- Email: [email protected]
An economic evaluation of early versus late referral of patients with progressive renal insufficiency
SPECIAL ARTICLE
An Economic Evaluation of Early Versus Late Referral of Patients With Progressive Renal Insufficiency Kevin McLaughlin, MB ChB, Braden Manns, MD, Bruce Culleton, MD, Cam Donaldson, PhD, and Kenneth Taub, MD ● Patients with progressive renal insufficiency (PRI) who start renal replacement therapy (RRT) within 4 months of seeing a nephrologist (late referral) have increased morbidity, mortality, and health care costs. We performed an economic evaluation of early versus late referral of patients with PRI to a multidisciplinary clinic. A decision analysis was performed from the perspective of the health care provider, using a Markov model to simulate progression of PRI and survival of patients on RRT. Our simulated patient cohort comprised 1,000 patients with PRI and estimated creatinine clearance of 20 mL/min. The study time horizon was 5 years. Clinical and cost data were taken from published Canadian and U.S. data, where available. Where published data were lacking, we used data from our prospectively maintained database. The study intervention was attendance at a PRI clinic where patients receive treatment to slow the rate of renal progression, receive treatment of complications of PRI, and are prepared for RRT. Endpoints were total cost of patient care, patient life-years, patient life-years free of RRT, and hospital admission days. Early referral resulted in cost savings and improved patient survival along with more life-years free of RRT and fewer hospital inpatient days. Cost-effectiveness was unaffected by univariate sensitivity analyses. Cost-effectiveness decreased as rates of renal function loss for patients referred early versus late approximated each other. In conclusion, early referral of patients with PRI to a multidisciplinary clinic appears cost-effective. © 2001 by the National Kidney Foundation, Inc. INDEX WORDS: Economic evaluation; progressive renal insufficiency (PRI); referral; end-stage renal disease (ESRD).
Appendix appears only on the web site (www.ajkd.org) electronic pages
A
PPROXIMATELY ONE third of patients with progressive renal insufficiency (PRI) have clinical features of uremia when they first are seen by a nephrologist and need to start renal replacement therapy (RRT) within 4 months of their first consultation.1 Patients referred to nephrologists at a late stage in the course of PRI appear to have increased morbidity and mortality compared with patients referred early.2-5 The potential benefits of early referral of patients From the Departments of Medicine, Community Health Sciences, and Economics, University of Calgary, Alberta, Canada. Received April 9, 2001, and accepted as submitted May 25, 2001. Supported in part by a Kidney Foundation of Canada start-up grant (K.M). Address reprint requests to Kevin McLaughlin, MB ChB, Department of Medicine, Division of Nephrology, Foothills Hospital & University of Calgary, 1403 29th Street NW, Calgary, Alberta T2N 2T9 Canada. E-mail: [email protected] © 2001 by the National Kidney Foundation, Inc. 0272-6386/01/3805-0026$35.00/0 doi:10.1053/ajkd.2001.28619 1122
with PRI to a nephrologist at a multidisciplinary PRI clinic include identifying and treating reversible causes of renal failure; slowing the rate of decline of renal function; managing the multiple coexisting conditions associated with PRI; and optimizing the biochemical, physical, and psychological state of the patient at the time of starting RRT.6-8 The increased morbidity of patients referred to a nephrologist at a late stage is likely to lead to greater usage of health care resources. It has been suggested that early referral would produce a cost saving in addition to the health benefits to the patient.9,10 This theory remains unproven, however. We performed an economic evaluation using decision analytic modeling to compare early versus late referral of patients with PRI to a nephrologist at a multidisciplinary PRI clinic. METHODS
Decision Analytic Model We constructed a decision analysis tree using ithink version 6.0 (High Performance Systems, Inc, Hanover, NH) to evaluate the costs and effects of two different clinical strategies for managing patients with PRI (Fig 1). The two clinical strategies were refer patients early (ie, patients are referred to a nephrologist at a multidisciplinary PRI clinic when their creatinine clearance is 20 mL/min) and refer patients late (ie, patients are referred when they are uremic
American Journal of Kidney Diseases, Vol 38, No 5 (November), 2001: pp 1122-1128
COST EFFECTIVENESS OF EARLY REFERRAL
Fig 1. Clinical strategies and health states in the Markov model. Abbreviations: RRT, renal replacement therapy; PRI, progressive renal insufficiency not yet requiring RRT; HD, hemodialysis; PD, peritoneal dialysis.
and require RRT). The intervention in this model is referral to a multidisciplinary team comprising a nephrologist, nurse, dietitian, social worker, and pharmacist, who provide comprehensive patient care. Four outcomes were considered: (1) average total cost of patient care, (2) patient life-years, (3) patient life-years free of RRT, and (4) hospital inpatient days. We used a simulated patient cohort of 1,000 average patients who had a mean estimated creatinine clearance of 20 mL/min. Subsequent to reaching this level of renal function, a patient can enter into one of five health states: (1) PRI not yet requiring RRT, (2) hemodialysis, (3) peritoneal dialysis, (4) renal transplantation, and (5) death. A Markov process was used to model transitions between these states on a monthly basis over a period of 5 years. The transition probabilities are outlined in Table 1.
General Assumptions of the Markov Model In our model, the following assumptions have been made: PRI progresses in all patients; there is a constant proportion of patients starting RRT each month (different for each strategy); there is a constant rate of hospital admission in the two limbs (different for each strategy); there is a constant proportion of patients switching between different modalities of RRT each month (equal for each strategy in the base-case analysis); and there is a constant proportion of patients dying each month depending on the health state (equal for each strategy in the base-case analysis). A discount rate of 5% (recommended rates for Canada) for cost and effect was used for the base-case analysis.11 It also was assumed that patients who were referred early continued to see their family physician (shared patient care).
Data Sources Published Canadian and U.S. data were taken where available to estimate base-case values and to provide ranges for sensitivity analyses. If no suitable published data were available, mean data over a 3-year period (March 31, 1997 through March 3, 2000) were taken from the prospectively maintained database at the Southern Alberta Renal Program (SARP). For patients followed by a nephrologist at a PRI clinic, the estimated rate of loss of
1123
residual renal function was 4.8 to 6.4 mL/min/y.12,13 Taking a mean of 5.6 mL/min/y (range, 4.8 to 6.4 mL/min/ y), from a baseline creatinine clearance of 20 mL/min/y, the estimated time to starting RRT (creatinine clearance, 12 mL/min) for this cohort was 17 months.14 It has been shown that patients followed by a nephrologist at a multidisciplinary PRI clinic may have an average extended duration of time without dialysis of 4.6 months compared with patients managed by standard care.8 Using this figure to estimate the time to RRT for patients not followed by a nephrologist at a PRI clinic (12.4 months), we estimated the median rate of loss of renal function in the late referral cohort to be 7.7 mL/min/y (range, 6.2 to 9.2 mL/min/y). These rates of deterioration corresponded to a constant proportion of patients starting RRT each month of 2.9% (2.5% to 3.3%) and 4% (3.3% to 4.8%). Data on choice of RRT depending on time of referral were taken from published Canadian and U.S. data, as were data on mortality rates for patients starting RRT with hemodialysis and peritoneal dialysis.3,9,15,16 When more than one study was available, the mean was taken for the base-case analysis (Table 1). Survival on transplantation and transition probabilities between RRT modality were taken from published Canadian and U.S. studies.15,17-19 Hospitalization rates for early and late referred patients over a 5-year period were taken from a U.S. study.20 Use of erythropoietin was taken from this study and from a personal communication (A. Levin).20 Mortality rate at PRI clinic and the preemptive transplant rate for early and late referred patients were taken from the SARP database.
Costs We performed the analysis from the perspective of a health care provider. The cost of outpatient care for patients in each of the health states was estimated from published Canadian data.21,22 Treatment costs were converted to 1999 Canadian dollars (1 CDN$ ⫽ 0.67 US$) using the Consumer Price Index for Canada from the years 1995 to 1999.23 These values were used for the base-case analysis. The cost for PRI care was taken from the SARP database and included total staffing costs (including physician costs), support services, supplies, and operational costs. Erythropoietin cost (Alberta, 1999) for patients with PRI was based on a mean dose of 4,000 U/wk (A. Levin, personal communication). The cost per day of inpatient care was estimated from the mean inpatient cost for all dialysis patients at Foothills hospital in 1999.
Sensitivity Analyses We performed sensitivity analyses for clinical variables, transition probabilities, costs, and discount rates (for costs and effects). When more than one study was available, the individual studies were used to provide the range for sensitivity analysis. When the 95% confidence intervals for cost were reported, these were used for sensitivity analysis.21 If confidence intervals were unavailable, a plausible range for cost (⫾25%) was used. For data that were derived from a single study or from the SARP database, a plausible range (⫾25%) was used for sensitivity analysis. To provide the range for discount rates for costs and effects, we used the
1124
MCLAUGHLIN ET AL Table 1.
Base-Case Clinical Values and Costs With Ranges for Sensitivity Analyses
Variable
Base-Case Value
Range in Sensitivity Analysis
EPREX use at PRI clinic (% patients) EPREX use at non-PRI (% patients) EPREX costs (CDN$/y)† PRI clinic costs (CDN$/y)§ HD costs (CDN$/y) PD costs (CDN$/y) Tx costs (CDN$/y) Admission costs (CDN$/d)§ PRI patient GFR loss (mL/min/y) Non-PRI patient GFR loss (mL/min/y) Start HD from PRI clinic (%) Start HD from non-PRI clinic (%) Start PD from PRI clinic (%) Start PD from non-PRI clinic (%) Start Tx from PRI clinic (%)§ Start Tx from non-PRI clinic (%)§ Transition from HD to Tx (%/y) Transition from HD to PD (%/y) Transition from PD to Tx (%/y) Transition from PD to HD (%/y) Transition from Tx to HD (%/y)¶ Transition from Tx to PD (%/y)¶ Death before starting RRT (%/y)§ Death on HD (%/y) Death on PD (%/y) Death on Tx (%/y) Hospitalization rate for PRI patients Hospitalization rate for non-PRI patients Discount rate for cost (%) Discount rate for effect (%)
55 9 3,240 1,366 94,198 47,628 37,106 792 5.6 7.7 57 76 32 24 11 0 10.35 12.7 15.4 15.95 1.03 1.03 5 22.2 21 2.45 13.6 19.6 5 5
40-70 0-17 2,340-4,050‡ 1,024-1,708‡ 87,017-101,381 42,216-53,059 27,830-46,383‡ 437-1,147 4.8-6.4 6.2-9.2 42-73 58-93 16-47 7-42 8.25-13.75‡ 0‡ 7.76-12.94‡ 9.52-15.88‡ 11.55-19.25‡ 11.96-19.94‡ 0.77-1.29‡ 0.77-1.29‡ 3.75-6.25‡ 21.3-23 16.6-25.3 1.1-3.8 10.2-17‡ 14.7-24.5‡ 3-6 3-5
References
3, * 3, *
20 20 21 8, 11, 12 8, 11, 12 3, 9 3, 9 3, 9 3, 9
14 14 14 14 18 18 14, 15 14, 15 16, 17, 18 19 19 23, 24, 25 23, 24, 25
NOTE. ⫾95% Confidence intervals used for sensitivity analysis. Abbreviations: CDN$, Canadian dollars; GFR, glomerular filtration rate; RRT, renal replacement therapy; PRI, progressive renal insufficiency not yet requiring RRT; HD, hemodialysis; PD, peritoneal dialysis; Tx, renal transplantation. *A. Levin: Personal communication. †Cost based on 4,000 units per week (A. Levin, personal communication). ‡⫾25% used for sensitivity analysis. §SARP database. ¶Assumes living-related transplant rate of 29.2% and equal proportions returning to PD and HD.
recommended rates for the United States and United Kingdom.11,24,25 Table 1 shows the ranges used in the sensitivity analyses along with the data source.
RESULTS
Base-Case Analysis Using the early referral strategy, the mean total cost per patient for a 5-year period was CDN$ 130,912 (US$ 87,711). The mean number of life-years per patient during this period was 3.52. Corresponding values for the late referral strategy were CDN$ 164,262 (US$ 110,056) and 3.36. The incremental cost-effective ratio for the
early versus late referral strategy was CDN$ ⫺208,438 (US$ ⫺139,653) per life-year gained. There is a cost saving and an improved patient survival associated with the early referral strategy. There also are more life-years free of RRT and fewer hospital inpatient days for the early referral strategy (Table 2). Sensitivity Analyses The cost-effectiveness of the strategy remained unaffected by univariate sensitivity analysis using the variables and ranges shown in Table 1. The cost-effectiveness of early referral
COST EFFECTIVENESS OF EARLY REFERRAL Table 2.
1125
Results of Base-Case Analysis for Early Versus Late Referral Strategies
Outcome Variable
Early Referral
Late Referral
Cost of nephrology care (CDN$) Patient life-years Life-years free of RRT Total hospital inpatient days
130,912 3.52 2.18 25
164,262 3.36 1.76 41
Incremental Cost-Effective Ratio (CDN$/Life-Year Gained)
⫺208,438
Abbreviations: CDN$, Canadian dollars; RRT, renal replacement therapy.
decreased as the rate of deterioration in renal function for patients referred early or late approached each other. When the rates were equal, there was, by definition, equal life-years off dialysis (2.01 years). There was, however, still a difference in overall patient survival (3.49 years versus 3.42 years), total costs (CDN$ 139,155 [U.S. $93,234] versus CDN$ 150,416 [U.S. $100,779]) and hospital days (27 days versus 37 days) in favor of early referral. The difference in mortality in this situation is largely due to the increased rate of preemptive transplantation in the early referral group. The cost difference is a result of early referral patients choosing less expensive forms of RRT (ie, peritoneal dialysis and preemptive transplantation). We examined the effect of varying the early referral rate (Table 3). These analyses were included to reflect the actual patterns of referral in North America (ie, a varying ratio of early-tolate referrals with no unit having all early or all late referrals). The results show a predictable increase in patient life-years, an increase in patient life-years off RRT, a decrease in the number of hospital inpatient days, and a decrease in total costs for patient care for each incremental increase in the early referral rate. Increasing the early referral rate from 50% to 75% would result Table 3.
in a cost saving of CDN$ 8,272 (U.S. $5,542) per patient over the 5-year period with a gain in life-years of 0.04 years, a gain in life-years off RRT of 0.11 years, and 4 fewer hospital inpatient days. DISCUSSION
This analysis strongly suggests that early referral of patients with PRI to a nephrologist at a multidisciplinary PRI clinic is cost-effective. Across the entire range of possible early referral rates, an incremental increase in the referral rate was associated with an incremental cost saving and a reduction in hospital days, time spent on RRT, and mortality. These findings appear to be robust because they were not reversed by sensitivity analysis using ranges of costs and clinical outcomes from published North American data (where available). The magnitude of cost-effectiveness for early versus late referral is influenced by the difference in the rate of progression in the two cohorts. Early referral remained cost-effective, however, albeit at a lower level, even if we assumed that there was no (renal or patient) survival benefit inherent to patients being followed by a nephrologist at a PRI clinic. We assumed that there was no survival benefit, either before or after starting
Results for Different Rates of Referral to PRI Clinic* Percentage of Patients Referred Early
Outcome Variable
0%
25%
50%
75%
100%
Cost of nephrology care (CDN$) per patient Patient life-years Life-years free of RRT Total hospital inpatient days
164,262 3.36 1.76 41
155,730 3.40 1.87 37
147,328 3.44 1.97 33
139,056 3.48 2.08 29
130,912 3.52 2.18 25
Abbreviations: CDN$, Canadian dollars; PRI, progressive renal insufficiency not yet requiring RRT; RRT, renal replacement therapy. *Using base-case estimates and assuming a 5-year time horizon.
1126
RRT, for patients who are followed at a PRI clinic. The reason for this was the lack of published prospective studies evaluating outcome as a function of late referral. There are, however, many retrospective studies suggesting that patients who are referred late have a poorer outcome on RRT. As such, we may have underestimated the cost-effectiveness of early referral in our analysis. We chose not to include data from these studies because the late referral group may have included patients with acute renal failure and patients with greater comorbidity, both of which may have biased the findings.1-5 There are conflicting data from Canada and the United States regarding the effect of the modality of RRT on patient survival.15,16 We took this into account in our analysis by using the mean survival between studies for the base-case analysis and the individual results for our sensitivity analysis. Varying survival for the different forms of RRT did not influence our findings, suggesting that the survival benefits of the early referral strategy are not due solely to choice of RRT modality. The cost of patient care is influenced heavily, however, by the choice of RRT. The finding of greater reliance on hemodialysis for patients referred late appears to be consistent across studies. Limitations Our study has several limitations. In our Markov model, we assumed that changes between the different health states occurred at a constant rate. A more realistic situation would have been a steadily increasing rate, but there are no published clinical data that allow us to calculate the change in rate with time. Other decision analyses have adopted an all-or-none approach to this problem, whereby no changes in health state occur until a given time point, at which point all patients change.26 This approach also has limitations and is no more realistic than our constant-rate approach. Our constant-rate assumption is also at odds with the observation that renal failure does not progress in all patients and that mortality is higher in the time immediately after starting a new modality of RRT. One way to avoid this problem would have been to use first-order Monte Carlo simulation to step patients one at a time within the model. This approach would have allowed us to use noncon-
MCLAUGHLIN ET AL
stant transitional probabilities between health states. We did not use this technique because we thought it unlikely that this would alter our findings, and it would have added extra complexity to the model, reducing the transparency of the results. The calculation of the rates of deterioration of renal function for early and late referral patients comes from a randomized controlled trial that compared enhanced patient education, within a multidisciplinary approach, with standard nephrology care.8 We considered this an appropriate choice of study because most centers now adopt a multidisciplinary approach to PRI patients. This study may underestimate, however, the benefit of early referral for three reasons. First, we assumed no difference between standard nephrology care and care provided by a nonnephrologist. Second, patients in this study had a mean serum creatinine of 4 mg/dL, which may represent an estimated creatinine clearance significantly lower than 20 mL/min in many patients. As such, the time delay in starting RRT for early referral patients may be longer than 4.6 months when starting at a creatinine clearance of 20 mL/min. Third, this study was published in 1993, and it is likely that the standard of care in multidisciplinary PRI clinics has improved since then. Consequently the difference between the rate of loss of renal function for patients at a PRI clinic and patients not referred to a nephrologist may be greater than we have assumed. Another limitation in our study was lack of published data on which to base our assumptions. Consequently, some of our data were based on single studies or registry data. This fact introduces a greater degree of uncertainty in our analyses. To minimize the bias related to these data, we assumed similar results for both groups, where applicable, and used a wide range (⫾25%) when the confidence intervals were not stated. It is reassuring that the outcomes of our study did not change with sensitivity analyses, and we can be more confident of the results despite the uncertainty in some of the data. Policy Considerations With the assumptions in our analysis, early referral of patients with PRI to a nephrologist at a multidisciplinary PRI clinic appears highly costeffective. Given the predicted rise in the inci-
COST EFFECTIVENESS OF EARLY REFERRAL
dence of end-stage renal disease, along with the burgeoning costs of providing RRT, the arguments for early referral would seem irresistible.21,27-29 We would suggest that referral of patients to a nephrologist with an estimated creatinine clearance of 20 mL/min should not be considered early referral. It is likely that referral of patients who have greater preservation of renal function at the time of referral offers a potentially greater therapeutic advantage and may be more cost-effective. At the same time, early referral of all patients with PRI is probably unattainable because of the insidious and asymptomatic loss of renal function that occurs in many patients. Certain groups are known, however, to be at a much higher risk of developing PRI (eg, patients ⬎50 years old with diabetes or hypertension). Studies that evaluate the cost-effectiveness of screening these high-risk patients and evaluate (true) early referral to a nephrologist are needed. Finally, it is important to consider the impact that an increase in the referral rate to nephrologists would have on the health care system. In the United States, 9.7% of males and 1.8% of females older than age 12 have a serum creatinine greater than 1.5 mg/dL.30 Referral of all patients with an elevated serum creatinine would overload the present model of health care delivery. With the increasing burden of renal disease, new models of health care delivery involving shared care of patients with other physicians and nonphysicians will be required. REFERENCES 1. Ellis PA, Reddy V, Bari N, Cairns HS: Late referral of end-stage renal failure. QJM 91:727-732, 1998 2. Innes A, Rowe PA, Burden RP, Morgan AG: Early deaths on renal replacement therapy: The need for early nephrological referral. Nephrol Dial Transplant 7:467-471, 1992 3. Arora P, Obrador GT, Ruthazer R, Kausz AT, Meyer KB, Jenuleson CS, Pereira BJ: Prevalence, predictors, and consequences of late nephrology referral at a tertiary care center. J Am Soc Nephrol 10:1281-1286, 1999 4. Jungers P, Zingraff J, Page B, Albouze G, Hannedouche T, Man NK: Detrimental effects of late referral in patients with chronic renal failure: A case-control study. Kidney Int 41:170-173, 1993 (suppl) 5. Metcalfe W, Khan IH, Prescott GJ, Simpson K, MacLeod AM: Can we improve early mortality in patients receiving renal replacement therapy? Kidney Int 57:25392545, 2000 6. Giatras I, Lau J, Levey AS: Effect of angiotensinconverting enzyme inhibitors on the progression of nondia-
1127
betic renal disease: A meta-analysis of randomized trials. Angiotensin-Converting-Enzyme Inhibition and Progressive Renal Disease Study Group. Ann Intern Med 127:337-345, 1997 7. Rasgon S, Schwankovsky L, James-Rogers A, Widrow L, Glick J, Butts E: An intervention for employment maintenance among blue-collar workers with end-stage renal disease. Am J Kidney Dis 22:403-412, 1993 8. Binik YM, Devins GM, Barre PE, Guttmann RD, Hollomby DJ, Mandin H, Paul LC, Hons RB, Burgess ED: Live and learn: Patient education delays the need to initiate renal replacement therapy in end-stage renal disease. J Nerv Ment Dis 181:371-376, 1993 9. Levin A, Lewis M, Mortiboy P, Faber S, Hare I, Porter EC, Mendelssohn DC: Multidisciplinary predialysis programs: Quantification and limitations of their impact on patient outcomes in two Canadian settings. Am J Kidney Dis 29:533-540, 1997 10. Schmidt RJ, Domico JR, Sorkin MI, Hobbs G: Early referral and its impact on emergent first dialyses, health care costs, and outcome. Am J Kidney Dis 32:278-283, 1998 11. Canadian Coordinating Office for Health Technology Assessment: Guidelines for Economic Evaluation of Pharmaceuticals: Canada (ed 2). Ottawa, Canada, Canadian Coordinating Office for Health Technology Assessment (CCOHTA), 1997 12. Hakim RM, Lazarus JM: Progression of chronic renal failure. Am J Kidney Dis 14:396-401, 1989 13. Tonelli M, Djurdjev O, Carlisle E: Factors affecting rate of progressive renal decline: Observations in a nonclinical trial setting. J Am Soc Nephrol 9:A80, 1998 (abstract) 14. Churchill DN, Blake PG, Jindal KK, Toffelmire EB, Goldstein MB: Clinical practice guidelines for initiation of dialysis. Canadian Society of Nephrology. J Am Soc Nephrol 10:289-291, 1999 (suppl 13) 15. Murphy SW, Foley RN, Barrett BJ, Murphy SW, Foley RN, Barrett BJ, Kent GM, Morgan J, Barre P, Campbell P, Fine A, Goldstein MB, Handa SP, Jindal KK, Levin A, Mandin H, Muirhead N, Richardson RM, Parfrey PS: Comparative mortality of hemodialysis and peritoneal dialysis in Canada. Kidney Int 57:1720-1726, 2000 16. Bloembergen WE, Port FK, Mauger EA, Wolfe RA: A comparison of cause of death between patients treated with hemodialysis and peritoneal dialysis. J Am Soc Nephrol 6:184-191, 1995 17. Rabbat CG, Thorpe KE, Russell JD, Churchill DN: Comparison of mortality risk for dialysis patients and cadaveric first renal transplant recipients in Ontario, Canada. J Am Soc Nephrol 11:917-922, 2000 18. Wolfe RA, Ashby VB, Milford EL, Ojo AO, Ettenger RE, Agodoa LY, Held PJ, Port FK: Comparison of mortality in all patients on dialysis, patients on dialysis awaiting transplantation, and recipients of a first cadaveric transplant. N Engl J Med 341:1725-1730, 1999 19. Hariharan S, Johnson CP, Bresnahan BA, Taranto SE, McIntosh MJ, Stablein D: Improved graft survival after renal transplantation in the United States, 1988 to 1996. N Engl J Med 342:605-612, 2000 20. Arora P, Kausz AT, Obrador GT, Ruthazer R, Khan S, Jenuleson CS, Meyer KB, Pereira BJ: Hospital utilization
1128
among chronic dialysis patients. J Am Soc Nephrol 11:740746, 2000 21. Goeree R, Manalich J, Grootendorst P, Beecroft ML, Churchill DN: Cost analysis of dialysis treatments for endstage renal disease (ESRD). Clin Invest Med 18:455-464, 1995 22. Laupacis A, Keown P, Pus N, Krueger H, Ferguson B, Wong C, Muirhead N: A study of the quality of life and cost-utility of renal transplantation. Kidney Int 50:235-242, 1996 23. Consumer Price Index for Canada (Health Care [Not Seasonally Adjusted], 1972-1996). Ottawa, Canada, Statistics Canada, Cat no 62-553, 2000 24. Gold MR, Sieleg JE, Russell LB, Weinstein MC: Cost-effectiveness in health and medicine. Oxford, UK, Oxford University Press, 1996. 25. Drummond MF, Jefferson TO: Guidelines for authors and peer reviewers of economic submissions to the BMJ. BMJ 313:275-283, 1996 26. Clark WF, Churchill DN, Forwell L, Macdonald G, Foster S: To pay or not to pay? A decision and cost-utility
MCLAUGHLIN ET AL
analysis of angiotensin-converting-enzyme inhibitor therapy for diabetic nephropathy. Can Med Assoc J 162:195-198, 2000 27. Schaubel DE, Morrison HI, Desmeules M, Parsons DA, Fenton SS: End-stage renal disease in Canada: Prevalence projections to 2005. Can Med Assoc J 160:1557-1563, 1999 28. Mendelssohn DC, Barrett BJ, Brownscombe LM, Ethier J, Greenberg DE, Kanani SD, Levin A, Toffelmire EB: Elevated levels of serum creatinine: Recommendations for management and referral. Can Med Assoc J 161:413417, 1999 29. Anonymous: Morbidity and mortality of renal dialysis: An NIH consensus conference statement. Consensus Development Conference Panel. Ann Intern Med 121:62-70, 1994 30. Jones CA, McQuillan GM, Kusek JW, Eberhardt MS, Herman WH, Coresh J, Salive M, Jones CP, Agodoa LY: Serum creatinine levels in the US population: Third National Health and Nutrition Examination Survey. Am J Kidney Dis 32:992-999, 1998
An Economic Evaluation of Early Versus Late Referral of Patients With Progressive Renal Insufficiency Kevin McLaughlin, MB ChB, Braden Manns, MD, Bruce Culleton, MD, Cam Donaldson, PhD, and Kenneth Taub, MD ● Patients with progressive renal insufficiency (PRI) who start renal replacement therapy (RRT) within 4 months of seeing a nephrologist (late referral) have increased morbidity, mortality, and health care costs. We performed an economic evaluation of early versus late referral of patients with PRI to a multidisciplinary clinic. A decision analysis was performed from the perspective of the health care provider, using a Markov model to simulate progression of PRI and survival of patients on RRT. Our simulated patient cohort comprised 1,000 patients with PRI and estimated creatinine clearance of 20 mL/min. The study time horizon was 5 years. Clinical and cost data were taken from published Canadian and U.S. data, where available. Where published data were lacking, we used data from our prospectively maintained database. The study intervention was attendance at a PRI clinic where patients receive treatment to slow the rate of renal progression, receive treatment of complications of PRI, and are prepared for RRT. Endpoints were total cost of patient care, patient life-years, patient life-years free of RRT, and hospital admission days. Early referral resulted in cost savings and improved patient survival along with more life-years free of RRT and fewer hospital inpatient days. Cost-effectiveness was unaffected by univariate sensitivity analyses. Cost-effectiveness decreased as rates of renal function loss for patients referred early versus late approximated each other. In conclusion, early referral of patients with PRI to a multidisciplinary clinic appears cost-effective. © 2001 by the National Kidney Foundation, Inc. INDEX WORDS: Economic evaluation; progressive renal insufficiency (PRI); referral; end-stage renal disease (ESRD).
Appendix appears only on the web site (www.ajkd.org) electronic pages
A
PPROXIMATELY ONE third of patients with progressive renal insufficiency (PRI) have clinical features of uremia when they first are seen by a nephrologist and need to start renal replacement therapy (RRT) within 4 months of their first consultation.1 Patients referred to nephrologists at a late stage in the course of PRI appear to have increased morbidity and mortality compared with patients referred early.2-5 The potential benefits of early referral of patients From the Departments of Medicine, Community Health Sciences, and Economics, University of Calgary, Alberta, Canada. Received April 9, 2001, and accepted as submitted May 25, 2001. Supported in part by a Kidney Foundation of Canada start-up grant (K.M). Address reprint requests to Kevin McLaughlin, MB ChB, Department of Medicine, Division of Nephrology, Foothills Hospital & University of Calgary, 1403 29th Street NW, Calgary, Alberta T2N 2T9 Canada. E-mail: [email protected] © 2001 by the National Kidney Foundation, Inc. 0272-6386/01/3805-0026$35.00/0 doi:10.1053/ajkd.2001.28619 1122
with PRI to a nephrologist at a multidisciplinary PRI clinic include identifying and treating reversible causes of renal failure; slowing the rate of decline of renal function; managing the multiple coexisting conditions associated with PRI; and optimizing the biochemical, physical, and psychological state of the patient at the time of starting RRT.6-8 The increased morbidity of patients referred to a nephrologist at a late stage is likely to lead to greater usage of health care resources. It has been suggested that early referral would produce a cost saving in addition to the health benefits to the patient.9,10 This theory remains unproven, however. We performed an economic evaluation using decision analytic modeling to compare early versus late referral of patients with PRI to a nephrologist at a multidisciplinary PRI clinic. METHODS
Decision Analytic Model We constructed a decision analysis tree using ithink version 6.0 (High Performance Systems, Inc, Hanover, NH) to evaluate the costs and effects of two different clinical strategies for managing patients with PRI (Fig 1). The two clinical strategies were refer patients early (ie, patients are referred to a nephrologist at a multidisciplinary PRI clinic when their creatinine clearance is 20 mL/min) and refer patients late (ie, patients are referred when they are uremic
American Journal of Kidney Diseases, Vol 38, No 5 (November), 2001: pp 1122-1128
COST EFFECTIVENESS OF EARLY REFERRAL
Fig 1. Clinical strategies and health states in the Markov model. Abbreviations: RRT, renal replacement therapy; PRI, progressive renal insufficiency not yet requiring RRT; HD, hemodialysis; PD, peritoneal dialysis.
and require RRT). The intervention in this model is referral to a multidisciplinary team comprising a nephrologist, nurse, dietitian, social worker, and pharmacist, who provide comprehensive patient care. Four outcomes were considered: (1) average total cost of patient care, (2) patient life-years, (3) patient life-years free of RRT, and (4) hospital inpatient days. We used a simulated patient cohort of 1,000 average patients who had a mean estimated creatinine clearance of 20 mL/min. Subsequent to reaching this level of renal function, a patient can enter into one of five health states: (1) PRI not yet requiring RRT, (2) hemodialysis, (3) peritoneal dialysis, (4) renal transplantation, and (5) death. A Markov process was used to model transitions between these states on a monthly basis over a period of 5 years. The transition probabilities are outlined in Table 1.
General Assumptions of the Markov Model In our model, the following assumptions have been made: PRI progresses in all patients; there is a constant proportion of patients starting RRT each month (different for each strategy); there is a constant rate of hospital admission in the two limbs (different for each strategy); there is a constant proportion of patients switching between different modalities of RRT each month (equal for each strategy in the base-case analysis); and there is a constant proportion of patients dying each month depending on the health state (equal for each strategy in the base-case analysis). A discount rate of 5% (recommended rates for Canada) for cost and effect was used for the base-case analysis.11 It also was assumed that patients who were referred early continued to see their family physician (shared patient care).
Data Sources Published Canadian and U.S. data were taken where available to estimate base-case values and to provide ranges for sensitivity analyses. If no suitable published data were available, mean data over a 3-year period (March 31, 1997 through March 3, 2000) were taken from the prospectively maintained database at the Southern Alberta Renal Program (SARP). For patients followed by a nephrologist at a PRI clinic, the estimated rate of loss of
1123
residual renal function was 4.8 to 6.4 mL/min/y.12,13 Taking a mean of 5.6 mL/min/y (range, 4.8 to 6.4 mL/min/ y), from a baseline creatinine clearance of 20 mL/min/y, the estimated time to starting RRT (creatinine clearance, 12 mL/min) for this cohort was 17 months.14 It has been shown that patients followed by a nephrologist at a multidisciplinary PRI clinic may have an average extended duration of time without dialysis of 4.6 months compared with patients managed by standard care.8 Using this figure to estimate the time to RRT for patients not followed by a nephrologist at a PRI clinic (12.4 months), we estimated the median rate of loss of renal function in the late referral cohort to be 7.7 mL/min/y (range, 6.2 to 9.2 mL/min/y). These rates of deterioration corresponded to a constant proportion of patients starting RRT each month of 2.9% (2.5% to 3.3%) and 4% (3.3% to 4.8%). Data on choice of RRT depending on time of referral were taken from published Canadian and U.S. data, as were data on mortality rates for patients starting RRT with hemodialysis and peritoneal dialysis.3,9,15,16 When more than one study was available, the mean was taken for the base-case analysis (Table 1). Survival on transplantation and transition probabilities between RRT modality were taken from published Canadian and U.S. studies.15,17-19 Hospitalization rates for early and late referred patients over a 5-year period were taken from a U.S. study.20 Use of erythropoietin was taken from this study and from a personal communication (A. Levin).20 Mortality rate at PRI clinic and the preemptive transplant rate for early and late referred patients were taken from the SARP database.
Costs We performed the analysis from the perspective of a health care provider. The cost of outpatient care for patients in each of the health states was estimated from published Canadian data.21,22 Treatment costs were converted to 1999 Canadian dollars (1 CDN$ ⫽ 0.67 US$) using the Consumer Price Index for Canada from the years 1995 to 1999.23 These values were used for the base-case analysis. The cost for PRI care was taken from the SARP database and included total staffing costs (including physician costs), support services, supplies, and operational costs. Erythropoietin cost (Alberta, 1999) for patients with PRI was based on a mean dose of 4,000 U/wk (A. Levin, personal communication). The cost per day of inpatient care was estimated from the mean inpatient cost for all dialysis patients at Foothills hospital in 1999.
Sensitivity Analyses We performed sensitivity analyses for clinical variables, transition probabilities, costs, and discount rates (for costs and effects). When more than one study was available, the individual studies were used to provide the range for sensitivity analysis. When the 95% confidence intervals for cost were reported, these were used for sensitivity analysis.21 If confidence intervals were unavailable, a plausible range for cost (⫾25%) was used. For data that were derived from a single study or from the SARP database, a plausible range (⫾25%) was used for sensitivity analysis. To provide the range for discount rates for costs and effects, we used the
1124
MCLAUGHLIN ET AL Table 1.
Base-Case Clinical Values and Costs With Ranges for Sensitivity Analyses
Variable
Base-Case Value
Range in Sensitivity Analysis
EPREX use at PRI clinic (% patients) EPREX use at non-PRI (% patients) EPREX costs (CDN$/y)† PRI clinic costs (CDN$/y)§ HD costs (CDN$/y) PD costs (CDN$/y) Tx costs (CDN$/y) Admission costs (CDN$/d)§ PRI patient GFR loss (mL/min/y) Non-PRI patient GFR loss (mL/min/y) Start HD from PRI clinic (%) Start HD from non-PRI clinic (%) Start PD from PRI clinic (%) Start PD from non-PRI clinic (%) Start Tx from PRI clinic (%)§ Start Tx from non-PRI clinic (%)§ Transition from HD to Tx (%/y) Transition from HD to PD (%/y) Transition from PD to Tx (%/y) Transition from PD to HD (%/y) Transition from Tx to HD (%/y)¶ Transition from Tx to PD (%/y)¶ Death before starting RRT (%/y)§ Death on HD (%/y) Death on PD (%/y) Death on Tx (%/y) Hospitalization rate for PRI patients Hospitalization rate for non-PRI patients Discount rate for cost (%) Discount rate for effect (%)
55 9 3,240 1,366 94,198 47,628 37,106 792 5.6 7.7 57 76 32 24 11 0 10.35 12.7 15.4 15.95 1.03 1.03 5 22.2 21 2.45 13.6 19.6 5 5
40-70 0-17 2,340-4,050‡ 1,024-1,708‡ 87,017-101,381 42,216-53,059 27,830-46,383‡ 437-1,147 4.8-6.4 6.2-9.2 42-73 58-93 16-47 7-42 8.25-13.75‡ 0‡ 7.76-12.94‡ 9.52-15.88‡ 11.55-19.25‡ 11.96-19.94‡ 0.77-1.29‡ 0.77-1.29‡ 3.75-6.25‡ 21.3-23 16.6-25.3 1.1-3.8 10.2-17‡ 14.7-24.5‡ 3-6 3-5
References
3, * 3, *
20 20 21 8, 11, 12 8, 11, 12 3, 9 3, 9 3, 9 3, 9
14 14 14 14 18 18 14, 15 14, 15 16, 17, 18 19 19 23, 24, 25 23, 24, 25
NOTE. ⫾95% Confidence intervals used for sensitivity analysis. Abbreviations: CDN$, Canadian dollars; GFR, glomerular filtration rate; RRT, renal replacement therapy; PRI, progressive renal insufficiency not yet requiring RRT; HD, hemodialysis; PD, peritoneal dialysis; Tx, renal transplantation. *A. Levin: Personal communication. †Cost based on 4,000 units per week (A. Levin, personal communication). ‡⫾25% used for sensitivity analysis. §SARP database. ¶Assumes living-related transplant rate of 29.2% and equal proportions returning to PD and HD.
recommended rates for the United States and United Kingdom.11,24,25 Table 1 shows the ranges used in the sensitivity analyses along with the data source.
RESULTS
Base-Case Analysis Using the early referral strategy, the mean total cost per patient for a 5-year period was CDN$ 130,912 (US$ 87,711). The mean number of life-years per patient during this period was 3.52. Corresponding values for the late referral strategy were CDN$ 164,262 (US$ 110,056) and 3.36. The incremental cost-effective ratio for the
early versus late referral strategy was CDN$ ⫺208,438 (US$ ⫺139,653) per life-year gained. There is a cost saving and an improved patient survival associated with the early referral strategy. There also are more life-years free of RRT and fewer hospital inpatient days for the early referral strategy (Table 2). Sensitivity Analyses The cost-effectiveness of the strategy remained unaffected by univariate sensitivity analysis using the variables and ranges shown in Table 1. The cost-effectiveness of early referral
COST EFFECTIVENESS OF EARLY REFERRAL Table 2.
1125
Results of Base-Case Analysis for Early Versus Late Referral Strategies
Outcome Variable
Early Referral
Late Referral
Cost of nephrology care (CDN$) Patient life-years Life-years free of RRT Total hospital inpatient days
130,912 3.52 2.18 25
164,262 3.36 1.76 41
Incremental Cost-Effective Ratio (CDN$/Life-Year Gained)
⫺208,438
Abbreviations: CDN$, Canadian dollars; RRT, renal replacement therapy.
decreased as the rate of deterioration in renal function for patients referred early or late approached each other. When the rates were equal, there was, by definition, equal life-years off dialysis (2.01 years). There was, however, still a difference in overall patient survival (3.49 years versus 3.42 years), total costs (CDN$ 139,155 [U.S. $93,234] versus CDN$ 150,416 [U.S. $100,779]) and hospital days (27 days versus 37 days) in favor of early referral. The difference in mortality in this situation is largely due to the increased rate of preemptive transplantation in the early referral group. The cost difference is a result of early referral patients choosing less expensive forms of RRT (ie, peritoneal dialysis and preemptive transplantation). We examined the effect of varying the early referral rate (Table 3). These analyses were included to reflect the actual patterns of referral in North America (ie, a varying ratio of early-tolate referrals with no unit having all early or all late referrals). The results show a predictable increase in patient life-years, an increase in patient life-years off RRT, a decrease in the number of hospital inpatient days, and a decrease in total costs for patient care for each incremental increase in the early referral rate. Increasing the early referral rate from 50% to 75% would result Table 3.
in a cost saving of CDN$ 8,272 (U.S. $5,542) per patient over the 5-year period with a gain in life-years of 0.04 years, a gain in life-years off RRT of 0.11 years, and 4 fewer hospital inpatient days. DISCUSSION
This analysis strongly suggests that early referral of patients with PRI to a nephrologist at a multidisciplinary PRI clinic is cost-effective. Across the entire range of possible early referral rates, an incremental increase in the referral rate was associated with an incremental cost saving and a reduction in hospital days, time spent on RRT, and mortality. These findings appear to be robust because they were not reversed by sensitivity analysis using ranges of costs and clinical outcomes from published North American data (where available). The magnitude of cost-effectiveness for early versus late referral is influenced by the difference in the rate of progression in the two cohorts. Early referral remained cost-effective, however, albeit at a lower level, even if we assumed that there was no (renal or patient) survival benefit inherent to patients being followed by a nephrologist at a PRI clinic. We assumed that there was no survival benefit, either before or after starting
Results for Different Rates of Referral to PRI Clinic* Percentage of Patients Referred Early
Outcome Variable
0%
25%
50%
75%
100%
Cost of nephrology care (CDN$) per patient Patient life-years Life-years free of RRT Total hospital inpatient days
164,262 3.36 1.76 41
155,730 3.40 1.87 37
147,328 3.44 1.97 33
139,056 3.48 2.08 29
130,912 3.52 2.18 25
Abbreviations: CDN$, Canadian dollars; PRI, progressive renal insufficiency not yet requiring RRT; RRT, renal replacement therapy. *Using base-case estimates and assuming a 5-year time horizon.
1126
RRT, for patients who are followed at a PRI clinic. The reason for this was the lack of published prospective studies evaluating outcome as a function of late referral. There are, however, many retrospective studies suggesting that patients who are referred late have a poorer outcome on RRT. As such, we may have underestimated the cost-effectiveness of early referral in our analysis. We chose not to include data from these studies because the late referral group may have included patients with acute renal failure and patients with greater comorbidity, both of which may have biased the findings.1-5 There are conflicting data from Canada and the United States regarding the effect of the modality of RRT on patient survival.15,16 We took this into account in our analysis by using the mean survival between studies for the base-case analysis and the individual results for our sensitivity analysis. Varying survival for the different forms of RRT did not influence our findings, suggesting that the survival benefits of the early referral strategy are not due solely to choice of RRT modality. The cost of patient care is influenced heavily, however, by the choice of RRT. The finding of greater reliance on hemodialysis for patients referred late appears to be consistent across studies. Limitations Our study has several limitations. In our Markov model, we assumed that changes between the different health states occurred at a constant rate. A more realistic situation would have been a steadily increasing rate, but there are no published clinical data that allow us to calculate the change in rate with time. Other decision analyses have adopted an all-or-none approach to this problem, whereby no changes in health state occur until a given time point, at which point all patients change.26 This approach also has limitations and is no more realistic than our constant-rate approach. Our constant-rate assumption is also at odds with the observation that renal failure does not progress in all patients and that mortality is higher in the time immediately after starting a new modality of RRT. One way to avoid this problem would have been to use first-order Monte Carlo simulation to step patients one at a time within the model. This approach would have allowed us to use noncon-
MCLAUGHLIN ET AL
stant transitional probabilities between health states. We did not use this technique because we thought it unlikely that this would alter our findings, and it would have added extra complexity to the model, reducing the transparency of the results. The calculation of the rates of deterioration of renal function for early and late referral patients comes from a randomized controlled trial that compared enhanced patient education, within a multidisciplinary approach, with standard nephrology care.8 We considered this an appropriate choice of study because most centers now adopt a multidisciplinary approach to PRI patients. This study may underestimate, however, the benefit of early referral for three reasons. First, we assumed no difference between standard nephrology care and care provided by a nonnephrologist. Second, patients in this study had a mean serum creatinine of 4 mg/dL, which may represent an estimated creatinine clearance significantly lower than 20 mL/min in many patients. As such, the time delay in starting RRT for early referral patients may be longer than 4.6 months when starting at a creatinine clearance of 20 mL/min. Third, this study was published in 1993, and it is likely that the standard of care in multidisciplinary PRI clinics has improved since then. Consequently the difference between the rate of loss of renal function for patients at a PRI clinic and patients not referred to a nephrologist may be greater than we have assumed. Another limitation in our study was lack of published data on which to base our assumptions. Consequently, some of our data were based on single studies or registry data. This fact introduces a greater degree of uncertainty in our analyses. To minimize the bias related to these data, we assumed similar results for both groups, where applicable, and used a wide range (⫾25%) when the confidence intervals were not stated. It is reassuring that the outcomes of our study did not change with sensitivity analyses, and we can be more confident of the results despite the uncertainty in some of the data. Policy Considerations With the assumptions in our analysis, early referral of patients with PRI to a nephrologist at a multidisciplinary PRI clinic appears highly costeffective. Given the predicted rise in the inci-
COST EFFECTIVENESS OF EARLY REFERRAL
dence of end-stage renal disease, along with the burgeoning costs of providing RRT, the arguments for early referral would seem irresistible.21,27-29 We would suggest that referral of patients to a nephrologist with an estimated creatinine clearance of 20 mL/min should not be considered early referral. It is likely that referral of patients who have greater preservation of renal function at the time of referral offers a potentially greater therapeutic advantage and may be more cost-effective. At the same time, early referral of all patients with PRI is probably unattainable because of the insidious and asymptomatic loss of renal function that occurs in many patients. Certain groups are known, however, to be at a much higher risk of developing PRI (eg, patients ⬎50 years old with diabetes or hypertension). Studies that evaluate the cost-effectiveness of screening these high-risk patients and evaluate (true) early referral to a nephrologist are needed. Finally, it is important to consider the impact that an increase in the referral rate to nephrologists would have on the health care system. In the United States, 9.7% of males and 1.8% of females older than age 12 have a serum creatinine greater than 1.5 mg/dL.30 Referral of all patients with an elevated serum creatinine would overload the present model of health care delivery. With the increasing burden of renal disease, new models of health care delivery involving shared care of patients with other physicians and nonphysicians will be required. REFERENCES 1. Ellis PA, Reddy V, Bari N, Cairns HS: Late referral of end-stage renal failure. QJM 91:727-732, 1998 2. Innes A, Rowe PA, Burden RP, Morgan AG: Early deaths on renal replacement therapy: The need for early nephrological referral. Nephrol Dial Transplant 7:467-471, 1992 3. Arora P, Obrador GT, Ruthazer R, Kausz AT, Meyer KB, Jenuleson CS, Pereira BJ: Prevalence, predictors, and consequences of late nephrology referral at a tertiary care center. J Am Soc Nephrol 10:1281-1286, 1999 4. Jungers P, Zingraff J, Page B, Albouze G, Hannedouche T, Man NK: Detrimental effects of late referral in patients with chronic renal failure: A case-control study. Kidney Int 41:170-173, 1993 (suppl) 5. Metcalfe W, Khan IH, Prescott GJ, Simpson K, MacLeod AM: Can we improve early mortality in patients receiving renal replacement therapy? Kidney Int 57:25392545, 2000 6. Giatras I, Lau J, Levey AS: Effect of angiotensinconverting enzyme inhibitors on the progression of nondia-
1127
betic renal disease: A meta-analysis of randomized trials. Angiotensin-Converting-Enzyme Inhibition and Progressive Renal Disease Study Group. Ann Intern Med 127:337-345, 1997 7. Rasgon S, Schwankovsky L, James-Rogers A, Widrow L, Glick J, Butts E: An intervention for employment maintenance among blue-collar workers with end-stage renal disease. Am J Kidney Dis 22:403-412, 1993 8. Binik YM, Devins GM, Barre PE, Guttmann RD, Hollomby DJ, Mandin H, Paul LC, Hons RB, Burgess ED: Live and learn: Patient education delays the need to initiate renal replacement therapy in end-stage renal disease. J Nerv Ment Dis 181:371-376, 1993 9. Levin A, Lewis M, Mortiboy P, Faber S, Hare I, Porter EC, Mendelssohn DC: Multidisciplinary predialysis programs: Quantification and limitations of their impact on patient outcomes in two Canadian settings. Am J Kidney Dis 29:533-540, 1997 10. Schmidt RJ, Domico JR, Sorkin MI, Hobbs G: Early referral and its impact on emergent first dialyses, health care costs, and outcome. Am J Kidney Dis 32:278-283, 1998 11. Canadian Coordinating Office for Health Technology Assessment: Guidelines for Economic Evaluation of Pharmaceuticals: Canada (ed 2). Ottawa, Canada, Canadian Coordinating Office for Health Technology Assessment (CCOHTA), 1997 12. Hakim RM, Lazarus JM: Progression of chronic renal failure. Am J Kidney Dis 14:396-401, 1989 13. Tonelli M, Djurdjev O, Carlisle E: Factors affecting rate of progressive renal decline: Observations in a nonclinical trial setting. J Am Soc Nephrol 9:A80, 1998 (abstract) 14. Churchill DN, Blake PG, Jindal KK, Toffelmire EB, Goldstein MB: Clinical practice guidelines for initiation of dialysis. Canadian Society of Nephrology. J Am Soc Nephrol 10:289-291, 1999 (suppl 13) 15. Murphy SW, Foley RN, Barrett BJ, Murphy SW, Foley RN, Barrett BJ, Kent GM, Morgan J, Barre P, Campbell P, Fine A, Goldstein MB, Handa SP, Jindal KK, Levin A, Mandin H, Muirhead N, Richardson RM, Parfrey PS: Comparative mortality of hemodialysis and peritoneal dialysis in Canada. Kidney Int 57:1720-1726, 2000 16. Bloembergen WE, Port FK, Mauger EA, Wolfe RA: A comparison of cause of death between patients treated with hemodialysis and peritoneal dialysis. J Am Soc Nephrol 6:184-191, 1995 17. Rabbat CG, Thorpe KE, Russell JD, Churchill DN: Comparison of mortality risk for dialysis patients and cadaveric first renal transplant recipients in Ontario, Canada. J Am Soc Nephrol 11:917-922, 2000 18. Wolfe RA, Ashby VB, Milford EL, Ojo AO, Ettenger RE, Agodoa LY, Held PJ, Port FK: Comparison of mortality in all patients on dialysis, patients on dialysis awaiting transplantation, and recipients of a first cadaveric transplant. N Engl J Med 341:1725-1730, 1999 19. Hariharan S, Johnson CP, Bresnahan BA, Taranto SE, McIntosh MJ, Stablein D: Improved graft survival after renal transplantation in the United States, 1988 to 1996. N Engl J Med 342:605-612, 2000 20. Arora P, Kausz AT, Obrador GT, Ruthazer R, Khan S, Jenuleson CS, Meyer KB, Pereira BJ: Hospital utilization
1128
among chronic dialysis patients. J Am Soc Nephrol 11:740746, 2000 21. Goeree R, Manalich J, Grootendorst P, Beecroft ML, Churchill DN: Cost analysis of dialysis treatments for endstage renal disease (ESRD). Clin Invest Med 18:455-464, 1995 22. Laupacis A, Keown P, Pus N, Krueger H, Ferguson B, Wong C, Muirhead N: A study of the quality of life and cost-utility of renal transplantation. Kidney Int 50:235-242, 1996 23. Consumer Price Index for Canada (Health Care [Not Seasonally Adjusted], 1972-1996). Ottawa, Canada, Statistics Canada, Cat no 62-553, 2000 24. Gold MR, Sieleg JE, Russell LB, Weinstein MC: Cost-effectiveness in health and medicine. Oxford, UK, Oxford University Press, 1996. 25. Drummond MF, Jefferson TO: Guidelines for authors and peer reviewers of economic submissions to the BMJ. BMJ 313:275-283, 1996 26. Clark WF, Churchill DN, Forwell L, Macdonald G, Foster S: To pay or not to pay? A decision and cost-utility
MCLAUGHLIN ET AL
analysis of angiotensin-converting-enzyme inhibitor therapy for diabetic nephropathy. Can Med Assoc J 162:195-198, 2000 27. Schaubel DE, Morrison HI, Desmeules M, Parsons DA, Fenton SS: End-stage renal disease in Canada: Prevalence projections to 2005. Can Med Assoc J 160:1557-1563, 1999 28. Mendelssohn DC, Barrett BJ, Brownscombe LM, Ethier J, Greenberg DE, Kanani SD, Levin A, Toffelmire EB: Elevated levels of serum creatinine: Recommendations for management and referral. Can Med Assoc J 161:413417, 1999 29. Anonymous: Morbidity and mortality of renal dialysis: An NIH consensus conference statement. Consensus Development Conference Panel. Ann Intern Med 121:62-70, 1994 30. Jones CA, McQuillan GM, Kusek JW, Eberhardt MS, Herman WH, Coresh J, Salive M, Jones CP, Agodoa LY: Serum creatinine levels in the US population: Third National Health and Nutrition Examination Survey. Am J Kidney Dis 32:992-999, 1998