Factors influencing antimicrobial prescription attitudes in bloodstream infections: susceptibility results and beyond. An exploratory survey

Abstract

Novel rapid antimicrobial susceptibility testing (RAST) methods promise quicker de-escalation of broad-spectrum antibiotics. However, other behavioural and situational factors influencing antimicrobial prescription are not well known. To explore factors associated with optimal antimicrobial prescription in patients with Gram-negative bloodstream infection and to propose specific scenarios in which a rapid antimicrobial susceptibility result may help to optimize prescribing. Exploratory survey (April-August 2018) in the UK and Spain using clinical case-related questions. Seniority, specialty and country of practice were recorded. Cases described patients with Gram-negative bloodstream infections, their empirical treatment and clinical course and the hypothetical RAST result. Respondents chose one of several options regarding antibiotic treatment management. Microbiologically optimal antibiotic choice (MOAC) was agreed by expert consensus beforehand. Responses were categorized as MOAC, request for support or sub-optimal choice. The relationship between the RAST result and the clinical course was defined as concordant (susceptible organism-clinical improvement; resistant organism-clinical deterioration) or as discordant otherwise. A total of 426 respondents (UK: 332; Spain: 94) and 1494 answers were analysed. Multivariate analysis identified that requests for support were 87% less likely in Spain; that antimicrobial resistance and clinical deterioration were associated with both increased request for support (odds ratio (OR) 7.66 and OR 4.26, respectively) and MOAC (OR 2.08 and OR 2.06, respectively). A discordant clinical course was associated with 82% lower odds for MOAC. Out-of-hours results, seniority and specialty did not have an effect. Antimicrobial choice is influenced by each country's type of practice, clinical course and susceptibility results. Antimicrobial resistance was associated with increased optimal treatment, suggesting RAST may be less useful for step-down decisions in settings with low baseline resistance rates.