Confounding factors for early death in incident end-stage renal disease patients: Role of emergency dialysis start.

Hemodialysis (HD) has been associated with higher 1-year mortality than peritoneal dialysis (PD) after dialysis start. Confounding effects of late referral, emergency dialysis start, or start with central venous catheter on this association have never been studied concomitantly. Survival was studied among the 495 incident dialysed patients in our department from 1995 to 2006 and followed at least 1 year until December 31, 2007. Nested Cox models adjusted on patient characteristics explored factors associated with 1-year and ≥1-year mortality. Hemodialysis patients were 332 (67.1%), 104 (21.0%) were late referred (<6 months), 167 (33.7%) started dialysis in emergency, and 144 (29.1%) started with central venous catheter. When adjusted only on age, sex, and comorbidities, HD was associated with poor 1-year outcome: adjusted hazard ratio (aHR) for death in HD vs. PD was 1.77, P=0.02. In fully adjusted model, among first dialysis feature variables, only emergency dialysis start was significantly associated with 1-year mortality: aHR 1.53, P=0.02. Dialysis modality was not associated with 1-year mortality rates in this fully adjusted model: aHR in HD vs. PD became 1.03, P=0.91. In ≥1-year period, HD was associated with lower mortality than PD (aHR 0.61, P=0.004), whereas other first dialysis features were not associated with death. Other factors associated with death were age, type 2 diabetes, peripheral vascular disease, heart failure, and hepatic failure. Negative association between HD and 1-year survival on dialysis was explained by confounders. Emergency dialysis start was strongly associated with early mortality on dialysis. Its prevention may improve patient survival.