Survival Outcomes and Mortality Risk Factors in Peritoneal Dialysis

Introduction

Kidney failure presents a growing health burden in India, driven by increasing rates of diabetes, hypertension, and an aging population.¹ Continuous Ambulatory Peritoneal Dialysis (CAPD) offers a viable renal replacement therapy, especially in resource-limited settings, due to its flexibility and lower dependence on healthcare infrastructure.² Despite its potential, peritoneal dialysis (PD) remains underutilized in India, and data on long-term outcomes and mortality risk factors among patients are limited.^3,4

Global studies have identified predictors of mortality in PD patients, such as age, diabetes, cardiovascular disease, nutritional status, and peritonitis, but these findings may not fully translate to the Indian context due to demographic and healthcare disparities.^3,5 Region-specific evidence is needed to guide clinical management and improve patient outcomes.

This retrospective study aims to assess survival and identify key predictors of mortality among PD patients at a tertiary care center, offering insights into the effectiveness of PD and highlighting potential targets for intervention in similar settings.

Materials and Methods

This study was a single-center, retrospective cohort analysis conducted at Christian Medical College Vellore, a prominent teaching hospital in Southern India. The study included all adult patients (≥18 years) who began PD between January 1, 2013, and December 31, 2023. This study was approved by the institutional ethics committee, which granted a waiver of informed consent due to retrospective nature of the study. Hence patient consent was not obtained. Patients who started PD for non-renal reasons were excluded from the analysis.

At our center, most PD catheters were inserted percutaneously by nephrologists trained in catheter placement, using a sterile operating room environment. The rest underwent open surgical insertion, typically due to prior abdominal surgery or anatomical considerations. A standard 2-week break-in period was observed before initiating full-volume exchanges. Early initiation was infrequent and not analyzed separately due to small numbers.

Demographic, clinical, and dialysis-related data at dialysis initiation were collected from electronic patient records. These included age, sex, circulating blood-borne viruses (HIV, hepatitis B virus, and hepatitis C virus), BMI, hypoalbuminemia, primary kidney disease, reason for opting PD, comorbid conditions, dialysis vintage and kidney transplant history, financial support, and history of PD peritonitis. Financial support encompassed all sources, including private insurance, government health schemes, and employer-based coverage. BMI was divided into four groups according to the World Health Organization Asian BMI classification: BMI <18.5 kg/m² (underweight), BMI 18.5–22.9 kg/m² (normal range) BMI 23–24.9 kg/m² (overweight), and BMI ≥25 kg/m² (obese). Hypoalbuminemia was defined as serum albumin <3.5 g/dL. Data on presence of comorbidities like diabetes mellitus, ischemic heart disease, and cardiac failure were included. The Modified Charlson Comorbidity Index (mCCI) was used to assess comorbidity burden. The following causes of death in PD patients were also collected retrospectively in this study: cardiovascular events, cerebrovascular events, peritonitis-related sepsis, sepsis due to other infections, malignancies, pulmonary events (respiratory failure secondary to primary pulmonary pathology like chronic obstructive pulmonary disease), other causes (include trauma, gastrointestinal bleed).

Results

A total of 550 patients were initiated on PD at our center during the study. Of these, 122 were excluded due to one or more of the following reasons: age <18 years at PD initiation, PD started for non-renal indications, or follow-up of <3 months. Twenty-three patients were lost within 3 months due to transfer to other centers, socioeconomic barriers, or unknown reasons. The final study cohort comprised 428 patients [Figure 1]. Nearly all patients in our cohort received manual PD, with only three on automated PD (APD).

Close

Figure 1:

Flowchart of patient selection. PD: Peritoneal dialysis.

Export to PPT

The mean age at initiation was 51.6 ± 14.5 years, and 67.5% were male. The median follow-up duration was 21.5 (IQR: 11.8–39.6) months, with the longest follow-up extending to 10.5 years. Around 53.5% had diabetes and 94.2% were hypertensive. A history of hemodialysis was present in 272 patients (63.6%). The median duration of hemodialysis prior to PD initiation in our cohort was 3.13 months (IQR: 1.65–7.55 months). Baseline demographic and clinical parameters, including age, sex, BMI distribution, comorbidities, and etiology of kidney disease, were largely similar between the PD and HD cohorts.

Patient's personal preference (74.9) was the most common reason for opting PD followed by multiple HD access failures (15%). Peritonitis was reported in 23.8% of patients. Among these, nine patients (8.8%) had multiple episodes, 14 (13.7%) developed fungal peritonitis, and five (4.9%) had tuberculous peritonitis [Table 1].

During the last follow-up, 227 patients died, and 97 (22.7%) were transitioned to HD. The number of patients initially started on PD was significantly higher in the survivors group compared to nonsurvivors [84 (53.8%) vs. 72 (46.2%); P = 0.035]. Conversely, patients transferred from HD to PD were more common in nonsurvivors [155 (57%) vs. 117 (43%); P = 0.035]. A greater proportion of patients had opted for PD due to multiple HD access failures and other associated medical conditions (like cardiac failure, chronic liver disease) in the mortality group.

Kaplan-Meier analysis revealed a cumulative patient survival of 86.4%, 67.0%, and 56.5% at 1, 2, and 3 years, respectively. Although survival estimates beyond 3 years showed a marked decline (42.1%, 31.8%, and 10.1% at 4, 5, and 10 years, respectively), these should be cautiously interpreted due to the small number of patients remaining at risk. Therefore, primary survival analyses and interpretations were restricted to the 3-year follow-up, which reflects the most statistically reliable segment of the cohort [Figure 2].

Close

Figure 2:

Kaplan-Meier survival curve of PD patients. Black line indicates the number of patients at risk (at every year point). PD: Peritoneal dialysis.

Export to PPT

Causes of mortality

Among the deceased, the cause of death was identified in 106 patients (47.1%), whereas it remained unknown in 119 (52.9%).

Among the former, cardiovascular events were the leading cause of death (32.1%). This was followed by sepsis due to infections other than peritonitis (20.8%) and sepsis secondary to PD-associated peritonitis (17.9%). CVA contributed to 9.4% of the deaths, while pulmonary infections, malignancy, and other causes accounted for 8.5%, 3.8%, and 7.5% of the mortality cases, respectively [Figure 3].

Close

Figure 3:

Causes of mortality in peritoneal dialysis patients.

Export to PPT

Discussion

The present study analyzed 10 years of experience with 428 PD patients at a tertiary care center, with a median follow-up duration of 21.5 months. We found the long-term survival to be modest, with a sharp decline after the third year. Multiple demographic, clinical, and socioeconomic variables were identified as predictors of mortality, with diabetes mellitus, hypoalbuminemia, diabetic kidney disease, and cerebrovascular disease showing consistent associations across univariate and multivariate analyses.

Our findings align with prior data from large PD registries and observational studies, which have consistently highlighted the significant impact of age and comorbid burden on PD outcomes. The average age of our cohort was 51.6 years, notably younger than those reported in Western PD populations such as the United States Renal Data System (USRDS) and the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA) cohorts, which often report mean initiation ages >60 years.^6,7 Despite this demographic advantage, long-term survival in our cohort was modest, possibly reflecting the high comorbidity burden, late referral patterns, and resource constraints prevalent in many low- and middle-income settings.⁸

The overall survival rates in our cohort at 1, 2, and 3 years were 86.4%, 67.0%, and 56.5%, respectively. They are lower than those reported in major international PD registries. For instance, the USRDS reports 1-year and 3-year survival rates of ~89% and 64%, respectively, among incident PD patients, while ANZDATA (2024) indicates similar benchmarks in high-resource settings, with 1-year survival exceeding 90% and 3-year survival above 70% in some subgroups.^7,9 Compared with other Indian studies [Table 3], the survival rates in the first 3 years in our cohort are better than those observed in Vikrant et al.'s cohort.^10–12 These disparities may be attributed to differences in healthcare infrastructure, timing of nephrology referral, nutritional status, and access to continuous follow-up and interventions such as assisted PD and infection surveillance. The overall trends of declining survival after 3 years are similar across the studies. The marked decline in survival beyond the third year may reflect a combination of cumulative comorbidity burden and resource limitations, which are commonly observed in lower-income settings.¹³

Diabetes and diabetic kidney disease were each independently associated with mortality, consistent with large observational studies from the USRDS, the European Renal Association-European Dialysis and Transplant Association (ERA-EDTA), and ANZDATA.^6,7 These findings reinforce the established risk conferred by both systemic and renal complications of diabetes in dialysis patients. Patients with diabetic kidney disease had nearly twice the risk of death in multivariable models.¹²

Hypoalbuminemia was present in ~40% of our cohort and was independently associated with mortality, echoing global evidence on the prognostic significance of serum albumin in dialysis patients.^14,15 Low albumin likely reflects a complex matrix of malnutrition, inflammation, and fluid overload. Studies such as CANUSA and the NECOSAD cohort have demonstrated the incremental predictive value of albumin for mortality, reinforcing its utility as a clinical surveillance marker.^16,17

In our cohort, BMI did not emerge as a statistically significant predictor of survival. This may be due to the influence of stronger prognostic factors such as age, cardiovascular comorbidities, or dialysis-related parameters. Additionally, the relatively low prevalence of extreme BMI categories in our cohort could have reduced the likelihood of detecting a significant association.

Cardiovascular disease remains the leading cause of mortality in PD patients worldwide. In our cohort, cardiovascular events accounted for 32.1% of known deaths, while cerebrovascular events were implicated in 9.4%. Notably, prior history of CVA, but not cardiac disease, remained independently predictive of mortality in multivariate analysis, potentially reflecting the disabling and progressive nature of cerebrovascular pathology.^18–20

Infectious mortality, including sepsis unrelated to PD and peritonitis-associated infections, accounted for 38.7% of known deaths. These rates are higher than those reported in high-income countries and highlight the need for enhanced infection control measures, prompt treatment protocols, and patient education.^21–23 Although our center follows standard PD training procedures, structural limitations and socioeconomic barriers likely contribute to infection risk.

A significant finding in our analysis was the protective association of financial support with patient survival. Having financial support was associated with a 60% lower risk of mortality (HR 0.396, P < .0001), underscoring the vital role of financial security in improving PD outcomes. The disproportionate burden of self-funded care in our cohort likely contributed to delayed interventions, suboptimal nutrition, medication non-adherence, and treatment discontinuity. Similar trends have been noted in other middle-income settings, where out-of-pocket expenditure remains a major barrier to equitable dialysis care.^3,24–27 These data support policy initiatives aimed at expanding insurance coverage for patients with ESKD, especially in low-resource settings.

Our study has several limitations. First, as a single-center retrospective analysis, the findings may not be generalizable to broader populations. Second, the cause of death remained unknown in >50% of the deceased, potentially introducing misclassification bias. Third, residual confounding due to unmeasured variables such as non-infectious complications, socioeconomic status, medication adherence, and health literacy cannot be excluded. Finally, the lack of granular data on PD technique survival, residual renal function, and inflammatory biomarkers limits deeper mechanistic insights. Despite the limitations, our study provides valuable insights into long-term outcomes in a real-world PD population from a resource-constrained setting and is likely the largest Indian cohort to date.

This study has several important limitations. As a single-center, retrospective analysis, the findings may not be generalizable to other populations or care settings. The median follow-up of 21.5 months restricts the reliability of long-term survival estimates beyond 3 years, due to the progressively smaller number of patients at risk. Although Kaplan-Meier methods allow extrapolation, we have limited interpretation to 3-year survival, with later estimates presented cautiously. The cause of death was unknown in >50% of the deceased patients due to limited documentation and out-of-hospital deaths, a common challenge in resource-limited contexts. Furthermore, data on urine output and residual renal function were not consistently recorded, preventing assessment of their impact on outcomes despite their known prognostic value. While peritonitis episodes were documented, precise timing and standardized peritonitis rates (e.g., per patient-year) could not be calculated due to incomplete records. Similarly, other non-infective PD complications, such as exit-site infections, hernias, and catheter-related issues, were not systematically captured. Finally, residual confounding from unmeasured factors (e.g., socioeconomic status, medication adherence, health literacy) cannot be excluded. Despite these limitations, this study contributes valuable insights into PD outcomes in a real-world, resource-constrained setting and represents one of the largest reported Indian cohorts to date.

In summary, this study highlights the multifactorial nature of mortality in patients undergoing PD, with metabolic, vascular, nutritional, and socioeconomic factors all contributing to adverse outcomes. The findings underscore the importance of early referral, comprehensive comorbidity management, infection prevention, and enhanced financial support systems to optimize long-term survival in PD patients, particularly in resource-constrained environments. Future research should focus on prospective multicenter registries and interventional studies targeting high-risk subgroups identified in our analysis.