UC Berkeley

This article continues from a prior commentary on evaluating the risk of bias in randomised controlled trials addressing nutritional interventions. Having provided a synopsis of the risk of bias issues, we now address how to understand trial results, including the interpretation of best estimates of effect and the corresponding precision (eg, 95% CIs), as well as the applicability of the evidence to patients based on their unique circumstances (eg, patients values and preferences when trading off potential desirable and undesirable health outcomes and indicators (eg, cholesterol), and the potential burden and cost of an intervention). Authors can express the estimates of effect for health outcomes and indicators in relative terms (relative risks, relative risk reductions, OR or HRs)-measures that are generally consistent across populations-and absolute terms (risk differences)-measures that are more intuitive to clinicians and patients. CIs, the range in which the true effect plausibly lies, capture the precision of estimates. To apply results to patients, clinicians should consider the extent to which the study participants were similar to their patients, the extent to which the interventions evaluated in the study are applicable to their patients and if all patient-important outcomes of potential benefit and harm were reported. Subsequently, clinicians should consider the values and preferences of their patients with respect to the balance of the benefits, harms and burdens (and possibly the costs) when making decisions about dietary interventions.