Annie-Claire Nadeau-Fredette Examination of Centre Effects using the ANZData registry ANZSN 53rd ASM Darwin 5th September Carmel Hawley On behalf of Annie-Claire Nadeau-Fredette Htay Htay David Johnson Elaine Pascoe Yeoungjee Cho PD working group ANZData Registry team The AKTN operates within the UQ School of Medicine’s Centre for Kidney Disease Research (CKDR).

Peritoneal Dialysis centre effects’ analyses Significant variation in PD outcomes is known to exist Identifying centre practices associated with outcomes is important as these practices may be modifiable and be relevant to quality improvement Aim: To determine the relative contributions of patient characteristics and dialysis centre characteristics Hypothesis: Centre factors are an important contributor to variation in important PD outcomes

Methods PD outcomes of interest examined Incidence of peritonitis: published Technique failure : published Outcome of peritonitis : under review Statistical method 2 level mixed effects models with random intercept for dialysis centres Fixed effects were patient and centre characteristics

Methods Study populations Australian incident PD patients ANZData registry 2003-2013 (peritonitis) 2004-2014 (technique failure; cure of peritonitis) 51 centres: 8711 (Peritonitis); 9362 (technique failure); 4428 (cure of peritonitis)

Model building Model 1 Model 2 Model 3 only centre (random effect) intercept only model obtain estimate for risk each centre & sd (sd1) relating to variation Model 2 Model 1+ add patient characteristics (fixed effects) obtain estimate for each centre & sd (sd2) relating to variation % reduction in variation = (sd1-sd2)/sd1 Model 3 Model 2 + add centre characteristics (as fixed effect) obtain estimate for each centre & sd (sd3) relating to variation Further % reduction = (sd2-sd3)/sd2

Methods Patient characteristics: Age, gender, race and diabetes + (patient level model) Centre characteristics: Centre size, PD %, APD exposure, icodextrin exposure, PET performance, and analysis-specific factors 1st and 4th quartile compared with combined 2nd and 3rd quartile

Peritonitis rates

Centre Variation in Peritonitis Rates Unadjusted Patient-adjusted Facility-adjusted 16% patient 34% centre Variation in incidence rate of peritonitis across 51 Australian peritoneal dialysis centers during the period of 2003-2013 in unadjusted (green diamond), patient-level adjusted (red triangle) and patient- and center-level adjusted models (blue circle) with standard errors. Dialysis centers are ranked by incidence rate Nadeau-Fredette et al PDI 2016 36: 509-518

Centre characteristics : Peritonitis risk Center-level characteristics Variation across centers (n=51) Center size (total patient-years of follow-up) 220 (113-425) Percentage of PD (versus overall dialysis population) 20% (16%-26%) Transplantation center (n, %) 19 (37%) APD treatments exposure (% patients in center) 73% (49%-85%) Icodextrin solution exposure (% patients in center) 46% (30%-56%) Biocompatible solutions exposure (% patients in center) 2% (0%-10%) PET performed at PD initiation (% patients in center) 54% (36%-62%) Anti-fungal prophylaxis tis (% peritonitis in center) 64% (24%-82%) Hospitalisation for peritonitis (% peritonitis in center) 49% (28%-62%) Catheter removal with peritonitis (% peritonitis in center) 20% (15%-24%) Nadeau-Fredette et al PDI 2016 36: 509-518

Centre characteristics associated with peritonitis risk Smaller centre size, higher % of PD, lower rates of PET performance at PD start and lower % of hospitalisation for peritonitis were assocaited with a lower peritonitis count. In contrast a lower % of APD exposure, greater icodextrin exposure and lower or higher use of antifungal prophylaxis at time of peritonitis were associated with a higher peritonitis risk Nadeau-Fredette et al PDI 2016 36: 509-518

Technique Failure

Centre Variation in Technique Failure 7 fold variation 28% patient 53% centre Unadjusted Patient-adjusted Facility-adjusted Hazard rate Figure 1. Variation in hazard of technique failure across 51 Australian peritoneal dialysis centers during the period of 2004-2014 unadjusted (green diamond), patient-level adjusted (red triangle) and patient- and center-level adjusted models (blue circle) with standard errors. Dialysis centers are ranked by hazard rate of peritonitis Average technique failure: 0.35 episodes /pt-yr Htay…Johnson Clin J Am Soc Nephrol 2017; 12: 1090-1099

Center characteristics associated with technique failure Adjusted analyses % Patients on PD also predictive Technique failure was significantly less in centres with higher % of PD patients and larger centres; and more likely in centres with lower proportions of patients achieving baseline serum phosphate levels in target Htay et al ; Clin J Am Soc Nephrol 2017; 12: 1090-1099

Peritonitis Cure with antibiotics Paper under review

Conclusions 1 Centre characteristics accounted for important variation in risk of peritonitis, technique failure, and cure of peritonitis with antibiotics Whether these factors are directly related to the outcomes or relate to surrogate markers for other characteristics of units is uncertain

Conclusions 2 More methodological research is required More information on centre factors that may influence important outcomes is required

Acknowledgements All registry contributors: patients and health professionals Annie Claire Nadeau-Fredette: Baxter Healthcare Clinical Evidence Council (CEC) research grant Htay Htay : Health Manpower Development Plan fellowship award co-sponsored by MOH and Singhealth (Singapore) BEAT CKD NHMRC program grant ANZData registry staff PD working group of ANZData NHMRC fellowships: Johnson and Cho