Computer aided prescribing
http://www.100md.com
《英国医生杂志》
EDITOR—The deficiencies of existing computerised prescribing decision support systems in the United Kingdom described by Fernado et al and Ferner are mirrored in Australia.1 2 Focus groups conducted by the Australian national prescribing service highlighted concerns that prescribing decision support prompts may not be evidence based or comprehensive.3
Subsequently, four prescribing packages were analysed, using the drug records of 20 elderly patients (N Sharma et al, Australian health and medical research congress, Melbourne, November 2002). There were 5-22 recommended drug-drug interaction prompts per patient. These interactions had been categorised by experts as clinically important (for example, ergotamine and erythromycin), clinically appropriate (for example, celecoxib, angiotensin converting enzyme inhibitor, and diuretic), or of low clinical importance (for example, tramadol and warfarin). The appropriateness of the information for a prescriber in general practice was also examined.
Large variations in the total number of prompts, clinical relevance, and appropriateness of the information were found in the prescribing packages. Between eight and 16 of the 32 recommended clinically significant interactions were not detected. Pharmacokinetic interactions were done well. The packages performed poorly in detecting pharmacodynamic and three way drug interactions, therapeutic duplications, when one drug treats an adverse effect induced by another, and promoting rational drug use.
The National Prescribing Service and General Practice Computing Group believes that the development of safe and effective decision support systems requires a formal information model based on an evidence based clinical model, which incorporates the logic and workflow needed to practice safely and effectively. The methods and models, using asthma as an example, and the general practice data model and core data set,4 are currently being developed (www.healthinformatics.unimelb.edu.au).
Siaw-Teng Liaw, professor
Department of Rural Health, University of Melbourne, Graham Street, Shepparton, VIC 3630, Australia t.liaw@unimelb.edu.au
Stephen Kerr, decision support officer
National Prescribing Service, Australia, Level 7, 418A Elizabeth Street, Surry Hills, NSW 2010, Australia
Competing interests: None declared.
References
Fernado B, Savelyich BSP, Avery AJ, Sheikh A, Bainbridge M, Horsfield P, et al. Prescribing safety features of general practice computer systems: evaluation using simulated test cases. BMJ 2004;328: 1171-2. (15 May.)
Ferner RE. Computer aided prescribing leaves holes in the safety net. BMJ 2004;328: 1172-3. (15 May.)
Ahearn M, Kerr SJ. General practitioner perceptions of the pharmaceutical decision support tools in the prescribing software. Med J Austr 2003; 179: 34-37.
Commonwealth Department of Australia and the General Practice Computing Group. General practice data model and core data set project final project report. September 2000. www.gpcg.org/publications/jointpubs.html (accessed 18 June 2004).
Subsequently, four prescribing packages were analysed, using the drug records of 20 elderly patients (N Sharma et al, Australian health and medical research congress, Melbourne, November 2002). There were 5-22 recommended drug-drug interaction prompts per patient. These interactions had been categorised by experts as clinically important (for example, ergotamine and erythromycin), clinically appropriate (for example, celecoxib, angiotensin converting enzyme inhibitor, and diuretic), or of low clinical importance (for example, tramadol and warfarin). The appropriateness of the information for a prescriber in general practice was also examined.
Large variations in the total number of prompts, clinical relevance, and appropriateness of the information were found in the prescribing packages. Between eight and 16 of the 32 recommended clinically significant interactions were not detected. Pharmacokinetic interactions were done well. The packages performed poorly in detecting pharmacodynamic and three way drug interactions, therapeutic duplications, when one drug treats an adverse effect induced by another, and promoting rational drug use.
The National Prescribing Service and General Practice Computing Group believes that the development of safe and effective decision support systems requires a formal information model based on an evidence based clinical model, which incorporates the logic and workflow needed to practice safely and effectively. The methods and models, using asthma as an example, and the general practice data model and core data set,4 are currently being developed (www.healthinformatics.unimelb.edu.au).
Siaw-Teng Liaw, professor
Department of Rural Health, University of Melbourne, Graham Street, Shepparton, VIC 3630, Australia t.liaw@unimelb.edu.au
Stephen Kerr, decision support officer
National Prescribing Service, Australia, Level 7, 418A Elizabeth Street, Surry Hills, NSW 2010, Australia
Competing interests: None declared.
References
Fernado B, Savelyich BSP, Avery AJ, Sheikh A, Bainbridge M, Horsfield P, et al. Prescribing safety features of general practice computer systems: evaluation using simulated test cases. BMJ 2004;328: 1171-2. (15 May.)
Ferner RE. Computer aided prescribing leaves holes in the safety net. BMJ 2004;328: 1172-3. (15 May.)
Ahearn M, Kerr SJ. General practitioner perceptions of the pharmaceutical decision support tools in the prescribing software. Med J Austr 2003; 179: 34-37.
Commonwealth Department of Australia and the General Practice Computing Group. General practice data model and core data set project final project report. September 2000. www.gpcg.org/publications/jointpubs.html (accessed 18 June 2004).