A moral Turing test: How belief and source shape detection of and agreement with LLM judgments.

The COVID-19 pandemic led to significant global mortality and morbidity. The World Health Organization (WHO) reported 5.42 million excess deaths worldwide due COVID-19 from January 2020 to December 2021; however, estimates suggest this number could be as high as 14.83 million over the same period. In 2022, COVID-19 was ranked third leading cause of death worldwide, after coronary artery disease and dementia. Secondary infections in COVID-19 patients may be categorised as co-infections, diagnosed at the time of initial COVID-19 presentation, or superinfections, diagnosed during the clinical course. In earlier work examining hospitalised patients with COVID-19 in Victoria, Australia, we identified secondary infections in 6.9% of admissions, driven largely by secondary bacterial infections, which occurred in 6.0% of admissions. Meta-analyses report pooled bacterial co-infection rates up to 27% and bacterial superinfection rates up to 20%. Concerns about secondary bacterial infections frequently prompt clinicians prescribing presumptive antibiotics before test results in hospitalised COVID-19 patients. In intensive care settings, a systematic review estimated bacterial co-infection prevalence to be 30.8%, yet more than half of the reviewed studies reported antibiotic prescribing for all patients. Inappropriate antibiotic use contributes significantly to antimicrobial resistance. A meta-analysis estimated the prevalence of resistant bacterial co-infection among COVID-19 patients at 24%.