Chronic renal failure

The purpose of this symposium is to introduce new approaches to analyze chronic disease data considering a specific example of organ failure, end-stage renal disease (ESRD) as an example. Four topics will be developed :
1. An introduction to the epidemiology of chronic kidney disease, a model of chronic disease, with a focus on the demand and offer of care for end-stage renal disease and the new challenges for a better aid to public health decision making. P. Landais.
2. Then an overview of Bayesian statistical methods will be presented related to geographical epidemiology for chronic diseases by L. Fortunato.
3. In a second part a new approach of modelling excess mortality (additive model) and relative mortality (multiplicative model) in the context of chronic renal failure introducing a combined model by C. Elie.
4. And finally, modelling the multi-state evolution of kidney transplant recipients for improving the prognostic decision by Y. Foucher.

1. Chronic kidney disease is very frequent, affecting almost 10% of the general population. Chronic kidney failure may progress towards end-stage renal disease (ESRD) and renal replacement therapy. Fortunately, dialysis or renal transplantation may replace kidney failure. More than 35,000 patients are dialyzed and nearly 26,000 live with a transplanted kidney in France. However, in the last years, patients entering ESRD became older and with an increasing number of associated morbidities, peculiarly cardiovascular. They are thus at high risk of morbidity and mortality while on replacement therapy. Moreover, this treatment is costly and requires a tight organization of the dialysis units all over the territory. Kidney procurement is not sufficient to transplant all the patients registered on the waiting list every year. Adapting the offer of care requires a tight understanding of the demand. There is thus a stake of public health in our countries. The topics developed in this mini-symposium are dedicated to better articulate clinical epidemiology, biostatistics and needs for a better public health adaptation of the offer of care.

2. Small area disease mapping studies have become an established technique in spatial epidemiology with the advent of routine health data indexed at a fine geographical resolution. In particular, they can highlight sources of heterogeneity underlying spatial patterns in the health outcomes and consequently are able to suggest important public health determinants or etiologic clues. Such models allow area-specific disease relative risks to be ‘smoothed’ towards global and/or local mean levels across the study region, e.g. using the convolution model proposed by Besag, York and Mollié. Several issues relating to disease mapping can occur such as the specification of the spatial dependency or the incorporation of covariates (risk factor or confounder). An overview of the main classes of models that have been used for disease mapping within a Bayesian estimation paradigm will be presented. Recent extensions to model the joint spatial distribution of multiple diseases or to analyse the spatio-temporal variations of disease will be also developed. The different methods will be illustrated.

3. End-stage renal disease requires renal replacement therapy. Dialysis or transplantation may be indicated. Often several types of treatment may be proposed consecutively in the personal history of a given patient for instance, peritoneal dialysis first, then transplantation and back to hemodialysis in case of progressing insufficiency of the transplanted kidney along with time. ESRD is associated with ageing and comorbidities and high cardiovascular morbidity. Background mortality is particularly relevant in the context of chronic diseases and requires appropriate methodological approaches. C. Elie proposes modelling ESRD mortality, comparing excess mortality (additive model) and relative mortality (multiplicative model). A model combining these two components is then proposed. A cohort of nearly 10,000 patients who started renal replacement therapy by dialysis from 2002 to 2007 for treatment of end-stage renal disease was studied. Generalized linear models were used to modelling relative and excess mortality. The combined model was designed on the basis of Aalen additive hazard model. This study underlines the interest of developing criteria to orient the choice of the most appropriate model, either additive or multiplicative. A new approach is proposed which combines both components in the same model.

4. If we concentrate the problematic on kidney transplant recipients, the excellent one year graft survival rate in kidney transplantation, due to the introduction of new drugs and new strategies after surgery, has led to a new challenge: the development of early surrogate endpoints for long-term graft survival. The first difficulty is the possibility of two terminal failures: back to dialysis in case of graft rejection, or patient death. These two events may be differentially correlated with surrogate markers. The second difficulty is the impossibility to discriminate deaths dependent on the transplantation itself from independent causes. In order to solve these difficulties, Y. Foucher proposed to develop the ROC theory in the context of two competing failure times with a relative survival part for the mortality modelling. The analysis was performed on the basis of the French data bank of kidney transplant patients (DIVAT). Moreover, in this study, he showed that this new marker predicts the long-term graft survival better than classical markers such as creatinine clearance.

Paul Landais (MD PhD) is epidemiologist and nephrologist at Paris Descartes University, Necker Enfants Malades hospital and created the national Renal Epidemiology and Information Network (REIN), which registers all the French patients entering ESRD with their follow-up.
Léa Fortunato (PhD) is Research Associate, Division of Epidemiology, Public Health and Primary Care, Imperial College, London, UK.
Caroline Elie (MD BSc) is research fellow of biostatistics and clinical research at Paris Descartes University.
Yohann Foucher (PhD) is biostatistician. His PhD degree was directed by Pr JP. Daurès Montpellier University, France. He is today involved in renal transplantation research and modelling at the Institute for transplantation and transplantation research (ITERT) at Nantes University, France.