Assisted peritoneal dialysis: a feasible KRT modality for frail older patients with end-stage kidney disease (ESKD)

Assisted PD is used as an alternative option for the growing group of frail, older ESKD patients unable to perform their own PD. This study was undertaken to investigate the outcomes of assisted PD in older patients by comparing assisted PD patients with self-care PD patients. This study included all patients aged 70 and above who started on PD in our hospital from 2009 to 2018. Patients were followed up until death, PD cessation or to the end of the study (December 31, 2019). Risk factors associated with mortality, peritonitis and technique failure were evaluated using both cause-specific hazards and subdistribution hazards models. 180 patients were enrolled, including 106 (58.9%) males with a median age of 77.5 (77.2–81.2) years. Among the 180 patients, 62 patients (34.4%) were assisted. Patients on assisted PD group were older, more likely to be female, more prevalent in DM and CVD, with a higher Charlson score than patients undergoing self-care PD (P all < 0.05). In the multivariable analysis, assisted patients had a comparable patient survival and peritonitis-free survival compared to self-care PD patients either in the Cox or in the FG models. According to a Cox model, the use of assisted PD was associated with a lower risk of technique failure (cs-HR 0.20, 95% CI 0.04–0.76), but the association lost its statistical significance in the Fine and Gray model. Our results suggest that assisted PD could be a safe and effective KRT modality for older ESKD patients who need assistance.

patients aged 70 and above who started on PD between 1 January 2009 and 31 December 2018 at Renji Hospital, Shanghai Jiao Tong University School of Medicine, China, were screened for eligibility. Patients had history of maintenance HD/transplantation, withdrew from PD within 3 months or with incomplete data were excluded from the study. All enrolled patients were dialyzed using lactate-buffered glucose-based PD solutions (Dianeal ® , Baxter) with twin-bag system. Patients and their caregivers had received standard training after catheterization by PD dedicated nurses 16 . The study was approved by the Human Research Ethics Committee of Renji Hospital, Shanghai Jiao Tong University School of Medicine. All individual information was securely protected and was made available to only the investigators.
Demographic and laboratory data. The demographic characteristics collected at baseline included age, gender, height, weight, underlying cause of ESKD and comorbid condition status such as diabetes mellitus (DM) and cardiovascular disease (CVD). Hypertension and diabetes were defined either as a comorbid disease or as the etiology of ESKD. CVD was defined as a previous history of any following condition: acute coronary syndrome, heart failure, cerebral infarction or hemorrhage, coronary artery atherosclerosis confirmed by percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) therapy. The Charlson comorbidity index was adopted to reflect the burden of comorbid conditions. Body mass index (BMI) was calculated as the weight (kg) divided by the square of height in meters (BMI = weight [kg]/height [m 2 ]).
Small solute clearance and peritoneal transport characteristics. All patients were evaluated small solute clearance and performed a standard peritoneal equilibration test (PET) 1-3 months after PD initiation. Small solute clearance was assessed by 24-h dialysate and urine collection, with the calculation of total weekly Kt/V and weekly CrCl normalized to 1.73 m 2 body surface area 17 . Residual renal function (RRF) was calculated as an average of 24-h urine urea and creatinine clearance 18 . Normalized protein catabolic rate (nPCR) was calculated by the methods described by Randerson, Chapman, and Farrell and normalized to standard body weight (total body water/0.58) 19 . Methodology. The enrolled patients were divided into assisted PD group (PD exchanges performed by a family member or a domestic helper) and self-care PD group according to the independence of bag exchange, and prospectively followed up until death, transfer to permanent hemodialysis, recovery of renal function, transfer to other centers, lost to follow-up or to the end of study (December 31st, 2019). All deaths, switches to HD and peritonitis episodes during the study period were carefully tracked and recorded. Detailed causes of death, switches to HD and outcome of peritonitis during PD were also collected. Causes of death were grouped in broad categories as follows: cardiovascular, including cardiac, cerebrovascular, peripheral vascular and sudden death; infection, including peritonitis and non-peritonitis infections; cancer; gastrointestinal hemorrhage; other and unknown causes. Causes of switch to HD were grouped into peritonitis; catheter complications; inadequate dialysis and other causes. Peritonitis was diagnosed and managed in accordance with guidelines of the International Society for Peritoneal Dialysis 20 , and peritonitis rate was calculated as number of peritonitis episodes per patient-year at risk.
Outcome measures. Outcome measures in our study included patient survival, peritonitis-free survival and technique survival. In patient and peritonitis-free survival analysis, the endpoint was death and first episode of peritonitis, respectively. In technique survival analysis, the endpoint was permanent transfer from PD to HD. For both patient and peritonitis-free survival analysis, the censored events were transfer to permanent hemodialysis, recovery of renal function, loss to follow-up, transfer to other dialysis centers, or to the end of study (December 31st, 2019). In technique survival analysis, the endpoint was permanent transfer from PD to HD, and death was regarded as censored event.
Statistics analysis. The Kolmogorov-Smirnov test was used to measure data normality. Parametric data were presented as mean ± standard deviation. Nonparametric data were described by the median value (first and third quartile). Categorical variables were presented by frequencies and percentages and were compared using chi-square tests. Normally distributed continuous variables and abnormally distributed continuous variables were compared using the independent sample t-tests and Mann-Whitney test, respectively. Kaplan-Meier and log-rank test methods were used to estimate and compare survival curves for each event of interest (death, peritonitis and transfer to HD) by comparing assisted PD group with self-care PD group. Considering the presence of competing events in this study, for multivariate analysis, risk factors for all-cause mortality, peritonitis and technique failure were evaluated by both cause-specific hazards and subdistribution hazards models 21 . When the event of interest was peritonitis, transfer to HD, renal transplantation, death and transfer to other centers were coded as competing events only when occurring before the first peritoneal infection. When the event of interest was death, the competing events included transfer to HD, renal transplantation and transfer to other centers. When the event of interest was technique failure, the competing events included death, renal transplantation www.nature.com/scientificreports/ and transfer to other centers. Demographic characteristics and important recognized risk factors that might be associated with outcomes (all-cause mortality, technical failure, and peritonitis) were first selected for univariate analysis. Variables with a P value < 0.05 in univariate analysis and important demographic characteristics were later entered into multivariate analysis except those with multicollinearity. Data analysis was carried out using the SPSS software package (version 22.0: SPSS, Chicago, IL, USA) and R 3.6.1 (R Foundation for Statistical Computing, Vienna, Austria; the 'cmprsk' library was used to fit the Fine and Gray regression models). All probabilities were two-tailed, and a P < 0.05 was considered statistically significant.
Ethics approval and consent to participate. All

Results
Study participants. A total of 180 patients were included in present study. Patient enrollment and followup were presented in Fig. 1. Patient characteristics were summarized in Table 1. Among the 180 patients, 62 needed assistance in performing bag exchanges ("assisted PD group"), and the remaining 118 patients were in the self-care PD group. Patients in the assisted PD group were older (80.7 (76.9-84.0) vs 75.6 (72.5-79.2) years, P < 0.001), less likely to be male (48.4% vs 64.4%, P < 0.05), more prevalent in diabetics (48.4% vs 33.1%, P < 0.05) and CVD (46.8% vs 29.7%, P < 0.05), with a higher Charlson score (7.0 (6.0-8.0) vs 6.0 (5.0-7.0), P < 0.001) than those in the self-care PD group, and other demographic and laboratory data were similar between the two groups.
The indices of small solute clearance, RRF, nPCR and peritoneal transport characteristics (D/Pcr) were shown in Table 2 and there was no difference between the two groups. Table 3. The median follow-up was 32.5 months (inter-quartile range, 20.7-43.7 months) for the assisted PD group and 33 months (inter-quartile range, 12.9-49.7 months) for the self-care PD group. By the end of the study, 100 (55.6%) patients died, 16 (8.9%) patients switched to HD, 6 (3.3%) patients were transferred to other centers, 1 (0.6%) patient was lost to follow-up, 1 (0.6%) patient was dialysis-independent and 54 (30.0%) patients were still on PD. The causes of death were similar in two groups and the leading cause of death was cardiovascular disease (32.0%), followed by infec-   Table 4) were independent predictors for all-cause mortality. However, for both models, the use of assisted PD was not associated with all-cause mortality. Fig. 2B, assisted PD patients had comparable peritonitis-free survival to self-care PD patients (Log-rank X 2 = 0.048, P = 0.827). In both Cox and Table 3. Outcomes of the patients. Failed treatment for peritonitis* was defined as discontinuation of PD including temporary or permanent transfer to hemodialysis or peritonitis-related deaths; Peritonitis-related deaths included death directly caused by active peritonitis or within 4 weeks of a peritonitis episode, or any death during hospitalization for peritonitis.   www.nature.com/scientificreports/ Fine-Gray (FG) models, there was no association between the use of assisted PD and peritonitis-free survival, and no variables were found to be significantly associated with peritonitis-free survival (P > 0.05). Fig. 2C, assisted PD patients had comparable technique survival to self-care PD patients (Log-rank X 2 = 1.888, P = 0.169). In the multivariable analysis, assisted PD (cs-HR 0.20, 95% CI 0.04-0.76, Table 4) was protective against the risk of transfer to HD in the Cox model, while advanced age (cs-HR 1.15, 95% CI 1.02-1.31, Table 4) and higher BMI (cs-HR 1.31, 95% CI 1.11-1.55, Table 4) were associated with an increased risk of technique failure. However, in the Fine-Gray (FG) model, higher BMI (sd-HR 1.25, 95% CI 1.02-1.54) was the only predictor that was associated with technique survival, while assisted PD was not associated with technique survival in this population.

Discussion
The present study compared the outcomes between assisted PD patients and self-care PD patients aged 70 and above to investigate the safety and effectiveness of assisted PD in older patients. The results showed that in our cohort, assisted PD patients had a comparable patient survival and peritonitis-free survival to self-care PD patients. Moreover, assisted PD might protect older patients incapable of self-care from technique failure. The demographic and clinical characteristics of the study cohort varied between the assisted group and selfcare group. Patients in the assisted PD group were older, more likely to be female, more prevalent in diabetics and cardiovascular disease and carried a heavier burden of comorbid diseases than patients in the self-care PD group. Similar to our study, Boyer et al. showed that patients starting PD with assistance were older than those starting unassisted (70.0 (61.5-78.3) vs 58.7 (43.8-69.2) years) 12 . In another study from France, Lobbedez et al. reported that assisted PD patients were older (74 ± 10.4 vs 52 ± 18.6 years, P < 0.001) and presented more comorbidity (CCI 7 ± 2.5 vs 4.3 ± 2.4, P < 0.05) compared with self-care patients 22 . These findings indicated that patients requiring assistance were often frail and older individuals, with physical disability or cognitive impairment, and had multiple comorbidities.
The causes of death were similar in assisted PD group and self-care PD group. It is well documented that cardiovascular disease is the most common cause of deaths in PD patients 23,24 . In our study, cardiovascular disease remained the leading cause of death in older PD patients, accounted for up to 32.0% of deaths. However, we found that infection was also a major cause of death, accounted for up to 26.0% of deaths, and the majority of which was due to non-peritonitis infections, most being pulmonary infection. Our finding indicated that older PD patients were prone to non-peritonitis infection, this might be a result of a high prevalence of DM, physical disabilities, poor nutrition and immunodeficiency. In concordance with our study, an analysis of elderly PD patients aged 70 and above found that infection constituted 26.6% of the causes of death 6 . Results of another study with a median age of 73  years showed that 37.4% of PD patients died of infection, mainly pulmonary infection 25 . Therefore, aggressive prevention and treatment of infection is essential for older PD patients. In patient survival, we found that assisted PD patients had similar survival rate compared to self-care PD patients. In concordance with our study, Smyth et al. reported that there was no difference in patient survival rates between assisted PD patients and self-care PD patients 17 . Querido et al. also found that assisted PD patients had similar survival rate compared to self-care PD patients 26 . However, in contrast with our results, some studies reported poorer survival rate was observed in assisted PD patients compared to self-care PD patients. Data from the French Peritoneal Dialysis Registry (RDPLF) for 1613 patients older than 75 years of age showed that the survival rate of assisted PD patients, whether assisted by family members or nurses, was lower than patients on self-care PD 27 . The potential causes for the differences in patient survival may be due to the fact that assisted PD in our cohort was provided by one trained dedicated person (e.g., spouse), so the training and daily assistance could be detailed and tailored, and caregivers were more aware of the condition of the patients. However, in the report from the RDPLF, patients were assisted by private community nurses and it is not patient-specific. Besides, several studies have demonstrated that family and social support is associated with improved outcomes in chronic conditions, including end-stage kidney disease (ESKD) [28][29][30] . In China, spouses and the younger generations are encouraged to take care of older PD patients. As PD exchanges were performed by their family members or domestic helper at home, patients have a high level of family support, which may be associated with better patient management and improved survival. Another retrospective PD study with patients over 65 years of age in Taiwan also suggested that older patients on assisted PD had a poorer patient survival rate than self-care PD patients 31 . As the author mentioned in discussion, the possible explanation may be that the assisted-care program for older patients was adopted as early as 1984 in Taiwan, the quality of the training system, which might determine the outcome of assisted PD, was worse than it is now. In the present study, we also identified that advanced age, comorbid with CVD and lower RRF were independent predictors for mortality, which were well-recognized prognostic factors for mortality in older PD patients demonstrated by numerous studies [32][33][34][35][36] .
The peritonitis rate was 0.155 episode per patient-year in the assisted PD group and 0.216 episode per patientyear in the self-PD group, respectively. In our cohort, peritonitis-free survival was comparable between assisted patients and self-care patients. Similarly, Xu et al. 24 reported that assisted PD patients overall had a similar peritonitis-free time compared with self-care PD patients. Smyth et al. 17 reported that there was no association between the use of assisted PD and peritonitis-free survival. In another report from the RDPLF, Benabed et al. 19 showed that in 3598 diabetic patients between 1 January 2002 and 31 December 2012, nurse-assisted PD patients had a lower risk of peritonitis compared with self-care PD patients while family-assisted PD had no protective effect against peritoneal infection. Verger et al. reported that nurse assistance was associated with a higher risk of peritonitis in APD patients, however, when home visits were made regularly by nurses from the PD center, assisted PD was not associated with a higher risk of peritoneal infection 18 . Taken together, these results demonstrated that the use of assisted PD was not associated with peritonitis-free survival. www.nature.com/scientificreports/ With regard to technique survival, a significant technique survival benefit was demonstrated in assisted patients compared to self-care patients in the Cox model, but the association lost its statistical significance in the Fine-Gray (FG) model. Consistent with our results, report from the RDPLF which analyzed 9822 incident patients starting PD between January 2002 and December 2010 suggested that assisted patients had a lower risk for transfer to HD compared with self-care patients 37 . Querido et al. also found that technique survival was better in assisted PD patients compared with self-care patients 26 . As older patients who engaged independently in PD usually suffer from poor physical strength, cognitive dysfunction, vision impairment and deafness, which are all conditions that may affect the PD procedure, we suggested that for some frail older patients unable to perform ideal self-dialysis, proper assistance should be provided to reduce the risk of PD technique failure, thereby prolong technique survival. Besides, in concordance with previous studies, higher BMI could predict technique failure in this population, which was independent predictor of technique failure reported by several studies 33,38 .
Our study also has several limitations. First, it was a retrospective design. Second, our study was a singlecentered study. Third, we did not collect the data regarding the quality of life (QoL) in our study, which is an important outcome measure in older patients. From the perspective of gaining high-quality evidence, better designed studies, such as prospective studies with larger sample sizes and multi-center participation, is clearly warranted.
In conclusion, our results showed that in a cohort of patients aged 70 and above, assisted PD patients had comparable patient survival and peritonitis-free survival to self-care PD patients. Moreover, assisted PD might protect older patients incapable of self-care from technique failure. Therefore, we suggested that poor self-care ability alone should not be used as a barrier to PD treatment and assisted PD could be a safe and effective modality of KRT for older patients incapable of self-care.

Data availability
The datasets analyzed during the current study are available from the corresponding author on reasonable request.