Staphylococcus aureus in a neonatal care center: methicillin-susceptible strains should be a main concern

The NCC of Montpellier is organized in 3 sectors: the paediatric reanimation and intensive care unit or PRICU (14 beds in 9 boxes for neonates including very preterm neonates and 6 beds in 5 rooms for infants), the NICU (24 beds in 10 boxes), and a mother-cum-child or “kangaroo” ward (9 beds in individual rooms, and 3 beds in a nursery).

Hospitalized patients in the PRICU are low birth-weight (<1500 g) preterm infants and newborns aged < 1 month with diseases or unstable states. The PRICU also hosts newborns having surgery and older children requiring intensive care. Newborns are transferred to the NICU once their clinical state is stabilised or improved. As soon as their condition allows, patients are transferred to the kangaroo ward before returning home.

The medical and paramedical teams from the PRICU, the NICU and the kangaroo ward, a microbiologist and a member of the infection control (IC) team meet weekly to discuss HAI cases, differentiating true HAIs from bacterial carriage or colonization by confronting microbiological data, biological and clinical contexts and initiation of antimicrobial treatment [18]. Infants are considered as colonized if a positive culture is obtained from a non-sterile site, and infected if a pathogen is isolated from a normally sterile site or if cultures are obtained for clinical purposes. In addition, in the NICU sector, the digestive carriage of SA and multi-drug resistant bacteria (MDRB) is screened for each patient upon admission and once a week thereafter. A positive culture from a digestive sample is considered as SA carriage [17]. Stool samples are cultured to determine methicillin and ceftazidime resistance in staphylococci and gram-negative bacilli, respectively. MDRB detection leads to supplementary infection control measures to prevent cross-contamination among patients, and to further environmental investigations. A real-time surveillance is implemented by a daily account of all newly MDRB colonized or infected patients by the microbiologist, using an antibiotic resistance information system (Sirweb®, i2a, Montpellier, France). Surveillance data is transmitted to the IC team via the hospital information system. This automatic surveillance system operates for MDRB only; hence outbreaks involving susceptible micro-organisms can be detected only through clinical observation of an increased number of cases over a given time span.

This clinical surveillance thus detected an increase in HAIs involving MSSA, leading to an outbreak investigation by the IC team including analysis of patients’ medical records, environmental investigation, and assessment of healthcare practices. Clinical strains were tested for their antimicrobial susceptibility by disk diffusion assay according to the French Committee for Antimicrobial Susceptibility Testing (Members of the Société Française de Microbiologie Committee, 2003). After extraction of the DNA performed according to Predari et al. [19], isolates were screened for genes encoding staphylococcal enterotoxins A (sea), toxic shock syndrome toxin 1 (tst), Panton-Valentine leukocidin (PVL; luk- PV) [2‐4]. Clinical MSSA strains were explored by genomotyping to determine genetic links between them. For this purpose, intact genomic DNA was extracted in agarose plugs and digested by the endonuclease Sma I, as described [20]. Macrorestriction fragments were separated by pulsed-field gel electophoresis (PFGE) by a ramp of pulses of 20s to 5 s at 6 V/min during 24 h on CHEF-DRII apparatus (Biorad).

Dry and humid surfaces were sampled, as well as health-care workers’ clothes (i.e.: white over-gowns used for the manipulation of central venous lines, especially the front part which comes into contact with patients); the cotton swabs used for sampling served to inoculate trypticase-soy, Chapman and McConkey agar plates (bioMérieux, France) which were incubated at 37°C for 48 h. Water was sampled at different points of use in the units in 250 mL sterile bottles containing sodium thiosulfate for chloride inhibition. Water samples were passed through a 0.22 μm nylon filters, which were then incubated at 30°C during 48 h on Chapman agar. Healthcare practices in the NCC were audited by direct observation of hand hygiene opportunities (a minimum of 30 observations) and central venous catheter manipulations (a minimum of 5 observations).

We studied bacterial isolates obtained during the daily care of preterm infants in our NICU. Therefore, this observational study was fully in line with the routine care of preterm infants and did not require the agreement of the ethical committee of our institution.

One hundred and thirty nine patients admitted to the NCC were included in the epidemiological study. Distribution of carriage, colonization and infection cases for each sector is summarized in Figure 1. Fifty-four patients were carriers, colonized or infected by SA, 22 in the summer period (4, 6, 4 and 8 respectively in June, July, August and September). In the NICU and kangaroo ward, most SA isolates were involved in carriage with a low frequency of colonization and/or infection. By contrast, patients in the PRICU with more instable clinical status were more frequently colonized and/or infected.

Distribution of the different types of SA positive clinical samples corresponding to colonization and/or infection by care sector is summarized in Table 1. Most of them occurred in the PRICU (70.9% of all positive samples), followed by the NICU (23.3%) and the kangaroo ward (5.8%). Respiratory samples were predominant in the NCC, overall representing 52.3% of all clinical samples, especially in the PRICU (63.9%). Ophthalmic samples were the second most frequent with about 17.5%, followed by blood cultures (10%) and catheter samples (7%). By care sector, ophthalmic colonization and infections predominated in the NICU, and came equal with respiratory samples (40% each) in the kangaroo ward. SA positive catheter samples and blood cultures were mainly present in the PRICU (8 of the 10 samples), 2 other cases were found in the NICU, and none in the kangaroo ward where patients are less often perfused.The evolution of SA monthly prevalence for each sector is represented on Figure 2. In the NICU, where digestive carriage of SA is routinely screened, average prevalence reached 12.9%, with increasing rates in March (20%), August (17%) and November (22%). In other sectors, because of the absence of systematic screening, prevalence rates obtained were certainly underestimated and did not reflect the real presence of SA. In the PRICU, average rates were around 5.5% and ranged generally between 5 and 10%, except in November and December when they fell to 3.1% and 1.1% respectively. In the kangaroo ward, rates were more variable, roughly around 6.2% with 2 peaks in June (16.2%) and November (14.3%). Monthly incidences of SA clinical samples and infections by care sector were also calculated. Evolution of incidences over the year was very similar to prevalence curves, especially in the NICU with increased incidence in March (26.1%), July (20%) and August (17.4%), and November (28.9%) (data not shown). Finally, epidemiological analyses showed several epidemic episodes on a background of SA endemicity: each care sector presented 2 or 3 prevalence peaks.Monthly MSSA and MRSA prevalence rates in the entire NCC are summarized in Figure 3. The overall MSSA prevalence did not decrease below 5.5% (in May and October), and reached 13.2% in March. Concerning MRSA, the mean prevalence in 2009 for all of the NCC was around 1.53%, with increasing rates in September (3.7%) and November (6%). These results show that MSSA endemicity largely exceeded MRSA endemicity in the NCC. Incidence rates confirmed that new cases of MSSA were also more frequent than MRSA in the NCC (data not shown).

Monthly analysis of infections showed an emergence of SA bacteraemia and catheter-linked infections in summer. For one of them, positive blood-culture was associated with a positive catheter. The origin of the other bacteraemia was not found.

The epidemic curve of SA infection cases between January and December 2009 is shown in Figure 4. Among the 17 SA infections, 15 were caused by MSSA strains versus 2 by MRSA. Outbreak alert was sounded because of an increase in MSSA case numbers observed in July (4 cases) and August (2 cases), contrasting with the average rate of one case every 2 months until then. Six other SA infections occurred in September and October with respectively 3 cases including one MRSA infection each month. Another MSSA infection in November and one in December were also reported.

The 6 cases of July and August led to the outbreak investigation. All 6 patients were hospitalized in the PRICU and 5 of them were preterm neonates. Principal characteristics of these patients and associated strains are given in Table 2. Antimicrobial susceptibility was tested for the 12 MSSA strains from the 6 patients (Table 2). The 12 strains were mostly penicillinase producers (n = 11) and susceptible to most antibiotic classes except one strain resistant to macrolides. Most strains lacked virulence factors since only one carried tst, the gene encoding for toxic shock syndrome toxin 1 and one other carried sea coding for enterotoxin A. PFGE-typing for the 12 strains revealed a predominant profile named pulsotype C and shared by 8 strains isolated from 4 patients (Table 2). Interestingly, some isolates with the same pulsotype displayed different content in virulence genes.

The synopsis of the IC team’s interventions and investigations is presented in Table 3. During the environmental investigation, a high level of surface contamination (109 surfaces sampled) in all areas (average rate of 82.6%) of the NCC was observed with the presence of pathogens such as Enterococcus faecalis, Pseudomonas aeruginosa and MSSA. MSSA were isolated from a rack of electric syringes, medical records of patient, 2 spare beds stored in the hallway, and an incubator. All of these environmental MSSA strains were typed by PFGE and differed from the profile C. Water samples (n = 19) at various points of use were also analyzed, but none was positive for SA.

The clinical audit of hand hygiene practices revealed compliance rates of 60% in the NICU (14 persons assessed) and 76% in the PRICU (16 persons assessed). Non-compliance with recommendations was mostly due to the concatenation of multiple care sequences for the same patient, for example the absence of hand hygiene between a contaminated care (nap change) and a clean one (catheter manipulation). The assessment of central venous catheter manipulations in the PRICU revealed failures in catheter monitoring: catheter insertion points were covered with opaque dressings preventing visual control, and the frequency of dressing changes was insufficient. Furthermore, health-care workers always wore white over-gowns whenever tending to a catheter or coming into direct contact with perfused patients. These gowns were stored in a drawer of the incubator until re-use. During the observations, these white gowns were pointed out as a potential bacterial reservoir or source of cross-contamination. Thereby, 35 of them were sampled in December: 25 gowns in the PRICU and 10 in the NICU. Many of them were positive (15/35) for several pathogens including SA, which was found on 8 separate gowns. Overall these gowns showed a high level of contamination (on average 50 CFU/25 cm²). SA strains were not typed due to the delay since the summer outbreak but the level of contamination suggested that misuse of these gowns could promote the transmission of pathogens.