- Email: [email protected]
Death during the first 90 days of dialysis: A case control study
Death During
the First 90 Days of Dialysis: A Case Control
Study
lzhar H. Khan, MRCP, Graeme R.D. Catto, DSc, Neil Edward, FRCP, and Alison M. MacLeod, MD 0 Comparison of survival data among centers may be used to assess performance, but may be influenced by the number of patients who die during the first 90 days of renal replacement therapy (RRT). Data published by registries in Europe do not detail these deaths, and US data generally exclude them from analysis for financial reasons. To study factors influencing such deaths we compared 42 patients who died within 90 days of first commencing RRT in one Scottish renal unit (group A) between 1971 and 1992 with 42 age- and sex-matched controls who started RRT over the same period and survived longer (group B). Patients who died within 90 days of RRT ranged in age from 25.3 to 83.7 years and had a mean age of 65.2 (SEM, 1.6; 95% confidence interval, 61.9 to 68.4). The proportion of patients who died during the first 90 days of RRT increased from 2% of all patients treated before 1981 to 12% in subsequent years. Thirty-three patients in group A received emergency dialysis via temporary venous access compared with only nine in group B (P < 0.055). There were more patients in group A with a diagnosis of arteriosclerotic renal artery stenosis (14 Y 1) and with a history of smoking (15 Y 2) than in group B (P < 0.0005). Median renal or nonrenal follow-up before RRT was 1.1 month in group A and 10.6 months in group B (P < 0.0001). Fewer patients in group A had no coexisting disease (1 v 17; P < 0.0001). Group A patients had a significantly lower mean serum albumin concentration at the time of commencing RRT than group B patients (31.4 g/L v 37.1 g/L) (P = 0.0006). Blood pressure and serum concentrations of creatinine and potassium were comparable in both groups. We conclude that go-day mortality in our center was associated with the presence of coexisting disease, renal artery stenosis, smoking, hypoalbuminaemia, and presentation with advanced renal failure. Furthermore, since such heterogeneous patients are difficult to classify, valid survival data can be obtained from different centers only when such early mortality is excluded from analysis. 0 1995 by the National Kidney Foundation, Inc. INDEX
WORDS:
Mortality;
morbidity;
dialysis;
end-stage
T
HE INCREASE in numbers of patients receiving renal replacement therapy (RRT) in developed countries over the years has resulted mainly from acceptance of patients with co-morbid illness and of those with advancing age. There is evidence that a number of patients who are treated for end-stage renal disease (ESRD) do not survive beyond the first few months of starting treatment. l-3 Factors that may influence very early mortality in RRT have not yet been studied. Survival data published by the United States Renal Data System exclude deaths that occur in the first 90 days of RRT in patients under the age of 65 years: whereas data published by the registry of the European Renal Association, which relies on voluntary submission of ques-
From the Renal Unit, Aberdeen Royal Infirmary, Aberdeen; and the Department of Medicine and Therapeutics, University of Aberdeen, Aberdeen, Scotland. Received March 28, 1994; accepted in revised form September 15, 1994. Supported by a grant from the Clinical Resource and Audit Group of the Scottish Ojhce Home and Health Department. Address reprint requests to Izhar H. Khan, MRCP, Department of Medicine and Therapeutics, Polwarth Sldg Foresterhill, Aberdeen, Scotland AB9 220. 0 1995 by the National Kidney Foundation, Inc. 0272.6386/95/2502-0009$3.00/O 276
American
Journal
renal
disease;
renal
replacement
therapy.
tionnaires from renal units in Europe, do not provide information on such early deaths. In a previous study’ we showed that in a 6.5year period 12% patients who commenced RRT in two Scottish renal units died during the first 90 days of dialysis. Furthermore, although the 2year actuarial survival rate was 65%, it improved to 75% after excluding deaths during the first 90 days of RRT. We therefore studied the influence of primary renal diagnosis, nonrenal co-morbidity, rapidity of onset of ESRD, mode of presentation, and predialysis renal or nonrenal follow-up on deaths during the first 90 days of commencing RRT in a regional renal unit. PATIENTS
AND
METHODS
The Aberdeen renal unit serves Grampian region and the Orkney and Shetland Islands (a population of just over half a million). We have estimated the annual incidence of advanced renal failure, defined as a persistent serum creatinine concentration greater than 500 pmol/L, to be 134 per million of the population in Grampian (unpublished data). The renal unit has been providing RRT since 1967, and prior to 197 1 there had been no recorded deaths in the first 90 days among the 41 patients who had received RRT. From the 459 patients who commenced RRT in the unit between January 1, 1971 and January 6, 1993 (22.5 years), 46 patients who died within 90 days of first RRT (early deaths) were identified. Case notes were available for 42 patients (group A). All patients were considered to have irreversible ESRD on the basis of of Kidney
Diseases,
Vol 25, No 2 (February),
1995:
pp 276-280
90-DAY
MORTALITY
Table
1. Numbers
ON
RRT
of Patients
277 and
Proportion
of Deaths
in 90 Days Annual
of Renal
Replacement
Period
No. of Patients Commencing RRT
Acceptance Rates Per Million
No. of Patients Dying in the First 90 Days (%)
1971-1981 1981-l 991 1991-l .6.93
101 263 95
20.1 52.6 76
2 (1.9) 32 (12.1) 12 (12.6)
* P < 0.02,
between
average
ages
in all three
Since
1971
Average
Age
W (SEW
40.2 49.3 54.4
(1.33)* (1 .l !?I)* (1.8)
periods.
clinical features and investigations. Renal diagnoses were based on clinical, radiologic, and histologic features. Diabetic nephropathy was defined as progressive renal failure in the presence of longstanding diabetes mellitus, and chronic pyelonephritis was diagnosed on the basis of histologic and/or radiologic features consistent with the diagnosis. In the majority of cases of renal artery stenosis, this diagnosis was based on arteriography. These patients were compared with a control group of 42 age-and sex-matched patients who commenced RRT during the same period and survived beyond 90 days of starting RRT (group B). Each control was of similar age (difference of age not more than 1.5 years) and started RRT on the nearest date to the corresponding patient in group A. Biographic details, date of commencing RRT, renal diagnoses, date of renal referral, duration of nephrologic and nonnephrologic follow-up, and coexisting disease were recorded for cases and controls. Causes of death and autopsy findings, when available, were recorded for the study group patients. Each patient was assigned to a risk group based on a previously described protoco13: low (age less than 70 years and no co-morbid illness), medium (age 70 to 80 years, age less than 70 years with presence of diabetes, or age less than 80 years with one co-morbid illness), and high (age greater than 80 years, age 70 to 80 years with presence of diabetes, any age with two organ diseases, diabetes with one or more comorbid diseases, or any age with history of visceral malignancy). These risk groups were assigned to the patients on the basis of age and co-morbidity at the time RRT was commenced.
all patients commencing RRT increased from 2% during the 1971 to 1980 period to 12.6% since 1991 (Table 1). All subsequent results relate to the 42 patients whose files were available for examination. Biographic Details The male to female ratio was 1: 1. Patients were matched with controls for age, sex, and time of commencement of RRT. The mean age was 65.2 years (95% confidence interval, 62.0 to 68.4) for group A patients and 64.1 years (95% confidence interval, 60.3 to 67.9) for group B patients (P = 0.7). There were more smokers in group A (35.7%) than in group B (4.7%) (P < 0.0002). Twenty-two patients in group A belonged to the high-risk group compared with only five in group B (P = 0.0001). Only one patient in group A belonged to the low-risk group compared with 17 in group B (P < 0.0001). There were similar numbers in the medium-risk group: 19 in group A and 20 in group B. The diastolic and systolic blood pressures at the time of dialysis were not significantly different between the two groups. Primary Renal Diagnoses
Statistics McNemar’s test ables. Wilcoxon’s son of continuous nificance was set
Therapy
was used for comparison of discrete varipaired rank sum test was used for comparinonparametric variables. The level of sigat 5%. RESULTS
The average annual patient acceptance rate per million population increased from 20.1 during the decade from 197 1 to 1980 to 76 during the 2.5year period since 1991. The average age of patients commencing RRT during the corresponding years increased from 40.2 to 54.4 years (P < 0.0001). The proportion of early deaths in
More patients in group A had atherosclerotic renal artery stenosis as the cause of renal failure. Polycystic disease and glomerulonephritis were more frequent in the control group (Table 2). The diagnosis of bilateral renal artery stenosis or stenosis in the artery of a single functioning kidney in the 14 group A patients was based on direct arteriography or digital subtraction angiography in nine patients, on isotope renography and intravenous urography in three, and on autopsy in two. In two patients, angiotensin-converting enzyme inhibitors were implicated in deterioration of pre-existing renal failure.
KHAN
278 Table
2. Primary
Diagnosis
Renal
Table 4. Mode Stage Renal
Diagnoses
Group
A
Group
of Presentation, Disease, and
Progression to EndPredialysis Follow-up
Group
Group
B (Controls) Variable
14* 12 5 3 3 0 0 2 3
Renal artery stenosis Uncertain Myeloma Pyelonephritis Diabetic nephropathy Glomerulonephritis Polycystic kidneys Vasculitis Others * P < 0.001,
group
A v group
1* 12 2 5 3 7* 6* 1 5 B.
To exclude the possibility of the lower number of controls with renal artery stenosis being due to chance, we searched for this diagnosis as a cause of renal failure in all patients commencing RRT since 1985; we discovered only one patient (the solitary patient in group B) who survived more than 90 days following first dialysis. Serum Biochemistry
There were no significant differences between the serum concentrations of creatinine, potassium, or bicarbonate at the time of first dialysis in either group (Table 3). The mean serum albumin (available for 38 patients in group A) was significantly lower in group A (3 1.4; 95% confidence interval, 29.5 to 33.3) than in controls (37.1; 95% confidence interval, 34.8 to 39.4) (P = .oooq. Mode of Presentation and Predialysis Follow-up
Thirty-three patients in group A required emergency dialysis via temporary vascular ac-
Table
3. Serum Parameters
Group A (SW
Variables
Potassium (mmol/L) Bicarbonate (mmol/L) Creatinine bmol/L) Albumin (g/L) * Wilcoxon’s
Concentrations at the Start
of Biochemical of Dialysis Group 6 (SW
Probability V&k?*
4.9 (0.15)
4.6 (0.13)
0.08
16.0 (0.79) 896 (55.6) (10.01 mg/dL) 31.4 (0.9)
14.8 (0.85) 988 (53.2) (11.2 mg/dL) 37.1 (1.1)
0.15
paired
rank sum test.
ET AL
0.25 0.0006
Median renal follow-up (mo) Median nonrenal follow-up (mo) Median time from first creatinine >300 to first creatinine >500 pmol/L (mo) Median time from first creatinine >500 to RRT 0-W No. of emergency dialyses
A (n = 42)
B (n = 42)
1 .12 (6.3-35.5)*
10.6 (15.9-38.9)
0 (11.7-45.4)*
12.5
(28.4-75.4)
0 (1.05-10.2)t
1.43
0.6 (2.4-14.5)’
6.0 (8.45-21.9)
33*
(4.5-l
3.4)?
g*
NOTE. Numbers in parentheses indicate 95% dence interval. * P < 0.0001, Wilcoxon paired rank test. t P = 0.053, Wilcoxon paired rank sum test. $ P = 0.055 McNemar’s test.
confi-
cess (subclavian catheter) compared with only nine in group B (P = 0.055). In group A, 13 patients had no follow-up by a hospital clinic (nephrologic or nonnephrologic), whereas in group B there were nine such patients (P = 0.236, group A v group B.) The median nephrologic and nonnephrologic predialysis follow-up period and interval from the first serum creatinine of 300 and 500 pmol/ L to dialysis were greater in group A (Table 4). Causes of Death
Cardiovascular problems were the most common cause of death (Table 5). Autopsies were carried out in 18 (43%) cases. Table
5. Causes
Cause
Myocardial infarction Septicaemia Uncertain Pulmonary infection Cardiac failure Cardiac arrest Others
of Death No. of Patients
11 11 10 3 3 2 2
90-DAY
MORTALITY
ON
RRT
DISCUSSION
This study showed that the proportion of patients who died within the first 90 days of RRT increased from 2% of all patients who commenced RRT in the decade between 1971 and 1981 to 12% thereafter. During this period the annual acceptance rate for RRT increased from 20.1 to 76 per million population. The average age of patients commencing RRT increased from 40.2 years in the decade between 1971 and 1981 to 54.4 since 1991. Furthermore, the vast majority of such early deaths were observed in patients who presented with advanced renal failure, had a history of smoking, and had not been followed by a nephrologist or nonnephrologist prior to dialysis. Emergency dialysis had been given to the majority of the patients who died early compared with controls. Therefore, the increase in the proportion of early deaths appears to be a feature associated with the increased availability and more liberal acceptance of patients for RRT in recent years. A striking feature was the majority of patients in the early death group who had a diagnosis of renal artery stenosis. There have been few published reports specifically addressing the outcome of patients with renal artery stenosis as a cause of ESRD. A recent retrospective review of surgical repair of renovascular disease from North America, however, showed that patients who had severe renal insufficiency prior to reconstructive surgery or those who progressed to dependence on dialysis had a relatively rapid rate of death during follow-up compared with those who had mild or no renal insufficiency.5 An earlier study from North America showed a median 5-year survival rate of 12% for patients with uncorrected atherosclerotic renal vascular disease who had progressed to ESRD.‘j As we could record only one case of renal artery stenosis who survived longer than 90 days on RRT, it is unlikely that the high number of patients in the early death group with this diagnosis was due to chance. Most patients with renal arterial disease suffer from widespread arteriosclerosis, which is associated with poor survival. Few studies have examined early mortality on RRT. In 1981, an uncontrolled study examining the deaths of patients aged less than 50 years in three British regions found that 8.5% patients
died within the first 90 days of dialysis.7 Most patients had significant nonrenal disease. At the time when the acceptance rate for RRT in the United Kingdom was 22 patients per million per year, the study was criticized for suggesting that RRT in such patients was perhaps not suitable in view of their high mortality.’ It used to be, and perhaps still is, a widely held belief in the United States that in Britain limitations of resources led nephrologists to impose an upper age limit for RRT.9 Our patients ranged in age from 23 to 83 years, and advancing age currently is not considered a contraindication to RRT in our center. With the increasing numbers of patients of advanced age and with co-morbid illness’O~” accepted for RRT, there is once more the need to address the problem of early mortality on RRT. A recent study from Nottingham examined deaths in the first year of dialysis and found that 44 of 253 patients (17.5%) who commenced RRT between January 1983 and January 1988 died during the first year of dialysis.’ Twelve percent of the patients (a proportion similar to our study) died in the first 90 days of RRT. These investigators suggested that late referral of patients to a renal unit was associated with death in the first year of RRT. Information as to whether these patients were followed by nonrenal physicians was not given; neither was the influence of coexisting nonrenal disease on early mortality considered. The US Renal Data System reports data on 90-day mortality for patients over the age of 65 years, as data concerning younger age groups is relatively incomplete prior to day 91 of commencing RRT. Data from the 199 1 US Renal Data System report showed a 90-day mortality rate of 11% for patients over 65 years of age. Our findings suggest that the majority of early deaths occurred in patients who presented with advanced renal failure for the first time to any discipline and who had little predialysis care (by nephrologists or nonnephrologists). Identification of early symptoms of renal impairment may lead to early referral, measures to retard the onset of ESRD, and, possibly, preparation for elective RRT, which may improve survival. Nevertheless, some patients present for the first time with advanced renal failure; these patients seem to have a worse prognosis than those who present earlier. Using a risk stratification method we have been able to study the influence of co-morbidity
280
on early survival on RRT. Only one patient in group A belonged to the low-risk group (under 70 years of age and with no coexisting disease) compared with 17 in the control group; 22 patients in group A were considered to be “high risk,” with only five in the control group. Thus, co-morbidity, which has been shown to influence long-term survival, also influenced survival during the first 90 days of dialysis. It has been suggested that a low serum albumin concentration is an independent predictor of mortality in RRT. “Jo Patients who died early had a significantly low serum albumin compared with controls at commencement of dialysis. Whether the low serum albumin predicted early mortality in the group of patients who died early or was simply a marker of preterminal illness in these patients is speculative. However, the degree of hypoalbuminaemia in patients who died early was striking, and its possible role as a predictor of mortality needs to be studied prospectively. Since patients who die in the first few weeks of commencing dialysis may influence overall survival data, valid comparisons of survival among different centers can be obtained if early mortality is excluded from analysis. Our study and the previous report from Nottingham suggest that a significant proportion (12%) of modemday dialysis patients fail to survive longer than 90 days. The European Renal Association does not publish data on such deaths, and data collected by the US Renal Data System allows such deaths to be studied only in those over the age of 65 years. We suggest that a larger prospective study should be carried out to further examine the factors which influence early mortality in RRT patients.
KHAN
ET AL
ACKNOWLEDGMENT The
authors
thank
CA.
Ritchie
for technical
assistance.
REFERENCES 1. Khan IH, Catto GRD, Edward N, Fleming LW, Henderson IS, MacLeod AM: Influence of coexisting disease on survival in renal replacement therapy. Lancet 34 I:41 5-418, 1993 2. Innes A, Rowe PA, Burden RP, Morgan AG: Early deaths on renal replacement therapy: The need for early nephrological referral. Nephrol Dial Transplant 6:467-671, 1992 3. Wright LF: Survival in patients with end-stage renal disease. Am J Kidney Dis 17:25-28, 1991 4. USRDS Annual Data Report, 1991 5. Hansen KJ, Starr SM, Sands RE, Burkart JM, Plonk GW, Dean RH: Contemporary surgical management of renovascular disease. J Vast Surg 16:319-331, 1992 6. Mailloux LU, Bellucci AG, Mossey RT, Lesser M, Napolitano B, Moore T, Wilkes BM, Bluestone PD: Predictors of survival in patients undergoing dialysis. Am J Med 84:855862, 1988 7. Clarke C: Deaths from chronic renal failure under the age of 50. BMJ 283:283-286, 1981 8. Audit in renal failure: The wrong target. BMJ 283:261262, 198 1 (editorial) 9. Berlyne GM: Over 50 and uremic = death: The failure of the British National Health Service to provide adequate dialysis facilities. Nephron 3 1: l89- 190, 1982 10. Collins AJ, Hanson G, Umen A, Kjellstrand C, Keshaviah P: Changing risk factor demographics in end-stage renal disease patients entering hemodialysis and the impact on long-term mortality. Am J Kidney Dis 15:422-432, 1990 11. Hylander B, Lundblad H, Kjellstrand CM: Changing patient characteristics in chronic hemodialysis. Stand J Urol Nephrol 25:59-63, 1991 12. Lowrie EG, Lew NL: Death risk in haemodialysis patients: The predictive value of commonly measured variables and an evaluation of death rate difference between facilities. Am J Kidney Dis 15:458-482, 1990 13. Gambia G, Mejia JL, Saldivar S, Correa-Rotter R, Pena JC: Death risk in CAPD patients: The predictive value of initial clinical and laboratory variables. Nephron 65:2327, 1993
the First 90 Days of Dialysis: A Case Control
Study
lzhar H. Khan, MRCP, Graeme R.D. Catto, DSc, Neil Edward, FRCP, and Alison M. MacLeod, MD 0 Comparison of survival data among centers may be used to assess performance, but may be influenced by the number of patients who die during the first 90 days of renal replacement therapy (RRT). Data published by registries in Europe do not detail these deaths, and US data generally exclude them from analysis for financial reasons. To study factors influencing such deaths we compared 42 patients who died within 90 days of first commencing RRT in one Scottish renal unit (group A) between 1971 and 1992 with 42 age- and sex-matched controls who started RRT over the same period and survived longer (group B). Patients who died within 90 days of RRT ranged in age from 25.3 to 83.7 years and had a mean age of 65.2 (SEM, 1.6; 95% confidence interval, 61.9 to 68.4). The proportion of patients who died during the first 90 days of RRT increased from 2% of all patients treated before 1981 to 12% in subsequent years. Thirty-three patients in group A received emergency dialysis via temporary venous access compared with only nine in group B (P < 0.055). There were more patients in group A with a diagnosis of arteriosclerotic renal artery stenosis (14 Y 1) and with a history of smoking (15 Y 2) than in group B (P < 0.0005). Median renal or nonrenal follow-up before RRT was 1.1 month in group A and 10.6 months in group B (P < 0.0001). Fewer patients in group A had no coexisting disease (1 v 17; P < 0.0001). Group A patients had a significantly lower mean serum albumin concentration at the time of commencing RRT than group B patients (31.4 g/L v 37.1 g/L) (P = 0.0006). Blood pressure and serum concentrations of creatinine and potassium were comparable in both groups. We conclude that go-day mortality in our center was associated with the presence of coexisting disease, renal artery stenosis, smoking, hypoalbuminaemia, and presentation with advanced renal failure. Furthermore, since such heterogeneous patients are difficult to classify, valid survival data can be obtained from different centers only when such early mortality is excluded from analysis. 0 1995 by the National Kidney Foundation, Inc. INDEX
WORDS:
Mortality;
morbidity;
dialysis;
end-stage
T
HE INCREASE in numbers of patients receiving renal replacement therapy (RRT) in developed countries over the years has resulted mainly from acceptance of patients with co-morbid illness and of those with advancing age. There is evidence that a number of patients who are treated for end-stage renal disease (ESRD) do not survive beyond the first few months of starting treatment. l-3 Factors that may influence very early mortality in RRT have not yet been studied. Survival data published by the United States Renal Data System exclude deaths that occur in the first 90 days of RRT in patients under the age of 65 years: whereas data published by the registry of the European Renal Association, which relies on voluntary submission of ques-
From the Renal Unit, Aberdeen Royal Infirmary, Aberdeen; and the Department of Medicine and Therapeutics, University of Aberdeen, Aberdeen, Scotland. Received March 28, 1994; accepted in revised form September 15, 1994. Supported by a grant from the Clinical Resource and Audit Group of the Scottish Ojhce Home and Health Department. Address reprint requests to Izhar H. Khan, MRCP, Department of Medicine and Therapeutics, Polwarth Sldg Foresterhill, Aberdeen, Scotland AB9 220. 0 1995 by the National Kidney Foundation, Inc. 0272.6386/95/2502-0009$3.00/O 276
American
Journal
renal
disease;
renal
replacement
therapy.
tionnaires from renal units in Europe, do not provide information on such early deaths. In a previous study’ we showed that in a 6.5year period 12% patients who commenced RRT in two Scottish renal units died during the first 90 days of dialysis. Furthermore, although the 2year actuarial survival rate was 65%, it improved to 75% after excluding deaths during the first 90 days of RRT. We therefore studied the influence of primary renal diagnosis, nonrenal co-morbidity, rapidity of onset of ESRD, mode of presentation, and predialysis renal or nonrenal follow-up on deaths during the first 90 days of commencing RRT in a regional renal unit. PATIENTS
AND
METHODS
The Aberdeen renal unit serves Grampian region and the Orkney and Shetland Islands (a population of just over half a million). We have estimated the annual incidence of advanced renal failure, defined as a persistent serum creatinine concentration greater than 500 pmol/L, to be 134 per million of the population in Grampian (unpublished data). The renal unit has been providing RRT since 1967, and prior to 197 1 there had been no recorded deaths in the first 90 days among the 41 patients who had received RRT. From the 459 patients who commenced RRT in the unit between January 1, 1971 and January 6, 1993 (22.5 years), 46 patients who died within 90 days of first RRT (early deaths) were identified. Case notes were available for 42 patients (group A). All patients were considered to have irreversible ESRD on the basis of of Kidney
Diseases,
Vol 25, No 2 (February),
1995:
pp 276-280
90-DAY
MORTALITY
Table
1. Numbers
ON
RRT
of Patients
277 and
Proportion
of Deaths
in 90 Days Annual
of Renal
Replacement
Period
No. of Patients Commencing RRT
Acceptance Rates Per Million
No. of Patients Dying in the First 90 Days (%)
1971-1981 1981-l 991 1991-l .6.93
101 263 95
20.1 52.6 76
2 (1.9) 32 (12.1) 12 (12.6)
* P < 0.02,
between
average
ages
in all three
Since
1971
Average
Age
W (SEW
40.2 49.3 54.4
(1.33)* (1 .l !?I)* (1.8)
periods.
clinical features and investigations. Renal diagnoses were based on clinical, radiologic, and histologic features. Diabetic nephropathy was defined as progressive renal failure in the presence of longstanding diabetes mellitus, and chronic pyelonephritis was diagnosed on the basis of histologic and/or radiologic features consistent with the diagnosis. In the majority of cases of renal artery stenosis, this diagnosis was based on arteriography. These patients were compared with a control group of 42 age-and sex-matched patients who commenced RRT during the same period and survived beyond 90 days of starting RRT (group B). Each control was of similar age (difference of age not more than 1.5 years) and started RRT on the nearest date to the corresponding patient in group A. Biographic details, date of commencing RRT, renal diagnoses, date of renal referral, duration of nephrologic and nonnephrologic follow-up, and coexisting disease were recorded for cases and controls. Causes of death and autopsy findings, when available, were recorded for the study group patients. Each patient was assigned to a risk group based on a previously described protoco13: low (age less than 70 years and no co-morbid illness), medium (age 70 to 80 years, age less than 70 years with presence of diabetes, or age less than 80 years with one co-morbid illness), and high (age greater than 80 years, age 70 to 80 years with presence of diabetes, any age with two organ diseases, diabetes with one or more comorbid diseases, or any age with history of visceral malignancy). These risk groups were assigned to the patients on the basis of age and co-morbidity at the time RRT was commenced.
all patients commencing RRT increased from 2% during the 1971 to 1980 period to 12.6% since 1991 (Table 1). All subsequent results relate to the 42 patients whose files were available for examination. Biographic Details The male to female ratio was 1: 1. Patients were matched with controls for age, sex, and time of commencement of RRT. The mean age was 65.2 years (95% confidence interval, 62.0 to 68.4) for group A patients and 64.1 years (95% confidence interval, 60.3 to 67.9) for group B patients (P = 0.7). There were more smokers in group A (35.7%) than in group B (4.7%) (P < 0.0002). Twenty-two patients in group A belonged to the high-risk group compared with only five in group B (P = 0.0001). Only one patient in group A belonged to the low-risk group compared with 17 in group B (P < 0.0001). There were similar numbers in the medium-risk group: 19 in group A and 20 in group B. The diastolic and systolic blood pressures at the time of dialysis were not significantly different between the two groups. Primary Renal Diagnoses
Statistics McNemar’s test ables. Wilcoxon’s son of continuous nificance was set
Therapy
was used for comparison of discrete varipaired rank sum test was used for comparinonparametric variables. The level of sigat 5%. RESULTS
The average annual patient acceptance rate per million population increased from 20.1 during the decade from 197 1 to 1980 to 76 during the 2.5year period since 1991. The average age of patients commencing RRT during the corresponding years increased from 40.2 to 54.4 years (P < 0.0001). The proportion of early deaths in
More patients in group A had atherosclerotic renal artery stenosis as the cause of renal failure. Polycystic disease and glomerulonephritis were more frequent in the control group (Table 2). The diagnosis of bilateral renal artery stenosis or stenosis in the artery of a single functioning kidney in the 14 group A patients was based on direct arteriography or digital subtraction angiography in nine patients, on isotope renography and intravenous urography in three, and on autopsy in two. In two patients, angiotensin-converting enzyme inhibitors were implicated in deterioration of pre-existing renal failure.
KHAN
278 Table
2. Primary
Diagnosis
Renal
Table 4. Mode Stage Renal
Diagnoses
Group
A
Group
of Presentation, Disease, and
Progression to EndPredialysis Follow-up
Group
Group
B (Controls) Variable
14* 12 5 3 3 0 0 2 3
Renal artery stenosis Uncertain Myeloma Pyelonephritis Diabetic nephropathy Glomerulonephritis Polycystic kidneys Vasculitis Others * P < 0.001,
group
A v group
1* 12 2 5 3 7* 6* 1 5 B.
To exclude the possibility of the lower number of controls with renal artery stenosis being due to chance, we searched for this diagnosis as a cause of renal failure in all patients commencing RRT since 1985; we discovered only one patient (the solitary patient in group B) who survived more than 90 days following first dialysis. Serum Biochemistry
There were no significant differences between the serum concentrations of creatinine, potassium, or bicarbonate at the time of first dialysis in either group (Table 3). The mean serum albumin (available for 38 patients in group A) was significantly lower in group A (3 1.4; 95% confidence interval, 29.5 to 33.3) than in controls (37.1; 95% confidence interval, 34.8 to 39.4) (P = .oooq. Mode of Presentation and Predialysis Follow-up
Thirty-three patients in group A required emergency dialysis via temporary vascular ac-
Table
3. Serum Parameters
Group A (SW
Variables
Potassium (mmol/L) Bicarbonate (mmol/L) Creatinine bmol/L) Albumin (g/L) * Wilcoxon’s
Concentrations at the Start
of Biochemical of Dialysis Group 6 (SW
Probability V&k?*
4.9 (0.15)
4.6 (0.13)
0.08
16.0 (0.79) 896 (55.6) (10.01 mg/dL) 31.4 (0.9)
14.8 (0.85) 988 (53.2) (11.2 mg/dL) 37.1 (1.1)
0.15
paired
rank sum test.
ET AL
0.25 0.0006
Median renal follow-up (mo) Median nonrenal follow-up (mo) Median time from first creatinine >300 to first creatinine >500 pmol/L (mo) Median time from first creatinine >500 to RRT 0-W No. of emergency dialyses
A (n = 42)
B (n = 42)
1 .12 (6.3-35.5)*
10.6 (15.9-38.9)
0 (11.7-45.4)*
12.5
(28.4-75.4)
0 (1.05-10.2)t
1.43
0.6 (2.4-14.5)’
6.0 (8.45-21.9)
33*
(4.5-l
3.4)?
g*
NOTE. Numbers in parentheses indicate 95% dence interval. * P < 0.0001, Wilcoxon paired rank test. t P = 0.053, Wilcoxon paired rank sum test. $ P = 0.055 McNemar’s test.
confi-
cess (subclavian catheter) compared with only nine in group B (P = 0.055). In group A, 13 patients had no follow-up by a hospital clinic (nephrologic or nonnephrologic), whereas in group B there were nine such patients (P = 0.236, group A v group B.) The median nephrologic and nonnephrologic predialysis follow-up period and interval from the first serum creatinine of 300 and 500 pmol/ L to dialysis were greater in group A (Table 4). Causes of Death
Cardiovascular problems were the most common cause of death (Table 5). Autopsies were carried out in 18 (43%) cases. Table
5. Causes
Cause
Myocardial infarction Septicaemia Uncertain Pulmonary infection Cardiac failure Cardiac arrest Others
of Death No. of Patients
11 11 10 3 3 2 2
90-DAY
MORTALITY
ON
RRT
DISCUSSION
This study showed that the proportion of patients who died within the first 90 days of RRT increased from 2% of all patients who commenced RRT in the decade between 1971 and 1981 to 12% thereafter. During this period the annual acceptance rate for RRT increased from 20.1 to 76 per million population. The average age of patients commencing RRT increased from 40.2 years in the decade between 1971 and 1981 to 54.4 since 1991. Furthermore, the vast majority of such early deaths were observed in patients who presented with advanced renal failure, had a history of smoking, and had not been followed by a nephrologist or nonnephrologist prior to dialysis. Emergency dialysis had been given to the majority of the patients who died early compared with controls. Therefore, the increase in the proportion of early deaths appears to be a feature associated with the increased availability and more liberal acceptance of patients for RRT in recent years. A striking feature was the majority of patients in the early death group who had a diagnosis of renal artery stenosis. There have been few published reports specifically addressing the outcome of patients with renal artery stenosis as a cause of ESRD. A recent retrospective review of surgical repair of renovascular disease from North America, however, showed that patients who had severe renal insufficiency prior to reconstructive surgery or those who progressed to dependence on dialysis had a relatively rapid rate of death during follow-up compared with those who had mild or no renal insufficiency.5 An earlier study from North America showed a median 5-year survival rate of 12% for patients with uncorrected atherosclerotic renal vascular disease who had progressed to ESRD.‘j As we could record only one case of renal artery stenosis who survived longer than 90 days on RRT, it is unlikely that the high number of patients in the early death group with this diagnosis was due to chance. Most patients with renal arterial disease suffer from widespread arteriosclerosis, which is associated with poor survival. Few studies have examined early mortality on RRT. In 1981, an uncontrolled study examining the deaths of patients aged less than 50 years in three British regions found that 8.5% patients
died within the first 90 days of dialysis.7 Most patients had significant nonrenal disease. At the time when the acceptance rate for RRT in the United Kingdom was 22 patients per million per year, the study was criticized for suggesting that RRT in such patients was perhaps not suitable in view of their high mortality.’ It used to be, and perhaps still is, a widely held belief in the United States that in Britain limitations of resources led nephrologists to impose an upper age limit for RRT.9 Our patients ranged in age from 23 to 83 years, and advancing age currently is not considered a contraindication to RRT in our center. With the increasing numbers of patients of advanced age and with co-morbid illness’O~” accepted for RRT, there is once more the need to address the problem of early mortality on RRT. A recent study from Nottingham examined deaths in the first year of dialysis and found that 44 of 253 patients (17.5%) who commenced RRT between January 1983 and January 1988 died during the first year of dialysis.’ Twelve percent of the patients (a proportion similar to our study) died in the first 90 days of RRT. These investigators suggested that late referral of patients to a renal unit was associated with death in the first year of RRT. Information as to whether these patients were followed by nonrenal physicians was not given; neither was the influence of coexisting nonrenal disease on early mortality considered. The US Renal Data System reports data on 90-day mortality for patients over the age of 65 years, as data concerning younger age groups is relatively incomplete prior to day 91 of commencing RRT. Data from the 199 1 US Renal Data System report showed a 90-day mortality rate of 11% for patients over 65 years of age. Our findings suggest that the majority of early deaths occurred in patients who presented with advanced renal failure for the first time to any discipline and who had little predialysis care (by nephrologists or nonnephrologists). Identification of early symptoms of renal impairment may lead to early referral, measures to retard the onset of ESRD, and, possibly, preparation for elective RRT, which may improve survival. Nevertheless, some patients present for the first time with advanced renal failure; these patients seem to have a worse prognosis than those who present earlier. Using a risk stratification method we have been able to study the influence of co-morbidity
280
on early survival on RRT. Only one patient in group A belonged to the low-risk group (under 70 years of age and with no coexisting disease) compared with 17 in the control group; 22 patients in group A were considered to be “high risk,” with only five in the control group. Thus, co-morbidity, which has been shown to influence long-term survival, also influenced survival during the first 90 days of dialysis. It has been suggested that a low serum albumin concentration is an independent predictor of mortality in RRT. “Jo Patients who died early had a significantly low serum albumin compared with controls at commencement of dialysis. Whether the low serum albumin predicted early mortality in the group of patients who died early or was simply a marker of preterminal illness in these patients is speculative. However, the degree of hypoalbuminaemia in patients who died early was striking, and its possible role as a predictor of mortality needs to be studied prospectively. Since patients who die in the first few weeks of commencing dialysis may influence overall survival data, valid comparisons of survival among different centers can be obtained if early mortality is excluded from analysis. Our study and the previous report from Nottingham suggest that a significant proportion (12%) of modemday dialysis patients fail to survive longer than 90 days. The European Renal Association does not publish data on such deaths, and data collected by the US Renal Data System allows such deaths to be studied only in those over the age of 65 years. We suggest that a larger prospective study should be carried out to further examine the factors which influence early mortality in RRT patients.
KHAN
ET AL
ACKNOWLEDGMENT The
authors
thank
CA.
Ritchie
for technical
assistance.
REFERENCES 1. Khan IH, Catto GRD, Edward N, Fleming LW, Henderson IS, MacLeod AM: Influence of coexisting disease on survival in renal replacement therapy. Lancet 34 I:41 5-418, 1993 2. Innes A, Rowe PA, Burden RP, Morgan AG: Early deaths on renal replacement therapy: The need for early nephrological referral. Nephrol Dial Transplant 6:467-671, 1992 3. Wright LF: Survival in patients with end-stage renal disease. Am J Kidney Dis 17:25-28, 1991 4. USRDS Annual Data Report, 1991 5. Hansen KJ, Starr SM, Sands RE, Burkart JM, Plonk GW, Dean RH: Contemporary surgical management of renovascular disease. J Vast Surg 16:319-331, 1992 6. Mailloux LU, Bellucci AG, Mossey RT, Lesser M, Napolitano B, Moore T, Wilkes BM, Bluestone PD: Predictors of survival in patients undergoing dialysis. Am J Med 84:855862, 1988 7. Clarke C: Deaths from chronic renal failure under the age of 50. BMJ 283:283-286, 1981 8. Audit in renal failure: The wrong target. BMJ 283:261262, 198 1 (editorial) 9. Berlyne GM: Over 50 and uremic = death: The failure of the British National Health Service to provide adequate dialysis facilities. Nephron 3 1: l89- 190, 1982 10. Collins AJ, Hanson G, Umen A, Kjellstrand C, Keshaviah P: Changing risk factor demographics in end-stage renal disease patients entering hemodialysis and the impact on long-term mortality. Am J Kidney Dis 15:422-432, 1990 11. Hylander B, Lundblad H, Kjellstrand CM: Changing patient characteristics in chronic hemodialysis. Stand J Urol Nephrol 25:59-63, 1991 12. Lowrie EG, Lew NL: Death risk in haemodialysis patients: The predictive value of commonly measured variables and an evaluation of death rate difference between facilities. Am J Kidney Dis 15:458-482, 1990 13. Gambia G, Mejia JL, Saldivar S, Correa-Rotter R, Pena JC: Death risk in CAPD patients: The predictive value of initial clinical and laboratory variables. Nephron 65:2327, 1993