Resistance to Co-Amoxiclav (Augmentin) in Community-Acquired and Nosocomial Pleural Infections

Empyema thoracis are rarely seen in the intensive care unit (ICU). They are reported to correspond to an unfavorable evolution of lung infection [1]. Bacterial ecology was largely described in ancient series [2-4]. The flora found in the condition of community-acquired was gladly polymorphic and often included Gram-positive cocci and anaerobes. This feature was acknowledged in more recent studies [5,6]. In empyema thoracis, systemic antibiotic therapy is recommended. However, the choice of the antibiotics is still matter of debate, including recommendation of the association of penicillin with a betalactamase inhibitor for community-acquired pleural infections [7,8]. However, recently, the resistant bacteria encountered in nosocomial thoracic empyema was emphasized, in up to 40% of the patients [6,9]. Consequently, owing to the expansion of the use of antibiotics and the frequent “multi-resistant” bacteria, increasing surveillance of susceptibility of the bacteria should be considered. Indeed, significant increase in mortality rate related to empyema was noticed in a study over 100 years. The authors speculated that changes in microorganisms, with perhaps more antibiotic resistance, might cause this increased mortality rate [10]. Accordingly, increased numbers of pleural infections due to Staphylococcus aureus [11,12], Streptococcus milleri [1], and Pseudomonas aeruginosa [11,13] was acknowledged.


Introduction
Empyema thoracis are rarely seen in the intensive care unit (ICU). They are reported to correspond to an unfavorable evolution of lung infection [1]. Bacterial ecology was largely described in ancient series [2][3][4]. The flora found in the condition of community-acquired was gladly polymorphic and often included Gram-positive cocci and anaerobes. This feature was acknowledged in more recent studies [5,6]. In empyema thoracis, systemic antibiotic therapy is recommended. However, the choice of the antibiotics is still matter of debate, including recommendation of the association of penicillin with a betalactamase inhibitor for community-acquired pleural infections [7,8]. However, recently, the resistant bacteria encountered in nosocomial thoracic empyema was emphasized, in up to 40% of the patients [6,9]. Consequently, owing to the expansion of the use of antibiotics and the frequent "multi-resistant" bacteria, increasing surveillance of susceptibility of the bacteria should be considered. Indeed, significant increase in mortality rate related to empyema was noticed in a study over 100 years. The authors speculated that changes in microorganisms, with perhaps more antibiotic resistance, might cause this increased mortality rate [10]. Accordingly, increased numbers of pleural infections due to Staphylococcus aureus [11,12], Streptococcus milleri [1], and Pseudomonas aeruginosa [11,13] was acknowledged.
Eight subjects had infection with at least a germ resistant to coamoxiclav, corresponding to 21% of the subjects with PF positivity (8/39) and 14% of our cohort (8/57). Microorganisms resistant to coamoxiclav were the following: Mycobacterium tuberculosis (n=2), E. cloacae (n=1), E. coli (n=1), Enterococcus faecium (n=1), S. epidermidis (n=1), and Candida albicans (n=2). Fungal infections were observed in one AIDS subject with nosocomial infection and in a subject with community-acquired polymicrobial infection (four microorganisms). Excluding fungi and Mycobacteria, "multi-resistant" bacteria were found (n=4 in 3 subjects) in the community-acquired group and during year 2011 and 2012 (i.e. two last years of survey): E. cloacae with derepressed cephalosporinase (n=1), E. coli (n=1) with extended-failure, chronic renal failure, cirrhosis, chronic respiratory failure, long-term corticosteroids treatment, transplantation, HIV/AIDS, malignant hematology disease, chemotherapy) or another past history (including chronic alcoholism, smoking, malnutrition, antecedent of thoracic surgical or trauma), previous antibiotic administration in the 14 days prior to admission, hospitalization, and nosocomial or community-acquired infection. Clinical data (respiratory functional signs, temperature, blood pressure, heart rate, respiratory rate, oxygen saturation), biological data (number of white blood cells, hemoglobin, albumin, red blood cell counts, distribution width, C-reactive protein, alkaline phosphatase, lactate dehydrogenase), and radiological findings (side of achievement, volume) were reported. PF: pH, lactate dehydrogenase, glucose, white cell and neutrophil counts, bacteriological findings (direct examination; cultures, and in vitro antimicrobial sensitivity with in particular, the 'first line' antibiotic resistance). Support: antibiotic therapy, clinical efficacy (apyrexia timeout defined as obtaining a temperature <37.8°C during 48 h), duration of stay in ICU, in-hospital survival, complications (loculation, sepsis, infection from another site...), pleural fibrinolysis, surgery.

Statistical analysis
Data are expressed as median (IQR: 25; 75 percentiles) for continuous variables and number (percent) for binary variables. Qualitative values were compared with Chi-2 tests and quantitative values with Mann-Whitney tests as appropriate. Odds ratios (OR) for hospital mortality were calculated with their 95% confidence intervals (CI). A p value <0.05 was considered significant. Statistical analysis was performed with Statview 5.0.
Prior to admission, 23 subjects (40%) received antibiotics for a median duration of 7 days (5; 11). Most of the subjects had communityacquired infections (91%). Pleural infection was classified nosocomial in 5 subjects (9%). PF was bilateral in two subjects. The side of pleural infection was not recorded in one subject. PF concerned the right and left side in 35 and 23 subjects (62% and 41%) respectively. The outpouring was considered of small (n=7), medium (n=34) or large (n=15) abundance, in 56 subjects for which chest imagery was available. PF was collected in all the subjects. The characteristics of the PF are summarized in Table 2.

Antibiotics and efficacy
All subjects received antibiotics from admission. Subjects were treated with co-amoxiclav (n=21), and/or another antibiotics (n=50). After admission, initiation of in vitro effective antibiotic therapy was <24 hours in 48 subjects (84%). Apyrexia was obtained after 1 day (0; 3). Noteworthy, despite administration of antibiotics and frequent chest tube drainage, unfavorable development of empyema (encystment, loculation, and lung abscess) was noted during the course in 45 subjects (79%).

Microbiology, antibiotics and prognosis
Refreshing or updating microbiological data seems essential to guide the probabilistic antibiotic therapy in pleural infection. Our study is consistent with the predominance of Gram-positive cocci in community-acquired pleural infections [4,6], even if 60% of our subjects had an underlying disease, increasing the risk of GNB and fungal infection [14]. The rate of PF bacterial "negativity" (32%) found in the present study is also in agreement with previous studies that consistently reported rates of negative PF of about 40 to 50% [1,12].
Our study found resistance to a first line antibiotic -co-amoxiclavin 14% of so-called community-acquired subjects and in 21% of the PF with bacteriological growth. The differences between nosocomial and community-acquired thoracic empyema included a predominance of infection with S. aureus and group 2 enterobacteria in the nosocomial empyema, also found by others [5,15]. We cannot recommend more judicious antimicrobial therapy in the light of our ecology. However, a probabilistic antibiotic with co-amoxiclav is questionable, even in community-acquired empyema. Accordingly, adequate coverage based on in vitro susceptibility of micro-organisms found in communityacquired and nosocomial pleural infections, was reported to be only 52% (49% to 53%) and 84% (78% to 86%) with the association of metronidazole and either penicillin (Scandinavian recommended antimicrobial therapy) or cefuroxime, respectively [5].
Tuberculosis infection did not appear to be a concern regarding the immediate risk of sepsis (mortality 0/3, 0% in a medical ICU) [9]. On the other hand, the fungal infection was a real threat, even   Table 3: Bacteriological data of pleural fluid according to the community-acquired (n=52 whose 34 germs were found) and nosocomial (n=5 whose 4 germs were found) origins of the pleural infection.
spectrum beta-lactamase, Enterococcus faecium resistant to ampicillin and a S. epidermidis resistant to Methicillin. As a whole 3 out of 57 (5%) subjects had microorganisms (not including candida and mycobacteria) resistant to co-amoxiclav.
Nosocomial thoracic empyema had more S. aureus, Group 2 among the immune-competent subjects (mortality 2/2 100%, in the same series [9]. One of our subjects with AIDS exhibited C. albicans empyema after staying for a long period in hospital and died rapidly. However, the appropriate antifungal treatment was delayed by 2 days. The role of resistant Gram-positive cocci (ampicillin-resistant E. faecium and Methicillin-resistant S. epidermidis in our cohort) and GNB (Enterobacteriaceae with derepressed cephalosporinase and extended-spectrum beta-lactamase) to front-line antibiotics should be highlighted in the community-acquired empyema in our study.
Before severe sepsis and septic shock related to purulent empyema, especially if the patient has received extended antibiotic therapy, broad-spectrum antibiotic therapy should be recommended, pending the results of the direct examination and culture of PF (and blood), as acknowledged by Tu and colleagues. In immunocompromised patients, the use of the antifungal therapy (fluconazole), especially for nosocomial empyema should be discussed [1,9,16]. Finally, an emergence of drug resistant bacteria is a possible mechanism, as those micro-organisms were found only in the last two years of our study [10].
However, despite delay in administration of adequate antimicrobial therapy, mortality remained lower in our subjects than that observed in severe complicated parapneumonic effusions (9% vs. 48%) and this discrepancy is questionable [9]. First, population may differ between the centers with different inclusion criteria. Our subjects rarely received catecholamines (11%) or mechanical ventilation (16%). Subjects were admitted in our ICU if chest tube drainage was needed. Though, our subjects did not represent a real ICU population with subsequent elevated mortality, but is in agreement with previous reports accounting for by mortality rates of about 4.7% to 27%, in patients with complicated pleural effusions [5,15,17]. Maskell found a 12-month mortality rate of about 22%, with a significant increase in patients with nosocomial infections (45%; OR 4.24), the mortality rates increased when S. aureus, GNB, or mixed aerobic bacteria where found, compared to anaerobes, Streptococcal, or culture-negative infections [6].
Although Tu et al did not report the adequation of "standard" antimicrobial therapy, they reported many "resistant" GNB (fermenting or group 3) along with fungi and a 36% rate of inadequate initial antimicrobial therapy in non-tuberculosis bacteriologically proven pleural infection [9]. The inadequacy between initial antibiotics and in vitro findings was associated with a 60% increase in the mortality rate. Though, their susceptibility to antibiotics antimicrobial therapy might have been even lower than ours. Finally, in contrast with lung infection [18], delay in adequate antimicrobial therapy in PF may not be the point [17], as chest tube drainage is an effective treatment to quickly decrease bacterial load and improve the patients' condition [19][20][21]. However, some authors recommend not draining every complicated pleural effusion [22]. Accordingly, although ICU patients aforementioned had small bore pleural pigtail catheter, their need for chest tube drainage, intra-pleural fibrinolysis and thoracic surgery was particularly low (9%, 4% and 4% respectively), rendering the characteristics of the pleural infection and its management different from ours. On the other hand, a study found a significant association between adequate empirical therapy and a decrease in hospital mortality (OR 0.43; p=0.02) in the multivariate analysis, consistent with the results of our study [5].
Intra-pleural fibrinolysis with streptokinase resulted in disappointing results [23]. However, the same team showed that adjunction of deoxyribonuclease with recombinant tissue plasminogen activator improved the drainage of infected PF and subsequently decreased the need for thoracic surgery and hospital stay with no change in the mortality rate [24]. Fibrinolysis was realized without deoxyribonuclease (not available in our hospital) and did not reduce the need for surgery, in our series. However, it was associated with a lower mortality rate, such as in sub-groups of proven loculated/septated effusions in a systematic review [25]. This review showed a significant reduction in surgical intervention. Nevertheless, our retrospective study did not allow concluding on the benefit of fibrinolysis, in terms of mortality or surgical requirements.
The other elements associated with hospital mortality were common. We did not find significant association between red blood cells distribution width or alkaline phosphatase and mortality [15]. However, those variables do not seem valuable tools for prognosis assessment in thoracic empyema. According to pulmonary vasculature and pleural drainage, we found a right predominance as expected [15].

Limitations
Our study is an overview of the ecology of a small series of subjects admitted for empyema thoracis and complicated pleural effusion in a single ICU. It is only a local vision and cannot be extended to the whole spectrum of purulent empyema. Diagnoses on admission were variable and population heterogeneous. Our hospitals having no facility for thoracic surgery, the rare nosocomial thoracic empyema were probably the consequences of complicated pulmonary infections and were not the image of post-operative direct inoculations. However, given the low rate of S. aureus resistant to Methicillin and Streptococcus pneumoniae with decreased susceptibility in France, our results might encourage other countries screening for bacterial resistance in communityacquired pleural infections.
PF had different characteristics than previous studies [6,17,20]. PF pH and glucose were higher, though, false positive of PF culture might account for misdiagnosis. However, previous studies reported that some patients with PF pH >7.2 do need tube thoracostomy or decortication [26]. Moreover, every subject had clinical signs of overt pleural infection in our series. Clinicians should take into account PF culture despite elevated PF pH and glucose, if clinically indicated [17].

Conclusion
Our study found a resistance to co-amoxiclav in 14% of the subjects with community-acquired thoracic pleural infection and in 21% of found microorganisms. There was no evidence of notable difference in spectrum's between nosocomial and community-acquired pleural infection.