Bastiaan Sallevelt

115 Conversion of STOPP/START version 2 into coded algorithms Figure 1. Flowchart illustrating the consensus procedure. First round A preparation panel consisting of 2 physicians (DdG; GP in training and CD; geriatric resident) prepared a draft algorithm together with a PhD in informatics (MM) for all 114 STOPP/START version 2 criteria. Therefore, the individual criteria needed to be itemized into ‘codable’ pieces. Roughly three categories were distinguished: (1) Diseases and/or medical conditions specified by ICPC 1, 2 and ICD-9 and 10; (2) drug (classes) (with or without specified doses or duration) at ATC 3, 4, or 5 level; (3) laboratory values and measurements (with or without cut-off values) specified in LOINC. After specifying all the codes, they were converted into separate logical algorithms per criterion. Second round For the second round, an expert panel was consulted to review the draft algorithms. The expert panel consisted of a geriatrician-clinical pharmacologist (RvM), a geriatrician (JvC), a clinical pharmacologist (JH), a hospital pharmacist (AV) and a general practitioner (MB). All members of the expert panel received a copy of the draft algorithms, with web links to the ICPC, ATC and ICD databases. The algorithms were accompanied by a code dictionary containing all incorporated codes, 2