Changes in Quality of Life during Hemodialysis andPeritoneal Dialysis Treatment: Generic and Disease SpecificMeasures

ALBERT W. WU,*†‡ NANCY E. FINK,†‡ JANE V.R. MARSH-MANZI,‡

KLEMENS B. MEYER,§ FREDERIC O. FINKELSTEIN,�

MICHELLE M. CHAPMAN,§ and NEIL R. POWE*†‡

*Department of Health Policy and Management, Johns Hopkins Bloomberg School of Public Health,Baltimore, Maryland; †Department of Medicine, Johns Hopkins University School of Medicine, Baltimore,Maryland; ‡Department of Epidemiology, Johns Hopkins Bloomberg School of Public Health, Baltimore,Maryland; §Department of Medicine, Tufts-New England Medical Center, Boston, Massachusetts; and �YaleUniversity, New Haven, Connecticut

Abstract. Despite more than 20 yr of use, relative differences inhealth-related quality of life (HRQOL) between hemodialysis(HD) and peritoneal dialysis (PD) are not clearly known. Theobjective of this study was to compare self-reported HRQOLand overall health status for HD and PD patients at the initia-tion of dialysis therapy and 1 yr later. A prospective cohort ofincident ESRD patients was enrolled between October 1995and June 1998 at 81 outpatient dialysis units in 19 states andincluded 698 HD and 230 PD patients who completed a base-line CHOICE Health Experience Questionnaire. The main out-come measured was change in qualify-of-life scores from startof dialysis to 1 yr on dialysis and overall health status. Of 928patients who completed the baseline questionnaire, 585 alsocompleted the 12-mo questionnaire; 101 had died, 55 hadreceived a kidney transplant, and 88 had moved to a newdialysis clinic. PD patients were slightly younger, were morelikely to be white, were well-educated, were employed, weremarried, had less comorbidity, and had higher hematocrit.

Unadjusted baseline scores showed better HRQOL for PDpatients in both generic and ESRD domains (bodily pain,travel, diet restrictions, and dialysis access [P � 0.05]). At 1yr, SF-36 scores improved, whereas some ESRD domainsimproved and others deteriorated. HD patients had greaterimprovements in two SF-36 domains (physical functioning andgeneral health perception) than PD patients, but results weremixed for ESRD domains (PD is better for finances, HD isbetter for sleep and overall quality of life). HD and PD patientsdid not differ in change in overall health status. HD and PD areassociated with similar HRQOL outcomes at 1 yr. GenericHRQOL in two domains improved more for HD patients.However, for ESRD-specific HRQOL, results were not con-sistent; some domains were better for PD patients whereasothers were better for HD patients. In advising patients aboutmodality choices, trade-offs should be discussed and individualpreferences for specific aspects of HRQOL should be elicited.

Every year in the United States, nearly 90,000 patients withchronic kidney disease progress to ESRD and begin renalreplacement therapy (1). Because the availability of kidneytransplants is limited, most patients must choose between he-modialysis (HD), usually performed at a dialysis center, andchronic ambulatory peritoneal dialysis (PD), usually performedat home. Careful investigations in recent years suggest that PDpatient survival equals or is slightly shorter than that of patientswho undergo HD (2–6). After consulting with their physicians,fewer than 10% of patients choose PD as their initial therapy

(1). Nonetheless, conclusions based on anecdote allow neithersystematic nor longitudinal comparison of representative pa-tients and fail to take account of case-mix factors.

It is now widely accepted that health-related quality of life(HRQOL) is an important outcome of health care and one onwhich patients base treatment decisions. HRQOL is a multidi-mensional concept that includes physical functioning, socialand role functioning, mental health, and general health percep-tions (7). In studies of specific conditions, it has becomecommon to measure aspects of quality of life typically affectedby the condition and its treatments in addition to generalHRQOL (8). Combining both generic and disease-specificmeasures allows comparisons to other populations and shouldincrease sensitivity to changes over time, whether in the naturalhistory of the disease or in response to treatment.

Cross-sectional studies have yielded most of the data on theHRQOL of people with ESRD. Comparisons of HRQOL onHD and PD have produced conflicting results. A review of thisliterature that focused on the relationship of dialysis modality

Received July 17, 2003. Accepted November 25, 2003.Correspondence to Dr. Albert Wu, Johns Hopkins Bloomberg School of PublicHealth, Hampton House 633, 624 North Broadway, Baltimore, MD 21205.Phone: 410-955-6567; Fax: 425-740-1650; E-mail: awu@jhsph.edu

1046-6673/1503-0743Journal of the American Society of NephrologyCopyright © 2004 by the American Society of Nephrology

DOI: 10.1097/01.ASN.0000113315.81448.CA

J Am Soc Nephrol 15: 743–753, 2004

to mental health domains suggested that patients experienceless distress and better psychologic well-being on PD (9).However, interpretation of this literature was limited becauseof small and convenient samples of patients, use of new in-struments or instruments not tested in ESRD, inadequate con-trol of case-mix, and a lack of repeated measures.

There have been just a few longitudinal studies of HRQOLin ESRD (10–13). The only longitudinal comparison of mo-dalities was a study of consecutive patients who began dialysisin 13 Dutch dialysis centers. The results suggested that HD wasassociated with a relative benefit in physical aspects of patient-reported health (14). However, this study was small and did notassess dialysis-specific aspects of HRQOL. We comprehen-sively examined HRQOL in an incident cohort of ESRD pa-tients who were treated by HD and PD in the United States.

Materials and MethodsStudy Design and Participants

We conducted a national, prospective cohort study, the Choices forHealthy Outcomes in Caring for ESRD (CHOICE) Study, of incidentdialysis patients who were enrolled at 81 clinics in the United Statesbetween 1995 and 1998 (15). Details of the study design have beenpublished (15–17). Briefly, to be eligible, patients had to be initiatingHD or PD for ESRD, to be 18 yr or older, and to speak either Englishor Spanish. Home HD patients were excluded from the study. Ap-proximately two thirds of eligible patients were enrolled; these pa-tients were similar to nonenrolled patients with regard to gender andage. The enrolled population was oversampled for PD patients toallow statistical comparisons by modality. PD accounted for 26% ofthe patients in CHOICE compared with 17% in nonenrolled patients.The CHOICE cohort had a larger percentage of white patients (72%versus 68%) and a larger percentage of patients with diabetes as theprimary cause of ESRD (47% versus 37%) than the nonenrolledpatients. The final CHOICE sample is representative of the UnitedStates Renal Data System (USRDS) population with respect to age,gender, and race but differs in the percentage PD. Diabetes andhypertension together accounted for ESRD in two thirds of the cohort,a figure similar to the USRDS; however, the percentage attributed tohypertension in the study population was lower than in the USRDS(17.5% versus 28.7%). Patients were enrolled a median of 45 d fromthe start of chronic dialysis (98% within 4 mo). The protocol wasapproved by Institutional Review Boards at all institutions, and par-ticipants gave written informed consent to participate in the study.

Data CollectionAt baseline, we obtained sociodemographic information from a

patient questionnaire, from the treatment clinic, and from the HealthCare Financing Administration Medical Evidence Form 2728. Base-line clinical assessments were reported from each treatment clinic.Participants completed the Choice Health Experience Questionnaire(CHEQ) that included information on generic and dialysis specificHRQOL (described below). Follow-up visits were scheduled at 3, 6,and 12 mo after enrollment. This article focuses on data collected atthe baseline and 12-mo visits.

We categorized sociodemographic variables as follows: age �65 yr(yes/no), race (white, black, other), currently married (yes/no), em-ployed part or full time (yes/no), obtained high school degree (yes/no), and residing �30 mi from treatment clinic (yes/no). Baselinelaboratory values included serum albumin (g/dl), serum creatinine(mg/dl), hematocrit (%), and blood urea nitrogen. We created baseline

and 12-mo Index of Co-existent Disease (ICED) scores as describedpreviously (18–20) using data from a comorbidity assessment form.From 19 disease categories, each graded according to four levels ofseverity, and 11 physical impairment categories, each graded accord-ing to three levels of severity, we calculated the four-level aggregateICED score, reflecting peak disease and physical impairment scores.A higher ICED score represents greater extent and/or severity ofcomorbid conditions.

HRQOL AssessmentThe CHEQ includes the SF-36 (21) and 14 dialysis-specific do-

mains (22) (Table 1). For the SF-36, eight domain scores (physicalfunctioning [PF], role limitations as a result of physical problems[RP], bodily pain [BP], general health perceptions [GH], social func-tioning [SF], role limitations as a result of emotional problems [RE],vitality [VT], and mental health [MH]) and summary Physical Com-ponent (PCS) and Mental Component (MCS) scores were calculatedat baseline and 1 yr. We calculated SF-36 domain scores ranging from0 to 100 according to published guidelines (23, 24). In addition, wecomputed 14 baseline and 1-yr dialysis-specific raw domain scores bysumming responses from CHEQ questionnaire items. Dialysis-spe-cific domains include time, freedom, travel, cognitive function, finan-cial concerns, diet restrictions, recreation, work, body image, symp-toms, sleep, sexual functioning, dialysis access, and global quality oflife. For example, one of the items on time asks, “Did you feel likeyou spent too much time dealing with your dialysis treatment?” Oneof the items on financial concerns asks, “Have you been bothered bybeing unable to work at a paying job?” We linearly converted dialysis-specific domain scores to a 0- to 100-point scale in the same manneras used for the SF-36 domain scores.

For multi-item scales (PF, RP, BP, GH, SF, RE, VT, MH, Time,Freedom, Cognitive Function, Symptoms, Sex, Sleep, Access, and Qual-ity of Life), we obtained internal consistency reliability estimates (Cron-bach’s �) using Multitrait/Multi-item Analysis Program-Revised version1 (25). For the SF-36 PCS and MCS, we used reliability estimates fromthe U.S. population. For single-item domains (Travel, Finance, Diet,Recreation, Work, and Body Image), we estimated test-retest reliabilityusing Pearson correlation between CHEQ domain scores from baselineand 3 mo later among 830 CHOICE patients who completed a CHEQ at3 mo after baseline. Although this may not be ideal given changes thatmight occur early in the course of dialysis, the coefficients that weobserved represent estimates for reproducibility.

Statistical AnalysesWe first performed an intention-to-treat analysis on patients who

completed the CHEQ at both baseline and 1 yr later. We compareddifferences in unadjusted mean CHEQ and SF-36 domain scores betweenHD and PD patients at baseline and 1 yr later using t tests (nonparametricWilcoxon rank sum tests yielded similar results). We calculated unad-justed mean 1-yr changes in every domain for each dialysis modality bysubtracting baseline from 1-yr scores and tested differences between HDand PD patients using t tests. To estimate and test differences betweenHD and PD patients in baseline and 1-yr mean domain scores and inmean 1-yr changes adjusting for confounders, we used generalized esti-mating equation models for repeated measures with a restructured dataset containing two observations per patient.

Each observation contained baseline risk adjustment variables se-lected from the literature, either the baseline or 1-yr domain scores,and a time variable equal to zero for baseline domain scores or thedifference between the date of the 1-yr CHEQ and the enrollment datefor the 1-yr scores (9, 26–34). Domain scores were modeled as

744 Journal of the American Society of Nephrology J Am Soc Nephrol 15: 743–753, 2004

dependent variables and baseline modality, the time variable, theinteraction of the time variable and baseline modality, and confound-ers (age, gender, race, education, baseline albumin, creatinine, hemat-ocrit, and ICED) as independent variables, specifying an equal within-patient correlation structure and the robust Huber/White/sandwichestimator of variance. The coefficients for the time variable and theinteraction term represented the adjusted mean 1-yr change in domainscore for HD patients and the adjusted mean difference in 1-yr changein domain score for PD versus HD patients, respectively. Because thebaseline domain score is part of the slope calculation in the repeatedmeasures model, it was not included as an adjustment variable.

To address the loss of information as a result of dropouts, we con-ducted a separate analysis that examined deteriorations, improvements,and constancy (no change) in domain HRQOL from baseline to 1 yr laterusing all patients with a baseline CHEQ. For the patients with both abaseline and 1-yr CHEQ, we assigned a score of “worsened” when theirdomain score decreased significantly in 1 yr, “no change” when theirdomain score did not change significantly, and “improved” when theirdomain score increased significantly. We defined significant 1-yr in-creases and decreases in dialysis-specific and SF-36 domains as changesin domain scores exceeding two standard errors of measurement (SEM)(24). We calculated the SEM as SD*(1 � R)1/2 where SD is the SD ofthe baseline domain score and R is the domain reliability. For patientswho were missing the 1-yr CHEQ, we assigned a domain score of“worsened” to those who died within the first 15 mo after enrollment and

a score of “improved” to those who received a kidney transplant duringthat period. For the remaining patients, we assigned a score of “wors-ened” when their ICED score increased in 1 yr, “no change” when theirICED score remained identical, and “improved” when their ICED scoredecreased. We tested differences in changes in domain HRQOL accord-ing to dialysis modality in bivariate (�2 tests) and multivariable (logisticregression comparing patients whose domain score improved with thosewhose score was unchanged or worsened, adjusting for baseline domainscore, age, gender, race, education and baseline albumin, creatinine andhematocrit, but not ICED, which was included in the outcome definition)analyses. Because not all patients continued on their baseline dialysismodality throughout the subsequent year as the above intention-to-treatanalyses assume, sensitivity analyses were conducted to assess the impactof modality switches during the first year after baseline. All analyses wererepeated excluding patients who switched modality before completing the1-yr CHEQ or, for patients without a 1-yr CHEQ, before the 1-yranniversary of the enrollment date. All analyses were conducted usingSAS (version 6.12) and Stata (version 7.0) statistical software.

ResultsPatients and Baseline Characteristics

Of the 1041 patients, 89% (698 HD patients and 230 PDpatients) completed a baseline CHEQ. Compared with the 113patients who did not complete the CHEQ, patients who com-

Table 1. Generic and dialysis-specific domains of health-related quality of lifea

Domain Description No. of Items in Scale

General (SF-36)PCS1 Physical health component scorephysical function Limitations in performing physical activities 10role physical Limitations in role functioning due to physical health 4bodily pain Somatic pain 2general health General perceptions about health 5MCS Mental health component scorevitality Energy/fatigue 4social function Social functioning 2role emotional Limitations in role functioning due to mental health 3mental health Psychological well-being 5

ESRD (CHEQ)time Rating of time spent on dialysis 2freedom Restrictions to lifestyle due to dialysis and kidney disease 4travel Limitations in ability to travel 1cognitive function Problems with cognitive functioning 3finance Financial problems and concerns 1diet Restrictions in eating and drinking 1recreation Limitations to pasttimes and recreation 1work Limitations to working 1body image Self-perceptions about physical appearance 1symptoms Somatic symptoms 13sex Sexual functioning 3sleep Problems with sleep and rest 3access Problems with dialysis access 2QOL Global quality of life 2

a PCS, Physical Component Score; MCS, Mental Component Score; CHEQ, CHOICE Health Experience Questionnaire; QOL, qualityof life. The PCS and MCS scores are based on a weighted combination of all eight subscales, with mental health subscales weightednegatively for PCS scores and physical health subscales weighted negatively for MCS scores.

J Am Soc Nephrol 15: 743–753, 2004 Dialysis and Changes in Quality of Life 745

pleted the CHEQ tended to be female (47% versus 35%; P �0.02), receiving HD (75% versus 60%; P � 0.001), at leasthigh school educated (71% versus 56%; P � 0.04), not cur-rently married (45% versus 30%; P � 0.01), and living fartherfrom the treatment clinic (13% versus 4%; P � 0.48). Therewere no other significant differences between the two groupsof patients. Of the 928 patients who completed a baselineCHEQ, 585 also completed a CHEQ 1 yr later. Of the 343patients with a baseline but no 1-yr CHEQ, 244 were no longerunder follow-up and were closed out before receipt of their1-yr CHEQ (within 15 mo of their enrollment date); the re-maining 99 remained in the study but failed to return a 1-yrCHEQ. Reasons for closing out included death (n � 101),kidney transplantation (n � 55), moving to a dialysis clinic notin CHOICE (n � 48), refusal (n � 17), and other (n � 23).There were no significant differences in demographics, comor-bidity, or clinical measures at baseline between participantswho did and did not complete the 12-mo follow-up. Quality oflife was better on a few domains for completers (C) versusnoncompleters (NC) of the 12-mo follow-up (mean scores andt test P values follow): social function 60.5 (C), 56.8 (NC), P� 0.05; mental health 69.5 (C), 66.2 (NC), P � 0.02; symp-toms 77.8 (C), 75.2 (NC), P � 0.02; travel 59.7 (C), 53.5 (NC),P � 0.01; recreation 60.0 (C), 54.0 (NC), P � 0.01; and access67.9 (C), 63.8 (NC), P � 0.02.

Baseline characteristics of cohort members who completedthe CHEQ at baseline (n � 928) and 1 yr later (n � 585) aredisplayed in Table 2, stratified by baseline dialysis modality.PD patients tended to be younger, white, college educated,employed, married, and living farther from the treatment clinicand to have less baseline comorbidity and higher hematocrit.There were no HD-PD differences according to gender or othermeasures such as baseline residual renal function. Mean Kt/Vamong HD patients was 1.3 at baseline (n � 534) and 1.3 1 yrlater (n � 344).

Internal Consistency and Test-Retest ReliabilityInternal consistency estimates were as follows: PF, 0.91; RP,

0.82; BP, 0.86; GH, 0.72; VT, 0.85; SF, 0.76; RE, 0.86; MH,0.84; PCS, 0.91; MCS, 0.87; symptoms, 0.81; cognitive func-tion, 0.86; sleep, 0.77; sex, 0.93; quality of life, 0.67; freedom,0.76; time, 0.58; and access, 0.72. Test-retest reliability corre-lations were as follows: travel, 0.62; finance, 0.79; diet, 0.70;recreation, 0.64; work, 0.72; and body image, 0.55.

Comparisons of Crude Quality of Life Domain ScoresBetween HD and PD

At baseline, SF-36 domain and summary scores were sub-stantially lower than in the general population (23). The onlyexception was for mental health, for which scores were only afew points less. When unadjusted mean dialysis-specific andSF-36 domain scores were compared (Table 3), PD patientshad significantly (P � 0.05) higher scores for SF-36 bodilypain (9 to 11% higher) and for the dialysis domains of travel(12 to 15% higher), diet restrictions (19 to 23% higher), anddialysis access (8 to 9% higher), both at baseline and 1 yr latercompared with HD patients. PD patients also had significantly

higher baseline scores for the SF-36 domains PF (17% higher)and RE (20% higher), as well as for the dialysis domain offinancial concerns (20% higher) at 1 yr. The only domains forwhich significant differences favored HD patients were VT(12% higher) and sexual functioning (15% higher) at 1 yr.

When unadjusted mean changes in SF-36 domain scoresfrom baseline to 1 yr later were compared between HD and PDpatients, HD patients tended to show greater improvement inSF-36 domain scores than did PD patients (Figure 1). HDpatients had significantly greater improvements in PF (5.1points greater) and GH perceptions (4.1 points greater) than PDpatients. Results for dialysis domains were mixed, with HDpatients exhibiting significantly greater improvement in sleep(8.1 points greater) and global quality of life (4.3 pointsgreater) over 1 yr, and PD patients showing greater improve-ment in the finance domain (8.1 points greater).

Comparisons of Adjusted Quality-of-Life DomainScores Between HD and PD

When mean change over 1 yr was adjusted for potentialconfounders (Figure 2), a similar pattern of results was ob-tained. In terms of generic HRQOL, HD patients showedgreater improvement in all SF-36 domains except mentalhealth, although only differences in PF and GH perceptionswere statistically significant. With regard to dialysis-specificdomains, after adjustment for confounders, HD patients had asignificantly greater improvement in sleep (7.4 points greater),whereas PD patients had significantly greater improvement inthe finance domain (6.6 points greater).

Comparisons of Deteriorations and Improvements inDomains by Modality in All Patients

Considering changes (worsened, no change, improved) inoverall domain health status, defined by changes in domainscore, death, kidney transplantation, or changes in extent ofcomorbidity, among all patients 20 to 31% had a worsening, 42to 60% had no change, and 19 to 28% had an improvement ingeneric domains of HRQOL. Among all patients, 19 to 30%had a worsening, 50 to 65% had no change, and 16 to 24% hadan improvement in dialysis-specific domains of HRQOL.There were no statistically significant differences between HDand PD patients in these changes for any of the generic ordialysis-specific domains (Table 4). Adjusted analyses tendedto confirm these results (Table 5). Sensitivity analyses exclud-ing the 55 patients (25 HD, 30 PD) who switched from theirbaseline modality during the first year produced virtually iden-tical results (data not shown).

DiscussionThis national study provides a comprehensive and detailed

description of the quality of life of patients who started HD andPD and their progress 1 yr later. The findings have importantimplications for physicians who evaluate and treat patientswith chronic kidney disease. They suggest that there is nosimple answer to the question of which dialysis modality canbe expected to provide better quality of life.

One year after starting dialysis, patients on both HD and PD

746 Journal of the American Society of Nephrology J Am Soc Nephrol 15: 743–753, 2004

reported improvements in nearly all aspects of general func-tioning and psychologic well-being. The surprising finding wasthat patients on HD improved more on aspects of generalHRQOL than patients on PD, with greater improvements in PFand GH perceptions. Despite lower scores at baseline, at 1 yr,patients on HD actually reported better scores in some do-mains, such as better physical role functioning. These findingsremained after adjustment for baseline patient characteristics,

although the differences were no longer statisticallysignificant.

Changes in dialysis-specific aspects of life were moremixed, and there were more differences between the twomodalities. HD patients improved more in some aspects, suchas sleep (which for PD patients actually became worse overtime) and body image. At the end of 1 yr, patients on HDreported significantly better sexual functioning than those on

Table 2. Baseline characteristics of CHOICE patients by baseline dialysis modality and CHEQ form completion status atbaseline and 1 year latera

Characteristic

Baseline 1 Year Later

HD (N � 698) PD (N � 230)Pb

HD (N � 452) PD (N � 133)Pb

n % n % n % n %

Age�40 71 10 39 17 0.007 36 8 20 15 0.0240–49 119 17 40 17 80 18 22 1750–59 127 18 51 22 86 19 33 2560–69 175 25 57 25 122 27 36 2770–79 161 23 35 15 108 24 20 1580� 45 6 8 3 20 4 2 1

Age (mean) 698 59 yr 230 54 yr 0.0001 452 59yr

133 54 yr 0.0007

Gender% female 332 48 105 46 0.6 216 48 68 51 0.5% male 366 52 125 54 236 52 65 49

Race% white 440 63 185 81 0.001 274 61 108 81 0.001% black 224 32 34 15 150 33 20 15% other 34 5 11 5 28 6 5 4

Education% � high school 224 32 41 18 0.001 142 32 24 18 0.001% high school graduate 374 54 133 59 250 56 75 57% college graduate 93 13 52 23 56 12 32 24

Employment% unemployed 636 91 170 74 0.001 411 91 99 74 0.001% employed 62 9 60 26 41 9 34 26

Marital status% not married 338 49 76 33 0.001 217 48 40 30 0.001% married 358 51 153 67 235 52 93 70

Distance from clinic% � 30 mi 623 91 155 72 0.001 408 92 91 71 0.001% 30 � mi 59 9 60 28 35 8 37 29

ICED% 0–1 217 31 111 48 0.001 148 33 66 50 0.001% 2 270 39 60 26 180 40 32 24% 3 210 30 59 26 124 27 35 26

Mean albumin (g/dl) 682 3.6 222 3.5 0.1 445 3.7 130 3.6 0.2Mean creatinine (mg/dl) 681 7.3 225 7.2 0.42 443 7.3 132 7.2 0.6Mean hematocrit (%) 678 31.9 221 34.0 0.0001 443 31.8 129 34.0 0.0001

a A total of 928 patients completed the CHEQ at baseline; 585 also completed a CHEQ at 1 yr after baseline; 99 remained in the studybut failed to complete a CHEQ at 1 yr later; 244 closed out of the study before 1 yr after baseline. HD, hemodialysis; PD, peritonealdialysis; ICED, Index of Co-existent Disease.

b Differences in proportions were tested using �2 or exact tests; differences in means were tested using ANOVA.

J Am Soc Nephrol 15: 743–753, 2004 Dialysis and Changes in Quality of Life 747

PD. PD patients improved more on other dialysis-specificaspects of life, such as financial well-being, and continued tohave higher scores for ability to travel, diet, and dialysisaccess.

Are the differences and changes in scores clinically impor-tant? Previous studies using the SF-36 suggest that scores inthe range of 2 to 3 points on the physical and mental healthsummary scores (equivalent to 0.2 to 0.3 SD units) and 5 ormore points for the individual subscale scores are likely to beclinically important. We observed changes that were 2 to 3points greater, suggesting that they are likely to be noticeableand meaningful to patients on dialysis (24). For example, 49%of patients on PD and 54% of HD patients reported moderateor severe financial problems at baseline, whereas 41% of PDpatients and 54% of HD patients reported such problems at 1yr. It should be noted, however, that there were substantialdifferences in only a few of the domains.

To strengthen the comparisons in the absence of an experi-

mental design, we assembled a large, national cohort of inci-dent dialysis patients and examined a comprehensive set ofdomains of quality of life allowing examination of the impactof dialysis modality on specific aspects of patients’ lives. Weattempted to account for selection bias by adjusting for a widerange of baseline clinical and sociodemographic characteristicsthat might affect patients’ ability to improve. We presentedboth actual scores reported by patients and changes in scores,because the former are more useful to describe what life is likefor patients, whereas changes are more likely to be attributableto the choice of therapy and may be more helpful for choosingmodality. In this study as well as many others, patients whostarted PD were healthier, more privileged, and more likely tobe working than those who started HD (27). Patients whostarted PD had significantly better general HRQOL, includingbetter PF, less BP, and better RE scores than patients whostarted HD. PD patients also reported better quality of life inareas specific to dialysis, with significantly greater ability to

Table 3. Unadjusted and adjusted mean quality-of-life domain scores at baseline and 1 year later by dialysis modalitya

Domain

Baseline 1 Year Later

Unadjusted Adjustedb Unadjusted Adjustedb

HD PD Pc HD PD Pc HD PD Pc HD PD Pc

General (SF-36)physical function 44.6 52.1 0.007 45.3 49.6 0.12 45.4 47.7 0.4 45.7 45.1 0.8role physical 23.4 25.1 0.6 23.7 25.2 0.7 28.1 27.5 0.9 28.5 26.5 0.6bodily pain 55.9 62.3 0.02 56.0 62.5 0.02 57.5 62.6 0.05 57.2 62.6 0.046general health 41.5 41.5 0.9 41.5 42.2 0.8 44.5 40.8 0.08 44.3 41.2 0.14vitality 42.2 39.7 0.3 42.2 40.7 0.5 44.4 39.7 0.03 44.3 39.9 0.05social function 60.1 61.9 0.5 60.3 62.9 0.4 64.5 64.9 0.9 64.2 64.1 0.9role emotional 51.9 62.4 0.02 52.3 60.9 0.06 57.0 63.7 0.13 57.4 63.1 0.20mental health 69.1 70.6 0.4 69.1 71.8 0.2 70.4 72.5 0.3 69.7 73.1 0.10PCS 32.5 33.8 0.2 32.5 33.5 0.4 33.1 33.1 0.9 33.2 32.5 0.5MCS 46.8 47.3 0.7 46.8 48.0 0.3 48.4 49.0 0.6 48.2 49.4 0.3

ESRDsymptoms 77.3 79.3 0.2 77.3 79.3 0.2 76.1 78.3 0.2 76.3 78.1 0.3cognitive function 76.2 77.1 0.7 76.2 77.5 0.6 74.7 75.7 0.7 74.3 75.8 0.5sleep 56.6 61.2 0.08 56.6 62.5 0.03 58.8 55.3 0.7 58.4 56.9 0.6sex 67.4 62.2 0.15 68.4 62.5 0.11 65.0 56.5 0.03 66.8 58.0 0.01body image 79.3 78.8 0.9 78.9 79.8 0.8 83.3 78.6 0.08 82.4 79.7 0.3quality of life 52.6 56.1 0.09 52.5 56.6 0.06 53.0 52.3 0.7 52.6 53.1 0.8freedom 58.6 58.1 0.8 57.9 60.1 0.4 59.2 60.9 0.5 58.5 61.7 0.2travel 58.1 64.8 0.04 58.1 66.6 0.02 54.1 62.0 0.02 54.3 63.0 0.01recreation 59.3 62.6 0.3 58.9 64.9 0.06 59.4 59.7 0.9 58.8 61.4 0.4finance 53.6 55.5 0.6 53.1 58.3 0.16 53.0 63.4 0.006 52.7 64.5 �0.01work 62.0 61.3 0.9 62.0 63.9 0.6 60.9 66.3 0.2 61.0 67.8 0.08diet 59.4 70.9 �0.001 58.7 72.3 �0.01 57.7 71.2 �0.001 57.2 72.9 �0.01time 57.4 56.6 0.8 57.3 58.7 0.6 55.2 58.5 0.2 55.0 59.8 0.05access 66.6 72.1 0.01 66.6 74.9 �0.01 73.1 79.4 0.01 72.3 80.7 �0.01

a A total of 585 patients completed a CHEQ at baseline and 1 yr later; HD, n � 452; PD, n � 133.b Adjusted for baseline age, gender, race, education, albumin, creatinine, hematocrit, and ICED score.c The differences in HD and PD means were compared using t tests (unadjusted) or Wald tests (adjusted); P values from Wilcoxon rank

sum tests for unadjusted comparisons were consistent except for travel (P � 0.06) and access (P � 0.1) at baseline, and bodily pain (P �0.04) and vitality (P � 0.05) for 1 yr later.

748 Journal of the American Society of Nephrology J Am Soc Nephrol 15: 743–753, 2004

travel, fewer dietary restrictions, and fewer concerns aboutdialysis access. Bivariate results provide the best description ofexpected quality-of-life scores for typical patients, whereas themultivariable results provide greater confidence in attributingchanges in scores to dialysis modality.

We also followed patients longitudinally and, with a thor-ough accounting of the disposition of patients, guarded againstsurvival biases that might be introduced by patients who leftthe cohort during the course of the study. We performed anoverall analysis that combined the major important outcomes,including HRQOL, transplantation, and survival. This analysisshowed that observed differences in quality of life were not dueto a higher mortality or transplant rate that might in turn beassociated with dialysis modality. Information on many of thedialysis-specific domains has not been collected systematicallyin previous evaluations of dialysis treatment modalities. Theresults highlight the importance of measuring dialysis-specificdomains. The results are also supported by hypotheses thatreflect clinical intuition and experience. Patients on HD wouldbe expected to have more problems with pain (e.g., needlesticks) and dialysis access (35). Patients on PD would beexpected to have more problems with sleep and body image butgreater ability to travel and work and better financial status.Measuring each of these domains separately allowed us to

identify specific aspects of life that differed by modality,information that might be useful to individual patients withspecific preferences as they attempt to decide betweenmodalities.

Our findings expand the results of previous studies, most ofwhich were cross-sectional and collectively were not conclu-sive as to impact of dialysis modality on quality of life. Severalstudies suggest advantages for PD in some domains (14, 35,36–38), advantages for HD in others (28, 35), or little differ-ence (27, 30, 32, 35–37, 39, 40). Cameron et al. (9) performeda meta-analysis to determine whether dialysis treatment mo-dality and case-mix could explain differences in emotionaldistress and well-being. PD was associated with greater well-being than HD, but several case-mix variables seemed impor-tant. The authors concluded by recommending a repeated mea-sures cohort study with careful identification anddocumentation of case-mix variables and statistical control,with a representative sampling of quality-of-life outcomes.

Merkus et al. (14) published a longitudinal comparison ofmodalities in their study of consecutive patients with incidentESRD in 13 Dutch dialysis centers. HRQOL was measuredusing the generic SF-36, which was administered 3, 6, 12, and18 mo after the start of dialysis. As only 139 of the 230 patientsstayed on their initial dialysis modality, an intention-to-treat

Figure 1. Crude mean differences in general (SF-36) and ESRD domain scores from baseline to 1 yr by baseline dialysis modality. �,hemodialysis (HD); �, peritoneal dialysis (PD); Symp, symptoms; Cog, cognitive function; Body, body image; Free, freedom; Trav, travel;Recr, recreation; Fin, finances; Acc, access. *P � 0.05.

J Am Soc Nephrol 15: 743–753, 2004 Dialysis and Changes in Quality of Life 749

analysis was also used. They found an overall decrease inSF-36 PCS over time of 1.9 points, with a significantly smallerdecrease for patients on HD compared with those on PD. It isinteresting that the mean difference in change score from 3 to18 mo was 2.3 points. We observed a small increase in PCS forpatients on HD and similar decrease for patients on PD, with adifference of 1.3 points over the 12 mo of study that favoredHD. Thus, our results are fairly comparable. However, in theirstudy, mental health scores did not change significantly overtime and were similar between modalities. Our subjects re-ported small increases in mental health, again with no differ-ence by modality. In their intention-to-treat analysis, therewere no significant treatment effects between HD and PD.Harris et al. (40), in their 12-mo prospective cohort study of174 dialysis patients 70 yr or older from four hospital-basedrenal units in London, found quality of life to be similar at 1year in PD and HD patients.

Our study had some limitations. First, we did not measurequality of life before patients began dialysis. Therefore, it islikely that we did not capture all of the improvements inHRQOL experienced by patients at the very start of dialysis.These are likely to be concentrated in domains of generalquality of life as assessed by the SF-36. If this initial improve-ment differed between dialysis modalities, then it is possiblethat we might fail to capture some of the subsequent changes.

In the future, attempts should be made to follow quality of lifein chronic kidney disease patients as kidney function declines,with particular attention to changes as they make the transitionfrom the predialysis state to dialysis treatment.

Second, it is possible that baseline patient characteristicsmight have been associated with a greater or lesser proclivityto improve or report improvement. It is possible that PDpatients, who were on average more advantaged than patientswho started HD, had more to lose and less opportunity tobenefit from dialysis. We also did not measure patient expec-tations about the outcomes of dialysis or their personalitycharacteristics, both of which might also influence treatmentselection and adherence. For example, optimistic people withgood social support and high self-efficacy might choose PD.HD patients might choose their therapy because they are lessindependent and receive more social support as part of theirtreatment. In a small Swedish sample, Lindqvist et al. (41)observed more avoidant coping among patients on HD andsuggested that there may be trait differences in patients whoselect different modalities. However, there are no data in thispatient population to suggest how such traits might affectchanges in HRQOL. Problems with measurement can some-times lead to similar results, but this does not seem to be thecase because ceiling effects were not evident for the majorityof scales for which there were differences between groups. If

Figure 2. Adjusted (for age; gender; race; education; Index of Co-existent Disease; and baseline albumin, creatinine, and hematocrit) meandifferences in general (SF-36) and ESRD domain scores from baseline to 1 yr by baseline dialysis modality. �, HD; �, PD; Symp, symptoms;Cog, cognitive function; Body, body image; Free, freedom; Trav, travel; Recr, recreation; Fin, finances; Acc, access. *P � 0.05.

750 Journal of the American Society of Nephrology J Am Soc Nephrol 15: 743–753, 2004

the functional form of recovery were nonlinear and sickerpatients were on a steeper portion of the functional status curvethan healthier patients, then it might also be easier to detect anincrease in score. This possibility would need to be addressedin a future study.

Third, despite our relatively large cohort, we had limitedpower to detect differences in subdomains of HRQOL. Evenlarger studies would be needed to determine whether some ofthe smaller observed differences truly exist, such as those thatfavor HD in most domains on the SF-36, the small differencein mental health favoring PD, and other small differences indialysis-specific domains. In addition, given the multiple sta-tistical tests performed, we cannot discount the possibility thatsome of the statistically significant findings are due to chance.

Finally, as this was an observational study rather than arandomized controlled trial, we cannot be absolutely certainthat differences in quality-of-life changes can be attributed to

dialysis modality. However, it is doubtful whether a random-ized controlled trial would be possible, because it would not bepossible to blind assignment to treatment groups. In addition,assignment to a specific dialysis modality might in itself de-crease the patient’s quality of life, threatening the internalvalidity of such a design (42).

Our study has important implications for practice. The goodnews for patients on both modalities is that health and generalwell-being should improve during the first year of dialysis.This information should be reassuring to patients who arereluctant to start dialysis. Conversely, there may be losses insome specific aspects of life. Although some of these losses arerelated to the dialysis modality, we cannot conclude that PDproduces a better quality of life than HD for patients whoinitiate renal replacement therapy. In fact, HD may result inmore positive changes in general and physical aspects ofHRQOL. There do seem to be distinct advantages and disad-

Table 4. Change in overall health statusa from baseline to 1 year after baseline in 928 CHOICE patients overall andstratified by modality

Domain

Overall HD PD

Pb

%Worsened

% NoChange

%Improved

%Worsened

% NoChange

%Improved

%Worsened

% NoChange

%Improved

General (SF-36)physical function 27 52 21 27 51 22 28 54 18 0.6role physical 22 58 21 21 58 21 23 56 21 0.8bodily pain 28 46 26 28 46 25 26 47 27 0.8general health 21 60 19 21 59 20 21 63 16 0.5vitality 27 49 24 27 49 24 29 46 25 0.7social function 20 58 21 20 59 21 22 57 21 0.9role emotional 30 42 28 30 41 28 30 43 27 0.9mental health 23 57 20 24 57 19 21 57 22 0.5PCS 31 42 27 31 42 27 32 43 25 0.9MCS 26 48 26 27 48 25 24 49 27 0.7

ESRDsymptoms 25 57 18 25 56 19 23 59 18 0.7cognitive function 26 56 17 27 56 17 25 56 19 0.8sleep 24 57 19 23 57 20 27 56 17 0.5sex 30 50 20 30 50 20 30 50 20 0.9body image 19 65 16 19 64 17 18 69 13 0.5quality of life 24 57 19 24 57 19 25 58 17 0.7freedom 22 59 19 22 59 19 20 59 21 0.6travel 25 58 17 25 58 17 23 60 17 0.7recreation 22 61 17 22 60 18 21 65 14 0.4finance 21 61 18 23 61 17 18 63 19 0.3work 23 58 19 24 58 18 19 58 23 0.1diet 24 60 17 25 58 17 21 62 17 0.5time 20 64 16 21 63 16 16 67 17 0.3access 22 54 24 23 53 24 19 58 23 0.4

a Change in overall health status was determined by increases (improved) or decreases (worsened) in domain scores during 1 yr afterbaseline or, if the 1-yr CHEQ was missing, by the occurrence of kidney transplantation (improved) or death (worsened) or by increases(worsened) or decreases (improved) in ICED score during 1 yr after baseline; no change was defined as changes in domain scores of �2SEMs or, if the 1-yr CHEQ was missing, by no change in ICED score from during 1 yr after baseline.

b Differences were tested using �2 tests.

J Am Soc Nephrol 15: 743–753, 2004 Dialysis and Changes in Quality of Life 751

vantages to each of the two modalities that should be exploredwith patients who are choosing between them. Physiciansshould be as explicit as possible in describing specific tradeoffsand attempt to elicit individual preferences for these aspects ofquality of life. For some patients, these preferences could bedecisive. For example, for individuals for whom work is a toppriority, PD may allow the best future functioning. Patients onPD can also anticipate less pain and problems with diet. How-ever, for those who are more concerned about other aspects oflife, such as sleep or sexual functioning, HD may offer relativeadvantages.

In conclusion, contrary to anecdotal reports, it does not seemthat PD produces a better quality of life than HD for patientswho initiate renal replacement therapy. In general, patients onHD may have maintained their general health status to a greaterextent than PD patients. However, there are distinct advantagesand disadvantages to each of the two modalities. These shouldbe explored with all patients who confront this choice.

AcknowledgmentSupported by grant R01-HS-08365 from the Agency for Health

Care Research and Quality (Rockville, MD) and grant RO1 DK 59616from the National Institute of Diabetes and Digestive and KidneyDiseases (Bethesda, MD). Dr. Powe is supported in part by grantK24DK02643 from the National Institute of Diabetes and Digestiveand Kidney Diseases (Bethesda, MD).

We thank the patients, staff, and medical directors of the partici-pating clinics at Dialysis Clinic, Inc. (DCI), St. Raphael’s Hospital,and New Haven CAPD who contributed to the study. We thankPatricia Bayton for help in preparing this manuscript.

References1. United States Renal Data System: USRDS 2002 Annual Data

Report, Bethesda, MD, National Institute of Diabetes, Digestiveand Kidney Diseases, National Institutes of Health, 2002

2. Foley RN, Parfrey PS, Harnett JD, Kent GM, O’Dea R, MurrayDC, Barre PE: Mode of dialysis therapy and mortality in end-stage renal disease. J Am Soc Nephrol 9: 267–276, 1998

3. Collins AJ, Hao W, Xia H, Ebben JP, Everson SE, ConstantineEG, Ma JZ: Mortality risks of peritoneal dialysis and hemodial-ysis. Am J Kidney Dis 34: 1065–1074, 1999

4. Locatelli F, Marcelli D, Conte F, D’Amico M, Del Vecchio L,Limido A, Malberti F, Spotti D: Survival and development ofcardiovascular disease by modality of treatment in patients withend-stage renal disease. J Am Soc Nephrol 12: 2411–2417, 2001

5. Winkelmayer WC, Glynn RJ, Mittleman MA, Levin R, PliskinJS, Avorn J: Comparing mortality of elderly patients on hemo-dialysis versus peritoneal dialysis: A propensity score approach.J Am Soc Nephrol 13: 2353–2362, 2002

6. Xue JL, Everson SE, Constantini EG, Ebben JP, Chen SC,Agodoa LY, Collins AJ: Peritoneal and hemodialysis: II. Mor-tality risk associated with initial patient characteristics. KidneyInt 61: 741–746, 2002

7. Croog SH, Levine S: Quality of life and health care evaluations.In: Handbook of Medical Sociology, 4th Ed., edited by FreemanHE, Levine S, Englewood Cliffs, NJ, Prentice Hall, 1989, pp508–528

8. Guyatt GH, Bombardier C, Tugwell PX: Measuring disease-specific quality of life in clinical trials. Can Med Assoc J 134:889–895, 1986

9. Cameron JI, Whiteside C, Katz J, Devins GM: Differences inquality of life across renal replacement therapies: A meta-ana-lytic comparison. Am J Kidney Dis 35: 629–637, 2000

10. Kutner NG, Brogan D, Fielding B: Employment status andability to work among working-age chronic dialysis patients.Am J Nephrol 11: 334–340, 1991

11. Parfrey PS, Vavasour HM, Gault HM: A prospective study ofhealth status in dialysis and transplant patients. Transplant Proc20: 1231–1232, 1988

12. Laupacis A: Changes in quality of life and functional capacity inhemodialysis patients treated with recombinant human erythro-poietin. Semin Nephrol 10: 11–19, 1990

13. Hashimoto Y, Matsubara T: Combined peritoneal dialysis andhemodialysis therapy improves quality of life in end-stage renaldisease patients. Adv Perit Dial 16: 108–112, 2000

14. Merkus MP, Jager KJ, Dekker FW, deHaan RJ, Boeschoten EW,Krediet RT for the Necosad Study Group: Quality of life overtime in dialysis: The Netherlands Cooperative Study on theAdequacy of Dialysis. Kidney Int 56: 720–728, 1999

15. Powe NR, Klag MJ, Sadler JH, Anderson GF, Bass EB, BriggsWA, Fink NE, Levey AS, Levin NW, Meyer KB, Rubin HR, Wu

Table 5. Adjusted odds ratiosa for improvement in overallhealth statusb on PD versus HD

Domain OddsRatio

95% ConfidenceInterval P Value

General (SF-36)physical function 0.72 0.45–1.15 0.17role physical 0.84 0.54–1.31 0.43bodily pain 1.12 0.74–1.71 0.59general health 0.65 0.41–1.04 0.07vitality 0.90 0.59–1.38 0.64social function 0.88 0.56–1.38 0.57role emotional 0.88 0.58–1.36 0.57mental health 1.19 0.76–1.89 0.45PCS 0.79 0.52–1.20 0.27MCS 0.95 0.62–1.45 0.81

ESRDsymptoms 0.86 0.53–1.41 0.55cognitive function 0.99 0.61–1.60 0.95sleep 0.84 0.52–1.35 0.48sex 0.65 0.40–1.05 0.08body image 0.57 0.33–0.99 0.05QOL 0.90 0.56–1.45 0.67freedom 1.03 0.66–1.63 0.88travel 1.07 0.66–1.75 0.77recreation 0.85 0.51–1.41 0.53finance 1.02 0.63–1.64 0.95work 1.17 0.74–1.85 0.51diet 1.22 0.74–2.02 0.43time 0.89 0.54–1.46 0.65access 0.97 0.61–1.55 0.91

a Adjusted for baseline domain score, age, gender, race,education, albumin, creatinine, and hematocrit.

b Defined as an increase in domain score 1 yr after baseline or,if the 1-yr CHEQ was missing, by the occurrence of kidneytransplantation or a decrease in ICED score 1 yr after baseline; noimprovement was defined as no change in or worsening of healthstatus, as previously described.

752 Journal of the American Society of Nephrology J Am Soc Nephrol 15: 743–753, 2004

AW for the CHOICE Study: Choices for healthy outcomes incaring for end stage renal disease. Semin Dial 9: 9–11, 1996

16. Longenecker JC, Coresh J, Klag MJ, Levey AS, Martin AA, FinkNE, Powe NR: Validation of comorbid conditions on the ESRDmedical evidence report: The CHOICE Study. J Am Soc Nephrol11: 520–529, 2000

17. Kinchen KS, Sadler JH, Fink NE, Brookmeyer R, Klag MJ,Levey AS, Powe NR: The timing of specialist evaluation inchronic kidney disease and mortality. Ann Intern Med 137:479–486, 2002

18. Nicolucci A, Cubasso D, Labbrozzi D, Mari E, Impicciatore P,Procaccini DA, Forcella M, Stella I, Querques M, Pappani A,Passione A, Strippoli P: Effect of coexistent diseases on survivalof patients undergoing dialysis. ASAIO J 41: M291–M295, 1992

19. Athienities NV, Miskulin DC, Fernandez GF, Bunnapradist S,Simon G, Landa M, Schmid CH, Greenfield S, Levey AS, MeyerKB: Comorbidity assessment in hemodialysis and peritonealdialysis using the Index of Coexistent Disease (ICED). SeminDial 13: 320–326, 2000

20. Miskulin DC, Meyer KB, Athienites NV, Martin AA, Terrin N,Marsh JV, Fink NE, Coresh J, Powe NR, Klag MJ, Levey AS forthe CHOICE Study: Comorbidity and other factors associatedwith modality selection in incident dialysis patients: TheCHOICE Study. Am J Kidney Dis 39: 324–336, 2002

21. Ware JE, Sherbourne CD: The MOS 36-Item short-form healthsurvey (SF-36): Conceptual framework and item selection. MedCare 30: 473–483, 1992

22. Wu AW, Fink NE, Cagney KA, Bass EB, Rubin HR, Meyer KB,Sadler JH, Powe NR for the CHOICE Study: Developing ahealth-related quality of life measure for end-stage renal disease:The CHOICE Health Experience Questionnaire (CHEQ). Am JKidney Dis 37: 11–21, 2001

23. Ware JE, Snow KK, Kosinski M, Gandek B: SF-36 HealthSurvey—Manual and Interpretation Guide, Boston, The HealthInstitute, New England Medical Center, 1993

24. Ware JE, Kosinski M, Keller SK: SF-36 Physical and MentalHealth Summary Scales: A User’s Manual, Boston, The HealthInstitute, New England Medical Center, 1994

25. Ware JE, Harris WJ, Gandek B, Rogers BW, Reese PR: MAP-Rfor Windows: Multitrait/Multi-item Analysis Program-Revised.Westwood, MA, Universal, 1997

26. Merkus MP, Jager KJ, Dekker FW, Boeschoten EW, Stevens P,Krediet RT and the Necosad Study Group: Quality of life inpatients on chronic dialysis: Self-assessment 3 months after thestart of treatment. Am J Kidney Dis 29: 584–592, 1997

27. Diaz-Buxo JA, Lowrie EG, Lew NL, Zhang H, Lazarus JM:Quality-of-life evaluation using Short Form 36: Comparison inhemodialysis and peritoneal dialysis patients. Am J Kidney Dis35: 293–300, 2000

28. Carmichael P, Popoola J, John I, Stevens PE, Carmichael AE:Assessment of quality of life in a single centre dialysis popula-tion using the KDQOL-SFJ questionnaire. Qual Life Res 9:195–205, 2000

29. Mingardi G for the DIA-QOL Group: Quality of life and endstage renal disease therapeutic programs. Int J Artif Organs 21:741–747, 1998

30. Bremer BA, McCauley CR, Wrona RM, Johnson J: Quality oflife in end-stage renal disease A re-examination. Am J KidneyDis 12: 200–209, 1989

31. Deniston OL, Carpentier-Alting P, Kneisley J, Hawthorne VM,Port FK: Assessment of quality of life in end-stage renal disease.Health Serv Res 24: 555–578, 1989

32. Evans RW: Recombinant human erythropoietin and the qualityof life of end-stage renal disease patients; a comparative analysis.Am J Kidney Dis 18: 62–70, 1991

33. Tell GS, Mittelmark MB, Hylander B, Shumaker SA, Russel GB,Burkart JM: Social support and health-related quality of life inblack and white dialysis patients. ANNA J 22: 301–310, 1995

34. Sehgal AR: Outcomes of renal replacement therapy amongblacks and women. Am J Kidney Dis 35: S148–S152, 2000

35. Majkowicz M, Afeltowicz Z, Lichodziejewska-Niemierko M,Debska-Slizien A, Rutkowski B: Comparison of the quality oflife in hemodialysis (HD) and peritoneally dialysed (CAPD)patients using the EORTC QLQ-C30 questionnaire. Int J ArtifOrgans 23: 423–428, 2000

36. Muthny FA, Koch U: Quality of life of patients with end-stagerenal failure. A comparison of hemodialysis, CAPD and trans-plantation. Contrib Nephrol 89: 265–273, 1991

37. Koch U, Muthny FA: Quality of life in patients with end-stagerenal disease in relation to the method of treatment. PsychotherPsychosom 54: 161–171, 1990

38. Simmons RG, Anderson CR, Abress LK: Quality of life andrehabilitation differences among four end-stage renal diseasetherapy groups. Scand J Urol Nephrol Suppl 131: 7–22, 1990

39. Tucker CM, Ziller RC, Smith WR, Mars DR, Coons MP: Qualityof life of patients on in-center hemodialysis versus continuousambulatory peritoneal dialysis. Perit Dial Int 11: 341–346, 1991

40. Harris SAC, Lamping DL, Brown EA, Constantinovici N: Clin-ical outcomes and quality of life in elderly patients on peritonealdialysis versus hemodialysis. Perit Dial Int 22: 463–470, 2002

41. Lindqvist R, Carlsson M, Sjödén PO: Coping strategies andquality of life among patients on hemodialysis and continuousambulatory peritoneal dialysis. Scand J Caring Sci 12: 223–230,1998

42. Szabo E, Moody H, Hamilton T, Ang C, Kovithavongs C,Kjellstrand C: Choice of treatment improves quality of life. ArchIntern Med 157: 1352–1356, 1997

J Am Soc Nephrol 15: 743–753, 2004 Dialysis and Changes in Quality of Life 753