Interpreting reported health-care benefits

Bias in treatment trials

Rigorous treatment trials exhibit a number of key features to guard against bias. Randomization, blinding (of patients, care staff and assessors), full follow-up on all patients, and 'intention-to-treat' analyses all contribute to removing bias so that any true treatment effect can be revealed. This article outlines the rationale behind these features and advises how bias in treatment trials can be assessed.

Aiding clinical decisions with decision analysis

As clinical decision making gets ever more complex, new analytical approaches are being developed to help. Decision analysis is used to structure complex decision problems in an uncertain environment by systematically linking decision choices with expected outcomes. Such models can include the probabilities of outcomes, patient preferences and costs. These models can help to advise about therapeutic avenues. This paper examines the nature of decision analysis, and explores the pitfalls that arise in interpreting the findings from published studies.