Factors associating with differences in the incidence of renal replacement therapy among elderly: data from the ERA-EDTA Registry

Abstract

INTRODUCTION

The European population is ageing and the proportion of individuals ≥80 years of age has increased from 3.9% in 2004 to 5.1% in 2014 [1]. Simultaneously, the number of patients on renal replacement therapy (RRT) ≥75 years of age has nearly doubled, comprising 30% of all patients entering RRT in Europe. However, in 2014 the incidence of RRT per million age-related population (PMARP) among inhabitants ≥75 years of age varied considerably between European countries, ranging from 177 in Finland to 898 in Greece [2 p.38, 3 p.33]. This has raised the question whether a high incidence of RRT in the elderly population is related to widespread use of dialysis in frail patients with multiple comorbidities and poor survival.

Elderly RRT patients have inferior survival compared with younger patients [2 p.64, 4] and cognitive decline as well as functional impairment and frailty are associated with adverse health outcomes [5]. The probability of receiving a kidney transplant is lower in the older age groups, whereas most elderly RRT patients are treated with in-centre haemodialysis [2 p.61], which is the most expensive treatment modality [6]. Among elderly patients with comorbidities, dialysis may not provide a survival advantage or improve health-related quality of life over conservative management of end-stage renal disease (ESRD) [7–9], so careful selection of patients who are likely to benefit from RRT is essential.

The aim of this study was to search for explanations for the notable differences in RRT incidence in the elderly population by studying the characteristics and survival of patients starting RRT in European countries.

MATERIALS AND METHODS

Patients ≥75 years of age who started chronic RRT between 2010 and 2013 were identified from 13 national (Austria, Bosnia and Herzegovina, Denmark, Estonia, Finland, France, Greece, Iceland, Norway, Romania, Serbia, Sweden and the Netherlands) and 16 regional (Dutch- and French-speaking Belgium, UK Scotland, and the Spanish regions of Andalusia, Aragon, Asturias, Basque Country, Cantabria, Castile and León, Castile-La Mancha, Community of Madrid, Catalonia, Extremadura, Galicia, Murcia and Valencia) registries providing individual patient-level data to the European Renal Association–European Dialysis and Transplant Association (ERA-EDTA) Registry. Data were only available from Bosnia and Herzegovina and Serbia in 2011–2013, from the Spanish region of Murcia in 2012–2013 and from Estonia in 2013. Twelve registries (Austria, Dutch- and French-speaking Belgium, Denmark, Finland, Iceland, Norway, Sweden, and the Spanish regions of Aragon, Catalonia, Galicia and Valencia) reported additional data on comorbidities at the start of RRT. Patients 20–74 years of age were selected as a control group.

The incidence of PMARP was defined as the number of patients starting RRT annually divided by the mid-year age-related general population. Patients were divided into two groups according to low (≤500 PMARP) or high (>500 PMARP) registry-level incidence of RRT in the age group ≥75 years. This division was selected to balance the number of patients and registries between the low- and high-incidence groups.

Demographic and clinical variables such as age, sex, primary renal disease, initial RRT modality, survival, comorbidity (diabetes, ischemic heart disease, peripheral vascular disease, cerebrovascular disease and malignancy) and body mass index (BMI) at the onset of RRT were correlated to the RRT incidence. Associations of life expectancy at birth [10] and gross domestic product (GDP) per capita [11] with RRT incidence were analysed. The reported causes of death were divided into eight groups: cardiovascular, cerebrovascular, infection, suicide, refusal or withdrawal from RRT, cachexia, malignancies, miscellaneous and unknown/unavailable.

When comparing distributions of variables between low- and high-incidence RRT registries, the chi-square test was used for categorical variables and the Mann–Whitney U test for continuous variables. Linear regression was used to assess the association between continuous variables and RRT incidence by registry. Survival probability was assessed using the Kaplan–Meier method and differences in survival by the log-rank test, while the relative risk of death as a function of risk factors was estimated using Cox proportional hazards regression. Comorbidities were entered into the model as single items. Patients from Romania were excluded from the survival analysis because complete RRT history was not available for all patients. The survival time was calculated from the first day of RRT (dialysis or pre-emptive kidney transplantation) and the patients were followed until death (n = 19 413), censoring at loss to follow-up (n = 486), recovery of renal function (n = 1079) or end of the follow-up period on 31 December 2014 (n = 16 632). Two-sided P-values <0.05 were considered statistically significant. Statistical analyses were performed using SPSS Statistics 23 (IBM, Armonk, NY, USA).

RESULTS

Incidence

In the European countries and regions reporting individual patient data to the ERA-EDTA Registry, a total of 38 457 patients ≥75 years of age entered RRT between 2010 and 2013. The mean annual incidence of RRT in the age group ≥75 years in 2010–2013 was 537 PMARP (Table 1). The incidence of RRT in the age group 20–74 years was 137 PMARP and varied considerably between countries and regions, ranging from 75 PMARP in Estonia to 175 PMARP in French-speaking Belgium. However, the difference between registries was much greater in the age group ≥75 years, with the incidence of RRT ranging from 157 PMARP in Estonia to 924 PMARP in Dutch-speaking Belgium (Figure 1). The variation further increased in the age group ≥85 years, in which the incidence ranged from 27 PMARP in Spain (Cantabria) to 755 PMARP in Dutch-speaking Belgium. In the age group ≥75 years the incidence rate ratio of RRT between men and women varied markedly, from 1.5 in Bosnia and Herzegovina to 4.4 in the Basque Country. The incidence rate ratio of RRT between the age groups ≥75 years and 20–74 years was considerably larger in registries with a higher incidence of RRT in the elderly (Table 1).

FIGURE 1

Incidence of RRT PMARP in 2010–2013 by age group.

Open in new tabDownload slide

Patient characteristics

Table 2 shows the characteristics of the patients in registries with low and high incidences of RRT. Twenty-two registries and 13 033 patients were included in the low-incidence group and seven registries and 25 424 patients were in the high-incidence group. The median age of the entire cohort of patients ≥75 years was 80.5 [interquartile range (IQR) 77.7–83.9] years at the start of RRT. Patients from the low-incidence areas were younger and more frequently had peritoneal dialysis as the initial treatment modality. Sex and BMI distributions were fairly similar in both incidence groups. Hypertension was a more common cause of ESRD in the high-incidence compared with the low-incidence group. The rate of kidney transplantation within 1 year from the onset of RRT was small in this cohort of elderly patients but was greater in the low-versus the high-incidence group.

Registries with an incidence of RRT in elderly patients in the upper range showed a significantly higher median age at the initiation of RRT, whereas the ratio between males and females, the life expectancy at birth and GDP per capita did not correlate with RRT incidence (Figure 2).

FIGURE 2

Association between explanatory factors and incidence of RRT in patients ≥75 years of age.

Open in new tabDownload slide

Survival

A cardiovascular cause of death was most common in both incidence groups, 28.0% in the low-incidence group and 31.4% in the high-incidence group. The distribution of causes of death was different between the low- and high-incidence groups (P < 0.001). The difference was mainly due to a greater percentage of patients dying from cardiovascular and cerebrovascular causes in the high-incidence registries and a lower percentage of patients dying due to suicide or refusal or withdrawal of treatment.

Survival among the 36 807 patients ≥75 years of age was worse in the older age groups. One-year survival from the onset of RRT was 74% [95% confidence interval (CI) 73–74] in the whole cohort and for the age groups 75–79 years, 80–84 years and ≥85 years the survival rates were 78% (95% CI 77–78), 73% (95% CI 72–74) and 66% (95% CI 65–68) (P < 0.001), respectively. Among patients ≥75 years of age the 1-year survival was 76% (95% CI 75–76) in the low-incidence group and 73% (95% CI 72–73) in the high-incidence group (P = 0.04) (Figure 3).

FIGURE 3

Survival from onset of RRT by low- and high-incidence RRT groups.

Open in new tabDownload slide

The relative risk (RR) of death was 1.03 (95% CI 1.002–1.065) in the high- compared with the low-incidence group. After adjustment for age and gender, there was no difference in survival between the low- and high-incidence groups [RR 0.97 (95% CI 0.94–1.00)]. Further adjustment for initial RRT modality and the cause of ESRD did not affect the RR of death [RR 1.00 (95% CI 0.97–1.03)].

Comorbidities

A total of 10 714 patients from 12 registries provided comorbidity data at the start of RRT. Basic characteristics of patients in registries with low and high incidence of RRT were similar compared with the whole cohort. Diabetes, ischaemic heart disease, cerebrovascular disease and malignancies were more common among patients in the low-incidence group, whereas peripheral vascular disease was more common in the high-incidence group (Table 2). In the low-incidence group, 76% of the patients had at least one comorbidity at the onset of RRT compared with 63% in the high-incidence group (P < 0.001).

In survival analysis of patients who had data available on all five reported comorbidities (n = 9014), the risk of death was lower in the high-incidence group [RR 0.76 (95% CI 0.71–0.80)] than in the low-incidence group. All of the comorbidities and older age at onset of RRT were associated with worse survival. When age at onset of RRT, sex and comorbidities were added into the multivariable model, the RR of death among patients in the high-incidence group did not change [RR 0.75 (95% CI 0.71–0.79)].

DISCUSSION

This study showed the incidence of RRT to vary significantly between European countries. Strikingly, in the age group ≥75 years the difference between the countries with the highest and lowest incidence was almost sixfold, whereas among patients 20–74 years of age the difference was only twofold. We were unable to find any plausible explanations for this difference. Notably, there was no association between wealth, life expectancy, comorbidities or other patient characteristics and the incidence of RRT among the elderly. The only exception was age at the onset of RRT, which was greater in high-incidence areas. Our study also revealed that the survival of elderly patients was remarkably similar in the low- and high-incidence areas. In the subset of patients with available comorbidity data, the risk of death was higher in the low-incidence group, and this did not change after adjustments. This reflects the fact that the comorbidities, all of which were associated with worse survival, were surprisingly more common in the low-incidence group, despite the older age in the high-incidence group. This is in line with a study from France showing lower comorbidity among older age groups of dialysis patients [12]. Taken together, our findings suggest that differences in patient selection and timing of RRT may account for at least part of the observed variation in incidence.

The main strength of this study is the large and comprehensive cohort from the 29 registries that includes complete data on age, sex, primary renal diagnosis, initial RRT modality and survival. Hence selection bias was minimal in the main analyses. In contrast, data on comorbidities were available from only 12 registries and were complete for only 84% of patients. Among these patients the risk of death was higher in the low-incidence registries, in contrast to the results from the whole study population. This limitation may have introduced a bias into the analyses. The higher incidence of comorbidities in the low-incidence registries might reflect a higher prevalence of comorbid conditions in the general population in these countries. There could also be differences in the practice of collecting and reporting comorbidity data or in evaluating pre-dialytic patients for comorbidities between countries and regions. The proportion of patients who receive conservative treatment of kidney failure might also vary. The timing of the onset of RRT is likely to have an impact on the incidence of RRT among the elderly, as their mortality rate is high and death is an important competing risk for starting dialysis [13–15]. Unfortunately we did not have information on estimated glomerular filtration rate before the start of RRT or on the prevalence of pre-dialytic renal disease to explore this issue. Notably, an early start of RRT has not been shown to improve survival [16], and in fact an early start has been reported to associate with worse survival in elderly patients because of their high rate of comorbidity [17, 18]. Variations in practices of acceptance of patients into RRT and conservative management, profit versus non-profit haemodialysis facilities, pre-dialysis management and competing mortality may all play an important role in explaining differences in RRT incidence, but these were not evaluated in this study.

Caskey et al. [19] reported that GDP per capita and money spent on health care and dialysis facilities were positively associated with RRT incidence rates across 46 countries worldwide. They also showed that a higher incidence of RRT associates with private for-profit haemodialysis facilities in developed countries, which is in line with a study of conservative care for ESRD in 11 European countries [20]. We were unable to show an association between GDP per capita and RRT incidence in the ageing population. This may potentially be due to smaller differences in GDP between the countries in our analysis. Furthermore, Visser et al. [21, 22] showed that the higher incidence of RRT in the elderly in Flanders compared with the Netherlands does not seem to be due to a more restrictive referral policy or lower access to RRT in the Netherlands. They suggested that the difference was more likely due to differences in comorbidity and lifestyle.

The differences in RRT incidence among the elderly could be partly explained by diverse usage of conservative management in the care of patients with ESRD. A study from the UK revealed that there is much variation in the way conservative management of ESRD is provided in 67 UK renal units [23]. This variability may be even more notable between countries. Several studies have reported that dialysis improves the survival of elderly patients [24–26]. In addition, better management of patients during the progression of chronic kidney disease could reduce the number of comorbidities and increase the number of healthier elderly patients who benefit from dialysis [27]. However, when older age is combined with greater comorbidity, the prognosis is similar with dialysis and conservative management [9, 28, 29]. Despite this, the level of comorbidity had no effect on the decision to initiate dialysis in elderly patients in a recent study from the USA [30]. Although dialysis improves the survival of elderly patients with ESRD, the treatment is associated with higher rates of hospitalization [31]. This, together with the burden of dialysis therapy, may explain why declining quality of life is frequently observed in these patients [25, 32]. In fact, the quality of life is often better preserved in patients receiving conservative care for ESRD [24]. Consequently, dialysis should not be expected to improve the quality of life in elderly patients with comorbidity [8, 25]. Information provided to patients by nephrologists and other renal unit staff plays an important role when older adults are choosing between dialysis and conservative management [33], but according to a study from the Netherlands, the decision seems to be based on personal values, beliefs and feelings towards life, suffering and expected difficulties on RRT rather than on the effectiveness of the treatment [34]. Dialysis is expensive, especially in the elderly, and is estimated to cost up to €90 000 annually for each patient [35]. Therefore the decision to start dialysis in an elderly patient with substantial comorbidity should only be made after careful consideration and the patient should be involved in the decision making [36, 37].

CONCLUSION

This European study shows that in areas with a high incidence of RRT among the elderly, the age at RRT start was slightly higher, while age-adjusted survival was similar compared with areas with low incidence. Data on comorbidities were available only for a subset of patients and comorbidities were more common in patients from the low-incidence areas. Factors explaining the differences in RRT incidence could not be elucidated and therefore further studies are reqiured. However, our findings may suggest that differences in patient selection and timing of RRT may explain some of the variation in incidence. An inappropriately high-acceptance rate to dialysis is very costly for society and may not improve the survival or quality of life of patients. On the other hand, the criteria for initiating dialysis in the elderly might be too strict in the low-incidence countries and regions. The fact that incidence of RRT among the elderly is not associated with survival raises the question whether the policy of high or low acceptance rate on RRT is correct. More information is needed to guide treatment decisions for ESRD in the elderly.

ACKNOWLEDGEMENTS

The authors would like to thank the patients and the staff of the dialysis and transplant units for contributing the data via their national and regional renal registries. Furthermore, the authors gratefully acknowledge the following registries and persons for their contribution of data: Austrian Dialysis and Transplant Registry [OEDTR] (R. Kramar); Dutch-speaking Belgian Society of Nephrology [NBVN] (B. De Moor and F. Schroven); French-speaking Belgian Society of Nephrology [GNFB] (JM. des Grottes and F. Collart); Renal Registry Bosnia and Herzegovina (L. Lukić, and S. Corić); Danish Nephrology Registry [DNS]; Estonian Society of Nephrology (M. Rosenberg and K. Lilienthal); Finnish Registry for Kidney Diseases; Epidemiology and Information Network in Nephrology [REIN] (M. Lassalle); Hellenic Renal Registry (N. Afentakis); Icelandic End-Stage Renal Disease Registry; Norwegian Renal Registry (T. Leivestad, A.V. Reisæter, and A. Åsberg); Romanian Renal Registry [RRR] (G. Mircescu, L. Garneata and E. Podgoreanu); Serbian Renal Registry (Working Group of Serbian RRT Registry and all of the Serbian renal units); Swedish Kidney Registry [SNR] (K.G. Prütz, M. Stendahl, S. Schön, L. Bäckman and M. Segelmark); Dutch End-Stage Renal Disease Registry [RENINE] (M. Hemmelder and A. Hemke); Scottish Renal Registry [SRR] (all of the Scottish renal units); and the regional registries of Andalusia [SICATA] (P. Castro de la Nuez), Aragon (J.I. Sanchez Miret), Asturias (R. Alonso de la Torre, J.R. Quirós, and RERCA Working Group), Basque Country [UNIPAR] (Á. Magaz, J. Aranzabal, M. Rodrigo and I. Moina), Cantabria (M. Arias Rodríguez and O. García Ruiz), Castile and León (R. González and C. Fernández-Renedo), Castile-La Mancha (G. Gutiérrez Ávila and I. Moreno Alía), Catalonia [RMRC] (E. Arcos, J. Comas and J. Tort), Extremadura (J.M. Ramos Aceitero and M.A. García Bazaga), Galicia (E. Bouzas-Caamaño and J. Sánchez-Ibáñez), Community of Madrid (M.I. Aparicio de Madre), Renal Registry of the Region of Murcia (I. Marín Sánchez) and Valencian region [REMRENAL] (C. Alberich Martí and M. Ferrer Alamar); and the other ERA-EDTA Registry committee members not mentioned above for their advice in the analysis and the drafting of this article: A. Więcek, F.J. Caskey, J.W. Groothoff, J. Harambat, F. Jarraya and I. Rychlik; and M. Pippias and V.S. Stel in the AMC Registry office for data collection and management. This article was written by J.H., A.K., J.M.A.-D., C.C., G.d.A., J.d.M., M.E., F.G., C.G.-R., J.G.H., V.L., M.N., R.P., U.P., H.R., R.S., C.S.d.P., Z.A.M., O.Z., K.J.J and P.F on behalf of the ERA-EDTA Registry, which is an official body of the ERA-EDTA.

FUNDING

The ERA-EDTA Registry is funded by the ERA-EDTA. J.H. was funded by the ERA-EDTA and Finnish Medical Foundation.

AUTHORS’ CONTRIBUTIONS

J.H., A.K., K.J. and P.F. contributed to the literature search, study design, data collection, data analysis, data interpretation and writing. All co-authors contributed to data collection, data interpretation and writing. All authors have seen and approved the final version of the article.

CONFLICT OF INTEREST STATEMENT

None declared.

REFERENCES