Hemodialysis Leads to Better Survival in Patients with Diabetes or High Comorbidity, Compared to Peritoneal Dialysis

Abstract

Peritoneal dialysis (PD) has some advantages, such as hemodynamic stability and volume regulation, compared with hemodialysis (HD). However, the influence of the dialysis modality on survival is still controversial. This study assessed the mortality of incident patients undergoing HD versus PD using a propensity score approach. This study enrolled 873 subjects who began dialysis therapy at Gachon University Gil Hospital in Korea between January 2000 and June 2009. A propensity score comprising demographic, clinical, and laboratory variables was used to select a 1:1 matched cohort. The overall 1-, 2-, 3-, and 5-year survival rates for the HD patients (n = 212) were 95.1, 89.6, 82.5, and 65.3%, respectively, whereas the equivalent survival rates for the PD patients (n = 212) were 93.6, 83.1, 73.9, and 48.4%, respectively (P = 0.002 by log rank test). In patients without diabetes or patients with a low modified Charlson comorbidity index (MCCI), including hypertension, cardiovascular disease, liver disease, etc., there was no difference in mortality between PD and HD. However, PD was associated with a higher mortality for diabetic patients (HR, 2.86; 95% CI, 1.73-4.74) and for patients with a high MCCI (HR, 2.54; 95% CI 1.57-4.10). These data suggest that survival for PD may be comparable with that for HD in incident dialysis patients without diabetes or high MCCI and that HD could be more beneficial in patients with diabetes or high MCCI in this propensity score-matched cohort.

Introduction

Recently, the size of the elderly population and the prevalence of diabetes have been skyrocketing in Korea, both of which are common causes of chronic kidney disease. The number of patients with end-stage renal disease (ESRD) is increasing (Fenton et al. 1995; Collins et al. 2005; Jin et al. 2012); therefore, ESRD is recognized as an emerging clinical problem.

Theoretically, peritoneal dialysis (PD) has some advantages, such as hemodynamic stability and volume regulation, compared with hemodialysis (HD), which leads to PD preference for patients with cardiovascular comorbidity. In contrast, mortality differences between HD and PD are still debated, although many studies have investigated the effect of HD versus PD on mortality (Weinhandl et al. 2010; Mehrotra et al. 2011; Chang et al. 2012b). These controversies are unavoidable due to the varying selection biases in nonrandomized studies (Quinn et al. 2011). The choice of renal replacement therapy (RRT) as a treatment modality depends not only on the patient’s clinical state and choice but also on the physician’s practice, health-care organization, and government policy. Because comorbidity (e.g., hypertension and cardiovascular disease) has an important effect on the differences in the survival of HD and PD patients, comparing the two modalities when considering comorbid conditions is necessary.

The mortality rates of dialysis modalities have been compared in many countries, including Canada, the United States, and Taiwan, where PD utilization is low. The smaller PD proportion in the study may affect worse patient outcomes compared with a larger PD proportion (Schaubel et al. 2001; Weinhandl et al. 2010; Mehrotra et al. 2011; Chang et al. 2012b). Although Korea has one of the highest incidences and prevalence of ESRD and relatively high proportion of PD (Jin et al. 2012), there were no data comparing the dialysis mortality between HD and PD in Korea. This study assessed the mortality of incident ESRD patients receiving HD versus PD using a propensity score approach. This approach was used to control for confounding by measured factors from non-random treatment assignment after initiating dialysis.

Materials and Methods

Study population

We included incident dialysis patients who began dialysis therapy at Gachon University Gil Hospital, South Korea, between 1 January 2000 and 30 June 2009 and were still on dialysis 3 months after inception (Chang et al. 2012a). We identified 873 subjects with 69.3% treated with HD and 30.1% with PD. Patients with a history of kidney transplantation (n = 10), who were younger than 18 years of age (n = 2), and with insufficient medical records (n = 25) were excluded from the study. Ultimately, 836 subjects were included in the analysis.

Methods

All of the patients’ medical records were reviewed retrospectively, and the patient’s age, sex, height, weight, cause of ESRD, laboratory data and comorbid conditions at the time of RRT initiation were collected. Comorbid conditions were assessed using the modified Charlson comorbidity index (MCCI) (Hemmelgarn et al. 2003). The comorbid conditions included hypertension, history of coronary artery disease, myocardial infarction, congestive heart failure, cerebrovascular accident, liver disease, chronic obstructive pulmonary disease, and malignancy. The Institutional Review Board approved the study protocol.

The participants were followed from the initiation of dialysis until the end of the study (December 31, 2010) or death (determined from the Korean National Statistical Office). The initial dialysis modality was defined as the modality 90 days after the first dialysis. In the analysis of patient survival, patients on dialysis during the follow-up period were censored on the date of renal transplantation or loss to follow-up. Patients were not censored in the main analyses at RRT modality switches.

Statistical Analysis

To reduce the selection bias and control for potential confounding factors, the estimated propensity score for being assigned to PD versus HD as the first dialysis modality was calculated for each patient using multiple logistic regression models with the following covariates: age, sex, cause of ESRD, medical history, and laboratory tests. We matched one PD patient to one HD patient based on a greedy 8-1 digit-matching algorithm. First, this algorithm attempted to match PD and HD subjects based on the first eight digits of the propensity score. For those that did not match, the PD subjects were then matched to HD subjects using seven digits of the propensity score. The algorithm proceeded sequentially to a one-digit match on the propensity score. Patients without a corresponding match were excluded. The propensity score matching was conducted using SAS 9.1.3 (SAS Institute, Cary, NC).

Categorical variables were compared using the χ² test, and continuous variables were compared using the t-test when appropriate. We performed Kaplan-Meier survival analyses to determine the patient survival rates and compared the survival rates between the HD and PD groups using the log-rank test. Cox regression analysis was used to assess the association between dialysis modality and mortality. The Cox regression model stratified on matched pairs was used in the matched cohort. Associations are presented as hazard ratios (HRs) with their corresponding 95% confidence intervals (CIs). All analyses excluding the propensity score matching were performed with SPSS ver. 12.0 (SPSS, Chicago, IL). A P-value < 0.05 was considered statistically significant.

Results

The clinical characteristics of the patients at the beginning of the RRT in the entire cohort and the matched cohort are presented in Table 1. In the entire cohort, patients undergoing HD were more likely to be older and male. Hypertension, diabetes, and moderate or severe liver disease were more common in patients undergoing HD. The patients undergoing HD had higher levels of BUN, creatinine, phosphorus, and CRP and lower levels of hemoglobin, albumin, and calcium. After we used the propensity method adjusted for the variables, 424 patients (212 in each group) remained. Their mean age was 52 years, and 62.7% were male. The prevalence of diabetes was 53.5% for HD and 57.1% for PD. Hypertension was the most common comorbidity. There was no difference in age, sex, cause of ESRD, hypertension, diabetes, hemoglobin, creatinine, or albumin between the two groups.

Fig. 1 shows a Kaplan-Meier actuarial survival plot by modality for the entire cohort (Fig. 1A) and for the propensity score-matched cohort (Fig. 1B). In the entire cohort, the HD group had a better survival rate over the follow-up period (P < 0.001 by log rank test). In the propensity score-matched cohort, the overall 1-, 2-, 3-, 5-, and 7-year survival rates for the HD patients were 95.1, 89.6, 82.5, 65.3, and 47.1%, respectively. The equivalent survival rates for the PD patients were 93.6, 83.1, 73.9, 48.4, and 28.3%, respectively. The patient survival rates differed significantly between modalities in the matched cohort (P = 0.002 by log rank test).

Fig. 2 shows the hazard ratios of PD versus HD for mortality and the results of the prespecified subgroup analyses. In patients without diabetes, PD and HD were comparable with respect to mortality (HR, 0.76; 95% CI 0.38-1.50). In contrast, in diabetic patients, PD had a significantly higher mortality compared with HD (HR, 2.86; 95% CI 1.73-4.74). The difference in the hazard ratio of diabetes was significant (Pint = 0.003). As shown in Fig. 3, PD patients with diabetes had the worst survival in the four combined groups. In other words, the presence of diabetes significantly affected the survival of the PD patients, but did not influence the survival of the HD patients.

The mortality risk of the PD patients with a MCCI ≤ 5 was similar to that of HD patients with a MCCI ≤ 5 (HR, 0.87; 95% CI 0.43-1.8). In patients with a MCCI > 5, PD had a higher risk of death than HD (HR, 2.54; 95% CI 1.57-4.1). The difference in the hazard ratio of MCCI was significant (Pint = 0.019) (Fig. 2). As shown in Fig. 4, PD patients with a high MCCI had the worst survival in the four combined groups, i.e., a high MCCI had an effect on the survival of the PD patients but did not affect the survival of the HD patients.

Table 1. Patients characteristics by dialysis modality

Characteristics	Before Matching			After Matching
	All HD (n = 579)	All PD (n = 257)	P	Matched HD (n = 212)	Matched PD (n = 212)	P
Age (years)	56.0 ± 14.2	50.2 ± 13.9	< 0.001	51.6 ± 13.8	52.1 ± 12.9	1
Male (%)	308 (53.2)	167 (65.0)	0.002	133 (62.7)	132 (62.3)	0.92
Cause of end stage renal disease (%)			< 0.001			0.829
Diabetes	321 (55.4)	128 (49.8)		112 (52.8)	118 (55.6)
Hypertension	55 (9.5)	47 (18.3)		34 (16.0)	28 (13.2)
Glomerulonephritis	34 (5.9)	30 (11.7)		18 (8.4)	20 (9.4)
Others	169 (29.2)	52 (20.2)		48 (22.6)	46 (21.7)
Follow-up (month)	41.2 ± 32.2	46.6 ± 29.4	0.018	45.1 ± 32.8	46.0 ± 29.5	0.826
Comorbidity (%)
Hypertension (%)	506 (87.4)	205 (79.8)	0.004	179 (84.4)	175 (82.5)	0.601
Diabetes (%)	341 (58.9)	131 (51.0)	0.033	113 (53.3)	121 (57.1)	0.435
Coronary artery disease (%)	104 (18.0)	50 (19.5)	0.607	40 (18.8)	45 (21.2)	0.544
Congestive heart failure (%)	110 (19.0)	54 (21.0)	0.499	33 (15.6)	43 (20.3)	0.205
Myocardial infarction (%)	22 (3.8)	16 (6.2)	0.122	5 (2.35)	13 (6.1)	0.089
Dysrhythmia (%)	20 (3.5)	6 (2.3)	0.39	5 (2.35)	5 (2.4)	1
Peripheral vascular disease (%)	19 (3.3)	6 (2.3)	0.458	7 (3.30)	5 (2.4)	0.558
Cerebrovascular disease (%)	58 (10.0)	22 (8.6)	0.509	18 (8.5)	19 (8.9)	0.863
Hemiplegia (%)	10 (1.7)	7 (2.7)	0.346	3 (1.4)	7 (3.3)	0.338
Peptic ulcer (%)	27 (4.7)	9 (3.5)	0.443	9 (4.3)	8 (3.8)	0.797
Mild liver disease (%)	36 (6.2)	7 (2.7)	0.034	11 (5.2)	6 (2.8)	0.216
Moderate or severe liver disease (%)	27 (4.7)	0 (0.0)	< 0.001	7 (3.3)	0 (0)	0.015
Chronic obstructive pulmonary disease (%)	9 (1.6)	3 (1.2)	0.664	1 (0.5)	3 (1.4)	0.623
Vasculitis (%)	4 (0.7)	1 (0.4)	0.602	2 (0.9)	1 (0.5)	1
Malignancy (%)	24 (4.2)	4 (1.6)	0.054	9 (4.3)	4 (1.9)	0.259
Any tumor (%)	20 (3.5)	3 (1.2)	0.062	7 (3.3)	3 (1.4)	0.338
Laboratory data
Blood urea nitrogen (mg/dL)	81.0 ± 47.5	64.3 ± 32.3	< 0.001	69.2 ± 35.7	66.8 ± 33.2	0.647
Creatinine (mg/dL)	11.9 ± 16.6	8.7 ± 7.4	< 0.001	11.0 ± 17.7	8.9 ± 8.0	0.729
Hemoglobin (g/dL)	8.2 ± 1.8	9 ± 1.7	< 0.001	8.8 ± 1.8	8.8 ± 1.6	0.297
Albumin (g/dL)	3.2 ± 0.6	3.3 ± 0.6	0.006	3.2 ± 0.7	3.3 ± 0.6	0.471
Total cholesterol (mg/dL)	162.0 ± 47.2	173.1 ± 47.9	0.002	169.0 ± 51.4	173.8 ± 49.1	0.109
Calcium (mg/dL)	8.1 ± 1.1	8.4 ± 1.0	0.002	8.3 ± 1.2	8.3 ± 1.0	0.841
Phosphorus (mg/dL)	5.5 ± 2.3	5.1 ± 1.9	0.008	5.3 ± 2.3	5.1 ± 2.0	0.595
C-reactive protein (mg/dL)	3.6 ± 5.7	1.6 ± 3.7	< 0.001	3.5 ± 5.8	1.5 ± 3.6	0.371

HD; Hemodialysis, PD; Peritoneal dialysis

Fig. 1.

Kaplan-Meier survival curve between peritoneal dialysis (PD) and hemodialysis (HD).

The HD group had a better survival rate than the PD group in the entire cohort (n = 836) (A) and in the propensity score-matched cohort (n = 424) (B).

Fig. 2.

The adjusted hazard ratio of peritoneal dialysis (PD) versus hemodialysis (HD) for mortality using a Cox proportional analysis in the matched cohort (n = 424).

Fig. 3.

Kaplan-Meier survival curve according to the dialysis modality and diabetes mellitus (DM) (n = 424). Patients on peritoneal dialysis (PD) with DM had the worst survival.

Fig. 4.

Kaplan-Meier survival curve according to the dialysis modality and comorbidity (n = 424).

Patients on peritoneal dialysis (PD) with a high score of the modified Charlson comorbidity index (MCCI) had the worst survival.

Discussion

We show that HD may lead to better survival in patients with diabetes or high comorbid conditions, although survival for PD was comparable with that for HD in patients without those conditions.

Winkelmayer et al. (2002) showed that PD appeared to be associated with higher mortality among older patients compared with HD. This observation is consistent with our results. Several reasons for the higher mortality of patients treated with PD were proposed. First, the serum lipoprotein (a) (Kimak et al. 2000; Kronenberg et al. 2003) and hyperlipidemia (Lee et al. 1998; Kronenberg et al. 2003) levels are reported to be higher in patients undergoing PD than in patients undergoing HD. Accelerated atherosclerosis may occur in the PD patients. Second, inadequate dialysis and fluid overload may occur because the residual renal function and ultrafiltration capacity of the peritoneal membrane in PD patients decrease over time (Churchill et al. 1998; Bargman et al. 2001; Menon et al. 2001). Inadequate dialysis and fluid overload may increase the overall mortality risk. However, Weinhandl et al. (Weinhandl et al. 2010) found that the overall mortality risk after dialysis initiation was 8% lower for PD than for matched HD patients. Patients starting dialysis in the United States in 2003 were enrolled in that study. However, PD patients accounted for only 6.76% of the ESRD patients, which differs from the rate in our study.

Other studies have shown that PD appears to be associated with equal or better survival during the first year compared with HD. In an analysis of 1,041 participants in the Choices for Healthy Outcomes in Caring for ESRD Patients study, Jaar et al. (Jaar et al. 2005) reported a propensity score-adjusted HR of death for PD versus HD of 1.47 (P = 0.32) in the first year and 2.05 (P = 0.03) in the second year, with little evidence of effect modification by age, cardiovascular disease, or diabetes. The cohort was notable for the relatively low mortality among the HD patients. In a propensity score-matched cohort of 16,791 patients initiating dialysis in Australia and New Zealand between 1991 and 2005, McDonald et al. (McDonald et al. 2009) reported an HR of death for PD versus HD of 0.99 (P = 0.80) from days 90 to 365 and 1.35 (P < 0.01) after day 366.

The ideal way to compare the difference is a randomized, controlled trial (RCT). However, conducting long-term RCTs is different. Korevaar et al. (Korevaar et al. 2003) attempted an RCT in the Netherlands, but the number of patients willing to participate was low; because only 38 patients enrolled, the trial was stopped prematurely. RCTs have failed to show the beneficial effect of dialysis modality in ESRD patients. Quinn et al. (Quinn et al. 2011) suggested that selection bias, rather than an effect of the treatment itself, likely explains the relative risk of death between dialysis modalities. Therefore, to increase the comparability between the PD and HD groups by reducing the selection bias and controlling for potential confounding, we matched the HD and PD patients based on the propensity score at the time of RRT initiation. Few studies have used propensity scores (Winkelmayer et al. 2002; Jaar et al. 2005; McDonald et al. 2009; Weinhandl et al. 2010; Chang et al. 2012b). Only one contemporary study has examined propensity score-matched comparisons of dialysis modality on survival in an Asian population. For 54,370 incident HD and 3,443 incident PD patients in the population-based insurance claims data for Taiwan, the PD and HD patients were observed to have similar long-term survival rates (Chang et al. 2012b). This result was inconsistent with ours. This discrepancy may be explained by the different characteristics of the study population. In our study, more patients were male (62.5% vs. 39.9%) and the rate of diabetes-related ESRD was higher (54.2% vs. 34.1%). Only 7.9% of ESRD patients had been treated with PD. Subgroup analyses revealed that younger patients who underwent PD had a better survival rate than younger patients who underwent HD.

This study has some limitations. First, despite the propensity score matching, the modest sample size limited the power to detect differences between the PD and HD cohorts. After propensity score matching from the pre-matching rates, the PD group showed a tendency to have higher cardiovascular disease-related risks, such as diabetes, congestive heart failure and myocardial infarction. In contrast, the HD group had a higher number of non-cardiovascular disease-related risks, such as liver disease. Considering the majority of mortalities in dialysis patients are due to cardiovascular diseases, this bias could affect the result significantly. The reason for this bias is most likely due to the initial patient selection bias at the time of the modality decision. Second, the generalizability was reduced because it was a single-center experience.

In conclusion, survival for PD may be comparable with that for HD in incident dialysis patients without diabetes or high comorbidity, and survival for HD could be more beneficial in patients with diabetes or high comorbidity in this propensity score-matched cohort.

Acknowledgements

This work was supported by a grant from the Seoul National University Bundang Hospital Research Fund (No. 03-2012-020), and grant No. 2011-0008605 from the National Research Foundation of Korea.

Conflict of Interest

The authors have no financial conflict of interest.

References

•  Bargman,  J.M.,  Thorpe,  K.E. &  Churchill,  D.N. (2001) Relative contribution of residual renal function and peritoneal clearance to adequacy of dialysis: a reanalysis of the CANUSA study. J. Am. Soc. Nephrol., 12, 2158-2162.
•  Chang,  J.H.,  Rim,  M.Y.,  Sung,  J.,  Ko,  K.P.,  Kim,  D.K.,  Jung,  J.Y.,  Lee,  H.H.,  Chung,  W. &  Kim,  S. (2012a) Early start of dialysis has no survival benefit in end-stage renal disease patients. J. Korean Med. Sci., 27, 1177-1181.
•  Chang,  Y.K.,  Hsu,  C.C.,  Hwang,  S.J.,  Chen,  P.C.,  Huang,  C.C.,  Li,  T.C. &  Sung,  F.C. (2012b) A comparative assessment of survival between propensity score-matched patients with peritoneal dialysis and hemodialysis in Taiwan. Medicine (Baltimore), 91, 144-151.
•  Churchill,  D.N.,  Thorpe,  K.E.,  Nolph,  K.D.,  Keshaviah,  P.R.,  Oreopoulos,  D.G. &  Page,  D. (1998) Increased peritoneal membrane transport is associated with decreased patient and technique survival for continuous peritoneal dialysis patients. The Canada-USA (CANUSA) Peritoneal Dialysis Study Group. J. Am. Soc. Nephrol., 9, 1285-1292.
•  Collins,  A.J.,  Kasiske,  B.,  Herzog,  C.,  Chavers,  B.,  Foley,  R.,  Gilbertson,  D.,  Grimm,  R.,  Liu,  J.,  Louis,  T.,  Manning,  W.,  Matas,  A.,  McBean,  M.,  Murray,  A.,  St Peter,  W.,  Xue,  J.,  Fan,  Q.,  Guo,  H.,  Li,  S.,  Roberts,  T.,  Snyder,  J.,  Solid,  C.,  Wang,  C.,  Weinhandl,  E.,  Arko,  C.,  Chen,  S.C.,  Dalleska,  F.,  Daniels,  F.,  Dunning,  S.,  Ebben,  J.,  Frazier,  E.,  Johnson,  R.,  Sheets,  D.,  Forrest,  B.,  Berrini,  D.,  Constantini,  E.,  Everson,  S.,  Frederick,  P.,  Eggers,  P. &  Agodoa,  L. (2005) Excerpts from the United States Renal Data System 2004 annual data report: atlas of end-stage renal disease in the United States. Am. J. Kidney Dis., 45, A5-7, S1-280.
•  Fenton,  S.,  Desmeules,  M.,  Copleston,  P.,  Arbus,  G.,  Froment,  D.,  Jeffery,  J. &  Kjellstrand,  C. (1995) Renal replacement therapy in Canada: a report from the Canadian Organ Replacement Register. Am. J. Kidney Dis., 25, 134-150.
•  Hemmelgarn,  B.R.,  Manns,  B.J.,  Quan,  H. &  Ghali,  W.A. (2003) Adapting the Charlson Comorbidity Index for use in patients with ESRD. Am. J. Kidney Dis., 42, 125-132.
•  Jaar,  B.G.,  Coresh,  J.,  Plantinga,  L.C.,  Fink,  N.E.,  Klag,  M.J.,  Levey,  A.S.,  Levin,  N.W.,  Sadler,  J.H.,  Kliger,  A. &  Powe,  N.R. (2005) Comparing the risk for death with peritoneal dialysis and hemodialysis in a national cohort of patients with chronic kidney disease. Ann. Intern. Med., 143, 174-183.
•  Jin,  D.C.,  Ha,  I.S.,  Kim,  N.H.,  Lee,  S.W.,  Lee,  J.S.,  Yoon,  S.R. &  Kim,  B.S. (2012) Brief Report: Renal replacement therapy in Korea, 2010. Kidney Res. Clin. Pract., 31, 62-71.
•  Kimak,  E.,  Solski,  J.,  Janicka,  L.,  Ksaziek,  A. &  Janicki,  K. (2000) Concentration of Lp(a) and other apolipoproteins in predialysis, hemodialysis, chronic ambulatory peritoneal dialysis and post-transplant patients. Clin. Chem. Lab. Med., 38, 421-425.
•  Korevaar,  J.C.,  Feith,  G.W.,  Dekker,  F.W.,  van Manen,  J.G.,  Boeschoten,  E.W.,  Bossuyt,  P.M. &  Krediet,  R.T. (2003) Effect of starting with hemodialysis compared with peritoneal dialysis in patients new on dialysis treatment: a randomized controlled trial. Kidney Int., 64, 2222-2228.
•  Kronenberg,  F.,  Lingenhel,  A.,  Neyer,  U.,  Lhotta,  K.,  Konig,  P.,  Auinger,  M.,  Wiesholzer,  M.,  Andersson,  H. &  Dieplinger,  H. (2003) Prevalence of dyslipidemic risk factors in hemodialysis and CAPD patients. Kidney Int. Suppl., S113-116.
•  Lee,  S.W.,  Kwon,  K.H. &  Kim,  M.J. (1998) Comparison of lipid profiles in long-term CAPD and hemodialysis patients. Perit. Dial. Int., 18, 435-437.
•  McDonald,  S.P.,  Marshall,  M.R.,  Johnson,  D.W. &  Polkinghorne,  K.R. (2009) Relationship between dialysis modality and mortality. J. Am. Soc. Nephrol., 20, 155-163.
•  Mehrotra,  R.,  Chiu,  Y.W.,  Kalantar-Zadeh,  K.,  Bargman,  J. &  Vonesh,  E. (2011) Similar outcomes with hemodialysis and peritoneal dialysis in patients with end-stage renal disease. Arch. Intern. Med., 171, 110-118.
•  Menon,  M.K.,  Naimark,  D.M.,  Bargman,  J.M.,  Vas,  S.I. &  Oreopoulos,  D.G. (2001) Long-term blood pressure control in a cohort of peritoneal dialysis patients and its association with residual renal function. Nephrol. Dial. Transplant., 16, 2207-2213.
•  Quinn,  R.R.,  Hux,  J.E.,  Oliver,  M.J.,  Austin,  P.C.,  Tonelli,  M. &  Laupacis,  A. (2011) Selection bias explains apparent differential mortality between dialysis modalities. J. Am. Soc. Nephrol., 22, 1534-1542.
•  Schaubel,  D.E.,  Blake,  P.G. &  Fenton,  S.S. (2001) Effect of renal center characteristics on mortality and technique failure on peritoneal dialysis. Kidney Int., 60, 1517-1524.
•  Weinhandl,  E.D.,  Foley,  R.N.,  Gilbertson,  D.T.,  Arneson,  T.J.,  Snyder,  J.J. &  Collins,  A.J. (2010) Propensity-matched mortality comparison of incident hemodialysis and peritoneal dialysis patients. J. Am. Soc. Nephrol., 21, 499-506.
•  Winkelmayer,  W.C.,  Glynn,  R.J.,  Mittleman,  M.A.,  Levin,  R.,  Pliskin,  J.S. &  Avorn,  J. (2002) Comparing mortality of elderly patients on hemodialysis versus peritoneal dialysis: a propensity score approach. J. Am. Soc. Nephrol., 13, 2353-2362.