Epidemiology of end-stage renal disease: International comparisons of renal replacement therapy

Kidney International, Vol. 57, Suppl. 74 (2000), pp. S-39 S-45 REGIONAL EXPERIENCE Epidemiology of end-stage renal disease: International comparisons of renal replacement therapy FRANCESCO PAOLO SCHENA Division of Nephrology, Dialysis and Transplant, Department of Emergency and Organ Transplantation, University of Bari, Policlinico, Bari, Italy important observations on the health of the patients with ESRD, the extent and accuracy of the data collected vary widely. For this reason, Kjellstrand [1] stated that international comparisons of data on ESRD patients and renal replacement therapy may not be valid because of the differences in acceptance of the treatment, patient demographics, socioeconomic burdens, and national health care legislation. Considering this important scien- tific bias, the renal registries decided to organize an inter- national collaborative study group called the Interna- tional Federation of Renal Registries (IFRR) to discuss different databases. The first meeting of the IFFR was held in 1997 during the 14th International Congress of Nephrology in Sydney, Australia, and the first report was published in the same year [2]. Delegates from the Asian-Pacific Registry, Australian and New Zealand Dialysis and Transplant Registry (ANZDATA), Canadian Organ Replacement Registry (CORR), European Renal Association Registry (ERA-EDTA), Japanese Society for Dialysis Therapy (JSDT), Latin American Registry, and the United States Renal Data System Registry (USRDS) participate in the IFRR. The duties of the federation include the development of a dialogue be- tween renal registries for the improvement and standardization of renal databases; the definition of a common terminology, the definition of a common methodology for analyzing data and finally the promotion of interna- tional collaborative studies. Epidemiology of renal replacement therapy: International comparisons of renal replacement therapy. Renal replacement therapy (RRT) remains the most important tool for all patients with end-stage renal disease (ESRD) in the world. Patients are usually listed in the national renal registries which report demographic data, incidence, prevalence and outcome. Differences in acceptance of the RRT, patient demographics, socioeconomic burdens and national health care legislature influ- ence the validity of international comparisons of ESRD therapy. For this reason, the renal registries decided to organize an international collaborative study group, the International Federation of Renal Registries (IFRR), for improving the standardization of renal databases and for promoting international collaborative studies. Demographic data from renal registries in 1996 showed that approximately 1,000,000 of ESRD patients received RRT throughout the world and approximately 200,000 new patients started RRT. The different distribution of patients in the world is strongly influenced by the level of gross domestic product (GDP), since economical crisis may influence the number of patients starting RRT. The reported incidence and prev- alence values of ESRD patients in RRT from the renal registries show these differences. The mortality rate of these patients is notably influenced by the percentage of enrolled diabetic RRT patients. Therefore, important differences in outcome of RRT depend on the underlying disease which is a determinant factor of prognosis. Other factors responsible for the outcome are represented by dialysis dose, type of membrane, erythropoietin and nutritional status. Renal replacement therapy (RRT) by hemodialysis, peritoneal dialysis, and renal transplantation became available in the 1960s; thereafter, a growing number of patients undergoing treatment of end-stage renal disease (ESRD) were listed in the renal registries. Dialysis continues to be the most frequent type of RRT because of the low rate of renal transplantation, strictly related to insufficient organ donation to meet demand. Renal transplantation is the best form of treatment for a limited number of ESRD patients. Although the use of data from renal registries permits Key words: renal registries, incidence, prevalence, outcome measures. 2000 by the International Society of Nephrology DEMOGRAPHIC DATA Table 1 shows the number of patients on RRT world- wide in 1996 and the number of new patients who started RRT for the first time. Data have been obtained from renal registries with 90% response rate (RR) from dialysis facilities, registries with 90% response rate, and from published papers (1999 pers. comm. from G. Lee, Asian-Pacific Registry; A. Disney, ANZDATA; S. Fenton, CORR; F. Berthoux, ERA-EDTA; K. Maeda, Japanese Society for Dialysis Therapy; N. Mazzuchi, S-39

S-40 Schena: Comparisons of RRT worldwide Table 1. RRT patients world-wide as of December 31, 1996 registries are an under-representation of patients receiving Prevalence Incidence Countries RRT. Third, in addition to the quantity and quality Registry (RR 90%) of the therapy, access to the different forms of RRT is ANZDATA 11,066 2,292 2 partly dependent on the health care system of the coun- CORR 18,874 3,413 1 try. This explains why in some areas of the world in JSDT 167,192 28,409 1 USRDS 282,847 73,091 1 which the GDP is low, a low percentage of subjects Total 479,979 107,205 with ESRD receives RRT. On the other hand, in some 488,000 a 108,000 a Registry (RR 90%) countries with a high GDP, renal transplantation is less Asian Pacific 46,105 12,938 10 frequent because organ donation is low due to religious ERA-EDTA 317,131 63,416 43 or social problems. Such variables cannot be easily ad- Latin American 81,796 21,894 21 Indian Ocean Soc. 767 125 4 justed for in most statistical analyses. Finally, reports Total 445,799 98,373 from the renal registries include ESRD patients receiv- 450,000 a 102,000 a ing RRT either dialysis or renal transplantation by No registry India 20,000 a 2,000 a 1 the first day of treatment. However, this is not consistent China 30,000 a 4,000 a 1 from registry to registry; for instance, the registry of Others 20,000 a 4,000 a Total 70,000 the Japanese Society for Dialysis Therapy includes only 10,000 a Total 1,000,000 a 220,000 a dialyzed patients, while the USRDS registry considers a Estimated all ESRD patients treated with RRT by the 91st day because the Medicare system does not achieve complete reporting of patients data before day 90. Thus, patients Latin American Registry; and L. Agodoa, USRDS). For who die before this time (early mortality within 90 days) some countries the number of RRT patients has been are not accounted for by the USRDS. In addition, in all estimated. In 1996, approximately 1,000,000 patients rebefore receiving RRT are not included, and data from renal registries world-wide, patients who die of ESRD ceived RRT (dialysis or renal transplant) in the world and 200,000 new patients started RRT. There are several the registries consequently do not express the true epide- limitations to this estimate because in the international miology of ESRD in those countries. renal registries and in the countries with low gross domestic product (GDP), some factors such as demography of the population, socioeconomic structures, organization INCIDENCE of health care systems, patient access to RRT, and Incidence represents the number of persons diagnosed the response rate of dialysis facilities to the questionnaire with ESRD in a specific population in a given time peinfluence the estimation of RRT population. These limits riod, usually a year. It is a useful measure for medical were documented by Mazzuchi et al in the Latin Ameri- and epidemiological research that examines the causes can Registry [3]. The most important factor, for the renal of the disease and the differences in how subpopulations registries, which influences the collection of annual data, are affected by these causes. The annual incidence rate is the response rate to the questionnaire. of treated ESRD patients is expressed by the number The database of renal registries world-wide should of patients beginning RRT related to the annual general validate the accuracy at the time of data collection, but population at the beginning of the year. It is expressed four important, strictly-correlated factors may influence as a number of patients per million of population (pmp) the quality of the data. First, the submission of the data per year. There is a bias of the incidence among the may be mandatory and linked to the reimbursement of renal registries because not all patients with ESRD are treatments performed in dialysis facilities. The USRDS included, as was mentioned in the discussion of demograregistry works in this manner because Medicare only reimburses the cost of dialysis treatment after it receives phy above. In addition, the incidence of ESRD does not patient data, which are then supplied to the USRDS. take into account those patients not treated; therefore This system covers more than 93% of all patients treated it does not measure the overall occurrence of ESRD. for ESRD in the United States. Second, other registries The incidence of ESRD patients on RRT continues work on the basis of voluntary submission of data. In such to increase world-wide. By considering the incidence cases, only a few renal registries, such as the ANZDATA rates of the four national renal registries with more than registry, CORR, and the Japanese Society for Dialysis 90% of response rate to the questionnaire during these Therapy receive data from more than 90% of all dialysis last four years (1994 1997), the number of new patients facilities. Other voluntary national and international regthan pmp listed in USRDS and Japanese Registries is higher istries have a response rate to the questionnaire of the in ANZDATA and CORR. The mean annual per- dialysis centers lower than 90%. For these reasons some centage of increase is 7.6% of incidence rate in USRDS

Schena: Comparisons of RRT worldwide S-41 Fig. 1. Overall incidence rates in new ESRD patients per million population. Symbols are:, 1994;, 1995;, 1996;, 1997. Fig. 2. Asian-Pacific Registry incidence data from selected countries (response rate 90%). Incidence rates are given as patients per million population (pmp). and 3.2 3.7% in the other three registries (Fig. 1). The rates of newly treated ESRD patients vary from registry of time (period prevalence). Point prevalence indicates to registry and may reflect differences in acceptance and the number of alive treated patients on a particular date, referral patterns among countries. The likelihood of ac- usually December 31 of the year. It is the most common ceptance into RRT program is influenced by many fac- definition of prevalence used by the renal registries. tors such as age, gender, distance of home from dialysis Point prevalence is a useful measure for public health facility, employment and transplantation program. The research, since it measures the overall burden of the acceptance rates for incident patients continues to in- disease on the health care delivery system. Period prevacrease in industrialized countries but not in those with lence differs from the above term because it counts the low GDP. In fact, in 1998 a reduction in incidence of number of patients receiving RRT anytime during a parnew RRT patients was reported in some Asian countries ticular interval of time, such as a calendar year. It is the (Singapore and Taiwan) as a result of the economic crisis sum of the point prevalence on the first day of the year in these areas of the world (Fig. 2), while a slight increase plus the incidence during the year. It is used less often was observed in some Latin-American countries in the than point prevalence. Period prevalence is a useful mealast two years (Fig. 3). In countries with high GDP, the sure for cost analysis, since it indicates the total burden mean age of incident patients is still increasing, due to of the disease over the course of the year. Its rate is the above-average proportion of older patients starting related to the annual general population count preferatherapy [4]. About half of the new patients are over the bly in the middle of the year and is expressed as number normal retirement age, and over the last decade there of patients per million population (pmp). For simplicity has been at least a four-fold increase in the number of both incidence and prevalence are referred as rates. patients entering dialysis over the age of 74 years [5]. A These two terms are adjusted to age, gender and other growing proportion of the RRT population is over the conditions. The Japanese registry records the highest age of 85 years. The pros and cons of RRT in patients mean prevalence rates (7.4%) during the last four over 80 years of age are continuously discussed [6]. How- years (1994 1998) followed by the USRDS, CORR and ever, elderly patients must be informed about the treat- ANZDATA (Fig. 4). This may be due to a better survival ments offered. Specific recommendations will help them among ESRD patients and low transplant activity in to decide whether to accept treatment, and the final Japan [7]. decision must be the patient s alone, with no pressure In other registries where incidence rates of ESRD to make choices on the grounds of financial limitations. patients in RRT are similar, in addition to patient demographics and transplant activity, a similar prevalence of PREVALENCE ESRD is observed (Figs. 5, 6, and 7). Prevalence represents the number of persons in a specific population who have ESRD at a given point. Both OUTCOME MEASURES incidence rates of the disease and the survival rates in- Outcome research is a rapidly evolving field that incorfluence it. The prevalence includes patients who are new porates epidemiology, health service research, health and those who are continuing RRT for ESRD. The prev- economics and psychometrics. The outcome measures alence rate per million population is the prevalence count have evolved from simple dichotomous ones such as divided by the population size in million on either a survival or occurrence of a clinical event to patient-oriented particular date (point prevalence) or during an interval measures such as satisfaction, quality of life and

S-42 Schena: Comparisons of RRT worldwide Fig. 3. Latin American Registry incidence data. (A) Overall incidence. (B) Incidence rates from selected countries (response rate 90%). due to differences in enrollment criteria or type of treatment. It can be adjusted among prevalent patients and differences in mortality could reflect innovations in treatment that affect all patients being treated in that year. Because large differences exist between countries on the acceptance of ESRD patients it is difficult to interpret observed differences in mortality rates. For dialysis patients factors influencing the mortality rate are represented by the type of disease causing ESRD, utilization of treatment modalities, different health care system and Fig. 4. Overall prevalence rates (patients pmp). Symbols are:, delivery. The USRDS registry excludes from some analy- 1994;, 1995;, 1996;, 1997. ses dialysis-unrelated deaths caused by AIDS, accidents unrelated to treatment, and street drug overdoses. These exclusions make the death rate more useful as a norm for evaluating the care given to ESRD patients. The functional status. The outcome measures, listed by Epadjusted annual death rates for all dialysis patients and stein and Sherwood, are grouped into three important for all ESRD patients, prevalent at any time during areas: clinical, characterized by clinical events, physioeach year, decreased progressively from 1989 to 1996 logic and metabolic measures and mortality; economic, in USRDS registry but remains higher than that of including direct and indirect medical measures; and huother renal registries [9]. The crude mortality rate in manistic, in which symptoms, quality of life, functional ANZDATA in these last three years remained approxistatus and patient satisfaction are included [8]. The mately 14% in dialysis patients and 7% in all ESRD choice of outcome measures depends on the priorities of those examining the data. Clinicians are more intertion, 1999) and ranged about 9% of dialysis patients in patients (A. Disney. ANZDATA, personal communica- ested in clinical and humanistic outcomes whereas pathe Japanese Registry [10]. Marcelli et al stated in a tients may be preferentially interested in humanistic measures. Health plan administrators are more interto develop a common standard data collection instru- recent paper that for international studies, it is necessary ested in economic ones. Current renal registry management in all ESRD registries including the recording of ment programs include a mixture of all measures. The mortality rate is an important outcome measure comorbid conditions and their severity [11]. Other facto characterize the impact of RRT on the ESRD patient tors responsible for the death rate among ESRD patients population in the national and international renal regiscellulose to synthetic membranes [13], administration of are represented by the dose of dialysis [12], shift from tries. The crude annual mortality rate is expressed by the number of deaths divided by the product of the erythropoietin dose [14], dialysis facility differences [15], number of ESRD patients and the time on treatment mode of dialysis therapy [16], hypertension [17], cardioduring the year (patient-years). It is expressed by the vascular diseases [18], Ca-P metabolism [19], and malnu- number of deaths per 100 patient-years at risk. Death trition [20]. rates can be adjusted for age categories, time spent on Important differences in outcome of RRT patients RRT and mortality of general population (standardized depend on the underlying disease, which is a determinant mortality ratio). Difference in mortality rate could be factor of prognosis. It is well known that diabetes reduces

Schena: Comparisons of RRT worldwide S-43 Fig. 5. Asian Pacific Registry prevalence data. (A) Overall prevalence. (B) Prevalence rates from selected countries (response rate 90%). Fig. 6. Latin American Registry prevalence data. (A) Overall prevalence. (B) Prevalence rates from selected countries (response rate 90%). Fig. 7. ERA-EDTA Registry prevalence data. (A) Overall prevalence. (B) Prevalence rates from 15 countries of the European Union (response rate 89%). the survival of ESRD patients; the prognosis is particu- receiving RRT in comparison with other countries [22]. larly severe for elderly patients. Some large differences For this reason, diabetes was the most common cause exist among countries in the utilization of dialysis modal- of ESRD in the United States in 1996, whereas glomerulonephritis ities and in acceptance of patients, particularly for diabetics was the primary cause of ESRD in other and elderly subjects. Therefore, it is difficult to registries patients (Table 2). Nevertheless, an increased interpret observed differences in mortality rates among incident number of diabetic patients has been evidenced registries [21]. A valid comparison of data in national in the last four years in all national and international registries should be attempted for the primary diagnosis. registries. The relatively high risk of death of these pa- The USRDS registry in the last report showed an in- tients is further increased by their poor nutritional status creased number of incident ESRD patients with diabetes and by factors related to their predialysis care. In fact,

S-44 Schena: Comparisons of RRT worldwide Table 2. The most common diseases responsible for ESRD in the dialysis than hemodialysis patients. Patient-reported renal registries (1996). a general health tended to decrease with age. Non-diabet- USRDS CORR ANZDATA JSDT ics had better health than diabetic hemodialysis patients. Incidence The majority of patients were retired or disabled. One- Diabetes 41.7 28.2 18.6 33.8 half of patients were driven by someone and one quarter Hypertension 25.7 9.9 9.7 6.5 Glomerulonephritis 10.5 16.1 34.5 38.2 drove themselves. After this study, in collaboration with Point prevalence Amgen, the University Renal Research and Education Diabetes 32.4 23.0 29.6 21.7 Association (URREA) located in Ann Arbor, Michigan, Hypertension 24.7 9.3 12.2 3.9 Glomerulonephritis 17.6 20.3 22.0 53.5 USA organized three important international dialysis a Values are expressed as percent of ESRD patients in RRT outcomes and practice patterns studies (DOPPS) in three continents of the world. The US-DOPPS involves 160 dialysis facilities as a national representative sample of US. The I-DOPPS randomly selected 100 dialysis facila prospective study on enrolled hemodialysis patients ities in five European countries (France, Germany, Italy, evidenced that at the start of dialysis diabetic patients Spain and the United Kingdom); 20 in each country. The had more left ventricular hypertrophy, ischemic heart Japan-DOPPS recruited 60 dialysis facilities in Japan as disease and cardiac failure than non-diabetics [23]. These a national representative sample. These three observafindings suggest that these events in the predialytic life tional and prospective studies follow a common design, should be notably reduced. protocol and analysis of the data for comparing practice The understanding of patient demographics and the patterns and their effects on patient outcomes. These proportion of patients with high mortality rate in predial- studies should lead to improve the general health of ysis life should be taken into account on the data base hemodialysis patients throughout the world. The ranof renal registries. It is possible to attempt to statistically domized choice of dialysis facilities and the accurate adjust differences in patient characteristics and comorbid instruction of the nursing staff and physicians participatconditions in a second phase. Consequently, interna- ing in the study will eliminate bias normally present in tional comparative studies of survival require multivari- the collection of data from retrospective studies. Thereate analysis with appropriate adjustment for differences fore, the adequate data collection methodology and the in patient demographics, known risk factors and comor- accuracy of information will give important insights on bid conditions. RRT in the world. Future collaborations among registries with more than 90% of response rate of the questionnaire from the dialysis facilities are well accepted for FUTURE COLLABORATIVE STUDIES prospective clinical studies. A few years ago the USRDS started a prospective observational study that produced very important preliminary results [24]. A substantial fraction of patients ACKNOWLEDGMENTS had had impaired nutrition for a prolonged period before I am indebted to the members of the International Federation of Renal Registries for their help: L. Agodoa (United States Renal Data starting dialysis; many reported nausea or vomiting for System); A. Disney (Australian and New Zealand Dialysis and Transplant Registry); S. Fenton (Canadian Organ Replacement Registry); 3 or more months. These data strongly suggest a need for appropriate medical care of the patients before ESRD K. Maeda (Japan Society of Dialysis Therapy); G. Lee (Asian Pacific Registry); F. Berthoux (ERA-EDTA Registry); N. Mazzuchi (Latintreatment, including early referral to nephrologist for American Renal Registry). patients with ESRD [20]. Patients receiving peritoneal dialysis were younger than those treated by hemodialy- Reprint requests to Prof. F. P. Schena, Division of Nephrology, Dialysis and Transplantation, Policlinico, Piazza Giulio Cesare 11, sis, and comorbid conditions were more prevalent among 70124 Bari, Italy. hemodialysis patients. In addition, the residual renal function among peritoneal dialysis patients was higher than REFERENCES hemodialysis patients. Poor compliance to RRT was regis- 1. Kjellstrand CM: International comparisons of dialysis survival tered by 10% of dialysis patients who skipped at least one are meaningless to evaluate differences in dialysis procedures. In: session, 15% who shortened at least one session, 12% Death on Dialysis, edited by Friedman EA. Hingham MA, Kluwer of CAPD patients who missed one exchange per week Academic, 1994, pp 55 68 and 6% of APD patients who skipped or shortened treat- 2. Schena FP: Report on the first meeting of the Chairmen of the National and International Registries. Kidney Int 52:1422, 1997 ment. Medication use was represented by a phosphate 3. Mazzuchi N, Schwedt E, Fernàndez JM, Cusumano AM, Anção binder containing calcium. Vitamin D was administered MS, Poblete H, Saldaña-Arévalo M, Espinosa NR, Centurión more frequently among HD treated than PD treated pa- C, Castillo H, González F, Milanés CL, Infante M, Ariza M: Latin American Registry of dialysis and renal transplantation: 1993 tients. Finally, considering the quality of life, the patient- annual dialysis report. Nephrol Dial Transplant 12:2521 2527, 1997 reported general health was slightly better for peritoneal 4. Molzahn M: Future evolution of ESRD patient population. A

Schena: Comparisons of RRT worldwide S-45 perspective for the year 2000. Nephrol Dial Transplant 11 (Suppl. Barre PE: The impact of anemia on cardiomyopathy, morbidity 8):59 62, 1996 and mortality in end-stage renal disease. Am J Kidney Dis 28:53 61, 5. Iglehart JK: The American health care system, the end stage 1996 renal disease program. N Engl J Med 328:328 366, 1993 15. McClellan WM, Soucie JM, Flanders WD: Mortality in end- 6. Schaefer K, Röhrich B: The dilemma of renal replacement therin stage renal disease is associated with facility-to-facility differences apy in patients over 80 years of age. Dialysis should not be withheld. adequacy of hemodialysis. J Am Soc Nephrol 9:1940 1947, 1998 Nephrol Dial Transplant 14:35 36, 1999 16. Foley RN, Parfrey PS, Harnett JD, Kent GM, O Dea R, Murray 7. Held PJ, Akiba T, Stearns NS, Marumo F, Turenne MN, Maeda DC, Barre PE: Mode of dialysis therapy and mortality in end- K, Port FK: Survival of middle-aged dialysis patients in Japan and stage renal disease. J Am Soc Nephrol 9:267 276, 1998 the U.S. In: Death on Dialysis, edited by Friedman EA. Hingham 17. Foley RN, Parfrey PS, Prichard SS, Snidermann SD: Impact of MA, Kluwer Academic, 1994, pp 12 23 hypertension on cardiomyopathy, morbidity and mortality in end- 8. Epstein RS, Sherwood LM: From outcomes research to disease stage renal disease. Kidney Int 49:1379 1385, 1996 management: a guide for the perplexed. Ann Intern Med 124:832 18. Silberg JS, Barre PE, Prichard SS, Snidermann SD: Impact of 837, 1996 left ventricular hypertrophy on survival in end-stage renal disease. 9. United States Renal Data System: 1998 Annual Data Report. Kidney Int 36:286 290, 1989 Patient mortality and survival. Am J Kidney Dis 32 (Suppl. 1): 19. Block GA, Hulbert-Shearon TE, Levin NW, Port FK: Associa- 69 80, 1998 tion of serum phosphorus and calcium x phosphate product with 10. Shinzato T, Nakai S, Akiba T, Yamagami S, Yamazaki C, Kitaoka mortality risk in chronic hemodialysis patients. A national study. T, Kubo K, Maeda K, Morii H: Report of the annual statistical Am J Kidney Dis 31:607 617, 1998 20. Pollack C, Ibels LS, Allen BJ: Nutritional markers and survival survey of the Japanese Society for Dialysis Therapy in 1996. Kidney in maintenance dialysis patients. Nephron 74:625 641, 1996 Int 55:700 712, 1999 21. Nelson CB, Port FK, Wolfe RA, Guire KE: The association of 11. Marcelli D, Stannard D, Conte F, Held PJ, Locatelli F, Port diabetic status, age and race to withdrawal from dialysis. JAm FK: ESRD patient mortality with adjustment for comorbid condi- Soc Nephrol 4:1608 1614, 1994 tions in Lombardy (Italy) versus United States. Kidney Int 50:1013 22. United States Renal Data System: 1999 Annual Data Report. 1018, 1996 XII: International Comparisons of ESRD therapy. Bethesda, MD, 12. Locatelli F, Manzoni C: Duration of dialysis sessions was Hegel USRDS, 1999 right? Nephrol Dial Transplant 14:560 563, 1999 23. Foley RN, Parfrey PS: Cardiac disease in the diabetic dialysis 13. Hakim RM, Held PJ, Stannard DC, Wolfe RA, Port FK, Dau- patient. Nephrol Dial Transplant 13:1112 1113, 1998 girdas JT, Agodoa L: Effect of the dialysis membrane on mortality 24. United States Renal Data System: 1996 Annual Data Report. of chronic hemodialysis patients. Kidney Int 50:566 570, 1996 The USRDS dialysis, morbidity and mortality study: wave 2. Am 14. Foley RN, Parfrey PS, Harnett JD, Kent GM, Murray DC, J Kidney Dis 30 (Suppl. 1):S67 S85, 1997