Williamson I. , MD Benge S. , PhD Moore M. , MSc Kumar S. , PhD Cross M. , BA Little P. , MD Department of Primary Medical Care, University of Southampton, Southampton, UK (IW, SB, MM, PL); University of Brighton, UK (SK); University of Portsmouth, UK (MC)
We found that practices with higher Townsend scores were under-represented in our sample. This means the absolute level of reported prescribing effects may have been slightly underestimated for the first vignette. However, this did not affect the main aim of our study, which was to assess the relative rather than absolute importance of variables in both diagnosis and prescribing. Including the Townsend score in the model did not significantly confound estimates for other variables.
The main limitation of the study, therefore, is that we were assessing reported decisions to diagnose and prescribe: in this context doctors may have felt pressure to report evidence-based practice rather than what they do in reality. However, family doctors do not have easy access to evidence about diagnostic criteria for sinusitis since it has been little summarized and not widely debated in the Family Practice/Primary Care literature.5,10,11,13 Diagnostic criteria are more widely reported in world literature, and include technology assessments.3,31 However reporting bias seems an unlikely explanation for our results.
FAST TRACK
Doctors’ beliefs in effectiveness of antibiotics for acute sinusitis contrasts with best evidence
Our case studies focus on acute purulent maxillary sinusitis and not pan-sinusitis or chronic sinusitis. The Berg and Carenfelt criteria are not known predictors of complications of acute sinusitis but these are rare. The criteria are, however, known to perform as well as radiology against a gold standard of microbiological diagnosis (receiver operator characteristic curves),3 and the included mucopurulent symptoms and signs have more recently been found to be predictive of treatment response.14
Acknowledgments
We wish to thank all those 557 Wessex Physicians who participated in the study. Funding was from the DOH (SE Regional Grant) although the views expressed in this paper are not necessarily those of the DOH.
This article was originally presented at NAPCRG (preliminary results) in 2004. Ethics approval Southampton: 358/02/w. Funding: Department of Health (former SE region) UK. Dr S. Benge, PhD, Research Fellow, was employed on this grant. The views in this paper are those of the authors and not necessarily those of the DOH. No conflicts of interest declared.
