We conducted a retrospective cohort study of 3073 incident CAPD patients from five PD centres in three provinces in China (The First Affiliated Hospital of Nanchang University, Nanchang, China; The First Affiliated Hospital of Zhengzhou University, Zhengzhou, China; Jiujiang No. 1 People’s Hospital, Jiujiang, China; Zhujiang Hospital of Southern Medical University, Guangzhou, China; and The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou, China), between January 1, 2005, and December 31, 2018. To maximally represent the real-world settings of the CAPD population, no patient was excluded from this study. The study was approved by the Human Ethics Committee of each research centre, consistent with the ethical principles of the Declaration of Helsinki. The data were anonymous, and the need for informed consent was therefore waived.

Data on demographics, comorbid conditions, medications, and laboratory values at the start of CAPD were abstracted from medical records by two trained investigators in each centre using uniform and standardized data collection tools: demographic characteristics (age, sex, body mass index [BMI], systolic BP, diastolic BP, 24-hour urine volume, current smoking, and current alcohol consumption); comorbidities (diabetes mellitus, preexisting CVD, hypertension, and hyperlipidaemia); underlying causes of ESRD; medications (calcium channel blockers, beta-blockers, angiotensin-converting enzyme inhibitors/angiotensin II receptor blockers [ACEIs/ARBs], diuretics, statins, and aspirin); and laboratory variables (haemoglobin, serum albumin, serum uric acid, estimated glomerular filtration rate [eGFR], residual renal function [RRF], cholesterol, triglyceride, high-density lipoprotein, low-density lipoprotein, and high-sensitivity C-reactive protein [hs-CRP]). Hypertension was defined as systolic BP $>$140 mmHg or diastolic BP $>$90 mmHg or the use of antihypertensive medications according to the 2016 Guidelines for the Management of Renal Hypertension in China [11]. The presence of CVD was defined as coronary heart disease, congestive heart failure, arrhythmias, cerebrovascular disease, or peripheral vascular disease [12]. Current smoking was defined as at least one cigarette a day, and current alcohol consumption was defined as $>$20 grams of ethanol a day [13]. The estimated glomerular filtration rate was calculated using the Chronic Kidney Disease Epidemiology Collaboration equation [14].

The primary and secondary outcomes were all-cause and CVD mortality, respectively. If the patients died in any hospital, the exact cause of death was available by death certificates, and if the patients died outside a hospital, experts would reach a consensus on the cause of death, with a comprehensive consideration of current health conditions provided by family members and the medical history and descriptions from each dialysis centre. All participants had conducted CAPD schedules produced by dialysis professionals following the International Standardized Peritoneal Dialysis Guidelines [15] and the patient’s health conditions. All patients were followed up until CAPD cessation, death, the end of an 8-year period, or June 30, 2019. Transferring to haemodialysis, renal transplantation, transferring to other centres, loss of follow-up, and survival after a follow-up period of 8 years or through June 30, 2019 were considered to be censored.

Variables with missing data before the data analysis were imputed using the missForest method, which handles different types of variables [16]. Incidence was calculated as the number of events divided by the total valid observational time at risk, scaled to episodes per 1000 years. Variables are presented as the mean $\pm$ standard deviation (SD), median (interquartile range, IQR) or number (%). Patients were divided into four groups: the control group (those without preexisting CVD or hypertension), CVD group (preexisting CVD patients), hypertension group, and CVD plus hypertension group (preexisting CVD patients coexisting with hypertension). Baseline variables were compared by one-way ANOVA or Kruskal‒Wallis tests according to variable distribution (normality tested with the Shapiro‒Wilk test) for quantitative variables and the chi-square test when appropriate for categorical variables among the groups.

We used Kaplan‒Meier curves to investigate the difference in cumulative mortality among the four groups over the observational period. To analyse the association between these interesting comorbidities and mortality, we constructed four Cox proportional hazards regression models adjusted for the following factors: Model 1, unadjusted; Model 2, Model 1 plus age, sex, body mass index, systolic BP, current smoking, current alcohol consumption, diabetes mellitus hyperlipidaemia, and underlying causes of ESRD; Model 3, Model 2 plus medications; and Model 4, Model 3 plus haemoglobin, serum albumin, serum uric acid, RRF, cholesterol, and hs-CRP. In addition, the association was also analysed among subgroups of men, women, diabetes mellitus, nondiabetes mellitus, hyperlipidaemia, and nonhyperlipidaemia. We tested for interactions of sex, diabetes mellitus, and hyperlipidaemia.

For all-cause mortality, haemodialysis, renal transplants, loss of follow-up, or transferring to other centres were considered competing risks. When using these competing risks, we evaluated the association between these interesting comorbidities and all-cause mortality using four Fine–Gray competing risk models. Similarly, for CVD mortality, non-CVD mortality, haemodialysis, renal transplants, loss of follow-up, or transfer to other centres were considered competing risks. Second, for adult patients with a short-term period of follow-up, interesting outcomes may not be completely observed, with underreporting of the incidence of mortality. To fully observe outcomes, we further analysed the effect of comorbidities at the start of dialysis on mortality in adult patients with at least a 24-month period of follow-up.

The results of the Cox proportional hazards models and Fine–Gray models are presented as the hazard ratio (HR) and the 95% confidence interval (CI). Statistical analyses were conducted using Stata 15.1 statistical software (StataCorp, College Station, TX, USA). The level of significance was set as 0.05 for all analyses.

All 3073 incident CAPD patients from five dialysis centres were included in the present study. All variables with less than 5% missing data were imputed before the data analysis, and there were no missing data for outcomes. Of 3073 patients with a median age of 49.0 (IQR 39.0–61.0), 1780 (57.9%) were men, 1986 (64.7%) had hypertension, 430 (13.9%) had preexisting CVD, and 567 (18.4%) had diabetes mellitus. Compared with the control group, the CVD plus hypertension group tended to be elderly, with higher BMI, systolic BP, haemoglobin, and cholesterol, as well as being more likely to be currently smoking; have diabetes mellitus, hyperlipidaemia, diabetic nephropathy, or hypertensive nephropathy; and taking calcium channel blockers, beta-blockers, diuretics, ACEIs/ARBs, aspirin, and statins but have lower diastolic BP. Compared with patients with hypertension, those with prior CVD were more likely to be older and female, have hyperlipidaemia, and have primary glomerulonephritis and be less likely to be taking calcium channel blockers and ACEIs/ARBs (Table 1).

The median observational period was 33.7 (IQR 15.7–60.9) months. During this period, 571 (18.6%) patients died, with 286 (9.3%) CVD deaths, 59 (1.9%) infection deaths, 10 (0.3%) gastrointestinal bleeding deaths, 17 (0.6%) tumour deaths, 101 (3.3%) other causes of death, and 89 (2.9%) unknown causes of death. In addition, 375 (12.2%) transferred to haemodialysis, 159 (5.2%) received renal transplants, 26 (0.8%) transferred to other dialysis centres, and 106 (3.4%) were lost to follow-up. The number of all-cause mortalities was 143 (38.6%), 15 (25.0%), 293 (18.1%), and 120 (11.7%) in the CVD plus hypertension, CVD, hypertension, and control groups, respectively. The number of CVD mortalities was 76 (20.5%), 7 (11.7%), 147 (9.1%), and 56 (5.5%) in the CVD plus hypertension, CVD, hypertension, and control groups, respectively.

The incidence of all-cause mortality was 55.7/1000 patient-years in the study population, with 27.9/1000 patient-years of CVD mortality incidence (Table 2). The incidence of all-cause mortality was 131.0, 74.4, 56.1, and 32.2/1000 patient-years, and CVD mortality incidence was 69.6, 34.7, 28.2, and 15.0/1000 patient-years among the CVD plus hypertension, CVD, hypertension, and control groups, respectively.

The survival analysis found that the CVD plus hypertension group had greater cumulative all-cause mortality (p 0.001) and CVD mortality (p 0.001) than the control group (Fig. 1). The association between comorbidities and mortality was evaluated by the different Cox proportional hazards regression models (Table 3). When compared with the control group, the CVD plus hypertension, CVD, and hypertension groups had a 3.97-fold (95% CI: 3.06 to 5.15), 2.21-fold (95% CI: 1.29 to 3.79), and 1.83-fold (95% CI: 1.47 to 2.29) higher risk of all-cause mortality and 4.68-fold (95% CI: 3.27 to 6.69), 2.10-fold (95% CI: 0.95 to 4.62), and 1.86-fold (95% CI: 1.36 to 2.54) higher risk for CVD mortality compared with the control group in Model 4, respectively. Similar trends were observed among the subgroups of men, women, diabetes mellitus, nondiabetes mellitus, hyperlipidaemia, and nonhyperlipidaemia (Fig. 2). There was no significant interaction between these subgroup variables and preexisting CVD coexisting with hypertension on all-cause and CVD mortality in the study population. The p values for interactions were $>$0.05 for all subgroups, suggesting that the increased risk of all-cause and cardiovascular mortality associated with interesting comorbidities was evident regardless of these subgroup variables.

Fig. 1.

Cumulative incidence of all-cause and CVD mortality. (A) presented the cumulative incidence of all-cause mortality, and (B) presented the cumulative incidence of CVD mortality. Cumulative survival was lowest in those with hypertension and prior CVD. Control group, patients without hypertension and prior CVD. Group 0, control group; Group 1, hypertension group; Group 2, prior CVD group; Group 3, hypertension plus prior CVD group. CVD, cardiovascular disease.

Fig. 2.

Adjusted HRs for all-cause and CVD mortality among subgroups. HRs were adjusted for variables in Model 4, except for the variable of the subgroup. Control group, patients without hypertension and prior CVD. HTN, hypertension; DM, diabetes mellitus; CVD, cardiovascular disease; HR, hazard ratio; CI, confidence interval.

Compared with patients with hypertension (reference), those with prior CVD had a higher risk of all-cause (HR: 1.38, 95% CI: 1.09 to 2.29) and CVD (HR: 1.59, 95% CI: 1.07 to 2.30) mortality (Table 4).

The joint associations showed similar patterns using the Fine–Gray competing risk models (Table 5). The CVD plus hypertension, CVD, and hypertension groups had a 4.15-fold (95% CI: 3.15 to 5.49), 1.99-fold (95% CI: 1.07 to 3.70), and 1.57-fold (95% CI: 1.25 to 1.98) higher risk of all-cause mortality than the control group, respectively. Similarly, compared with the control group, the CVD plus hypertension, CVD, and hypertension groups had a 2.90-fold (95% CI: 1.92 to 4.38), 1.85-fold (95% CI: 0.82 to 4.16), and 1.30-fold (95% CI: 1.08 to 1.79) higher risk of CVD mortality, respectively.

A total of 42 (1.4%) patients aged 18 years at the start of dialysis were excluded, with six deaths at the end of the study. By the end of the study, 810 (26.3%) adult patients were followed up for less than 24 months, and 282 (9.2%) adult patients survived for less than 24 months. The remaining 1939 (63.1%) adult patients were followed up for at least 24 months, with 283 (14.6%) all-cause mortality and 135 (7.0%) CVD mortality. We found that compared with the control group, the CVD plus hypertension, CVD, and hypertension groups had a 3.92-fold (95% CI: 2.76 to 5.57), 1.99-fold (95% CI: 1.47 to 2.71), and 1.58-fold (95% CI: 1.17 to 2.13) higher risk of all-cause mortality and 4.47-fold (95% CI: 2.68 to 7.46), 1.78-fold (95% CI: 0.54 to 5.90), and 1.48-fold (95% CI: 1.28 to 1.71) higher risk of CVD mortality in Cox regression Model 4, respectively, among adult adults with at least a 24-month follow-up period (data not shown).