included 17,926 patients either receiving dialysis on January 1, 1992 or starting dialysis during 1992. PD patients were younger and more likely white. An intent-to-treat approach was used with a Cox proportional hazards model. Additional information is provided in Appendix C, Table 1. There were 3 reports from ANZDATA11,12,38 including 2 cited above because they also included an HHD group. 11,12 The study dates ranged from 1991 to 2011 with maximum follow-ups of 15 years. Two of the studies included approximately 25,000 incident patients12,38 while the third included 6,419 patients.11 Two studies reported that PD patients were older and less likely male. 11,38 As noted for the HHD/HD comparison above, one study use an “as-treated” approach with a MSM model12 and another used an “as-treated” approach with a Cox proportional hazards model.11 The third study used an intent-to-treat approach with Cox regression models.38 Appendix C, Table 1 provides more information about these studies. Four of the registry studies were from Canada – 3 from the Canadian Organ Replacement Register (CORR)39-41 and one from the Institute for Clinical Evaluative Sciences (ICES). 42 The studies enrolled patients between 1990 and 2006 with maximum follow-up periods ranging from 5 years41 to 17 years.39 Sample sizes ranged from 6,57342 to 46,83939 incident patients. One study reported that the HD patients were older than the PD patients41 and another reported that there was a higher percentage of HD patients in the age 65 and older category while more PD patients were in the age 35 to 64 year category.39 All of the studies used an intent-to-treat approach with Cox models (Appendix C, Table 1). 28 Home-based versus In-center Dialysis Evidence-based Synthesis Program The remaining 8 registry reports were from Europe or the UK. Included were reports from the Dutch End-Stage Renal Disease Registry (RENINE), 43 the European Renal AssociationEuropean Dialysis and Transplant Association (ERA-EDTA), 44 the Finnish Registry for Kidney Diseases, 45 the French Renal Epidemiology and Information Network (REIN),46 the Lombardy Dialysis and Transplant Registry, 47 the Romanian Renal Registry, 48 the Scottish Renal Registry, 49 and the United Kingdom Renal Registry (UKRR).14 The studies included incident patient data from 1987 to 2011 with follow-up periods ranging from a mean of 2.4 years43 to a maximum of 25 years. 49 Sample sizes ranged from 2,47514 to 16,643.43 Three studies reported that PD patients were younger43-45 while another reported that PD patients were older.46 Three studies reported that PD patients were less likely male46,48,49 while a third reported that PD patients were more likely male. 43 One study reported that a higher percentage of PD patients were on the transplant wait list. 45 All of the studies used an intent-to-treat approach with Cox models. Additional information about the studies is reported in Appendix C, Table 1. One RCT and 2 clinical cohort studies also compared HD and PD. The RCT, completed in the Netherlands, enrolled 38 patients new to dialysis and randomized them to HD or PD.50 Patients were followed for a maximum of 5 years. The trial was stopped because of low enrollment, failing to reach the goal of 100 patients. Despite randomization, the HD patients were older. The clinical cohort studies included the Choices for Healthy Outcomes in Caring for ESRD (CHOICE) study completed in the US51 and the Netherlands Cooperative Study on Adequacy of Dialysis (NECOSAD). 52 The CHOICE study enrolled 1,041 incident patients between 1995 and 1998 and followed them for a maximum of 7 years.51 The PD patients in this study were younger and more likely white. The NECOSAD cohort enrolled 1,222 incident patients and followed them for a maximum of 4 years. 53 PD patients were younger and more likely male. Both studies used an intent-to-treat approach with Cox proportional hazards models. Mortality Mortality outcomes are summarized in Table 2 with more detailed reporting in Appendix C, Tables 1 and 2. Of the 27 registry studies, 22 reported overall mortality with 12 finding no significant difference in mortality between HD and PD, 4 finding a more favorable outcome for PD, and 6 finding a more favorable outcome for HD. Of the 11 CMS/USRDS studies, one reported overall mortality (maximum follow-up of 4 years) finding no difference (HR 1.05 [95% CI 0.96, 1.16]).29 By year of follow-up, the difference was not significant during the first year but there was a difference, favoring HD, at 2 years (HR 1.19 [95% CI 1.02, 1.38]).29 Another study with over 23,000 patients reported results at one year and 2 years (but no overall results).26 In that study, there was significantly reduced mortality at both one (HR 0.59 [95% CI 0.44, 0.78]) and 2 years (HR 0.52 [95% CI 0.34, 0.80]) for the PD group. Another study reported reduced overall mortality (maximum follow-up of 6 years) for PD (HR 0.88 [95% CI 0.81, 0.95]).27 Two older studies33,36 found increased mortality in the PD group. Follow-up periods were 2 years33 and one year. 36 One study found no difference in mortality over a maximum follow-up of 5 years between PD and HD (HR 1.03 [95% CI 0.99, 1.06]).28 The remaining study did not report overall mortality results. 31 The other US study reported an overall