Bastiaan Sallevelt

127 Conversion of STOPP/START version 2 into coded algorithms testing of potassium [27,28]. Whether this value is already an indication to stop a presumed indicated medication like an ACE inhibitor in a clinical setting, remains debatable. Therefore, future applicators of the algorithm might decide differently, depending on their own expertise. Additionally, the expert panel consulted for this study comprised a limited number of professionals from one country. This might restrict the extrapolation of the results to other countries. Supplementary international validation through a Delphi method could be considered. STOPP/START related restrictions The majority of STOPP/START criteria are designed for clinicians facing the difficulties of polypharmacy in individual patients, presuming knowledge or at least accessible documentation of this patient’s medical history and prior treatment regimens. However, converting these criteria into coded algorithms is challenging and sometimes even infeasible. In STOPP D2; ‘initiation of TCAs as first-line antidepressant treatment’ for example, a clinician might know immediately how to act, but ‘first-line treatment’ is not convertible into a code. The same reasoning applies to START G1 and G2; Alpha1- receptor blocker/-5alpha reductase inhibitor with symptomatic prostatism, where prostatectomy is not considered necessary.’ This restriction cannot be coded, let alone be extracted from a database or health record if it were codable. Consequently, leaving incodable elements out of the algorithm, led to a simplification of certain criteria. Moreover, when all STOPP/START criteria based algorithms are implemented together in a database or CDSS to detect PIP, one must keep in mind that several criteria addressing overlapping diagnoses can result in conflicting recommendations. In STOPP L2 for example, the use of opioids without concomitant laxative is undesirable and the opioid is identified here as PIM, while in START H2 laxatives are recommended for the same patient using opioids. In START F1, an ACE-inhibitor is recommended in patients with type 2 diabetes mellitus with renal disease, while in case of concurrent hyperkalemia this is contra-indicated according to STOPP A11 and STOPP A12. Additionally, in START A7 and A8, a beta blocker is recommended in patients with ischemic heart disease and/or stable systolic heart failure. However, in patients already using verapamil or diltiazem or in case of present bradycardia, this is undesirable because of the increased risk of (total) heart block according to STOPP B3 and B4. In this same hypothetical patient, the use of verapamil or diltiazem will also trigger STOPP B2:’ Verapamil or diltiazem with NYHA Class III or IV heart failure’. If this recommendation is followed, starting a beta blocker will most likely be appropriate advice after all. This illustrates the complexity of applying (coded or non-coded) criteria to both databases and individual patients without 2