Change of Hospitalization Status Among Patients With End-Stage Kidney Disease Before and After Receiving Chronic Hemodialysis

Compared with other diseases, end-stage kidney disease (ESKD) carries a greater risk of comorbidities including diabetes and anemia and has a higher hospital admission rate. The cause of hospital admission appears to be a common factor affecting the prognosis of patients with ESKD. Therefore, this study conducted a retrospective cohort analysis on all patients diagnosed with ESKD and receiving hemodialysis, investigating whether the type of their diagnosis for hospital admission changed before and after they started hemodialysis. This study recruited 592 and (ICD-9-CM) The patients’ demographic data and hospitalization one hemodialysis were analyzed. A McNemar test was conducted to analyze the diagnostic changes from before to after hemodialysis in patients with ESKD. investigating patients receiving dialysis, this study offers valuable insights that are conducive to analyzing epidemiological and relevant risk factors among Taiwanese patients receiving dialysis; thus, it will aid in the future prevention of renal diseases, enhance the survival of these patients, and reduce the incidence of associated comorbidities.


Background
End-stage Kidney disease (ESKD) has been a global public health problem over the past few decades [1,2], and research teams around the globe have investigated ESKD's causes and effects [3,4]. Taiwan has higher incidence and prevalence rates of ESKD that do developed countries. Compared with other diseases, ESKD results in a relatively large healthcare burden and high hospital admission rate [5,6]. According to the 2016 Annual Data Report published by the United States Renal Data System, admissions and readmissions to hospitals constituted the major burden to patients with ESKD, where the average patient with ESKD is admitted to hospital twice a year [7]. The high medical cost of ESKD treatment has been an increasing burden on Taiwan's healthcare system. According to the National Health Research Institutes' 2018 Annual Report on Kidney Disease in Taiwan, the 2016 incidence and prevalence rates of ESKD requiring dialysis were 493 and 3,392 people per million population, respectively; these gures equate to an annual increase of 1.9% over the previous three years. The admission rate per 1,000 dialysis patients increased from 923 admissions in 2000 to 1,015 admissions in 2016, indicating the increase in admission rate among patients receiving dialysis [8], and this increasing trend was primarily attributable to the admission of patients with ESKD. Accordingly, reducing the admission rate of and monitoring dialysis quality among patients with ESKD have become urgent issues for health care policy makers in Taiwan and abroad.
Compared with other diseases, ESKD has a higher mortality rate and involves a greater utilization of outpatient, emergency, and inpatient services due to its more complex clinical comorbidities (e.g., diabetes, heart diseases, anemia, cardiovascular diseases [stroke and peripheral vascular disease], lower albumin levels, and lower hematocrits) and higher risks of stroke, myocardial infarction, congestive heart failure, infectious disease, and sepsis [3,9,10]. Patients with ESKD are most burdened by admissions and readmissions, which become more frequent and severe as the disease progresses [5]. Observational studies conducted by a single medical institution or health maintenance organization have also demonstrated that patients with ESKD have a high burden of comorbidities [11]. In Taiwan, a 2016 survey of the prevalence rate of comorbidities in dialysis patients revealed that the top three comorbidities were as follows (percentage of all dialysis patients with the comorbidity written in parentheses): hypertension (83.3%), cardiovascular diseases (57.1%), and diabetes (50.4%).
The early determination of the hospitalization risk among patients with ESKD potentially reduces hospital admissions [12]. Studies have noted that in patients with ESKD, (1) patient condition and hospitalization rate are positively correlated and (2) the frequency and cause of hospital admissions affect ESKD prognosis [7,13]. Studies on patients with ESKD have mostly focused on trends for admission and death rate in the rst year of dialysis; these studies have rarely compared patient status before and after they received dialysis [5]. This study conducted a retrospective cohort analysis on all patients diagnosed with ESKD and receiving hemodialysis, investigating whether the type of their diagnosis for hospital admission changed before and after they started hemodialysis.

Ethics statement
This study has been approved by the Institutional Review Board Committee, En Cku Kong Hospital (reference number: ECKIRB1061202), and all experimental methods in this study conformed to the Declaration of Helsinki. Considering the low-risk nature of this study and its retrospective research design, the institutional review board, En Cku Kong Hospital requirements regarding patient informed consent were waived.

Data collection
This study collected data from the electronic medical records of patients with ESKD and from the dialysis patient database Taiwan Society of Nephrology: Kidney Dialysis, Transplantation (TSN-KiDiT). The Taiwan Society of Nephrology-Kidney Transplantation (TSN-KiDiT) database, having received reports from dialysis centers in Taiwan since 1997, integrates all medical records of patients with ESRD in Taiwanese hospitals and dialysis clinics who have required chronic hemodialysis. The data in TSN-KiDiT include demographic statistics, related diseases, date of rst dialysis treatment, type of dialysis, residual renal function, and laboratory data of all patients receiving dialysis in Taiwan. This study collected annual reports of dialysis facilities including dialysis dose, treatment quality, laboratory data, and clinical results. Because the TSN-KiDiT facilitates access to existing patient data systems, this study collected all the medical records of patients with ESKD from it. A teaching hospital in Taiwan was the study site, and this study conducted a retrospective analysis on the dialysis data of 1873 patients with ESKD who received hemodialysis at this hospital at any period between January 2005 and November 2017. This study aimed to investigate whether the patients were assigned different International Classi cation of Diseases Ninth Revision Clinical Modi cation (ICD-9-CM) codes one year before versus two years after they received hemodialysis. Patients were included if they satis ed all of the following criteria: assigned the ICD-9-CM code for ESKD, aged 20 years or older, received hemodialysis, and had been hospitalized only in the teaching hospital. Accordingly, this study excluded 37 patients who had received either peritoneal dialysis for a long time or a kidney transplant; 62 patients who had no record of hospitalization; 35 patients who had started their hemodialysis since 2000; 59 patients whose data did not satisfy the inclusion criteria (one year before and two years after the hemodialysis); 445 patients whose follow-up data spanned shorter than the inclusion criteria (hemodialysis data for <2 years); 244 patients with a hospitalization record only for before their hemodialysis; and 399 patients (1) who had a hospitalization record only for after their hemodialysis, (2) who had a period of less than two years when their hemodialysis began and ended, and (3) who had been transferred between hospitals. Finally, 592 patients were recruited as participants (Fig. 1).
All diagnoses of ESKD involved a nephrologist and were based on the 2012 Kidney Disease Improving Global Outcome guidelines; the nephrologist evaluated the disease condition, con rmed the diagnosis, and determined the stage of chronic kidney disease according to the cause, glomerular ltration rate (GFR), and albuminuria category (CGA) [14] . All recruited patients in the present study were in the fth stage of chronic kidney disease, having a GFR <15 ml/min/1.73m 2 [15].
This study collected the demographic data of patients with ESKD and their hospitalization data for one year before and two years after they received hemodialysis. The demographic and clinical data comprised age, sex, primary disease category, comorbidities, and reason for hospitalization. ICD-9-CM codes were used to record all primary and secondary diagnoses of patients with ESKD in the hospitalization data les. The recruited patients were assigned a total of 2995 ICD-9-CM diagnosis codes, 1% of which (100 codes) were retained in the study for analysis. Hemodialysis quality indices were established according to the On-Site Inspection Standards for Hemodialysis and Peritoneal Dialysis Practices and the clinical indices in the dialysis database published by the Taiwan Society of Nephrology [8,16]. Six comorbidities (ICD-9-CM) were analyzed in this study: diabetes, hypertension, congestive heart failure, ischemic heart disease, cerebrovascular accident, and gout [8, 16-18].

Statistical analysis
The McNemar test employed in this study is a simple and common method used for binary matched-pairs data. This test can be used in longitudinal observational research, and it involves a 2 × 2 contingency table for testing the correlation between randomly paired variables [19][20][21]. This study rst used the McNemar test to evaluate how the 100 primary and secondary diagnoses changed before versus after the patients received hemodialysis. The signi cance level was set at 0.05. Consulting the literature and considering which diagnoses signi cantly differed between before and after hemodialysis, this study selected 11 reasons for hospitalization: diabetes; essential hypertension; anemia; native atherosclerosis; urinary tract infection; gastric ulcer without mention of hemorrhage, perforation, or obstruction (hereafter gastric ulcer); pneumonia; re ux esophagitis; duodenal ulcer without mention of hemorrhage, perforation, or obstruction (hereafter duodenal ulcer); hyperkalemia; and bacteremia. The ICD-9-CM codes of these conditions were then used in a chi-square test to determine the association of the diagnosis changes with the six comorbidities and demographic variables [8, [22][23][24][25]. An analysis of variance (ANOVA) was then conducted to investigate the associations of age with those ICD-9-CM codes that exhibited signi cant differences before versus after hemodialysis. SAS for Windows version 9.4 (SAS Institute, Cary, NC, USA) was used to process and analyze the data. Table 1 presents the distributions of the demographic data. A retrospective analysis was conducted on 592 patients who were  Table 2 presents the relationship of diagnoses (ICD-9-CM codes) and patients' hospitalization status before and after they received hemodialysis. The McNemar test was used to examine how the 100 ICD-9-CM diagnosis codes [1, 8, 26, 27] changed before versus after hemodialysis. This study analyzed only 11 conditions that exhibited a statistical difference before versus after the treatment as follows: type 2 (non-insulin-dependent and adult-onset) diabetes (hereafter type 2 diabetes), essential hypertension, anemia, native atherosclerosis, urinary tract infection, gastric ulcer, pneumonia, re ux esophagitis, duodenal ulcer, hyperkalemia, and bacteremia. With regard to the causes of admission, the probabilities of being admitted for the following conditions decreased signi cantly after the patients received hemodialysis: type 2 diabetes, native atherosclerosis, urinary tract infection, gastric ulcer, pneumonia, re ux esophagitis, duodenal ulcer, and bacteremia. Comparing the numbers of patients admitted to the hospital for the 11 conditions before and after patients received hemodialysis, the result revealed a signi cant decrease from 65 admissions for type 2 diabetes (10.98%) before hemodialysis to 41 admissions (6.93%) after hemodialysis (p = 0.020); from 74 (12.50%) to 28 (4.73%) for native atherosclerosis (p < 0.001); from 90 (15.20%) to 58 (9.80%) for urinary tract infection (p = 0.009); from 91 (15.37%) to 34 (5.74%) for gastric ulcer (p < 0.001); from 113 (19.09%) to 58 (9.80%) for pneumonia (p < 0.001); from 68 (11.49%) to 40 (6.76%) for re ux esophagitis (p = 0.007); from 60 (10.14%) to 33 (5.57%) for duodenal ulcer (p = 0.005); and from 52 (8.78%) to 10 (1.69%) for bacteremia (p < 0.001). However, the probabilities of admission for essential hypertension, anemia, and hyperkalemia increased signi cantly after hemodialysis. The result revealed a signi cant increase in the number of admissions from 20 (3.38%) before hemodialysis to 112 (18.92%) after hemodialysis for essential hypertension (p < 0.001); from 76 (12.84%) to 124 (20.95%) for anemia (p < 0.001); and from 41 (6.93%) to 69 (11.66%) for hyperkalemia (p = 0.008). Table 3 presents the relationship between age and the 11 ICD-9-CM codes that had exhibited signi cant difference before versus after hemodialysis. According to the ANOVA model, age was correlated with status changes in native atherosclerosis, urinary tract infection, pneumonia, and hyperkalemia. Younger patients were less likely to be admitted to the hospital for native atherosclerosis, urinary tract infection, and pneumonia before and after hemodialysis. Patients with hyperkalemia aged between 65 and 66 years and those who received hemodialysis were more likely to be admitted to the hospital for hyperkalemia (average age: 64 years old). Patients who had experienced urinary tract infection and hyperkalemia before they received hemodialysis tended to be older (average age: 70 years old). Older patients aged approximately 69-70 years old were more likely to develop pneumonia either before or after hemodialysis. Older patients were also more likely to have a urinary tract infection relative to their younger counterparts.

Results
This study conducted a chi-square test to analyze the correlation between the six comorbidities and the changes in hospitalization diagnosis (ICD-9-CM) before versus after hemodialysis. Among patients diagnosed with type 2 diabetes, the probabilities of admission for diabetes, hypertension, and gout increased signi cantly after hemodialysis. Speci cally, the number of admissions increased from 38 (9.34%) before hemodialysis to 50 (12.29%) after hemodialysis for diabetes (p < 0.001); from 35 (7.13%) to 52 (10.59%) for hypertension (p < 0.001); and from 5 (5.21%) to 9 (9.38%) for gout (p < 0.004). The number of patient admissions to the hospital for ischemic heart disease decreased signi cantly from 7 (5.60%) before hemodialysis to 6 (4.80%; Appendix 1) after hemodialysis. Among patients diagnosed with essential hypertension, the probability of admission for hypertension signi cantly decreased after hemodialysis; speci cally, the number of admissions decreased from 99 (20.16%) before hemodialysis to 15 (3.05%) after hemodialysis for hypertension (p < 0.001; Appendix 2).

Discussion
External arteriovenous shunts, invented in 1960, facilitate the application of hemodialysis in patients with ESKD. Hemodialysis has had a considerably improved prognosis, having undergone advances over the past 40 years. This improvement is mainly attributable to improved monitoring of medical quality, improved decision-making processes, more streamlined data analysis methods, and new dialysis technologies [8, 25,28,29]. The present study on patients with ESKD included patients' demographic characteristics and medical history in its analysis of the reasons for admission one year before and two years after hemodialysis was received. This study employed a cross-sectional research design in analyzing data on six comorbidities and 12 causes of admission for 592 patients. This study's sample female (307) and male (285) patients with approximately equal populations for all age groups. Among the 592 patients, hypertension was the most common comorbidity, which is consistent with ndings in the literature; accordingly, hypertension was a prevalent disease in patients receiving hemodialysis.
Demographic statistics from the TSN-KiDiT were analyzed as categorical variables in the McNemar test for the cross-sectional analysis. According to cross-sectional datasets, this study revealed age and comorbidities as key variables affecting changes in the type of diagnosis for patients' hospital admission before and after they started hemodialysis. Additionally, according to diagnoses in relation to age, the possibility of patients being admitted to the hospital for urinary tract infection, pneumonia, and hyperkalemia decreased signi cantly among all age groups. Cardiovascular diseases are considered the most in uential risk factor for patients with ESKD during hospitalization. Most patients with ESKD have one or more comorbidities, and almost two-thirds of patients develop ESKD because of diabetes and hypertension; cardiovascular is also a prevalent comorbidity of ESKD [30]. Accordingly, the present study analyzed the six most common comorbidities of ESKD (prevalence rate written in parentheses), as follows: diabetes (68.75%), hypertension (82.94%), congestive heart failure (27.20%), ischemic heart disease (21.11%), cerebrovascular accident (21.11%), and gout (16.22%). Having multiple comorbidities increases the treatment burden (and, by implication, physical burden) on of patients, increasing the risk of noncompliance. Notably, anemia was found to be the second most common comorbidity of ESKD, echoing the nding of another study [31]. Among the patients receiving hemodialysis, the high prevalence of anemia as a comorbidity was attributable to the fact that ESKD increases the risk of anemia [32]. Moreover, the following 11 diseases may also increase the treatment burden and number of hospital admissions: type 2 diabetes, essential hypertension, anemia, native atherosclerosis, urinary tract infection, gastric ulcer, pneumonia, re ux esophagitis, duodenal ulcer, hyperkalemia, and bacteremia.
In terms of age, the older population exhibited a greatly increased prevalence of pathogenic comorbidities [33]. Three related studies have indicated that the incidence of ESKD increases with age. According to the present study's results, younger participants (aged 65-66 years) were less prone to hospital admission due to atherosclerosis of the native coronary artery, urinary tract infection, or pneumonia. Participants who were admitted to hospitals for hyperkalemia after dialysis had an average age of 64 years. Those who had a urinary tract infection and hyperkalemia before dialysis tended to be older (average age: 70 years old). Finally, participants who developed pneumonia only before or after dialysis were also older (69-70 years on average), and urinary tract infection was more common among older participants.
A clinical epidemiological study reported that uremia in ESRD changed cell and humoral immunity, and hence, increased patients' susceptibility to widespread infection. Although a low-grade urinary tract infection does not reduce renal functions, the recurrence of said infection can affect the progression of existent renal diseases, thereby causing renal functions to decline [34]. ESRD is a chronic in ammatory condition that increases patients' susceptibility to widespread infection, and its comorbidities can lead to urinary tract infection due to conditions such as diabetes and incomplete bladder emptying [27].
ESKD is a prevalent condition, particularly in Taiwan, which has the highest incidence of ESKD and has performed the most renal replacement therapies worldwide. Prone to immunode ciency, patients with ESKD have a higher risk of and mortality from infectious diseases than do those without ESKD [5,6]. The high demand for inpatient treatment due to pneumonia and sepsis remains of concern.
Infections constitute a primary factor that drives admission and death among patients receiving dialysis [27]. For patients with ESKD, inpatient treatment for pneumonia is a common practice [26,35]. With regard to the causes of admission, admissions for pneumonia were more common among older patients; speci cally, patients who developed pneumonia before or after hemodialysis tended to be older (range: 69-70 years old). Compared with younger patients with ESKD, older patients cannot position themselves or walk on their own, which increase their risk of pneumonia during hospitalization [36]. The present study revealed a signi cant correlation between pneumonia and ESKD.
Common among patients receiving dialysis, acute nursing care and hospital admissions account for the largest proportion (approximately 40%) of the cost of ESKD treatment [7]. To enhance nursing care quality and resource use e ciency for patients with ESKD, countries should prioritize investigations into the causes or results of these patients' hospital admissions. One study revealed a positive correlation between the health status and admission rate of patients with ESKD and that the frequency and cause of admission affected their prognosis. Typically, a general patient with ESKD is admitted to hospital twice annually on average, and the two leading causes for admission are cardiovascular events and infectious diseases [3,37].
ESKD exhibits complex clinical comorbidities, and care for the disease is thus challenging. Such care involves the participation of not only a hemodialysis team but also other medical teams from various specializations (e.g., diabetes, ophthalmology, and neurology); it also involves the provision of health education on chronic kidney disease that is personalized according to the patient's comorbidities, dialysis type, and lifestyle. Nursing care for ESKD is aimed at improving the patient's quality of life, at arresting the deterioration of kidney functions, and at preventing the development of comorbidities. Therefore, such nursing care should be practiced according to the following foci: (1) improve patients' understanding of the comorbidities of kidney diseases; (2) provide personalized health education (administered by kidney health educators) in accordance with the patient's comorbidities; (3) communicate the proper health-seeking behaviors and the importance of regular follow-ups to patients to prevent patients from using over-the-counter or folk medicine and to reduce patient drop-outs from dialysis; (4) reduce the admission rate, which is realized by (4a) early contact with patients with ESKD, (4b) cross-team collaboration among medical professionals, and (4c) continual communication with caregivers and patients; (5) provide patients with the appropriate nutritional knowledge for ESKD and understand the dietary compliance of patients to foster appropriate dietary habits in patients. These nursing care foci facilitate comprehensive, continual, professional, and integrative medical care for patients to improve their quality of life.
The present study revealed correlations of comorbidities of ESRD with the duration and cause of hospital admission. It conducted comprehensive data collection on risk factors and comorbidities of ESRD and, with a retrospective design, followed up on the cause of hospital admission of patients with ESRD 1 year before they started dialysis and 2 years after dialysis. The study model involved investigating changes in 100 primary and secondary diagnoses of patients with ESKD before and after dialysis, which could aid the precise recording of patients' renal functions and prognosis. The data used in this study are of high quality because they were collected from a large-scale renal registration system [38]. The study results highlight the need to control comorbidities associated with ESKD and affect clinical and research practice. By integrating test data on patients receiving dialysis, this study provides comprehensive information on patients with ESKD. Having made a breakthrough in investigating patients receiving dialysis, this study offers valuable insights that are conducive to analyzing epidemiological and relevant risk factors among Taiwanese patients receiving dialysis; thus, it will aid in the future prevention of renal diseases, enhance the survival of these patients, and reduce the incidence of associated comorbidities. This study has several limitations. The rst lies in its retrospective design and limited sample size. In this study's retrospective analysis, patients with missing data and records were excluded. Second, the data were collected from the TSN-KiDiT, which lacked clinical indicators critical to the research topic; therefore, this study could not obtain information regarding the severity of certain diseases (e.g., blood pressure and the severities of diabetes and left ventricular hypertrophy). Additionally, other comorbidities, such as alcohol or substance dependence, have been reported to be associated with ESKD [22] , but they were not included in this study's analysis. Considering the research context of this study, the lack of disease data can result in results for a disease to be underestimated, which yields the false conclusion that the disease is unrelated to ESKD. Third, the inclusion criteria in this study were patients with ESKD who received hemodialysis and had complete medical data one year before and two years after their hemodialysis. Therefore, patients whose medical data did not conform to the criteria were excluded; this constituted another limitation of this study. Furthermore, this study analyzed only data collected by one hospital. The sample comprised only patients in the hospital who met the inclusion criteria, and some variables may have been overlooked. Because this study focused on the hemodialysis data of only one hospital, this single-center characteristic limits the generalizability of the study results to patients with ESKD with various combinations of comorbidities from other hospitals. Therefore, prospective research involving a larger patient sample size is required to verify the present study's ndings.

Conclusions
This study investigated the causes of ESKD patients' admissions to a hospital one year before and two years after they received hemodialysis. This study results show that (1) hypertension to be the most common comorbidity and (2) cardiovascular diseases to be the most in uential risk factor for patients with ESKD during hospitalization. Most patients with ESKD exhibited one or more comorbidities, and two-thirds of them developed ESKD because of diabetes and hypertension.
With regard to the causes of admission, pneumonia was more prevalent among older patients. Additionally, changes in the diagnosis codes of native atherosclerosis, urinary tract infection, pneumonia, and hyperkalemia were signi cantly correlated with age. Therefore, when administering comprehensive nursing care and treatment for ESKD, clinicians should not focus only on comorbidities but also consider factors (e.g., age) that can affect patient prognosis. No before and Yes after: no hospital admission before hemodialysis, hospital admission after hemodialysis.

Abbreviations
Yes before and No after: hospital admission before hemodialysis, no hospital admission after hemodialysis.
Yes before and Yes after: hospital admission before hemodialysis, hospital admission after hemodialysis.