Empiric combination antibiotic therapy (ECAT) is commonly used to treat healthcare-associated bloodstream infections (HA-BSIs) and sepsis. However, the level of supporting evidence is low and clinical practice varies significantly. We conducted a post hoc analysis using the EUROBACT-2 international cohort study database, which contained data on 2406 adult patients from 328 intensive care units (ICUs) across 52 countries, collected between June 2019 and January 2021. The main outcome was the proportion of patients receiving ECAT for HA-BSIs. Patient and institutional factors influencing the use of ECAT were examined using Markov-Chain Monte Carlo estimation. Three quarters of patients (75.2%; n = 1810) received empiric antibiotic therapy, with ECAT used in approximately half of cases (52.5%; n = 950). Most patients receiving ECAT (70.4%; n = 669) were treated with two antibiotics, beta-lactams plus glycopeptides being the most common combination (40.2%; n = 382). The odds of ECAT were increased by immune deficiency (OR 1.35 [95% CrI 1.03-1.75]), SOFA scores > 11 (OR 1.77 [95% CrI 1.28-2.46]), uncommon sources of infection (OR 1.63 [95% CrI 1.02-2.59]), and admission to ICUs where > 25% of Enterobacteriaceae isolates produce carbapenemases (OR 2.46 [95% CrI 1.37-4.41). The intra-class correlation coefficients at the ICU and country levels were 23.2% and 4.4%, respectively. In conclusion, factors at the individual, institutional, and national levels may affect the use of ECAT to treat HA-BSIs. Given the impact of institutional variables on the use of ECAT and the inconclusive evidence regarding its potential risks, it is of great importance that treatment is tailored based on local antibiotic stewardship programs and the needs of the individual patient.