IMPACT OF EARLY DIALYSIS INITIATION ON SURVIVAL OUTCOMES IN END‑STAGE RENAL DISEASE PATIENTS

Abstract

The evidence of the impact of dialysis on survival are inconclusive and the timing at which patients need to commence dialysis is a controversial issue. It is a systematic review of the available evidence since the IDEAL trial and it is geared towards the establishment of the relationship between the dialysis start time and mortality outcomes e.g. those that employ an advanced causal inference strategy.  Fourteen searches were performed in the MEDLINE, Embase, Cochrane Central and Web of science databases to identify 43 eligible studies that encompassed 1,247,892 patients. The critical discussion of the methodology of the studies and the significance of methods of bias reduction was provided such as propensity score matching, time-dependent covariate and instrumental variable. The estimation of the effect modification by patient characteristics, and estimate the pooled hazard ratios and subgroup analyses were done with the help of random-effects meta-analysis. The overall hazard ratio of all-cause death in the early and late initiation was 1.07 and it was very heterogeneous (I 2 = 89.4 percent). This was however found to be diluted to no significance when restricted to studies which used propensity score matching and not randomized controlled trials. Non-linear analysis revealed that there was the J-shaped exposure-response curve with the nadir at the estimated glomerular filtration rate of 8.5 mL/min/1.73m 2. Hypoalbuminemic patients (70 years and above), diabetes patients (they used to treat it much more in the past) were identified to have an improved nutritional status and age and diabetes effect were managed. Mortality due to an untimely initiation was skewed during the first 24 months of initiation, and then converged thereafter, in the time-varying test.Routine early initiation of dialysis is not supported by the evidence on the basis of estimated thresholds of glomerular filtration rate. The most suitable is the consideration of personal which is based on the nutritional well-being, age, diabetes and clinical symptoms. The correlations that would be found would heavily depend on the methodological rigor as well as the application of advanced techniques of causal inference would be critical in future research.