Q J Med 2002; 95: 585-590
© 2002 Association of Physicians
High incidence of renal failure requiring short-term dialysis: a prospective observational study
From the Renal Unit, and 1 Department of Audit and Clinical Effectiveness, Dumfries and Galloway Royal Infirmary, Dumfries, and 2 Robertson Centre for Biostatistics, University of Glasgow, Glasgow, UK
Received 12 February 2002 and in revised form 10 May 2002
 |
Summary |
|---|
 Top Summary IntroductionMethodsResultsDiscussionReferences |
|---|
Background: Previous estimates of incidence of acute renal failure (ARF) requiring renal replacement therapy have varied from 18 to 75 patients per million per year, but have been beset by problems of definition.
Aim: To investigate whether the ‘90-day rule’ provides a more reliable, reproducible and robust estimate of the need for short-term dialysis.
Setting: District general hospital serving a population of 147 000.
Design: Prospective observational study.
Methods: Patients who received renal replacement therapy in Dumfries and Galloway between 01/01/94 and 31/12/2000 were divided into two groups: long-term dialysis (
90 days) and short-term dialysis (<90 days).
Results: Of 302 patients, 193 received short-term dialysis, giving an incidence for short-term dialysis of 187 patient episodes per million per year (95%CI 170–203). Use of a more conventional definition for ARF (including all ARF and acute-on-chronic renal failure, but excluding patients with chronic renal failure who present acutely) produced a similar estimate at 176 patients per million per year (95%CI 160–193).
Discussion: The 90-day rule estimated the incidence of short-term dialysis/ARF at nearly twice the incidence of chronic renal failure requiring dialysis, and more than twice the most recent estimate of the incidence of ARF in the UK. The main attraction of the 90-day rule is its simplicity. If the high level of short-term dialysis/ARF found in our study is confirmed by other centres, there are significant resource implications for the provision of renal care.
|
Introduction |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
The rate of acceptance of chronic renal failure on to a dialysis programme in the UK is around 100 per million population (pmp) per year.1,2 Much less is known about the incidence of acute renal failure (ARF). Previous estimates have varied from 18 pmp/year requiring dialysis in Exeter3 to 50 pmp/year requiring dialysis in Aberdeen4 and 83 pmp/year requiring dialysis in Kent.5 These are likely to be underestimates, as it is probable that only a minorityof those with ARF were referred to nephrologists in these studies.
Classification of renal failure is facilitated by the availability of a pre-morbid serum creatinine concentration. When, as is often the case, no recent measurements of serum creatinine are available, the distinction between ARF, acute-on-chronic renal failure and chronic renal failure can be difficult. Patients with chronic renal failure are often unknown to medical services until their renal failure becomes advanced and uraemic symptoms arise, or they may develop an intercurrent illness such that their already compromised renal function deteriorates rapidly. These patients then present ‘acutely’ for the first time in advanced renal failure requiring urgent dialysis.
In an attempt to provide a more reliable, reproducible and robust classification of ARF requiring short-term dialysis, we have prospectively studied 302 consecutive patients who received renal replacement therapy in Dumfries and Galloway from the beginning of 1994 through to the end of 2000. We classified patients into one of two treatment groups: those who dialysed for <90 days (short-term dialysis) and those who dialysed for 90 days (long-term dialysis). Our objectives were first, to estimate the incidence of patients requiring short-term dialysis and second, to compare this with the number of patients requiring treatment for ARF defined by conventional diagnostic criteria.
|
Methods |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Details of all consecutive patients requiring dialysis in Dumfries and Galloway were recorded prospectively between 01/01/1994 and 31/12/2000. The indications for dialysis were the same in 1994, as they were in 2000, and consisted of one or more of the following: uraemia, fluid overload, hyperkalaemia, acidosis or pericarditis. Patients with obstructive uropathy, myeloma and malignancy who required dialysis were included. Patients who developed renal failure after cardiothoracic surgery do not appear in our analyses, because there is no cardiothoracic unit in Dumfries. The following information was then extracted for analysis: gender, age at first dialysis, urea and creatinine at first dialysis, cause of renal failure if known, need for ventilation, number of treatment sessions if <90 days, renal outcome, survival status at 31/03/2001 with date of death if relevant. The cut point used to define short- and long-term dialysis was a treatment time of 90 days. This was chosen because ‘the 90 day rule’ is already used in the US as the basis for determining which patients with chronic renal failure are eligible for reimbursement of costs. The conventional criteria used for classifying ARF, acute-on-chronic renal failure and chronic renal failure are shown in Table 1
. We defined an episode of treatment as the period of dialysis dependence between first dialysis and renal recovery or death.
|
Statistical methods
The incidence rates (pmp/year) and associated 95%CIs were calculated assuming a constant population in Dumfries and Galloway of 147 000 over the 7-year study period. The 95%CIs for the proportions having ARF by different definitions were calculated assuming binomial variability. All analyses were performed using SAS 8.2 for Windows NT4.
|
Results |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Overall, 302 new patients (187 males, 59%) required 315 patient episodes of treatment during the observation period (Figure 1
). Their mean age was 63.6 years (range 21–89). An additional 42 patients already on dialysis who were transferred temporarily (n=31) or permanently (n=11) from peritoneal dialysis to haemodialysis were excluded from the analysis, as were patients referred from other centres already on dialysis (n=7), who received plasma exchange only (n=9) or who were treated for lithium intoxication (n=1) or hypothermia (n=1). Of these new patients, 109 received dialysis for 90 days. Ten of these had experienced one or more episodes of short-term dialysis followed by temporary renal recovery before requiring long-term dialysis: for the purpose of analysis these patients have been assigned to the >90 days group. One patient who dialysed for 90 days recovered sufficient renal function to discontinue renal replacement therapy after 106 days: the underlying diagnosis was polyarteritis nodosa, which eventually responded to immunosuppressive therapy. The other 108 patients continued to require renal replacement therapy giving an incidence for chronic renal failure of 105pmp/year (95%CI 89–122).
|
Overall, 193 patients required 195 patient episodes of treatment lasting <90 days. This was a heterogeneous group, comprising patients with previously normal renal function who had ARF and recovered renal function (n=54) or who died without recovering renal function (n=65); patients with acute-on-chronic renal failure who recovered renal function (n=25) or who died without recovering renal function (n=38), and patients with chronic renal failure who died within 90 days of starting renal replacement therapy (n=11). Sixty-six (34%) of the patients who underwent short-term dialysis also required ventilation for respiratory failure.
Comparison with other methods of estimating incidence
An analysis of the incidence of renal failure requiring dialysis, obtained by applying different diagnostic criteria, is shown in Table 2. The lowest estimate of the need for dialysis is obtained if the number of cases is limited to those with ARF (see Table 1
for definition) and serum creatinine >500 µmol/l. Incidence increases progressively as the criteria are ‘relaxed’ and peaks at 187 pmp/year if all cases requiring dialysis for <90 days, irrespective of serum creatinine concentration, are included. Use of conventional criteria for defining ARF, which also includes those with acute-on-chronic renal failure but excludes patients with chronic renal failure presenting acutely, leads to an estimate of incidence that is only slightly lower, at 176 pmp/year.
|
Severity of renal failure and number of dialysis treatments
Median levels of blood urea and serum creatinine immediately before dialysis in those undergoing short-term dialysis (<90 days) were 34 mmol/l and 430 µmol/l, respectively. Corresponding median values of urea and creatinine for patients who were dialysed for 90 days were 34 mmol/l and 638 µmol/l. Seventy-nine (41%) of those treated for <90 days recovered sufficient renal function to become independent of dialysis. Figure 2 shows the number of dialysis treatments required by patients in this group. The median number of treatments was six (range 1–47) in those who recovered renal function, and five (range 1–39) in those who did not. Of the 79 who recovered renal function, 11 did so after one dialysis. If these patients are excluded from the analysis, on the basis that they might have recovered without treatment, 182 patients still required short-term dialysis in this survey. This gives an incidence of 176 pmp/year (95%CI 160–193), a figure which is still considerably in excess of that recorded in earlier studies.
|
|
Discussion |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
Previous studies of the incidence of ARF requiring dialysis have suggested that it is considerably less common than we have found. In a widely quoted paper, Feest and colleagues reported their experience of severe ARF in Devon between 1986 and 1988.3 Acute renal failure in this study was defined as a serum creatinine concentration >500 µmol/l, with recovery or death during the acute illness. Patients with malignancy were excluded. There was no cardiac surgery unit in the districts of Devon covered by the study. Overall, 125 adults (140 pmp/year) had ARF as defined, and 16 patients (18 pmp/year) received acute dialysis.
Khan4 studied the incidence of ARF in Grampian, Scotland between 1989 and 1990. The definition of ARF used in this study was serum creatinine concentration >500 µmol/l with recovery or death during the acute illness. Patients with malignancy were included, as were those who had undergone cardiac surgery. Fifty-one patients (100 pmp/year) had ARF as defined, and 25 patients (50 pmp/year) received dialysis. These figures must underestimate the true need for dialysis, as only a minority of patients in Grampian with ARF were referred to a nephrologist.4
Liano6 reported the incidence of ARF in adult patients admitted by any of the 13 tertiary care hospitals in Madrid during a nine-month period in 1991–2. ARF was defined as a sudden rise in serum creatinine of >177 µmol/l in subjects with previously normal renal function; elevated serum creatinine on admission with at least 50% recovery before discharge; elevated serum creatinine on admission with no suspicion of chronic renal failure and normal or increased renal size on ultrasound; and subjects with baseline serum creatinine <264 µmol/l who experienced a sudden rise in serum creatinine of 50% or more. Patients with myeloma were excluded, as were those with hydronephrosis with cortical atrophy. Overall, 665 patients (209 pmp/year) had ARF as defined, and 36% of these (75 pmp/year) required renal replacement therapy. Again, these data must underestimate the true need for dialysis, as only patients who were referred to tertiary centres were included in the analysis.
The most recent study of the incidence of ARF is a 12-month prospective analysis of patients presenting to three district general hospitals in East Kent between March 1997 and February 1998.5 Inclusion criteria were a temporary rise in serum creatinine to 
300 µmol/l and/or a rise in blood urea above 40 mmol/l, or clinical features indicating an acute deterioration of previously normal renal function. Patients with acute-on-chronic renal failure were also included. Patients with myeloma, hydronephrosis or renal failure in the context of untreatable terminal illness were excluded. These criteria were fulfilled by 288 patients (486 pmp/year) of whom 48 (81 pmp/year) received renal replacement therapy.5 These data must also underestimate the true need for dialysis, as not all patients who may have benefited from renal replacement therapy were able to receive it (P. Stevens, personal communication).
An analysis of hospital discharge diagnoses relating to ARF, routinely collected on SMR1 in Scotland shows a consistent year-on-year increase in the number of patients with ARF, however defined, between 1990 and 1999 (Figure 3).7 In 1990, 408 patients had a principal discharge diagnosis of ARF (ICD9:584 or ICD10:N17). By 1999, this figure had risen to 965 patients. Only a small proportion were recorded as having received dialysis or haemofiltration (OPCS4:X40) (Figure 3). Despite the year-on-year increase, an element of under-reporting seems likely for two reasons: only principal diagnoses were used in this analysis, and clinicians were not obliged to record a diagnosis of ARF or an episode of dialysis in patients during the course of an acute illness.
|
Against this background, the results of our study show a progressive rise between 1994 and 2000 in the need for short-term dialysis, the incidence of which is nearly twice the incidence of chronic renal failure accepted for dialysis and more than twice the most recent estimate of the incidence of ARF in the UK. Our results cannot be explained by inappropriate dialysis of patients who would have recovered without treatment, and cannot be due to misclassification of patients with chronic renal failure, as our estimate of the incidence of patients requiring long-term dialysis is close to that reported by the Scottish and UK Renal Registries for chronic renal failure.1,2 We also compared the incidence of short-term dialysis with that obtained using a more conventional definition for ARF, which includes all ARF and acute on chronic renal failure but excludes patients with chronic renal failure who present acutely. Given the ease of classification by duration of dialysis and the difficulties associated with the conventional classification of renal failure, there was a remarkable degree of concordance between the two estimates of incidence. This must reflect the fact that the large majority of our patients undergoing short-term dialysis had ARF or acute-on-chronic renal failure. Indeed, only 11 were judged by conventional criteria to be cases of chronic renal failure presenting acutely.
It remains to be determined whether these results are peculiar to patients living in south-west Scotland or typical of the need for treatment of ARF generally. For reasons that are not entirely clear, the epidemiology of ARF has thus far attracted less attention than that of chronic renal failure. Nevertheless, a recently completed study conducted by colleagues in north-east Scotland in a larger population but over a shorter period of time, has found a similar high incidence of ARF requiring renal replacement therapy.8 Additional circumstantial evidence supports the view that the requirement for treatment is greater than before. In Dumfries, the assessment of patients with multi-organ failure in the intensive care unit was undertaken jointly by intensivists and nephrologists: treatment was by nephrologists and all patients treated had ARF. By contrast, an international survey of severe ARF in critically ill patients in 345 centres, most of which were in Europe and North America, found that renal replacement therapy was frequently used in the absence of ARF for fluid control, congestive heart failure, adult respiratory distress syndrome and sepsis.9 If anything, therefore, these data lead us to believe that our results may underestimate rather than overestimate the need for treatment.
As argued earlier, the reason why the traditional classification of renal failure fails is that the assignment of patients to one or other category often requires a pre-morbid measurement of serum creatinine. Patients who do not have such measurements, who present for the first time in advanced renal failure requiring dialysis, and who then die during the acute illness or require long-term dialysis, are difficult to classify. The advantages of the method we propose are that incidence is much easier to estimate and entirely objective. There is no room for misinterpretation or inconsistency in judgement. The fact that our estimate of incidence for short-term dialysis is closely similar to that obtained by application of more conventional criteria for ARF further commends its use. If the high level of short-term dialysis/ARF found in our study is confirmed by other centres, then this will have significant resource implications for the provision of renal care.
|
Acknowledgments |
|---|
We thank Susan Frame of the Information and Statistics Division of the NHS in Scotland for providing the record linkage data, and Mrs Josephine Campbell for her help in preparation of the manuscript. This study was supported in part by an Infrastructure Grant from the Chief Scientist Office.
|
Notes |
|---|
Address correspondence to Dr Sue Robertson, Renal Unit, Dumfries and Galloway Royal Infirmary, Dumfries DG1 4AP. e-mail: c.isles{at}dgri.scot.nhs.uk
|
References |
|---|
|
TopSummaryIntroductionMethodsResultsDiscussionReferences |
|---|
1. Simpson K, Metcalfe W, Prescott G. The Scottish Renal Registry Report, 1999. The Scottish Renal Registry, Glasgow 2001 [www.show.scot.nhs.uk/srr/].
2. Ansell D, Feest T. Third Annual Report of the UK Renal Registry. UK Renal Registry, South Mead Hospital, Bristol, 2000 [www.renalreg.com].
3. Feest TJ, Round A, Hamad S. Incidence of severe acute renal failure in adults: results of a community based study. Br Med J1993; 306:481–3.
4. Khan IH, Catto GRD, Edward N, Macleod AM. Acute renal failure: factors influencing nephrology referral and outcome. Q J Med1997; 90:781–5.
5. Stevens PE, Tamimi NA, Al-Hasani MK, et al. Non specialist management of acute renal failure. Q J Med2001; 94:533–40.
6. Liano F, Pascual J. The Madrid Acute Renal Failure Study Group. Epidemiology of Acute Renal Failure: a prospective multi-centre community based study. Kidney Int1996; 50:811–18.[Web of Science][Medline]
7. Information and Statistics Division of NHS in Scotland. Trinity Park House, South Trinity Road, Edinburgh EH5 3SQ.
8. Metcalfe W, Simpson M, Khan IH, Prescott G, Simpson K, Smith C, MacLeod AM on behalf of the Scottish Renal Registry. Acute renal failure requiring renal replacement therapy: incidence and outcome. Q J Med2002; 95:579–83.
9. Ronco C, Zanella M, Brendolan A, et al. Management of severe acute renal failure in critically ill patients: an international survey in 345 centres. Nephrol Dial Transplant2001; 16:230–7.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Ishani, J. L. Xue, J. Himmelfarb, P. W. Eggers, P. L. Kimmel, B. A. Molitoris, and A. J. Collins Acute Kidney Injury Increases Risk of ESRD among Elderly J. Am. Soc. Nephrol., January 1, 2009; 20(1): 223 - 228. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. M. Cohen, E. G. Tessier, M. J. Germain, and N. B. Levy Update on Psychotropic Medication Use in Renal Disease Psychosomatics, February 1, 2004; 45(1): 34 - 48. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. METCALFE, M. SIMPSON, I.H. KHAN, G.J. PRESCOTT, K. SIMPSON, W.C.S. SMITH, and A.M. MacLEOD Acute renal failure requiring renal replacement therapy: incidence and outcome QJM, September 1, 2002; 95(9): 579 - 583. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||