The Association of Social Determinants of Health on Mortality among Adults with End-Stage Kidney Disease: A Systematic Review and Meta-Analysis

Abstract

Rationale/Objective:

Synthesize evidence from published studies on end-stage kidney disease (ESKD) mortality and determine associations between social determinants of health (SDOH) and ESKD mortality.

Study Design:

This PRISMA-guided study examined all published research across four electronic databases.

Study Populations:

Adults with ESKD living in the United States.

Selection Criteria for Studies:

Clinical trials, observational, cross-sectional, and prospective/retrospective cohort studies were considered if one or more of the five domains of SDOH were present in the analysis.

Extraction/Analysis:

Extraction was performed independently by two reviewers using a standardized form that captured study purpose and design, outcomes, SDOH domain, setting, participant characteristics, and description of results. A systematic review of all eligible studies was completed. A meta-analysis was conducted using the restricted maximum likelihood method for the random effects model to assess the association of SDOH with mortality. Subgroups were analyzed by SDOH domains.

Results:

86 studies met inclusion for the systematic review which were categorized by SDOH domain. Meta-analysis showed that those with higher levels of education (pooled HR 0.83 [95% CI, 0.69–0.1.00]) and those living in urban areas (pHR 0.92 [0.88–0.96]) had lower mortality risk. People from various minoritized backgrounds (Black-pHR 0.89 [0.85–0.93]; Asian-pHR 0.59 [0.55–0.64]; and American Indian/Alaska Native-pHR 0.82 [0.77–0.88]) had lower mortality risk compared to White populations.

Conclusions:

SDOH have a significant impact on mortality outcomes in persons with ESKD. The reductions in mortality risk reveal a survival paradox where it appears that minoritized groups have improved survival outcomes, but they are diagnosed with ESKD earlier than White populations and live on dialysis longer. It is critical to conduct research to develop strategies to mitigate individual social needs.

Registration:

Registered at PROSPERO: CRD42023483453

INTRODUCTION

End-stage kidney disease (ESKD) occurs when a person’s kidneys cease functioning permanently, with options of dialysis or receiving a kidney transplant for continued survival [1]. It affects nearly 800,000 persons in the United States (US) [2]. Despite an overall decrease in adjusted ESKD incidence by 8.9% between 2001 and 2019 and by an additional 3.1% in 2020, ESKD incidence count increased in 2021 to 135,972 [2]. Furthermore, there was a staggering financial impact of ESKD. In 2019, total Medicare fee-for-service spending for those with ESKD (including prescription drugs) was $37.3 billion or $86,400 per person per year, accounting for about 7% of the Medicare paid claims costs [3].

ESKD disproportionately affects minoritized groups; minoritized describes a group that is devalued, oppressed, and given less resources and representation than the dominant population, yet the balance of power remains with the dominant population [4]. In 2021, Black, American Indian/Alaska Native (AI/AN), and Hispanic individuals experienced incidence rates of 3.8, 2.3, and 2 times that of White individuals, respectively [2]. Additionally, all-cause mortality from ESKD abruptly increased for those receiving hemodialysis, with age-adjusted mortality rates increasing by 12% for White, 23% for Black, and 29% for Hispanic individuals in 2020 [2].

Minoritized persons experience varied health outcomes based on their race and ethnicity as a result of the complex and interacting mix of social, cultural, and biological factors [5]. For instance, AI/AN and Hispanic individuals with ESKD who resided in a rural state such as South Dakota (SD) in the United States experience high levels of health disparities. SD’s rural and frontier counties have high levels of adverse social determinants of health (SDOH), such as chronic underfunding and poor proximity to healthcare services [6,7], racial discrimination [8,9], geographic placement onto reservations (for American Indians) [10], and extensive poverty [11,12]. These factors likely play a significant role in health outcomes. A recent literature review suggested that SDOH had clear and strong associations with both chronic kidney disease (CKD) and ESKD and substantially contributed to CKD and ESKD disparities across racial and ethnic groups [5]. However, the literature has not been examined systematically, and it has primarily focused on socioeconomic disparities in various kidney health outcomes, e.g., prevalence and mortality [5]. In addition, most studies included in their review were not specifically related to ESKD [5]. While socioeconomic factors provide a basic view of the complex web of SDOH, biological and clinical factors, with social constructs of race, ethnicity, and other individual SDOH (e.g., food insecurity, housing insecurity, lack of transportation, limited social support, and neighborhood crime) could also impact ESKD disparities. On the other hand, various review and meta-analysis studies identified other factors, including comorbidities, frailty, physical function, and SDOH, that contributed to the mortality of patients with ESKD [13–22]. However, these studies did not investigate the relationships of these factors across racial and ethnic groups.

The purpose of this systematic review and meta-analysis was to synthesize evidence from published studies on ESKD mortality related to SDOH among adults living in the US. The review identified SDOH from the five domains outlined in the Healthy People 2030 framework: economic stability, education access and quality, healthcare access and quality, neighborhood and built environment, and social and community context, to determine the relationships among these domains and ESKD mortality. In addition to the synthesis based on SDOH, we also aim to analyze the included literature by race and ethnicity groups. Prior to carrying out the review, a search of PROSPERO, MEDLINE, the Cochrane Database of Systematic Reviews, and JBI Evidence Synthesis was conducted, and no current or in-progress systematic reviews on the topic were identified.

METHODS

Study Design

This systematic review and meta-analysis was conducted in accordance with the Joanna Briggs Institute Methodology for systematic reviews [23] and is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines.

Search Strategy

A comprehensive search was performed across multiple databases, including MEDLINE (PubMed), CINAHL (EBSCOhost), Web of Science Core Collection (which includes Science Citation Index Expanded, Social Sciences Citation Index, and Arts & Humanities Citation Index), and the Cochrane Central Register of Controlled Trials (CENTRAL). The search strategy outlining the terms and syntax for MEDLINE is detailed in Table S1 in the supplementary materials and was modified to match the proper syntax for all searched databases. The search was developed in partnership with a Health Sciences Librarian with expertise in evidence synthesis methods. An initial search included studies published from inception to December 19, 2023, and were considered for screening. The same search was re-run and updated before the final extraction. Reference lists of identified articles were also screened for additional relevant studies.

Eligibility Criteria

Eligible study designs included randomized and nonrandomized clinical trials, observational studies, cross-sectional, prospective cohort, and retrospective cohort studies that took place in the US. These were considered for inclusion based on the following criteria: 1) adults 18 years or older with a diagnosis of ESKD (with or without prior transplant); 2) one or more of the five domains of SDOH was included in the modeling for study analysis. The SDOH domains specifically reflect those identified within the priority areas of the Healthy People 2030 [24]. The five domains of SDOH, definitions, and specific categorical elements of each domain are provided in Table 1. If a study did not include data from any of the five domains of SDOH but reported on race and ethnicity differences only, this study was not included. While a person’s race and ethnicity are not considered a SDOH, racism is considered a SDOH by the World Health Organization [25] and there are correlations between a person’s health-related social needs and their race and ethnicity [26]. Therefore, all articles that reported race and ethnicity needed to also include an additional SDOH domain in their analysis to be included.

Table 1.

SDOH Definitions and Categorical Elements[24]

SDOH Domain	Domain Description	Categorical Elements
Healthcare Access and Quality	constructs relevant to healthcare access and quality, including access to providers, services, and healthcare systems, as well as the perceived or documented quality of care.	Access to hospitals Provider Availability Health Coverage/ Insurance Distance to dialysis centers Provider Linguistic and Cultural Humility Quality of care financial constraints provider shortages geographic accessibility. prolonged wait times for specialist care limited provider availability Underinsurance limitations impacting access to services delays in diagnosis and treatment interruptions
Education Access and Quality	constructs related to education access and quality, often in the context of participants’ educational backgrounds, experiences with schooling, and the ways these intersected with other factors such as language, literacy, and engagement with care.	Literacy Health literacy Access to education Early childhood education Language Higher education Vocational Education Educational attainment
Economic Stability	constructs tied to participants’ financial circumstances, including how economic resources—or lack thereof—affected access to care, stability in daily life, and the broader ability to maintain health over time.	Reliance on public transportation vs personal transportation Employment Access to affordable housing Income Medical Bills Expenses employment instability financial insecurity lack of paid leave delays related to potential lost wages limited access to public transit reliance on emergency services for primary care lack of flexible employment
Social and Community Context	constructs tied to participants’ social and community environments, including how they impact health-related experiences, interactions, or sense of connection or disconnection within health systems.	Social support systems Community Event Involvement Social stressors/ discrimination/ stigma/ medical mistrust Social isolation Community-based interventions
Neighborhood and built environment	constructs related to participants’ surrounding environments, highlighting how features of the built and physical landscape influenced exposure, access, and day-to-day conditions shaping health-related experiences	Access to public transportation Access to public outdoor spaces (green space) Parks/ playgrounds Zip code tabulation area/ Geography Air quality Income by neighborhood/ zip code Food deserts Safe Housing Safety Walkability Rural/ Urban Employment rates environmental hazards Healthcare deserts Urban healthcare facilities were reported as having a disproportionate ratio of patients to providers

Data Extraction and Management

All members of the author team participated in article screening, with at least two reviewers independently assessing each article for eligibility. Discrepancies were resolved through discussion with the full review team until consensus was reached. Full texts were obtained for all relevant articles and eligible articles were extracted using a standard extraction template in Covidence [27] that captured study purpose and outcome, study design, analysis method, SDOH domain(s), study setting/context, participant characteristics, comorbidities, participant subgroups, and a description of the main results. For studies eligible for inclusion in the meta-analysis, we extracted effect size metrics, risk ratios (RR), odds ratios (OR), and hazard ratios (HR).

Risk of Bias Assessment

Two members of the review team assessed each included article for risk of bias using one of two checklists: 1) the JBI Critical Appraisal Checklist for Cohort Studies [28], and 2) the JBI Critical Appraisal Checklist for Prospective/Quasi-Experimental Studies [29]. All eligible studies were included in the review regardless of the risk of bias assessment results for two reasons: 1) due to the lack of eligible studies for the meta-analysis, and 2) intentional inclusion and specific analysis of SDOH variables and how they are associated with ESKD mortality is relatively new in scientific literature.

Assessment of Heterogeneity and Publication Bias

Heterogeneity was assessed by calculating I² statistic [30]. An I² value of greater than 50% (considered moderate heterogeneity) warranted further investigation through subgroup analysis to explore potential reasons for the variation. Publication bias was calculated using Egger’s [31] and Begg’s [32] tests.

Statistical Analysis

All analyses were conducted using R (v4.4.1) [33] and the “meta” package [34]. Hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs) were extracted directly from the included studies when available. The HRs and their confidence intervals were log-transformed for analysis. The standard error of the log-transformed HR was computed using the upper and lower bounds of the confidence interval. The resulting log HRs and standard errors were then pooled using the inverse variance method. Given the heterogeneity across studies in terms of design, population, and setting, a random-effects model was employed. In this model, study weights were calculated as the inverse of the within-study variance plus the estimated between-study variance. The between-study variance component was estimated using the DerSimonian–Laird [35] method and incorporated into the weighting.

To ensure consistency across studies, hazard ratios were harmonized by switching reference groups where necessary. This was done by taking the reciprocal of the reported HR and its confidence interval bounds, so that all comparisons reflected the same reference group. Where there are three or more studies with a specific SDOH variable and harmonized reference group then a meta-analysis was done. When this is not possible, the summaries of the effects of SDOH within studies are presented. These results were presented as forest plots.

Subgroup and Sensitivity Analysis

Subgroup analysis was completed on predefined groups comprised of the five SDOH domains (healthcare access & quality; education access & quality; economic stability; social & community support; and neighborhood & built environment). Analysis of a subsect of minoritized racial and ethnic groups (people identifying as either Black, Asian, AI/AN, and/or Hispanic) and their survival differences were compared to White populations.

Data Synthesis

Data were synthesized using a narrative approach and meta-analytical techniques. The narrative synthesis provides an overview of each domain of SDOH in the included studies. These five overarching domains each encompass multiple, interrelated social determinants of health, and some degree of overlap between domains is both expected and appropriate. For example, limited access to public transportation can simultaneously reflect aspects of the built environment, barriers to healthcare access, and economic instability. Because of this cross-domain relevance, certain findings were included under more than one domain in the narrative synthesis to fully reflect their multidimensional impact. This approach acknowledges that the Healthy People 2030 domains serve as organizing frameworks rather than mutually exclusive categories. We used forest plots to illustrate the meta-analysis estimates and summarize the findings of relevant studies that were not able to be analyzed with meta-analysis.

Quality of Evidence

The overall quality of evidence for each outcome was evaluated using the GRADE Framework [36]. This approach involves assessing the risk of bias, the directness of the evidence found, the consistency of the results, precision of estimates, and the likelihood of publication bias.

RESULTS

Study Selection and Characteristics

An initial search across five databases – MEDLINE, CINAHL, Web of Science, PubMed, and CENTRAL – yielded 1,955 records, of which 23 were duplicate records and were removed prior to screening. Of the 1,932 titles/abstracts screened, 328 were identified for full-text screening, of which 86 met our inclusion criteria and were included in the systematic review and meta-analysis. The detailed PRISMA flow diagram is available in Figure 1. The categorical elements of SDOH are provided in Table 1 within their corresponding domains. Data extracted from the included articles is available in Supplemental Table S2. The Quality Appraisals for cohort studies and prospective/quasi-experimental studies can be found in Supplemental Tables S3 and S4, respectively.

FIGURE 1. PRISMA Flow Diagram.

Source: Page MJ, et al. BMJ 2021;372:n71. doi: 10.1136/bmj.n71.

This work is licensed under CC BY 4.0. To view a copy of this license, visit https://creativecommons.org/licenses/by/4.0/

Systematic Review of Included Studies

Of the 86 included studies, 59 explored the domain of healthcare access and quality (A&Q).[37–95] Articles within this domain explored a variety of SDOH constructs relevant to healthcare access and quality, including access to providers, services, and healthcare systems, as well as the perceived or documented quality of care. Reported barriers to healthcare A&Q included financial constraints, provider shortages, and geographic accessibility.^{[38,39,47,51,53,56,62,64,68–72,75,88,89,93,96]} Several studies indicated prolonged wait times for specialist care due to limited provider availability, including in rural areas [38,53,56,62–64,68,69,75,84,89]. Underinsurance was cited as a recurring issue, with coverage limitations impacting access to services [41,47,51,53,61–63,73,75,88]. Reports included delays in diagnosis and treatment interruptions [45,47,63,72,75].

Education A&Q was discussed in 18 studies [63,68,72,73,77,84,97–108]. Within this domain, articles examined a variety of constructs related to education access and quality, often in the context of participants’ educational backgrounds, experiences with schooling, and the ways these intersected with other factors such as language, literacy, and engagement with care. Studies identified health literacy, language barriers, and disparities in educational resources [63,77,97,99,100,102,103,105]. Differences in education levels were reported in relation to healthcare navigation and engagement with preventive care [68,72,97,99–101,105,106]. Some studies cited the limited availability of culturally competent health education [63,100]. Misinformation, low literacy, and delayed treatment-seeking behaviors were also reported [63,97,102]. One study identified potential education-related bias impacting provider care and patient outcomes [100]. Some studies reported a correlation between higher education and improved adherence to treatment recommendations, preventive care use, and chronic disease management outcomes [72,100].

Economic stability was examined in 32 studies [40,47–49,53,55,56,62,68,70–72,75–78,81,83,84,89,93,101,103,105,106,109–115]. Articles within this domain included constructs tied to participants’ financial circumstances, including how economic resources—or lack thereof—affected access to care, stability in daily life, and the broader ability to maintain health over time. Reported barriers included employment instability, financial insecurity, and lack of paid leave [47,53,56,62,70,71,77,101,103,105,112]. Studies described delays in treatment due to financial constraints and potential lost wages [47,53,56,62,70,71,77,101,103,105,112]. Underinsurance limited access to necessary medical care [41,47,51,53,61–63,75,88,96,112]. Transportation difficulties were reported, with limited access to public transit identified as a barrier in lower-income areas reflecting both infrastructural gaps and the financial burdens of unreliable or unaffordable transportation options [55,106,109,112]. Some studies reported reliance on emergency departments for primary care [49,72,103,105]. Several studies detailed how individuals lacking paid sick leave or flexible employment were less likely to attend scheduled medical visits, delaying preventive or follow-up care [47,62,72,83,100,101,111].

Social and community context was analyzed in 31 studies [37,39–43,45–51,54,68,76,84,96–99,102,103,105,107,109,110,116–119]. Within this domain, constructs tied to participants’ social and community environments, including how they impact health-related experiences, interactions, or sense of connection or disconnection within health systems. Studies additionally reported on social support networks, stigma, and medical mistrust [37,39,40,54,68,76,96,102,103,105,107,118]. Reports noted adherence to treatment among individuals with strong support systems [96,103,105,118]. Social isolation, experiences of discrimination, and fear of stigma were described as deterrents to healthcare engagement [37,118]. Studies identified historical mistreatment as contributing to medical distrust [54,68,103]. Community-based interventions were noted in some studies, with variability in access [54,109]. Individuals with strong social networks were more likely to access healthcare services consistently and adhere to prescribed treatments, while social isolation was linked with lower healthcare engagement [37,39,40,54,68,76,96,102,103,105,107,118].

Neighborhood and built environment was discussed in 33 studies [40,47,49,52–54,56,62,68,69,71,73,75,76,79,80,84,89,93–96,102,108,109,113–116,119–122]. This domain addressed constructs related to participants’ surrounding environments, highlighting how features of the built and physical landscape influenced exposure, access, and day-to-day conditions shaping health-related experiences. Studies reported on healthcare facility availability, environmental hazards, and community safety [40,47,49,53,56,62,68,75,79,84,89,93,94,108,109,114,115,120,122]. Healthcare deserts were identified with limited primary and specialty care [49,68,75,89,113]. Urban healthcare facilities were reported as having a disproportionate ratio of patients to providers [49,93]. Environmental exposures, such as air pollution, were linked to disease incidence [79,108,109,115]. Several studies indicated that individuals living in neighborhoods with high rates of environmental pollutants, substandard housing, or limited green spaces experienced higher rates of respiratory diseases, cardiovascular conditions, and other chronic illnesses [68,79,108,109,114].

Meta-Analysis of Eligible Studies

Effect sizes represented as hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were extracted from articles meeting inclusion for the meta-analysis. The random effects model was employed to calculate pooled estimates to accommodate the between-study heterogeneity as very high I² values (>70%) were observed in most meta-analysis. This is due to the small number of studies included in each study as well as large sample sizes within each study. The ${\tau }^2$ statistics measuring the between-study variance was estimated and reported [123]. Figure 2 presents the pooled HRs (pHRs) and CIs for ESKD mortality among different demographic groups and based on SDOH domain, where applicable.

Figure 2.

Forest plot for meta-analysis of SDOH factors of ESKD mortality restricted to publications that included one or more SDOH variables in their analysis

Education Access and Quality.

Three studies reported data within the Education Access and Quality domain [97,100,102]. Analysis of pooled data (Figure 2a) indicated that as education level attained increases, mortality risk decreases (pHR 0.83 [0.69–1.00]). Heterogeneity was moderately high and significant in the included studies (Q=15.19, P<0.001). The between study variability was found to be ${\tau }^2=0.0266$. Overall, the meta-analysis summary effect indicates that higher education levels generally correspond to a 10% lower risk of ESKD mortality after starting dialysis.

Neighborhood / Built Environment.

Three studies reported on the neighborhood and built environment that fit eligibility for meta-analysis [64,68,76]. Analysis of the pooled data (Figure 2b) suggests that those living in urban areas have lower mortality than those living in rural areas (pHR 0.92 [0.88–0.96]). Heterogeneity was significantly high among included studies (Q=71.73, P<0.001). The between study variability was found to be ${\tau }^2=0.0022.$ The meta-analysis shows a summary effect where urban patients generally experienced an 8% lower risk of ESKD mortality compared to rural patients, highlighting a geographic disparity.

Some articles in the systematic review reported on mortality risk based on distance to healthcare facilities. These findings have been summarized in a forest plot (Figure 3a), but due to the lack of consistent reference groups, conducting a meta-analysis for this data was not possible.

Figure 3.

Forest plot of SDOH factors of ESKD mortality restricted to publications that included one or more SDOH variables in their analysis

Healthcare Access & Quality.

Due to a lack of common reference groups in this SDOH domain, we were unable to perform a meta-analysis. However, Figure 3b provides a forest plot of the HRs and CIs of the four studies that explored this SDOH domain, specifically insurance coverage [49,65,71,76].

Social / Community Support and Economic Stability.

Due to a lack of common reference groups in these SDOH domains, we were unable to perform a meta-analysis.

Race & Ethnicity Differences.

There are several subgroup analyses that were completed to determine mortality differences among different race/ethnicity groupings. Ten studies were eligible for meta-analysis for Black versus White [37,42,49,52,62,64,65,69,87,122], four studies reported Asian versus White [52,64,69,122], five studies for AI/AN versus White [49,52,64,122], three studies for each Hispanic versus non-Hispanic White [42,69,87], and Hispanic versus non-Hispanic groups [52,71,76]. Black versus White (Figure 4a) analysis shows significantly reduced mortality risk (pHR 0.89 [0.85–0.93]) and significantly high heterogeneity (Q=1348.84, P<0.001). The between study variability was found to be ${\tau }^2=0.0071.$ Asian versus White (Figure 4b) shows significantly reduced mortality risk (pHR 0.59 [0.55–0.64]) and significantly high heterogeneity (Q=302.77, P<0.001). The between study variability was found to be ${\tau }^2=0.005.$ AI/AN versus White (Figure 4c) analysis shows significantly reduced mortality risk (pHR 0.82 [0.77–0.88]) and moderate (but significant) heterogeneity (Q=23.38, P<0.001). The between study variability was found to be ${\tau }^2=0.0023.$ Differences in mortality for Hispanic versus non-Hispanic White (Figure 4d) and non-Hispanic (Figure 4e) were not significantly different, but heterogeneity was significantly high in each analysis.

Figure 4.

Forest plots for meta-analysis of racial and ethnic comparisons of mortality in ESKD patients restricted to publications that included one or more of the five domains of SDOH in their modeling analysis

Publication Bias.

Publication bias was assessed to determine its impact on the meta-analyses using Egger’s and Begg’s tests. The results indicate that there is no significant evidence of publication bias in most comparisons, as evidenced by the p-values for Egger’s and Begg’s tests being well above 0.05 for most groups (Table 2). The exception is the SDOH domain Education Access and Quality, where Egger’s test yielded a p-value of 0.0227, suggesting potential publication bias in studies examining the impact of education on ESKD mortality.

Table 2.

Publication Bias

Meta-Analysis	P-Value
	Egger’s Test	Begg’s Test
Black vs White	0.62	0.95
Native American vs White	0.57	0.35
Asian vs White	0.94	0.46
Hispanic vs Non-Hispanic White	0.87	1.00
Rurality	0.88	1.00
Education	0.02	0.05

DISCUSSION

The findings of this systematic review and meta-analysis indicate that mortality risk in ESKD is not evenly distributed and that economic stability, education access and quality, healthcare access and quality, neighborhood and built environment, and social and community context have variable impact on mortality. Specifically, socioeconomic status, insurance coverage, education level, geographic location, and psychosocial context influence mortality.

White persons diagnosed later in life had the shortest survival times on dialysis, and previous research suggests that age at diagnosis plays a critical role in survival analyses differences between minoritized populations and White persons [124]. AI/AN, Black, and Asian populations have 16%, 19%, and 38% lower mortality risk than White populations, respectively. White populations have historically had improved access to care during CKD stages [125] that can delay progression to ESKD. These results suggest significant barriers to accessing healthcare services during early kidney disease in AI/AN, Black, and Asian populations and highlights a survival paradox. Superficially, it appears that White persons are at a survival disadvantage, but they are receiving more healthcare services, thus delaying the initiation of dialysis. Uninsured adults have severely limited access to healthcare and generally suffer worse health outcomes compared to their insured counterparts [126,127], resulting in underdiagnosed diabetes and hypertension, the two primary causes of ESKD [128]. People who lack insurance may not know they have kidney disease until they are seriously ill and present to an emergency room, where they are found to be in kidney failure and in need of dialysis. Despite eligibility for Medicare on dialysis initiation, coverage for most people starts after 90 days [129]. Furthermore, Medicare is only available for US citizens and non-citizens who meet specific eligibility criteria; for people who are undocumented, receiving proper care if dialysis is needed would pose significant hardship. There is a significant need for increased access to kidney disease screenings for under- or uninsured people, so efforts can be taken to help identify options to obtain insurance to delay dialysis initiation.

Our meta-analysis of rurality showed that people living in urban areas have lower mortality risk than those in rural or frontier areas. This is true for people receiving hemodialysis and peritoneal dialysis (PD) [68]. People residing in rural areas and are on PD are more likely to have gaps in their PD training, dietary education, and rapid responses to complications than those in urban areas [68,130]. Additionally, rural hospitals often lack the resources to effectively manage PD complications, resulting in higher mortality rates than people receiving care at urban centers [131]. People on hemodialysis also have a higher risk of mortality the farther they live from their dialysis unit, especially those living more than 20 miles from the facility [68]. Distance to their facility has other ramifications, such as poorer access to nephrologists and transplant centers, as people residing in rural areas are less likely to receive a kidney transplant regardless of dialysis modality [68].

Lower education levels correlate with reduced health literacy, affecting both treatment adherence and care navigation.[97] Many patients struggle to understand treatment options, contributing to decreased health engagement and compromised disease management. Further, persons with higher education levels are more likely to engage in effective self-management practices and adhere to prescribed treatment regimens, improving their outcomes [132,133]. Additionally, higher education is correlated with higher socioeconomic status, which results in better access to insurance, healthcare services, and generally more secure living conditions that are conducive to improved health [134]. This was confirmed in our meta-analysis, with higher educational attainment being associated with better long-term survival rates among persons on dialysis.

Assessing the direct impact of SDOH on health outcomes is relatively new in the medical literature, despite controlling for confounding variables (such as insurance status, education level, etc.) using multiple models in survival analyses. This is particularly challenging using large datasets, such as the USRDS, which was the database of choice for 45 of the included studies. While the USRDS database does give a snapshot of some SDOH variables at the start of dialysis, there is a lack of longitudinal data to better support assertions made about the true long-term impact of SDOH on outcomes in ESKD. Furthermore, SDOH measures are not yet consistently collected on individual levels, and many SDOH are community-level variables or highly overlap with individual-level SDOH. Because of this, understanding the cumulative effects of SDOH and intersectionality between multiple SDOH and their impact on mortality are difficult to discern. This also makes it difficult to reproduce interventions across communities to improve SDOH for individuals with ESKD. Therefore, researchers should move beyond including demographic variables as a “one and done” solution. Rather, we need to intentionally assess SDOH through use of validated instruments (e.g., Accountable Health Communities Health-Related Social Needs Screening Tool [135,136]) using prospective designs to better assess the individual health-related social needs of people with ESKD at multiple points in time, striving to improve social needs alongside medical treatment delivery.

Limitations

This systematic review is not without limitations. First, many of the included studies did not have sufficient data in some domains to meet inclusion in the meta-analysis. Further, categorizing race and ethnicity posed significant challenges as these categories are often overly broad, potentially hiding disparities within small subgroups within each race and ethnicity. Additionally, the impact of race-based glomerular filtration rate (GFR) calculations on diagnoses and outcomes was not evaluated due to historical variability in use throughout the timeline for included studies. In some of the domains of SDOH, there were not common group comparisons limiting our ability to conduct a meta-analysis, such as for insurance and distance to facility. Finally, many of the included studies had methodological limitations such as small sample sizes and limited variation in SDOH factors, which may have influenced the reliability of the results. Further, included studies did not necessarily measure SDOH domains consistently. Some studies also reported odds ratios instead of HRs, which are not suitable for our meta-analysis, which further complicated the aggregation and comparison of data across the SDOH domains. Finally, meta-analysis results showed large values of I² indicating high heterogeneity between studies. This was due to the small number of studies included within each meta-analysis and large sample sizes within each study.

CONCLUSION

Our systematic review and meta-analysis highlights the significant association between SDOH and mortality among people with ESKD. The findings of our systematic review and meta-analysis indicate that factors such as lack of adequate insurance, rural residence, and people identifying as a minority race and/or ethnicity are associated with reduced mortality risk compared to White peers. However, it is no longer sufficient to review a short list of SDOH variables as one-time inclusions in clinical analyses. Instead, a deliberate and ongoing assessment of clinical outcomes, comorbid conditions, treatment patterns, and SDOH is required. The new dialysis enrollment form now gathers SDOH data, but this it is still incumbent on healthcare providers to take action on any identified concern [137]. By broadening the focus beyond conventional medical interventions, the healthcare researchers and clinicians can make substantive progress in addressing the complex social needs of individuals living with ESKD.

Data Sharing Agreement

All data used to conduct this systematic review and meta-analysis are available in the supplemental materials or referenced in the reference list.