An informatics approach to monitor the coverage of antibiotic regimens using open-source software

Viggo Holten, Henrik Carl Schønheyder

Publikation: Konferencebidrag uden forlag/tidsskriftPosterFormidling

9 Downloads (Pure)

Abstract

Background: Antibiotic stewardship programs (ASPs) are important local countermeasures to the global problem of antibiotic resistance. An ASP can reserve broad-spectrum antibiotics for patients most in need of effective treatment or prophylaxis. Current recommendations for ASPs emphasize collaboration with departments of clinical microbiology to keep guidelines up-to-date with prevalence of antibiotic resistance. However, antibiotic susceptibility testing (AST) primarily serves the need of the individual patient, and there may be additional constrains. Therefore, AST data can be cumbersome to apply to ASP purposes. We describe here an informatics approach to monitor the coverage of variousempirical antibiotic regimens using open source software.Material/methods: We obtained AST data (SIR classification according to EUCAST) for all blood culture isolates deemed clinically significant from 2011 through 2015 in North Denmark Region(population app. 650,000 inhabitants). Data was anonymized keeping information on the time sequence of isolates in individual patients. The following tasks were implemented by use of algorithms in this order: expert rules for intrinsic resistance, inference from an antibiotic tested to members of the same class if appropriate for the species, and exclusion of repeat isolates with identical antibiogram within a given time interval. Antibiotic susceptibility was defined by SIR = S, and this rule was not relinquished for combination therapy. Coverage was calculated by the proportion of susceptible isolates among all unique isolates within the time window. Algorithms were implemented in R (www.r-project.org). We report three estimates of coverage based on 1) isolates with AST information (CR-AST), 2) isolates with AST data emended by algorithms (CR-ALGO), and 3) isolates with emended AST data, but assuming resistance for an entire class if no representative antibiotic was tested (‘worst case scenario’) (CR-WCS).
Results: First we assessed the time window for exclusion of repeat isolates. We found a distinct threshold when selecting a 5-day period, and subsequently this interval was used. During the 5-year period 10,267 blood isolates were retrieved including 1,305 repeat isolates resulting in 8,962 unique isolates. We analyzed 8 antibiotic regimens (Table). Notably, CR-ALGO included information from approximately 50% more isolates than CR-AST and showed the combination of ampicillin, gentamicin and metronidazole to be comparable with piperacillin/tazobactam combined with gentamicin and even with meropenem. The figure depicts the coverage of three regimens during the years 2011-2015. A decrease in coverage in 2013 was followed by an increase 2014-2015 as shown by all three measures of coverage.Conclusions: The implementation of algorithms augments consistency of AST data, expands the data set, and can be a practical tool to inform ASPs with up-to-date coverage for frequently used antibiotic regimens. Our informatics approach should also be adaptable to microorganisms from other sources especially if clinically significant isolates can be distinguished from colonizing flora.
OriginalsprogEngelsk
Publikationsdato2017
StatusUdgivet - 2017
Begivenhed27th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID) - Vienna, Østrig
Varighed: 22 feb. 201725 feb. 2017
Konferencens nummer: 27
https://2017.eccmid.org/

Konference

Konference27th European Congress of Clinical Microbiology and Infectious Diseases (ECCMID)
Nummer27
Land/OmrådeØstrig
ByVienna
Periode22/02/201725/02/2017
Internetadresse

Citationsformater