According to our knowledge, few studies have focused on post-operative infections caused by ESBLs in paediatric cardiac surgery, especially with regards to previous colonization with ESBLs. No data is available for ESBL colonization and/or infection in children with congenital heart disease in Morocco.
In our country, some studies showed that the occurrence of ESBLs was high on faecal carriage (42.8%) , 28.6% in blood culture, superficial and deep pus  and catheters and 20.7% in blood cultures, urine, pus and bronchial sampling .
Multidrug-resistant bacteria (MDRB) screening is recommended as a standard protocol for all cardiac surgery patients . Kim et al.  suggest that in a non-epidemic situation, systematic detection of ESBLs in ICU patients is not cost-effective, and that stringent contact precaution for infected patients might be adequate. However, our institution applies selection criteria based on previous hospitalization or prior use of antibiotics to perform MDRB screening. Screening of all MDRBs, including MRSA and carbapenemases, was performed for all patients with a risk criterion. In this study, we only reported ESBL-E screening because no MRSA strain was isolated and only one strain of Klebsiella pneumoniae was both carbapenemases- and ESBL-producing. Therefore, it appears more relevant to focus on ESBLs.
We found 15% ESBL colonization, similar to a Polish study on the same population (16% MDRB)  and to a reported colonization rate in a 2012 report from France (15%) . However, it was lower than those reported by Kim et al. in a Korean study of intensive care unit patients (28.2%) ; in another recent Korean study, the authors reported a rate of 42.5% in ICU patients and 20.3% even in healthy persons .
The majority of ESBL strains in this study were isolated from anal swab. This finding is similar to those reported in colonization screening by other authors . It could be interesting to discuss whether screening for MDRB in our setting should be done only with anal swab (just one site of sampling), with the objective of enhancing the cost-effectiveness of MDRB screening. The most common isolated species were Klebsiella pneumoniae in our study, while others reported higher frequencies of Escherichia coli isolation .
Otherwise, nosocomial infections in children after heart surgery occur in 12.9% to 30.8% of cases [24, 25]. Bloodstream infections (BSI) are one of the most common infections [26–28] seen in children undergoing cardiac surgery, with studies reporting incidence as high as 65% of total nosocomial infections .
The most common organisms detected in infectious endocarditis after surgery for congenital heart defect were Streptococcus viridans and Staphylococcus aureus (each 23% of total), as reported by Cynthia et al. . However, other studies have reported that Gram-negative bacilli (GNB) were the most frequent isolates in nosocomial pneumonia (NP) in infants in a paediatric surgical ICU after cardiac surgery (86.1%) . The main GNB were Acinetobacter baumannii isolates (13.9%), Pseudomonas aeruginosa (10 isolates, 12.7%), Klebsiella pneumoniae (7 isolates, 8.9%), and E. coli (6 isolates, 7.6%). The bacteria-producing extended spectrum beta-lactamases were mainly K. pneumoniae and Escherichia coli .
In our study, infection with ESBLs was found in five patients, four of them likely infected with a strain of the same species isolated upon colonization screening at admission, with the same antibiotic susceptibility features. We could not confirm the clonality of colonization and infection isolates with molecular study due to lack of resources. Only one case was infected with E. coli after being colonized by K. pneumoniae. This case was probably a nosocomial infection contracted during hospitalization in the ICU. We could not confirm this due to a lack of techniques and reagents. In addition, the conventional technique used for the detection of ESBL-E is not very sensitive; possible colonization by E. coli could not be identified with this technique. On the other hand, since this case was a newborn, a possible carriage in the mother could be at the origin of this contamination with E. coli.
Carriers of ESBLs appear to be at higher risk of subsequent bacteraemia compared with non-carriers . Due to the spread of extended-spectrum β-lactamase-producing E. coli (ESBL-EC) carriage, a practitioner may be tempted to modify empirical antibiotic therapy in known carriers with signs of infection.
Our results, even with a lower rate (23%), demonstrate that previous colonization with ESBLs may be a risk factor for the occurrence of infections after cardiac surgery in paediatric patients. In fact, the difference in the incidence of those infections between the two groups (colonized and non-colonized) was significant.
So, should we prescribe antibiotic prophylaxis systematically on previously colonized children undergoing cardiac surgery?
Actually, antibiotic prophylaxis for cardiac surgery is a controversial area. Recent systematic reviews and meta-analyses of randomized controlled trials have concluded that surgical site infection can be reduced by prolonging prophylaxis for 24–48 h. In addition, post-operative pneumonia and every cause of mortality can be reduced by giving agents with both anti-Gram-negative and anti-Gram-positive activity . The choice of the most appropriate regimen remains open to debate, and yet no recommendations exist for paediatric cardiac surgery patients.
Our goal was to gain an insight into occurrence rate of colonization and infection with ESBLs in this specific population. Our findings were very concerning, and may influence our screening and antibiotic prophylaxis policies.
It is clear that infection control for ESBL-producing strains would be achieved through active surveillance of high-risk patients, such as those in the ICU . Some authors claim that strict contact precautions might be more appropriate and cost-effective than systematic MDRB screening. Nevertheless, ESBL colonization rate in paediatric cardiac surgery patients is still relatively low, allowing the prevention of infection through systematic screening. This screening will provide epidemiological data to guide antibiotic therapy policies. It must be applied for all MDRBs, even if ESBLs are the most common.
In fact, besides ESBLs, carbapenemase-producing Enterobacteriaceae seem to be of great concern in our geographic area. Their recent emergence and the potential for horizontal transfer of resistance genes make critical the control of their spread while the frequencies are still low . This will require rapid and accurate testing of carbapenem-resistant isolates from clinical specimens, as well as the adoption of appropriate infection control procedures if carbapenemase-producing isolates are found. In addition, the incidence of CRE should be monitored continuously. If CRE incidence increases, more aggressive management, such as active surveillance of high-risk patients, will be necessary .
Our study has many limits. Due to the retrospective design, we could not have all clinical data that would be of interest with regard to colonization and infection with ESBLs, such as hospitalization duration before surgery, mortality, detailed antibiotic regimen before and after surgery, and other infections apart from ESBL infections.