Improvements in immunosuppressive regimens are likely to be moderate in size. Therefore, studies to detect them reliably require a guarantee of small biases (so randomization is essential) and small random errors. Even after adjustment, nonrandomized studies may have large biases, so are not a suitable alternative to large randomized trials.

Original publication

DOI

10.1097/TP.0b013e31817d4f14

Type

Journal article

Journal

Transplantation

Publication Date

15/07/2008

Volume

86

Pages

26 - 27

Keywords

Bias (Epidemiology), Evidence-Based Medicine, Humans, Immunosuppressive Agents, Observation, Organ Transplantation, Randomized Controlled Trials as Topic, Registries, Treatment Outcome