Determination of a Profile of Patients with Renal Failure to Be Supported As a Priority in the Care Pathway in Order to Avoid Emergency Hospitalizations.

This observational study on retrospective data does not justify a specific calculation of numbers. Over the 27-month observation period, have been identified in the department an active file of approximately 440 stage 4 or 5 chronic kidney failure patients eligible for the study. The qualitative variables will be expressed in numbers and corresponding percentages, and compared by Fisher's exact test. Quantitative variables will be expressed by their mean and standard deviation or by their median and interquartiles, depending on their distribution. They will be compared by the Student test in the case of a normal distribution, by the Wilcoxon test otherwise. The main outcome criterion, namely the occurrence of hospitalization during the observation period, will be analyzed using a logistic regression model. The covariates tested in the model will be : * Age * Gender * Helpers among: Parent, Sibling, Child, Spouse, neighbors, friend, Social care (social security (SSC), mutual insurance or universal health coverage (CMU) * Active, disabled, retired * MRC Stage (4 or 5) in 2020 * Mode of entry into nephrology consultation (after hospitalization in Nephrology, attending physician, diabetologist, cardiologist, other) * History and associated comorbidities including high blood pressure, diabetes, dyslipidemia, vascular pathologies, heart disease, sleep apnea syndrome, obesity, other chronic pathologies. * Number of chronic pathologies of the patient * The pathology(s) retained by the nephrologist as responsible for the kidney disease (several pathologies can be retained and reported as such) * Number of treatments on the patient's prescription * Treatments: hypotensive, antidiabetic, hypolipidemic, diuretic, potassium chelator, cardiac treatment, antithrombotics, others * Number of years of follow-up in nephrology at the CHSF before the period studied. * Status of known patients at the end of follow-up (it can be before 03/31/2022 if the patient has moved to another status before) among: consultation, dialysis (urgent or normal), transplant, death, lost to follow-up . * Did the patient benefit from pre-dialysis information, at what Glomerular Filtration Rate (GFR) and what was his choice (peritoneal dialysis, hemodialysis or conservative treatment) * Did the patient benefit from an arteriovenous fistula, at what GFR * Did the patient benefit from a pre-transplant assessment * The number of consultations by a nephrologist, in consultation or in day hospital. * The number of consultations canceled by the patient. * The number of consultations not honored by the patient. * Le nombre de consultations sans bilan récent. * The GFR at the consultation * Unscheduled hospitalizations at the CHSF during follow-up, whether in nephrology or in another hospital department. * The reason(s) for unscheduled hospitalizations related to renal pathology (cardiac decompensation, anemia, sodium and water overdose, acute lung edema (APO), high blood pressure (hypertension), emergency dialysis starts, renal failure acute (ARI), and ionic disorders,) * The number of days for each hospitalization. The log-linearity of the quantitative variables will be evaluated by graphical method. In the event of non-log-linearity, the variable will be discretized according to its median or according to the inflection points of the curve. Covariates whose unadjusted Odds Ratio has a p value less than 0.20 will be retained in the multivariate model. All second-order interactions will be tested. Multicollinearity phenomena will be evaluated by measuring variance inflation factors, with a positivity threshold set at 2.5. A variable selection using a descending step-by-step method minimizing the Akaiké criterion will then be carried out. The area under the ROC curve of the final model as well as the p value of the Hosmer-Lemeshow test will be specified. In secondary analyses, patients who were not hospitalized during the observation period will be compared to those who had 1 to 3 hospitalizations and to those who had 4 or more hospitalizations. Qualitative variables will be analyzed using Fisher's exact test. The quantitative variables will be compared by analysis of variance if its conditions of application are met, by the Kruskal-Wallis test otherwise. In the event of significance, comparisons 2 by 2 will be carried out, respectively by the Student test or the Wilcoxon test, with adjustment of the value of p by the Bonferroni-Holm correction in order to take account of the multiple comparisons. All the tests will be carried out bilaterally, with a 1st type risk set at 5%. The analysis will be performed using R software (version 4.3.1 © 2023 The R Foundation for Statistical Computing). Management of missing data: no plan to replace missing data for this observational data study, the design of which in any case makes the presence of missing data in numbers unlikely, particularly concerning the main criterion.