Controlling antibiotic usage-A national analysis of General Practitioner/Family Doctor practices links overall antibiotic levels to demography, geography, comorbidity factors with local discretionary prescribing choices.

INTRODUCTION

Ecological studies show association between antimicrobial resistance (AMR), and inappropriate oral antibiotics use. Moderating antibiotic prescribing requires an understanding of all drivers of local prescribing. The aim was to quantify how much is determined by external factors compared with discretionary clinical choices.

METHODS

Oral antibiotic usage taken from England General Practitioner/Family Doctor practice prescribing data was aggregated using WHO/ATC defined daily doses (DDDs). The average annual antibiotic daily prescribing rate (AAADPR) in each practice was the total DDD of oral antibiotics divided by registered population and 365. The AAADPR of English practices in 2017_18 was linked by regression to factors including demographics, geography, medical comorbidities, clinical performance, patient satisfaction, medical workforce characteristics and prescribing selection. The regression coefficients for modifiable prescribing selection factors were applied to the difference between the median and top decile practice values to establish overall reduction opportunities through changing prescribing behaviour.

RESULTS

Twenty five factors accounted for 58% of the AAADPR variation in 5889 practices supporting 49.8 million patients. Non-modifiable factors linked increased AAADPR to more northerly location, higher prevalence of diabetes, COPD, CHD, and asthma; higher white ethnicity; higher patient satisfaction and lower population density. Modifiable behaviour accounted for 11% of the variation in AAADPR, with increases associated with a wider range of antibiotics, higher proportion taken as liquids, higher doses in each prescription, lower guideline compliance, lower targeted antibiotics, lower spend/dose, and less seasonal variation. If all practices achieved the level of modifiable factors of the top decile, this model suggests that overall AAADPR could reduce by 31%.

CONCLUSION

Such analysis is associative and does not infer causation. However, demographics, location, medical condition of the population, and prescribing selection are drivers of overall antibiotic prescribing. This analysis provides benchmarks for both non-modifiable and modifiable factors against which practices could evaluate their opportunities to reduce antibiotic prescribing.