Patient and physician predictors of peritoneal dialysis technique failure: a population based, retrospective cohort study.

Background: The use of peritoneal dialysis (PD) has been declining over the past decade in Canada, and high technique failure rates have been implicated. Studies have examined clinical risk factors for PD technique failure, but few studies have addressed sociodemographic factors driving technique failure. There are no studies examining the effect of physician factors on technique failure. Methods: We conducted a retrospective cohort study using Ontario healthcare databases from 1 April 1995 to 31 March 2005 to examine the effects of patient sociodemographic and physician characteristics on PD technique failure. The primary outcome was time to technique failure. Secondary outcomes included the proportion of patients experiencing technique failure during the first year and the proportion of patients experiencing death during the study period. A competing risks analysis was applied to the Cox proportional hazards model to determine the predictors of technique failure, death, and kidney transplantation. Results: In 5162 incident PD patients, the probability of technique success and patient survival at 5 years was 58.2% and 46.9% respectively. Of patients failing PD, 43.5% failed during the first year of treatment. Statistically significant predictors of technique failure included increasing age [hazard ratio (HR) 1.02], diabetes mellitus (HR 1.32), lower neighborhood education level (HR 2.93), and receiving transient (<= 3 months) hemodialysis before starting PD (HR 1.24). Predictors of patient death included increasing age (HR 1.05), diabetes mellitus (HR 1.44), coronary artery disease (HR 1.26), congestive heart failure (HR 1.58), and late referral to the nephrologist (HR 1.27). Distance from treating dialysis center and residing in a rural area did not impact the risk of technique failure or death. Male physician gender increased the risk of technique failure (HR 1.31). Increased PD patient volume decreased the risk of technique failure (HR 0.98). None of the physician factors were predictors of patient death. Conclusion: These findings support the need for implementing strategies to reduce technique failure, which could include increasing educational resources for patients initiating PD, aggressive risk factor modification in patients with multiple comorbidities, and increasing physician awareness regarding the detrimental outcomes associated with late referral and late PD start.