Background
Chronic kidney disease (CKD) is a growing public health concern. At the end of 2005, an estimated 1 in 1,000 Canadians had been diagnosed with end-stage renal disease (ESRD) with almost 61% (19,721 of 32,375) receiving dialysis [
1]. Between 1996 and 2005 the incident rate of renal replacement therapy (RRT) rose 36% in Canada from 119 to 162 per million.
Despite advances in nephrological care prior to and after dialysis is initiated, ESRD patients continue to have a high morbidity and mortality, and a significant decline in quality of life. In Canada, the five-year survival of ESRD patients on dialysis ranges from 20% for diabetics over age 65 years to 59% for non-diabetics aged 18-65 years [
1]. The mean number of co-morbid conditions in dialysis patients is approximately four per patient, the mean hospital days/patient/year is approximately 15, and self-reported quality of life is far lower than the general population [
2].
The cost of treating ESRD in Canada is significant. In 2000, the direct health-care expenditures for ESRD were estimated at $1.3 billion [
3]. Patients on dialysis were responsible for approximately two-thirds (69%) of all ESRD expenditures. Although only 0.1% of Canadians had ESRD, these costs represented 1.3% of Canada's total health-care spending [
3].
Canadian guidelines recommend referral to a nephrologist for patients with: acute kidney failure; eGFR < 30 mL/min/1.73 m
2; progressive decline of eGFR; persistent proteinuria; or, inability to achieve treatment targets or other difficulties in the management of CKD [
4]. Similarly, the National Kidney Foundation's Kidney Disease Outcome Quality Initiative (NKF/KDOQI) guidelines recommend co-management with a nephrologist at stage 3 CKD (eGFR 30-59 mL/min/1.73 m
2), and referral at stage 4 (eGFR <30 mL/min/1.73 m
2) [
2].
A recent meta-analysis evaluating timing of referral before starting RRT, has shown that patients referred late to nephrologists have a two-fold higher risk of death compared with early referral (relative risk 1.99; 95% confidence interval (CI), 1.66 to 2.39, p < 0.001) [
5]. The duration of hospital stay at the time of initiation of RRT was also greater in the late referred group by an average of 12 days (95% CI, 8.0 to 16.1; p = 0.0007) [
5]. A Canadian economic evaluation of early versus late referral of patients with progressive renal insufficiency to a multidisciplinary clinic, showed that early referral is cost-effective and is associated with an incremental cost savings and a reduction in hospital days [
6].
Early referral should increase the likelihood that patients initiate dialysis electively, outside of the hospital, with a mature arteriovenous fistula (AVF) or peritoneal dialysis (PD) catheter on the optimal chronic modality of choice. However, unplanned dialysis continues to occur in patients both known and unknown to nephrology services and in both late and early referrals. Recently, Mendelssohn
et al [
7] found that 70% of incident HD patients in Canada start with a central venous catheter (CVC).
The objective of this article is to review the available published literature that examines the clinical and socioeconomic outcomes of unplanned dialysis initiation. The secondary objective is to explore the potential cost implications of reducing the rate of unplanned first dialysis in Canada.
Methods
The literature review included studies examining the clinical, economic or quality of life outcomes in patients with an unplanned, compared to those with a planned, first dialysis. Unplanned dialysis was broadly defined as any patient who received unanticipated dialysis regardless of location or previous referral status to nephrologists. Terms considered synonymous to unplanned dialysis were unscheduled, unprogrammed, urgent, and emergent. The search strategy included MESH headings "kidney failure, chronic" or "dialysis." These headings were combined with the non-MESH headings "planned" or "unplanned" or "emergent" or "unscheduled" or "non-programmed" and the MESH heading "time factors". Databases used were MEDLINE and EMBASE from inception to 2008. The reference lists of published papers examining the impact of early referral status were examined for additional relevant studies. Reviews, editorials, practice guidelines, and studies conducted in children were not included. Data from each study were described in a qualitative manner. Crude cost impact estimates of unplanned dialysis in Canada were performed based on data from the Canadian Organ Replacement Registry Report (CORR) and the Canadian Institute for Health Information (CIHI).
Discussion
This review of eight studies comparing clinical, economic and patient reported outcomes in 5,805 European dialysis patients has shown that duration of hospitalization and mortality is higher for the unplanned versus planned population. Patients undergoing a first unplanned dialysis have significantly worse laboratory parameters at baseline compared with patients undergoing planned dialysis. As well, QoL was significantly worse in unplanned dialysis patients both at baseline, and as early as eight weeks after starting dialysis. The studies reviewed in this article reported rates of unplanned dialysis ranging from 24-49%.
Some patients may have primary care physicians that underestimate the potential benefits of dialysis and/or the length of time required to optimally prepare a patient for dialysis [
8]. Educational efforts targeting primary care givers, explaining the clinical, economic, and quality of life benefits of the timely dialysis planning are needed to ensure that all patients with renal failure that potentially require dialysis are referred to a nephrologist in a timely manner. The Canadian Society of Nephrology has a policy document and implementation strategy intended to achieve timely referral of appropriate patients [
4].
Pre-dialysis education seems to be significant in determining whether patients have an optimal or suboptimal dialysis start [
9]. Patients attending multidisciplinary pre-dialysis clinics were more likely to present with a functioning permanent vascular access at dialysis initiation (48% vs. 5%; p < 0.01) [
17]. Moreover, they had fewer hospitalizations at 1-year (7.0 vs. 69.7 d/patient/y; p < 0.01), with fewer deaths at 1-year (2% vs. 23%; p < 0.01) [
17]. In a Cox-adjusted linear regression model, non-clinic pre-dialysis care was shown to be an independent predictor of death during therapy (RR 2.9; p = 0.011) [
17]. A matched-cohort study evaluating the effectiveness of multidisciplinary care (MDC), showed a 50% mortality risk reduction for MDC compared with non-MDC (HR 0.50; 95% CI, 0.35 to 0.71) [
18]. A trend towards a reduction in risk for all-cause and cardiovascular-specific hospitalizations was observed (p = NS). The opportunity to educate patients/caregivers leading to informed decisions may improve QoL and decrease economic resource utilization since patients referred early may be more likely to choose peritoneal rather than hemodialysis [
19].
Whether to advise a patient to undergo an attempt to place an AVF is a complex decision that requires a nephrologist's anxious consideration. On the one hand, it is documented in the literature that there is a subset of patients who die with a functional AVF or PD catheter who never started dialysis [
20]. On the other hand, there is a compelling moral and ethical obligation to avoid long term catheter based vascular access [
21]. Selecting only patients with progressive nephropathy who will need to initiate HD is an imprecise science at present and requires considerable clinical judgement.
One challenge with comparing outcomes across the studies reviewed in this article is the variability in the definition of unplanned dialysis. Unplanned dialysis occurs in patients both known and unknown to nephrology services and in both late and early referrals. Some researchers defined unplanned dialysis only if it was started in a life threatening situation [
8,
10,
11]. Other researchers also included an element of timing of nephrology referral (e.g. 1 to 4 months) [
8,
13,
14] or the lack of a ready to use vascular or peritoneal access [
8,
9,
12] in the definition of unplanned dialysis. We suspect that our search has missed articles because the terms planned and unplanned dialysis starts are difficult to define and inconsistently used in the literature.
To improve evaluation of the effectiveness of strategies for reducing the incidence of unplanned dialysis, a more consistent definition is required. We propose the term suboptimal initiation to include all patients starting in hospital and/or with a central venous catheter, and/or not starting on their chronic modality of choice. In contrast, an optimal start occurs when patients initiate dialysis electively in an outpatient setting with a mature AVF or PD catheter, on the patient's chosen chronic dialysis modality. We believe this definition is simple and precise, will be accepted by clinicians and researchers, and can be more consistently applied.
Sub-optimal dialysis initiation is surprisingly common. In Canada, 70% of incident patients start with a CVC [
7]. In addition, 55% of patients attending a multidisciplinary pre-dialysis clinic at a Toronto hospital did not have a functioning permanent vascular access at the time of starting hemodialysis [
17]. Another Canadian cohort study conducted in 15 dialysis centres across 7 provinces during 1998-9, revealed that only 66% of those known to nephrologists had a permanent access in place [
22]. Based on these studies, the rate of sub-optimal initiation in Canada is potentially between 55-70%. The estimated cost impact of suboptimal dialysis initiation in this analysis (which is based on a literature rate of 30%) is likely a marked underestimation.
Once referred to a nephrologist, Canadian clinical practice guidelines recommend monitoring renal function and nutritional status every three months, although the actual frequency of clinical evaluation should still be based on clinical judgment [
4]. This monitoring frequency is thought to be sufficient to detect patients with a more rapid rate of decline in renal function. It may also permit specific and targeted interventions to slow the decline in renal function or, alternatively, speed preparation for dialysis. However, even in the studies including only patients already referred to a nephrology service, suboptimal dialysis initiation was associated with worse laboratory parameters at baseline and an increase in hospitalization and mortality, suggesting that early dialysis referral does not guarantee optimal care [
4,
13,
14,
23].
It is disturbing that suboptimal initiation occurs commonly even when patients are referred to a nephrologist early. A preliminary list of causes of suboptimal dialysis initiation despite early referral includes a) acute on chronic kidney disease, b) patient induced delays and indecision, c) barriers to surgical resources, d) suboptimal nephrology care and e) lack of dialysis resources to accommodate new patients. We recommend that the factors on this list (and possibly others) need to be investigated and quantified so that new approaches can be developed to overcome them.
Conclusion
This review of eight studies comparing outcomes in 5,805 European dialysis patients has shown that duration of hospitalization and mortality is higher patients undergoing suboptimal initiation. These patients have significantly worse laboratory parameters at baseline, and lower QoL compared with patients initiating dialysis in an optimal fashion.
In Canada, costs associated with suboptimal dialysis initiation are significant. By reducing the estimated rate of unplanned dialysis by one-half, a projected $13.1 to 16.1 million in hospital costs could be avoided. This estimate is conservative since it did not include direct medical costs outside of the hospital and indirect costs such as loss of quality of life and productivity. Further research and initiatives to reduce the rate of suboptimal dialysis initiation in Canada are needed.
Competing interests
DM has served on advisory boards and has received honorariums for lecturing from Amgen, Baxter, Genzyme, Ortho Biotech, Roche, Shire and several BP medication companies. CM and BH have acted as consultants for Ortho Biotech Canada. The results of this paper have not been presented or published elsewhere, in whole or in part.
Authors' contributions
CM and BH performed the integrative review and wrote the first draft. DM reviewed the analysis, and wrote the final draft. All authors contributed to the conception and design, and read and approved the final manuscript.