Comparison of Patient Survival Between Hemodialysis and Peritoneal Dialysis Among Patients Eligible for Both Modalities

Original Investigation

Comparison of Patient Survival Between Hemodialysis and Peritoneal Dialysis Among Patients Eligible for Both Modalities Ben Wong, Pietro Ravani, Matthew J. Oliver, Jayna Holroyd-Leduc, Lorraine Venturato, Amit X. Garg, and Robert R. Quinn Background: Although peritoneal dialysis (PD) costs less to the health care system compared to in-center hemodialysis (HD), it is an underused therapy. Neither modality has been consistently shown to confer a clear benefit to patient survival. A key limitation of prior research is that study patients were not restricted to those eligible for both therapies. Study Design: Retrospective cohort study. Setting & Participants: All adult patients developing end-stage renal disease from January 2004 to December 2013 at any of 7 regional dialysis centers in Ontario, Canada, who had received at least 1 outpatient dialysis treatment and had completed a multidisciplinary modality assessment. Predictor: HD or PD. Outcomes: Mortality from any cause. Results: Among all incident patients with endstage renal disease (1,579 HD and 453 PD), PD was associated with lower risk for death

among patients younger than 65 years. However, after excluding approximately one-third of all incident patients deemed to be ineligible for PD, the modalities were associated with similar survival regardless of age. This finding was also observed in analyses that were restricted to patients initiating dialysis therapy electively as outpatients. The impact of modality on survival did not vary over time. Limitations: The determination of PD eligibility was based on the judgment of the multidisciplinary team at each dialysis center.

Correspondence to B. Wong (bcw@ualberta. net) Am J Kidney Dis. 71(3): 344-351. Published online November 22, 2017. doi: 10.1053/ j.ajkd.2017.08.028

© 2017 by the National Kidney Foundation, Inc.

Conclusions: HD and PD are associated with similar mortality among incident dialysis patients who are eligible for both modalities. The effect of modality on survival does not appear to change over time. Future comparisons of dialysis modality should be restricted to individuals who are deemed eligible for both modalities to reflect the outcomes of patients who have the opportunity to choose between HD and PD in clinical practice.

M

anagement of end-stage renal disease (ESRD) is resource intensive, and the cost of caring for patients with ESRD is largely driven by the provision of maintenance.1 In 2013, the United States spent more than $25 billion (w7% of all Medicare expenditure) caring for pa-

Editorial, p. 309 tients who had ESRD, although this cohort made up <0.5% of the Medicare population.1 Peritoneal dialysis (PD) affords cost savings to the health care system compared to conventional in-center hemodialysis (HD),2,3 but remains an underused therapy.1 As a result, many jurisdictions have introduced strategies to increase the use of PD, including implementation of the ESRD Prospective Payment System.4,5 However, there remains equipoise regarding the impact of dialysis modality on patient survival. The single randomized controlled trial comparing the 2 therapies was unable to meet recruitment targets because patients developed strong preferences for a particular dialysis modality when they were educated about their treatment options, and it is unlikely that another randomized controlled trial will be completed.6 Numerous observational studies comparing PD and HD have been conducted in this area with mixed results.6-15 Given that randomized 344

Complete author and article information provided before references.

comparisons do not appear to be possible, strategies to minimize the potential for bias in observational studies comparing different dialysis therapies are important to inform clinical practice. One key limitation of previous works is that comparisons were not restricted to patients who were eligible for both therapies—the population faced with a decision between the 2 therapies in clinical practice.16 This may have led to biased results because patients who are not eligible for PD tend to have worse health status than those who are eligible for PD.6,17-20 The primary objective of this study was to compare survival in incident patients with ESRD treated with HD or PD who were eligible for both therapies. As a secondary objective, we attempted to quantify the impact of including ineligible patients on mortality in prior comparisons of HD and PD. Methods Overview We conducted a retrospective cohort study using deidentified administrative health data stored at the Institute for Clinical Evaluative Sciences (ICES) in Toronto, Canada. The Conjoint Health Research Ethics Board at the University of Calgary and the Institutional Review Board at Sunnybrook AJKD Vol 71 | Iss 3 | March 2018

Original Investigation Health Sciences Center, Toronto, Canada, approved the study. Informed consent was waived due to deidentified information. Data Sources We used data collected from January 2004 through December 2013 from the 7 centers that participated in the Dialysis Measurement Analysis and Reporting (DMAR) system (Sunnybrook Health Sciences Center, Halton Healthcare, London Health Sciences Center, Grand River Hospital, Sault Area Hospital, William Osler Health Center, and The Ottawa Hospital). The DMAR system prospectively collects high-quality data for the purposes of quality improvement, using a web-based data collection platform. Data elements include detailed information for demographics, comorbid conditions, laboratory values, history of predialysis care, and acuity of dialysis therapy initiation. All dialysis modality changes, hospitalizations, vascular access procedures, transplantations, patients lost to follow-up, transfers out of the program, and deaths are captured. Data are entered by trained front-line staff using a standardized coding scheme. To ensure data quality, all data elements collected at the participating sites are double-reviewed by the same 2 investigators (R.R.Q. and M.J.O.) and queries are communicated to end users to be rectified. Using encrypted versions of patients’ unique health insurance numbers, data from the Canadian Organ Replacement Register and Registered Persons Database were linked to DMAR to obtain additional information (self-reported race, primary kidney disease, and socioeconomic status) that may confound the relationship between dialysis modality and survival. Patient Population Consecutive incident patients 18 years or older initiating dialysis therapy at each participating center were included if they: (1) had a diagnosis of ESRD confirmed by a nephrologist, (2) had received at least 1 outpatient dialysis treatment, and (3) had completed a multidisciplinary modality assessment. A multidisciplinary team including a nephrologist, predialysis nurse(s), PD and/or acute care nurse(s), and sometimes a social worker met every 2 weeks at the respective regional dialysis programs to review incident dialysis patients and determine eligibility for HD and PD using a structured assessment (Item S1). The team ensured that patients were educated about their treatment options and offered the therapies they were candidates for so that they could make an informed choice. Patients who had previously received a kidney transplant and those who had recovered kidney function within 180 days of dialysis therapy initiation were excluded. We also excluded those with less than 6 months of potential follow-up, those who had had significant gaps (>1 month) in follow-up due to temporary transfer out of the program, and those who had had significant gaps in dialysis treatment of more than 31 days. Potential follow-up for 6 months means that we were able to report on a patient’s AJKD Vol 71 | Iss 3 | March 2018

status 6 months following the initiation of dialysis therapy regardless of whether they were still receiving dialysis or had died, recovered, or transferred out of a program. We excluded patients with less than 6 months of “potential follow-up” to ensure that minimum follow-up on dialysis therapy was 6 months. However, if patients died or stopped dialysis therapy early for transplantation, they were included. This is in contrast to actual follow-up time, which is measured from the initiation date of dialysis therapy until a patient dies or has a censoring event. A minimum of 6 months of potential follow-up was chosen to ensure adequate time to declare modality choice, observe for clinical outcomes, and minimize the risk for bias introduced by urgent dialysis therapy starts, for which patients are preferentially treated with HD and have a poor prognosis. We constructed 3 different patient cohorts: (1) all patients who had completed modality assessment, regardless of their eligibility for PD, to mirror the population used in traditional analyses (traditional cohort); (2) patients deemed eligible for both dialysis modalities to reflect those faced with a modality choice in clinical practice (eligible cohort); and (3) patients deemed eligible for both modalities who initiated dialysis therapy electively as outpatients to determine whether exclusion of patients who initiated dialysis therapy in the hospital affected results (eligible outpatient cohort). The process of determining PD eligibility has been described previously in detail.18 Statistical Analyses We used standard methods for descriptive statistics and group comparison (PD vs HD). Frequency was reported for qualitative variables and mean ± standard deviation or median with range were reported, as appropriate, for quantitative variables. We screened for collinearity using the variance inflation factor.21 The initial outpatient dialysis treatment modality (PD vs HD) was the primary exposure of interest. Patients were followed up from the date of the first outpatient dialysis treatment for all-cause mortality. Follow-up was censored at the occurrence of transplantation, loss to follow-up, recovery of kidney function, or end of the study period, but not with changes in dialysis modality. To account for patients who initiated HD therapy urgently but subsequently switched to PD therapy because PD was their modality choice, we performed a sensitivity analysis on the eligible cohort in which patients who converted from HD to PD within 6 months of initiating dialysis therapy were removed from the analysis. Survival methods were used to compare mortality according to dialysis modality in the 3 separate cohorts. We used Cox regression to adjust for covariates that may have confounded the relationship between dialysis modality and all-cause mortality. These included demographic variables (age stratified into <65, 65-74, and ≥75 years; sex; and selfreported race), socioeconomic status as defined by neighborhood income quintile, primary kidney disease (diabetes, 345

Original Investigation hypertension, glomerulonephritis, polycystic kidney disease, and other), inpatient dialysis therapy initiation, comorbid medical conditions (presence of diabetes, coronary artery disease, congestive heart failure, cerebrovascular disease, malignancy, and peripheral vascular disease), receipt of a minimum of 4 months of predialysis care, and baseline laboratory variables (hemoglobin, creatinine, and albumin). Neighborhood income quintile is a household size–adjusted measure of household income on the basis of the 2006 Canadian Census data. Quintiles were defined within each neighborhood and not across the entire province to minimize the effect of large differences in housing costs and ensure an equal percentage of the population in each income quintile. We also included prespecified interaction terms (age × modality, sex × modality, and diabetes × modality) containing variables that may modify the relationship between dialysis modality and survival.7-12 Interaction terms were retained only if they were significant at P < 0.05. We tested the proportional hazards assumption using formal tests based on Schoenfeld residuals, and graphically, using log-log plots for categorical covariates, and plotting the slope of the scaled Schoenfeld residuals over (log) time for quantitative variables. For the patient cohort in which the proportional hazards requirement was violated, we performed further Cox analysis stratified by age and if necessary by follow-up time (before and after year 3).

All analyses were conducted using Stata, version 13.1 (StataCorp LP). Results Study Population There were 2,146 patients who had confirmed ESRD and had received at least 1 outpatient dialysis treatment after the exclusion of certain patients. Patients were excluded for the following reasons: previous transplant (148 [6%]: 133 HD, 15 PD), less than 6 months of potential follow-up (160 [6%]: 120 HD, 40 PD), significant gaps (>1 month) in follow-up due to temporary transfer out of the program (110 [4%]: 100 HD, 10 PD), significant gaps in dialysis treatment of more than 31 days (13 [0.5%]: 3 HD, 10 PD), and recovered kidney function within 6 months (65 [2%]: 64 HD, 1 PD). Of these, 114 (5%) could not be assessed for PD (Fig 1). Of the remaining patients, 460 (23%) had a contraindication to PD and an additional 196 (10%) had barriers to PD that could not be overcome with support. These barriers are categorized into 4 groups: medical (eg, nocturia), physical (eg, frailty), social (eg, small living space), and cognitive (eg, learning disability; Fig 1). The traditional cohort consisted of 2,032 patients, including 1,579 HD and 453 PD patients. Patients were followed up for a median of 520 days. Overall, HD patients

2146 paƟents with confirmed ESRD and had received at least one outpaƟent dialysis treatment 114 (5%) could not complete modality assessment • • • • •

44 37 20 7 6

Refused to parƟcipate Transferred out of program Died Other PalliaƟve

460 (23%) had a contraindicaƟon to PD

TradiƟonal cohort (N=2032)

196 (10%) had at least one barrier to PD that could not be overcome with support (categories are not mutually exclusive) • 138 Medical • 142 Physical • 113 Social • 109 CogniƟve

Eligible cohort (N=1376)

Medical 86 Abdominal scarring 50 Obesity 20 Abdominal aneurysm 19 Cirrhosis 16 Hernia not repairable 15 Ileostomy 14 Colostomy 13 Large polycysƟc kidneys 12 Bowel cancer 10 DiverƟculiƟs 10 Ileal conduit 8 Future abdominal surgery 8 Inflammatory bowel disease 48 Other * Social 64 18 12 10 27

Nursing home ReƟrement home No permanent residence Complex conƟnuing care Other %

OutpaƟent Eligible cohort (N=874) Figure 1. A total of 2,146 patients met inclusion criteria; 124 patients were unable to complete dialysis modality assessment. Of the remaining patients, 460 (23%) had a contraindication to peritoneal dialysis (PD) and 196 (10%) had barriers to PD that could not be overcome with support. *Bowel obstruction, diarrhea, gastrostomy tube, gastroparesis, incontinence, ischemic gut, and nephrotic syndrome. %Rehabilitation, small living space, and employment. Abbreviation: ESRD, end-stage renal disease.

346

AJKD Vol 71 | Iss 3 | March 2018

Traditional Cohort (n = 2,032) Variable

HD

PD

Total no. of patients Age, y Female sex Race White Asian Black Other Unknown Income quintile 1 2 3 4 5 Primary kidney disease Diabetes Glomerulonephritis Other Polycystic kidney disease Renal vascular Inpatient start Diabetes Coronary artery disease Congestive heart failure Cerebrovascular disease Malignancy Peripheral vascular disease Received predialysis care ≥4 mo Hemoglobin, g/dL eGFR, mL/min/1.73 m2 Albumin, g/dL

1,579 66.8 ± 15.3 40.5%

453 64.8 ± 14.8 40.2%

72.7% 4.6% 4.4% 10.1% 8.2%

68.4% 5.1% 6.0% 12.8% 7.7%

23.2% 18.8% 19.6% 19.6% 18.8%

19.0% 22.5% 21.0% 19.0% 18.5%

37.0% 18.2% 23.9% 3.2% 17.7% 54.1% 55.0% 37.3% 31.0% 19.1% 21.6% 19.7% 78.1% 9.7 ± 2.0 9.3 ± 10.1 3.3 ± 1.8

34.2% 18.1% 19.6% 6.2% 21.9% 9.9% 46.4% 26.3% 16.8% 9.9% 14.1% 10.6% 94.9% 10.7 ± 1.4 8.6 ± 3.3 3.8 ± 0.5

Eligible Cohort (n = 1,376) P 0.01 0.9 0.3

HD

PD

926 65.3 ± 16.0 38.3%

450 64.7 ± 14.8 40.4%

68.6% 5.9% 4.8% 12.6% 8.1%

68.2% 5.1% 6.0% 12.9% 7.8%

22.6% 20.4% 19.0% 17.5% 20.5%

19.1% 22.7% 21.1% 18.7% 18.4%

35.5% 19.3% 23.9% 2.9% 18.4% 49.8% 52.7% 35.2% 28.3% 17.3% 17.6% 16.7% 80.2% 9.6 ± 1.7 8.8 ± 9.1 3.5 ± 2.2

34.4% 17.8% 19.6% 6.2% 22.0% 9.1% 46.0% 26.0% 16.0% 9.8% 14.0% 10.7% 95.1% 10.7 ± 1.4 8.5 ± 3.1 3.8 ± 0.5

0.2

P 0.5 0.5 0.9

HD

PD

465 64.1 ± 16.2 35.7%

409 64.1 ± 14.4 40.6%

70.8% 4.5% 3.2% 11.8% 9.7%

68.0% 5.4% 6.1% 13.2% 7.3%

23.0% 17.6% 19.4% 18.1% 21.9%

18.1% 23.0% 22.0% 19.3% 17.6%

32.9% 19.1% 24.7% 4.7% 18.5% — 50.3% 30.8% 17.2% 13.8% 18.7% 14.2% 91.2% 10.0 ± 1.6 8.6 ± 3.4 3.6 ± 0.6

35.2% 18.3% 19.6% 6.1% 20.8% — 45.7% 24.5% 13.5% 9.1% 14.2% 9.8% 96.6% 10.8 ± 1.4 8.4 ± 3.0 3.8 ± 0.5

0.4

0.006

<0.001 0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 <0.001 0.02 <0.001

Eligible Outpatient Cohort (n = 874)

Note: Values for categorical variables are given as percentages; values for continuous variables, as mean ± standard deviation. Abbreviations: eGFR, estimated glomerular filtration rate; HD, hemodialysis; PD, peritoneal dialysis.

0.9 0.1 0.2

0.07

0.01

<0.001 0.02 0.001 <0.001 <0.001 0.1 0.003 <0.001 <0.001 0.4 <0.001

P

0.4

— 0.2 0.04 0.1 0.03 0.08 0.05 0.001 <0.001 0.3 <0.001

Original Investigation

AJKD Vol 71 | Iss 3 | March 2018

Table 1. Baseline Patient Characteristics as per Cohort Definition

347

Original Investigation were older and had a higher frequency of diabetes, coronary artery disease, congestive heart failure, cerebrovascular disease, malignancy, and peripheral vascular disease (Table 1). They also initiated dialysis therapy with lower hemoglobin and albumin values, had more inpatient dialysis therapy starts, and were less likely to have received at least 4 months of predialysis care. The eligible cohort consisted of 1,376 patients (926 HD and 450 PD patients), representing 68% of those who completed modality assessment and were deemed eligible for both dialysis modalities (Table 1). Patients were followed up for a median of 547 days. Among eligible patients, HD patients had a higher burden of comorbid conditions and were more likely to initiate dialysis therapy in an inpatient setting. There was a high proportion of central venous catheter (CVC) use, with 84% of HD patients dialyzing via a CVC. The eligible outpatient cohort consisted of 874 patients (465 HD and 409 PD) who initiated dialysis therapy electively, as outpatients. Patients were followed up for a median of 564 days. Baseline characteristics between HD and PD patients in this cohort were more homogeneous (Table 1). Use of a CVC remained high; 73% of HD patients dialyzed via a CVC. Survival Data In the traditional cohort (n = 2,032), there were 628 (31%) deaths, 530 of which occurred in the HD group and 98 of which occurred in the PD group, corresponding to event rates of 0.65 and 0.42 deaths per 1,000 patient-days, respectively (Table 2). We found a significant interaction between age and dialysis modality (P = 0.02) and a timevarying association between dialysis modality and mortality (before and after year 3). There was no statistically significant difference in all-cause mortality between HD and PD in the elderly (65-74 and ≥75 years; Figs 2 and S1). However, in those younger than 65 years, PD was associated with significantly lower risk for death when compared to HD in the first 3 years of dialysis therapy (adjusted hazard ratio for PD vs HD [HRPD:HD] = 0.60; 95% confidence interval [CI], 0.42-0.86; Figs 2 and S2). In the eligible cohort (n = 1,376), there were 333 (24%) deaths, 239 of which occurred in the HD group and

94 of which occurred in the PD group, corresponding to an event rate of 0.47 and 0.38 deaths per 1,000 patientdays, respectively (Table 2). We found the same significant interaction between age and dialysis modality (P = 0.02). Age was again identified as an effect modifier. The effect of dialysis modality on survival did not vary over time, and PD and HD were associated with a similar risk for death (adjusted HRPD:HD, 1.08; 95% CI, 0.82-1.42; Figs 2 and S3). In the eligible outpatient cohort (n = 874), there were 186 (21%) deaths, 107 of which occurred in the HD group and 79 of which occurred in the PD group, corresponding to event rates of 0.41 and 0.34, respectively (Table 2). We found that none of the prespecified interaction terms were significant (age × modality: P = 0.07; sex × modality, P = 0.4; and diabetes × modality, P = 0.3). People treated with PD and HD had similar risks for allcause mortality (adjusted HRPD:HD, 1.19; 95% CI, 0.861.65), with constant estimates over time (Figs 2 and S4). Sensitivity Analysis When patients in the eligible cohort who converted from HD to PD therapy within 6 months of initiating dialysis therapy were removed from the analysis, none of the prespecified interaction terms were significant. Again, the effect of dialysis modality on survival did not vary over time, and PD and HD were associated with a similar risk for mortality (adjusted HRPD:HD, 1.09; 95% CI, 0.82-1.45).

Discussion In this analysis comparing HD and PD, we attempted to make the 2 populations more comparable by including only patients deemed eligible for PD after a multidisciplinary assessment. We initially found that there was no difference in survival between the 2 therapies in patients older than 65 years, but that PD was associated with lower risk for death in younger patients in analyses in the traditional cohort. However, approximately one-third of all incident dialysis patients were deemed ineligible for PD. When these patients were excluded, we found that HD and PD were associated with similar survival in incident dialysis patients, regardless of age. The effect was robust in

Table 2. Survival Data Pertaining to Various Cohorts Cohort Traditional Overall Age < 65 y, before y 3 Age < 65 y, after y 3 Age ≥ 65 y, before y 3 Age ≥ 65 y, after y 3 Eligible Eligible outpatient

HD Event Ratea

PD Event Ratea

Unadjusted HRPD:HD (95% CI)

0.65

0.42

0.59 (0.48-0.73)

0.47 0.41

0.38 0.34

0.76 (0.60-0.97) 0.83 (0.62-1.11)

Adjusted HRPD:HD (95% CI)

0.60 1.45 0.91 1.54 1.08 1.19

(0.42-0.86) (0.74-2.86) (0.69-1.19) (0.93-2.50) (0.82-1.42) (0.86-1.65)

Abbreviations: CI, confidence interval; HD, hemodialysis; HR PD:HD, hazard ratio for PD vs HD; PD, peritoneal dialysis. a Number of deaths per 1,000 patient-days.

348

AJKD Vol 71 | Iss 3 | March 2018

Original Investigation

Figure 2. Adjusted risk for death (peritoneal dialysis vs hemodialysis) according to cohort definition. Abbreviation: HR, hazard ratio.

analyses that were restricted to patients initiating dialysis therapy electively, as outpatients. In addition, the impact of modality on survival did not vary over time. In this study, we assessed the association between patient survival and dialysis modality restricting the analysis to patients who were eligible for both PD and HD, which is to our knowledge has not previously been studied, aside from the prior attempt at a randomized controlled trial.6 This is a relevant consideration because only patients who are eligible for both therapies are faced with a choice between the 2 in clinical practice, and studies should ideally be restricted to this population.16 Unfortunately, most registries do not contain information about eligibility. In our study, PD eligibility was determined using a structured approach that was centrally reviewed to enhance transparency and minimize subjectivity. Patients who did not undergo modality assessment and those deemed ineligible for PD made up 36% of the total dialysis population, were older, and had a higher burden of comorbid conditions. Prior estimates6,17-20 of PD eligibility have ranged from 64% to 83% among incident dialysis patients. As a consequence, PD eligibility may have been an important contributor to the underlying case-mix differences between HD and PD patients in many of the previous comparisons. The population of patients who are not eligible for PD may naturally carry a poor prognosis, and inclusion of these patients in prior analyses may have led to biased results. Some have posited that there may be an initial survival advantage with PD that dissipates over time.7,8,10,12 This has historically been explained by better preservation of residual kidney function and urine output in PD patients followed by loss of ultrafiltration capacity, inadequate volume control, and elevated risk for death the longer a patient spends on therapy.22,23 However, we did not find that the relative risk (RR) for death on PD therapy, as compared to HD, changes over time when we restricted AJKD Vol 71 | Iss 3 | March 2018

our analyses to only patients deemed eligible for both dialysis modalities. Selection bias may better explain the observed change in the RR for death over time between HD and PD in previous studies.11 The inclusion of sicker patients and urgent starts may bias analyses against HD early in the treatment course because they have a much higher initial mortality rate and are treated almost exclusively with HD. Consistent with our findings, the increase over time in the RR for death on PD compared to HD therapy has been shown to disappear when acute-start patients are excluded.11 It may be that patients who are not eligible for PD are also more likely to start dialysis therapy urgently while admitted to the hospital. The presence of diabetes has been shown to modify the relationship between dialysis modality and survival in previous studies.2,8,9,12,13 Similar findings were reported for older patients compared with younger patients8,9,12-14 and for females compared with males.12-14 In our study, after we excluded patients who were not eligible for PD, only age was found to modify the relationship between modality choice and survival. Of note, older age was not found to be associated with lower PD eligibility after accounting for barriers to self-care and family support in a previous analysis.18 Our study has important strengths. First, it was based on high-quality data collected prospectively in clinical practice with rigorous oversight. Second, we restricted our analysis to patients who were considered eligible for both HD and PD using a structured assessment. This information is not routinely available in prior studies of patients with ESRD, but is important and allows us to focus on the patient group that is faced with a choice between HD and PD in clinical practice. Our study also has limitations. First, the determination of PD eligibility was based on the consensus of the respective multidisciplinary team at each dialysis center. There may have been subjective variations across the various participating centers, but the process reflected actual decision making in everyday clinical practice. Second, we did not perform an as-treated analysis that takes into account changes in dialysis modality and/or vascular access over time for each respective incident patient. It is not uncommon for patients with ESRD to switch from one dialysis modality to another, and our analysis does not show the association between current dialysis modality and mortality. However, we performed sensitivity analysis in which patients who converted from HD to PD within 6 months of initiating dialysis therapy were removed from the analysis and we reached similar findings. Third, our cohort was smaller than most registry studies comparing HD and PD, but similar in size compared with other cohort studies done to date.10,14,15 Fourth, despite adjusting for comorbid conditions, the analysis does not necessarily account for the severity of patients’ comorbid diseases. Taken together, we acknowledge that there is residual confounding given the observational nature of this study, and the association found between mortality and dialysis modality in this study does not imply causality. 349

Original Investigation In conclusion, we have shown that HD and PD are associated with similar mortality among incident dialysis patients who are eligible for both modalities. The effect of modality on survival does not appear to change over time. Ideally, future studies should be restricted to individuals who are deemed eligible for both modalities when possible, in an attempt to reflect the outcomes of patients faced with a choice between HD and PD therapy in clinical practice. Supplementary Material Figure S1: Survival curves for traditional cohort for those age 65 and older. Figure S2: Survival curves for traditional cohort for those younger than 65. Figure S3: Survival curves for eligible cohort. Figure S4: Survival curves for eligible outpatient cohort. Item S1: Structured assessment to determine eligibility for HD and PD.

Article Information Authors’ Full Names and Academic Degrees: Ben Wong, MD, MSc, Pietro Ravani, MD, PhD, Matthew J. Oliver, MD, MHS, Jayna Holroyd-Leduc, MD, Lorraine Venturato, PhD, Amit X. Garg, MD, PhD, and Robert R. Quinn, MD, PhD. Authors’ Affiliations: Department of Community Health Sciences, University of Calgary, Calgary, Alberta (BW, PR, JH-L, RRQ); Department of Medicine, Headwaters Health Care Center, Orangeville, Ontario (BW); Cumming School of Medicine, University of Calgary, Calgary, Alberta (PR, JH-L, RRQ); Department of Medicine, Sunnybrook Health Sciences Center, University of Toronto, Toronto, Ontario (MJO); Faculty of Nursing, University of Calgary, Calgary, Alberta (LV); Department of Medicine, Western University, London (AXG); and Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada (AXG). Address for Correspondence: Ben Wong, MD, MSc, Headwaters Health Care Center, 100 Rolling Hills Drive, Orangeville, Ontario, Canada L9W 4X9. E-mail: [email protected] Authors’ Contributions: Research idea and study design: BW, RRQ; data acquisition: BW; data analysis/interpretation: BW, PR, AXG, RRQ; statistical analysis: BW, PR, RRQ; supervision or mentorship: PR, MJO, JH-L, LV, RRQ. Each author contributed important intellectual content during manuscript drafting or revision and accepts accountability for the overall work by ensuring that questions pertaining to the accuracy or integrity of any portion of the work are appropriately investigated and resolved. Support: This study was supported by the ICES Western site. ICES is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). Core funding for ICES Western is provided by the Academic Medical Organization of Southwestern Ontario (AMOSO), the Schulich School of Medicine and Dentistry (SSMD), Western University, and the Lawson Health Research Institute (LHRI). The research was conducted by members of the ICES Kidney, Dialysis and Transplantation team, at the ICES Western facility, who are supported by a grant from the Canadian Institutes of Health Research (CIHR). The funders of this study did not have any role in study design; collection, analysis, and interpretation of data; writing the report; and the decision to submit the report for publication. Financial Disclosure: Drs Oliver and Quinn are co-inventors of the DMAR system. They receive support from Baxter Healthcare to build

350

a PD catheter registry for the North American Research Consortium of the International Society of Peritoneal Dialysis. Disclaimer: The opinions, results, and conclusions are those of the authors and are independent from the funding sources. No endorsement by ICES, AMOSO, SSMD, LHRI, CIHR, or the MOHLTC is intended or should be inferred. Parts of this material are based on data and/or information compiled and provided by Canadian Institute for Health Information (CIHI). However, the analyses, conclusions, opinions and statements expressed in the material are those of the author, and not necessarily those of CIHI. Peer Review: Received February 26, 2017. Evaluated by 2 external peer reviewers, with direct editorial input from a Statistics/Methods Editor, an Associate Editor, and the Editor-in-Chief. Accepted in revised form August 27, 2017.

References 1. Saran R, Li Y, Robinson B, et al. US Renal Data System 2015 Annual Data Report: epidemiology of kidney disease in the United States. Am J Kidney Dis. 2016;67(3)(suppl 1):Svii, S1-S305. 2. Winkelmayer WC, Weinstein MC, Mittleman MA, Glynn RJ, Pliskin JS. Health economic evaluations: the special case of end-stage renal disease treatment. Med Decis Making. 2002;22(5):417-430. 3. Lee H, Manns B, Taub K, et al. Cost analysis of ongoing care of patients with end-stage renal disease: the impact of dialysis modality and dialysis access. Am J Kidney Dis. 2002;40(3): 611-622. 4. Hornberger J, Hirth RA. Financial implications of choice of dialysis type of the revised Medicare payment system: an economic analysis. Am J Kidney Dis. 2012;60(2):280-287. 5. Liu FX, Walton SM, Leipold R, Isbell D, Golper TA. Financial implications to Medicare from changing the dialysis modality mix under the bundled prospective payment system. Perit Dial Int. 2014;34(7):749-757. 6. Korevaar JC, Feith GW, Dekker FW, et al. Effect of starting with hemodialysis compared with peritoneal dialysis in patients new on dialysis treatment: a randomized controlled trial. Kidney Int. 2003;64(6):2222-2228. 7. Fenton SS, Schaubel DE, Desmeules M, et al. Hemodialysis versus peritoneal dialysis: a comparison of adjusted mortality rates. Am J Kidney Dis. 1997;30(3):334-342. 8. Collins AJ, Hao W, Xia H, et al. Mortality risks of peritoneal dialysis and hemodialysis. Am J Kidney Dis. 1999;34(6): 1065-1074. 9. Vonesh EF, Snyder JJ, Foley RN, Collins AJ. The differential impact of risk factors on mortality in hemodialysis and peritoneal dialysis. Kidney Int. 2004;66(6):2389-2401. 10. Jaar BG, Coresh J, Plantinga LC, et al. Comparing the risk for death with peritoneal dialysis and hemodialysis in a national cohort of patients with chronic kidney disease. Ann Intern Med. 2005;143(3):174-183. 11. Quinn RR, Hux JE, Oliver MJ, Austin PC, Tonelli M, Laupacis A. Selection bias explains apparent differential mortality between dialysis modalities. J Am Soc Nephrol. 2011;22(8): 1534-1542. 12. Yeates K, Zhu N, Vonesh E, Trpeski L, Blake P, Fenton S. Hemodialysis and peritoneal dialysis are associated with similar outcomes for end-stage renal disease treatment in Canada. Nephrol Dial Transplant. 2012;27(9):3568-3575. 13. van de Luijtgaarden MW, Noordzij M, Stel VS, et al. Effects of comorbid and demographic factors on dialysis modality choice and related patient survival in Europe. Nephrol Dial Transplant. 2011;26(9):2940-2947.

AJKD Vol 71 | Iss 3 | March 2018

Original Investigation 14. Winkelmayer WC, Glynn RJ, Mittleman MA, Levin R, Pliskin JS, Avorn J. Comparing mortality of elderly patients on hemodialysis versus peritoneal dialysis: a propensity score approach. J Am Soc Nephrol. 2002;13(9):2353-2362. 15. Termorshuizen F, Korevaar JC, Dekker FW, et al. Hemodialysis and peritoneal dialysis: comparison of adjusted mortality rates according to the duration of dialysis: analysis of the Netherlands Cooperative Study on the Adequacy of Dialysis 2. J Am Soc Nephrol. 2003;14(11):2851-2860. 16. Quinn RR, Austin PC, Oliver MJ. Comparative studies of dialysis therapies should reflect real world decision-making. J Nephrol. 2008;21(2):139-145. 17. Jager KJ, Korevaar JC, Dekker FW, Krediet RT, Boeschoten EW; Netherlands Cooperative Study on the Adequacy of Dialysis (NECOSAD) Study Group. The effect of contraindications and patient preference on dialysis modality selection in ESRD patients in the Netherlands. Am J Kidney Dis. 2004;43(5):891-899. 18. Oliver MJ, Garg AX, Blake PG, et al. Impact of contraindications, barriers to self-care and support on incident peritoneal dialysis utilization. Nephrol Dial Transplant. 2010;25(8):2737-2744.

AJKD Vol 71 | Iss 3 | March 2018

19. Little J, Irwin A, Marshall T, Rayner H, Smith S. Predicting a patient’s choice of dialysis modality: experience in a United Kingdom renal department. Am J Kidney Dis. 2001;37(5): 981-986. 20. Prichard SS. Treatment modality selection in 150 consecutive patients starting ESRD therapy. Perit Dial Int. 1996;16(1): 69-72. 21. Kutner M, Nachtsheim C, Neter J, eds. Applied Linear Regression Models. 4th ed. Columbus, OH: McGraw-Hill Irwin; 2004. 22. Churchill DN, Thorpe KE, Nolph KD, Keshaviah PR, Oreopoulos DG, Page D. 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. 1998;9(7):1285-1292. 23. Menon MK, Naimark DM, Bargman JM, Vas SI, Oreopoulos DG. Long-term blood pressure control in a cohort of peritoneal dialysis patients and its association with residual renal function. Nephrol Dial Transplant. 2001;16(11):22072213.

351