Timothy Miles Rawson
A case-based reasoning, or CBR, algorithm offered appropriate antibiotic recommendations significantly narrower in spectrum than choices made currently in clinical practice, according to results from a study published in Clinical Infectious Diseases.
“Antimicrobial resistance is one of the greatest threats to patient safety. To reduce the impact of antimicrobial resistance, we need to find methods of reducing inappropriate use of antimicrobials,” Timothy Miles Rawson, PhD, MBBS, BSc, MRCP, honorary clinical research fellow at Imperial College London’s Hammersmith Hospital, told Healio. “Our case-based reasoning tool was able to make antibiotic recommendations that were appropriate but of a narrower spectrum than those currently being prescribed.”
The decision support tool used by Rawson and colleagues has the potential “to reduce antimicrobial selection pressure placed on microorganisms and thus help in the fight against antimicrobial resistance,” he added.
The researchers compared recommendations provided by the CBR algorithm with decisions from physicians in clinical practice. Prescriber recommendations were also compared with recommendations based on the Antimicrobial Spectrum Index and the WHO Essential Medicine List Access, Watch, Reserve system. They examined prescribing recommendations for 145 patients with Escherichia coli bloodstream infections and 79 ward-based patients with various infections. They also defined the appropriateness of a prescription as the spectrum of the prescription covering the known or most likely organism antimicrobial sensitivity profile.
The researchers found that CBR recommendations were appropriate in 202 (90%) of cases vs. 186 (83%) in practice (OR = 1.24; 95% CI, 0.392-3.936). The CBR recommendations (49%) were more likely than physicians’ prescriptions (35%) to be classified as access class antimicrobials (OR = 1.77; 95% CI, 1.212-2.588). The results were similar for ward patients and patients with E. coli infection in a subgroup analysis.
The algorithm suggested therapies that were of a narrower spectrum compared with those of clinical practice and more likely to adhere to WHO access classification.
“This was surprising, as the system had not been designed or weighted to make decisions based on the spectrum of an antimicrobial,” Rawson said. “A possible explanation is that prescribers may tend to prescribe broader than required empiric antimicrobial therapy, especially when a clinical presentation does not fit neatly within a clinical guideline.”
Rawson also emphasized that one of the study’s limitations was that it did not address the impact the CBR tool would have on antimicrobial prescribing behavior.
“Given that a range of behavioral factors, such as the role of team dynamics and ‘prescribing etiquette’ have a significant influence on antimicrobial prescribing, it is important to evaluate this system experimentally,” he said. “We are currently in the process of setting up a prospective evaluation of the system to address this limitation.”
“Our hope is that, through the adoption of decision support tools containing artificial intelligence, we will be able to support the clinician in making more appropriate, narrower spectrum antimicrobial choices when faced with similar prescribing challenges in the future,” Rawson said. – by Eamon Dreisbach
Disclosure: Rawson reports no relevant financial disclosures. Please see the study for all other authors’ relevant financial disclosures.