An integrated review of "unplanned" dialysis initiation: reframing the terminology to "suboptimal" initiation

Background Ideally, care prior to the initiation of dialysis should increase the likelihood that patients start electively outside of the hospital setting with a mature arteriovenous fistula (AVF) or peritoneal dialysis (PD) catheter. However, unplanned dialysis continues to occur in patients both known and unknown to nephrology services, and in both late and early referrals. The objective of this article is to review 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 MEDLINE and EMBASE from inception to 2008 were used to identify studies examining the clinical, economic or quality of life (QoL) outcomes in patients with an unplanned versus planned first dialysis. Data were described in a qualitative manner. Results Eight European studies (5,805 patients) were reviewed. Duration of hospitalization and mortality was higher for the unplanned versus planned population. Patients undergoing a first unplanned dialysis had significantly worse laboratory parameters and QoL. Rates of unplanned dialysis ranged from 24-49%. The total annual burden to the Canadian healthcare system of unplanned dialysis in 2005 was estimated at $33 million in direct hospital costs alone. Reducing the rate of unplanned dialysis by one-half yielded savings ranging from $13.3 to $16.1 million. Conclusion The clinical and socioeconomic impact of unplanned dialysis is significant. To more consistently characterize the unplanned population, the term suboptimal initiation is proposed to include dialysis initiation in hospital and/or with a central venous catheter and/or with a patient not starting on their chronic modality of choice. Further research and implementation of initiatives to reduce the rate of suboptimal initiation of dialysis in Canada are needed.


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 dis-ease (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 1865 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 selfreported 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 3059 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).

Definition of Unplanned Dialysis
The majority of studies used the term "unplanned". Two studies used the terms "known acute" and "non-programmed" [8,14]. With the exception of three studies that included only patients referred to nephrology prior to first dialysis [8,13,14], most enrolled patients regardless of referral status to nephrology. Three studies defined unplanned dialysis as starting in a life threatening situation [8,10,11]. Two other studies defined unplanned as those patients that did not have a permanent access device in place [12,13].

Baseline Clinical Status
Two studies were designed to examine the baseline characteristics of patients starting unplanned dialysis. The first study was a retrospective survey conducted at a large regional renal network [8]. Elective patients commenced RRT in the outpatient setting using either an AVF, PD catheter or by pre-emptive transplantation. Known Acute patients were known to renal services for more than four months and started dialysis with a hemodialysis catheter or during an emergency hospital admission. Patients had been known for a median time of three years with no statistical difference in the length of time between the two groups. At the start of RRT, known acute patients had statistically significant lower concentrations of serum albumin and hemoglobin than elective patients. This group also had higher serum creatinine, urea and phosphate concentrations compared with the elective group.
A multicentre retrospective review examined the factors affecting a planned versus non-planned start of dialysis [9]. A planned start was a scheduled outpatient initiation of dialysis with the use of a permanent vascular or peritoneal access. Conversely, a non-planned start was unscheduled and included patients with and without a permanent dial- ysis access in place. Forty-six percent started dialysis in a non-planned fashion, although half of these patients had been followed by a nephrologist for at least 3 months. Planned starts were associated with a younger age, longer renal and pre-dialysis follow-up, more patient education, more medical visits, more follow-up by specific ESRD units, more permanent access (including PD), and better renal function and biochemical status at the start of dialysis.
Three other studies evaluated baseline clinical status and outcomes over a longer period of time. Similar to previous findings [8,9], planned dialysis was associated with higher baseline levels of serum hemoglobin, calcium and albumin and lower baseline levels of serum urea, creatinine and phosphate [10,12,13]. Gorriz et al (2002) also showed that unplanned dialysis initiation was associated with uremic symptoms, fluid overload and increased transfusion requirements [12]. Unplanned patients were older, and had a shorter follow-up period of 3 months [13]. In both studies, unplanned patients had a higher comorbidity index [12,13].

Morbidity and Mortality
Four studies examined morbidity and mortality outcomes over study periods ranging from 6 months to three years. Duration of hospitalization was consistently longer for unplanned dialysis. Mortality was also higher in the unplanned dialysis group in all [11][12][13] but one study [10].
A retrospective Spanish study classified 117 patients as either programmed or non-programmed [10]. A patient was programmed when dialysis was planned with time, and non-programmed when the first dialysis was started for an emergency condition or was not appropriate to delay for >24 hours. Metcalfe et al [13] performed a prospective nationwide study of all patients commencing RRT in Scotland over a one-year period. Patients were classified as planned when a permanent vascular or peritoneal access was ready for use at the first RRT, and unplanned when follow-up by a nephrologist occurred for <1 month, with steady progression to ESRD, and no permanent access. An unplanned presentation occurred in 129/532 patients, and relative to those with a planned presentation, these patients were more likely to die within 90 days (12.4% vs. 3.1%; p = 0.001), with a longer hospital admission (median, 9 vs. 3 days; p < 0.001) at the time of RRT initiation.

Economic Outcomes
The Gorriz et al study described above evaluated the direct medical costs (dialysis, hospital, and physician costs) of unplanned dialysis initiation [12]. On a per patient basis, the direct costs associated with an unplanned initiation were 4.4 times that of a planned initiation during the first six months of dialysis. This difference was due mainly to a higher number of dialysis sessions during hospitalization and a higher incidence of hospitalization.
As shown in Table 3, by conservatively assuming that planned first dialysis requires a planned hospital admission ($3,485) and a significantly lower rate of re-hospitalization within one year (10%), the annual hospital costs avoided by preventing one unplanned dialysis start is an estimated $16,791. Reducing the rate of unplanned dialysis by one-half (30% to 15%) or by two-thirds (30% to 10%), yields savings of $13.3 and $17.7 million, respectively (Table 4). Reframing the analysis without an initial planned hospitalization, shows that the annual hospital costs avoided by preventing one unplanned start is $20,276 (Table 3) with even greater savings of $16.1 and $21.4 million when that rate is reduced by one-half and two-thirds, respectively (Table 4).

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 2449%.
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 predialysis 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, nonclinic 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 reduc-  tion 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 19989, 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 5570%. 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.
Publish with Bio Med Central and every scientist can read your work free of charge