In the present study, the proportion of patients on hemodialysis who continued to use the catheter for a period > 3 months was still above the recommended level, as reported in Sesso et al. (2008) [2, 11]. Several countries have made efforts to reduce this proportion, such as the United States, where one third of patients are still on dialysis with a catheter after 6 months from the beginning of treatment [6, 8, 15]. Among European countries, differences in catheter use can also be observed, though some are close to, or have already reached, the established practice standards [8].
The average time until AVF placement in our study was greater than that found in the study by Linardi et al. (2004), whose estimate varied from 1 to 3 weeks for Brazil [16]. The study by Either et al. (2008) reported that the minimum time was 5 days for Italy and 43 days for the United Kingdom [8].
The results of this analysis indicated 4 individual factors that were associated with the time until the first AVF procedure: age; having hypertension and cardiovascular diseases or undetermined cause as causes of CKD; residing in capital cities; and residence in certain regions of Brazil (South, Midwest, North, and Northeast regions). These associations remained significant even after the inclusion of the variable related to renal unit (type of renal unit).
Increasing age was associated with longer times with temporary catheters, possibly because older patients were more likely to have vascular disease, making fistulas more difficult to establish, as suggested by several authors [4, 15, 17]. Reddan et al. (2002) suggest that AVF failure rates may be increased in older patients [17]. On the other hand, Stehman-Brenn et al. (2000) point out that older age does not necessarily result in poorer patency outcomes [15].
Hypertension was associated with a lower probability for the AVF procedure in relation to glomerulonephritis, in agreement with other studies that address hypertension [18]. This proposition is plausible for Brazil, where hypertensive patients face difficulties in obtaining pre-dialysis care, which reduces their chances of having a mature access at the time of dialysis [19].
The present study also found that patients who lived in capital cities had a lower probability of having timely performance of fistula surgery. A study by Osis et al. (1993) found a higher use of services by the inhabitants of the interior cities (medical appointments) which may be associated with differences in the access to care [20]. Therefore, for our study, residing in interior cities could result in less competition for the performance of the outpatient procedure. However, another study diverges from this finding and reports easier access for the inhabitants of the capitals, indicating that these cities may have a higher technological density and capacity [21].
Brazil is divided into five regions: Southeast, South, Midwest, Northern and Northeast. This division emphasizes a historical and spatial perspective, referring to economic and social characteristics and political organization of the national space [22]. The differences found among the regions of Brazil point to differences in access to health care in the country. Coelho et al. (2006) point to a large discrepancy in the health indicators among the South and Southeast regions, which present better results, and the Northern and Northeast regions, which may suggest differences in access to health services [21]. The analysis carried out by Cazelli et al. (2002) also revealed differences among the regions in the availability of high complexity equipment, where the Northern and Northeast regions showed lower coverage than the national average [23].
Lima et al. (2002) suggests that one of the factors related to the differences found in the access to and use of health services among the regions may be the differences in the size and complexity of the service networks, which influence their availability [24].
Studies conducted in England and the United States also found a wide variability in the use of the AVF among different geographic regions of the countries, suggesting that this variation reflects differences in the practice standards adopted by nephrologists and vascular surgeons or renal unit teams; it may also suggest discrepancies between the offer and demand for renal units and the variations in the type or quality of the services offered [4, 25, 26].
The variations observed in the time until placement of AVF were associated with the type of renal units. This allows for the identification of an association between the renal service and the provision of the fistula, which can be related to differences in the practice standards of these services [14]. The variation found among the renal units confirms the importance of including evaluations of these services to identify points of reference for public policies that can impact the quality of health care for ESRD patients.
Linardi et al. (2003) found variations in catheter use among renal units that were distributed among 7 Brazilian states and suspected that structural differences among units, such as the location of the clinic in relation to the regional reference hospital, could influence the patients' profiles [27].
A study by Allon et al. (2000) also showed differences in the prevalence of fistulas among different renal units in a metropolitan area of the United States. However, the data collected in this study did not allow an evaluation the reasons of this variability, and stated that specific renal units have been able to increase the fistula frequency through the implementation of their own efforts [4].
Limitations inherent in the use of an administrative database should be considered for the present investigation because factors such as socioeconomic level, race or ethnicity, and co-morbid conditions could not be investigated. In addition, we used the dependent variable of the date of fistula surgery, and not the date of catheter removal, because in the National Database of RRT there are only the dates of the creation of the first AVF. However, we must consider that the date of catheter removal would be more clinically relevant, because even a successful fistula will not go into service for 6-8 weeks and many fistulas do not mature. Thus, the proportion of those with the first AVF surgery included some patients who have had their first surgeries but for whom the maturation of the AVF "failed".
The patients who had an undetermined cause for CKD were not excluded from the analysis because, as suggested by Moura et al. (2009), the exclusion of patients with this condition (42%) could compromise the present investigation [28]. Moreover, this study showed a deficiency in this subsystem of the Brazilian Public Health System for recording the causes of CKD. The proportion found in this study and in Sesso et al. (2008) was practically double the proportion reported in Spain and is higher than the proportion in the United States [11, 15, 18, 29].