Surveillance cultures as a tool for choosing empirical antibiotic therapy

Background A daily challenge for the multidisciplinary team in intensive care units (ICUs) is balancing broad-spectrum antibiotics with the appropriate empirical antibiotic therapy. Aim To establish the carbapenem-resistant Gram-negative bacilli screening cultures predictives values. Methods We conducted a retrospective study. We included patients admitted to the intensive care unit for at least 48 hours. We measured carbapenem-resistant negative and positive predictive values, sensitivity, and speci�city in Gram-negative bacilli screening cultures. Results


Introduction
Antibiotic therapy evaluation is essential for critical patient pharmaceutical care [1].A daily challenge for the multidisciplinary team in intensive care units (ICUs) is balancing broad-spectrum antibiotics with the appropriate empirical antibiotic therapy [2].In addition, expert consensus recommends early appropriate antibiotics administration in septic patients [3,4].However, the consumption of high-spectrum antimicrobials is a major global concern [5][6][7].
The patient's previous colonization and clinical factors guide the choice of empirical treatment [8].Although screening cultures are not diagnostic tests, they are often used as guides in choosing antibiotic therapy until knowledge of the etiological agent [9].However, there is disagreement regarding the predictive values of cultures from screenings of extended beta-lactamase-producing Gram-negative bacilli (K.pneumoniae and Enterobacter spp.).Positive predictive value (PPV) is the probability of a speci c diagnostic test nding positive values for really ill individuals.On the other hand, the negative predictive value (NPV) is the test's ability to present a negative result for individuals who do not have the disease [10].
Two robust studies with more than three thousand patients each conducted in the ICU differed in their respective ndings [11,12].The study conducted in the French ICUs demonstrated that previous colonization by third-generation cephalosporin-resistant Enterobacteriaceae was the leading risk factor for subsequent infection [12].Previously, Rottier et al. (2015) stated that preceding colonization with Enterobacteriaceae resistant to third-generation cephalosporins has low positive predictive value for infections caused by these pathogens, and strict adherence to guidelines would unnecessarily encourage the use of broad-spectrum antibiotics [11].In addition, we have no data about the carbapenem-resistant Gram-negative bacilli as A. baumannii and P. aeruginosa.
Therefore, the multidisciplinary intensive team must understand the real role of screening cultures in predicting the etiologic agent responsible for subsequent infections.Thus, this study aimed to establish the predictive value of screening carbapenem-resistant Gram-negative bacilli cultures and producing extended-spectrum β-lactamase Enterobacteriaceae spp.

Methods
The retrospective observational study was carried out in a tertiary hospital in Rio de Janeiro with 52 intensive care beds.The six clinical pharmacists are part of the ICUs multidisciplinary team.It included patients admitted in ICUs who presented infection in 2019.We excluded patients with an ICU stay of fewer than 48 hours, aged less than 18 years old, and who did not use antimicrobials.Microbial samples were collected weekly for screening for clinically relevant bacterial cultures.
We characterized the study population based on data from medical records.We collected age, gender, Simpli ed Acute Physiology Score 3 (SAPS 3), Charlson Comorbidity Index (CCI), mechanical ventilation, renal replacement therapy [13], vasoactive amine, previous colonization, blood transfusion up to seven days before the infection [14,15], parenteral nutrition[16], and previous exposure to systemic antimicrobials up to 90 days before the infection [17].We adopt the infection de nition as the case that a pathogen isolated in culture for diagnosis with clinical interpretation of the infectious process [18].The outcome was infection by carbapenem-resistant Gram-negative bacilli (GNB) or extended-spectrum betalactamase-producing GNB in patients previously colonized by these.
We de ned prior colonization positive results for extended-spectrum beta-lactamase-producing Enterobacteriaceae, carbapenem-resistant Pseudomonas aeruginosa, carbapenem-resistant Acinetobacter baumannii, carbapenem-resistant Enterobacteriaceae spp. in surveillance cultures before infection [12].We used the Rstudio® program for statistical analysis.A 95% con dence interval was adopted with a pvalue <0.05 to be considered statistically signi cant.First, we performed a descriptive analysis.The quantitative variables were expressed as median or mean, and data dispersion was estimated using the interquartile range (25%-75%) or standard deviation.The categorical variables were expressed as absolute and relative frequencies.
We assessed the data using the non-parametric MannWhitney test for continuous variables.The chisquare tests or Fisher's exact test compared categorical variables.We submitted the variables with pvalues less than 0.2 to the logistic regression model We calculated the relative infection risk in patients previously colonized with extended-spectrum betalactamase-producing Enterobacteriaceae spp., carbapenem-resistant Pseudomonas aeruginosa, carbapenem-resistant Acinetobacter baumannii, isolated carbapenem-resistant Enterobacteriaceae spp.Finally, we measured the positive and negative predictive values, sensitivity, speci city, and likelihood ratio by epiR package.The local research ethics committee approved the study under CAAE: 25683019.4.0000.5249.

Results
Six hundred and fty-one patients had at least one microorganism isolated during the study period of 4,250 admissions in 2019.Of these, 282 were excluded from admission to the ICU for less than 48 hours (Figure 1).The characterization of the study population is presented in Table 1.Enterobacteriaceae,;. # The categorical variables were expressed in absolute frequencies and, in between parenthesis, the relative frequencies.The continuous variables were expressed in median and, in between parenthesis, the 25%-75% interquartile range.SAPS 3 = Simpli ed Acute Physiology Score 3.
There was no methicillin-resistant Staphylococcus aureus isolated in the screening culture.Nine vancomycin-resistant Enterococcus spp were isolated in screening cultures and none in the diagnosis culture.The previous colonization by carbapenem-resistant Pseudomonas aeruginosa, carbapenemresistant Acinetobacter baumannii, and carbapenem-resistant Enterobacteriaceae showed to be associated with risk factors for subsequent infection for these pathogens.However, previous colonization by extended-spectrum beta-lactamase-producing Enterobacteriaceae was not found as a risk factor for subsequent infection by ESBL pathogens in this study (Figure 2).

Discussion
The PPVs found were low and the NPVs high, suggesting that screening cultures were e cient in establishing that carbapenem-resistant Gram-negative bacilli and extended-spectrum beta-lactamaseproducing GNB rarely infect patients not colonized by these pathogens.On the other hand, previously colonized patients will not necessarily be infected by the pathogens.The sensitivity and speci city values reinforced this nding.In addition, the observed accuracy of predicting etiologic agents by screening cultures was low.
The positive predictive values and sensitivity for subsequent infections by extended beta-lactamaseproducing Enterobacteriaceae found by Massart et al. (2020) [12] were twice as high compared to the study by Rottier et al. (2015) [11].Both studies were carried out in ICU.Our ndings are consistent with those stipulated by Rottier and colleagues [11].However, these authors did not determine the negative predictive values.Therefore, we compared our NPV and speci city with those pointed out by Massart and collaborators, and the results are similar [12] (NPV >85% and speci city greater than 90% in both studies).
We did not previously nd the predictive carbapenem-resistant Gram-negative screening cultures values in the literature.However, prior colonization by carbapenem-resistant GNB is a signi cant risk factor for subsequent infection by these pathogens [19], and our data showed this relationship.Although, we observed that patients previously colonized with carbapenem-resistant GNB did not present infection by these pathogens most of the time.
The pharmacist is an essential member of the antimicrobial stewardship program within hospitals [20].
The clinical pharmacist participation in the antibiotic choice contributes to more appropriate use, especially in developing countries [21].We suggest that the screening cultures analysis can be an important tool for pharmaceutical intervention regarding empirical antimicrobials.Our results indicate that previously colonized septic patients should receive antibiotic therapy considering the previous colonization (e.g., carbapenem-resistant Pseudomonas aeruginosa, carbapenem-resistant Acinetobacter baumannii, and carbapenem-resistant Enterobacteriaceae).A delay in administering adequate antibiotic treatment is a factor for mortality in this population [3,9].However, as most of the patients are not infected by the pathogens by which they are previously colonized, it would be reasonable to reserve broad-spectrum antibiotics for unstable patients with organ dysfunction.Antibiotics used to treat carbapenem-resistant non-fermenting Gram-negative bacilli are considered a last therapeutic resort [22] and should only be reserved if these pathogens have a strong suspicion of infection.
Although low and middle-income countries (LMIC) publish fewer studies and are less robust than highincome countries [23], LMIC has the highest carbapenem-resistant GNB infection prevalence.This study involved only one center and retrospective data collection.Despite these limitations, we measured for the rst time the predictive values of non-fermenter carbapenem-resistant Gram-negative bacilli.These pathogens are a major public health problem [6], especially in LMIC [24].The results obtained provide evidence on the role of culture screening in predicting etiological agents responsible for infections in critically ill patients.In addition, they may contribute to choosing appropriate empirical antibiotic therapy for patients in the ICU, promoting more rational antimicrobials use.

Conclusion
Previous colonization by carbapenem-resistant Pseudomonas aeruginosa, carbapenem-resistant Acinetobacter baumannii, and carbapenem-resistant Enterobacteriaceae showed risk factors for subsequent infection.However, the screening cultures' negative predictive values and observed speci city were high, indicating that uncolonized patients will rarely become infected by these pathogens.This result may contribute to the choice of empirical antibiotic therapy, discouraging the prescription of antibiotics against carbapenem-resistant Pseudomonas aeruginosa, Acinetobacter baumannii, and Enterobacteriaceae spp.

Figures
Figures

Figure 2 Relative
Figure 2

Table 2
presents each previous colonization's positive and negative predictive values, sensitivity, speci city, likelihood ratio, and accuracy.Screening cultures showed high negative predictive values and speci city and low positive predictive values and sensitivity.Finally, we present the odds ratio adjusted for the covariates in Table3.All carbapenem-resistant Gram-negative bacilli displayed values with statistical signi cance.