increased risk of death over one year follow-up with PD (RR 1.32, P = .005).37 29 Home-based versus In-center Dialysis Evidence-based Synthesis Program Two of the ANZDATA studies favored HD. One reported an increased overall mortality risk (maximum follow-up of 11 years and 9 month) in the PD group (HR 1.10 [95% CI 1.06, 1.16]).12 The other reported an increased risk in the PD group at one or more years follow-up (HR 1.32 [95% CI 1.26, 1.38]).38 The third study, focused on patients from New Zealand, found no difference between PD and HD in overall mortality although mortality was lower in the PD group during the first 3 years and greater in the PD group at greater than 3 years.11 Among the 4 studies from Canada, 3 found no difference in overall mortality between HD and PD. 39,41,42 Follow-up periods ranged from maximums of 541 to 1739 years. One reported reduced overall mortality (maximum follow-up of 6 years) with PD (mortality rate ratio 0.93 [95% CI 0.87, 0.99]).40 All but one of the studies42 reported an early survival advantage for PD patients with no difference after 2 to 3 years of treatment. Of the 8 studies from Europe or the UK, 6 reported no difference in mortality between HD and PD. 14,43,45,47-49 Follow-up periods were up to 25 years. These studies enrolled patients from 1982 to 2011. In one of the studies, all of the patients were on a renal transplant list at some point after the start of dialysis indicating comparable baseline characteristics. 49 One study reported reduced mortality (mean follow-up of 1.6 years) in the PD group (HR [PD vs HD] 0.82 [95% CI 0.75, 0.90])44 and one reported reduced morality (maximum follow-up of 7 years) in the HD group (HR [PD vs HD] 1.48 [95% CI 1.33, 1.65]).46 Although it is difficult to assess temporal trends due to differences in study populations, length of follow-up reported, and methods of data analysis, publication dates would suggest that a trend may exist. All but 2 studies showing increased mortality with PD compared to in-center HD were published before 2003 while all but 3 studies showing no difference or reduced mortality with PD were published after 2003. The RCT reported no difference in mortality (HD vs PD) with a maximum follow-up of 5 years. 50 The adjusted hazard ratio was 3.6 (95% CI 0.08, 15.4, P = .09) with higher mortality in the HD group. In the CHOICE study, the relative hazard of death (PD vs HD) was 1.61 (95% CI 1.13, 2.30) using a multivariate model and adjusting for demographic characteristics, clinical/treatment factors, and laboratory values. 51 By year of treatment, the relative hazard was 1.39 (95% CI 0.64, 3.06) in the first year and 2.34 (95% CI 1.19, 4.59) in the second year indicating that the risk of death did not differ significantly between PD and HD in the first year of treatment but during the second year, the risk of death for PD patients was significantly higher than for HD patients. In the NECOSAD study, the one year mortality risk ratio (HD vs PD) was 1.32 (0.80, 2.18).53 There was no difference in mortality for the first 2 years of dialysis. After 2 years, the adjusted risk ratio decreased and favored HD. The authors concluded that long-term use of PD was associated with increased mortality. 30 Home-based versus In-center Dialysis Evidence-based Synthesis Program Table 2. Mortality – In-center Hemodialysis (HD) vs Peritoneal Dialysis (PD) – Registry and Trial Data Country/ Region: Number of Reports Study Years Patients: Number of Reports or Sample Size Overall Mortality: Number of Reports Number of Studies Reporting Effects by: No difference Favor PD Favor HD Age Gender Race BMI DM CVD ESRD Duration REGISTRY STUDIES USA: 12a 1987-2006 Incident: 11 Prevalent: 1 (Matched: 2) 2 2b 3 4 2 3 3 5e 3 Australia/ New Zealand: 3 1991-2007 Incident: 2 1 2c 3 2 2 2 1 2 Canada: 4 1990-2006 Incident: 4 3 1 1 Europe/ UK: 8 1987-2011 Incident: 8 (Matched: 1) 6 1 1 3 2 2 1 RANDOMIZED CONTROLLED TRIALS Netherlands: 1 1997-2000 N=38 1 CLINICAL COHORT STUDIES USA: 1 1995-1998 Incident, N=1041 1 Netherlands: 1 1997-2002 Incident, N=1222 1d a 5 studies reported mortality in subgroups but no overall mortality b One study favored PD at 1 year and at 2 or more years (no overall results reported) c After 1st year for 1 of the 2 studies d Favored HD after 2 years e 5 datasets (reported in 7 publications) Interactions (Appendix C, Table 3) Age. Ten registry studies assessed interactions between dialysis modality (HD, PD) and age. Significant interactions were reported for 5 studies.12,29,32,36,43 In one of the US studies, which demonstrated an overall increase in the risk of death with PD, the risk of death was significantly higher for PD patients than for HD for patients older than 55 years but not for those younger than 55 years. 36 Two other US studies evaluated risk above or below age 65 years with a significant interaction favoring HD for patients age 65 and older.29,32 A study from Australia/New Zealand reported a significant interaction by age at dialysis inception. 12 A study from the Netherlands reported an age by modality interaction with the survival benefit of PD decreasing with age.43 Five other studies reported either non-significant interactions11,26,45,46 or a significant interaction in the first year of dialysis but not after one year. 38 Gender. Four studies assessed interactions between modality and gender. One