Nephrol Dial Transplant (1991) 6: 444-451 © 1991 European Dialysis and Transplant Association-European Renal Association
Nephrology Dialysis Transplantation
Original Article
M. A. Gentil, A. Carriazo, M. I. Pavon, M. Rosado, D. Castillo, B. Ramos, G. R. Algarra, F. Tejuca, V. P. Banasco and J. A. Milan Andalusian Renal Patients Registry, Office of Information and Informatics, Andalusian Health Service, Sevilla, Spain
Abstract. We compared the survival of 842 patients on centre haemodialysis to 272 patients on continuous ambulatory peritoneal dialysis (CAPD). All patients selected had begun treatment between 1 January 1984 and 30 June 1988 and were from six centres which participate in a regional renal patients registry. Patients on CAPD were older and had a greater proportion of diabetes and other associated diseases. Age, diabetes, and cardiovascular diseases were associated with a shorter survival on treatment in all the patients studied. Without adjustment for risk factors, patient 3-year survival was higher in centre haemodialysis than in CAPD, 80% versus 64% respectively. However, no significant differences could be shown in the survival rates of the two treatment modalities after accounting for the heterogeneity of the patients in the two groups, either by stratification or by multivariate analysis (Cox). Age was the main predictive factor for CAPD patient survival, while the influence of diabetes and cardiovascular diseases was less clear. Technique survival was much better in centre haemodialysis (94% versus 56% in CAPD, 3-year survival). Older age and diabetes mellitus were associated with a greater risk of switching from centre haemodialysis to CAPD and a trend to retain those patients on CAPD.
Key words: End-stage renal disease; Continuous ambulatory peritoneal dialysis; Haemodialysis; Renal replacement therapy; Survival
Introduction Ten years after the description of continuous ambulatory peritoneal dialysis (CAPD), its role in end-stage renal disease (ESRD) treatment remains controversial. Different studies tend to present CAPD as equivalent to hospital haemodialysis [1-5] or even better for certain groups of patients [6]. Many studies have been conducted in centres or regions with a broad experience in CAPD and where utilisation frequency approached or even surpassed centre haemodialysis. However, this situation is uncommon in Spain and in Europe as a whole [7], where the proportion of CAPD patients in ESRD treatment is no more than 10%. In Andalusia, CAPD was instituted early [8], but has had a limited dissemination, a development similar to the rest of the European community. A regional registry on renal data, begun in December 1983, has allowed us to analyse our experience with these dialysis methods.
Methods Correspondence and offprint requests to: Dr Ana Carriazo, Oficina de Informaci6n e Informatica, Servicio Andaluz de Salud, c/Federico Sanchez Bedoya, 3, 41001 Sevilla, Spain.
The data used here are from the Andalusian Renal Patients Registry, compiled from all dialysis and trans-
Downloaded from http://ndt.oxfordjournals.org/ at Fudan University on May 13, 2015
Comparison of Survival in Continuous Ambulatory Peritoneal Dialysis and Hospital Haemodialysis: A Multicentric Study
Survival in HD and CAPD
starting dialysis, year of entering the programme and previous stage in intermittent peritoneal dialysis as potential risk factors. We classified as high-risk patients those who were in at least one of the following groups: age greater than 65, suffering from diabetes mellitus, and those with cardiovascular diseases. In the multivariate analysis we used these five risk factors as covariates, plus the use of CAPD or centre haemodialysis as the first treatment method. For the statistical analysis, we used the BMDP software. For the univariate analysis, we applied the Kaplan—Meier survival curves, using the Mantel and Breslow tests. In the multivariate analysis, Cox's proportional hazards regression function [11] was used, with stepwise regression and the Wald test for the elimination of other factors, to test the independent effect of each covariate. Except for the year of entrance into the programme, patients were classified in dichotomous categories for each variate, i.e. 0 for the absence and 1 for the presence of the factor (age greater than 60, diabetes, cardiovascular disease, previous intermittent peritoneal dialysis or use of CAPD as first option). In the Cox model, ezP represents the influence of each covariate on the hazard function, where (3 is the estimated coefficient and z the covariate value. Thus, the exponent takes the value 1 for the absence of the study factor (z = 0) and for z = 1 represents the factor by which the basic hazard rate is multiplied (relative risk).
Results Patient Characteristics The characteristics of the patients in the two treatment groups were substantially different. The distribution by age groups was significantly different, with a clear predominance of older patients among those on CAPD (Fig. 1). Twenty-three per cent of the CAPD patients were over 65 years of age by the time they started the technique, versus 9% of the centre haemodialysis patients. There was no difference in the sex distribution (56% males in centre haemodialysis and 57% in CAPD). Most diabetic patients were included in CAPD: 78 versus only 19 in centre haemodialysis (28.7% and 2.3% of the total, P < 0.0001). Moreover, diabetes mellitus was an associated disease, not defined as cause of the ESRD, in an additional 1.4% of CAPD and 1.7% of centre haemodialysis patients. The differences observed in the other aetiologies (Table 1) were attributable to the difference in the age distribution of the two populations, with the exception of the frequency of polycystosis in the CAPD group, which was less than half.
Downloaded from http://ndt.oxfordjournals.org/ at Fudan University on May 13, 2015
plantation centres in the region (population in 1988, 7 million) since December 1983. A record of basic data for each patient with ESRD starting renal replacement therapy in the region had been kept and updated twice yearly. The contents and codes are similar to those used by the European Dialysis and Transplant Association (EDTA) Registry, and have been described elsewhere [9], along with our data-processing methodology. The presence of any chronic disease or condition affecting patient's survival was registered. As all centres reported regularly, we can assume that the registry covers almost 100% of the area's patients. This study is limited to those six main nephrology centres (and their satellites), which, during the study period (1 January 1984 to 30 June 1988), admitted a minimum of 100 new patients into a dialysis programme with at least 30 in CAPD. We ensured the existence of a basic organisation and a certain level of experience in the use of CAPD by choosing only centres with at least two years experience prior to the beginning of the study. We included patients who were first dialysed during the study period, considering their first stage in either of two methods: centre haemodialysis (n = 842) or CAPD (n = 272). Seven per cent of the centre haemodialysis cases and 45% of the CAPD patients had had previous treatment by hospital intermittent peritoneal dialysis, almost always lasting less than a month. In each survival analysis study, individuals are followed until death, change to other treatment modality, end of study, or loss to follow-up due to transfer from Andalusia, or renal function recovery. Since the registry covers all centres in Andalusia, transfers from one dialysis unit to another did not disrupt follow-up. In each separate analysis the events considered as response and as loss to follow-up vary. For patient survival 'on treatment', change of treatment modality is considered as loss to follow-up; for survival 'after treatment', the patient is followed until he or she dies, regardless of the dialysis method used. In both cases, transplantation is counted as loss to follow-up. This separate analysis allows us to take into account early deaths occurring after changes in treatment methods [10]. For analysis of technique survival, we first considered patient death or transplantation as loss to follow-up; in a second analysis the change to home haemodialysis was considered as loss to follow-up as well. For the treatment success curves, both the patient's death and the change of dialysis method are considered as failure. Again transplantation, and in a second analysis, the change to home haemodialysis, was counted as loss to follow-up. We examined the influence of age at the time of starting dialysis, the presence of diabetes mellitus as a cause of ESRD or as co-morbidity, chronic cardiovascular disease (including all cardiopathies and arteriopathies with significant manifestations) at the time of
445
M. A. Gentil et al
446
CAPO -
HHD
0%
25% 25%
50% 50% % patients
0-14 y
H M 15-34 y
55-64 y
M
75% I
100%
I 35-54 y
> 65 y
Fig. 1. Per cent distribution by age groups. Comparison between CHD and CAPD patients.
First treatment modality
Glomerulonephritis Interstitial Cystic Hereditary Vascular Systemic Other Unknown
Table 2. Patients' outcome
CAPD
CHD n
(%)
n
(%)
190 161 99 16 51 52 32 222
(23) (20) (12)
38 43 11 3 14 11 3 71
(20) (22)
(2)
(6) (6) (4)
(27)
(6) (2) (7) (6) (2)
(35)
The percentages have been calculated over the total of non-diabetic patients.
A greater proportion of the CAPD patients presented with other chronic conditions at the time of entry into the dialysis programme (38% versus 30% of the centre haemodialysis patients, P = 0.047). But above all, there was a higher frequency of concomitant cardiovascular diseases: 19.9% in CAPD and 10.7% in centre haemodialysis (P = 0.0001). Cardiovascular disease was highly correlated with age distribution and varied among all the patients from 0% in the under-15 group to 28% in the over-65s. Once adjusted for the age distribution, cardiopathy remained 30% greater in the CAPD group, but without reaching statistical significance (P < 0.9).
Patients Outcome Table 2 reflects the patients' situation at the end of the study period. A greater proportion of the CAPD patients had died or had changed their treatment method, whereas the centre haemodialysis patients were transplanted in a higher proportion than the CAPD patients (14% versus 4%). The characteristics of
First treatment modality CAPD
CHD
Death Transplant Change to ' Home HD CAPD-PD CHD
Lost to follow-up Still on first treatment
n
(%)
107 117
(13) (14)
48 12
(18)
42 44 — 30
(5) (5)
(2) (2)
502
(60)
5 5 73 7 122
(3)
n
(%)
(4)
(27) (2)
(45)
'22 HHD patients and 23 CAPD patients died after the change of treatment.
The comparison of causes of death (Table 4) demonstrates a twofold risk of infectious processes in CAPD (21% versus 9%, P = 0.048), particularly of acute peritonitis (13% of the deaths in CAPD and only 1% in centre haemodiaiysis, P < 0.006). All deaths due to acute perionitis occurred in the first 2 years of the study and within the over-55 age group. Five deaths were from the same centre, whereas four centres had no deaths due to acute peritonitis in the same period. Two lethal episodes of sclerosing peritonitis were reported in patients in CAPD, one of them after changing to centre haemodialysis. There were no marked differences between the two groups with respect to other causes of mortality, including cardiovascular mortality, the leading cause of death in both groups. However, thee were almost three times more undetermined deaths in centre haemodialysis (22% versus 8%, P = 0.046). Most of these were sudden
Downloaded from http://ndt.oxfordjournals.org/ at Fudan University on May 13, 2015
Table 1. ESRD aetiology (diabetes mellitus excluded)
the patients remaining in treatment were very similar to those at the beginning in each group (Table 3). Among the deceased, older age, diabetes mellitus and cardiovascular pathology was twice as frequent as among those alive at initiation of the study; in contrast, these frequencies are lower in the patients leaving the technique. Both selections resulted in a group that remained in treatment with characteristics similar to the starting one (with a slight increment of diabetic patients in CAPD). We must take into account (Table 2) that the mainflowof patients in centre haemodialysis is elective, towards transplantation or home haemodialysis, treatments rarely used in our region for patients older than 60 or with comorbid conditions. On the other hand, in CAPD the movement of lower-risk patients was towards centre haemodialysis, as a consequence of technique dropout.
Survival in HD and CAPD
447
Table 3. Patient characteristics by outcome
Table 5. Patient survival by characteristics
Frequency of character (%)'i Treatment
>65 years Cardiovascular Diabetes
Patients («)
842
9
11
4
Age at entry (yr)** 0-15 15-34 35-54 55-64
233 107
5 21
8 26
2 8
>64 Sex
502
8
8
4
272
23
20
30
Male Female Diabetes* No Yes
Patients («)
6 months
1 year
2 years
18 222 412 311 134
100 98 98 92 85
100 96 95 78
*** 95 91 74 58
629 481
95 94
90 91
83 82
998 116
95 91
91 87
84 68
748 222 144
96 94 88
92 93 81
86 84 68
87
Comorbidity** No 102 48
12 48
15 27
15 35
122
21
21
38
Other diseases Cardiovascular
*P
Nephrology Dialysis Transplantation
Original Article
M. A. Gentil, A. Carriazo, M. I. Pavon, M. Rosado, D. Castillo, B. Ramos, G. R. Algarra, F. Tejuca, V. P. Banasco and J. A. Milan Andalusian Renal Patients Registry, Office of Information and Informatics, Andalusian Health Service, Sevilla, Spain
Abstract. We compared the survival of 842 patients on centre haemodialysis to 272 patients on continuous ambulatory peritoneal dialysis (CAPD). All patients selected had begun treatment between 1 January 1984 and 30 June 1988 and were from six centres which participate in a regional renal patients registry. Patients on CAPD were older and had a greater proportion of diabetes and other associated diseases. Age, diabetes, and cardiovascular diseases were associated with a shorter survival on treatment in all the patients studied. Without adjustment for risk factors, patient 3-year survival was higher in centre haemodialysis than in CAPD, 80% versus 64% respectively. However, no significant differences could be shown in the survival rates of the two treatment modalities after accounting for the heterogeneity of the patients in the two groups, either by stratification or by multivariate analysis (Cox). Age was the main predictive factor for CAPD patient survival, while the influence of diabetes and cardiovascular diseases was less clear. Technique survival was much better in centre haemodialysis (94% versus 56% in CAPD, 3-year survival). Older age and diabetes mellitus were associated with a greater risk of switching from centre haemodialysis to CAPD and a trend to retain those patients on CAPD.
Key words: End-stage renal disease; Continuous ambulatory peritoneal dialysis; Haemodialysis; Renal replacement therapy; Survival
Introduction Ten years after the description of continuous ambulatory peritoneal dialysis (CAPD), its role in end-stage renal disease (ESRD) treatment remains controversial. Different studies tend to present CAPD as equivalent to hospital haemodialysis [1-5] or even better for certain groups of patients [6]. Many studies have been conducted in centres or regions with a broad experience in CAPD and where utilisation frequency approached or even surpassed centre haemodialysis. However, this situation is uncommon in Spain and in Europe as a whole [7], where the proportion of CAPD patients in ESRD treatment is no more than 10%. In Andalusia, CAPD was instituted early [8], but has had a limited dissemination, a development similar to the rest of the European community. A regional registry on renal data, begun in December 1983, has allowed us to analyse our experience with these dialysis methods.
Methods Correspondence and offprint requests to: Dr Ana Carriazo, Oficina de Informaci6n e Informatica, Servicio Andaluz de Salud, c/Federico Sanchez Bedoya, 3, 41001 Sevilla, Spain.
The data used here are from the Andalusian Renal Patients Registry, compiled from all dialysis and trans-
Downloaded from http://ndt.oxfordjournals.org/ at Fudan University on May 13, 2015
Comparison of Survival in Continuous Ambulatory Peritoneal Dialysis and Hospital Haemodialysis: A Multicentric Study
Survival in HD and CAPD
starting dialysis, year of entering the programme and previous stage in intermittent peritoneal dialysis as potential risk factors. We classified as high-risk patients those who were in at least one of the following groups: age greater than 65, suffering from diabetes mellitus, and those with cardiovascular diseases. In the multivariate analysis we used these five risk factors as covariates, plus the use of CAPD or centre haemodialysis as the first treatment method. For the statistical analysis, we used the BMDP software. For the univariate analysis, we applied the Kaplan—Meier survival curves, using the Mantel and Breslow tests. In the multivariate analysis, Cox's proportional hazards regression function [11] was used, with stepwise regression and the Wald test for the elimination of other factors, to test the independent effect of each covariate. Except for the year of entrance into the programme, patients were classified in dichotomous categories for each variate, i.e. 0 for the absence and 1 for the presence of the factor (age greater than 60, diabetes, cardiovascular disease, previous intermittent peritoneal dialysis or use of CAPD as first option). In the Cox model, ezP represents the influence of each covariate on the hazard function, where (3 is the estimated coefficient and z the covariate value. Thus, the exponent takes the value 1 for the absence of the study factor (z = 0) and for z = 1 represents the factor by which the basic hazard rate is multiplied (relative risk).
Results Patient Characteristics The characteristics of the patients in the two treatment groups were substantially different. The distribution by age groups was significantly different, with a clear predominance of older patients among those on CAPD (Fig. 1). Twenty-three per cent of the CAPD patients were over 65 years of age by the time they started the technique, versus 9% of the centre haemodialysis patients. There was no difference in the sex distribution (56% males in centre haemodialysis and 57% in CAPD). Most diabetic patients were included in CAPD: 78 versus only 19 in centre haemodialysis (28.7% and 2.3% of the total, P < 0.0001). Moreover, diabetes mellitus was an associated disease, not defined as cause of the ESRD, in an additional 1.4% of CAPD and 1.7% of centre haemodialysis patients. The differences observed in the other aetiologies (Table 1) were attributable to the difference in the age distribution of the two populations, with the exception of the frequency of polycystosis in the CAPD group, which was less than half.
Downloaded from http://ndt.oxfordjournals.org/ at Fudan University on May 13, 2015
plantation centres in the region (population in 1988, 7 million) since December 1983. A record of basic data for each patient with ESRD starting renal replacement therapy in the region had been kept and updated twice yearly. The contents and codes are similar to those used by the European Dialysis and Transplant Association (EDTA) Registry, and have been described elsewhere [9], along with our data-processing methodology. The presence of any chronic disease or condition affecting patient's survival was registered. As all centres reported regularly, we can assume that the registry covers almost 100% of the area's patients. This study is limited to those six main nephrology centres (and their satellites), which, during the study period (1 January 1984 to 30 June 1988), admitted a minimum of 100 new patients into a dialysis programme with at least 30 in CAPD. We ensured the existence of a basic organisation and a certain level of experience in the use of CAPD by choosing only centres with at least two years experience prior to the beginning of the study. We included patients who were first dialysed during the study period, considering their first stage in either of two methods: centre haemodialysis (n = 842) or CAPD (n = 272). Seven per cent of the centre haemodialysis cases and 45% of the CAPD patients had had previous treatment by hospital intermittent peritoneal dialysis, almost always lasting less than a month. In each survival analysis study, individuals are followed until death, change to other treatment modality, end of study, or loss to follow-up due to transfer from Andalusia, or renal function recovery. Since the registry covers all centres in Andalusia, transfers from one dialysis unit to another did not disrupt follow-up. In each separate analysis the events considered as response and as loss to follow-up vary. For patient survival 'on treatment', change of treatment modality is considered as loss to follow-up; for survival 'after treatment', the patient is followed until he or she dies, regardless of the dialysis method used. In both cases, transplantation is counted as loss to follow-up. This separate analysis allows us to take into account early deaths occurring after changes in treatment methods [10]. For analysis of technique survival, we first considered patient death or transplantation as loss to follow-up; in a second analysis the change to home haemodialysis was considered as loss to follow-up as well. For the treatment success curves, both the patient's death and the change of dialysis method are considered as failure. Again transplantation, and in a second analysis, the change to home haemodialysis, was counted as loss to follow-up. We examined the influence of age at the time of starting dialysis, the presence of diabetes mellitus as a cause of ESRD or as co-morbidity, chronic cardiovascular disease (including all cardiopathies and arteriopathies with significant manifestations) at the time of
445
M. A. Gentil et al
446
CAPO -
HHD
0%
25% 25%
50% 50% % patients
0-14 y
H M 15-34 y
55-64 y
M
75% I
100%
I 35-54 y
> 65 y
Fig. 1. Per cent distribution by age groups. Comparison between CHD and CAPD patients.
First treatment modality
Glomerulonephritis Interstitial Cystic Hereditary Vascular Systemic Other Unknown
Table 2. Patients' outcome
CAPD
CHD n
(%)
n
(%)
190 161 99 16 51 52 32 222
(23) (20) (12)
38 43 11 3 14 11 3 71
(20) (22)
(2)
(6) (6) (4)
(27)
(6) (2) (7) (6) (2)
(35)
The percentages have been calculated over the total of non-diabetic patients.
A greater proportion of the CAPD patients presented with other chronic conditions at the time of entry into the dialysis programme (38% versus 30% of the centre haemodialysis patients, P = 0.047). But above all, there was a higher frequency of concomitant cardiovascular diseases: 19.9% in CAPD and 10.7% in centre haemodialysis (P = 0.0001). Cardiovascular disease was highly correlated with age distribution and varied among all the patients from 0% in the under-15 group to 28% in the over-65s. Once adjusted for the age distribution, cardiopathy remained 30% greater in the CAPD group, but without reaching statistical significance (P < 0.9).
Patients Outcome Table 2 reflects the patients' situation at the end of the study period. A greater proportion of the CAPD patients had died or had changed their treatment method, whereas the centre haemodialysis patients were transplanted in a higher proportion than the CAPD patients (14% versus 4%). The characteristics of
First treatment modality CAPD
CHD
Death Transplant Change to ' Home HD CAPD-PD CHD
Lost to follow-up Still on first treatment
n
(%)
107 117
(13) (14)
48 12
(18)
42 44 — 30
(5) (5)
(2) (2)
502
(60)
5 5 73 7 122
(3)
n
(%)
(4)
(27) (2)
(45)
'22 HHD patients and 23 CAPD patients died after the change of treatment.
The comparison of causes of death (Table 4) demonstrates a twofold risk of infectious processes in CAPD (21% versus 9%, P = 0.048), particularly of acute peritonitis (13% of the deaths in CAPD and only 1% in centre haemodiaiysis, P < 0.006). All deaths due to acute perionitis occurred in the first 2 years of the study and within the over-55 age group. Five deaths were from the same centre, whereas four centres had no deaths due to acute peritonitis in the same period. Two lethal episodes of sclerosing peritonitis were reported in patients in CAPD, one of them after changing to centre haemodialysis. There were no marked differences between the two groups with respect to other causes of mortality, including cardiovascular mortality, the leading cause of death in both groups. However, thee were almost three times more undetermined deaths in centre haemodialysis (22% versus 8%, P = 0.046). Most of these were sudden
Downloaded from http://ndt.oxfordjournals.org/ at Fudan University on May 13, 2015
Table 1. ESRD aetiology (diabetes mellitus excluded)
the patients remaining in treatment were very similar to those at the beginning in each group (Table 3). Among the deceased, older age, diabetes mellitus and cardiovascular pathology was twice as frequent as among those alive at initiation of the study; in contrast, these frequencies are lower in the patients leaving the technique. Both selections resulted in a group that remained in treatment with characteristics similar to the starting one (with a slight increment of diabetic patients in CAPD). We must take into account (Table 2) that the mainflowof patients in centre haemodialysis is elective, towards transplantation or home haemodialysis, treatments rarely used in our region for patients older than 60 or with comorbid conditions. On the other hand, in CAPD the movement of lower-risk patients was towards centre haemodialysis, as a consequence of technique dropout.
Survival in HD and CAPD
447
Table 3. Patient characteristics by outcome
Table 5. Patient survival by characteristics
Frequency of character (%)'i Treatment
>65 years Cardiovascular Diabetes
Patients («)
842
9
11
4
Age at entry (yr)** 0-15 15-34 35-54 55-64
233 107
5 21
8 26
2 8
>64 Sex
502
8
8
4
272
23
20
30
Male Female Diabetes* No Yes
Patients («)
6 months
1 year
2 years
18 222 412 311 134
100 98 98 92 85
100 96 95 78
*** 95 91 74 58
629 481
95 94
90 91
83 82
998 116
95 91
91 87
84 68
748 222 144
96 94 88
92 93 81
86 84 68
87
Comorbidity** No 102 48
12 48
15 27
15 35
122
21
21
38
Other diseases Cardiovascular
*P
