Association of ward-level antibiotic consumption with healthcare-associated Clostridioides difficile infections: an ecological study in five German university hospitals, 2017-2019.

ABSTRACT

OBJECTIVES: To analyse the influence of antibiotic consumption on healthcare-associated healthcare onset (HAHO) Clostridioides difficile infection (CDI) in a German university hospital setting. METHODS: Monthly ward-level antibiotic consumption measured in DDD/100 patient days (pd) and CDI surveillance data from five university hospitals in the period 2017 through 2019 were analysed. Uni- and multivariable analyses were performed with generalized estimating equation models. RESULTS: A total of 225 wards with 7347 surveillance months and 4 036 602 pd participated. With 1184 HAHO-CDI cases, there was a median incidence density of 0.17/1000 pd (IQR 0.03-0.43) across all specialties, with substantial differences among specialties. Haematology-oncology wards showed the highest median incidence density (0.67/1000 pd, IQR 0.44-1.01), followed by medical ICUs (0.45/1000 pd, IQR 0.27-0.73) and medical general wards (0.32/1000 pd, IQR 0.18-0.53). Multivariable analysis revealed carbapenem (mostly meropenem) consumption to be the only antibiotic class associated with increased HAHO-CDI incidence density. Each carbapenem DDD/100 pd administered increased the HAHO-CDI incidence density by 1.3% [incidence rate ratio (IRR) 1.013; 95% CI 1.006-1.019]. Specialty-specific analyses showed this influence only to be valid for haematological-oncological wards. Overall, factors like ward specialty (e.g. haematology-oncology ward IRR 2.961, 95% CI 2.203-3.980) or other CDI cases on ward had a stronger influence on HAHO-CDI incidence density (e.g. community-associated CDI or unknown association case in same month IRR 1.476, 95% CI 1.242-1.755) than antibiotic consumption. CONCLUSIONS: In the German university hospital setting, monthly ward-level carbapenem consumption seems to increase the HAHO-CDI incidence density predominantly on haematological-oncological wards. Furthermore, other patient-specific factors seem to be equally important to control HAHO-CDI.