Bloodstream Infections in the Intensive Care Unit: a Single-Center Retrospective Bacteriological Analysis Between 2007 and 2019

Hospital-acquired bloodstream infections (BSI) are a severe care problem worldwide, associated with significant mortality (Massart et al. 2021). According to Watson et al. (2019), hospital-acquired bloodstream infections result in a 1.3- to 4.3-fold increase in the hospitalization duration, 2.8- to 5.3-fold higher hospitalization costs and a 4.3- to 8-fold increase in the percentage of patients who died.

Observations from the last few years have shown an increased incidence of bloodstream infections (Salm et al. 2018; Tajima et al. 2021). Moreover, data from Denmark revealed a considerable increase in bacteremia between 2000 and 2014, although the all-cause 30-day mortality after first-time bacteremia decreased (Holm et al. 2021).

Over 50% of patients admitted to the intensive care unit (ICU) require central vein catheterization (CVC) (Blot et al. 2015). Central lines are necessary for multiple purposes, including drug infusion, blood sample collection, and hemodynamic monitoring. The presence of a catheter inside a central vein is associated with the risk of central line-associated bloodstream infections (CLABSI). In the USA, five million CVCs are inserted each year, leading to approximately 200,000 cases of CLABSI, and the number of deaths attributable to such infections may be as high as 25,000 (Blot et al. 2015). However, these estimates may be understated because some cases of BSI secondary to urinary, respiratory, and gastrointestinal infections may be assigned the International Classification of Disease (ICD) codes for those illnesses (Goto and Al-Hasan 2013).

The objective of this study was to analyze bloodstream infections in patients hospitalized in the intensive care unit of the Military Institute of Medicine between 2009 and 2017 and to determine the changes in the incidence of pathogens and their susceptibility/resistance to antimicrobial agents with particular attention paid to the incidence of CVC infections in the analyzed cohort.

The study analyzed data collected between 2007 and 2019. General information regarding the number of patients hospitalized in the ICU, the reasons for their hospitalization, the number of microbiological investigations of blood samples, and the results of these examinations are shown in Tables I and II.

Study population and blood samples. Between 2007 and 2019, 3,502 patients were hospitalized in our ICU. Among these, 2,581/3,502 were men (73.7%), and 921/3,502 (26.3%) were women. Among 46,253 microbiological specimens, 12,619 (27.3%) were obtained from blood samples. One thousand five hundred nine samples (12%) were positive, and 1,557 pathogens were isolated. Among these were 770/1,557 (49.5%) Gram-negative bacteria, 733/1,557 (47.1%) Gram-positive bacteria, and 54/1,557 (3.5%) fungal species. A CVC was inserted in 3,460 of the 3,502 (98.8%) patients. The CVC was inserted into a subclavian vein in 2,588/3,460 cases (74.8%), into the internal jugular in 685/3,460 cases (19.8%), and the femoral vein in 187/3,460 cases (5.4%). A CVC was implicated in 278/1,509 (18.4%) of positive blood cultures, and the number of infected central vein catheters was 278/632 (44.6%). The pathogens isolated from positive blood cultures are shown in Table II. Among all Gram-negative bacteria, A. baumannii was isolated quite often, in 312/770 (40.5%) of cases. It represented 312/1,557 (20%) of all isolated pathogens.

Isolated pathogens. Gram-negative bacilli were the most common isolated pathogens in our materials. The most frequently isolated pathogen was A. baumannii, followed by K. pneumoniae, and P. aeruginosa. Among Gram-positive cocci, the most frequent was MRCNS, followed by E. faecalis. A. baumannii was isolated from 312 samples (20% of all isolated pathogens and 40.5% (312/770) of all Gram-negative isolates). P. aeruginosa was obtained from 111 blood samples (7.4% (111/1,509)), representing 7.1% of all isolated bacterial species and 14.4% (111/770) of Gram-negative pathogens. Only in two A. baumannii isolates with the MBL mechanism of antibiotic resistance were found. K. pneumoniae was detected in 165 blood samples (11% (165/1,509)), and represented 10.6% (165/1,557) of all isolated pathogens and 21.4% (165/770) of Gram-negative bacterial isolates. Among all K. pneumoniae strains, the following antibiotic resistance mechanisms were identified: ESBL in 67 (40.6%) isolates, NDM in five, KPC – four, and OXA-48 in one isolate. E. coli was isolated from 69 samples and comprised 4.6% (69/1,509) of all and 9% (69/770) of Gram-negative isolates. Eleven of them (15.9%) exhibited an ESBL mechanism of resistance.

Staphylococci were the most frequent (74.4% (545/ 733)) isolates among the Gram-positive bacteria. The majority (368/545, 67.5%) of all isolated staphylococci were MRCNS. More than 78% (426/545) of staphylococci were resistant to methicillin (86.4% (368/426) of them were MRCNS). Enterococci were isolated from 184/1509 (12.2%) samples; 41/184 (22.3%) strains exhibited a high-level aminoglycoside resistance (HLAR), and 11 (6%) E. faecium species were resistant to vancomycin.

The resistances of isolated pathogens to antimicrobial agents are shown in Tables III and IV. The graphical presentation of these data is shown in Fig. 1, 2, and 3. The above data include the number of Gram-positive (Table III, Fig. 2) and Gram-negative (Table III, Fig. 3) alert pathogens with their resistance mechanisms. All MRSA and MRCNS strains were susceptible to vancomycin during the analyzed period. From 2010, all MRSA strains were susceptible to gentamycin and trimethoprim/sulfamethoxazole. However, MRCNS strains were rather (mean 57%, median 61%) susceptible to sulphonamide and exhibited high resistance (up to 90% in 2019) to gentamycin. MRSA was more susceptible to the other than mentioned above antimicrobials than MRCNS. All MSSA strains were susceptible to isoxazolyl penicillins.

Fig. 1

Fig. 2

Fig. 3

Amongst the enterococci, all E. faecalis isolates (21% of isolates with the HLAR mechanism of resistance) were susceptible to ampicillin, E. faecium to vancomycin and teicoplanin, and VRE to linezolid and tigecycline.

The most problematic Gram-negative alert pathogen was A. baumannii due to its high prevalence and resistance to antimicrobial agents. In such cases, colistin was virtually the only therapeutic choice. In the last two years of the period analyzed, all P. aeruginosa isolates were susceptible to ceftazidime, cefepime, and gentamycin.

The most frequently isolated species of Enterobacterales were K. pneumoniae (165 isolates) and E. coli (69 isolates). More than 40% (67/165) of K. pneumoniae strains and more than 16% of E. coli (11/69) exhibited an ESBL mechanism of resistance. Fortunately, all these pathogens were susceptible to meropenem, and only six of these isolates produced carbapenemases.

We had a broad spectrum of therapeutic options for fungal infection treatment. Candida albicans isolates were completely susceptible to fluconazole and non-albicans Candida species were susceptible to echinocandins and amphotericin.

Bloodstream infections may lead to metastatic infections, severe sepsis, and multiorgan failure. The prevalence of bloodstream infections is estimated at 174–204/100,000 in North America and 166–189/100,000 in Europe, corresponding to 73,349–84,823 cases in the USA and 157,750–276,318 in Europe (Goto and Al-Hasan 2013). CVCs may be related to BSIs in 90% of cases (Polderman and Girbes 2002).

Possible reasons for this strong correlation include lack of asepsis upon catheter insertion and use, skin changes (e.g., burn-related), the biofilm formation on the catheter’s inner and outer surfaces, and the development of central vein thrombosis at the site of the catheter insertion. The only way to decrease the number of CLABSI cases is to ensure rigorous asepsis when the catheter is placed and used (Ling et al. 2016; Bell and O’Grady 2017; Ielapi et al. 2020).

Our work has been one of the most extensive single-center analyses performed in Poland in recent years. The study was conducted over a long period in a multibed tertiary hospital, a regional trauma center that gives credibility to the results and allows their generalization. The authors of this study believe that analyzing the microbiological situation in single centers may help make up a complete picture of both microbiological hazards and therapeutic possibilities on the national and regional levels. Our analysis was performed at a multidisciplinary ICU, with most of the cases being surgical and trauma patients. Some of them may be immunocompromised; in such patients, the transfer of pathogens from the site of infection to the lumen of the catheter via different connectors may be reasonably easy, resulting in more facile development and progression of a disease.

The data suggest that one in 20 ICU patients develop bloodstream infections. The longer the period of central vein catheterization, the higher the incidence of catheter infection (Gunst et al. 2011). We found that the incidence of positive microbiological results in blood samples was 12.8%. Those values are relatively low compared to the results of other studies which report positive results in 18.9% to 31.3% of cases (Rani et al. 2012; Musicha et al. 2017; Rani et al. 2017). This discrepancy may be due to the procedures at our center, where patient blood samples are analyzed microbiologically every time the patient’s body temperature exceeds 38°C. Some studies from ICUs in southern Poland report a lower incidence of BSI (Wałaszek et al. 2018). However, the spectrum of pathogens isolated from blood samples was similar to our results, including A. baumannii, coagulase-negative staphylococci, E. coli, P. aeruginosa, K. pneumoniae, and C. albicans (Kołpa et al. 2018; Wałaszek et al. 2018).

The etiology of bloodstream infection depends on the geographical localization of the hospital. For example, in Malawi (Africa), the most common BSI pathogens are Salmonella Typhi and Streptococcus pneumoniae (Musicha et al. 2017). The Surveillance Network data from Korea collected from 2006 to 2017 showed that the most common pathogens were E. coli and S. aureus. In Japan, E. coli, S. aureus, and K. pneumoniae are the most common isolates (Rani et al. 2017; Lee et al. 2018).

Data from the European Antimicrobial Resistance Surveillance Network study performed in 22 countries between 2000 and 2009 indicated that E. coli and S. aureus were the most common BSI pathogens (Gagliotti et al. 2011). The results of a EUROBACT study spanning 162 ICUs in 24 European countries and Turkey, Brazil, and China showed that in 53%, 32%, 7.8%, and 1.2% of cases, bacteremia was caused by Gram-negative bacteria, Gram-positive pathogens, fungi, and anaerobic bacteria, respectively. Multidrug-resistant pathogens were isolated in 47.8% of cases (Tabah et al. 2012). Our results were similar, although there were some differences in the exact numbers of isolated pathogens (49.5% Gram-negative strains, 47.1% Gram-positive strains, and 3.5% fungi).

Gram-positive pathogens were responsible for half of all bloodstream infections. Most of them were staphylococci. Their resistance to methicillin ranges from 40 to 60%. of importance is that MRCNS are a more likely etiological factor in infection than MRSA (Gunst et al. 2011; Louzi et al. 2016; Tian et al. 2019; Tomaszewski et al. 2019). Our results supported this observation. Moreover, MRCNS exhibit greater resistance to antimicrobials, except against vancomycin. We also noted the difference in resistance to gentamycin and trimethoprim/sulfamethoxazole. Such differences may result from the different proportions of branched-chain and straight-chain fatty acids in the membrane lipids of bacteria (Tiwari et al. 2020). However, susceptibility to sulphonamide is significant clinically because the drug can be administered orally, which may be necessary in patients who suffer from staphylococcal bone infections and require proper therapy for an extended time.

In many studies, Gram-negative pathogens responsible for more than half of bloodstream infections pose a greater therapeutic challenge than Gram-positive bacteria, partially due to their increasing antibiotic resistance. Enterobacterales produce ESBL and CPE mechanisms of resistance. Both Acinetobacter spp. and Pseudomonas spp. exhibit mechanisms of resistance that are multidirectional and difficult to treat.

In our study, Gram-negative bacteria were isolated in 49.5% of cases. The most frequently isolated was A. baumannii (40.5%), followed by K. pneumoniae (21.4%), P. aeruginosa (14.1%), and E. coli (9%). These results differ significantly from those found in an analysis of 9,334 strains of A. baumannii performed between 1994 and 2011 (Munoz-Price et al. 2013), during which time this pathogen was responsible for 11% of blood infections. This situation is alarming because the only effective antimicrobial against this pathogen is colistin.

Among the Gram-negative bacteria, 30.4% were K. pneumoniae or E. coli. The ESBL mechanism of resistance was found in 40.6% of K. pneumoniae and 16% of E. coli. These results are similar to those found by other researchers. According to the literature, the ESBL mechanism of resistance is present in 34–75% of K. pneumoniae and is increasingly common (Gunst et al. 2011; Louzi et al. 2016; Tian et al. 2019). Fortunately, we found this pathogen to be entirely susceptible to meropenem. Moreover, only a small number of the isolated K. pneumoniae exhibited a KPC resistance mechanism.

There are a few significant limitations to our study. The first group of constraints is related to other observational studies with routinely collected electronic data. They include missing data, potential bias, and possible misclassification or inconsistencies in medical coding. The second one is related to the indication for blood sample collection. In our center, blood samples were collected when the infection was suspected clinically. It cannot be ruled out that some patients’ pathogens were present in their bloodstream with no clinical and bacteriological signs and symptoms of infection. The third one is related to the setting of this study: our analysis is connected to only one hospital, although a large one. Finally, assessing the results with some differences in patient-level factors, such as the severity of illness, may have influenced the conclusions.

Despite the limitations mentioned earlier, we believe that our results may be informative and exciting for those interesting in the epidemiology of BSI. Our study estimated bloodstream infections in a large tertiary university hospital and trauma center for 13 years, as incidence rates and pathogens’ distribution, and their susceptibility/ resistance to antimicrobial agents. The presented data may constitute material for further analyses assessing the microbiological situation in Poland and the region to improve the quality of care and outcome in critically ill patients.

Our 13-year study shows that among 12,619 micro-biologically analyzed blood samples, 12% (1,509/12,619) were positive. In 278/1,509 (18.4%) of cases, a central line catheter infection was confirmed, and 1,557 pathogens were isolated.

The most frequently (770/1,557; 49.5%) isolated species were Gram-negative bacteria. Among the Gram-negative bacteria, the most frequently isolated were A. baumannii (312/770; 40.5%), K. pneumoniae (165/770 (24.1%); 67/165 (40.6%) of them produced ESBL, 5/165 (3%) – produced NDM, 4/165 (3.4%) – expressed KPC, 1/165 (0.6%) – produced OXA48), P. aeruginosa (111/770 (14.4%); 2/111 (1.8%) strains produced MBL), and E. coli (69/770 (9%); 11/69 (15.9%) strains expressed ESBL).

Gram-positive pathogens were isolated in 733/1,557 (47.1%) cases. The majority were staphylococci (545/733; 74.4%), mainly coagulase-negative (368/545; 67.5%). MRSA species represented 58/545 (10.6%) of all staphylococci. Fungi were isolated from 3.5% of samples.

Using these data to improve clinical practice, one can say that the MRSA and MRCNS species are susceptible to vancomycin, MSSA is susceptible to isoxazolyl penicillins, and VRE is susceptible to linezolid and tigecycline. However, colistin remains the only therapeutic option in some infections caused by A. baumannii and KPC-producing K. pneumoniae. P. aeruginosa is still susceptible to cefepime and ceftazidime. Echinocandins are effective therapeutics in treating fungal infections caused by C. albicans and non-albicans Candida species.

The bacteriological situation in our department, assessed by the incidence of infections, is similar to other ICUs in Poland and Eastern Europe. However, the number of A. baumannii and P. aeruginosa among the isolated pathogens is disturbing. We implemented and followed the infection control procedures. They include the decontamination of the patient environment, prevention of secondary contamination, personnel training, compliance control, hand hygiene, and clothing policy. We also follow the central catheter procedures. They emphasize full aseptic, and the additional connections are reduced to a minimum. A separate catheter lumen is reserved for parenteral nutrition, and the administration of drugs takes place after the disinfection of the connectors.

The applied infection reduction procedures follow the existing recommendations (Hryniewicz et al. 2013). In addition to the guidelines, it is essential to be aware of the infections, severity of the problem, prevention methods, and the real benefits of compliance with the recommendations that translate into a reduction in the frequency of infections in intensive care units.