Robin E Ferner1, Jamie J Coleman. 1. West Midlands Centre for Adverse Drug Reactions, City Hospital, Birmingham, UK. r.e.ferner@bham.ac.uk
Abstract
BACKGROUND: Contraindications to medicines define circumstances in which the medicines must not be given. Computerized prescribing systems with decision support should display alerts to warn prescribers of contraindications. However, for such systems to be effective, alerts should only be displayed when relevant. OBJECTIVE: We set out to construct an algorithm that would classify contraindications according to the data available to a computerized system, and allow them to be displayed in context as far as possible. METHOD: We drafted an initial algorithm from first principles, refined it by classifying further datasets, and then tested it on a further set of 95 phrases for contraindications. RESULTS: We were able to classify 94 of the 95 phrases; 13 related to age or sex and four related to allergies, but the majority depended on co-morbid conditions. CONCLUSIONS: We have constructed a practicable algorithm for classifying alerts to contraindications. The classification used will allow alerts to be displayed when relevant. However, most contraindications relate to co-morbid conditions, and prescribing systems will only be able to display these in context if they have access to relevant clinical data.
BACKGROUND: Contraindications to medicines define circumstances in which the medicines must not be given. Computerized prescribing systems with decision support should display alerts to warn prescribers of contraindications. However, for such systems to be effective, alerts should only be displayed when relevant. OBJECTIVE: We set out to construct an algorithm that would classify contraindications according to the data available to a computerized system, and allow them to be displayed in context as far as possible. METHOD: We drafted an initial algorithm from first principles, refined it by classifying further datasets, and then tested it on a further set of 95 phrases for contraindications. RESULTS: We were able to classify 94 of the 95 phrases; 13 related to age or sex and four related to allergies, but the majority depended on co-morbid conditions. CONCLUSIONS: We have constructed a practicable algorithm for classifying alerts to contraindications. The classification used will allow alerts to be displayed when relevant. However, most contraindications relate to co-morbid conditions, and prescribing systems will only be able to display these in context if they have access to relevant clinical data.
Authors: Gilad J Kuperman; Anne Bobb; Thomas H Payne; Anthony J Avery; Tejal K Gandhi; Gerard Burns; David C Classen; David W Bates Journal: J Am Med Inform Assoc Date: 2006-10-26 Impact factor: 4.497
Authors: Anna V Medem; Hanna M Seidling; Hans-Georg Eichler; Jens Kaltschmidt; Michael Metzner; Carina M Hubert; David Czock; Walter E Haefeli Journal: Eur J Clin Pharmacol Date: 2017-02-14 Impact factor: 2.953
Authors: Laura Légat; Sven Van Laere; Marc Nyssen; Stephane Steurbaut; Alain G Dupont; Pieter Cornu Journal: J Med Internet Res Date: 2018-09-07 Impact factor: 5.428
Authors: Jamie J Coleman; Heleen van der Sijs; Walter E Haefeli; Sarah P Slight; Sarah E McDowell; Hanna M Seidling; Birgit Eiermann; Jos Aarts; Elske Ammenwerth; Ann Slee; Robin E Ferner; Robin E Ferner; Ann Slee Journal: BMC Med Inform Decis Mak Date: 2013-10-01 Impact factor: 2.796