DynaMed includes information about ultrafiltration in the Acute Heart Failure article.

For the outcome of readmission rate:
DynaMed references ACC/AHA 2009 practice guideline recommending use in patients with refractory congestion. Evidence suggests that ultrafiltration may be associated with reduction in hospital readmission. For a summary of evidence, in the DynaMed article, use links on left side of page – Treatment section>Other management>ultrafiltration.  For details of the evidence, including cited studies see 2009 focused update: ACCF/AHA guidelines for the diagnosis and management of heart failure in adults.

For the outcome of length of stay:
A search of PubMed’s Clinical Queries (Therapy filter, heart failure AND length of stay AND ultrafiltration) retrieved 2 RCTs (Hanna, 2012 and Costanzo, 2005) that included length of stay as an outcome. Both studies reported evidence that ultrafiltration may be associated with reduction in length of stay. See individual studies for details on population size and other aspects of the methodology.

For the outcome of kidney function:
Revising the PubMed Clinical Queries to  Therapy/narrow filter, heart failure AND ultrafiltration AND (kidney function OR renal function OR creatinine)  Retrieves studies and reviews discussing kidney function. If there are other actual markers that would be used to measure renal function, those can be added to the Clinical Query.

Use in cardiac step down units:
Searches of PubMed (“step down” OR “progressive care”) AND ultrafiltration AND heart failure and CINAHL (step down OR progressive care) AND heart failure, did not identify any studies specifically discussing ultrafiltration on step down units.

For risks PubMed Clinical Queries for Therapy/Broad filter, heart failure AND (adverse effects OR risk) AND ultrafiltration. There are reviews of evidence by Freda (2011) and Dandamudi (2011) and a retrospective analysis (Flythe, 2011) of data from an RCT to examine morbidity and mortality.  These results include systematic reviews as well.