Abstract
BACKGROUND: Differences in prescribing behaviour among general practitioners (GPs). AIM: To formulate and validate clinical prescribing indicators based on general practice guidelines. DESIGN: Validatory study. SETTING: Pharmacies and general practices in the Netherlands in 2003. PARTICIPANTS: A total of 379 pharmacies, 947 general practices and 3.8 million patients. METHODS: A total of 51 potential indicators were formulated, based on medicinal recommendations from the evidence-based guidelines of the Dutch College of General Practitioners and the corresponding recommendations from the Commission Pharmaceutical Help of the Health Care Insurance Board. These indicators were submitted to an expert panel to assess content validity. The panel assessment was analysed using the RAND-UCLA appropriateness method (RAM). Then, for the remaining indicators, it was assessed to what extent these could be used to determine the prescribing behaviour of GPs and the level to which this behaviour varies among GPs. This was done using a prescribing analyses and cost (PACT) database that was compiled from prescription databases from 379 pharmacies, with all prescriptions from 1,434 GPs over an entire year to 3.8 million patients. RESULTS: The panel considered 34 of the 51 potential indicators to be valid with respect to providing an adequate reflection of the central recommendations in the guideline and in terms of relevance with respect to health gain and/or efficiency. Of these 34 indicators, 20 revealed considerable differences in the prescribing behaviour of GPs. CONCLUSION: On the basis of existing general practice guidelines, 20 prescribing indicators could be formulated that were assessed by an expert panel to be sufficiently valid and which could also discriminate the prescribing behaviour of GPs as reflected in the prescription databases of pharmacies.
