- Email: [email protected]
Quality-of-life evaluation using Short Form 36: Comparison in hemodialysis and peritoneal dialysis patients
Quality-of-Life Evaluation Using Short Form 36: Comparison in Hemodialysis and Peritoneal Dialysis Patients Jose A. Diaz-Buxo, MD, Edmund G. Lowrie, MD, Nancy L. Lew, SM, Hongyuan Zhang, PhD, and J. Michael Lazarus, MD ● Short Form 36 (SF-36) is a well-documented health-related quality-of-life (HRQOL) instrument consisting of 36 questions compressed into eight scales and two primary dimensions: the physical and mental component scores. This tool was used to evaluate QOL among peritoneal dialysis (PD) and hemodialysis (HD) patients. The results of 16,755 HD and 1,260 PD patients (728 continuous ambulatory PD [CAPD] and 532 continuous cycling PD [CCPD]) completing an SF-36 during 1996 were analyzed. Three analyses of variance were performed, consisting of (1) no adjustment, (2) case mix (age, sex, race, and diabetes), and (3) case mix plus laboratory parameters. PD patients were younger (P F 0.001), a larger fraction were white (P F 0.001), fewer had diabetes (P F 0.001), and had lower serum albumin concentrations (P F 0.001) and higher creatinine, hemoglobin, and white blood cell count values (P F 0.001) than HD patients. Diabetes was present in a larger fraction of CCPD than CAPD patients (P F 0.001). HD and PD patients scored similarly for scales reflecting physical processes. PD patients scored higher for mental processes, but only after statistical adjustment for the laboratory measures. Scores on scales reflecting physical processes were worse, and those reflecting mental processes were better among CCPD than CAPD patients. HD and CAPD scores were similar. CCPD patients perceived themselves as more physically impaired but better adjusted than HD or CAPD patients. These descriptive data show that perception of QOL among PD and HD patients is similar before adjustment, but PD patients score higher for mental processes with adjustment. CCPD patients score worse for physical function and better for mental function than either CAPD or HD patients. We cannot, however, exclude the influence of therapy selection. 娀 2000 by the National Kidney Foundation, Inc. INDEX WORDS: Quality of life (QOL); Short Form 36 (SF-36); outcome; hemodialysis (HD); peritoneal dialysis (PD).
S
HORT FORM 36 (SF-36) is both a standalone measure of functioning, appropriate for use with any and all patients, and the core component of the Kidney Disease Quality of Life survey instrument.1,2 This instrument has been widely used and validated as a quality-oflife (QOL) assessment tool for the general population and for patients with congestive heart failure, kidney disease, and mental illness.3-5 Fresenius Medical Care North America started performing pilot studies of the SF-36 during mid-1995. In 1996, the company embarked on a second phase, offering the survey to all affiliated centers. By year’s end, more than 20,000 forms had been received and analyzed for hemodialysis (HD) patients. Preliminary results suggested that, in general, the physical component scores (PCSs) among dialysis patients were less than those observed in the US population and similar to those in persons with congestive heart failure.6 The mental component scores (MCSs) were similar to or slightly less than those found in the US population, but substantially better than among persons with clinical depression. Because of the lack of data on the potential differences in healthrelated QOL (HRQOL) among HD and peritoneal dialysis (PD) patients in large populations, we performed the present analysis comparing
SF-36 scores of patients undergoing each of these therapies. METHODS
Description of the SF-36 The SF-36 instrument consists of 36 questions, 35 of which are compressed into eight scales. The eight multi-item scales are as follows: (1) Physical Functioning (PF) is a 10-question scale that captures abilities to deal with the physical requirement of life, such as attending to personal needs, walking, and flexibility. (2) Role-Physical (RP) is a four-item scale that evaluates the extent to which physical capabilities limit activity. (3) Bodily Pain (BP) is a two-item scale that evaluates the perceived amount of pain experienced during the most recent 4 weeks and the extent to which that pain interfered with normal work activities. (4) General Health (GH) is a five-item scale that evaluates general health in terms of personal perception. (5) Vitality (VT) is a four-item scale that evaluates feeling of pep, energy, and fatigue. (6) Social Functioning (SF) is a two-
From Fresenius Medical Care North America, Lexington, MA. Received May 13, 1999; accepted in revised form August 20, 1999. Address reprint requests to Jose A. Diaz-Buxo, MD, Fresenius Medical Care North America, Peritoneal Dialysis Services, 1051 E Morehead St, Suite 250, Charlotte, NC 28204. E-mail: [email protected]
娀 2000 by the National Kidney Foundation, Inc. 0272-6386/00/3502-0028$3.00/0
American Journal of Kidney Diseases, Vol 35, No 2 (February), 2000: pp 293-300
293
294
DIAZ-BUXO ET AL
item scale that evaluates the extent and amount of time, if any, that physical health or emotional problems interfered with family, friends, and other social interactions during the most recent 4 weeks. (7) Role-Emotional (RE) is a threeitem scale that evaluates the extent, if any, to which emotional factors interfere with work or other activities. (8) Mental Health (MH) is a five-item scale that evaluates feelings principally of anxiety and depression. The eight scales of the SF-36 have been compressed into two primary dimensions: physical (PCS) and mental (MCS). The PF, RP, and BP scales are the primary members of the physical dimension, whereas the SF, RE, and MH are the primary members of the mental dimension. The GH and VT are considered members of both dimensions.7
of case mix and laboratory variables. These adjustments have been used in many of our previous analyses and are described elsewhere.9,10 The laboratory variables selected for analysis were those found most closely associated with the SF-36 during previous studies6 and included serum albumin, serum creatinine, blood hemoglobin, serum potassium, serum phosphorus, serum iron, bicarbonate, ferritin, and aspartate aminotransferase concentrations and white blood cell (WBC) count. The laboratory parameters were measured during the 3 months before administration of the SF-36, and the average value was reported. The urea reduction ratio, found to be weakly associated with the SF-36 scales,6 was not included because similar measures were not available for PD.
Data Collection and Statistical Analysis Data included 16,755 HD patients and 1,260 PD patients who completed an SF-36 form during 1996. Patients receiving PD were further subclassified into continuous ambulatory PD ([CAPD] n ⫽ 728) and continuous cycling PD ([CCPD] n ⫽ 532). Only complete forms were accepted for analysis. Three analyses of variance or covariance8 were performed for each of the eight primary and the two composite scales. The statistical model treated CAPD and CCPD patients as nested within the overall PD classification. Thus, we were able to evaluate potential differences between HD and PD patients in the aggregate and as they were divided into the two therapy types. We were also able to evaluate potential differences between the two classes of PD patients. Analyses of possible differences of predictor covariates (eg, age, sex, albumin concentration) were evaluated by similarly structured analysis of variance. The three analyses involved adjustments made to the SF-36 scores and consisted of (1) none, (2) case mix variables (age, sex, race, and diabetes), and (3) the combination
RESULTS
Table 1 shows the distribution of measures among patients divided by type of therapy. In general, PD patients were younger than patients receiving HD (P ⬍ 0.001), and larger fractions were white (P ⬍ 0.001). A larger fraction of HD than PD patients had diabetes (P ⬍ 0.001). However, within the PD category, a larger fraction receiving CCPD than CAPD had diabetes (P ⬍ 0.001). Serum albumin concentration tended to be less (P ⬍ 0.001) but creatinine concentration greater (P ⬍ 0.001) among PD than HD patients. Blood hemoglobin concentration was greater among PD than HD patients (P ⬍ 0.001), as was WBC count (P ⬍ 0.001). Other values are listed in Table 1. One must consider the nature of the treatments and the usual blood sampling
Table 1. Patient Characteristics and Distribution of Measures Among Patients Divided by Type of Therapy Variable
HD
PD
CAPD
CCPD
No. of patients Age (y) Sex (% men) Race (% white) Diabetes (%) Laboratory variables Albumin (g/dL) Creatinine (mg/dL) Hemoglobin (g/dL) WBC (103/µL) Potassium (mEq/L) Phosphorus (mEq/L) Iron (µg/dL) Bicarbonate (mEq/L) Ferritin (ng/dL) Aspartate aminotransferase (U/L)
16,755 59.44 ⫾ 15.28 51.78* 52.93 45.46
1,260 53.45 ⫾ 15.31 50.48* 68.97 37.98
728 53.15 ⫾ 15.10 47.25 67.31† 34.03
532 53.86 ⫾ 15.61 54.89 71.24† 43.37
3.85 ⫾ 0.38 10.16 ⫾ 3.48 10.29 ⫾ 1.15 7.17 ⫾ 2.31 4.88 ⫾ 0.65 5.90 ⫾ 1.65 57.96 ⫾ 24.42 20.25 ⫾ 2.95 381.10 ⫾ 450.62 19.48 ⫾ 14.67
3.58 ⫾ 0.48 10.80 ⫾ 4.38 10.76 ⫾ 1.52 7.84 ⫾ 2.50 4.18 ⫾ 0.71 5.49 ⫾ 1.56 64.39 ⫾ 27.11 23.97 ⫾ 2.99 306.19 ⫾ 461.22 21.58 ⫾ 11.95
3.58 ⫾ 0.48† 10.87 ⫾ 4.52† 10.70 ⫾ 1.58† 7.77 ⫾ 2.46† 4.14 ⫾ 0.73† 5.39 ⫾ 1.55 63.04 ⫾ 25.84† 24.09 ⫾ 3.11† 290.15 ⫾ 419.63† 22.03 ⫾ 12.77†
3.58 ⫾ 0.48† 10.70 ⫾ 4.19† 10.84 ⫾ 1.43† 7.92 ⫾ 2.57† 4.22 ⫾ 0.68† 5.63 ⫾ 1.56 66.21 ⫾ 28.66† 23.81 ⫾ 2.83† 327.47 ⫾ 510.92† 20.97 ⫾ 10.70†
NOTE. Means accompanied by the same symbol were not significantly different from each other (P ⬎ 0.05). *HD patients were compared with PD and both classes of PD patients. †CAPD and CCPD patients were compared (*).
SF-36 QOL EVALUATION IN HD AND PD
295
Table 2. Distribution of SF-36 Scores Among the Patient Groups and the General US Population Scale
HD
PD
CAPD
CCPD
US
PCS MCS PF RP BP GH VT SF RE MH
33.3 ⫾ 10.4 47.5 ⫾ 11.7 41.4 ⫾ 30.0 33.1 ⫾ 39.4 57.2 ⫾ 28.6 43.7 ⫾ 22.0 44.7 ⫾ 23.1 64.1 ⫾ 29.4 53.0 ⫾ 44.3 68.7 ⫾ 21.2
33.7 ⫾ 10.6 47.9 ⫾ 11.6 44.8 ⫾ 28.7 33.3 ⫾ 38.4 60.1 ⫾ 28.0 42.3 ⫾ 22.1 42.3 ⫾ 23.9 65.2 ⫾ 28.5 58.4 ⫾ 43.9 69.7 ⫾ 20.4
34.6 ⫾ 10.6 47.4 ⫾ 11.5 46.1 ⫾ 28.3 36.3 ⫾ 39.5 60.7 ⫾ 28.1 43.4 ⫾ 22.4 43.7 ⫾ 24.1 65.6 ⫾ 28.7 56.9 ⫾ 43.9 69.0 ⫾ 20.3
32.4 ⫾ 10.5 48.5 ⫾ 11.6 43.1 ⫾ 29.3 29.1 ⫾ 36.6 59.4 ⫾ 27.8 40.8 ⫾ 21.6 40.2 ⫾ 23.5 64.7 ⫾ 28.2 60.4 ⫾ 43.9 70.8 ⫾ 20.4
50.0 ⫾ 10.0 50.0 ⫾ 10.0 84.2 ⫾ 23.3 81.0 ⫾ 34.0 75.2 ⫾ 23.7 72.0 ⫾ 20.3 60.9 ⫾ 21.0 83.3 ⫾ 22.7 81.3 ⫾ 33.0 74.7 ⫾ 18.1
NOTE. Values are expressed as mean ⫾ SD.
times between them when evaluating the relative values of such measures as creatinine, potassium, phosphorus, and bicarbonate. Table 2 shows the mean ⫾ SD of SF-36 scores among the patient groups. In general, patients receiving dialysis had lower scores reflecting physical function (eg, the PCS and PF and RP scores). The scores reflecting mental functioning (eg, MCS, MH, RE, and SF) tended to be close to those observed in reference populations without kidney failure.4,5 Figure 1 compares the unadjusted PCS scores among the four groups of patients (left cluster of bars). The scores are also compared after adjustment for case mix variables (middle cluster) and case mix plus laboratory variables (right cluster). Before adjustment (left cluster of bars), there was no significant difference between scores for
Fig 1. The PCS among HD and PD patients. P over the bars indicate the significance of the difference from HD. Values in the CCPD bar denote the difference from CAPD. NS means P G 0.10.
HD (mean, 33.28) and PD patients (mean, 33.66). Conversely, CAPD patients reported significantly better PF (mean score, 34.61) than HD patients, but CCPD patients reported significantly worse scores (mean, 32.35). CCPD patients reported significantly worse PF than CAPD patients (P ⬍ 0.001). PF, however, was strongly associated with age among dialysis patients6; both the PF and PCS scores deteriorate as a linear function of age. PD patients were younger than HD patients, and, thus, one would expect better scores among PD than HD patients, all else being equal. Adjustment for case mix (middle cluster) measures, including age, suggested no significant difference among HD patients and either PD patients as a class or CAPD patients. CCPD patients, however, reported significantly worse PF than
296
either HD patients or patients receiving CAPD after adjustment for age, sex, race, and diabetes. Serum albumin concentration is also associated with PF among dialysis patients,6 in which a higher concentration is associated with better functioning. PD patients had noticeably lower serum albumin concentrations than HD patients. Therefore, adjustment for serum albumin concentration should improve the apparent scores of PD patients relative to HD. Conversely, blood hemoglobin concentration is favorably associated with PF,6 and PD patients tend toward higher blood hemoglobin levels than HD patients. Therefore, adjustment for hemoglobin should reduce the score for PD patients relative to HD. Other laboratory measures can be evaluated similarly. Such analysis is more difficult for serum creatinine concentration, however. Whereas higher levels of creatinine tend to be associated with better PF in the population at large, serum creatinine concentration is the likely product of two processes. The first is nutritional, in which a higher creatinine level is associated with better outcome, including PF. The second is the intensity of dialysis exposure, in which a higher creatinine level is associated with lower exposure. We did not (and could not) adjust for differences in treatment exposure (either dialysis or residual renal function) between these populations of patients. Therefore, the evaluation of creatinine is somewhat difficult in this model. Nonetheless, after adjustment for laboratory
DIAZ-BUXO ET AL
variables (right cluster), there appeared again no significant difference between HD patients and either PD patients as a class or CAPD patients. The scores reported by CCPD patients remained significantly less favorable than those reported by either HD patients or CAPD patients. Figure 2 is a chart of similar format showing differences in the MCS scale among the patient groups. The implications are quite different from those of the PCS. In all cases, patients receiving CCPD reported a significantly better MCS than patients receiving HD. After statistical adjustment for case mix and other variables, the score was significantly better than for patients receiving CAPD. There were no differences between HD patients and patients receiving CAPD. Figures 3 and 4 contain similar charts for the eight primary SF-36 scales. In general, we notice that adjustment for case mix or case mix plus laboratory values reduced to not significant differences in the PF scale among the groups. Evaluation of the RP scale suggested that CCPD patients regarded their physical limitations as interfering with their life to a much greater extent than either HD patients or CAPD patients. Patients on HD generally perceived greater bodily pain (BP scale) than patients on PD, but the perceptions of general health (GH scale) appear worse only among CCPD patients; the difference disappeared after statistical adjustment for laboratory measures. Vitality (VT scale) also appeared noticeably
Fig 2. MCS among HD and PD patients. The figure uses the same format as Fig 1.
SF-36 QOL EVALUATION IN HD AND PD
Fig 3.
297
Four scales reflecting physical function among HD and PD patients.
worse among CCPD than other patients, but SF appeared no different among the therapy groups. The MH scores were noticeably better among CCPD patients than either CAPD patients or patients on HD. The RE scale was also significantly better among patients receiving CCPD. We examined other files to determine the availability of residual creatinine clearance tests measured contemporaneously with laboratory tests and the SF-36 surveys. We were able to identify only 70 such tests (0.4% of patients) for HD patients, 165 tests (22.7%) for CAPD patients, and 113 tests (21.2%) for CCPD patients. Thus, estimates of residual renal function were available for only a negligible fraction of HD patients but were available for approximately one fifth of PD patients. Table 3 lists the distribution of creatinine clearance values among CAPD and CCPD patients. We evaluated the association of the PCS, MCS, and PF scales with residual creatinine clearance
for CAPD patients, CCPD patients, and all PD patients. In general, the associations were direct. A significant association (P ⫽ 0.046) was observed between the MCS and residual creatinine clearance for CAPD patients only. Considering the number of statistical tests, however, and the marginal value of the finding, it should be discounted as not very important. DISCUSSION
Generally, we give greatest weight to the case mix–adjusted models. Age and diabetes in particular are associated with many of the SF-36 scales,6 and there were differences in those parameters among these groups. Therefore, it is appropriate to make adjustments for those measures. Whereas such measures as serum albumin concentration, blood hemoglobin concentration, and WBC count are associated with the SF-36 scales,6 the values of those measures could result in part
298
DIAZ-BUXO ET AL
Fig 4.
Four scales reflecting mental function among HD and PD patients.
from the particular treatments. Therefore, one may or may not wish to make adjustments for them when contemplating possible differences of the score profiles between the treatments. It appears that there was little difference between patients receiving HD and CAPD for scales reflecting either physical or mental processes after adjustment for the differences in case Table 3. Distribution of Creatinine Clearance Among CAPD and CCPD Patients Statistic
CAPD
CCPD
No. of patients Mean SD First quartile Median Third quartile
165 4.52 4.21 1.7 3.5 5.9
113 4.90 5.35 1.4 3.7 6.4
NOTE. Creatinine clearance values are expressed in milliliters per minute.
mix measures between the treatments. The sole exception was the BP scale, which was significantly better in CAPD than in HD patients before but not after adjustment for the laboratory measures. By and large, HD patients completed the survey in conjunction with their treatment. The extent to which that timing influenced the responses to the questions is not known. The scales reflecting physical processes tended to be noticeably worse among CCPD patients than either CAPD or HD patients. That was particularly true for the composite scale (PCS) and those scales incorporating a perceptive component, such as the RP, GH, and VT scales. Adjustment for laboratory values tended to reduce but not extinguish those differences. The scales reflecting mental processes, however, tended to be noticeably better among CCPD patients. The finding is not unlike the divergent associations of the PCS and MCS with age.6
SF-36 QOL EVALUATION IN HD AND PD
Older patients tended to have worse PF than younger patients, but the mental process scales tended to be better in older patients. There were no differences in the SF scale among the case mix–adjusted groups. Adjustment for the different laboratory values (eg, lower albumin level and higher WBC count) improved estimated SF among CCPD patients so that the difference approached statistical significance. If the albumin level were higher and WBC count lower, SF would theoretically improve. Both the RE and MH scale scores, as well as the MCS, were higher in CCPD patients than among either HD or CAPD patients, and adjustment for the differences in laboratory values tended to strengthen those differences. If the laboratory values among CCPD patients were similar to those of HD patients, their MH would improve, exceeding HD and CAPD by even greater margins. We cannot evaluate the extent, if any, to which treatment selection rather than the treatments themselves influenced the mental process scale differences among the groups. Independent people, for example, with a good and assertive outlook and a high internal locus of control could choose CCPD because of the nature of that treatment. HD patients may choose HD because they are less independent and receive more external support as part of the treatment process. Such a dynamic could explain better MH and RE scales among CCPD patients and worse perceived PF (RP, GH, and VT scales), whereas objective PF was not different. Because CCPD patients perceive themselves as being more physically impaired but report better emotional adjustment than either HD or CAPD patients, the possibility exists of selection of patients with more severe physical impairments for CCPD. Automated PD is often preferred for those with limiting visual or neuromuscular disabilities. The proportion of patients with diabetes among CCPD patients was greater than that observed with CAPD. However, patients receiving HD tended to be older than patients on CCPD, and a greater fraction had diabetes. Serum albumin concentration was lower among CCPD patients than patients on HD but was not different from patients on CAPD. Hence, the supplemental case mix and laboratory measures
299
do not support this possibility. The difference in PCS between CCPD and HD and between CCPD and CAPD originates from the significant difference in the RP scale. Scores on the other three scales that primarily compose the PCS were not significant between the three dialysis modalities. That CCPD frees patients from procedures during the day may have influenced the selection of patients who experienced more role limitations in their daily life because of their physical health. Merkus et al11 observed a positive correlation between QOL and residual renal function among new end-stage renal disease patients. We could not show meaningful associations for several scales with residual renal function in small subsets of PD patients. However, one would not be surprised to find differences in native renal function associated with QOL in the same manner that age is, for example. Persons with normal renal function should have better HRQOL than persons with failed renal function, all else being equal, particularly on the physical process scales. Unfortunately, as with many epidemiological studies of dialysis patients, a compelling analysis was not possible. Furthermore, it is possible that patients were selected for CAPD or CCPD by their physicians. For example, the CCPD patients could have faster membrane transport than CAPD patients. High transporters are known to have worse outcomes than patients with lower peritoneal membrane transport capacity.12,13 However, that would not explain the disparate physical and mental component findings. When comparing HD with PD patients as a group, we could not find significant differences in either mental or physical processes, although PD patients scored higher for mental processes after adjustment. Two other cross-sectional studies failed to show differences in perceived HRQOL between HD and PD patients after adjustment for demographic characteristics, comorbidity, and duration of therapy.14,15 Other investigators have correlated the results of the SF-36 with mortality and morbidity and found it to be a good screening tool for patients at high risk for death, hospitalization, poor attendance to dialysis, and depression.16 Our data, however, are descriptive only. QOL data must be interpreted cautiously because patient selection, as well as treatment, may have a role. One should
300
DIAZ-BUXO ET AL
not conclude from them either that CCPD patients have worse PF than either HD or CAPD patients or that CCPD makes patients happier and better adjusted to end-stage renal disease. One can reasonably conclude from the data, however, that CCPD patients see themselves as more physically impaired but better adjusted than HD or CAPD patients. REFERENCES 1. Hays RD, Kallich JD, Mapes DL, Coons SJ, Carter WB: Development of the Kidney Disease Quality of Life (KDQOLTM) Instrument. Qual Life Res 3:329-338, 1994 2. Kutner NG: Assessing end-stage renal disease patients’ functioning and well-being: Measurement approaches and implications for clinical practice. Am J Kidney Dis 24:321333, 1994 3. Lowrie EG, Zhang H, LePain N, Lew NL, Lazarus JM: The association of SF-36 quality of life scales with patient mortality. Fresenius Medical Care Memorandum to Dialysis Services Division Medical Directors, Lexington, MA, January 16, 1998 4. Ware JE, Snow KK, Kosinski M, Gandek B: SF-36 health survey: Manual and interpretation guide, in The Health Data Institute of New England Medical Center. Boston, MA, Nimrod, 1993 5. Ware JE, Kosinski M, Keller SD: SF-36 physical and mental health summary scales: A user’s manual, in The Health Data Institute of New England Medical Center. Boston, MA, Nimrod, 1994 6. Lowie EG, LePain L, Zhang H, Lew NL, Lazarus JM: Preliminary evaluation of the SF-36 among hemodialysis patients. Fresenius Medical Care Memorandum to Dialysis Services Division Medical Directors, Lexington, MA, October 6, 1997. Memorandum provided online (www.ajkd.org) as ‘‘Supplementary Material’’ to the February 2000 AJKD contents. 7. McHorney CA, Ware JE, Raczek AE: The MOS 36-
item short-form health survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care 31:247-263, 1993. 8. Moson RL, Gust RF, Hess JL: Statistical Design and Analysis of Experiments. New York, NY, Wiley, 1989 9. Lowrie EG: Chronic dialysis treatment: Clinical outcome and related processes of care. Am J Kidney Dis 24:255-266, 1994 10. Owen WF, Lew NL, Liu Y, Lowrie EG, Lazarus JM: The urea reduction ratio and serum albumin concentration as predictors of mortality in patients undergoing hemodialysis. N Engl J Med 329:1001-1006, 1993 11. Merkus MP, Jager KJ, Dekker FW, Boeschoten EW, Stevens P, Krediet RT, The NECOSAD Study Group: Quality of life in patients on chronic dialysis: Self-assessment 3 months after the start of treatment. Am J Kidney Dis 29:584-592, 1997 12. Selgas R, Bajo MA, Fernandez-Reyes MJ, Bosque E, Lopez-Revuelta K, Jimenez C, Borrego F, de Alvaro F: An analysis of adequacy of dialysis in a selected population on CAPD for over 3 years: The influence of urea and creatinine kinetics. Nephrol Dial Transplant 8:1244-1253, 1993 13. 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. J Am Soc Nephrol 9:1285-1292, 1998 14. Evans RW, Manninen DL, Garrison LP, Hart LG, Blagg CR, Gutman RA, Hull AR, Lowrie EG: The quality of life of patients with end-stage renal disease. N Engl J Med 312:553-559, 1985 15. Bremer BA, McCauley CR, Wrona RM, Johnson JP: Quality of life in end-stage renal disease: A reexamination. Am J Kidney Dis 13:200-209, 1989 16. DeOreo P: Hemodialysis patient assessed functional health status predicts continued survival, hospitalization, and dialysis-attendance compliance. Am J Kidney Dis 30: 204-212, 1997
S
HORT FORM 36 (SF-36) is both a standalone measure of functioning, appropriate for use with any and all patients, and the core component of the Kidney Disease Quality of Life survey instrument.1,2 This instrument has been widely used and validated as a quality-oflife (QOL) assessment tool for the general population and for patients with congestive heart failure, kidney disease, and mental illness.3-5 Fresenius Medical Care North America started performing pilot studies of the SF-36 during mid-1995. In 1996, the company embarked on a second phase, offering the survey to all affiliated centers. By year’s end, more than 20,000 forms had been received and analyzed for hemodialysis (HD) patients. Preliminary results suggested that, in general, the physical component scores (PCSs) among dialysis patients were less than those observed in the US population and similar to those in persons with congestive heart failure.6 The mental component scores (MCSs) were similar to or slightly less than those found in the US population, but substantially better than among persons with clinical depression. Because of the lack of data on the potential differences in healthrelated QOL (HRQOL) among HD and peritoneal dialysis (PD) patients in large populations, we performed the present analysis comparing
SF-36 scores of patients undergoing each of these therapies. METHODS
Description of the SF-36 The SF-36 instrument consists of 36 questions, 35 of which are compressed into eight scales. The eight multi-item scales are as follows: (1) Physical Functioning (PF) is a 10-question scale that captures abilities to deal with the physical requirement of life, such as attending to personal needs, walking, and flexibility. (2) Role-Physical (RP) is a four-item scale that evaluates the extent to which physical capabilities limit activity. (3) Bodily Pain (BP) is a two-item scale that evaluates the perceived amount of pain experienced during the most recent 4 weeks and the extent to which that pain interfered with normal work activities. (4) General Health (GH) is a five-item scale that evaluates general health in terms of personal perception. (5) Vitality (VT) is a four-item scale that evaluates feeling of pep, energy, and fatigue. (6) Social Functioning (SF) is a two-
From Fresenius Medical Care North America, Lexington, MA. Received May 13, 1999; accepted in revised form August 20, 1999. Address reprint requests to Jose A. Diaz-Buxo, MD, Fresenius Medical Care North America, Peritoneal Dialysis Services, 1051 E Morehead St, Suite 250, Charlotte, NC 28204. E-mail: [email protected]
娀 2000 by the National Kidney Foundation, Inc. 0272-6386/00/3502-0028$3.00/0
American Journal of Kidney Diseases, Vol 35, No 2 (February), 2000: pp 293-300
293
294
DIAZ-BUXO ET AL
item scale that evaluates the extent and amount of time, if any, that physical health or emotional problems interfered with family, friends, and other social interactions during the most recent 4 weeks. (7) Role-Emotional (RE) is a threeitem scale that evaluates the extent, if any, to which emotional factors interfere with work or other activities. (8) Mental Health (MH) is a five-item scale that evaluates feelings principally of anxiety and depression. The eight scales of the SF-36 have been compressed into two primary dimensions: physical (PCS) and mental (MCS). The PF, RP, and BP scales are the primary members of the physical dimension, whereas the SF, RE, and MH are the primary members of the mental dimension. The GH and VT are considered members of both dimensions.7
of case mix and laboratory variables. These adjustments have been used in many of our previous analyses and are described elsewhere.9,10 The laboratory variables selected for analysis were those found most closely associated with the SF-36 during previous studies6 and included serum albumin, serum creatinine, blood hemoglobin, serum potassium, serum phosphorus, serum iron, bicarbonate, ferritin, and aspartate aminotransferase concentrations and white blood cell (WBC) count. The laboratory parameters were measured during the 3 months before administration of the SF-36, and the average value was reported. The urea reduction ratio, found to be weakly associated with the SF-36 scales,6 was not included because similar measures were not available for PD.
Data Collection and Statistical Analysis Data included 16,755 HD patients and 1,260 PD patients who completed an SF-36 form during 1996. Patients receiving PD were further subclassified into continuous ambulatory PD ([CAPD] n ⫽ 728) and continuous cycling PD ([CCPD] n ⫽ 532). Only complete forms were accepted for analysis. Three analyses of variance or covariance8 were performed for each of the eight primary and the two composite scales. The statistical model treated CAPD and CCPD patients as nested within the overall PD classification. Thus, we were able to evaluate potential differences between HD and PD patients in the aggregate and as they were divided into the two therapy types. We were also able to evaluate potential differences between the two classes of PD patients. Analyses of possible differences of predictor covariates (eg, age, sex, albumin concentration) were evaluated by similarly structured analysis of variance. The three analyses involved adjustments made to the SF-36 scores and consisted of (1) none, (2) case mix variables (age, sex, race, and diabetes), and (3) the combination
RESULTS
Table 1 shows the distribution of measures among patients divided by type of therapy. In general, PD patients were younger than patients receiving HD (P ⬍ 0.001), and larger fractions were white (P ⬍ 0.001). A larger fraction of HD than PD patients had diabetes (P ⬍ 0.001). However, within the PD category, a larger fraction receiving CCPD than CAPD had diabetes (P ⬍ 0.001). Serum albumin concentration tended to be less (P ⬍ 0.001) but creatinine concentration greater (P ⬍ 0.001) among PD than HD patients. Blood hemoglobin concentration was greater among PD than HD patients (P ⬍ 0.001), as was WBC count (P ⬍ 0.001). Other values are listed in Table 1. One must consider the nature of the treatments and the usual blood sampling
Table 1. Patient Characteristics and Distribution of Measures Among Patients Divided by Type of Therapy Variable
HD
PD
CAPD
CCPD
No. of patients Age (y) Sex (% men) Race (% white) Diabetes (%) Laboratory variables Albumin (g/dL) Creatinine (mg/dL) Hemoglobin (g/dL) WBC (103/µL) Potassium (mEq/L) Phosphorus (mEq/L) Iron (µg/dL) Bicarbonate (mEq/L) Ferritin (ng/dL) Aspartate aminotransferase (U/L)
16,755 59.44 ⫾ 15.28 51.78* 52.93 45.46
1,260 53.45 ⫾ 15.31 50.48* 68.97 37.98
728 53.15 ⫾ 15.10 47.25 67.31† 34.03
532 53.86 ⫾ 15.61 54.89 71.24† 43.37
3.85 ⫾ 0.38 10.16 ⫾ 3.48 10.29 ⫾ 1.15 7.17 ⫾ 2.31 4.88 ⫾ 0.65 5.90 ⫾ 1.65 57.96 ⫾ 24.42 20.25 ⫾ 2.95 381.10 ⫾ 450.62 19.48 ⫾ 14.67
3.58 ⫾ 0.48 10.80 ⫾ 4.38 10.76 ⫾ 1.52 7.84 ⫾ 2.50 4.18 ⫾ 0.71 5.49 ⫾ 1.56 64.39 ⫾ 27.11 23.97 ⫾ 2.99 306.19 ⫾ 461.22 21.58 ⫾ 11.95
3.58 ⫾ 0.48† 10.87 ⫾ 4.52† 10.70 ⫾ 1.58† 7.77 ⫾ 2.46† 4.14 ⫾ 0.73† 5.39 ⫾ 1.55 63.04 ⫾ 25.84† 24.09 ⫾ 3.11† 290.15 ⫾ 419.63† 22.03 ⫾ 12.77†
3.58 ⫾ 0.48† 10.70 ⫾ 4.19† 10.84 ⫾ 1.43† 7.92 ⫾ 2.57† 4.22 ⫾ 0.68† 5.63 ⫾ 1.56 66.21 ⫾ 28.66† 23.81 ⫾ 2.83† 327.47 ⫾ 510.92† 20.97 ⫾ 10.70†
NOTE. Means accompanied by the same symbol were not significantly different from each other (P ⬎ 0.05). *HD patients were compared with PD and both classes of PD patients. †CAPD and CCPD patients were compared (*).
SF-36 QOL EVALUATION IN HD AND PD
295
Table 2. Distribution of SF-36 Scores Among the Patient Groups and the General US Population Scale
HD
PD
CAPD
CCPD
US
PCS MCS PF RP BP GH VT SF RE MH
33.3 ⫾ 10.4 47.5 ⫾ 11.7 41.4 ⫾ 30.0 33.1 ⫾ 39.4 57.2 ⫾ 28.6 43.7 ⫾ 22.0 44.7 ⫾ 23.1 64.1 ⫾ 29.4 53.0 ⫾ 44.3 68.7 ⫾ 21.2
33.7 ⫾ 10.6 47.9 ⫾ 11.6 44.8 ⫾ 28.7 33.3 ⫾ 38.4 60.1 ⫾ 28.0 42.3 ⫾ 22.1 42.3 ⫾ 23.9 65.2 ⫾ 28.5 58.4 ⫾ 43.9 69.7 ⫾ 20.4
34.6 ⫾ 10.6 47.4 ⫾ 11.5 46.1 ⫾ 28.3 36.3 ⫾ 39.5 60.7 ⫾ 28.1 43.4 ⫾ 22.4 43.7 ⫾ 24.1 65.6 ⫾ 28.7 56.9 ⫾ 43.9 69.0 ⫾ 20.3
32.4 ⫾ 10.5 48.5 ⫾ 11.6 43.1 ⫾ 29.3 29.1 ⫾ 36.6 59.4 ⫾ 27.8 40.8 ⫾ 21.6 40.2 ⫾ 23.5 64.7 ⫾ 28.2 60.4 ⫾ 43.9 70.8 ⫾ 20.4
50.0 ⫾ 10.0 50.0 ⫾ 10.0 84.2 ⫾ 23.3 81.0 ⫾ 34.0 75.2 ⫾ 23.7 72.0 ⫾ 20.3 60.9 ⫾ 21.0 83.3 ⫾ 22.7 81.3 ⫾ 33.0 74.7 ⫾ 18.1
NOTE. Values are expressed as mean ⫾ SD.
times between them when evaluating the relative values of such measures as creatinine, potassium, phosphorus, and bicarbonate. Table 2 shows the mean ⫾ SD of SF-36 scores among the patient groups. In general, patients receiving dialysis had lower scores reflecting physical function (eg, the PCS and PF and RP scores). The scores reflecting mental functioning (eg, MCS, MH, RE, and SF) tended to be close to those observed in reference populations without kidney failure.4,5 Figure 1 compares the unadjusted PCS scores among the four groups of patients (left cluster of bars). The scores are also compared after adjustment for case mix variables (middle cluster) and case mix plus laboratory variables (right cluster). Before adjustment (left cluster of bars), there was no significant difference between scores for
Fig 1. The PCS among HD and PD patients. P over the bars indicate the significance of the difference from HD. Values in the CCPD bar denote the difference from CAPD. NS means P G 0.10.
HD (mean, 33.28) and PD patients (mean, 33.66). Conversely, CAPD patients reported significantly better PF (mean score, 34.61) than HD patients, but CCPD patients reported significantly worse scores (mean, 32.35). CCPD patients reported significantly worse PF than CAPD patients (P ⬍ 0.001). PF, however, was strongly associated with age among dialysis patients6; both the PF and PCS scores deteriorate as a linear function of age. PD patients were younger than HD patients, and, thus, one would expect better scores among PD than HD patients, all else being equal. Adjustment for case mix (middle cluster) measures, including age, suggested no significant difference among HD patients and either PD patients as a class or CAPD patients. CCPD patients, however, reported significantly worse PF than
296
either HD patients or patients receiving CAPD after adjustment for age, sex, race, and diabetes. Serum albumin concentration is also associated with PF among dialysis patients,6 in which a higher concentration is associated with better functioning. PD patients had noticeably lower serum albumin concentrations than HD patients. Therefore, adjustment for serum albumin concentration should improve the apparent scores of PD patients relative to HD. Conversely, blood hemoglobin concentration is favorably associated with PF,6 and PD patients tend toward higher blood hemoglobin levels than HD patients. Therefore, adjustment for hemoglobin should reduce the score for PD patients relative to HD. Other laboratory measures can be evaluated similarly. Such analysis is more difficult for serum creatinine concentration, however. Whereas higher levels of creatinine tend to be associated with better PF in the population at large, serum creatinine concentration is the likely product of two processes. The first is nutritional, in which a higher creatinine level is associated with better outcome, including PF. The second is the intensity of dialysis exposure, in which a higher creatinine level is associated with lower exposure. We did not (and could not) adjust for differences in treatment exposure (either dialysis or residual renal function) between these populations of patients. Therefore, the evaluation of creatinine is somewhat difficult in this model. Nonetheless, after adjustment for laboratory
DIAZ-BUXO ET AL
variables (right cluster), there appeared again no significant difference between HD patients and either PD patients as a class or CAPD patients. The scores reported by CCPD patients remained significantly less favorable than those reported by either HD patients or CAPD patients. Figure 2 is a chart of similar format showing differences in the MCS scale among the patient groups. The implications are quite different from those of the PCS. In all cases, patients receiving CCPD reported a significantly better MCS than patients receiving HD. After statistical adjustment for case mix and other variables, the score was significantly better than for patients receiving CAPD. There were no differences between HD patients and patients receiving CAPD. Figures 3 and 4 contain similar charts for the eight primary SF-36 scales. In general, we notice that adjustment for case mix or case mix plus laboratory values reduced to not significant differences in the PF scale among the groups. Evaluation of the RP scale suggested that CCPD patients regarded their physical limitations as interfering with their life to a much greater extent than either HD patients or CAPD patients. Patients on HD generally perceived greater bodily pain (BP scale) than patients on PD, but the perceptions of general health (GH scale) appear worse only among CCPD patients; the difference disappeared after statistical adjustment for laboratory measures. Vitality (VT scale) also appeared noticeably
Fig 2. MCS among HD and PD patients. The figure uses the same format as Fig 1.
SF-36 QOL EVALUATION IN HD AND PD
Fig 3.
297
Four scales reflecting physical function among HD and PD patients.
worse among CCPD than other patients, but SF appeared no different among the therapy groups. The MH scores were noticeably better among CCPD patients than either CAPD patients or patients on HD. The RE scale was also significantly better among patients receiving CCPD. We examined other files to determine the availability of residual creatinine clearance tests measured contemporaneously with laboratory tests and the SF-36 surveys. We were able to identify only 70 such tests (0.4% of patients) for HD patients, 165 tests (22.7%) for CAPD patients, and 113 tests (21.2%) for CCPD patients. Thus, estimates of residual renal function were available for only a negligible fraction of HD patients but were available for approximately one fifth of PD patients. Table 3 lists the distribution of creatinine clearance values among CAPD and CCPD patients. We evaluated the association of the PCS, MCS, and PF scales with residual creatinine clearance
for CAPD patients, CCPD patients, and all PD patients. In general, the associations were direct. A significant association (P ⫽ 0.046) was observed between the MCS and residual creatinine clearance for CAPD patients only. Considering the number of statistical tests, however, and the marginal value of the finding, it should be discounted as not very important. DISCUSSION
Generally, we give greatest weight to the case mix–adjusted models. Age and diabetes in particular are associated with many of the SF-36 scales,6 and there were differences in those parameters among these groups. Therefore, it is appropriate to make adjustments for those measures. Whereas such measures as serum albumin concentration, blood hemoglobin concentration, and WBC count are associated with the SF-36 scales,6 the values of those measures could result in part
298
DIAZ-BUXO ET AL
Fig 4.
Four scales reflecting mental function among HD and PD patients.
from the particular treatments. Therefore, one may or may not wish to make adjustments for them when contemplating possible differences of the score profiles between the treatments. It appears that there was little difference between patients receiving HD and CAPD for scales reflecting either physical or mental processes after adjustment for the differences in case Table 3. Distribution of Creatinine Clearance Among CAPD and CCPD Patients Statistic
CAPD
CCPD
No. of patients Mean SD First quartile Median Third quartile
165 4.52 4.21 1.7 3.5 5.9
113 4.90 5.35 1.4 3.7 6.4
NOTE. Creatinine clearance values are expressed in milliliters per minute.
mix measures between the treatments. The sole exception was the BP scale, which was significantly better in CAPD than in HD patients before but not after adjustment for the laboratory measures. By and large, HD patients completed the survey in conjunction with their treatment. The extent to which that timing influenced the responses to the questions is not known. The scales reflecting physical processes tended to be noticeably worse among CCPD patients than either CAPD or HD patients. That was particularly true for the composite scale (PCS) and those scales incorporating a perceptive component, such as the RP, GH, and VT scales. Adjustment for laboratory values tended to reduce but not extinguish those differences. The scales reflecting mental processes, however, tended to be noticeably better among CCPD patients. The finding is not unlike the divergent associations of the PCS and MCS with age.6
SF-36 QOL EVALUATION IN HD AND PD
Older patients tended to have worse PF than younger patients, but the mental process scales tended to be better in older patients. There were no differences in the SF scale among the case mix–adjusted groups. Adjustment for the different laboratory values (eg, lower albumin level and higher WBC count) improved estimated SF among CCPD patients so that the difference approached statistical significance. If the albumin level were higher and WBC count lower, SF would theoretically improve. Both the RE and MH scale scores, as well as the MCS, were higher in CCPD patients than among either HD or CAPD patients, and adjustment for the differences in laboratory values tended to strengthen those differences. If the laboratory values among CCPD patients were similar to those of HD patients, their MH would improve, exceeding HD and CAPD by even greater margins. We cannot evaluate the extent, if any, to which treatment selection rather than the treatments themselves influenced the mental process scale differences among the groups. Independent people, for example, with a good and assertive outlook and a high internal locus of control could choose CCPD because of the nature of that treatment. HD patients may choose HD because they are less independent and receive more external support as part of the treatment process. Such a dynamic could explain better MH and RE scales among CCPD patients and worse perceived PF (RP, GH, and VT scales), whereas objective PF was not different. Because CCPD patients perceive themselves as being more physically impaired but report better emotional adjustment than either HD or CAPD patients, the possibility exists of selection of patients with more severe physical impairments for CCPD. Automated PD is often preferred for those with limiting visual or neuromuscular disabilities. The proportion of patients with diabetes among CCPD patients was greater than that observed with CAPD. However, patients receiving HD tended to be older than patients on CCPD, and a greater fraction had diabetes. Serum albumin concentration was lower among CCPD patients than patients on HD but was not different from patients on CAPD. Hence, the supplemental case mix and laboratory measures
299
do not support this possibility. The difference in PCS between CCPD and HD and between CCPD and CAPD originates from the significant difference in the RP scale. Scores on the other three scales that primarily compose the PCS were not significant between the three dialysis modalities. That CCPD frees patients from procedures during the day may have influenced the selection of patients who experienced more role limitations in their daily life because of their physical health. Merkus et al11 observed a positive correlation between QOL and residual renal function among new end-stage renal disease patients. We could not show meaningful associations for several scales with residual renal function in small subsets of PD patients. However, one would not be surprised to find differences in native renal function associated with QOL in the same manner that age is, for example. Persons with normal renal function should have better HRQOL than persons with failed renal function, all else being equal, particularly on the physical process scales. Unfortunately, as with many epidemiological studies of dialysis patients, a compelling analysis was not possible. Furthermore, it is possible that patients were selected for CAPD or CCPD by their physicians. For example, the CCPD patients could have faster membrane transport than CAPD patients. High transporters are known to have worse outcomes than patients with lower peritoneal membrane transport capacity.12,13 However, that would not explain the disparate physical and mental component findings. When comparing HD with PD patients as a group, we could not find significant differences in either mental or physical processes, although PD patients scored higher for mental processes after adjustment. Two other cross-sectional studies failed to show differences in perceived HRQOL between HD and PD patients after adjustment for demographic characteristics, comorbidity, and duration of therapy.14,15 Other investigators have correlated the results of the SF-36 with mortality and morbidity and found it to be a good screening tool for patients at high risk for death, hospitalization, poor attendance to dialysis, and depression.16 Our data, however, are descriptive only. QOL data must be interpreted cautiously because patient selection, as well as treatment, may have a role. One should
300
DIAZ-BUXO ET AL
not conclude from them either that CCPD patients have worse PF than either HD or CAPD patients or that CCPD makes patients happier and better adjusted to end-stage renal disease. One can reasonably conclude from the data, however, that CCPD patients see themselves as more physically impaired but better adjusted than HD or CAPD patients. REFERENCES 1. Hays RD, Kallich JD, Mapes DL, Coons SJ, Carter WB: Development of the Kidney Disease Quality of Life (KDQOLTM) Instrument. Qual Life Res 3:329-338, 1994 2. Kutner NG: Assessing end-stage renal disease patients’ functioning and well-being: Measurement approaches and implications for clinical practice. Am J Kidney Dis 24:321333, 1994 3. Lowrie EG, Zhang H, LePain N, Lew NL, Lazarus JM: The association of SF-36 quality of life scales with patient mortality. Fresenius Medical Care Memorandum to Dialysis Services Division Medical Directors, Lexington, MA, January 16, 1998 4. Ware JE, Snow KK, Kosinski M, Gandek B: SF-36 health survey: Manual and interpretation guide, in The Health Data Institute of New England Medical Center. Boston, MA, Nimrod, 1993 5. Ware JE, Kosinski M, Keller SD: SF-36 physical and mental health summary scales: A user’s manual, in The Health Data Institute of New England Medical Center. Boston, MA, Nimrod, 1994 6. Lowie EG, LePain L, Zhang H, Lew NL, Lazarus JM: Preliminary evaluation of the SF-36 among hemodialysis patients. Fresenius Medical Care Memorandum to Dialysis Services Division Medical Directors, Lexington, MA, October 6, 1997. Memorandum provided online (www.ajkd.org) as ‘‘Supplementary Material’’ to the February 2000 AJKD contents. 7. McHorney CA, Ware JE, Raczek AE: The MOS 36-
item short-form health survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care 31:247-263, 1993. 8. Moson RL, Gust RF, Hess JL: Statistical Design and Analysis of Experiments. New York, NY, Wiley, 1989 9. Lowrie EG: Chronic dialysis treatment: Clinical outcome and related processes of care. Am J Kidney Dis 24:255-266, 1994 10. Owen WF, Lew NL, Liu Y, Lowrie EG, Lazarus JM: The urea reduction ratio and serum albumin concentration as predictors of mortality in patients undergoing hemodialysis. N Engl J Med 329:1001-1006, 1993 11. Merkus MP, Jager KJ, Dekker FW, Boeschoten EW, Stevens P, Krediet RT, The NECOSAD Study Group: Quality of life in patients on chronic dialysis: Self-assessment 3 months after the start of treatment. Am J Kidney Dis 29:584-592, 1997 12. Selgas R, Bajo MA, Fernandez-Reyes MJ, Bosque E, Lopez-Revuelta K, Jimenez C, Borrego F, de Alvaro F: An analysis of adequacy of dialysis in a selected population on CAPD for over 3 years: The influence of urea and creatinine kinetics. Nephrol Dial Transplant 8:1244-1253, 1993 13. 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. J Am Soc Nephrol 9:1285-1292, 1998 14. Evans RW, Manninen DL, Garrison LP, Hart LG, Blagg CR, Gutman RA, Hull AR, Lowrie EG: The quality of life of patients with end-stage renal disease. N Engl J Med 312:553-559, 1985 15. Bremer BA, McCauley CR, Wrona RM, Johnson JP: Quality of life in end-stage renal disease: A reexamination. Am J Kidney Dis 13:200-209, 1989 16. DeOreo P: Hemodialysis patient assessed functional health status predicts continued survival, hospitalization, and dialysis-attendance compliance. Am J Kidney Dis 30: 204-212, 1997