All-Cause Standardized Mortality Ratio in Hemodialysis and Peritoneal Dialysis Patients: A Nationwide Population-Based Cohort Study

Patients with end-stage renal disease (ESRD) are at a higher mortality risk compared with the general population. Previous studies have described a relationship between mortality and patients with ESRD, but the data on standardized mortality ratio (SMR) corresponding to different causes of death in patients undergoing hemodialysis (HD) and peritoneal dialysis (PD) are limited. This study was designed as a nationwide population-based retrospective cohort study. Incident dialysis patients between January 2000 and December 2015 in Taiwan were included. Using data acquired from the Taiwan Death Registry, SMR values were calculated and compared with the overall survival. The results showed there were a total of 128,966 patients enrolled, including 117,376 incident HD patients and 11,590 incident PD patients. It was found that 75,297 patients (58.4%) died during the period of 2000–2017. The overall SMR was 5.21. The neoplasms SMR was 2.11; the endocrine, nutritional, metabolic, and immunity disorders SMR was 13.53; the circulatory system SMR was 4.31; the respiratory system SMR was 2.59; the digestive system SMR was 6.1; and the genitourinary system SMR was 27.22. Therefore, more attention should be paid to these diseases in clinical care.


Introduction
According to the International Society of Nephrology 2019 Global Kidney Health Atlas cross-sectional survey, the global average number of newly diagnosed cases of end-stage renal disease (ESRD) is 144 per million general population, and the incidence of ESRD varies greatly among different countries. Hemodialysis (HD) is the most common form of renal replacement therapy [1]. In most countries, ≥80% of chronic dialysis patients receive HD [2].
Patients with ESRD are at a higher mortality risk compared with the general population. Although previous studies have described a relationship between mortality and patients with ESRD, these studies had some limitations . First, some studies consisted of small sample sizes. Second, peritoneal hemodialysis (PD) patients usually had better baseline characteristics than HD patients; however, some studies did not separate these two groups. Third, many studies had short follow-up periods. Fourth, some studies also lacked general population data as the comparison group. Fifth, some studies only compared the overall mortality rate without including the specific cause of death. Most importantly, most of these studies investigated the risk of mortality. Only a few studies investigated the standardized mortality ratio (SMR) in a large representative cohort compared with the general population [37][38][39][40]. In addition, the study periods of the above studies were different from our study, and dialysis technologies have changed in recent years.
Therefore, the objectives of this study were to describe the causes of death and determine the all-cause SMR for a nationwide cohort of patients on dialysis.

Study Design
This study was designed as a population-based retrospective cohort study. HD patients and PD patients during a specific period (1 January 2000, to 31 December 2015) were included in the study. We then monitored the clinical outcomes (deaths) in these groups over time until the end of the study (31 December 2017).

Data Source
Taiwan National Health Insurance is a compulsory social health insurance plan that started in 1995. Approximately 99% of Taiwan's 23 million people participate in the program. The National Health Insurance Research Database (NHIRD) of Taiwan has been used by many parties for epidemiological research and research on prescription drug use. The accuracy of the disease diagnosis recorded in NHIRD has been verified, and the recorded data are of high quality [41][42][43][44][45][46][47][48][49][50][51].
All NHIRD data sets can be externally linked to the Taiwan Death Registry (TDR). In Taiwan, the law requires that all deaths must be registered within 10 days. TDR is considered accurate and complete because registering a death in Taiwan is necessary for doctors to issue a death certificate [52].

Dialysis Cohort
All disease diagnosis codes are assigned according to the ninth and tenth editions of the International Classification of Diseases, Clinical Modification (ICD-9-CM or ICD-10-CM). First, patients with ESRD (who have catastrophic illness cards for ESRD, ICD-9-CM code 585 or ICD-10-CM code N18.6) who started receiving dialysis between 1 January 2000, and 31 December 2015, were included in the dialysis cohort. In Taiwan, if a patient is diagnosed with a disease classified as a catastrophic illness by the Ministry of Health and Welfare, the patient can submit relevant disease information and apply for a catastrophic illness certificate [53]. Patients with catastrophic illness certificates do not need to pay deductibles for outpatient or inpatient treatment related to this disease during the validity period of the certificate. For example, a patient with a catastrophic illness certificate of ESRD does not need to pay the deductible for dialysis treatment. These dialysis patients were then divided into PD and HD patients according to their dialysis modality. Patients who had received both PD and HD were classified as HD group if the HD duration was longer than PD in the first six months, and vice versa for the PD group. Patients who received dialysis for less than six months were excluded. Follow-ups on all individuals continued until death, renal transplantation, a change of dialysis modality, or the end of the study (2017).

Definition of Study Outcomes
For each cohort participant, the TDR data link was contacted to determine any deaths of cohort members. The TDR has kept records of all death certificates in Taiwan since 1971. The TDR provides cause-specific mortality data, classified by the International Classification of Diseases (ICD-9 and ICD-10) (Supplementary Table S1). The observation period of this study was from 2000 to 2017.

Reference Population
Mortality data of the entire country's general population of Taiwan were used for reference. Data were also obtained from TDR data.

Validation
We validated our method of identifying ESRD (catastrophic illness card for ESRD, ICD-9-CM code 585 or ICD-10-CM code N18.6) by analyzing the medical records (charts) of 100 patients in E-DA Hospital, a teaching hospital with 1100 beds in Taiwan. We randomly selected 50 patients with ESRD major illness registration cards from the patient claims database from 2008 to 2010 and 2015 to 2016. The positive predictive value of ESRD was estimated. The results showed that the positive predictive value of ESRD was 100%.

Statistical Analysis
To examine whether differences in mortality between dialysis (PD and HD) and the general population were present, we calculated the overall SMR and determined the underlying causes of death in these patients. The expected number of deaths was calculated according to the average death incidence rate of the general population from 2000 to 2017, standardized for sex and age, and then multiplied by the person-years of the HD or PD cohort. SMR confidence intervals were calculated using Byar's method [54].

Ethics Statement
The access to the research data has been reviewed and approved by the National Institutes of Health Review Board.
The study was approved by the ethics committee/Institutional Review Board of E-DA Hospital (IRB number: EMRP-108-061).

Descriptive Statistics
There were 128,966 patients enrolled in this study, including 117,376 incident HD patients and 11,590 incident PD patients. Table 1 shows the demographic characteristics and clinical comorbidities of the HD and PD cohorts.

Follow-Up
During the follow-ups, 5516 patients (4.2%) were censored because of kidney transplantation, and 75,297 (58.4%) patients (38,699 men and 36,598 women) died. No causes of death were missing or unknown.
To the best of our knowledge, there are only a limited number of studies investigating the SMR in dialysis patients [37][38][39][40]. Villar  Except for genitourinary system diseases, our results showed that cardiovascular disease is the principal cause of mortality in patients with ESRD. One reason is that patients with ESRD have high comorbidity rates, including diabetes mellitus, hypertension, and hyperlipidemia, and these comorbidities are all cardiovascular disease risk factors. In addition to these traditional cardiovascular risk factors, there were non-traditional risk factors in dialysis patients, such as chronic volume overload, anemia, endothelial dysfunction, hyperparathyroidism, inflammation, malnutrition, and uremic toxin [55]. All these factors promote a high cardiovascular disease mortality rate in dialysis patients.
The reasons for a high neoplasms mortality rate are possible as follows: First, antioxidant capacity decreases in dialysis patients, which may lead to deoxyribonucleic acid damage because of the increase in reactive oxygen species [56]. Second, the increased production of cytokines during dialysis due to the bio-incompatibility of the dialysis membrane or dialysate has also been suggested to predispose to neoplasm [57]. Third, patients with ESRD are at risk for the accumulation of carcinogenic agents due to reduced renal elimination. Finally, the chronic inflammation status of dialysis patients may act together to accelerate neoplasm formation [58].
This study had some limitations. First, we did not have the personal information of enrolled cohorts, such as smoking history, family history, or laboratory parameters, which may be associated with specific causes of death. Second, misclassification of diseases may occur in an administrative database study. However, in our study, we included only patients with ESRD. The diagnoses of patients with ESRD are reliable because the catastrophic illness card for ESRD needs a formal review to confirm the diagnoses. Furthermore, longterm dialysis treatments are needed for patients with ESRD. Patients with ESRD can be exempted from copayment, and all these patients apply for catastrophic illness cards for ESRD. Most importantly, in our validation study, the results showed that the coding of PD and HD patients is credible. Third, misclassification of mortality causes may occur in TDR because most mortality causes depend on clinical diagnosis. Fourth, although obtained in a nationwide population-based cohort with clear SMR corresponding to different causes of death, our study findings do not add much to current knowledge, and the novelty may be limited.
In summary, this large nationwide population-based study showed that the overall SMR of dialysis patients was 5.21 compared with the general population. Except for genitourinary system diseases, the most common causes of death in dialysis patients were circulatory, endocrine/metabolic, and neoplasms diseases. Therefore, more attention should be paid to these diseases in clinical care.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/ijerph20032347/s1 Table S1: International Classification of Diseases Code; Table S2: All-cause standardized mortality ratio (SMR) of peritoneal dialysis (PD) stratified by gender; Table S3: All-cause standardized mortality ratio (SMR) of hemodialysis (HD) stratified by gender; Table S4: All-cause standardized mortality ratio (SMR) of peritoneal dialysis (PD) stratified by age; Table S5: All-cause standardized mortality ratio (SMR) of hemodialysis (HD) stratified by age. Funding: This was not an industry-supported study. This study was supported by EDAD2001, NCKUEDA11101, and EDAHP111023 from the Research Foundation of E-DA Hospital and National Cheng Kung University Hospital, Taiwan.

Institutional Review Board Statement:
The study was approved by the ethics committee/Institutional Review Board of E-DA Hospital (IRB number: EMRP-108-061).
Informed Consent Statement: Informed consent is waived by the ethics committee. The administrative data is held by governments; the data are used in compliance with local regulatory and legal frameworks that govern data use. The access to the research data has been reviewed and approved by the National Institutes of Health Review Board.

Data Availability Statement:
The data that support the findings of this study are available from National Health Insurance Research Database, but restrictions may apply to the availability of these data. However, processed datasets can be requested and made available from the authors with the permission of National Health Insurance Ad-ministration and Ministry of Health and Welfare.