Geographical variation in antimicrobial use and multiresistant pathogens in Brazilian intensive care units: a nationwide study

Introduction: Geographical analyses of antibiotic use identify regions with the highest consumption and help design policies for strategic patient groups. Methodology: We conducted a cross-sectional study based on official data available in July 2022 from Brazilian Health Surveillance Agency (Anvisa). Antibiotics are reported as a defined daily dose (DDD) per 1,000 patient-days, and central line-associated bloodstream infection (CLABSI) is defined according to Anvisa criteria. We also considered multi-drug resistant (MDR) as the critical pathogens the World Health Organization listed. We measured antimicrobial use and CLABSI trends per ICU bed using the compound annual growth rate (CAGR). Results: we evaluated the regional variation in CLABSI by multidrug-resistant pathogens and the antimicrobial use in 1,836 hospital intensive care units (ICUs). In 2020, the leader in use in intensive care units (ICUs) in the North was piperacillin/tazobactam (DDD = 929.7) in the Northeast. Midwest and South were meropenem (DDD = 809.4 and DDD = 688.1, respectively), and Southeast was ceftriaxone (DDD = 751.1). The North has reduced polymyxin use (91.1%), and ciprofloxacin increased (439%) in the South. There was an increase in CLABSI by carbapenem-resistant Pseudomonas aeruginosa in the North region (CAGR = 120.5%). Otherwise, CLABSI by vancomycin-resistant Enterococcus faecium (VRE) increased in all regions except the North (CAGR = -62.2%), while that carbapenem-resistant Acinetobacter baumannii increased in the Midwest (CAGR = 27.3%). Conclusions: we found heterogeneity in antimicrobial use patterns and CLABSI etiology among Brazilian ICUs. Although Gram-negative bacilli were the primary responsible agent, we observed a notable increase trend of CLABSI by VRE.


Introduction
The World Health Organization (WHO) recommends establishing healthcare-associated infection (HAIs) surveillance programs at national and institutional levels [1].Approximately 70% of all patients admitted to intensive care units (ICU) are treated with antibiotics [2,3], which makes them more prone to infections by multidrug-resistant microorganisms [4].Several factors influence the rapid spread of multidrug-resistant pathogens in the ICU [4], including antibiotic exposure [5].The Global Antimicrobial Resistance Surveillance System (GLASS) Report presented an alarming scenario for AMR rates in Brazil [6].However, the data in this report encompasses the entire hospital, including community and nosocomial infections.
Antibiotic consumption surveillance contributes to the underuse, misuse, or overuse identification [7], and it is crucial to compare antibiotic use patterns in different contexts [8].Many studies reported substantial geographic variation in antibiotics consumption in the community, especially for broad-spectrum antibiotics [9][10][11][12][13].Thus, geographical comparison studies may identify regions with the highest consumption [14].For example, Latin America consumes more reserve antibiotics in adult inpatients compared to other continents, followed by West and Central Asia [15].
European countries showed significant differences in antibiotic consumption in hospitals [16].Brazilian area is equivalent to about 81% of the European continent [17], and we hypothesized that heterogeneity in antimicrobial use trends and the HAIs epidemiology between Brazilian regions might exist.In 2011, the Brazilian Health Surveillance Agency (Agência Nacional de Vigilância Sanitária -Anvisa) implemented a national HAI surveillance system for reporting these nosocomial infections and antibiotic use in Brazilian ICUs [18].Therefore, our objective was to investigate the geographic variation in HAIs related to central line-associated bloodstream infection (CLABSI) by MDR pathogens and antimicrobial profile use in Brazilian ICUs, analyzing the notifications from this database.

Study design
We performed a cross-sectional study with secondary data from the CLABSI and antimicrobial use in ICUs Brazilian database following the RECORD-PE guidelines.The RECORD-PE statement was derived from rigorous methodology and endorsed by the International Society for Pharmacoepidemiology [19].

Setting and Participants
Brazil is constituted of 27 states grouped into five regions: North, Northeast, Midwest, Southeast, and South (Supplementary Table 1).
We included all hospitals that adhered to the national epidemiological surveillance system for HAIs.Anvisa establishes that hospitals with more than ten ICU beds must report monthly the main HAIs and antimicrobials used [20,21].We also evaluated the state reporting compliance by considering how many services registered in the National Registry of Health Establishments (Cadastro Nacional de Estabelecimento de Saúde -CNES) were eligible for CLABSI notification and how many services reported CLABSI per year.Anvisa considers as adherent those hospitals that have reported at least ten months each year.

Variables and outcomes
Antibiotics are reported as a defined daily dose (DDD) per 1,000 patient-days.We classified antimicrobials by the Anatomical Therapeutic Chemical (ATC) at the 4 th level granularity [22] and the AWaRe classification [23].Three groups compose the AWaRe classification: Access antibiotics include antibiotics with a lower potential for resistance than the other groups.Then, these antimicrobials are first or second-choice empirical treatments for infectious syndromes.Watch antibiotics should be monitored closely, and Reserve antibiotics should be considered the last therapeutic option.Both groups should be the focus of antimicrobial stewardship programs [23,24].We listed the antimicrobials monitored by Anvisa in Supplementary Table 2.
Anvisa defines CLABSI with the following criteria: (i) central catheter use for a period longer than two days (D1 being the day the device was installed) and that on the date of infection, the patient was using the device, or it was removed previous day; (ii) pathogen identified in one or more blood cultures; and (iii) the microorganism identified is not related to another infectious focus [25].

Data sources
Between July and August 2022, we extracted data on CLABSI and antimicrobial use in intensive care units from Brazilian hospitals through the Anvisa website from 2012 to 2020 [27] and the ICU beds for this period from the CNES database [28].However, due to data availability, we could only evaluate the antimicrobial use in the ICU from 2018 onwards and at a regional level from 2019 onwards.

Data analysis
We calculated the ratio between CLABSI prevalence and ICU beds for each state and region in the respective year.We also evaluated antimicrobial use and CLABSI trends per ICU bed using the compound annual growth rate (CAGR).CAGR reflects the average annual change as a proportion (%) of use in the initial year [16].We calculated the relative frequencies using the total pathogen isolated regardless of the sensitivity profile as the denominator.All the analyses were conducted at the state, regional, and national levels.We use the R program and Epitools package for data analysis.Since we used public, no ethical appraisal was required.

Results
Data on CLABSI prevalence and antimicrobial use from 1,836 hospitals were included in the analysis.The main pathogen responsible for CLABSI during the study period was 3 rd /4 th generation Cephalosporinresistant Klebsiella pneumoniae, followed by carbapenem-resistant K. pneumoniae, carbapenemresistant Acinetobacter baumannii, and MRSA.All carbapenem-resistant Gram-negative bacilli showed a negative trend, except Klebsiella pneumoniae.However, CLABSI by VRE increased by 19.3% (Table 1).
Table 2 presents the absolute and relative CLABSI frequencies among Brazilian regions.The prevalence of pathogens showed a heterogeneous pattern, with the North presenting the highest CRPA and the South presenting the lowest.
A heterogeneous CLABSI was found among the different Brazilian regions (Table 3).Only the North region presented an increase in CLABSI per CRPA (CAGR = 120.5%);CLABSI per VRE increased in all regions except in the North (CAGR = -62.2%);PSBI and MRSA showed a negative trend in all regions, and CLABSI per CRAB increased only in the Midwest (CAGR = 27.3%).Table 4 shows the prevalence of CLABSI produced by the pathogens under analysis from 2013 to 2020.All pathogens showed growth except oxacillin-and vancomycin-resistant Staphylococcus aureus.
In 2020, of the 1856 hospitals that should report CLABSI to Anvisa's HAIs surveillance system, 1,720 did (93% adherence).Complete data on adherence from 2013 to 2020 by the Brazilian state is available in Supplementary Table 4.

Discussion
To our knowledge, our study is the first to investigate antimicrobial use and the prevalence of CLABSI by MDR in ICUs throughout Brazil.We used official data collected by Anvisa.The results demonstrated significant geographic variation in both parameters analyzed in Brazilian ICUs.Unlike Brazil, Switzerland showed a consistent decline in MRSA across the country's ICUs [29], and Swedish researchers noted a consistent increase in 3 rd and 4 th -generation cephalosporin-resistant K. pneumoniae and E. Coli [30].Geographical variations in antimicrobial consumption are well described in the literature [16,[31][32][33].However, regional epidemiological differences do not always explain these variability [34].We observed heterogeneous trends in the growth of antimicrobials use across Brazilian regions that did not follow the CLABSI pathogen trends.For example, CLABSI by MRSA declined in all regions, but the use of glycopeptides increased in all locations, with a considerable increase in the South region.However, the Midwest, which had the lowest MRSA reduction rate, was the only region that reduced teicoplanin use and also had the lowest vancomycin growth rate.According to the European Center for Disease Prevention and Control (ECDC), glycopeptide consumption has increased in France, Croatia, Estonia, and Hungary [35].All these countries showed an increase in MRSA isolates, which may explain the higher glycopeptide consumption [36].ECDC data covered the entire hospital sector, while our study was ICU-specific.
Sjövall et al. conducted a study in Swedish ICUs where they described antimicrobial consumption.The authors report that the most consumed antibiotics were isoxazolyl penicillins (ATC group J01CF); penicillins with beta-lactamase inhibitors, mainly piperacillin/tazobactam (J01CR); 3 rd or/and 4 th generation cephalosporins (J01DD or DE); and carbapenems (J01DH) [30].Previous studies reported discrepant prevalence points in Brazilian ICUs t [37,38].Porto and colleagues described ceftriaxone, meropenem, and vancomycin as the most used antimicrobials [38].On the other hand, Nunes Castro et al.
Our result showed that the most used antibiotics in Brazilian ICUs were meropenem (J01DH), followed by ceftriaxone (J01DD), piperacillin/tazobactam (J01CR), and polymyxin (J01XB).Additionally, meropenem utilization increased between 2018 and 2020 in Brazilian ICUs, which the high prevalence of cephalosporin-resistant pathogens may explain.Furthermore, we noticed that the incidence of ICUacquired infection, as well as the prevalence of MDR pathogen infections, are higher in LMICs than in developed countries [39] and, consequently, the Reserve antimicrobials consumption is also higher in LMICs [15,40].As expected, Brazilian ICU patients are more exposed to Reserve antibiotics than others: We found 40% of Reserve antibiotics in Brazilian ICUs, compared to 16% in Sweden in 2018.
Gram-negative MDR bacilli are the main pathogens responsible for HAIs in low and middle-income countries (LMIC) like Brazil [39,41].However, we identified that CLABSI by VRE increased in all regions except in the North.The South region presented the most significant increase in CLABSI by VRE but the lowest growth in overall CLABSI by MDR.Although VRE is not primarily responsible for CLABSI in Brazilian ICUs, excessive glycopeptide utilization could be among the causes of the growth of CLABSI by VRE.
The South region presented the most favorable scenario for overall CLABSI by MDR, probably because one of its three states (i.e., Paraná) reported having fully implemented a GLASS antimicrobial surveillance program [42].In contrast, the South was the region with the most significant increase in linezolid and polymyxin B use, which is not following CLABSI by MDR trends, especially when we identified a considerable increase in CLABSI by polymyxinresistant Klebsiella pneumoniae.Although the susceptibility in vitro assays for polymyxins has several limitations and does not always provide reliable results [43], this potentially inappropriate antimicrobial selection discordance should be considered.
Brazil is still lagging in epidemiological HAIs surveillance compared to countries such as the United States [44] or Sweden [33].Our study used data from the national HAI surveillance system where not all regions had 100% reporting adherence.In addition, Brazil monitors HAIs through passive surveillance, which usually has low sensitivity and can lead to misclassification and underreporting [45].Anvisa has recently implemented policies to reduce the MDR prevalence in hospitals [46].

Conclusions
This paper summarizes the antimicrobials use and the CLABSI by MDR prevalence with the respective growth trends from 2018 to 2020 in Brazilian ICUs.In summary, we identified and visualized an important heterogeneity in the antimicrobial utilization in Brazilian ICUs and a discrepant CLABSI microbiological profile across Brazilian regions.Some inappropriate antibiotic use trends include regions where MRSA prevalence decreased while glycopeptide use increased.Despite the increasing overall VRE prevalence, Gram-negative bacilli remain the main responsible for CLABSI in Brazilian ICUs.Aiming to reduce the prevalence of HAI by MDR in all regions, Anvisa has recently implemented new policies.

Table 1 .
Critical pathogens responsible for central line-associated bloodstream infection in a Brazilian ICU weighted per 1000 bed.
CAGR: compound annual growth rate.

Table 2 .
Clinically relevant pathogens distribution responsible for central line-associated bloodstream infection among Brazilian regions in 2020.North, N (%

Table 3 .
Central Line-associated Bloodstream Infection density caused by clinically relevant pathogens in Brazilian regions per thousand ICU beds between 2019 -2020.

Table 4 .
Distribution of clinically relevant pathogens responsible for primary bloodstream infections during 2013 -2020 in Brazilian ICUs.
*p values were obtained using the chi-square test or Fisher's exact test; ¹ Anvisa data grouped in Enterococcus spp., it is not possible to measure rates for Enterococcus faecium; ² Anvisa did not monitor vancomycin-resistant Staphylococcus aureus in the period; ³Anvisa did not monitor polymyxin resistance during this period.;4Anvisaonly released the relative frequencies.CAGR: compound annual growth rate.

Table 5 .
Anti-infective agents use (DDD) notified to Anvisa between 2019 and 2020.Agência Nacional de Vigilância Sanitária; CAGR: compound annual growth rate, *p values were obtained using the chi-square test or Fisher's exact test.