Background.: This study investigated the centre effect on the risk of peritonitis in peritoneal dialysis (PD) patients.
Methods.: This was a retrospective cohort study based on data from the French Language Peritoneal Dialysis Registry. We analysed 5017 incident patients starting PD between January 2008 and December 2012 in 127 PD centres. The end of the observation period was 1 January 2014. The event of interest was the first peritonitis episode. The analysis was performed with a multilevel Cox model and a Fine and Gray model.
Results.: Among the 5017 patients, 3190 peritonitis episodes occurred in 1796 patients. There was significant heterogeneity between centres (variance of the random effect: 0.11). The variance of the centre effect was reduced by 9% after adjusting for patient characteristics and by 35% after adjusting on centre covariate. In the multivariate analysis with a multilevel Cox model, centre with a nurse specialized in PD or centre providing home visits before dialysis initiation decreased the centre effect on peritonitis. Patients treated in centres with a nurse specialized in PD or in centres providing home visits before dialysis initiation had a lower risk of peritonitis [cause-specific hazard ratio (cs-HR): 0.75 (95% confidence interval, CI, 0.67-0.83) and cs-HR: 0.87 (95% CI 0.76-0.97), respectively]. The data show that neither centre type nor centre volume influenced peritonitis risk. In the competing risk analysis, centre with a nurse specialized in PD and centre with home visits had a protective effect on peritonitis [sub-distribution HR (sd-HR): 0.77 (95% CI 0.70-0.85) and sd-HR: 0.85 (95% CI 0.77-0.94), respectively].
Conclusion.: There is a significant centre effect on the risk of peritonitis that can be decreased by home visits before dialysis initiation and by the presence of a nurse specialized in PD.
Keywords: centre effect; multilevel modelling; peritoneal dialysis.
© The Author 2017. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved.