There has been a significant increase in the number of incident dialysis patients over the past decade in Korea. According to the 2012 annual report from the Korean nationwide registry program, the number of new end-stage renal disease (ESRD) patients with hemodialysis (HD), peritoneal dialysis (PD), or kidney transplantation (KT) was 8,811 (169.8 per million population [PMP]), 923 (17.8 PMP), or 1,738 (33.5 PMP), respectively [1].

Notably, the number of new patients on PD has been decreasing since 2007, whereas the number of those on HD and KT has been increasing continuously, consistent with the trend observed in the United States [2]. Possible reasons for the decreasing rate of choosing PD as an initial dialysis therapy could include increased number of HD centers, better outcome in HD patients [3], and concerns about PD-related complications including peritonitis. Moreover, high technique failure rate in PD patients has contributed to the decreasing number of prevalent PD [1].

ESRD patients on dialysis suffer not only from higher mortality compared with the general population [4] but also from high morbidity due to various cardiovascular diseases, cancer, and technique failure. Technique failure can induce medical stress including subsequent mortality and additional cost, especially in the earlier period after initiation of PD [5], [6], [7]. Accordingly, there has been a continuous effort to analyze the causes and predictors of technique failure. Nevertheless, there are still controversial issues for technique failure risk factors because of different medical and social environments across countries. Although a population-based study for technique failure was reported in Western countries, there has not been a similar study in Asian countries. Because there are significant differences between Western and Asian ESRD patients in etiology of renal failure and mortality rates [8], it is likely that there are different characteristics for technique failure in Asian patients.

In addition, it is known that there are various competing risks in the research for outcomes in PD, and the risk of the outcome of interest may be overestimated without consideration of these competing risks [9], [10].

In this study, we aimed to analyze technique failure rate in PD in detail and to determine the specific predictors for technical failure by both conventional analysis and competing risks analysis, using the Korean Health Insurance Review and Assessment Service (HIRA) database, which is a nationwide, population-based data set.

In this study, we performed a national population-based study to evaluate technique failure in Korean patients initiating dialysis, using conventional survival analyses and competing risks analysis. The crude incidence rate of technique failure among PD patients was 54.1 per 1,000 patient-years. Because the incidence rate represented as cases per 1,000 patient-years is considered numerically equivalent to future 10-year incidence of events, thus it is recently regarded as essential for risk stratification and guiding therapy [14].

The annual technique failure rate among PD patients varies according to the region and the period when the study was conducted. The 2-year technique failure rate was 36% in an earlier Dutch study and in the recent US study [15], [16], which is higher than our rate of 10.3%. However, our 3-year technique failure rate is 15.6% and is similar to other studies that reported a rate of 16% in the early 1980s and a rate of 16.3% in a Korean single-center study [17], [18]. Those studies were mainly conducted in centers with particular experiences for treating PD patients, and a center's experience is associated with technique failure and mortality in PD [19]. Therefore, our results that included all Korean PD centers, regardless of centers' experiences, may not be inferior. In fact, a population-based Canadian study, which has a similar design to our study, reported a higher incidence of technique failure [20]. However, despite the decreasing rate of peritonitis in Korea [21], there has been no change in technique failure incidence over several decades. In addition, when we compared the crude technique failure–free survival rates among PD patients according to the dialysis initiation year between 2005 and 2008, there were no significant changes across the years. Although the period was not enough to investigate sequential changes, the survival rate among Korean PD patients initiating dialysis significantly improved from 2005 to 2008, as previously reported [22]. Therefore, technique failure in PD persists as a major problem in Korea.

In survival analysis of dialysis patients, there are several competing risks, such as death, transplantation, and technique failure [10]. The Kaplan–Meier method is likely to overestimate the incidence of technique failure. In this study, the 3-year cumulative incidence of technique failure in PD was 2.8% higher in Kaplan–Meier analysis compared to competing risks analysis. When comparing cumulative incidence function in competing risks analysis, we found that younger patients aged below 65 years, diabetics, and Medical Aid beneficiaries showed higher incidences of technique failure in PD patients.

Next, we also identified prognostic predictors for future technique failure events in PD, using both the Cox proportional hazard model and Fine and Gray subdistribution model. Of interest, independent predictors were not the same between the 2 modeling methods. In Cox analysis, Medical Aid and diabetes mellitus were significant factors, whereas Medical Aid, diabetes mellitus, and cancer were significant in the Fine and Gray model. Similarly to the previous report [23], the results in this study were also different between the Cox model and Fine and Gray model, thus we suggest that a competing risks analysis should be considered for the research on technique failure among PD patients. In another previous report, female gender, lower serum albumin, and higher body mass index were found to be independent risk factors [24]. However, because our database did not include laboratory data, we could not elucidate more valuable prognostic factors.

There were conflicting results in regards to age. In this study, age was not a significant predictor for technique failure, whereas several studies have suggested that old age was associated with an increased risk of technique failure in PD [15], [20]. In a comparison of cumulative incidence by 2 age groups, however, the technique failure rate was significantly higher in younger patients than in elderly patients. Fear to change and low life expectancy in elderly patients are the possible reasons for a lower technique conversion rate. In addition, more active decision-making in younger patients and the easy accessibility of HD in Korea could contribute to a higher modality conversion rate. However, we need further investigation to evaluate the exact reason.

We also found that diabetes mellitus was an independent factor for predicting technique failure, which is consistent with several previous studies [25], [26], [27]. Recently, however, Cotovio et al [28] reported that diabetes was a significant predictor for death in PD patients but was not a predictor for technique failure. Moreover, another retrospective cohort study indicated that baseline HbA1c levels greater than 6.5% did not show a statistically significant correlation with technique failure in PD [29].

In addition, Medical Aid beneficiaries had higher technique failure rates than National Health Insurance patients. Although low socioeconomic status did not predict fundamental outcomes such as patient survival or technique survival for PD patients in previous studies [16], [30], it might not be a major determinant to outcomes because the included subjects were in a managed-care setting or had taken a special care service from centers with a wide range of PD experiences. On the contrary, we evaluated a technique failure rate in claims database which comprised all Korean PD patients irrespective of hospital characteristics or center experiences. In addition, poor socioeconomic status generally put ESRD patients in a more disadvantageous position in terms of higher mortality and more frequent laboratory abnormalities [31], [32]. Moreover, limited health literacy, which was significantly related to socioeconomic status, is associated with health outcomes including mortality in patients with chronic kidney disease [33], whereas its association with adverse outcomes was not found in a study with 36 PD patients [34]. Although Medical Aid and National Health Insurance are a crude dichotomous classification, complex and rigorous criteria are applied into the classification process in Korea, including lack of family support and/or incapability of labor as well as low income. Therefore, the type of health security system could serve as a rough estimate for each patient's socioeconomic status in Korea, and it could be postulated that economic and functional status of patients and family support for them would generally affect the clinical outcomes in PD patients.

In the present study, cancer was a negative predictor for technique failure in the Fine and Gray model. Frequent medical checkup and paying more attention to their own health status would be associated with lower technique failure rates in PD patients with cancer. On the other hand, cancer is a well-known risk factor for mortality in PD patients [3]. When considering poor survival rate and low activity in ESRD patients with cancer, it is unlikely to change dialysis modalities without any inevitable reason.

Our study had several limitations. First, the database used in this study provided insufficient individual characteristics, inadequate clinical information, and no specific laboratory data. We could not collect data on membrane characteristics, residual renal function, health literacy level, and timing of referral to the nephrologists. Therefore, more valuable independent predictors could not be identified. In addition, the database did not include data on the cause of technique failure such as peritonitis, thus we could not recommend any specific strategy to decrease the technique failure rate in the future. Second, we excluded cases of technique failure that occurred within 3 months from commencing dialysis therapy because we would like to include only patients on “long-term dialysis” in this study. Descoeudres et al [35] emphasized high technique failure rate within the first 6 months on PD. Therefore, the technique failure rate in this study is likely to be underestimated. Last, the present study did not evaluate the effect of renal center characteristics, payment policy, physicians' acknowledgment, and patients' awareness for dialysis modality selection at dialysis initiation on technique failure in PD patients.

In spite of database-related inherent limitations, this is the first study on technique failure among Asian dialysis patients that used a national population-based data set. Technique failure was still a major concern in patients initiating PD in Korea, especially in diabetic patients and Medical Aid beneficiaries. The results of our study offer a basis for risk stratification for technique failure.

All authors have no conflicts of interest to declare.