.

 

3.8 Impact of SES adjustment on racial and ethnic disparities

The change in significance between univariable and multivariable analysis is important to note. In addition, the change in the point estimate is also worth to be summarized given the significance level may depend on the sample size, use of appropriate statistical models, etc.

 

Limitations

The authors need to acknowledge that conference abstracts are not comprehensively captured in the listed databases and there is a possibility of missing some of them if a separate grey literature search was not conducted.

 

Supplementary materials

The overall quality score was listed for each study, however, given that multiple outcomes and exposures (i.e., SES factors) were extracted and summarized, the ascertainment methods of each of them should be assessed. In other words, each included analyses should have a NOS score.

Author Response

Authors would like to thank the reviewer for their time providing constructive criticism of our initial submission draft. Please find the attached comments for our point-by-point response. 

Comment 1:
Overall, it is a well-written literature review. Although a formal meta-analysis was not conducted, the authors did well in summarizing a large volume of studies for multiple SES exposures and HCC outcomes. Clarifications on the methods and criteria used to come to the conclusions will strengthen the manuscript.

The introduction started with racial/ethnicity disparities seen in the care continuum in HCC patients and then focused on how SES factors could potentially contribute to these disparities. However, it turns out that understanding the contribution of SES factors to racial/ethnicity disparities is only part of the analysis. It is confusing to start with such introduction then end up with a different focus in your main analysis.

Response 1:
Authors appreciate this point of feedback. We agree that race and ethnicity is not the primary focus, but wanted to highlight that existing studies have focused primarily on race and ethnicity despite a clear relationship between racial and ethnic minorities and lower SES. This is the rationale for focusing on SES in this review.

Comment 2:
Line 24-26 “Worse outcomes were observed with lower indicators across all SES domains and HCC outcomes, particularly amongst analyses utilizing national cancer databases.”

It is a bit too simple to put this SLR into this sentence of conclusion without considering the heterogeneity we saw in the results, especially when a formal meta-analysis was not conducted. Please clarify what criteria were used to come to this conclusion. Was it simply based on the number of studies which reported worse SES was associated with worse HCC outcomes?

Response 2:
Authors appreciate this useful point of feedback and agree that this sentence needs to be qualified and less definitive given the lack of meta-analysis. The original statement was present due to a majority of studies reporting a trend of worse outcomes across SES domains- the updated manuscript reflects a more attenuated version of this sentence, reflected in lines 26-27 in the revised paper.

Comment 3:
Line 95-96 should be moved up to study selection criteria

Response 3:
Authors appreciate this feedback and agree. This sentence has been moved up to the study selection subsection alongside other exclusion criteria. This is now reflected in lines 85-87 in the revised manuscript document.

Data extraction

Comment 4:
Please clarify how data across multiple years were extracted and summarized, given that the publication years were between 2000 and 2021 and it is possible that one study reported relevant data for multiple years. From table 1, it seems that if a study reported relevant results for multiple years, all data years were combined in the final report. Please clarify it in the Methods.

Response 4:
The “Date range” column of Table 1 indicates the time range of the study cohort reported by each individual study. We used aggregate data reported by each study. The publication year reflects the year that the article was published.

Quality assessment

Comment 5:
Please clarify that assessment items were modified for cohort and case-control studies. For example, in case-control studies, ascertainment of exposure should be assessed, not outcome as listed in supplemental table 6.

Response 5:
After re-review of the included studies, all were classified as cohort studies, in which case the outcome was assessed as stated in the supplemental table.  

Comment 6:
What scoring and classification criteria were used for the quality assessment? Per supplemental table 6, it seems that score above 7 is considered high quality. Please clarify it in the Methods.

Response 6:
Authors appreciate this observation and agree that this needs clarification. An additional sentence defining a “good quality” study as a NOS score of 7 or greater has been added to the methods section.

Comment 7:
Line 123-125 please remove. They seem to be irrelevant.

Response 7:
Authors would appreciate clarification on this comment. Line 123 is an important subtitle (3.1. Study characteristics and thematic design) Lines 124-125 explain the first few boxes of the PRISMA flow diagram (Figure 1). Authors have opted to retain these lines in the revised manuscript but would appreciate further clarification on which parts are irrelevant.

PRISMA

Comment 8:
It says 19 studies were excluded due to ineligible study design, however, in the study selection section, the criteria based on study design is not clearly described. 

Response 8:
The 19 studies excluded in the PRISMA diagram (Figure 1) refer to review articles, which is an exclusion criteria. Figure 1 has been modified to explicitly state that these excluded items were review papers.

Comment 9:
Six studies were excluded due to duplicate cohort or study. If the same cohort or study resulted in a different publication, it might be due to an updated data or a different study question. If it is the latter case, it is fine not to include this publication as no new data was reported. Please provide explanation of this exclusion criteria and also add this criteria to the study selection section.

Response 9:
Duplicate studies were abstracts that ended up being published as a manuscript, thus leading us to replace the abstract with data from the manuscript.

Line 366

Comment 10:
Please clarify in the Methods what criteria are used to determine good quality studies. As the main study objective is to understand how each SES factor affects the HCC healthcare outcomes, bias assessment is very important. However, quality assessment is very brief and lacking essential details of how bias was assessed and taken into account in the narrative summary.

Response 10:
Authors appreciate this feedback and agree with our reviewer. An additional sentence has been added to the methods section clearly defining a “good quality” study as a NOS score of seven or greater, reflected in lines 122-123 of the updated manuscript document.

3.8 Impact of SES adjustment on racial and ethnic disparities

Comment 11:
The change in significance between univariable and multivariable analysis is important to note. In addition, the change in the point estimate is also worth to be summarized given the significance level may depend on the sample size, use of appropriate statistical models, etc.

Response 11:
We believe it would be misleading to readers to focus on point estimate changes, as most of the confidence intervals in the multivariable analysis overlap with those seen in the univariable analysis. As a result, we cannot conclude that changes in point estimates are meaningful. As such, we have elected to refrain from summarizing point estimate changes in the revised manuscript.

Limitations

Comment 12:
The authors need to acknowledge that conference abstracts are not comprehensively captured in the listed databases and there is a possibility of missing some of them if a separate grey literature search was not conducted.

Response 12:
Lack of a grey literature search is now included as a limitation.

Supplementary materials

Comment 13:
The overall quality score was listed for each study, however, given that multiple outcomes and exposures (i.e., SES factors) were extracted and summarized, the ascertainment methods of each of them should be assessed. In other words, each included analyses should have a NOS score.

Response 13:
Authors appreciate this feedback from our reviewer. On further review of several studies including Ha et al (2016), Hoehn et al (2015), and Robbins et al (2011), the ascertainment methods of each individual analyses are consistent with the general methodology of each paper across SES factors. As such, authors feel that individual NOS scores for every analysis would not add additional meaning and detract from the readability of the NOS supplemental table. As such, authors have opted to refrain from including NOS scores for every analysis.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have thoroughly addressed all of my previous comments and provided satisfactory clarifications to the raised queries. The revised manuscript has been substantially improved in clarity, structure, and scientific rigor. I find that the current version meets the journal’s standards and is now suitable for publication in its present form.