EARLY INITIATION OF DIALYSIS AND LATE IMPLANTATION OF CATHETERS ADVERSELY
AFFECT OUTCOMES OF PATIENTS ON CHRONIC PERITONEAL DIALYSIS
Chih-Chung Shiao,
1Jenq-Wen Huang,
2Kuo-Liong Chien,
2Hsueh-Fang Chuang,
4Yung-Ming Chen,
2and Kwan-Dun Wu
2Renal Division,
1Department of Internal Medicine, St. Mary’s Hospital, Lo Tung; Renal Division,
2Department
of Internal Medicine, National Taiwan University Hospital and College of Medicine, National Taiwan
University; College of Medicine,
3Institute of Preventive Medicine, National Taiwan University;
Department of Nursing,
4National Taiwan University Hospital, Taipei, Taiwan
Correspondence to: Y.M. Chen, Renal Division, Department of Internal Medicine, National Taiwan University Hospital, No. 7 Chung-Shan South Road, Taipei, 100, Taiwan.
Received 13 June 2007; accepted 10 September 2007.
♦ ♦♦ ♦
♦ Objectives: Predialysis nephrology care is thought to af-fect morbidity and mortality in hemodialysis patients. This study evaluated the impact of different patterns of predial-ysis care on outcomes of patients undergoing chronic peri-toneal dialysis (PD).
♦ ♦♦ ♦
♦ Design: Retrospective cohort.
♦ ♦♦ ♦
♦ Setting and Participants: 275 patients enrolled from Janu-ary 1997 to March 2005 in a medical center in North Taiwan who recently initiated dialysis were classified according to early or late referral to nephrologists (≥≥≥≥≥6 or <6 months of dialysis), planned or late implantation of Tenckhoff catheters (absence or presence of preceding emergent hemodialysis), and early or late start of dialysis [glomerular filtration rate (GFR) ≥≥≥≥≥5 or <5 mL/minute/1.73 m2].
♦ ♦♦ ♦
♦ Main Outcome Measures: All-cause mortality and hospitalization.
♦ ♦♦ ♦
♦ Results: Dur ing a median follow-up of 2.5 years, 41 deaths, 38 transfers to hemodialysis, and 26 renal trans-plantations occurred. Late start of dialysis was associated with a significant survival benefit (log rank, p = 0.012) and, along with planned implantation of catheters, exhibited a reduced risk for all-cause hospitalization (log rank, p = 0.025, 0.013). The predictors of overall mortality included baseline GFR [hazard ratio (HR) 1.18, p = 0.023], age (HR 1.07, p < 0.001), and diabetes (HR 3.64, p = 0.001); whereas the risk factors for all-cause hospitalization included age (HR 1.02, p = 0.012), late implantation of catheters (HR 1.78, p = 0.011), and diabetes (HR 1.92, p = 0.005). The timing of nephrology referral did not affect either death or hospitalization.
♦ ♦♦ ♦
♦ Conclusions: Our data do not support earlier initia-tion of PD, but underscore the importance of planned
implantation of catheters before commencement of chronic PD.
Perit Dial Int 2008; 28:73–81 www.PDIConnect.com
KEY WORDS: End-stage renal disease; hospitalization; mortality.
T
he incidence of treated end-stage renal disease (ESRD) in Taiwan is among the highest in the world (1). The resultant reimbursement for chronic dialysis therapy absorbs a large proportion of the nation’s healthcare budget (2). More than 90% of ESRD patients choose hemodialysis (HD) as the modality of chronic renal replacement therapy, which is being looked at with concern by policymakers and healthcare providers. As a result, the government is promoting the implementation of chronic peritoneal dialysis (PD) for incident ESRD pa-tients due mainly to its inexpensiveness relative to HD. However, while official detailed figures are lacking, only a small proportion of ESRD patients undergoing chronic PD had a planned start of dialysis, that is, without pre-ceding acute HD through temporary HD catheters. This would greatly offset the economic advantage of PD, since any complications associated with the use of HD cath-eters could invariably lead to an increase in morbidity and mortality, let alone medical costs. The reasons for failure to initiate chronic PD in time are likely multifac-torial but could be related, at least in part, to inadequate predialysis care.Predialysis nephrology care has been shown to af-fect morbidity and mortality af ter patients begin chronic renal replacement therapy, particularly HD. For instance, delayed nephrology referral has been impli-cated in increased mortality and hospitalization after ini-tiation of chronic HD (3,4), and timely referral allows
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
University Hospital’s Research Ethics Committee (No. 9461700833).
OUTCOMES
The primary end point of the study was overall mor-tality; the secondary end point was all-cause hospital-ization at termination of follow-up. The baseline demographic characteristics and laboratory data of each patient were documented by reviewing medical charts. The earliest (performed within 3 months after start of chronic PD) and the last available PET, including urine amount, total Kt/V, total weekly creatinine clearance (WCCr), peritoneal Kt/V, peritoneal WCCr, renal Kt/V, renal WCCr, and normalized protein catabolic rate (nPCR) were also recorded. The last PET data were unavailable in 16 patients due to insufficient follow-up period, that is, less than 6 months. All parameters were calculated according to standard formulas presented in the Euro-pean best practice guidelines for PD (12).
COVARIATES
The patients were categorized according to the tim-ing of nephrology referral, implantation of Tenckhoff catheters, or initiation of PD. Early referral was defined as referral to nephrologists at least 6 months before the initiation of chronic PD; late referral was defined as re-ferral less than 6 months before commencement of chronic PD. The planned implantation category included patients that received implantation of Tenckhoff cath-eters before initiation of chronic PD, without preceding emergent HD; the late implantation category included patients that had undergone at least one session of emer-gent HD via temporary non-cuffed catheters, before im-plantation of Tenckhoff catheters. Early start was defined as initiation of chronic PD when glomerular fil-tration rate (GFR) was ≥5 mL/minute /1.73 m2; late start
was defined as initiation of chronic PD with GFR <5 mL/ minute/1.73 m2. The cutoff level to define early and late
start of dialysis has been used by Wilson et al. (13). The GFRs were estimated according to the Modification of Diet in Renal Disease equation: GFR (mL/min/1.73 m2) =
186 × sCr–1.154 × age–0.203 × 1.212 (if black) × 0.742 (if
female) (7).
STATISTICAL ANALYSES
Student’s t-test and Wilcoxon signed rank test were performed to compare differences in continuous and non-normally distributed variables between subgroups. Chi-square test was used to compare the nominal vari-implementation of multidisciplinary predialysis care,
which can lead to improved patient survival after initia-tion of dialysis (5). Further, timely creainitia-tion of a perma-nent vascular access before starting chronic HD is associated with a lowered risk of sepsis and death from the use of temporary HD catheters (6), and earlier cre-ation allows the fistula to become mature before cannu-lation, thus providing long-term complication-free vascular access (3,7,8). Finally, recent studies suggest that earlier initiation of dialysis may not be linked with better patient outcomes (9,10), and could even cause a paradoxically increased risk for death (11). By contrast, information is limited regarding the impact of different predialysis care on outcomes of patients undergoing chronic PD. This study was performed to evaluate over-all mortality and hospitalization in a retrospective co-hort of incident PD patients stratified by different care patterns (early vs late nephrology referral, catheter im-plantation, and dialysis initiation) at a medical center in North Taiwan where ESRD is endemic.
METHODS
STUDY POPULATION
This study screened 310 ESRD patients started on PD at the National Taiwan University Hospital between 1 January 1997 and 31 March 2005. These patients were treated at the same unit by the same team of physicians and nurses. Patients were treated with continuous am-bulatory PD or automated PD using conventional dialy-sates (Dianeal 1.5%, 2.5%, or 4.25% glucose; Baxter Healthcare, Deerfield, Illinois, USA). Other types of di-alysates, such as Nutrineal (1.1%; Baxter) and Extraneal (7.5%; Baxter), were prescribed whenever needed for patients with significant hypoalbuminemia (serum al-bumin <3.5 g/dL measured by the bromcresol green assay) or poor ultrafiltration (<0.5 L/day despite the use of high osmolality solutions) respectively. Seventeen patients that discontinued PD treatment within 3 months of commencement were excluded at initial screening. The reasons for discontinuation included death (7 patients: 5 died from cardiovascular disease, 2 from infection), transfer to HD (3 patients: 2 due to catheter dysfunction or leakage, 1 due to lack of family support), renal transplantation (3 patients), transfer to other hospitals (3 patients), and recovery of renal func-tion (1 patient). Another 18 patients were excluded for lack of initial biochemical data or peritoneal equilibra-tion test (PET). Finally, a total of 275 persons (124 males, 151 females) were selected and followed until 30 June 2005. This study was approved by the National Taiwan
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
ables between subgroups. Logistic regression analysis was used to determine the factors associated with planned implantation of catheters. Patient survival and event probability were evaluated using Kaplan–Meier survival analysis, with comparisons between survival curves made by the log-rank test. The backward stepwise model of the Cox proportional hazards method was ap-plied to calculate hazard ratio (HR) and adjusted sur-vival curves for time to event, adjusting for age, gender, level of education, occupational activity, nephrology re-ferral, implantation of catheters, initiation of dialysis, comorbidities (diabetes, coronary artery disease, con-gestive heart failure), and laboratory data that had prob-ability values less than 0.2 on univariate analysis or were considered to be important. Data were analyzed using the Scientific Package for Social Science (SPSS, version 12.0; SPSS Inc, Chicago, Illinois, USA) for Windows op-erating system (Microsoft Corp., Redmond, Washington, USA). Any probability values less than 0.05 were consid-ered significant.
RESULTS
STUDY POPULATION
Baseline characteristics of the 275 patients enrolled in this study are shown in Table 1. The patients were mostly female (55%), middle-aged (mean age 51.3
years), occupationally active (67%), and had a higher educational level (62%). Among the participants, 43% of patients were referred early and 21% received planned implantation of catheters before the start of PD. Mean duration of PD treatment was 30.1 months. One hundred and ten patients dropped out due to various causes, in-cluding death, modality shift to HD, kidney transplan-tation, and transfer to other hospitals (Table 2). At initiation of chronic PD, mean serum creatinine level and GFR were 12.0 mg/dL and 4.8 mL/minute/1.73 m2
re-spectively (Table 3).
TIMING OF INITIATION OF DIALYSIS AND PATIENT OUTCOMES AFTER START OF CHRONIC PD
Kaplan–Meier plots show that late start of dialysis was associated with a significant survival benefit (log rank, p = 0.012) and a reduced risk for all-cause hospitaliza-tion (log rank, p = 0.025) (Figure 1). Compared with the early start group, patients in the late start group were younger (p < 0.001), occupationally more active (p < 0.001), and had fewer comorbidities (for diabetes, p < 0.001; for congestive heart failure, p = 0.017) (Table 1). The late start group had higher serum albumin (p = 0.005) and calcium-phosphate product (p = 0.001), but lower hemoglobin (p = 0.004) and renal WCCr (p = 0.001) at study baseline (Table 3). They also had a lower rate of dropout due to death (p = 0.023) and a higher rate of TABLE 1
Comparison of Baseline Demographic Characteristics of Patients with Different Patterns of Initiation of Dialysis and Implantation of Catheters
Initiation of dialysisc Implantation of cathetersd Total (n=275) Early (n=110) Late (n=165) p Value Planned (n=58) Late (n=217) p Value
Age (years) 51.3±17.2 56.1±18.5 48.1±15.5 <0.001 51.0±15.3 51.4±17.7 NS
Female gender 151 (55%) 39 (35%) 112 (68%) <0.001 35 (60%) 116 (53%) NS
Higher education levela 171 (62%) 67 (61%) 104 (63%) NS 44 (76%) 127 (59%) 0.015 Occupationally activeb 184 (67%) 57 (52%) 127 (77%) <0.001 49 (85%) 135 (62%) 0.001 Referral time>6 months 119 (43%) 48 (44%) 71 (43%) NS 47 (81%) 72 (33%) <0.001
Planned implantation 58 (21%) 24 (22%) 34 (21%) NS — — — Comorbidity Diabetes 53 (19%) 33 (30%) 20 (12%) <0.001 7 (12%) 46 (21%) NS Hypertension 136 (49%) 56 (51%) 80 (48%) NS 32 (55%) 104 (48%) NS CAD 15 (5.5%) 6 (5%) 9 (5%) NS 0 (0%) 15 (7%) 0.046 CHF 18 (6.5%) 12 (11%) 6 (4%) 0.017 4 (7%) 14 (6.5%) NS
CAD = coronary artery disease; CHF = congestive heart failure; NS = not significant. aHigher education level denotes junior high school and above.
bOccupationally active is defined by being a student or having a job, including housekeeping. cEarly or late start of dialysis means glomerular filtration rate ≥5 or <5 mL/minute/1.73 m2.
dPlanned or late implantation of Tenckhoff catheters means absence or presence of preceding emergent hemodialysis. Values are presented as mean±SD or number (%) unless otherwise specified.
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
TABLE 2
Dropouts Among Patients with Different Patterns of Initiation of Dialysis and Implantation of Catheters Initiation of dialysisd Implantation of catheterse Total (n=275) Early (n=110) Late (n=165) p Value Planned (n=58) Late (n=217) p Value Follow-up period (months)a 30.1±20.8 28.5±19.4 31.2±21.7 NS 32.0±20.7 29.7±20.9 NS
Dropout rateb 110 (40%) 49 (45%) 61 (37%) NS 17 (29%) 93 (43%) 0.049
Causes of dropoutc
Death 41 (38%) 23 (47%) 18 (30%) 0.023 7 (41%) 34 (37%) NS
Shift to hemodialysis 38 (35%) 21 (43%) 17 (28%) 0.039 6 (35%) 32 (34%) NS
Transplant 26 (24%) 3 (6%) 23 (38%) 0.002 4 (24%) 22 (24%) NS
Transfer to other hospital 5 (4.5%) 2 (4%) 3 (5%) NS 0 (0%) 5 (5%) NS
NS = not significant.
aFrom the start of peritoneal dialysis to the end of follow-up. bParentheses denote percentage among total participants. cParentheses represent percentage among dropout patients.
dEarly or late start of dialysis means glomerular filtration rate ≥5 or <5 mL/minute/1.73 m2.
ePlanned or late implantation of Tenckhoff catheters means absence or presence of preceding emergent hemodialysis. Values are presented as mean±SD or number (percentage) unless otherwise specified.
TABLE 3
Baseline Laboratory Data of Patients with Different Patterns of Initiation of Dialysis and Implantation of Catheters Initiation of dialysisa Implantation of cathetersb Total (n=275) Early (n=110) Late (n=165) p Value Planned (n=58) Late (n=217) p Value BUN (mg/dL) 115.5±41.4 99.3±28.1 126.3±45.2 <0.001 111.1±23.9 116.7±45.0 NS Creatinine (mg/dL) 12.0±4.7 8.4±1.8 14.3±4.6 <0.001 11.1±3.5 12.2±5.0 0.048 GFR (mL/min/1.73 m2) 4.8±2.2 6.8±2.1 3.5±0.9 <0.001 5.0±2.4 4.8±2.1 NS Potassium (mmol/L) 4.7±1.2 4.6±0.8 4.8±1.4 NS 4.6±0.9 4.8±1.2 NS Albumin (g/dL) 3.6±0.7 3.4±0.7 3.7±0.6 0.005 3.9±0.7 3.5±0.6 <0.001 Phosphate (mg/dL) 5.8±2.5 5.0±1.8 6.3±2.7 <0.001 6.3±2.0 5.6±2.6 NS Ca×P 47.5±19.1 43.0±16.3 50.4±20.2 0.001 51.8±14.8 46.3±19.9 0.049 iPTH (pg/mL) 339.3±311.8 258.1±249.3 382.3±333.9 0.010 338.1±271.5 339.6±321.5 NS Triglycerides (mg/dL) 145.4±90.7 131.8±64.1 155.7±105.6 NS 145.7±71.5 145.3±95.5 NS Total cholesterol (mg/dL) 183.2±49.0 179.8±52.8 185.6±49.1 NS 180.7±42.7 183.8±50.7 NS WBC (×109/L) 6.9±2.3 7.3±2.8 6.6±1.8 0.026 6.1±1.7 7.1±2.4 0.001 Hemoglobin (g/dL) 8.0±2.1 8.5±1.6 7.7±2.4 0.004 8.5±2.1 7.9±2.1 NS CRP (mg/L) 2.2±2.4 2.0±2.4 2.4±2.5 NS 1.7±3.2 2.1±3.5 NS
Urine amount (L/day) 0.63±0.55 0.63±0.58 0.62±0.54 NS 0.9±0.6 0.6±0.5 <0.001
Total Kt/V 2.4±1.3 2.3±0.6 2.4±1.6 NS 2.5±0.6 2.3±1.5 NS Total WCCr (L/week) 63.0±20.2 66.9±24.6 60.3±16.2 0.014 70.5±21.6 61.0±19.4 0.001 Peritoneal Kt/V 1.8±0.4 1.7±0.4 1.8±0.5 0.011 1.7±0.6 1.8±0.4 NS Peritoneal WCCr (L/week) 39.5±10.7 39.9±11.2 39.3±10.3 NS 38.7±13.5 39.7±9.8 NS Renal Kt/V 0.6±0.9 0.7±0.8 0.5±0.9 NS 0.7±0.5 0.5±1.0 NS Renal WCCr (L/week) 19.9±18.0 24.7±21.9 16.7±14.1 0.001 27.3±17.5 17.9±17.7 <0.001 nPCR (g/kg/day) 1.12±0.26 1.10±0.28 1.13±0.24 NS 1.19±0.26 1.10±0.25 0.010
BUN = blood urea nitrogen; GFR = glomerular filtration rate; Ca×P = product of calcium and phosphate; iPTH = intact parathyroid hormone; WBC = white blood cell count; CRP = C-reactive protein; Kt/V = urea clearance; WCCr = weekly creatinine clearance; nPCR = normalized protein catabolic rate; NS = not significant.
Values are presented as mean±SD unless otherwise stated.
aEarly or late start of dialysis means glomerular filtration rate ≥5 or <5 mL/minute/1.73 m2.
bPlanned or late implantation of Tenckhoff catheters means absence or presence of preceding emergent hemodialysis.
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
renal transplantation (p = 0.002) during the period of observation (Table 2).
TIMING OF IMPLANTATION OF CATHETERS AND PATIENT OUTCOMES AFTER START OF CHRONIC PD
The timing of implantation of the PD catheter did not affect patient survival, but was associated with a reduced risk for all-cause hospitalization (log rank, p = 0.013) (Figure 1). This difference in hospitalization rate per-sisted throughout the period of observation. Compared with the late implantation group, patients in the planned implantation group had higher education lev-els (p = 0.015) and suffered less comorbidity of coro-nary heart disease (p = 0.046). They were referred to nephrologists earlier (p < 0.001) and were occupation-ally more active (p = 0.001) (Table 1). The planned im-plantation group had higher serum albumin (p < 0.001), nPCR (p = 0.01), calcium–phosphate product (p = 0.049), and renal WCCr (p < 0.001) and lower white blood cell count (p = 0.001) at study baseline (Table 3). They also had a marginally lower rate of dropout during the
fol-low-up although no apparent causes could be identified (Table 2).
TIMING OF NEPHROLOGY REFERRAL AND PATIENT OUTCOMES AFTER START OF CHRONIC PD
Unlike initiation of dialysis and implantation of cath-eters, the timing of referral to nephrologists did not have an appreciable impact on either patient survival or hos-pitalization rate (Figure 1). By contrast, early nephrol-ogy referral (p < 0.001) and higher occupational activity (p = 0.009), but not early initiation of dialysis, were sig-nificantly associated with planned catheter implantation (Table 4).
COX PROPORTIONAL HAZARDS MODEL FOR PREDICTORS OF PD PATIENT OUTCOME
The results of Cox regression analysis for overall mor-tality and hospitalization are shown in Tables 5 and 6 respectively. The tables show the results when variables were analyzed first in the univariate analysis and also in
Figure 1 — Kaplan–Meier survival curves for overall mortality, as cumulative patient survival (A), and hospitalization, as cumula-tive hazards of hospitalization (B), in patients with different patterns of predialysis care.
A
B
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
the final backward stepwise model; p < 0.2 on univariate analysis was considered important. The predictors for overall mortality include baseline GFR (HR 1.18, p = 0.023), age (HR 1.07, p < 0.001), and diabetes (HR 3.64, p = 0.001). The independent factors associated with all-cause hospitalization include age (HR 1.02, p = 0.012),
late catheter implantation (HR 1.78, p = 0.011), and dia-betes (HR 1.92, p = 0.005). Although baseline GFR was not statistically significant in the backward stepwise model: [HR 1.08, 95% confidence interval (CI) 0.999 – 1.160; p = 0.054], it was significant when using the single-step model (HR 1.09, 95% CI 1.001 – 1.181; p = 0.048) (data not shown).
DISCUSSION
This study demonstrates that early start of PD, to-gether with diabetes and old age, are strong predictors for overall mortality after commencement of chronic PD. Additionally, late implantation of Tenckhoff catheters, diabetes, old age, and probably early start of PD are in-dependent risk factors for all-cause hospitalization after initiation of chronic PD.
The timing of initiation of dialysis and its impact on patient outcomes has been a subject of debate. Earlier observations suggested that higher GFR at initiation of dialysis was associated with better patient outcomes, including hospitalization and patient survival (14). Sub-TABLE 5
Independent Predictors for Overall Mortality Using Cox Proportional Hazards Model
Univariate Backward logistical regression
Variablea (reference group) HR 95% CI p Value HR 95% CI p Value
Late implantation of catheterb (hazard for
planned implantation=1.0) 1.643 0.689–3.915 NS — — —
Late referral to nephrologistc (hazard for early referral=1.0) 1.901 0.949–3.809 NS — — —
Male (hazard for female=1.0) 1.877 1.008–3.533 0.047 — — —
Diabetes (hazard for nondiabetic=1.0) 6.733 3.579–12.668 <0.001 3.644 1.697–7.825 0.001
CAD (hazard for non-CAD patient=1.0) 3.628 1.600–8.225 0.002 — — —
Age (per 1 year) 1.083 1.058–1.108 <0.001 1.072 1.043–1.095 <0.001
Albumin (per g/dL)d 0.471 0.286–0.776 0.003 — — —
GFR (per mL/minute/1.73 m2)d 1.168 1.061–1.287 0.002 1.182 1.023–1.366 0.023
Hemoglobin (per g/dL)d 1.152 1.034–1.284 0.010 — — —
WBC (per 1×109/L)d 1.191 1.067–1.330 0.002 — — —
Total Kt/V (per 1 unit)e 0.426 0.260–0.697 0.001 — — —
Total WCCr (per L/week)e 0.977 0.960–0.994 0.007 — — —
Renal Kt/V (per 1 unit)e 0.312 0.119–0.819 0.018 — — —
Renal WCCr (per L/week)e 0.978 0.957–1.000 NS — — —
nPCR (per g/kg/day)e 0.055 0.012–0.254 <0.001 — — —
CAD = coronary artery disease; GFR = glomerular filtration rate; WBC = white blood cell; Kt/V = urea clearance; WCCr = weekly creatinine clearance; nPCR = normalized protein catabolic rate; HR = hazard ratio; 95% CI = 95% confidence interval; NS = not significant.
aThe independent variables were selected for multivariate analysis if they had p < 0.2 on univariate analysis or because they were considered to be important.
bPlanned or late implantation of Tenckhoff catheters means absence or presence of preceding emergent hemodialysis. cEarly or late referral to nephrologist means ≥6 or <6 months of dialysis.
dThe last biochemical data before initiation of chronic peritoneal dialysis.
eThe earliest data of peritoneal equilibration tests after initiation of chronic peritoneal dialysis.
TABLE 4
Factors Associated with Planned Implantation of Catheters (Logistic Regression Analysis)
Variable B SE χ2 p Value
Age at enrollment –0.010 0.012 0.990 NS
Female gender 0.191 0.354 1.210 NS
Higher education level 0.651 0.376 1.918 NS Occupationally active 1.285 0.490 3.615 0.009 Early nephrology referral 2.030 0.383 7.616 <0.001 Early dialysis start 0.618 0.358 1.856 NS
Diabetes –0.004 0.556 0.996 NS
CAD –19.079 9729.741 <0.001 NS
CAD = coronary artery disease; B = estimated coefficient; SE = standard error; NS = not significant.
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
sequent studies, however, argued that any survival ben-efit from early initiation of dialysis could be accounted for by lead-time bias, and they did not support initiat-ing dialysis therapy for patients with significant residual renal function (9,10). In this report, the patients start-ing chronic PD at higher GFR levels (i.e., earlier start of PD) exhibited a higher dropout rate and an increased risk of overall death. They also tended to have more hospi-talization during the follow-up. A plausible explanation for this seeming paradox could be inferred from Tables 1, 2, and 3. Noticeably, patients in the early start group were older and occupationally less active, suffered more comorbidity (diabetes and congestive heart failure), and had poorer nutritional status (lower serum albumin and phosphate) at initiation of dialysis. Some of these fac-tors, such as older age and diabetes, are well-known prognostic factors for death in patients on chronic PD (15). These features of the early-start patients also may be causally related to fewer dropouts from kidney trans-plantation in this group. Collectively, our results were consistent with the report by Beddhu et al. (11), which showed that patients initiating chronic dialysis with sig-nificant residual renal function were associated with
higher mortality, even when other confounders were taken into consideration. They suggested that low serum creatinine levels, as seen in those patients, could be a marker of reduced muscle mass and unhealthy nutri-tional status, which in turn might contribute to the poor survival. Future research should elucidate whether this phenomenon may represent a part of the “reverse epi-demiology” in the dialysis population (16).
The present study shows for the first time that patients receiving planned implantation of Tenckhoff catheters, namely those that started chronic PD without preceding emergent HD, were associated with reduced risk of over-all hospitalization after initiation of PD. Because emer-gent dialysis via temporary HD catheters has been linked with an increased risk of sepsis (6), one might argue that the lower hospitalization rate seen in the planned im-plantation group was due to the lack of emergent HD before the start of chronic PD. However, our cohort re-cruited patients that had been on PD for more than 3 months, at which point none still had a HD catheter in place. Further, the difference in hospitalization rates retained significance not only immediately but also months after the start of chronic PD (Figure 1). Thus, it TABLE 6
Independent Predictors for All-Cause Hospitalization Using Cox Proportional Hazards Model
Univariate Backward logistical regression
Variablea (reference group) HR 95% CI p Value HR 95% CI p Value
Late implantation of catheterb (hazard for
planned implantation=1.0) 1.694 1.111–2.583 0.014 1.783 1.142–2.784 0.011
Late referral to nephrologistc (hazard for early referral=1.0) 1.294 0.939–1.784 NS — — —
Male (hazard for female=1.0) 1.297 0.944–1.784 NS — — —
Diabetes (hazard for nondiabetic=1.0) 2.654 1.832–3.844 <0.001 1.922 1.225–3.017 0.005
CAD (hazard for non-CAD patient=1.0) 1.065 0.560–2.023 NS — — —
Age (per 1 year) 1.022 1.011–1.032 <0.001 1.015 1.003–1.028 0.012
Albumin (per g/dL)d 0.784 0.606–1.013 NS — — —
GFR (per mL/minute/1.73 m2)d 1.075 1.011–1.142 0.020 — — —
Hemoglobin (per g/dL)d 0.996 0.923–1.076 NS — — —
WBC (per 1×109/L)d 1.102 1.032–1.176 0.004 — — —
Total Kt/V (per 1 unit)e 0.806 0.615–1.058 NS — — —
Total WCCr (per L/week)e 0.995 0.987–1.003 NS — —
Renal Kt/V (per 1 unit)e 0.876 0.669–1.147 NS — — —
Renal WCCr (per L/week)e 0.995 0.986–1.004 NS — — —
nPCR (per g/kg/day)e 0.700 0.363–1.350 NS — — —
CAD = coronary artery disease; GFR = glomerular filtration rate; WBC = white blood cell count; Kt/V = urea clearance; WCCr = weekly creatinine clearance; nPCR = normalized protein catabolic rate; HR = hazard ratio; CI = 95% confidence interval; NS = not significant. aThe independent variables were selected for multivariate analysis if they had p < 0.2 on univariate analysis or because they were
considered to be important.
bPlanned or late implantation of Tenckhoff catheters means absence or presence of preceding emergent hemodialysis. cEarly or late referral to nephrologist means ≥6 or <6 months of dialysis.
dThe last biochemical data before initiation of chronic peritoneal dialysis.
eThe earliest data of peritoneal equilibration tests after initiation of chronic peritoneal dialysis.
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
seems unlikely that complications associated with the use of HD catheters could explain the different hospi-talization rates between patients with planned and pa-tients with late catheter implantation. Our logistic regression analysis showed that early nephrology refer-ral and higher occupational activity were the only two factors correlated significantly with planned implanta-tion of catheters. Based on Table 1, patients in the planned implantation group had higher levels of educa-tion and suffered less comorbidity of coronary heart dis-ease. These features could be associated with higher motivation for self-care and better compliance to dialy-sis prescriptions and nursing advice, which might con-tribute to fewer hospitalization events during the follow-up. Additionally, as shown in Table 3, the planned implantation group exhibited higher residual renal func-tion (higher urine output and renal WCCr), better nutri-tional indices (higher albumin and nPCR), and reduced inflammatory status (lower baseline white cell count). These variables have all been implicated in better pa-tient outcomes after initiation of chronic PD (17).
Early referral for patients with advanced chronic kid-ney disease has been associated with reduction of mor-bidity, mortality, and economic burden (18). The present study, however, failed to show such benefits of early nephrology referral. The retrospective nature of this study may account in part for the lack of such an asso-ciation. Conversely, the association of timing of nephrol-ogy referral with patient outcomes remains controversial (19). In the current analysis, early referral was found to be the most important factor associated with planned implantation of catheters. Indeed, as shown in Table 1, more patients in the planned implantation group (81%), compared with the late implantation group (33%), were referred earlier (>6 months) to nephrologists. It is con-ceivable that early nephrology referral allowed patients multidisciplinary pre-ESRD care, including planned im-plantation of catheters, which then leads to reduced hospitalization after the start of chronic PD. Thus, planned implantation of catheters may represent the adequacy of predialysis care, which is more important than the timing of nephrology referral. Consistent with this notion, previous studies have shown that patients with chronic kidney disease that are referred earlier to nephrologists can receive more consistent pre-ESRD care, including earlier creation of vascular access and timely use of fistulas, which is associated with a reduced risk of sepsis and death by avoiding the use of HD cath-eters (5,6).
There are limitations to the present study. First, there exists potential selection bias since 35 patients were excluded from the study after initial screening. Among
those, 7 deaths were caused by cardiovascular or infec-tious diseases, which arguably might have affected the outcome. Nevertheless, even when these patients were taken into consideration, our findings as described in the earlier sections remain the same (data not shown). Second, this study was a single-center experience. There-fore, observations accrued here might not be extrapo-lated to PD patients elsewhere. Further multicenter prospective studies using a randomized clinical trial de-sign [such as the ongoing IDEAL trial (20)] are needed to confirm our findings.
In summary, this study shows that early start of di-alysis adversely affects patient outcomes, whereas planned implantation of catheters reduces hospitaliza-tion. Our data do not support earlier initiation of PD, but underscore the importance of planned implantation of catheters to reduce hospitalization after the start of chronic PD.
ACKNOWLEDGMENTS
This work was supported by grants from the Ta-Tung Kidney Foundation and the Mrs. Hsiu-Chin Lee Kidney Research Fund, Taipei, Taiwan.
REFERENCES
1. U.S. Renal Data System. Excerpts from the USRDS 2006 Annual Data Report. Am J Kidney Dis 2007; 49(Suppl 1): S1–296.
2. Department of Health. National Health Insurance Annual Statistical Report. Taiwan, ROC: Department of Health; 2005. Available at: http://www.doh.gov.tw/ufile/doc/ 940506.xls [Accessed 12 February 2007].
3. Stack AG. Impact of timing of nephrology referral and pre-ESRD care on mortality risk among new pre-ESRD patients in the United States. Am J Kidney Dis 2003; 41:310–18. 4. Kazmi WH, Obrador GT, Khan SS, Pereira BJ, Kausz AT. Late
nephrology referral and mortality among patients with end-stage renal disease: a propensity score analysis. Nephrol Dial Transplant 2004; 19:1808–14.
5. Ravani P, Marinangeli G, Tancredi M, Malberti F. Multi-disciplinary chronic kidney disease management improves survival on dialysis. J Nephrol 2003; 16:870–7.
6. Oliver MJ, Rothwell DM, Fung K, Hux JE, Lok CE. Late cre-ation of vascular access for hemodialysis and increased risk of sepsis. J Am Soc Nephrol 2004; 15:1936–42. 7. National Kidney Foundation. K/DOQI Clinical practice
guidelines for chronic kidney disease: evaluation, classi-fication, and stratif ication. Am J Kidney Dis 2002; 39(Suppl 1):S1–266.
8. Ravani P, Brunori G, Mandolfo S, Cancarini G, Imbasciati E, Marcelli D, et al. Cardiovascular comorbidity and late referral impact arteriovenous fistula survival: a
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
tive multicenter study. J Am Soc Nephrol 2004; 15:204–9. 9. Korevaar JC, Jansen MA, Dekker FW, Jager KJ, Boeschoten EW, Krediet RT, et al. Netherlands Cooperative Study on the Adequacy of Dialysis Study Group. When to initiate dialysis: effect of proposed US guidelines on survival. Lancet 2001; 358:1046–50.
10. Traynor JP, Simpson K, Geddes CC, Deighan CJ, Fox JG. Early initiation of dialysis fails to prolong survival in pa-tients with end-stage renal failure. J Am Soc Nephrol 2002; 13:2125–32.
11. Beddhu S, Samore MH, Roberts MS, Stoddard GJ, Ramkumar N, Pappas LM, et al. Impact of timing of initia-tion of dialysis on mortality. J Am Soc Nephrol 2003; 14: 2305–12.
12. Dombros N, Dratwa M, Feriani M, Gokal R, Heimbürger O, Krediet R, et al. EBPG Expert Group on Peritoneal Dialy-sis. European best practice guidelines for peritoneal di-alysis. 2 The initiation of didi-alysis. Nephrol Dial Transplant 2005; 20(Suppl 9):ix3–ix7.
13. Wilson B, Harwood L, Locking-Cusolito H, Chen SJ, Heidenheim P, Craik D, et al. Optimal timing of initiation of chronic hemodialysis? Hemodial Int 2007; 11:263–9. 14. Churchill DN. An evidence-based approach to earlier
ini-tiation of dialysis. Am J Kidney Dis 1997; 30:899–906. 15. Paniagua R, Amato D, Vonesh E, Correa-Rotter R, Ramos
A, Moran J, et al. Mexican Nephrology Collaborative Study Group. Effects of increased peritoneal clearances on mor-tality rates in peritoneal dialysis: ADEMEX, a prospective, randomized, controlled trial. J Am Soc Nephrol 2002; 13: 1307–20.
16. Kalantar-Zadeh K, Block G, Humphreys MH, Kopple JD. Reverse epidemiology of cardiovascular risk factors in maintenance dialysis patients. Kidney Int 2003; 63: 793–808.
17. Lo WK. Serum parameters, inflammation, renal function and patient outcome. Contrib Nephrol 2006; 150:152–5. 18. Powe NR. Early referral in chronic kidney disease: an
enor-mous opportunity for prevention. Am J Kidney Dis 2003; 41:505–7.
19. Roubicek C, Brunet P, Huiart L, Thirion X, Leonetti F, Dussol B, et al. Timing of nephrology referral: influence on mor-tality and morbidity. Am J Kidney Dis 2000; 36:35–41. 20. Cooper BA, Branley P, Bulfone L, Collins JF, Craig JC,
Dempster J, et al. IDEAL Study Steering Committee. The Initiating Dialysis Early and Late (IDEAL) Study: study ra-tionale and design. Perit Dial Int 2004; 24:176–81.
at NTU MEDICAL LIBRARY on April 29, 2009
www.pdiconnect.com
