Researchers have found there is no evidence of substantial risk for drug–drug interactions involving antibiotics, despite warnings of severe outcomes in national formularies and electronic health record software.
The authors of the study, which was published in Clinical Pharmacology and Therapeutics on 30 November 2022, said there was a need for the evidence base for the inclusion of drug–drug interactions in national formularies to be “strengthened” and to reduce “indiscriminate flagging”, which can lead to alert fatigue among healthcare professionals.
The study evaluated drug–drug interactions between antibiotic and non-antibiotic drugs listed with a warning of severe outcomes in the British National Formulary (BNF) based on adverse drug reactions (ADRs) detectable with routine International Classification of Diseases (Tenth Revision) coding.
Around 6.5% of hospital admissions are thought to be related to ADRs, with drug–drug interactions accounting for 16.6% of these, the paper states. However, the researchers say there is limited evidence showing which pairs of medicines increase drug–drug interaction risks.
To investigate, the authors collated primary care data from the Clinical Practice Research Databank and anonymised electronic health records from general practices in England that were linked to hospital admission records. Next, they analysed 121,546 ADR-related admission cases matched to 638,238 controls.
Several drug–drug interactions were found to have adjusted odds ratios (ORs) that were statistically significantly increased compared with non-exposure, but not statistically significant compared with single exposure of one of the medicines.
Of the 51 drug–drug interactions (antibiotic/non-antibiotic) evaluated for ADR-related hospital admissions, 38 (74.5%) had adjusted ORs that were significantly increased when compared with non-exposure.
However, single exposure to one of the two drugs, particular the antibiotic, also had increased ORs, and the difference between concomitant use and single exposure was small. This indicated that the presence of substantial effects was associated with antibiotic use rather than drug–drug interaction effects.
“No evidence of substantial risk was found for multiple drug–drug interactions with antibiotics, despite warning of severe outcomes in a national formulary and automated flagging, which compounds alert fatigue,” the authors concluded, adding that publicly available evidence should be provided before including drug–drug interactions into national formularies and electronic health record prescribing alerts.
Both the BNF and related website do not routinely publish the evidence for including drug–drug interaction warnings, other than brief classifications,” the authors said.
“We were also unable to find reasons why DDIs were selected for electronic health record flagging (including the alerts in EMIS, the UK’s most widely used primary care software), or what the effectiveness is of such flagging.”
Claire Preston, content manager for interactions, ADRs and monitoring at Pharmaceutical Press, which publishes the BNF [as well as The Pharmaceutical Journal], said that timely access to information about drug interactions was a “vital part” of patient safety.
“Much of the data around drug interactions comes from clinical pharmacokinetic studies in patients under 65 years without comorbidities,” she said.
“As such, we agree with the authors that there is a lack of systematically collected evidence on the impact of drug–drug interactions in susceptible patient groups in usual clinical practice.
“BNF interactions content is primarily focused on pharmacokinetic drug–drug interactions, with some pharmacodynamic interactions also listed.
“Severity ratings present the possible worst-case outcome if an interaction is not managed.”
Preston added that content within the BNF around drug interactions should be viewed alongside cautions, contraindications and side effects for the drugs in question, in order to make a clinical decision on treatment depending on an individual patient’s circumstances.