Back

Community Antibiotic Prescribing in Patients with COVID-19 Across Three Pandemic Waves: A Population-Based Cohort Study

Ciaccio, L.; Donnan, P. T.; Parcell, B. J.; Marwick, C. A.

2023-07-01 infectious diseases
10.1101/2023.06.29.23291797 medRxiv
Show abstract

BackgroundReported changes in antibiotic prescribing during the COVID-19 pandemic have focused on hospital prescribing or community population trends. Community antibiotic prescribing for individuals with COVID-19 are less well described. MethodsData covering a complete geographic population ([~]800,000) were utilised. SARS-CoV-2 virus test results from February 1, 2020-March 31, 2022 were included. Anonymised data were linked to prescription data +/-28 days of the test, GP data for high-risk comorbidities, and demographic data. Multivariate binary logistic regression examined associations between patient factors and the odds of antibiotic prescription. ResultsData included 768,206 tests for 184,954 individuals, identifying 16,240 COVID-19 episodes involving 16,025 individuals. There were 3,263 antibiotic prescriptions +/-28 days for 2,385 patients. 35.6% of patients had a prescription only before the test date, 52.5% of patients after, and 11.9% before and after. Antibiotic prescribing reduced over time: 20.4% of episodes in wave one, 17.7% in wave two, and 12.0% in wave three. In multivariate logistic regression, being female (OR 1.31, 95% CI 1.19,1.45), older (OR 3.02, 95% CI 2.50, 3.68 75+ vs <25 years), having a high-risk comorbidity (OR 1.45, 95% CI 1.31, 1.61), a hospital admission +/-28 days of an episode (OR 1.58, 95% CI 1.42, 1.77), and health board region (OR 1.14, 95% CI 1.03, 1.25, board B versus A) increased the odds of receiving an antibiotic. ConclusionCommunity antibiotic prescriptions in COVID-19 episodes were uncommon in this population and likelihood was associated with patient factors. The reduction over pandemic waves may represent increased knowledge regarding COVID-19 treatment and/or evolving symptomatology.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.