Background
Chronic kidney disease (CKD) is a growing health problem; the population prevalence has increased to an estimated 13% in the United States [
1]. Although CKD affects all ages there is a strong positive association with age: in individuals over 70 years, prevalence is almost 50% [
2]. With an aging population, it may be difficult to identify which patients would benefit most from referral to and follow-up by a nephrologist [
3]. Many diagnosed with stable CKD stage 3 can be followed up by primary health care after the initial nephrology evaluation. However, patients with advanced kidney disease and a high risk for renal replacement therapy (RRT) may benefit from nephrology care to facilitate the necessary preparations for RRT such as timely fistula surgery, if consistent with their preferences.
One risk factor for both RRT and death is a rapid decline in renal function. It was recently found that patients with renal function loss >30% over an initial period of two years had a higher risk of end-stage renal disease (ESRD) regardless of age and estimated glomerular filtration rate (eGFR) [
4]. However, this risk may differ in the “real-life” setting of unselected patients referred to nephrology care as opposed to patients included in various studies and general population cohorts. Many elderly patients can after a drop in eGFR, have stable eGFR levels for several years without the need for dialysis initiation [
5,
6]. In this study we aimed to investigate the absolute risk of RRT associated with a slow versus rapid disease progression in a nationally representative sample of nephrology-referred patients taking into account the competing risk of death.
Discussion
In this nationwide study of unselected, representative nephrology care patients in Sweden with eGFR <45 ml/min/1.73 m2 we found that clinical characteristics such as age, sex and CKD stage greatly influenced the risk of RRT and mortality before RRT initiation. Further, the preceding one-year progression rate was an important indicator of the future RRT risk. In general, elderly patients classified as slow progressors in our study had a low risk of RRT (about 25% within 5 years in CKD stage G5), while elderly with a fast progression still had much lower RRT risk than the younger patient of similar CKD stage.
Some previous studies have shown that both age and CKD stage influence the risk of RRT, but also that there is a strong interaction between them because of the competing risk of death [
3,
17]. The two largest of these studies were performed in different clinical settings (O’Hare [
17] used a cohort of American Veterans, whereas De Nicola [
3,
18] used nephrology- referred patients in a geographical region in Italy). In the US-based study, the risk of mortality before RRT initiation was higher compared with the Italian study. The results of our study in Swedish nephrology-referred patients, however, demonstrated a high mortality before RRT initiation in line with the study by O’Hare [
17]. It has been suggested that differences in RRT risk are attributed to the population background; general population cohorts have overall lower end-stage kidney disease risk as compared with specialized care cohorts [
4,
19]. However, our results suggest that the risk of RRT and end-stage kidney disease does not only depend on general versus referred population but also on other factors. These factors could be underlying population mortality (there is a strong north-south cardiovascular disease risk gradient across Europe [
20] which may explain the higher mortality before RRT in Sweden versus Italy), or they could instead reflect differences in attitude towards initiating dialysis in elderly individuals [
21], timing of dialysis initiation [
22] or referral pattern [
23]. For example, a recent comparison of the performance of the kidney failure risk equation demonstrated that there were quite substantial differences in the risk for end-stage kidney disease between various cohorts in Europe and the US [
24].
Decline in renal function has been associated with increased cardiovascular mortality [
25] and initiation of RRT [
26]. In younger individuals, low eGFR
per se is more predictive of end-stage kidney disease since it is more often the result of a specific kidney disease as compared with elderly where it is more often a marker of age-related conditions [
17]. Recently Coresh and coworkers demonstrated the importance of a rapid progression rate and suggested that a decline in GFR of 30% over two years could be used as an end-point in clinical trials [
4]. Our study is in agreement with these data, and indicates that also in elderly individuals the progression rate influences the risk of RRT but not the competing risk of death to the same extent. However, the absolute risk of RRT is still extensively influenced by the competing risk of death in elderly individuals, also under the circumstance of a fast progression rate - in our data shown as a decreasing absolute risk difference with age.
It has been proposed that trajectories of renal function could be used to predict timely RRT planning [
27]. In addition, early referral to a nephrologist is associated with better survival [
24,
28] and more timely preparation for dialysis care. However, it is equally important that the resources of nephrology care are directed to where it is most needed. Our results indicate that continued follow-up at the nephrology unit is most important for younger CKD patients, especially those with a fast progression rate who also exhibits a high absolute risk of RRT start within 5 years. In elderly patients (>75 years of age) the 5-year risk of RRT is low-moderate also at advanced stages of CKD and the 5-year risk of RRT is especially low in those with a slow progression rate. Thus, elderly slow progressors in CKD stage 3b-4 who do not suffer any major metabolic disturbances could be referred to primary health care with the proviso that they should be returned to the nephrology clinic if they progress to CKD stage 5 where the risk of future RRT increases. In these elderly patients, timely access planning also has to be weighed against the risk of dying with a functioning fistula before or shortly after RRT initiation since mortality continues to be high also the year after dialysis start [
29].
Our study has several strengths. It is the first study to use a national, representative cohort under stable nephrology care, where the treating nephrologists provided the primary renal diagnoses, and where clinical information was available both at inclusion and follow-up. Due to the use of linkages to national registries and the unique personal identity number of all Swedish citizens there were no losses of follow-up. The health-care system in Sweden is tax-funded, and there are no major differences in incidence of RRT throughout the country [
30]. Our study also has some limitations. Although 96% of the nephrology units in Sweden send data to the SRR-CKD, it has been estimated that about 76% of all nephrology-referred CKD stage 4–5 patients are included in the registry [
7]. In earlier CKD stages (3b), the completeness is unknown, since it is not mandatory to use the registry until eGFR drops below 30 ml/min/1.73 m
2. However, registrations should be systematic within each clinic, and bias due to selection would be possible only if clinics registering patients already at CKD stage 3 differ in their RRT initiation policy or mortality from those who do not. Another limitation is that we excluded patients with less than two creatinine measurements or who died within one year. Thus, our results are only generalizable to patients under stable nephrology care for more than one year. When considering all patients referred to nephrology care, the risk of death before RRT initiation may be even higher. Furthermore, ACR was not measured in all the patients since it was not a mandatory variable in the registry until 2013. The associations between ACR, progression and mortality are well described [
31]. However, in the sensitivity analysis in our study, adding ACR did not substantially affect our results regarding mortality. In addition, the progression rate would in itself consider some of the effect of ACR. Further research is needed to investigate whether adding progression rate as a predictor to the kidney failure risk equation could improve its performance and maybe explain some of the variation in incidence of RRT between the different CKD cohorts.
Conclusions
In this nationwide registry study of 8,771 referred patients with CKD stage 3b-5, the risk of RRT was high in younger patients with fast initial progression rate. Furthermore, the cumulative incidence of RRT was generally low among elderly, slowly progressing patients even in advanced CKD stages. Thus, for planning, treating and preparing the right patients for RRT, following the slope of eGFR is important.
Acknowledgement
The authors thank the Swedish Renal Registry for contribution of data and Anna-Lena Blom and Susanna Gabara, administrators at the SRR, for their help with data management.