Depending on whether peritoneal dialysis was the initial therapy or the therapy used following failed in-center dialysis? Key Question 3. What are the a) health care system, b) provider, and c) patient factors associated with selection of and technique survival for home-based dialysis (including peritoneal dialysis)? Key Question 4. In the published literature, what are the costs of home hemodialysis or peritoneal dialysis compared to in-center hemodialysis? 14 Home-based versus In-center Dialysis Evidence-based Synthesis Program METHODS TOPIC DEVELOPMENT This topic was nominated by Susan Crowley, MD, VHA National Program Director for Kidney Disease and Dialysis and Rudolph Rodriguez, MD, Chair, VA Renal Field Advisory Committee. Key questions and outcomes were developed with input from a Technical Expert Panel. SEARCH STRATEGY MEDLINE and the Cochrane Library were searched from 1995 to December 2013 for randomized controlled trials (RCTs), controlled clinical trials (CCTs), and observational studies. The search strategy included MeSH terms and keywords for HD and peritoneal dialysis (Appendix A). STUDY SELECTION We included studies of adults with chronic kidney disease receiving dialysis (in-center HD, HHD, or PD) as outpatients. We excluded studies that did not report our outcomes of interest. Primary Outcomes: KQ1, KQ2 – All-cause mortality KQ3 – Health system organizational factors, provider knowledge, patient factors (age, race, gender, caregiver support, social support, comorbidities, cognitive function, physical abilities, rural vs urban [distance from dialysis center], home vs assisted living or skilled care facility) KQ4 – Costs (from literature) Secondary Outcomes: KQ1, KQ2 – Cardiovascular events (MI, stroke, cardiovascular death); hospitalizations; clinically diagnosed depression or cognitive impairment; clinically meaningful difference in quality of life scale scores; conversion to a different type of dialysis (eg, from peritoneal to incenter hemodialysis) Intermediate Outcomes: KQ1, KQ2 – Quality of life (EuroQolEQ, Kidney Disease QOL) scale scores; depressive symptoms; cognitive function; total and mental- and physical-health subscale scores Harms: KQ1, KQ2 – Complications related to vascular access including button hole technique (access failure, infection requiring procedure, thrombectomy, angioplasty, fibrin striping of catheters, replacement of catheters); complications of dialysis (fluid and electrolyte disorders requiring hospitalization, additional dialysis, or both, symptomatic hypotension) We included all RCTs or CCTs that met eligibility criteria. For Key Questions 1 and 2, we required registry studies to enroll at least 1,000 patients and have a mean or median follow-up of 15 Home-based versus In-center Dialysis Evidence-based Synthesis Program at least one year if they reported outcomes of mortality, cardiovascular events, technique failure, or transplantation. For all other outcomes and for Key Question 3, a minimum enrollment of 100 and a mean or median follow-up of at least one year was required. Additionally, for Key Question 3, we included studies of dialysis modality selection only if they followed patients to determine the dialysis modality the patient received. DATA ABSTRACTION From registry studies we extracted study characteristics (dialysis modalities, study purpose, cohort years, country, sample size, and patient inclusion criteria), patient characteristics (age, gender, and race), data analysis technique (factors adjusted for, modeling technique, analysis approach), length of follow-up, and outcomes. If reported, we also extracted data on interactions between mortality and age, gender, race, body mass index (BMI), diabetes, cardiovascular disease, and duration of ESRD therapy. For mortality outcomes, most of the registry studies presented more than one analysis approach (different statistical model, different adjustment factors, etc). We extracted the most-adjusted model. Many studies reported outcomes at multiple time points during the follow-up period. We focused on data at one year, 2 years, and at maximum follow-up time, if provided. QUALITY ASSESSMENT For included RCTs and CCTs, trained research methodologists rated the risk of bias of individual studies as low, moderate, or high risk. Risk of bias ratings were based the following criteria: allocation sequence generation, allocation concealment, blinding, incomplete outcome data, and selective outcome reporting – a modification of the Cochrane approach to determining risk of bias. 4 For observational studies, trained methodologists assessed risk of bias using criteria suggested in the AHRQ Methods Guide: selection bias (use of appropriately comparable control group, design/analysis accounted for important confounding and modifying variables); masking of the outcome assessment (outcome assessor); use of intention-to-treat principles (ie, inclusion of all comparison group participants in outcomes analyses); attrition bias (if overall or differential dropout/loss to follow-up or exclusions a concern, missing data appropriately handled); and selective reporting of prespecified outcomes. 5 Observational studies were considered high risk of bias unless all 5 criteria were addressed by the study authors. Studies that addressed all 5 criteria were considered moderate or low risk of bias depending on how completely the criteria