Epidemiological profi le of patients on renal replacement therapy in Brazil , 2000-2004

OBJECTIVE: To describe the clinical and epidemiological profi le of patients under renal replacement therapies, identifying risk factors for death. METHODS: This is a non-concurrent cohort study of data for 90,356 patients in the National Renal Replacement Therapies Database. A deterministicprobabilistic linkage was performed using the Authorization System for High Complexity/Cost Procedures and the Mortality Information System databases. All patients who started dialysis between 1/1/2000 and 12/31/2004 were included and followed until death or the end of 2004. Age, sex, region of residence, primary renal disease and causes of death were analyzed. A proportional hazards model was used to identify factors associated with risk of death. RESULTS: The prevalence of patients under renal replacement therapies increased an average of 5.5%, while incidence remained stable during the period. Hemodialysis was the predominant initial modality (89%). The patients were majority male with mean age 53 years, residents of the Southeast region and presented unknown causes as the main cause of chronic renal disease, followed by hypertension, diabetes and glomerulonephritis. Of these patients, 42% progressed to death and 7% underwent kidney transplantation. The patients on peritoneal dialysis were older and had higher prevalence of diabetes. The death rate varied from 7% among transplanted patients to 45% among non-transplanted patients. In the fi nal Cox proportional hazards model, the risk of mortality was associated with increasing age, female sex, having diabetes, living in the North and Northeast region, peritoneal dialysis as a fi rst modality and not having renal transplantation. CONCLUSIONS: There was an increased prevalence of patients on renal therapy in Brazil. Increased risk of death was associated with advanced age, diabetes, the female sex, residents of the North and Northeast region and lack of renal transplant. DESCRIPTORS: Renal Insuffi ciency, Chronic, epidemiology. Renal Replacement Therapy. Hospital Information Systems. Mortality Registries.


INTRODUCTION
The aging of the population and increased life expectancy, resulting from the demographic transition over the last decades in Brazil, contributed to changes in the morbidity and mortality profi le and the increase in the prevalence of chronic diseases, including chronic kidney disease (CKD). 7Hypertension and diabetes are the main risk factors for CKD and are becoming more common in the general population, contributing to the increased incidence of CKD. 4 CKD is global public health problem.Urinal tract and renal diseases account for approximately 850 thousand deaths every year and 15 million disability-adjusted life years lost, constituting the 12th leading cause of death and 17th cause of disability. 23The fi nal stage of CKD is called end-stage renal disease (ESRD), when the patient needs a renal replacement therapy (RRT) to survive.The available RRT methods are dialysis (hemodialysis [HD] and peritoneal dialysis [PD]) and renal transplantation (RT).The prevalence of ESRD in the global population increased 6% between 2003 and 2004. 7At the end of 2004, approximately 1.8 million patients were undergoing RRT in the world, a prevalence of 280 patients per 1 million of population (pmp).Of those, 77% were undergoing some form of dialysis and 23% were renal transplanted. 7e growth of the population with CKD has substantial implications on public policies in health, especially due to the high cost of patients on RRT, with 85% to 95% of this therapy subsidized by the National Health System (SUS). 18The systematic collection of information about dialysis patients has been a challenge for the majority of countries.In Brazil, there is a lack of national-level data to assist the monitoring of the population on RRT. 18A possibility for overcoming this diffi culty is the use of information from administrative data systems, whose primary objective is the documentation of payments for procedures performed by the SUS for patients on RRT.
The objective of the present study was to describe the clinical and epidemiological profi le of patients on RRT in Brazil, identifying risk factors for death.

METHODS
This is a non-concurrent cohort study, stemming from a large study called the "RRT Project -Economicepidemiological evaluation of renal replacement therapies in Brazil" conducted by the Research Group in Health Economics of the Universidade Federal de Minas Gerais. 5The data source was the National RRT Database, developed by probabilistic linkage performed on the Authorization System for High Complexity/Cost Procedures (APAC) database from the SUS Ambulatory Information System (SIA) and on the Mortality Information System (SIM) database, with the goal of following a cohort. 5,15nsidering that RRT procedures are continuous, a process was developed for inputting information in the documentation gaps in RRT modality between the fi rst month of observation, the occurrence of death or the a Therneau T. Survival analysis, including penalized likelihood: package version 2.34-1 [internet].R-Forge Statistics.[cited 2008 Oct  20].Available from: https://r-forge.r-project.org/search/?type_of_search=soft&group_id=0&atid=0&forum_id=0&group_project_ id=0&words=package+version+2.34-1&Search=Searchb R Development Core Team (2008).R: A language and environment for statistical computing.R Foundation for Statistical Computing, Vienna, Austria.ISBN 3-900051-07-0, URL http://www.R-project.org.end of follow up, due to project end (12/31/2004) or by loss of follow up.The data entry was done randomly when dealing with intervals between different treatment modalities, and when the same modality was at the either end of the interval, the modality was repeated.
The population studied included all the patients registered for RRT in the National Database, who began dialysis in the period from 1/1/2000 to 12/31/2004, with at least three consecutive months of procedure registered.For the survival analysis, patients who died in the fi rst three months under RRT and patients under 18 years were excluded.
The dependent variable was the elapsed time between the day the fi rst treatment modality began until the day of patient death.The independent variables were: demographic (age, sex, municipality and region of residence); clinical (initial diagnosis of cause for chronic kidney disease [International Classifi cation of Diseases -10th Edition -ICD 10]; treatment modality [HD, PD and RT)], length of treatment) and outcome (death, continued treatment or loss to follow up).The initial treatment modality was defi ned as the fi rst modality in which the patient remained for at least three consecutive months, without considering the subsequent changes to modality.
A descriptive analysis was performed through frequency distributions, measures of central tendency and variability of the characteristics studied.The χ 2 test was used to verify differences of proportion between categorical variables, and the Student's t test used for comparison of continuous variables.
The analysis of survival considered the total time in RRT (HD or PD) and the length of survival after performing renal transplantation, independent of changes between modalities.Death was considered as a fi nal event, and the patients were only censored through loss of follow up or at the end of the study period. 10 order to indentify an effect independent of the explanatory variables for survival, a multivariate Cox proportional hazard model was used.The proportionality assumption was evaluated by the graphic logminus-log method.The quality adjustment for the fi nal model was evaluated by graphic analysis of martingale and deviance residuals.The analyses were done using the survival package a of the open software R b 2.7.2, with a signifi cance level of 5%.
The RRT Project was approved by the Research Ethics Committee of the Universidade Federal de Minas Gerais (Process: ETIC 397/2004).

RESULTS
In Brazil, 68,467 patients underwent RRT in 2000, of which 8,501 progressed to death and 1,190 were lost to follow up until 12/31/2004.Among the 58,746 patients in RRT at the end of the year, there was a prevalence of 354 pmp (Table 1).Another 17,114 patients began treatment in 2000, which corresponds to an incidence rate of 103 pmp.Between 2000 and 2004, the prevalence of patients on RRT increased by a mean of 5.5%.The incidence rate was stable during this period, and the lethality rate increased.
Between 2000 and 2004, 90,356 patients began dialysis in Brazil, with hemodialysis being the most common treatment (Table 2).The main primary cause of CKD was indeterminate for the majority of patients, followed by hypertension and other cardiovascular diseases, diabetes and glomerulonephritis.Of these patients, 7% underwent renal transplantation (live or deceased donor) and 42% progressed to death.The three main causes of death among these patients were related to CKD.A gradual increase in mean age at initiation of RRT was observed over the study period, from 52 years in 2000 to 54 years in 2004.In the fi ve years of follow up, the mean time on RRT was 19 months.
In the fi ve regions of the country, the youngest age at initiation of RRT was found among patients in the Central-West region (mean=51; SD 17.2 years) and the greatest age in the South region (mean=55; SD=17.5 years), which also had the greatest percentage of elderly patients (31%).In all fi ve regions, the majority of patients (from 60% to 69%) began RRT at an intermediate age (20 to 64 years).
The main cause of CKD at enrollment was hypertension and other cardiovascular diseases (ranging from 42% to 54%), except for the North region where the most frequent cause was diabetes (36%).In all regions, the majority of patients with ESRD began RRT with hemodialysis, with the greatest percentage in the Northeast region (92%).The highest percentage of PD occurred among patients residing in the Southeast (12%).In the North region, 4% of the patients underwent transplant, while the greatest percentages were observed in the Southeast and Central-West regions (both 8%).The North region had the greatest percentage of deaths (47%), and the Central-West region had the lowest (39%).The longest mean time on RRT was in the Southeast region (20 months), and the lowest was in the North region (16 months).
The majority of the patients that initiated RRT with hemodialysis (Table 2) were men, with mean age of 53 years, at the age group of 45 to 64 years.For those who began with PD, the sex distribution was similar, with mean age of 55 years and the majority of patients in the age group above 65 years.A greater number of patients were observed in the Southeast, Northeast and South regions.
Of the patients beginning on hemodialysis, 10% began RRT with an access point through an arteriovenous fi stula.In PD, 28% performed the training procedures recommended at treatment enrollment.At the end of the follow up period, 47% of the patients on PD and 42% on hemodialysis progressed to death, principally do to diabetes mellitus and cardiovascular diseases.The mean time on RRT was similar for both modalities.
In Table 3, a greater percentage of male patients can be observed in the transplanted group.The mean age was 37 years, and 2% were above 65 years.Among patients that did not undergo transplantation, the mean age was 55 years and 28% of them were above 65 years old.
Among the transplanted patients, the main causes of CKD were glomerulonephritis, hypertension and other cardiovascular diseases and diabetes.For the nontransplanted patients, hypertension and other cardiovascular diseases, diabetes and glomerulonephritis were observed.Forty-fi ve per cent of the non-transplanted and 7% of the transplanted patients progressed to death.The mean duration of treatment was 19 months among the non-transplanted and 41 months for the transplanted.
The 76,949 patients for the survival analysis were selected from a total of 90,356 patients that initiated dialysis in Brazil between 2000 and 2005, excluding 2,727 patients under 18 years and 10,680 who died in the fi rst three months of treatment.The Figure shows the survival curves for the patients that initiated RRT between 2000 and 2004 in Brazil, according to selected characteristics.The graphic (b) with the Kaplan-Meier cumulative survival probabilities by region showed an intersection in the curves for the Southeast and Central-West regions.Since there was not a difference in the Log-rank test (p = 0.66) for these regions, they were grouped in order to avoid violating the supposition of risk proportionality, thus producing curves that did not cross and that were signifi cantly different.4).
The analysis of the plots for the Martingale and Deviance residuals suggests that there were no outliers in the analysis, which could infl uence the relative risk estimate.There were no residuals greater than 3 or less than -3 in any of the plots.

DISCUSSION
The current study was performed with population data of patients in renal replacement therapy through the SUS and showed an increased prevalence in the number of patients on RRT, even though the incidence has remained constant.The majority of the patients initiated hemodialysis treatment at a productive age, are residents of the Southeast region and have hypertension and diabetes as the primary cause of ESRD.At the end of the follow up period, 42% of patients progressed to death and only 7% were transplanted.Greater survival was found for patients who underwent renal transplantation.
The mean increase in the prevalence ESRD, estimated at 5% in Brazil, also follows the internationally observed trend (6%) for the world population. 7The prevalence of 431 pmp found for Brazil is greater than the average of Latin American countries (349 pmp in 2001), 6 but less than in developed countries (700 pmp in Europe and 1403 pmp in the USA) 3 and even those described in countries like Uruguay (809 pmp), Chile (662 pmp) or Argentina (571 pmp). 6These fi gures suggest that in Brazil a portion of people with ESRD have not been diagnosed or do not have access to health services. 12esides this, it is probable that many patients with CKD died from complications of diabetes and hypertension before reaching end stage renal disease.The incidence rate remained stable in the period, as has occurred in the USA and in several developed countries. 20The small  15 In Brazil, men represent the majority (57%) of new patients, a fi nding similar to international studies. 8,9,19he main cause of CKD identifi ed by our study was hypertension, followed by diabetes, in contrast to other  studies performed in countries of America, 6,8,9,19 in which diabetes was identifi ed as the leading cause.The differences observed between the studies may be the result of diffi culties in establishing a precise diagnosis of this disease in Brazil, which can be inferred from the high percentage of indeterminate causes.There was an important proportion of glomerulonephritis observed, which can be associated with infectious agents, c with the Brazilian profi le approaching the profi le of African countries and diverging from what occurs in more developed countries. 1e greater concentration of RRT patients in the Southeast and South regions may be the result of the greater population density of these regions and the greater availability of health services of high technological complexity, which favors patient access. 21Moura et al, 11 only used the enrollment archive of the APAC for analysis and found similar prevalence, incidence rate and primary cause for ESRD, as well as a similar distribution of patients between the Brazilian regions.
The distribution of the patients between the modalities of hemodialysis and PD, respectively 89% and 11%, is in accordance with the global experience. 7In agreement with the global picture, the results of our study showed that the population in dialysis has become more elderly.In the United States, in 1996, patients with 65 years or more were 46% of new patients on renal replacement therapy. 10spite not being the main primary cause of CKD, diabetes accounts for a greater proportional contribution for patients that began PD than those that in hemodialysis, suggesting a differential allocation of patients between these modalities.In developed countries, the patients on PD are more autonomous, have higher schooling, perform better preparation before initiating RRT and have less comorbidities. 19The majority of patients with ESRD can be treated with any RRT, according to their clinical conditions, with each one having advantages and disadvantages.Nonetheless, it is thought that in Brazil, as in other countries, the choice of RRT modality is based on clinical condition, as well as on non-medical reasons such as: fi nancial compensation, lack of consistent information to the patient regarding the options, availability of resources and moral, social and cultural aspects. 3 the present study, 10% of patients began hemodialysis did so with defi nitive vascular access (arteriovenous fi stula); in European countries this rate was 66% and in the USA 15%. 14In Brazil, the high percentage of people initiating hemodialysis with temporary access suggests the possibility of diffi cult access to nephrologists before the end stage of the disease, late diagnosis or even under diagnosis of CKD. 12 In Brazil, 7% of new patients were transplanted in fi ve years of follow up, in the United Kingdom this percentage was 6.2%, 17 5.7% in Australia and 3.7% in New Zealand.d The transplanted population initiated RRT at a younger age and with a greater proportion of glomerulonephritis (47%) as the primary cause of CKD.Despite the existence of a single waiting list, patients residing in the Southeast region had greater access to renal transplant, revealing geographic disparities reported in other studies. 13Despite the increased demand for kidney transplantation and the marked growth of the waiting list, it has been reported that access to renal transplant is very unequal between regions in the country, especially for the low income population without private health insurance plans and that live far from transplantation centers, mainly concentrated in the South and Southeast regions.e In accordance with other studies, 9,16 we found greater survival for patients that underwent renal transplant.
In addition, greater mortality of patients on RRT was associated with increased age and the presence of diagnosed diabetes.International studies show that besides these factors, a long time on the transplant wait list contributes to greater mortality. 22Besides this, a long-term survival advantage has been reported for transplanted patients, even in diabetic patients. 9 this study, lower survival was observed for patients who initiated RRT on PD compared to those on HD.This fi nding may be due to the probable selection bias in the allocation of patients to PD as an initial modality, since a greater proportion of patients beginning on this modality were of older mean age and diabetic, as previously described.There is no consensus in the literature about the benefi ts to survival between the different dialyses.Nonetheless, recent studies suggest greater survival on PD in the fi rst year of RRT, followed by an equivalent survival between the modalities. 8,22 is necessary to have reliable and available databases about RRT in Brazil, since they are fundamental for recognizing various aspects of chronic renal disease, such as prevalence, incidence, mortality and factors that infl uence its development.The systematic collection of data about the patients on dialysis remains a challenge in the majority of countries.Besides this, they can provide data to characterize the reality of dialysis treatment, the identifi cation of problems in the provisioning of therapy and the analysis of patient survival, morbidity and quality of life.The combination of this information can inform the planning of actions and support the more rational use of fi nancial resources devoted to this high cost therapy. 18spite the possibilities that the National Database of RRT provides, the utilization of an administrative database, whose purpose is billings and not research, results in some incomplete, inconsistent and inexistent information.These characteristics limit the possibility of making larger inferences.For example, the lack of fi nancial, social (race, education and income) and clinical (comorbidity) data deserves mention, since these are fundamental for evaluating the equity in access to RRT, as well as associated risk factors.Besides this, we highlight the high proportion of indeterminate causes (44%), which negatively impacted the distribution of causes of CKD; in developed countries this percentage is at most 19%. 2 Other important information that are also missing from this database concern hospitalization (cause, duration of stay) and type of donor for the renal transplant (live or deceased).
In conclusion, there was an increase in the prevalence of patients on renal therapy in Brazil, with a high mortality during the fi ve year study period.The factors associated with increased mortality risk were increased age, female sex, diabetes, living in the North and Northeast regions, peritoneal dialysis as an initial modality and not having renal transplant.These fi ndings can contribute to improved care for people with chronic kidney disease in the country.

Figure .
Figure.Survival curves for patients on RRT in Brazil from 2000 to 2004, according to selected characteristics: (a) Age group, (b) Region of residence, (c) Sex, (d) Cause of CKD, (e) Initial treatment modality and (f) Renal transplantation (curves with p<0.05 for Long-Rank test).

Table 1 .
Prevalence, incidence and lethality rate of patients on renal replacement therapy.Brazil, 2000 to 2004.
Source: National Database of RRT IBGE: Instituto Brasileiro de Geografi a e Estatística; pmp: patients per million population

Table 2 .
Demographic and clinical characteristics of incident patients on renal replacement therapy according to initial treatment modality.Brazil, 2000-2004.
a Test by independent sample assuming unequal variance * p<0.001RRT: Renal Replacement Therapy; HD: Hemodialysis; PD: Peritoneal Dialysis; CKD: Chronic Kidney Disease

Table 4 .
Cox proportional risk model for survival analysis of patients on renal replacement therapy according to demographic and clinical variables.Brazil, 2000-2004.
increase in the lethality rate between 2000 and 2004 can be attributed to the under notifi cation of identifying patient data in the Mortality Information System (SIM), as reported byQueiroz et al (2009).