Delayed referral to a nephrologist: outcomes among patients who survive at least one year on dialysis

The national renal registries of Australia/New Zealand, the United States, Canada and Japan have reported an increasing incidence of end-stage renal disease (ESRD) of between 3% and 8% per year between 1993 and 1997.1 Despite improvements in treatment, the mortality of people on dialysis remains high.2 Annual mortality rates in the various renal registries range from 10% to 20%.3

The proportion of ESRD patients referred "late" to a nephrologist (ie, patients needing to commence dialysis within three to four months of referral) varies widely. In developing countries, the proportion is as high as 62%,4 while in developed countries it is normally 25%–40%.5-8 Previous US and UK reports have shown that increasing age and coexisting illnesses,9 ethnicity10 and membership of a health maintenance organisation5 are associated with late referral.

Late-referral (LR) patients on dialysis experience greater early morbidity and higher early mortality. Late referral, associated with advanced uraemic symptoms, metabolic acidosis, hypertension, pulmonary oedema and pericarditis, frequently results in emergency haemodialysis using central venous catheters.6-8 It is also associated with longer6,7 and more costly6 initial hospitalisation. Early mortality, during the initial 6–12 months on dialysis, is higher for LR patients,4,11 but little is known about survival differences beyond the first year.

The aim of our study was to investigate whether or not late referral to a nephrologist influences the rates of transplantation and mortality among patients who have survived at least one year's dialysis.

Of the 4243 patients included in the study, 1141 (26.9%) were LR patients (Box 1). Hypertensive renal disease, other types of primary renal disease and uncertain diagnoses were more common in the LR group. Primary glomerulonephritis, polycystic disease, analgesic nephropathy and reflux nephropathy were more common in the non-LR group. The LR group had a greater burden of comorbid illness.

There were no significant differences in age or sex between LR and non-LR patients, but a significantly higher proportion of the LR group was of Indigenous origin (13.4% v 7.7%; P < 0.001).

LR patients were also less likely to receive a transplant in their second and subsequent years on RRT (unadjusted rate ratio, 0.71; 95% CI, 0.58–0.86). This difference remained significant after adjustment for age, sex, number of comorbidities, primary cause of renal disease and Indigenous status (adjusted rate ratio, 0.78; 95% CI, 0.63–0.95).

Kaplan–Meier survival curves, according to timing of referral, showed a significant difference in survival after the first year on RRT (P value for log rank test, < 0.001) (Box 1). The mortality rate was 20 deaths (95% CI, 18–22 deaths) per 100 patient-years for the LR group and 15 deaths (95% CI, 14–16 deaths) for the non-LR group. The unadjusted hazard ratio for death in the LR group compared with the non-LR group was 1.30 (95% CI, 1.14–1.48). After adjusting for known predictors of mortality (age, sex, number of comorbidities, presence of primary renal disease and Indigenous status), the hazard ratio for death in the LR group was still significant (1.19; 95% CI, 1.04–1.35). The hazard ratio for death was significant even when Indigenous patients were excluded from the analysis. Inclusion of a variable that described the dialysis modality in the first year of treatment made no significant difference.

It is possible that the difference in survival rate between LR and non-LR patients is related to the level of renal function at the start of dialysis, which has been shown to be a determinant of patient survival.16 If non-LR patients start treatment earlier in the course of their chronic renal disease (at a stage when renal function is significantly less impaired), subsequent survival on dialysis may be longer. ANZDATA has, since 1998, collected data regarding the level of renal function at the start of dialysis. Of patients starting RRT during 1998, the LR patients (n = 358) had a mean creatinine clearance of 8.0 mL/min (SD, ± 7.1) and non-LR patients (n = 1133) had a mean clearance of 7.9 mL/min (SD, ± 3.6), a non-significant difference.14 Assuming that the results would have been similar for the other study years (1995–1997), the observed survival difference would not appear to be directly related to the level of renal function at the start of dialysis.

Incomplete adjustment for intervening and confounding variables may be part of the explanation for the survival difference attributed to late referral. Chandna et al11 have shown that a total comorbidity severity score is a better predictor of mortality on RRT than the number of comorbidities. We were unable to explore this possibility, as ANZDATA does not collect data on the severity of comorbid illnesses. However, if late referral is an indicator of suboptimal pre-ESRD care, it is plausible that worse outcomes might be due to inadequate management of comorbid illnesses, including vascular disease and heart failure.

Two previous studies7,8 of the effect of late referral on long-term survival (at least five years' follow-up) found no significant difference in long-term survival between LR and non-LR patients. Another study11 found that unplanned presentation (which is not the same as late referral) adversely affected survival. However, all three studies involved fewer than 300 patients and had limited power to detect a significant difference between LR and non-LR groups. By contrast, our study, based on a national cohort, has much greater power. Moreover, because our study (unlike previous studies) excluded all patients who died in their first year on dialysis, we were able to separate the short-term effects of unplanned commencement of RRT from the long-term disadvantage arising from suboptimally managed chronic renal insufficiency.

Our results suggest that improving the quality of pre-ESRD care, through timely referral, might improve long-term survival on RRT. Angiotensin-converting enzyme inhibitors,17,18 angiotensin-II-receptor antagonists,19 rigorous blood pressure control20 and rigorous glycaemic control21 have all been proven effective in slowing the progression of chronic renal insufficiency. However, there have been no definitive studies demonstrating methods to reduce mortality in people with chronic renal impairment.

The continuing high incidence of late referral seems to indicate that nephrologists are failing to communicate to physicians and general practitioners the importance of optimal pre-ESRD care. Primary care doctors may be unaware of the severity of renal insufficiency in some patients, particularly if serum creatinine level is the only measure used to monitor renal function.22 Late referral may stem from uncertainty about the appropriateness of RRT for a given patient,22 or the perception that treatment services are not easily accessible to the patient. In a US survey of GPs who referred new ESRD patients to a renal unit, Campbell et al23 found that key factors delaying referral were lack of knowledge of guidelines (relating to timing and indications for referral) and inadequate communication between nephrologists and GPs.

Levin13 contends that "many specialists (and general practitioners) perceive nephrologists only as providers of dialysis therapy . . . [and fail to appreciate] the utility of nephrological care during early stages of renal insufficiency". Although the Australian Kidney Foundation has issued guidelines for the care of people with renal impairment,24 to our knowledge no attempt has been made to assess the awareness of these guidelines among GPs and non-nephrologists.

The AusDiab study25 found that 2.5% of Australian adults aged 25 years or over had significant proteinuria and that 1.1% had a serum creatinine level of over 120 μmol/L. This suggests that several hundred thousand Australians have indicators of renal disease. Screening studies in Japan indicate that people with proteinuria are 15 times more likely than those without proteinuria to develop renal failure within 10 years.26 We believe that there is sufficient evidence that progression of chronic renal insufficiency to ESRD can be prevented to suggest that targeted screening for renal disease among people in high-risk groups should be undertaken in general practice. The most important potential benefit, despite the absence of strong evidence, could be reduced mortality among people with chronic renal insufficiency.

We suggest that GPs use dipstick urinalysis for proteinuria to screen patients who have any one of the following risk factors: age over 50, hypertension, diabetes, smoking, family history of renal disease, or being of Indigenous origin. Serum creatinine level should be measured to calculate the glomerular filtration rate using the Cockcroft–Gault equation.27 Where appropriate, patients should be treated aggressively for hypertension, proteinuria and other vascular risk factors, and, if the glomerular filtration rate falls below 30mL/min, referred promptly to a nephrologist.24

We must urgently address the lack of a strong evidence base in the management of patients with chronic renal insufficiency. These patients need a continuum of care from the time of diagnosis to the onset of ESRD. Achieving optimal treatment will require a true collaboration between GPs and specialist nephrologists.

2: Kaplan–Meier curves of cumulative survival, by timing of referral

Number of patients at risk at beginning of each year of study

	Year 1	Year 2	Year 3	Year 4
LR patients	1141	704	384	132
Non-LR patients	3102	1869	909	318
LR = late referral.