Despite the lack of difference on the primary endpoint, the main finding of our study is that patients treated in a dedicated renal care network were less frequently hospitalized both during the year before and after dialysis start. Moreover, they enter dialysis with better control of serum phosphate, higher proportion of normal serum albumin and received more frequently educations and information’s sessions on CKD and dialysis techniques. Finally, they progressed more slowly towards ESRD. This could have a major impact on the quality of life of the patients, and on the economical burden of renal care .
Regarding the impact of MDRC on the early survival of incident dialysis patients, our results differ from previous studies. In a case–control study, Curtis and colleagues showed that exposure to MDRC of CKD patients was associated with a better survival . However, in Curtis et al. Study the mortality rate was rather high likely because of a greater proportion of diabetic patients. Despite the lack of information regarding other comorbidities, it cannot be excluded that patients from the study by Curtis et al. were at higher risk than ours, resulting in an enhanced effect of therapeutic intervention . Similarly, in a cohort study, Goldstein and colleagues found a better survival and fewer hospitalizations in patients followed in MDRC before dialysis, as compared to patients receiving standard care . This difference could be also due to a higher basal risk of death and cardiovascular events, because of a higher proportion of patients with diabetes and/or previous cardiovascular diseases. In line with the hypothesis that the effect of network care was minimized by the recruitment of patients at relatively low risk, the comparison with the whole population of contemporary incident dialysis patients within the REIN Registry showed a lower prevalence of diabetes (22.5% versus 41%) and heart failure (17.5% versus 28.1%) amongst the study patients, resulting in a lower mortality rate during the first year on dialysis (13.8% versus 17%) . In addition, in our study, the rate of use of cardiac and renal protective medications was higher (almost 93% for blockers of RAS) than previously observed , and did not differ between the two groups. This may have contributed to the lack of differences on the primary endpoint. The percentage of patients taking an anti-platelet agent was similar between the two groups, close from the figures reported previously [26–28]. Finally, the length of follow-up in our study could have been too short to detect differences on major outcomes. It is noteworthy that Curtis and colleagues found only a slight difference on survival curves after one year of follow up  between the MDRC and the control groups.
Network patients required fewer individual hospitalizations before and after the beginning of dialysis, as well as for first dialysis session. There was a trend towards a decrease of unplanned first dialysis and an increase of the proportion of sustainable access used for this first dialysis (Table 4). This occurred despite a higher proportion of patients with usable and sustainable access for first dialysis in the control group comparatively to the mean proportion in the REIN registry (69.2% vs 63%) , that suggests a high quality of care for the creation of dialysis access.
The eGFR at dialysis initiation in network patients was similar to that observed in the REIN registry (9.6 ml/min/1,73m2) , but it was lower in the control group. This difference in average eGFR at dialysis initiation has had probably no impact on the primary outcome and on the one-year mortality rate. Indeed, the IDEAL study showed that there were no difference in the one-year mortality rate between two randomized groups of patients who started dialysis respectively with 9.0 and 7.2 ml/min/1,73m2 of average eGFR . Even if the impact of this difference is not known on the occurrence of cardiovascular events during, it seems unlikely that there could be an impact on the primary outcome of our study. So patients from the control group were theoretically more prone to develop metabolic and nutritional disorders, as shown by the lower proportion of patients with normal serum phosphate and albumin levels at dialysis initiation (Table 2). However, some others relevant biological parameters were not different between groups. The decreased rate of hospitalizations is likely related to a lower occurrence of complications, including those requiring unplanned dialysis like electrolytic disorders or acute pulmonary oedema. The incidence of these complications could be positively affected by educational sessions about lifestyle changes and by visits with trained dieticians, which were more frequent in the network patients.
An interesting result of our study is that network patients had a slower decline of renal function (Figure 3), with the dialysis start virtually postponed by more than 4 months. Those results are consistent with previous findings by Devins et al. and more recently by Bayliss et al. [14, 30]. Devins and colleagues showed in a randomized controlled trial that a psycho-educational intervention added to usual care allowed delaying dialysis start by 3 months. Educational sessions were also included in our network care and may contribute to the observation of close results between the study by Devins et al. and ours . The results of the observational study by Bayliss and colleagues showed a slow-down of eGFR decrease. This study included patients with a large proportion of diabetes and obesity, a higher baseline eGFR and slower decline than in our study. The control of diabetes and BP was similar in the two groups, suggesting an impact of the MDRC intervention on other progression factors such as dietary habits or use of nephrotoxic medications .
The proportion of patients with hereditary diseases was higher in the network group (Table 1). Differences in the distribution of primary nephropathies might have had an influence on the results. However, in advanced stages of CKD, the nature of primary renal disease has only a minor influence on the slope of eGFR decrease . Therefore, the differential progression of CKD is more likely related to the difference in renal care organization.
Our study was designed to assess the effects of coordinated renal care within a distributed network involving healthcare professionals working in the community. At the difference of centralized MDRC in which the services are provided within a same centre [15, 16], the network allows the patients to maintain ambulatory encounters with family practitioners. In addition, it provides a biological monitoring closer to CKD management guidelines than usual renal care (Table 2). This could explain the favourable impact of network care on CKD progression. A recent study by Hotu et al. showed that in diabetic and hypertensive CKD patients, a community based model of care leads to a higher decrease in proteinuria and BP than in patients receiving usual care . To our knowledge, this issue was addressed by only one observational study . The authors showed that a distributed network could improve the progression of CKD in patients with stage 3–5 CKD comparatively to usual care. However, in this study, renal follow up was only performed by the GP, without visits with the nephrologist who gave the treatment adaptations remotely . This is quite different from our approach because the absence of direct contact between patients and nephrologist may affect the clinical evolution of the patients.
This study describes for the first time in France the effects of coordinated renal care provided by a dedicated network. Although accurate epidemiological data on the treatment of stages 4–5 CKD are lacking, data from the REIN registry show a trend towards a stabilization of the incidence rate of ESRD over the five last years. However, the prevalence of CVD remains high in incident patients, contributing to their early mortality: French data network REIN showed that the risk of death is highest in the first year after the start of dialysis treatment (17% for the whole population), especially in older patients . In a context of a public social insurance allowing virtually unlimited access to renal caregivers, a way to improve the results of the health system could be to implement new therapeutic strategies based on care coordination through renal care networks. Moreover, the French public health system, as in several countries, is actually subject to an economic pressure because the resources allocated tend to become limited and care cost is becoming more expensive (due to the aging of the population and to several others reasons like incoordination of some provided care). Therefore care network could also help to reduce the costs related to the management of ESRD, which is presently crucial because it is a costly disease.
These results must be interpreted in the light of some limitations of the study. The two main limitations are related to the design of the study. This was an observational matched cohort study and although patients were matched on important confounders, it cannot be excluded that some others confusion parameters may have influenced the results. Another limitation was the difficulty to collect retrospective data from 24h urine measurements that precluded any analysis of the relationship between proteinuria and outcomes. However, the predictive value of proteinuria on cardiovascular events has been mainly assessed in earlier CKD stage patients  and it is unclear if the inclusion of proteinuria would have changed the meaning of the results in our late 4–5 CKD stage patients. Finally, one other limitation to mention is that our study was only interested by patients with stages 4–5 CKD. It can be speculated that the results could have been different with patients who wouldn’t have started dialysis (being treated only conservatively) or in patients with less severe CKD (i.e. stage 3 CKD).