How to measure success?
Outcome selection for an interventional study will inevitably stem from the intervention scope and target; in considering what would be a success, it is important to reflect on the goals of the intervention, and the audience for whom evidence is being produced. Particularly for deprescribing trials, no change in clinical or patient-reported outcomes may be considered a success, assuming success is judged as a reduction in the number of medications. This should be decided a priori, with the study design and analysis planned appropriately to evaluate non-inferiority in clinical outcomes. In contrast for medicines optimisation trials, a reduction in medicines may not always occur (or be appropriate), and therefore some benefit in other outcomes may be the goal and basis for designing and powering the study. Process outcomes are important regardless of intervention focus, to assess whether the behavioural strategies have been delivered. For deprescribing, process outcomes could also include the number of participants who had a medicine (or medicines) deprescribed or, similar to above, the mean reduction in the number of medicines. However, using similar measures in a medicines optimisation trial may not fully capture the degree to which actions have been taken, and a small or negligible reduction in the mean number of medicines could be underpinned by a large, balanced number of medicines stopped and started.(7)
Outcomes should be selected with regard to what is meaningful for stakeholders, including patients, healthcare professionals, and decision-makers who may wish to implement the intervention being assessed.(8) Core outcome sets can be drawn on as a means of enhancing consistency and cumulative evidence across trials.(9) Arguably these should be tied to trials with the same intervention scope and targets (i.e. core outcomes for medicines optimisation interventional studies may not all be relevant or useful for a deprescribing interventional study of a single medication class).
Focusing on single medications may facilitate selection of outcomes that are highly specific to that context (and potentially more meaningful to patients), both for clinical and patient-reported outcomes e.g. sleep quality as an outcome if deprescribing hypnotic agents, cognition for anticholinergic agents. It may be difficult to assess success using measures that reflect general medicines optimisation, particularly where such generic or global health measures may not be sensitive to changes in medications.(8) In such cases, outcomes often include measures of prescribing appropriateness, adverse drug reactions, and medicines-related hospitalisations, where the attributable risk due to inappropriate prescribing, and preventability via medications optimisation, may be low.
Costs and cost-effectiveness are important for decision-makers. For interventions targeting many medications, estimating the indirect costs saved through averting adverse outcomes is challenging when these events are likely rare, diverse, and potentially difficult to measure. Although costs relating to occurrence of a generic adverse drug reaction or medicines-related hospitalisation could be estimated, these may have limited acceptability among decision-makers, and add significant uncertainty to the cost-effectiveness estimate. In contrast, for single medication interventions, evidence on the known risks of continued prescribing can be integrated with the effectiveness of the intervention in reducing such prescribing to estimate the incremental costs and benefits of implementing the intervention, and can be extrapolated over a timeframe beyond the trial duration.(10)