Antibiotic overuse and mortality in patients hospitalised for COVID-19: a global cohort study.

BACKGROUND: Global antibiotic use during the COVID-19 pandemic and its impact remain unclear. This study analysed antibiotic use, bacterial co-infections, and the association between antibiotic use and in-hospital mortality among patients hospitalised for COVID-19, using a WHO database.

METHODS: We conducted a multicountry cohort study using anonymised patient-level data from The WHO Global Clinical Platform for COVID-19. Data were collected across 64 countries in all six WHO regions between January, 2020, and October, 2023. The study included hospitalised patients of all ages with suspected or confirmed SARS-CoV-2 infection. We obtained patient-level data from standardised WHO case report forms completed prospectively or retrospectively and from harmonised national registries, research studies, and hospital databases. Patients without a suspected or confirmed COVID-19 diagnosis or missing information on hospital admission date, antibiotic use, or outcome were excluded from the analysis. Patients were followed up from hospital admission until death, discharge, transfer, or 28 days, whichever occurred first. Follow-up data were collected at hospital discharge or death by the treating facility and entered into the platform. The primary outcome was 28-day in-hospital mortality, and we analysed this outcome using Cox proportional hazards models with a shared frailty term to account for country-level clustering.

FINDINGS: The median age of the study population was 61·9 years (±21·7), with 198 223 women and 251 990 men with data available on antibiotic use. Of the 592 898 patients enrolled, 451 291 (76·1%) had data on antibiotic use and were included in the descriptive analysis; 122 854 had complete data on empirical treatment type, of whom 47 359 (38·5%) were included in the regression analysis. Among the 451 291 patients with data on antibiotic use, 341 125 (75·6%) received antibiotics, more frequently in severe to critical cases than in mild to moderate cases (80·5% vs 71·6%). Prescribing was highest in the Eastern Mediterranean (83·0%) and African (82·7%) regions and lowest in the Western Pacific region (32·8%), declining in the European region and region of the Americas but increasing in the African region between 2020 and 2022. Treatment was predominantly empirical and involved Watch antibiotics of the WHO Access, Watch, and Reserve classification. Clinically suspected co-infections were reported in 47 050 (7·9%) of the 592 898 patients; among those with positive cultures, 1815 (42·8%) of 4237 had multidrug-resistant bacteria. In the unadjusted analyses, empirical antibiotic use was associated with higher mortality across severity strata. After adjustment for age, sex, comorbidities, and disease severity, empirical antibiotic use remained associated with increased mortality in mild to moderate cases (adjusted hazard ratio 1·80, 95% CI 1·36-2·38) and severe to critical cases (1·16, 1·08-1·24).

INTERPRETATION: Despite clinically suspected co-infections being reported in only 47 050 (7·9%) of the 592 898 enrolled patients, 341 125 (75·6%) of the 451 291 patients with data on antibiotic use received antibiotic therapy, predominantly broad-spectrum agents. As antibiotic use was not associated with improved clinical outcomes and is known to increase antimicrobial resistance, it should be rationalised through strengthened antimicrobial stewardship and behavioural interventions, particularly during respiratory viral epidemics.

FUNDING: None.