Objectives - (1) To describe interpractice variation in diagnosis of respiratory infections at consultation. (2) To test the hypotheses that: (a) The decision to prescribe an antibiotic in respiratory infection is influenced by the diagnosis, the perceived certainty of diagnosis, and whether or not a consultation takes place. (b) The choice of antibiotic is influenced by the diagnosis. Design - A regional survey of prescribing and associated morbidity in general practice, over a 2-week period in April 1994. Setting - Stratified quota sample of 22 Northern Ireland practices. Results - There was wide interpractice variation in diagnosis of common respiratory infections at consultation, especially tonsillitis (5.0-157.5/1000 consultations). Overall, different diagnoses predicted the decision to prescribe an antibiotic at different levels (coryza 42.3%, tonsillitis 84.8%), but there was wide interpractice variation in the decision to prescribe for most diagnoses. With the exception of coryza and sinusitis, the perceived certainty of diagnosis did not significantly influence the decision to prescribe. The decision to prescribe was not significantly influenced by whether or not a consultation took place. Overall, broad spectrum penicillins were the therapeutic group most frequently prescribed for a given diagnosis with the exception of tonsillitis (phenoxymethylpenicillin) and sinusitis (tetracyclines), but there was wide interpractice variation in choice of antibiotic. Conclusions - Little consensus exists among practices regarding rational prescribing decisions in respiratory illness. The absence of a consultation was no deterrent to antibiotic prescribing. At one extreme, it is suggested that some practices are avoiding consultations for respiratory infections. There is wide variation in choice of antibiotic, despite existing guidelines. Copyright (C) 1999 John Wiley & Sons, Ltd.
