The use of serial point-prevalence studies to investigate hospital anti-infective prescribing

Abstract

Abstract Objectives To develop and test an efficient, reproducible method for the surveillance of hospital anti-infective use; to use this method to investigate patterns of anti-infective prescribing. Method A series of three standardised point-prevalence studies were carried out in which pharmacists recorded details of all inpatients prescribed systemic anti-infectives. Time taken to collect, enter and analyse these data was documented. Parameters examined included: percentage of patients prescribed anti-infectives, percentage of anti-infectives that were for “reserved” use, percentage of these with appropriate approval, percentage of anti-infectives administered intravenously, duration of therapy, and combinations of anti-infectives used. Setting All hospital inpatients in a large National Health Service (NHS) Trust comprising four sites. Key findings For each study, an estimated 35 additional hours of pharmacists' time was required for data collection, cleaning and analysis, and 15 hours for data entry. The method developed was easily reproducible and results from the three studies were very similar. Overall, 33 per cent of 2,656 inpatients were prescribed at least one anti-infective (mean 1.7 per patient); 48 per cent of anti-infectives were given intravenously (IV), of which 34 per cent could have been given orally. Of the anti-infectives used, 21 per cent were for “reserved” use. Of these, 65 per cent were used for an approved indication, and 11 per cent were not. The remaining 24 per cent had no indication documented in the medical notes. Conclusion This is a practical method for studying hospital anti-infective use in the absence of computerised prescribing. The database produced provides a wealth of information and various targets for intervention have been identified; these can now be evaluated against the baseline data collected. The methods developed could be used in other hospitals to provide benchmarking data.