Facility-level comorbidity burden and cardiovascular outcome in patients undergoing hemodialysis
Article information
Abstract
Background
The impact of facility-level comorbidity burden on the prognosis of hemodialysis patients remains unclear. This study aimed to investigate the association between facility-level comorbidity burden and hemodialysis outcomes.
Methods
We examined 15,481 participants receiving hemodialysis at primary clinics participating in the Periodic Hemodialysis Quality Assessment by Health Insurance Review and Assessment Service in Korea. Facility-level comorbidity burden, defined as the sum of the Charlson Comorbidity Index of all patients divided by the number of nurses in each hemodialysis center, was the primary predictor. The primary outcome was major adverse cardiac and cerebrovascular events (MACCE).
Results
During a median follow-up of 6.8 years, MACCE and all-cause mortality occurred in 9,797 (63.3%) and 8,513 participants (55.0%), respectively. Participants in the highest facility-level comorbidity burden had the highest incidence rates of both MACCE and all-cause mortality. Hazard ratios (HRs) of MACCE and all-cause mortality in the highest vs. the lowest quartile were 1.10 (95% confidence interval [CI], 1.03–1.18) and 1.14 (95% CI, 1.06–1.22), respectively. Applying the facility-level comorbidity burden as a continuous variable, each 10-unit increase in facility-level comorbidity burden was associated with a 4% and 8% higher risk of MACCE and all-cause mortality, respectively. These associations remained consistent across subgroups.
Conclusion
Our findings revealed that the facility-level comorbidity burden in hemodialysis centers was associated with a higher hazard of poor outcomes in patients undergoing hemodialysis.
Introduction
End-stage kidney disease (ESKD) is associated with increased risks of hospitalization, cardiovascular events, and death [1–3]. To improve the prognosis of patients with ESKD receiving hemodialysis, several guidelines recommend optimizing structural indicators of the hemodialysis center, such as staffing ratios and medical equipment quality, while assessing individual patient risk factors [4,5]. Nevertheless, ESKD has been identified as a global issue owing to the substantial social and financial burdens it places on the medical system. Therefore, efforts to develop novel structural indicators in hemodialysis centers remain crucial for improving the outcomes of patients with ESKD.
In Korea, the adequacy of hemodialysis is routinely assessed through the national Hemodialysis Quality Assessment Program conducted by the Health Insurance Review and Assessment Service (HIRA) [6]. Although the program includes structural metrics such as the number and clinical experience of physicians and nurses, it remains unclear whether these factors alone adequately reflect the quality of care.
Evidence from other clinical settings suggests that facility-level patient severity can significantly affect clinical outcomes. For instance, increased occupancy in intensive care units (ICU) and greater stroke severity have been linked to higher mortality rates [7,8]. However, to date, no study has examined patient outcomes in relation to staffing levels adjusted for facility-level severity in the hemodialysis centers. This is particularly important in hemodialysis patients, given the high comorbidity burden observed in this population [9]. Moreover, because most patients select hemodialysis centers based on proximity to their home or workplace, there can be substantial variation in comorbidity burden across facilities, which may in turn influence both staffing needs and patient outcomes [10].
Therefore, in this study, we aimed to investigate the association between facility-level comorbidity burden and the outcomes of patients undergoing hemodialysis.
Methods
Study participants
The HIRA is a national institution in Republic of Korea that evaluates the quality and cost of medical services. HIRA oversees the following two major healthcare delivery systems. The National Health Insurance System (NHIS) is a publicly funded healthcare system in Korea that covers >98% of the population. The remaining portion of the population includes low-income individuals supported by the government’s Medical Aid Program. Since 2010, the HIRA has conducted a mandatory quality assessment program biennially for patients undergoing hemodialysis in Korea, known as the Korean National Periodic Hemodialysis Quality Assessment Program. The data collected through this program are linked by HIRA to the Korean National Health Insurance claims data, which is managed by the same institution. These data are regularly made available to selected researchers. This study utilized patient- and facility-level data from this program, which were linked using unique hospital identifiers. During the 3-month quality assessment period, this program evaluates the quality of care provided to all patients aged ≥18 years who receive maintenance hemodialysis at least twice weekly (eight times per month) in outpatient hemodialysis centers. Patients hospitalized or lost to follow-up during the assessment period were excluded from the periodic hemodialysis quality assessment.
This study initially screened 21,662 individuals undergoing hemodialysis at primary clinics who participated in the fourth and fifth Korean National Periodic Hemodialysis Quality Assessment Programs from 2013 to 2015. Patients were excluded if they experienced an event within 3 months (n = 341); had missing values in center characteristics (n = 32), dialysis vintage (n = 1,465), single pooled Kt/V (spKt/V, n = 4,737), or body mass index (BMI, n = 9); or lacked key laboratory values, including hemoglobin, serum albumin, calcium, and phosphorus levels. A total of 15,481 patients receiving maintenance hemodialysis were included in the final analysis (Fig. 1).
Flow diagram of the study cohort.
BMI, body mass index.
Data collection and measurements
The start date of each hemodialysis quality assessment was considered the baseline date for the study. If a patient participated in both the fourth and fifth assessments, the date of the fourth assessment was used as the baseline date. Baseline demographic data and medical history—including age, sex, BMI, predialysis systolic blood pressure (SBP), primary kidney disease, type of health insurance, and comorbidities—were collected. Hemodialysis-related factors were also collected, including dialysis vintage, type of vascular access (e.g., arteriovenous fistula, arteriovenous graft, or central catheter), and spKt/V. Laboratory assessments of hemoglobin, serum albumin, calcium, and phosphorus levels were performed before each hemodialysis session according to standardized predialysis blood sampling procedures recommended by clinical guidelines [11]. The average values of clinical measurements obtained during the three months following study enrollment were used in the analysis. Comorbidities were identified from the NHIS claims database and defined as having one or more medical claim records for a specific disease in the past year. Comorbidities were diagnosed according to the International Classification of Diseases, 10th Revision (ICD-10) codes (Supplementary Table 1, available online).
Primary predictors and study outcomes
The primary predictor was facility-level comorbidity burden in the hemodialysis center, calculated as the sum of the Charlson Comorbidity Index (CCI) scores of all patients in each hemodialysis center divided by the number of nurses in the same center. In Korea, the medical staff of hemodialysis centers consists solely of doctors and licensed nurses. Notably, nurses at hemodialysis centers in Korea perform not only nursing duties but also take on additional roles as caregivers and technicians. Therefore, this study divided the sum of the comorbidity burden according to the number of nurses in a hemodialysis center. For the primary analysis, patients were categorized according to the quartiles of facility-level comorbidity burden in the hemodialysis centers. The facility-level comorbidity burden was also evaluated as a continuous variable (per 10-unit increase).
The primary outcome was major adverse cardiac and cerebrovascular events (MACCE), defined as a composite of nonfatal myocardial infarction, nonfatal stroke, revascularization, and all-cause mortality. All diagnoses were defined using ICD-10 diagnostic codes, related procedure or operation codes, and Main ingredient code (Supplementary Table 2, available online). The secondary outcome was all-cause mortality. The patients were followed from baseline until the outcome date or the end of the study period (May 31, 2024).
Statistical analyses
Continuous variables are represented as means and standard deviations, while categorical variables are represented as counts and proportions. Variables with skewed distributions are presented as medians with interquartile ranges (IQRs) and compared using the Kruskal-Wallis test. The Kolmogorov-Smirnov test was used to confirm the normality of the distribution. Cumulative incidence rates of MACCEs according to facility-level comorbidity burden were estimated using the Kaplan-Meier analysis. Hazard ratios (HRs) were calculated using multivariate Cox frailty models with random effects for hemodialysis centers to investigate the associations between facility-level comorbidity burden and the risks of MACCE and all-cause mortality. Model 1 was adjusted for age and sex. Model 2 was further adjusted for medical aid beneficiaries, BMI, predialysis SBP, and medical history, including diabetes, myocardial infarction, congestive heart failure, and cerebrovascular disease, in addition to the factors included in Model 1. In Model 3, primary kidney disease, vintage year, vascular access type, spKt/V, and laboratory parameters, including hemoglobin, serum albumin, calcium, and phosphorus, were further adjusted. To evaluate the prognostic value of facility-level comorbidity burden beyond individual-level clinical risk factors, we conducted a likelihood ratio test (LRT) comparing two nested Cox proportional hazards models. The base model included individual-level covariates, while the full model additionally incorporated the facility-level comorbidity burden. Additionally, we assessed the relative contribution of each factor by reporting HRs from the full-adjusted multivariable model. For sensitivity analysis, we recalculated the facility-level comorbidity burden using the age-adjusted CCI instead of the original CCI [12]. We additionally examined subgroup analyses across prespecified subgroups by sex (male vs. female), age (<65 years vs. ≥65 years), dialysis vintage (<4.0 years vs. ≥4.0 years), and nurse caseload (<3.6 vs. ≥3.6). Nurse caseload was defined as the average number of hemodialysis cases per day at each hemodialysis facility divided by the number of nurses at the same facility [13]. Conversely, we also assessed whether the association between nurse caseload and clinical outcomes differed by facility-level comorbidity index (<28.8 vs. ≥28.8). The results are presented as HRs and 95% confidence intervals (CIs). Statistical significance was defined as p < 0.05, and all statistical analyses were performed using R (version 3.5.1; R Foundation for Statistical Computing) and SAS Enterprise Guide (version 6.1; SAS Institute).
Ethics statement
The study procedures adhered to the tenets of the Declaration of Helsinki, and the research protocol was approved by the Institutional Review Board of Yonsei University Health System (No. 4-2021-1631). The requirement for informed consent was waived owing to the retrospective nature of the study using a fully anonymized dataset provided by the HIRA. This study followed the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines.
Results
Baseline characteristics
Baseline characteristics of the 15,481 patients from 312 hemodialysis centers according to the four facility-level comorbidity burden groups are presented in Table 1. The median age of the patient was 59.0 years, and 6,277 patients (40.2%) were female. The medians of facility-level comorbidity burden were 18.0 (IQR, 14.7–20.0), 25.4 (IQR, 24.0–27.0), 31.8 (IQR, 30.0–33.6), and 39.6 (IQR, 37.7–44.7) for the lowest to highest quartiles, respectively. Patients in the lowest quartile for facility-level comorbidity burden were more likely to be younger, attend hemodialysis centers with higher numbers of nursing staff, and have a lower BMI, longer dialysis vintage, and more frequent use of arteriovenous fistulas.
Baseline characteristics of patients according to the quartiles (Q) of facility-level comorbidity burden
Association of facility-level comorbidity burden with major adverse cardiac and cerebrovascular events
During a median follow-up period of 6.8 years, MACCE occurred in 9,797 patients (63.3%; 102.1 per 1,000 person-years). Facilities with higher quartiles of comorbidity burden tended to experience high incidence rates of MACCE: 91.1, 97.4, 105.1, and 117.4 per 1,000 person-years from the lowest to the highest quartiles, respectively (Table 2). The Kaplan-Meier curves showed a similar pattern (all p < 0.001) (Fig. 2). In the fully adjusted model, the risk of MACCE was highest among participants in the highest quartile of facility-level comorbidity burden. Compared to the first quartile, the adjusted HRs were 1.05 (95% CI, 0.98–1.12), 1.03 (95% CI, 0.97–1.10), and 1.10 (95% CI, 1.03–1.18) for the second, third, and fourth quartiles, respectively (Table 3). When the facility-level comorbidity burden was treated as a continuous variable, a 4% (HR, 1.04; 95% CI, 1.01–1.06) higher hazard of MACCE was observed per 10-unit increase in the facility-level comorbidity burden.
Incidence rate of outcomes according to facility-level comorbidity burden
Kaplan-Meier cumulative event rate of MACCE.
(A) MACCE. (B) All-cause death.
MACCE, major adverse cardiac and cerebrovascular event; Q, quartile.
Hazard ratios for all-cause mortality and MACCE based on the facility-level comorbidity burden
Association of facility-level comorbidity burden with all-cause mortality
All-cause mortality occurred in 8,513 patients (55.0%; 77.2 per 1,000 person-years). The incidence rates gradually increased from the lowest to the highest quartiles of facility-level comorbidity burden: 69.4, 74.3, 78.5, and 88.0 per 1,000 person-years, respectively (Table 2). After adjustment for confounding factors, the highest quartile of facility-level comorbidity burden showed the highest risk of all-cause mortality, with an adjusted HR of 1.14 (95% CI, 1.06–1.22) compared with the first quartile (Table 3).
Prognostic value of facility-level comorbidity index
Among individual-level factors, age, sex, medical aid, predialysis SBP, patient comorbidities, and dialysis vintage were significantly associated with both MACCE and all-cause mortality in fully adjusted models (Supplementary Table 3, available online). LRT was performed to assess whether the addition of the facility-level comorbidity burden significantly improved the model fit. We compared a model that included both individual patient risk factors and facility-level comorbidity burden with a nested model that excluded the facility-level comorbidity burden. The addition of the facility-level comorbidity burden significantly improved the model fit (LRT χ² = 17.08, p < 0.001).
Sensitivity and subgroup analysis
In the sensitivity analysis using the age-adjusted CCI to define the facility-level comorbidity burden, the associations with both MACCE and all-cause mortality remained robust. Compared to the lowest quartile, the highest quartile of the facility-level comorbidity burden was significantly associated with higher risks of MACCE and all-cause mortality, with adjusted HRs of 1.11 (95% CI, 1.04–1.19), 1.14 (95% CI, 1.06–1.22), respectively, compared with the first quartile (Supplementary Table 4, available online).
Subgroup analyses were conducted by sex (male vs. female), age (<65 years vs. ≥65 years), dialysis vintage (<4 years vs. ≥4 years), and nurse caseload (<3.6 vs. ≥3.6). There were no significant interactions observed for any of the subgroups (all p for interaction > 0.05) (Fig. 3). Across all subgroups, patients in facilities with the highest comorbidity burden consistently exhibited the highest hazard of MACCE. Similarly, no meaningful interaction was found for all-cause mortality across the same subgroups (Supplementary Fig. 1, available online). In addition, we assessed whether the association between caseload and outcomes varied by facility-level comorbidity burden (<28.8 vs. ≥28.8). Although the p for interaction was not statistically significant, a trend toward higher risk of adverse outcomes with higher caseloads was observed in high facility-level comorbidity burden centers (Supplementary Table 5, available online).
Forest plots for the associations of facility-level comorbidity burden with major adverse cardiac and cerebrovascular event.
CI, confidence interval; HR, hazard ratio; Q, quartile.
Discussion
In this nationwide cohort study of 15,481 patients undergoing maintenance hemodialysis in primary clinics, patients who attended hemodialysis centers in the highest quartile of facility-level comorbidity burden showed significantly higher incidence of MACCE and all-cause mortality. Subgroup analyses showed similar results, with higher risks observed in the highest quartile of facility-level comorbidity burden. These findings imply that excessive facility-level comorbidity burden may compromise patient outcomes, highlighting its potential as a structural indicator of prognosis and care quality in hemodialysis centers.
Recent studies have investigated the association between medical personnel caseloads and patient outcomes in various medical situations using different definitions of ‘caseload.’ Some studies found that an increased patient-to-nurse ratio was associated with higher mortality rates in the ICU [14–16]. This may be due to inadequate monitoring, delayed response to emergencies, limited resources, and time constraints, all of which may lead to adverse patient outcomes [15,17,18]. Other studies reported that a higher Nursing Activity Score was associated with a higher mortality risk and an increased risk of hospital-acquired infections in ICU settings [19,20]. However, few studies have examined the association between the patient-to-nurse ratio and the prognosis of hemodialysis patients. One study found a positive association between the patient-to-nurse ratio and all-cause mortality in hemodialysis centers. [13] When the patient-to-nurse ratio increased above a certain level, the mortality risk of hemodialysis patients substantially increased. A higher patient-to-nurse ratio was also associated with a higher likelihood of not receiving appropriate hemodialysis treatment [21].
Hemodialysis is a complex and demanding treatment that requires specialized skills and knowledge from healthcare providers [22]. Patients undergoing hemodialysis require unique assessments, including dialysis adequacy, vascular access monitoring, and cardiac function evaluation, which differ from those required for other patient groups. Hemodialysis for patients with multiple comorbidities, which are known to impact mortality substantially [23–25], increases the nursing workload owing to additional tasks required for their care. These include frequent checking of vital signs and symptoms, frequent blood tests, and increased need for medication administration during hemodialysis. Therefore, the patient-to-nurse ratio, which considers patients with diverse comorbidities as having equal weight in terms of workload, may fail to accurately reflect the intensity of nursing care in each hemodialysis center.
The current study used facility-level comorbidity burden to represent nurses’ workload, considering the severity of patients’ conditions in each hemodialysis center. Using this approach, we found that a higher facility-level comorbidity burden was significantly associated with elevated risk of MACCE and all-cause mortality, underscoring the need for systemic improvements in hemodialysis care. First, current nurse staffing models, which are typically based on absolute patient numbers, should be revised to incorporate center complexity, as centers with a higher burden of comorbidities may require greater staffing resources to ensure adequate care. Second, facility-level comorbidity burden allows for more equitable comparisons across facilities and prevents potential penalization of centers treating medically complex patients. Third, the findings may provide a policy rationale for targeted governmental support to high-burden centers, such as financial incentives, staffing subsidies, or inclusion in quality improvement initiatives. Collectively, these strategies may help optimize staffing and quality monitoring in hemodialysis centers. Further studies are warranted to validate facility-level comorbidity burden as a structural indicator and to guide its implementation in health policy and clinical management. In the subgroup analysis, the association between facility-level comorbidity burden and adverse outcomes was consistently observed across all subgroups, with the most pronounced risks noted among patients in the highest quartile of facility-level burden. Notably, these associations remained consistent regardless of patients’ age, sex, dialysis vintage, or nurse caseloads underscoring the robustness of the findings. These results highlight the critical importance of evaluating not only patient-level factors but also facility-level characteristics when assessing prognosis and designing quality improvement strategies in hemodialysis care.
The findings of this study should be interpreted considering the following limitations. First, because this was an observational study, causality was uncertain, and potential confounding factors may not have been completely controlled. While we attempted to reduce potential heterogeneity by restricting the analysis to primary hemodialysis centers without inpatient wards or ICU settings, the possibility of residual and unmeasured confounding cannot be entirely ruled out. Second, this study lacked a specific cause of mortality owing to the nature of the data sources. Finally, this study included only Korean hemodialysis centers; thus, the generalization of these findings to other countries with different healthcare systems is limited. However, there were fewer racial or geographic differences compared to other countries’ studies.
In conclusion, a markedly elevated facility-level comorbidity burden in hemodialysis centers was significantly associated with higher hazards of MACCE and all-cause mortality. These findings may provide insights into the need for appropriate staffing adjustments based on facility-level comorbidity burden to improve patients’ ESKD outcomes. Further well-designed studies adapted to each healthcare system are required to evaluate the impact of facility-level comorbidity burden on patients in each hemodialysis center as a measure of hemodialysis quality assessment.
Supplementary Materials
Supplementary data are available at Kidney Research and Clinical Practice online (https://doi.org/10.23876/j.krcp.25.118).
Notes
Conflicts of interest
Tae-Hyun Yoo is the Editor-in-Chief of Kidney Research and Clinical Practice and was not involved in the review process of this article. All authors have no other conflicts of interest to declare.
Funding
This research was supported by a grant from the Joint Project on Quality Assessment Research, Republic of Korea.
Acknowledgments
This study used Health Insurance Review and Assessment Service (HIRA) research data (M20220125787) acquired by the HIRA. The views expressed are those of the authors and not necessarily those of the HIRA and the Ministry of Health and Welfare. The epidemiologic data used in this study were obtained from the Periodic Hemodialysis Quality Assessment by HIRA.
Data sharing statement
Data sharing does not apply to this article, as the original data is available after getting approval from HIRA. Datasets are available at https://opendata.hira.or.kr with the permission of HIRA. The additional information is available from Dr. Kim upon request (e-mail: drhwint@yuhs.ac).
Authors’ contributions
Conceptualization: HJK, SJH, HWK
Data curation, Formal analysis: HJK, HWK
Investigation: HJK, JTP, GYH, KWK, YUK, SHK, GOK, SHH, THY, SWK, HWK
Methodology: HJK, SJH, JTP, HWK
Supervision: JTP, HWK
Writing–original draft: HJK
Writing–review & editing: HWK
All authors read and approved the final manuscript.