LAUSR

As explained in previous essays in this series, it is impossible to know what the effects of a treatment are without comparing it to what would have happened without the treatment. Predicting what the likely effects of a treatment will be depends on treatment comparisons – typically comparisons between groups of people who are treated differently. The trustworthiness of estimates of treatment effects from treatment comparisons depends on the extent to which the design, conduct, analysis, reporting and interpretation of the comparisons have minimised the risk of systematic errors (biases) that distort effect estimates away from the actual effects. Because it is generally not possible to know the degree to which an effect estimate is biased, judgements must be made about the risk of bias using criteria that assess sources of bias. In this essay, we explain seven sources of bias that should be considered when making judgements about the trustworthiness of treatment comparisons: