Bastiaan Sallevelt

125 Conversion of STOPP/START version 2 into coded algorithms Figure 3. START criterion C3 as a flowchart. Discussion Main findings For this study STOPP/START criteria version 2 were converted into coded algorithms implementable in software applications. During four multidisciplinary consensus rounds we converted all 34 START criteria and 76 STOPP criteria into algorithms. Consensus based decisions on interpretation are necessary to convert STOPP/ START elements requiring clinical context and knowledge of individual patients’ history into coded algorithms. Five principles for universal coding were formulated to prevent essentially altering the content of criteria by elucidating the underlying intention of a criterion and to minimize the risk of bias. Strengths To the best of our knowledge, this is the first study providing implementable algorithms for software applications based on STOPP/START version 2. For the development of these algorithms, experts, trained in the use of STOPP/START in daily practice and familiar with international guidelines regarding pharmacotherapy in older people, were consulted. Experts from both general practices and hospital settings were involved, of which the majority also cooperated in the specification of STOPP/START version 1 [19]. Additionally, the experience and resources of two researchers involved in the development and application of a STOPP/START version 2 based CDSS were used. This allowed for evaluating our developed algorithms within this CDSS. 2