Catheter-related bloodstream infections (CRBSI) are associated with increased rates of morbidity in intensive care unit patients and in neonates [1]. The prevention of the avoidable part of CRBSIs is a public health priority [2, 3]. In this context, since 2019, all French hospitals and clinics are encouraged to participate in an annual 3-month survey of CRBSI coordinated by the national infection control SPIADI network. Over the last two decades, multidrug-resistant Staphylococcus capitis has been increasingly reported as a major agent responsible for CRBSI in preterm babies [4]. Therapeutic failures likely due to heteroresistance to vancomycin in this bacteria [5] and local epidemics have been identified and investigated in NICUs [5,6,7]. S. capitis seems to be particularly well-adapted to the NICU environment, possibly in connection with its ability to produce biofilm [8, 9]. However, the neonate contamination routes remain obscure. Recent studies performed in distinct parts of the world have demonstrated a single lineage within the S. capitis species, named NRCS-A, responsible for invasive neonatal infections worldwide [10, 11]. The mechanisms that have driven the global dissemination of this clone have not yet been elucidated. We report the results of the 3-month nationwide BSI survey conducted during the first quarter of 2019 in the largest series of NICUs located in 38 French hospitals. We present clinical data related to the neonates suffering from BSI, and the incidence rates and major characteristics of the neonatal BSIs. In addition, using molecular methods, we characterized the isolates responsible for S. capitis BSIs to establish whether or not they belong to the NRCS-A clone. We provide new data that increase the knowledge about S. capitis in the current neonatal setting.

Materials and methods

BSI epidemiological survey method

Study population

Thirty-eight maternity hospitals comprising neonatal intensive beds participated in the study (Fig. 1). The 447 beds surveyed represented 56.6% of French neonatal intensive beds (

Fig. 1
figure 1

Location of the 38 participating centers and number of neonatal intensive care beds

Study design

The surveillance program involved a 3-month survey of all cases of nosocomial BSI between January 1 and April 30 2019. The survey covered 33,971 intensive care patient-days (PD). Nosocomial BSIs were defined according to international definitions (CDC). The variables studied included clinical data (i.e., sex, gestational age, birth weight, death within 7 days of BSI diagnosis), major characteristics of the BSI such as the portal of entry (skin [primitive cutaneous form or superinfection of a skin breach], lungs, urine, intravascular device, or digestive tract), and for catheter-related BSI, the time lag between the insertion of the catheter, and the appearance of the clinical signs of the BSI. The BSI incidence rates were calculated per 1000 PD. Ethical approval of the surveillance program was obtained at the national level from the Réseau de Prévention des Infections Associées aux Soins.

Microbiological study

PFGE was used as a typing technique [12].

Statistical data

The data were analyzed with R software. Chi-square tests and Fisher’s exact test (two-tailed) were used to test associations, and a P value of 0.05 was considered significant.


Epidemiology of neonatal BSI

During the study period, 141 nosocomial BSIs were diagnosed in 81 male and 60 female neonates. The mean BSI incidence rate was 4.15 per 1000 PD (Table 1). The most frequently isolated micro-organisms were S. epidermidis (39.0%), S. aureus (17.0%), S. haemolyticus (15.6%), and S. capitis (14.2%). Twenty BSIs were polymicrobial (14.2%).

Table 1 BSI, B-cvc, and B-uvc incidence rates per 1000 PD according to the participating centers

The portal of entry of the BSIs was suspected or proven in 83.7% of the cases. The digestive tract (12.1%), the skin (8.5%), and the pulmonary tract (6.4%) were minor portals of entry. Most of the BSIs were catheter-related (70 CRBSIs; 50.0 %) (Table 2). The CRBSI involved a central venous catheter (CVC) in 47 cases (67.1%), all but one associated with staphylococci (97.9%), and an umbilical venous catheter (UVC) in 23 cases (32.9%). The UVC-related BSIs were more diverse than those related to CVC: enterococci-, Enterobacteriaceae-, and B. cereus-BSIs were more frequent with UVC-BSIs (26.1%) rather than with CVC-BSIs (4.3%) (p = 0.022). The median time lag between the insertion of the catheter and the appearance of the clinical signs of the BSI was significantly longer for S. capitis (63.6%, ≥ 10 days) rather than for S. aureus (7.7%), S. epidermidis (16.1%), S. haemolyticus (30.8%), enterococci, and Enterobacteriaceae (no case) (p = 0.018; Table 3).

Table 2 Major characteristics of the BSIs and infected neonates according to the micro-organism
Table 3 Time lag between the insertion of the catheter and the appearance of the clinical signs of the CRBSI

Characteristics of the infected neonates

The gestational age of the infected neonates ranged between 24 and 41 weeks (median value 28), and their birth weight ranged between 455 and 4050 g (median value 1100); 15.6% of the neonates died during the 7-day period after the diagnosis of the BSI. BSIs involving S. aureus, Enterobacteriaceae, and Enterococci were associated with the highest prevalence of early death among infected neonates (29.4, 29.2, and 14.3% for Enterobacteriaceae-, S. aureus-, and Enterococci-associated BSIs, respectively). The prevalence of BSI in the neonates with the a gestational age ≥ 33 weeks and a birth weight > 1500 g differed according to the bacteria (Table 2): it was the highest for Enterococci (42.9%), Enterobacteriaceae (35.3%), and S. aureus (29.2%), lower for S. capitis (20.0%) and E. epidemidis (18.2%) and nil for S. haemolyticus and B. cereus (p = 0.056).

S. capitis BSI characteristics and antibiotic susceptibility of S. capitis strains

Twenty BSIs were associated with S. capitis (14.2%), resulting in a mean incidence of 0.59 per 1000 PD, ranging between 0 and 2.24 according to centers (Table 1); 39.5% of the NICUs reported at least one S. capitis-BSIs. The S. capitis-BSIs were significantly associated with the largest NICUs: at least one S. capitis-BSIs was reported in 15 of the 22 NICUs with ≥ 10 beds, whereas none was reported in the 14 NICUs with < 10 beds (p < 0.001). Four NICUs documented two (n = 3) or three (n = 1) S. capitis-BSIs during the survey period. The antibiotic susceptibility patterns of 18 strains were available (90.0%). Most of the strains were resistant to multiple antibiotics, i.e., methicillin (100%), gentamicin (100%), rifampicin (61.1%), fosfomycin (55.5%), erythromycin (44.4%), fluoroquinolones (33.3%), and fusidic acid (22.2%). Vancomycin and teicoplanin MIC values ranged between 0.25 and 4 mg/L (Table 4). Data regarding antibiotic treatment were available for 18 cases: 17 neonates received vancomycin over 2–24 days (median value: 8 days) and the remaining neonate received linezolid (11 days). A favorable outcome was observed in all but one case. An early death was observed for a preterm infected neonate (gestational age 25 weeks; birth weight 455 g), who received vancomycin over 3 days following the detection of a S. capitis and S. haemolyticus-associated CRBSI.

Table 4 Antibiotic susceptibility of the S. capitis strains

Twelve S. capitis BSI strains from 8 NICUs were available for molecular typing. A considerable homogeneity was demonstrated among the strains, and PFGE pattern analysis demonstrated that all strains belonged to the NRCS-A clone [10] (Fig. 2). Regarding the three NICUs that reported several S. capitis-BSI cases, the strains isolated in a same center shared the same pattern in two cases. In addition, the strains isolated from three distinct centers located in two distant French regions shared the same pattern.

Fig. 2
figure 2

SmaI PFGE patterns of the S. capitis strains responsible for neonatal BSI


This nationwide study adds several elements to the available data on S. capitis responsible for neonatal BSI.

We provide a first mean incidence of S. capitis BSIs in French NICUs. S. capitis BSIs currently involve an average of one neonate per 1700 PD, which is lower than that observed for S. aureus and S. epidermidis, but higher than that of Enterobacteriaceae in the population of neonates surveyed. Our findings confirm S. capitis as a significant agent responsible for nosocomial BSI in the neonatal setting [10, 11, 13].

Second, such as S. epidermidis and S. haemolyticus, we showed that S. capitis preferentially infects the more fragile neonates and thus confirmed that S. capitis is an opportunistic pathogen, devoid of great virulence potential. Concordant with previous studies [13], all the S. capitis strains responsible for BSIs displayed resistance to methicillin and gentamicin, but remained susceptible to vancomycin. S. capitis-BSIs have been taken into account by the clinicians, and vancomycin probably played a crucial role in the recovery of neonates.

Third, we identified one particularity distinguishing S. capitis among the bacteria associated with CRBSI cases. Our study reveals a doubled lag time between insertion of the catheter and the first signs of the BSI involving S. capitis when compared with other bacteria. The absence of early infection likely excludes a contamination of the catheter at the time of its insertion, but rather indicates that the contamination of the catheter may have occurred following catheter manipulations among neonates presenting the longest periods of catheterization.

Finally, the molecular analysis of a large part of the S. capitis strains indicates that they belong to the multidrug-resistant NRCS-A clone and highly suggests likely epidemic phenomena among the NICUs presenting the highest incidence rates of S. capitis BSIs.


Our data confirm the clone NRCS-A particularly well-suited to the neonatal setting and its cumbersome epidemiology [10, 11, 13]. In most NICUs, S. capitis BSIs remain relatively infrequent among neonates, but concern primarily the most fragile ones. In order to better determine the factors involved in the occurrence of these infections, monitoring of BSIs should be continued and complemented by a systematic investigation when several cases are identified over a 3-month period in the same NICU.