Cumulative Risk for Developing End-Stage Renal Disease in the US Population
Bryce A. Kiberd* and
Catherine M. Clase
*Departments of Medicine, Dalhousie University, Halifax, Nova Scotia; and McMaster University, Hamilton, Ontario, Canada
Correspondence to: Dr. Bryce A. Kiberd, 5077 AC Dickson Building, Queen Elizabeth II Health Sciences Center, Halifax, Nova Scotia, Canada B3H 2Y9. Phone: 902-473-7058; Fax: 902-473-2675; E-mail: bkiberd{at}is.dal.ca
ABSTRACT. The individual risk of developing end-stage renaldisease (ESRD) and its overall impact on life expectancy isnot known. This study’s objectives were to determine theeffect of ESRD on life expectancy for a cohort of 20-yr-oldsand to compare this impact to that of several cancers for whichpopulation-based screening programs exist. A computer simulation,stratified by race (white, black) and by gender was used tocalculate cumulative lifetime risk of ESRD, life-years lostto ESRD, and cumulative Medicare payments for ESRD. Similarcalculations were made for breast, prostate, and colorectalcancer. The cumulative lifetime risk of ESRD for a 20-yr-oldblack woman is 7.8%. Equivalent risks for black men are 7.3%,white men 2.5%, and white women 1.8%. Lost years of life attributableto ESRD are 1.09, 1.10, 0.40, and 0.32 yr for black women, blackmen, white men, and white women, respectively. In blacks, ESRDis responsible for nearly as much loss of life-years as breastcancer in women and more loss of life-years than colorectalor prostate cancer in men. In addition, treatment costs forESRD in this population are many-fold more expensive than cumulativetreatment costs of these cancers. Exploring new screening andtreatment strategies may be warranted to prevent ESRD, particularlyin the US black population.
In the United States, 86,825 patients began treatment for end-stagerenal disease (ESRD) in 1998, and there were 326,217 prevalentpatients. With the current rate of growth, there will be 172,667incident and 661,330 prevalent patients by the year 2010 (1).The Medicare costs for care of ESRD will increase from $12 billionto $28.3 billion over this time period (1). However, survivalon ESRD is relatively short; therefore, incidence and prevalencedata do not capture the full impact that the disease has onoverall health in the population. Estimates of average yearsof life lost to ESRD and cumulative lifetime risks and treatmentcosts for a cohort of young individuals would better capturethe societal, individual, and economic impact of this problem.Similarly, comparing these risks to those associated with commonmalignancies for which established screening programs existwould help relate the importance of ESRD to overall health.
The life expectancy of 20-yr-old men and women of black or whiterace was calculated using a Markov model in a commercial softwarepackage (Decision Analysis by TreeAge [DATA] 3.5, TreeAge, Williamstown,MA) (Figure 1). Annual (1998) death rates were obtained fromthe Report of the United States Department of Vital Statistics(2). Few individuals survive beyond 100 yr; therefore, the analysiswas truncated at this age. Cumulative risk of ESRD was thencalculated by the Markov analysis output, (see Figure 1, uppertree) created in TreeAge using the following methodology inExcel 4.0 (Microsoft, Redmond, WA). The proportion of the populationat each time interval x, is at risk of developing ESRD. Theproportion developing ESRD can be calculated by multiplyingthe proportion of the population surviving to interval x bythe annual probability of developing ESRD, denoted here as pESRDx.The probability, pESRDx, was calculated from the age-, gender-,and race-specific ESRD incidence rate per million population(rESRDx), taken from the United States Renal Data System 2000Annual Report (see Appendix for input values), by the formulaused to convert rates to probabilities (3,4),
Figure 1. Data 3.5 (TreeAge) medical decision analysis software was used to estimate life expectancy for a 20-yr-old black male from general population mortality rates (upper tree). This upper tree was used to calculate (see text) the cumulative risk of disease (end-stage renal disease [ESRD] or cancer). The lower tree was used to create a life table that included an estimate of patients with ESRD (or cancer) at each time interval. This information was used to calculate the excess death from disease (ESRD or cancer) at each interval for adjustments in a disease elimination mortality rate calculation.
The proportion of the population developing new-onset ESRD foreach interval (x to x + 1) was summed over the cohort lifetime(truncated at 100 yr) to estimate the cumulative lifetime riskof ESRD.
To calculate the years of life lost for a disease, a diseaseelimination strategy was performed (3). This estimates the increasein life expectancy if ESRD is eliminated from the general populationand takes into account death from competing diseases. Figure 1also shows the model created in TreeAge. The Markov analysisoutput (life table) from TreeAge was exported to an Excel spreadsheet.The total mortality, dx, during the interval x to x + 1 is calculatedas the reduction in population over the 1-yr interval. Thisincludes patients with ESRD. To eliminate those who die dueto the excess risks associated with ESRD, several steps arerequired.
First the excess mortality or disease-specific mortality dueto ESRD is determined by subtracting the general populationmortality rate from the observed mortality rate with ESRD (forall modalities, including dialysis and transplantation) (5).
(2)
Disease-specific ESRD mortality rate is converted to a probabilityby an equation similar to Equation 1. This disease-specificprobability is then multiplied by the proportion of patientswith ESRD from the life table to calculate excess ESRD death(dESRDx) for the interval x.
To calculate the new population mortality rate with ESRD eliminated,Qx, the formula used is,
(3)
Where qx is the mortality probability for interval x with ESRDincluded, dx is the total deaths in the interval x, and dESRDxis excess deaths from ESRD. The new life expectancy for thefour (black/white, male/female) cohorts, starting at age 20,are then recalculated using the new probabilities (Qx) in themodel. Loss of life-years was calculated by subtracting lifeexpectancy calculated with published general mortality ratesfrom the life expectancy calculated with the new ESRD eliminatedrates (Figure 2).
Figure 2. Data 3.5 (TreeAge) medical decision analysis software was used to estimate life expectancies derived from general population mortality rates (upper tree) and from disease-specific (ESRD or cancer) death eliminated mortality rates (lower tree). Here, an example for black males is shown.
For comparative purposes, we performed a similar analysis forbreast cancer in women and for prostate and colorectal cancerin men. Incidence rates for malignancy (breast, prostate, andcolorectal) were taken from Surveillance, Epidemiology and EndResults (SEER) Cancer Statistics Review, 1973–1998 (6).Five-year observed survival data (gender-, age-, and race-specific)from this registry (1992 to 1998) were converted to annual mortalityprobabilities by the declining exponential approximation oflife expectancy method (5). As above, the model calculated thecumulative risk and reduced life expectancy of cancer.
Data from this model were also used to calculate the cumulativecosts from a third party payer’s perspective for ESRDusing Medicare payments as reported by the USRDS Annual report(1). Lifetime direct treatment costs for the three malignancieswere taken from the literature (7). These costs were convertedto 1998 US$ by the medical component of the consumer price indexfrom the original 1992 US$ values (8). This source of cancertreatment cost information has been used in recent publishedcost-effectiveness analyses (9,10), provides cost estimatesthat are similar to those reported elsewhere (11,12), and providesmore complete follow-up costs than other sources (13,14). Mostof these cost derivations are based on Medicare cost estimatesand are therefore comparable with ESRD Medicare treatment costsin terms of the third party payer’s perspective (12,13,14).
In a sensitivity analysis, we explored the uncertainty of thesevalues on our conclusions. The model incorporated incidencerates, mortality rates, and cancer costs that were ±2 standard errors. The standard errors for ESRD incidence andmortality rates were taken directly from USRDS registry (1).The SEER*Stat program was used to calculate standard errorsfor 1-yr cancer mortality rates and incidence rates from the1992 through 1998 submission (6). Cumulative risks, years oflife lost, and costs require the summation of 80 intervals (age,20 to 100); therefore, the reported upper and lower bounds aremuch broader than a 95% confidence interval (3). Future costs,and to the same degree future outcomes, should be discounted;however, the recommended base discount rate differs among experts(15). Therefore, the results are presented as undiscounted (0%)and discounted at 3%, 5%, and 7%.
This model assumes that incidence and mortality rates for thegeneral population and diseases will not change over the next80 yr. There will be inevitable growth in the rates of ESRDand cancer along with an aging population. We did not modelgrowth due to the speculative nature of these projections. Webelieve that even a conservative analysis such as this willbe informative.
Cumulative risk and lost years of life and for each cohort areshown in Tables 1 and 2. Figure 3 graphically depicts the cumulativerisk of ESRD. By age 56, the cumulative risk of ESRD in blackmen and women already exceeds the lifetime risk of ESRD in whitemen and women, respectively. The loss of life due to ESRD forwhites is small compared with that for blacks. The lost life-yearsfrom ESRD in black women is 2.9 (1.09/0.32) times greater thanthat of white women. The lost life-years from ESRD in blackmen is 2.75 (1.10/0.40) times greater than that of white men.
Table 2. Reduction in life expectancy (years of life lost) for ESRD and common malignanciesa
In our model of breast cancer, loss of 1.25 yr and 0.98 yr oflife in black and white women is respectively predicted (Table 2).The impact of ESRD in black women is comparable with thatof breast cancer (1.25 ± 0.41 lost years of life frombreast cancer versus 1.09 ± 0.08 from ESRD). The largerange in the estimate for lost life from breast cancer is dueto the high standard errors for mortality in black women withbreast cancer (Table 2). In white women, the impact of breastcancer is 3.1 (0.98/0.32) times greater than that of ESRD. Forboth groups, the modeled cumulative risk for breast cancer isgreater than the risk for ESRD.
Tables 1 and 2 also show the comparable data for prostate andcolorectal cancer in men. Though cumulative risk for prostatecancer is very high in both white and black men, its impactassessed as years of life lost to prostate cancer is less thanthat attributable to ESRD in both groups. In black men, colorectalcancer is responsible for slightly more than half the loss ofyears of life expected from ESRD; in white men, lost years dueto colorectal cancer exceed those lost to ESRD.
Cancer and ESRD may occur at different ages; therefore, we examinedthe impact of discount rates up to 7%, to weight early outcomesmore heavily than distant outcomes. The relative importanceof ESRD and breast cancer in white and black women was relativelystable over the range of discount rates (Table 3). In blackmen, ESRD is responsible for 1.2 times more undiscounted lostyears of life than that attributable to prostate cancer. Thisincreases to a ratio of 2.75 with a discount rate of 7%. Theeffect of ESRD on loss of life in white men is 1.42 to 2.8 timesgreater than prostate cancer over the range of discount rates.
Table 3. Relative loss of life from ESRD compared with cancera
To estimate of costs of healthcare resource use for ESRD, wecalculated the cumulative Medicare payments, discounted andundiscounted, for maintenance therapy per US citizen on thebasis of the model’s predictions of lifetime ESRD probability.In discounted models, the cumulative costs for whites are relativelymodest when calculated for 20-yr olds, whereas the costs perblack citizen are substantially higher (Table 4). Differencesbetween whites and blacks, and absolute costs, are all morepronounced at lower discount rates.
Table 4. Cumulative Medicare payments (1998 US dollars) for ESRD per US citizen
Table 5 shows the relative cumulative treatment costs of ESRDrelative to the cancers for each of the four cohorts adjustedfor discount rates and over the mean ± 2 standard errorsfor the cancer treatment costs. For black women and men, cumulativeESRD costs were several to many-fold higher than breast, colorectal,and prostate cancer. The cumulative costs of breast cancer treatmentin white women were the only cancer to exceed the cumulativecosts of ESRD therapy.
The ESRD risk, assuming no future increase in ESRD incidencerates, is 1 in 40 for white men and nearly 1 in 50 for whitewomen. For black men and women, the cumulative risk of ESRDis nearly 1 in 12. We are not aware of any previously publishedestimates of the cumulative risk of ESRD. We suspect that theabove estimates of cumulative risk are much higher than appreciatedby most nephrologists, and this risk will increase in the future(1).
This study also shows that white men and women are at a comparativelyhigher risk of cancer in this analysis than ESRD. Colorectaland breast cancer also cause more loss of life than ESRD forwhite men and women, respectively. On the other hand the cumulativeincidence of ESRD is greater than colorectal cancer for blackmen. ESRD also causes more loss of life than either colorectalor prostate cancer in black men and nearly as much loss of lifeas breast cancer in black women.
The incidence rates for both cancer and ESRD increase in theolder persons but not always at the same rate. Diseases thatdevelop early may have a greater impact on potential life lostthan more prevalent diseases in the aged. To account for theimpact of early events relative to late events, the effectsof various discount rates were examined. Prostate cancer isa good example of a very common malignancy in white men witha relatively low impact on life years lost. Even without discounting,ESRD causes more life lost than prostate cancer for both blackand white men. With discounting, this difference is accentuated,as quantified by the higher ratio of lost life-years for ESRDto those lost to prostate cancer. Simply put, ESRD occurs earlyand has a higher disease-specific mortality than prostate cancer.A small effect with discounting is also seen for colorectalcancer in black and white men (and breast cancer in white women),with an increase in the ratio of lost life for ESRD to lostlife for cancer with higher discount rates. On the other handthis ratio decreases only slightly with higher discount ratesfor black women, and it is stable in white women. This lackof an effect on life loss with higher discount rates suggeststhat breast cancer and ESRD impact the life of women proportionatelyat similar time points.
ESRD therapy is expensive and is a replacement therapy ratherthan curative. This study shows that the cumulative ESRD costsin blacks (including the mix of dialysis and transplantation)are much greater than the direct treatment costs (initial, maintenance,and terminal care) of cancer. Depending on the discount rate,ESRD is about 5 times more expensive than breast cancer forblack women, fourfold higher than for prostate cancer for blackmen and 10- to 20-fold higher than for colorectal cancer. Evenlarge errors in cost estimates cannot account for these differences.The cumulative costs of colorectal and prostate cancer are alsosignificantly less than those incurred due to ESRD for whitemen. Only the cumulative costs of breast cancer in white womenare higher than (or equivalent to) the cumulative costs of ESRD.
Cancer screening programs for breast, prostate, and colon havetraditionally had a high public profile, and special effortshave been made to increase screening in blacks. Not all diseasesthat have a significant impact on life are suitable for screening.Lung cancer is a prime example. Screening for renal diseaseis not presently recommended by the US Preventive Services TaskForce (16). The types of evidence required for such a recommendationare not available (17). However, ESRD has a comparable or greaterimpact on loss of life in black people than the cancers studiedin this report. This impact on life, together with high cumulativeESRD costs, suggest that concerted efforts at prevention maybe warranted in this group.
A recently published analysis suggests that screening for diabetesmellitus (a significant cause of ESRD) could be very cost-effectivein young (<45 yr old) black Americans (18), but this hasnot been tested clinically. Cost-effective models examiningthe effect of delaying diabetic complications rely heavily onpreventing ESRD (19). Renal disease is screened for in nondiabeticsubjects detected to have hypertension, but not all patientswith an elevated creatinine have hypertension (20). It has beensuggested renal impairment may predate overt hypertension andthat hypertensive nephrosclerosis is over-diagnosed in blackAmericans (21). Furthermore renal disease may cluster in families,especially black Americans (22). In an attempt to model renalloss in the US population (Third National Health and NutritionExamination Survey), we found that there are a small portionof young blacks in their forties (and probably younger) withmild renal impairment, who may be at great risk of progression(23). From the above and because the cumulative risk of ESRDin blacks already exceeds the lifetime risk in whites by age56, appropriately timed and more novel screening strategiesshould be studied in this population.
Screening alone is not, however, sufficient. A more recent reportsuggests that a large part of renal disease progression in diabeticblacks is related to socioeconomic status, suboptimal health-relatedbehavior, and poor BP and glucose control rather than race perse (24). Not only is research into cost-effective identificationof individuals at risk therefore lacking, but there is the lackof information on the efficacy of aggressive therapy in earlyrenal disease. This should soon be answered by the African AmericanStudy of Kidney Disease and Hypertension (25). Even if thisstudy is favorable, developing strategies to implement earlytreatment effectively into the general population will be criticalto reduce the burden of illness due to ESRD.
This analysis has several limitations. Its assumptions are afixed age and demographic structure for the general populationand stable ESRD and cancer incidence rates and mortality. TheESRD incidence rates do not include those who are not referredfor therapy, but could benefit or those who decline therapy.Each of these assumptions leads to bias in the direction ofunderestimation of cumulative risk of ESRD. If mortality ratesfor ESRD patients decrease over time, the impact of ESRD onlife expectancy may be less than that predicted. This will notaffect cumulative incidence rates and will increase cost projections,as each patient survives longer on renal replacement therapy.We have chosen this conservative approach rather than attemptingto project future incidence and mortality rates in view of thedifficulties associated with making accurate predications inthis area for both ESRD and cancer.
This article is not intended to be a self-serving message tonephrologists that a new screening strategy is required forthe prevention of ESRD over current preventive health recommendations.Nor is it intended to support a shift in research support fromcancer to renal disease. Our recommendations are more measured.Given the higher cumulative risk, early onset of disease andlarge cumulative costs of ESRD, we suggest greater attentiontoward implementing existing treatment strategies and more researchinto novel screening strategies in the black population. Dependingon the growth of ESRD and the feasibility, investigating newscreening strategies in other populations may then be warranted.
US Renal Data System: USRDS 2000 Annual Data Report, Bethesda MD, National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases, 2000
National Center for Health Statistics: US Vital and Health Statistics: Hyattsville, MD, National Center for Health Statistics, 2000
Steve Selvin: Monographs in Epidemiology and Bioststistics: Statistical Analysis of Epidemiologic Data,Vol. 17, Oxford, Oxford University Press, 1991,pp. 241–277
Anderson RN: A method for constructing complete annual US life tables. National Center for Health Statistics. Vital Health Stat 2, 1999,pp. 1–27
Beck JR, Kassirer JP, Pauker SG, Gottlieb JE, Klein K: A convenient approximation of life expectancy. Am J Med 73: 883–897, 1982[CrossRef][Medline]
Ries LAG, Eisner MP, Kosary CL, Hankey BF, Miller BA, Clegg L, Edwards: SEER Cancer Statistics Review, 1973-1998, Bethesda, MD, National Cancer Institute, 2001
Taplin SH, Barlow W, Urban N, Mandelson MT, Timlin DJ, Ichikawa L, Nefcy P: Stage, age comorbidity and direct costs of colon, prostate and breast cancer care. J Natl Cancer Inst 87: 417–426, 1995[Abstract/Free Full Text]
Stewart KJ, Reed SB: Consumer price index research series using current methods, 1978–1998. Monthly Labor Review 122 (6): 29–38, 1999
Salzman P, Kerlikowske K, Phillips K: Cost-effectiveness of extending screening mammography guidelines to include 40 to 49 years of age. Ann Intern Med 127: 955–965, 1997[Abstract/Free Full Text]
Frazier AL, Colditz GA, Fuchs CS, Kuntz KM: Cost-effectiveness of screening for colorectal cancer in the general population. JAMA 284: 1954–1961, 2000[Abstract/Free Full Text]
Krahn MD, Mahoney JE, Eckman MH, Trachtenberg J, Pauker SG, Detsky AS: Screening for prostate cancer: A decision analysis. JAMA 272: 773–780, 1994[Abstract]
US Congress, Office of Technology Assessment: Cost-effectiveness of colorectal cancer screening in average-risk adults. OTA-BP-H-146. Washington, DC, US Government Printing Office, 1995
US Congress, Office of Technology Assessment: Costs and effectiveness of prostate cancer screening in elderly men. OTA-BP-H-145. Washington, DC, US Government Printing Office, 1995
Kattlove H, Liberati A, Keeler E, Brook RH: Benefits and costs of screening and treatment for early breast cancer. JAMA 273: 142–148, 1995[Abstract]
Weinstein MC, Siegal JE, Gold KR, Kamlet MS, Russell LB: Recommendations of the panel on cost-effectiveness in health and medicine. JAMA 276: 1253–1258, 1996[Abstract]
Harris RP, Helfand M, Woolf SH, Lohr KN, Mulrow CD, Teutsch SM, Atkins D, for the Methods Work Group, Third US Preventive Services Task Force: Current methods of the US Preventive Services Task Force: A review process. Am J Prev Med 20 [Suppl 3]: 21–35, 2001[Medline]
CDC Diabetes Cost-Effectiveness Study Group: The cost-effectiveness of screening for type 2 diabetes mellitus. JAMA 280: 1757–1763, 1998[Abstract/Free Full Text]
Kiberd B, Larsen T: Estimating the benefits of solitary pancreas transplantation in nonuremic patients with type I diabetes mellitus. Transplantation 70: 1121–1127, 2000[CrossRef][Medline]
Coresh J, Wei GL, McQuillan G, Brancati FL, Levey AS, Jones C, Klag MJ: Prevalence of high blood pressure and elevated serum creatinine level in the United States: findings from the third National Health and Nutrition Examination Survey, 1988-1994. Arch Intern Med 161: 1207–1216, 2001[Abstract/Free Full Text]
Schlessinger SD, Tankersley MR, Curtis JJ: Clinical documentation of end-stage renal disease due to hypertension. Am J Kidney Dis 23: 655–660, 1994[Medline]
Freedman BI, Soucie JM, McClellan WM: Family history of end-stage renal disease among incident dialysis patients. J Am Soc Nephrol 8: 1942–1945, 1997[Abstract]
Kiberd BA, Clase CM: Modelling renal function loss in the US population [Abstract]. J Am Soc Nephrol 11: 153A, 2000
Krop JS, Coresh J, Chambless LE, Shahar E, Watson RL, Szklo M, Brancati FL: A community-based study of the explanatory factors for the excess risk for early renal function decline in blacks and whites with diabetes. Arch Intern Med 159: 1777–1783, 1999[Abstract/Free Full Text]
Wright JT Jr, Kusek JW, Toto RD, Lee JY, Agodoa LY, Kirk KA, Randall OS, Glassock R: Design and baseline characteristics of participants in the African American Study of Kidney Disease and Hypertension (AASK) Pilot Study. Control Clin Trials 17 [Suppl 4]: 3S–16S, 1996[Medline]
Received for publication December 13, 2001.
Accepted for publication February 8, 2002.
This article has been cited by other articles:
B. Kiberd The Chronic Kidney Disease Epidemic: Stepping Back and Looking Forward
J. Am. Soc. Nephrol.,
November 1, 2006;
17(11):
2967 - 2973.
[Abstract][Full Text][PDF]
L. C. Walter, K. Lindquist, A. M. O'Hare, and K. L. Johansen Targeting screening mammography according to life expectancy among women undergoing dialysis.
Arch Intern Med,
June 12, 2006;
166(11):
1203 - 1208.
[Abstract][Full Text][PDF]
T. Tillin, N. Forouhi, P. McKeigue, and N. Chaturvedi Microalbuminuria and Coronary Heart Disease Risk in an Ethnically Diverse UK Population: A Prospective Cohort Study
J. Am. Soc. Nephrol.,
December 1, 2005;
16(12):
3702 - 3710.
[Abstract][Full Text][PDF]
T. Ortega, F. Ortega, C. Diaz-Corte, P. Rebollo, J. Ma Baltar, and J. Alvarez-Grande The timely construction of arteriovenous fistulae: a key to reducing morbidity and mortality and to improving cost management
Nephrol. Dial. Transplant.,
March 1, 2005;
20(3):
598 - 603.
[Abstract][Full Text][PDF]
B. M. Curtis, P. Ravani, F. Malberti, F. Kennett, P. A. Taylor, O. Djurdjev, and A. Levin The short- and long-term impact of multi-disciplinary clinics in addition to standard nephrology care on patient outcomes
Nephrol. Dial. Transplant.,
January 1, 2005;
20(1):
147 - 154.
[Abstract][Full Text][PDF]
W. M. McClellan and W. D. Flanders Risk Factors for Progressive Chronic Kidney Disease
J. Am. Soc. Nephrol.,
July 1, 2003;
14(90002):
S65 - 70.
[Abstract][Full Text][PDF]
Top
Abstract
Introduction
