Discussion

Improving the use of antibiotics and the knowledge of bacterial resistance represent key objectives of the Global Action Plan on Antimicrobial Resistance, as presented by the WHO and endorsed in Brazil.^1,15 A significant strategy involves investigating the relationship between antimicrobial use and resistance. This study aimed to explore the patterns of antibiotic utilization in the ICUs and their correlation with bacterial resistance in bloodstream isolates within the Health Surveillance System of a Brazilian state. Furthermore, we examined antibiotic use across different groups of hospitals categorized by administrative type – public, private, and philanthropic – a unique feature of Brazil’s healthcare sector.

Until 2015, antibiotic use data at the hospital level based on a systematic national surveillance system were restricted to a few countries,^{Reference Plüss-Suard, Pannatier, Kronenberg, Mühlemann and Zanetti7,Reference Dumartin, L’Hériteau and Péfau16–Reference Qu, Yin and Sun18} while antibiotic use reports based on regional statistics^{Reference Grau, Fondevilla and Mojal5,Reference Domínguez, Rosales, Cabello, Bavestrello and Labarca17,Reference Versporten, Bolokhovets and Ghazaryan19} or restricted to local intensive care units^{Reference Hanberger, Arman and Gill20–Reference Meyer, Schwab, Schroeren-Boersch and Gastmeier22} were relatively more common. However, the establishment and expansion of GLASS have significantly increased surveillance of AMR and antibiotic use with national data from different regions of the globe. In 2021, 109 countries enrolled in GLASS.²³ Our study presents BSIs data from an AMR surveillance period before BR-GLASS, describing antibiotic use in intensive care units in the State of São Paulo, whose SHD was a national pioneer in the composition of an information surveillance system for hospital infections.

The global use of antibiotics in the ICU found in our study is lower than previously reported in ICU by other studies: 916 DDD/1000 patients-day in Porto Alegre, Brazil, 1140 DDD/1000 patients-day in Germany, 1143 DDD/1000 patients-day in Switzerland, 1466 DDD/1000 patients-day in France and 1471 DDD/1000 patients-day in Catalonia, Spain.^{Reference Jacoby, Kuchenbecker, Dos Santos, Magedanz, Guzatto and Moreira3,Reference Grau, Fondevilla and Mojal5,Reference Plüss-Suard, Pannatier, Kronenberg, Mühlemann and Zanetti7,Reference Dumartin, L’Hériteau and Péfau16,Reference de With, Meyer, Steib-Bauert, Schwab, Daschner and Kern21} However, different drug surveillance methods may explain multinational antibiotic use discrepancies.

In our study, cephalosporins were the most used therapeutic class, which differs from the practice observed in European ICUs in which penicillins, especially using amoxicillin-clavulanate, represent the group of antibiotics most used in intensive care units.^{Reference Grau, Fondevilla and Mojal5,Reference Plüss-Suard, Pannatier, Kronenberg, Mühlemann and Zanetti7,Reference Dumartin, L’Hériteau and Péfau16,Reference Agodi, Auxilia and Barchitta24} A Brazilian point-prevalence study showed that ceftriaxone was the most prescribed antibiotic in the sample evaluated, mainly for treating respiratory and urinary infections.^{Reference Porto, Goossens, Versporten and Costa25} The reasons for these differences in prescription patterns, favoring either amoxicillin-clavulanate or ceftriaxone as the first choice, remain unclear and warrant further investigation, possibly involving cultural factors or cost considerations.

Otherwise, cephalosporins use exhibited a decreasing trend over the observed period, possibly associated with an increase in the proportion of gram-negative bacteria resistant to these drugs. Globally, there is a trend toward increasing MDRO gram-negative in intensive care units,^{Reference Mascarello, Simonetti and Knezevich26–Reference Fowler and Lee28} which is generally associated with an increase in hospital length of stay, mortality, and hospital costs.^{Reference Chen, Lee, Su, Tang, Chang and Liu27} Our study also observed an increase in the proportion of CRAb and CRKp over the 10-year period.

Few studies have evaluated the difference in antibiotic use by administrative type. A French study,^{Reference Dumartin, L’Hériteau and Péfau16} showed a higher antibiotic use in private hospitals than in public institutions, a trend not observed in our study. Public hospitals exhibited the highest antibiotic use, while private institutions used more antibiotics than philanthropic hospitals. A previous Brazilian survey on community and hospital antibiotic use observed a greater burden of resistant bacteria in public hospitals, possibly attributable to patient overload in these institutions, which may affect healthcare quality.^{Reference Boszczowski, Neto and Blangiardo29}

Moreover, among public hospitals, we observed higher use of carbapenems, glycopeptides, and polymyxins in SHO hospitals. Notably, there was a positive correlation between glycopeptide use and MRSA incidence in SHO hospitals but not in DPA hospitals. For both types of administration, there was a positive correlation between polymyxins and the incidence of CR-GNB and no correlation between the use of meropenem and the incidence of 3GC-resistant Enterobacterales. Because these are unique administration profiles in other countries, there are no comparative data for these subgroups. SHO represents the private administration of public institutions with governmental funding.^{Reference Rodrigues and Spagnuolo30,Reference Barbosa and Elias31} The correlation of antibiotic use with bacterial resistance incidence suggests more efficient resource utilization, which has significant implications for Brazilian public health policy. However, further studies are needed to evaluate this hypothesis.

Typically, comparisons of antibiotic utilization among ICU types are based on the specialty of care, such as cardiology, medical and surgical, and there are limited studies describing differences based on the level of care complexity.^{Reference Fridkin, Edwards and Pichette32} In our study, ICUs were stratified into high or low complexity based on the rate of mechanical ventilation. Although it is not a formally validated parameter for the definition of care complexity, the percentage of mechanical ventilation usage allows inference of patient severity in the unit and has been described in another study as a method of ICU stratification for data benchmarking.^{Reference Wetzker, Bunte-Schönberger, Walter, Schröder, Gastmeier and Reichardt33}

We observed higher antibiotic utilization and incidence of MRSA, carbapenem-resistant gram-negative bacteria (CR-GNB), and third-generation cephalosporin (3GC)-resistant Enterobacterales in high-complexity ICUs. C Critically ill patients in the ICU, along with the associated risks of delayed therapy, contribute to a lower threshold for initiating antimicrobial treatment, and this hypothesis helps to explain the observed result.

We found a weak negative correlation between the proportion of 3GC-resistant Enterobacterales and carbapenems use. His finding was unexpected; however, one possible explanation is the concurrent administration of carbapenems in the treatment of CR-GNB,, which may be associated with the observed negative correlation between the proportion of 3GC-resistant Enterobacterales and CR-GNB.

Our study has several limitations. First, the ecological study design cannot establish the cause-and-effect relationship between antibiotic use and resistance. Second, during the period observed, there was a change in the notification criteria of bacteremias, and this does not allow for continuous analysis of the resistance incidence over the entire period. Third, although a significant difference in antibiotic use was found based on the complexity of the ICU, the criteria employed for stratification require validation by other studies. Fourth, significant differences in antibiotic use and MDRO incidence were observed in the different hospital administrative categories. However, the study design does not allow us to establish the causes related to these findings.

In summary, we found substantial variations in antibiotic utilization and MDRO incidence across different hospital administrative categories, highlighting the importance of tailored interventions based on healthcare facility characteristics. Notably, we observed a higher proportion of MDRO in public hospitals compared to private and philanthropic institutions, suggesting potential areas for targeted interventions to mitigate resistance emergence. Furthermore, our findings revealed an unexpected negative correlation between the proportion of third-generation cephalosporin (3GC)-resistant Enterobacterales and carbapenem use, underscoring the complexity of AMR dynamics in healthcare settings. Despite these insights, our study has inherent limitations, including its ecological design and the need for validation of ICU complexity stratification criteria. Future research efforts should focus on elucidating the causal relationships between antibiotic use and resistance patterns, thereby facilitating the development of evidence-based antimicrobial stewardship strategies tailored to diverse healthcare contexts.