Technique Failure in a Multicenter Canadian Home Hemodialysis Cohort.

Abstract

RATIONALE & OBJECTIVE\nIncreasing uptake of home hemodialysis (HD) has led to interest in characteristics that predict discontinuation of home HD therapy for reasons other than death or transplantation. Recent reports of practice pattern variability led to the hypothesis that there are patient- and center-specific factors that influence these discontinuations.\n\n\nSTUDY DESIGN\nRetrospective cohort study.\n\n\nSETTING & PARTICIPANTS\nIncident home HD patients at 7 centers in Canada between 2000 and\xa02010.\n\n\nPREDICTOR\nTreatment center, case-mix, and process-of-care variables.\n\n\nOUTCOMES\nTechnique failure (defined as discontinuation of home HD therapy for any reason other than training failure, death, or transplantation) and mortality.\n\n\nANALYTICAL APPROACH\nRegression modeling of technique failure using Cox proportional hazard models adjusting for treatment center and modifiable and nonmodifiable patient-level variables, censored for death and transplantation.\n\n\nRESULTS\nThe cohort consisted of 579 patients. Mean age was 49.9±14.1 years, 74% were of European ancestry, median dialysis vintage was 1.9 (IQR, 0.6-5.2) years, and 68% used an arteriovenous access. Mean duration of dialysis was 31.2±12.6 hours per week. Unadjusted 1- and 2-year technique survival and overall survival were 90% and 83% and 94% and 87%, respectively. Treating center was a strong predictor of technique failure and mortality, with HRs ranging from 0.37 to 5.11 for technique failure (1 of 6 centers with P<0.05 relative to the reference) and 0.17 to 8.73 for mortality (3 of 6 centers with P<0.05 relative to the reference). With baseline adjustment for center, only older age and more than 3 treatments per week remained significant predictors of technique failure, while no individual-level variables remained as significant predictors of survival.\n\n\nLIMITATIONS\nLimited statistical power.\n\n\nCONCLUSIONS\nHome HD treating centers may influence technique failure and patient mortality independent of case-mix. The relationship between processes of care and patient outcomes requires further investigation.