County-Level Association Between Social Vulnerability and Rheumatoid Arthritis-Related Mortality in the United States

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This is a well-conducted study addressing a relevant question regarding the role of social context, as measured by the Centers for Disease Control and Prevention Social Vulnerability Index (SVI), in age-adjusted mortality from rheumatoid arthritis (RA) among U.S. patients from 2010 to 2019. However, some aspects require further clarification.

1. Because the four SVI thematic domains may partially overlap conceptually, reporting multicollinearity diagnostics (such as variance inflation factors) could further support the stability and interpretation of the mutually adjusted estimates.
2. While the proposed explanations related to Medicaid expansion, healthcare access, and specialist availability are plausible, some interpretations may extend beyond what can be directly inferred from the ecological design. The Discussion could benefit from slightly more cautious wording, emphasizing the speculative nature of these mechanisms.
3. Throughout the manuscript, terms such as “independent association” may be interpreted as implying stronger causal separation between domains than can be established in an ecological observational study. Consider using more neutral formulations such as “mutually adjusted associations” or “associations after adjustment for the remaining domains.”
4. The Discussion appropriately highlights potential differences between urban and rural healthcare access. However, urbanicity itself may represent an important contextual factor influencing both SVI domains and mortality patterns. Clarifying whether this was considered analytically or discussing it more explicitly as a potential source of residual confounding would strengthen the interpretation of the findings.
5. Considering disability was one of the most important predictors for RA-related mortality, a brief discussion regarding rehabilitation would further add nuance.
6. The present findings should also be interpreted in the context of prior population-level studies examining social vulnerability and mortality in other immune-mediated rheumatic diseases.

Author Response

Response to Reviewer #1

This is a well-conducted study addressing a relevant question regarding the role of social context, as measured by the Centers for Disease Control and Prevention Social Vulnerability Index (SVI), in age-adjusted mortality from rheumatoid arthritis (RA) among U.S. patients from 2010 to 2019. However, some aspects require further clarification.

Response:

We thank the reviewer for the thoughtful and constructive evaluation of our manuscript. We appreciate the reviewer’s recognition that the study addresses an important question regarding social vulnerability and RA-related mortality. We have revised the manuscript to clarify the interpretation of mutually adjusted models, to temper causal language, to address multicollinearity and urbanicity, and to expand the discussion of rehabilitation and prior population-level studies in immune-mediated rheumatic diseases.

Comment 1

Because the four SVI thematic domains may partially overlap conceptually, reporting multicollinearity diagnostics, such as variance inflation factors, could further support the stability and interpretation of the mutually adjusted estimates.

Response:
We thank the reviewer for this important suggestion. Because the four SVI thematic domains may be conceptually related, we performed multicollinearity diagnostics for the multivariable model that included all four domains simultaneously. Variance inflation factors were low for all domains: RPL_THEME1 = 1.002, RPL_THEME2 = 1.052, RPL_THEME3 = 1.203, and RPL_THEME4 = 1.253. These values are well below commonly used thresholds for concern, suggesting that multicollinearity was unlikely to materially affect the stability or interpretation of the mutually adjusted estimates.

We also examined pairwise Pearson correlations among the four SVI domains. Correlations were generally weak, with the strongest correlation observed between minority status/language and housing type/transportation (r = 0.400), followed by household composition/disability and housing type/transportation (r = 0.201). Other pairwise correlations were close to zero. These findings further support that the four domains provided largely distinct information in the analytic dataset.

We have added the VIF assessment to the Statistical Analysis section (2.3. Statistical Analysis) and reported the results in the revised Results section (3.3. Multivariable Associations of Continuous SVI Measures with RA Mortality).

Revision added to Methods:

To assess potential multicollinearity among the four SVI thematic domains included in the multivariable model, variance inflation factors (VIFs) were calculated. Pairwise Pearson correlations among the four domains were also examined to characterize the degree of overlap among SVI domains.

Revision added to Results:

Multicollinearity diagnostics supported the stability of the multivariable model. VIF values were low for all four SVI domains: RPL_THEME1 = 1.002, RPL_THEME2 = 1.052, RPL_THEME3 = 1.203, and RPL_THEME4 = 1.253. Pairwise correlations among domains were generally weak, with the highest correlation observed between RPL_THEME3 and RPL_THEME4 (r = 0.400), followed by RPL_THEME2 and RPL_THEME4 (r = 0.201) (Supplement Table 1). These findings suggest that multicollinearity among the SVI domains was limited and unlikely to materially affect the mutually adjusted estimates.

Comment 2

While the proposed explanations related to Medicaid expansion, healthcare access, and specialist availability are plausible, some interpretations may extend beyond what can be directly inferred from the ecological design. The Discussion could benefit from slightly more cautious wording, emphasizing the speculative nature of these mechanisms.

Response:
We agree. We revised the Discussion to more clearly state that mechanisms related to Medicaid expansion, healthcare access, specialist availability, and mortality documentation are hypotheses rather than direct findings from the present ecological analysis. We also added wording emphasizing that individual-level mechanisms cannot be inferred from county-level data.

Revised Discussion text:

One possible explanation relates to the study period from 2010 to 2019, which coincided with major changes in U.S. health policy, including implementation and expansion of Medicaid under the Affordable Care Act. These policy changes may have improved insurance coverage and access to care among socioeconomically disadvantaged populations, potentially attenuating traditional socioeconomic gradients in RA outcomes at the population level. However, this interpretation remains hypothesis-generating. Because the present study used county-level ecological data, we could not directly assess individual insurance status, healthcare utilization, rheumatology access, DMARD use, or changes in treatment continuity.

Comment 3

Throughout the manuscript, terms such as “independent association” may be interpreted as implying stronger causal separation between domains than can be established in an ecological observational study. Consider using more neutral formulations such as “mutually adjusted associations” or “associations after adjustment for the remaining domains.”

Response:
We appreciate this important point and revised the manuscript accordingly. We replaced language such as “independent association” with more neutral terms, including “mutually adjusted association” or “association after adjustment for the remaining domains”. These revisions were made throughout the revised manuscript to avoid implying causal separation among SVI domains.

Comment 4

The Discussion appropriately highlights potential differences between urban and rural healthcare access. However, urbanicity itself may represent an important contextual factor influencing both SVI domains and mortality patterns. Clarifying whether this was considered analytically or discussing it more explicitly as a potential source of residual confounding would strengthen the interpretation of the findings.

Response:
We agree that urbanicity is an important contextual factor that may influence both SVI domains and RA-related mortality. Urbanicity was not included as a covariate in the present analysis because the primary objective was to evaluate the domain-specific associations of SVI with RA-related mortality. We have now explicitly acknowledged this as a potential source of residual confounding in the Discussion and Limitations. We also revised the language regarding urban tertiary care access to make clear that this mechanism is speculative.

Revised Discussion text:

Because urbanicity was not directly incorporated into the present models, residual confounding by urban–rural context cannot be excluded. Future analyses incorporating rural–urban continuum codes, provider density, and rheumatology access measures may help clarify the extent to which urbanicity modifies or confounds the association between SVI domains and RA mortality.

Comment 5

Considering disability was one of the most important predictors for RA-related mortality, a brief discussion regarding rehabilitation would further add nuance.

Response:
We agree and have added a brief discussion of rehabilitation, functional support, and multidisciplinary care. Because the household composition and disability domain was the most consistent predictor of mortality, we now emphasize that rehabilitation and functional preservation may be important components of risk reduction in vulnerable communities.

Revised Discussion text:

The strong association of household composition and disability vulnerability with RA mortality also highlights the potential importance of rehabilitation and functional support. Beyond pharmacologic control of inflammation, multidisciplinary management that includes physical therapy, occupational therapy, assistive devices, fall prevention, and caregiver support may help preserve mobility, reduce disability progression, and improve resilience during acute illness. At the population level, counties with high disability-related vulnerability may benefit from integrated care models that combine rheumatology treatment with rehabilitation services and community-based functional support.

Comment 6

The present findings should also be interpreted in the context of prior population-level studies examining social vulnerability and mortality in other immune-mediated rheumatic diseases.

Response:
We agree with the reviewer that our findings should be more directly interpreted in the context of studies examining social vulnerability and mortality in other immune-mediated rheumatic diseases. We revised the Discussion to focus specifically on studies linking SVI or neighborhood-level social vulnerability to mortality outcomes in systemic lupus erythematosus (SLE), which appears to be the closest available literature. We also clarified that, compared with SLE, evidence regarding SVI and mortality in RA remains limited, supporting the novelty of the present study.

In particular, Pamuk et al. reported that state- and county-level SVI was associated with SLE-related age-adjusted mortality rates in the United States, with county-level SLE mortality significantly correlated with SVI scores. Kim et al. similarly reported, in a hospitalized SLE cohort, that higher SVI was associated with increased risk of death, with several SVI subdomains contributing to mortality risk. Reed et al. further found that low economic and household instability factors were associated with early mortality among patients with SLE after adjustment for race, sex, and disease severity. Carter et al. also reported increased mortality among lupus patients living in socially vulnerable census tracts. We have added these comparisons to the revised Discussion.

Added to Discussion:

Our findings should be interpreted in the context of emerging studies linking social vulnerability to mortality in other immune-mediated rheumatic diseases, particularly systemic lupus erythematosus (SLE). In a state- and county-level analysis of SLE-related mortality in the US, Pamuk et al. reported that counties with higher SVI scores had higher SLE-related age-adjusted mortality rates, with a significant correlation between county-level SVI and SLE mortality. They also found that declines in SLE-related mortality over time were less pronounced in regions with higher social vulnerability, suggesting that social vulnerability may influence not only mortality burden but also improvement in mortality trends. In a hospitalized SLE cohort, Kim et al. reported that patients in the highest SVI quartile died at a younger mean age than those in the lowest quartile and that higher SVI was associated with increased risk of death. Similarly, Reed et al. found that neighborhood-level low economic and household instability factors were associated with early mortality among patients with SLE after adjustment for race, sex, and disease severity. Carter et al. also reported that mortality was increased among lupus patients living in socially vulnerable census tracts, particularly among patients with disease damage. Together, these findings indicate that social vulnerability is increasingly recognized as a mortality-relevant contextual factor in SLE. Our study extends this emerging literature to RA and suggests that, while social vulnerability is relevant across immune-mediated rheumatic diseases, the domain-specific mortality patterns may differ by disease, outcome definition, population, and analytic scale.

Reviewer 2 Report

Comments and Suggestions for Authors

We are grateful to the authors for having the strength to undertake this research. The basic message is that there is a significant relationship between social vulnerability (SVI - Social Vulnerability Index) and mortality. This is evident from the very beginning of the disease (early or late diagnosis), the availability of educated rheumatologists, continuity in treatment and the possibility of applying the most modern therapy. In addition to functional disorders, the low social status of the patient contributes to poorer functional capacity, but also to earlier mortality. SVI is compared with patients suffering from interstitial lung pathology, cardiovascular diseases and coronary heart disease, but it can also be transferred to other chronic diseases of the cardiovascular system. The research was done thoroughly. The basic message from my point of view is that when dealing with chronic diseases (such as all connective tissue diseases as well as most cardiovascular diseases) (from congenital heart defects, arrhythmias and coronary heart disease), doctors must also be concerned about the social status of the patient.

 

Author Response

Response to Reviewer #2

We are grateful to the authors for having the strength to undertake this research. The basic message is that there is a significant relationship between social vulnerability (SVI - Social Vulnerability Index) and mortality. This is evident from the very beginning of the disease (early or late diagnosis), the availability of educated rheumatologists, continuity in treatment and the possibility of applying the most modern therapy. In addition to functional disorders, the low social status of the patient contributes to poorer functional capacity, but also to earlier mortality. SVI is compared with patients suffering from interstitial lung pathology, cardiovascular diseases and coronary heart disease, but it can also be transferred to other chronic diseases of the cardiovascular system. The research was done thoroughly. The basic message from my point of view is that when dealing with chronic diseases (such as all connective tissue diseases as well as most cardiovascular diseases) (from congenital heart defects, arrhythmias and coronary heart disease), doctors must also be concerned about the social status of the patient.

Response:

We sincerely thank the reviewer for the positive assessment of our work and for recognizing the importance of social vulnerability in chronic disease outcomes. We agree that social context may influence RA outcomes through multiple pathways, including timing of diagnosis, access to rheumatology care, continuity of treatment, functional capacity, and the ability to receive modern therapies. We also agree that these considerations are relevant not only to RA but also to other chronic immune-mediated and cardiovascular diseases.

In response to the reviewer’s comments, we have revised the Discussion to more explicitly emphasize the clinical implications of the study. Specifically, we added text noting that clinicians caring for chronic diseases should consider social vulnerability and functional support as part of patient-centered management. We also expanded the discussion of rehabilitation and multidisciplinary care, given the strong association between household composition/disability vulnerability and RA-related mortality.

Revised Discussion text:

Clinically, these findings reinforce the importance of considering social context when managing RA and other chronic inflammatory diseases. Social vulnerability may influence not only access to diagnosis and specialist care, but also treatment continuity, ability to adhere to therapy, functional capacity, and resilience during comorbid illness. Accordingly, patient-centered management should incorporate assessment of functional limitations, caregiving resources, transportation barriers, insurance coverage, and access to rehabilitation or community support services.

We appreciate the reviewer’s encouragement and believe these revisions strengthen the clinical relevance of the manuscript.

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript is well written and highlights current issues affecting the population with rheumatoid arthritis. It is a well-structured article, and the authors clearly and effectively acknowledge the limitations of the study. I only have a few minor comments:

In the Introduction section, it would be useful to update the data regarding the current mortality rates among patients with rheumatoid arthritis in US.

In the Discussion section, it would also be interesting to include a comparison of AR related mortality data worldwide, providing specific numbers and comparing them with the results of the present study.

Author Response

Response to Reviewer #3

The manuscript is well written and highlights current issues affecting the population with rheumatoid arthritis. It is a well-structured article, and the authors clearly and effectively acknowledge the limitations of the study. I only have a few minor comments:

Response:

We thank the reviewer for the positive and constructive comments. We appreciate the reviewer’s recognition that the manuscript is well structured and that the limitations are clearly acknowledged. We have revised the Introduction and Discussion as suggested.

Comment 1

In the Introduction section, it would be useful to update the data regarding the current mortality rates among patients with rheumatoid arthritis in the U.S.

Response:
We agree. We added updated U.S. mortality context to the Introduction to better frame the public health significance of RA-related mortality. Specifically, we incorporated recent national data showing that U.S. age-adjusted mortality rates related to RA have declined over recent decades but remain clinically important, with persistent demographic and geographic disparities.

Proposed revised Introduction text:

Recent U.S. analyses using CDC WONDER data indicate that RA-related mortality has declined over the past two decades but remains clinically meaningful. One study reported that the age-adjusted RA mortality rate decreased from 5.65 per 100,000 population in 1999 to 3.33 per 100,000 in 2019, followed by an increase to 4.07 per 100,000 in 2020. Differences in reported absolute mortality estimates may reflect differences in case definitions and death-certificate coding, including whether RA was recorded as the underlying cause or as a contributing cause of death. These persistent mortality patterns support the need to examine contextual and social factors associated with RA-related mortality.

Comment 2

In the Discussion section, it would also be interesting to include a comparison of RA-related mortality data worldwide, providing specific numbers and comparing them with the results of the present study.

Response:
We thank the reviewer for this helpful suggestion. We revised the Discussion to place the mortality burden observed in our study in the context of prior U.S. and global RA mortality estimates. We also clarified that direct numerical comparisons should be interpreted cautiously because studies differ in outcome definition, including whether RA was recorded as any cause of death or only as the underlying cause of death.

Proposed revised Discussion text:

The mortality burden observed in this study should be interpreted alongside prior U.S. and global estimates. In our analysis, 354,280 deaths from 2010 to 2019 were iden-tified among individuals with RA recorded as a cause of death. By comparison, a recent CDC WONDER-based U.S. analysis reported that RA-related mortality declined from 1999 to 2019 but increased again in 2020, suggesting that RA mortality remains a per-sistent public health concern. At the global level, the GBD 2021 RA study estimated approximately 38,300 deaths due to RA in 2020. Differences in absolute estimates across studies likely reflect differences in outcome definition and analytic scale, including whether RA was recorded as any cause of death, as in our study, or only as the underlying cause of death. Therefore, direct numerical comparisons should be interpreted cautiously, but these data collectively support RA-related mortality as a continuing global and U.S. public health burden.