Decreased Vancomycin Susceptibility of Coagulase-Negative Staphylococci in a Neonatal Intensive Care Unit: Evidence of Spread of Staphylococcus warneri

Abstract

Coagulase-negative staphylococci (CoNS) are important pathogens in premature neonates; decreasing glycopeptide susceptibility has been observed among these isolates. The epidemiology of colonization with CoNS, the organisms' vancomycin susceptibilities, and genetic relatedness were studied over 6 months in a tertiary-care neonatal unit. A total of 321 isolates of CoNS were isolated. Seventy-five percent of the infants were colonized at admission, and virtually all were colonized thereafter. Common species were Staphylococcus epidermidis (69%), S. warneri (12%), S. haemolyticus (9.7%), and S. hominis (5.6%). A total of 3.9% of CoNS isolates had decreased vancomycin susceptibility (DVS) (MICs > 2.0 μg/ml); isolate recovery was associated with a stay in a neonatal intensive care unit for >28 days (P = 0.039), vancomycin exposure (P = 0.021), and S. warneri colonization (P < 0.0001). Nine of 12 (75%) CoNS with DVS were S. warneri, had enhanceable high-level resistance in vitro, were indistinguishable or closely related by pulsed-field gel electrophoresis, and were different from 29 vancomycin-susceptible S. warneri isolates. Epidemiological analysis suggested unsuspected nosocomial spread. Species determination in certain settings may aid in the understanding of emerging nosocomial problems.

Coagulase-negative staphylococci (CoNS), once dismissed as culture contaminants, have become the predominant bloodstream pathogen in patients requiring prolonged neonatal intensive care (3, 21, 35, 36). The number of infections caused by these pathogens has increased dramatically as management of patients in the neonatal intensive care unit (NICU) has become more dependent on invasive procedures and indwelling devices. The organ sites of CoNS infections vary, but virtually all infections involve contamination of an indwelling medical device (3, 31, 35). A prerequisite for infection is CoNS colonization, which the neonate largely acquires from environmental sources, including health care workers (16, 30). Treatment of CoNS infections is complicated by certain unique characteristics of these organisms, most notably, their ability to express resistance to multiple antibiotics. Resistance to methicillin is almost universal among isolates recovered from hospitalized individuals. For this reason, vancomycin is usually the antibiotic of choice in the treatment of CoNS infections. However, it has been reported that the glycopeptide susceptibilities of clinically significant staphylococci are decreasing (5, 6, 13, 17, 38). While transfer of vanA from enterococci is responsible for the emergence of vancomycin-resistant Staphylococcus aureus (6, 27), the mechanism for reduced the glycopeptide susceptibilities of vancomycin-intermediate S. aureus (12) and CoNS is not entirely known (37, 38).

Noting rising MICs of vancomycin for isolates from patients in the NICU at St. Christopher's Hospital for Children, we undertook a prospective, longitudinal study to examine the hypothesis of the nosocomial spread of CoNS in this tertiary-care NICU during a 6-month period. The goals of this study were to identify the species and determine the vancomycin susceptibilities of CoNS populating infants in the NICU, to determine the molecular relatedness of CoNS species to various vancomycin susceptibilities, and to identify the clinical risk factors associated with decreased vancomycin susceptibility.

MATERIALS AND METHODS

Patient population and specimen collection.

St. Christopher's Hospital for Children is a tertiary-care medical center serving the Philadelphia, Pa., metropolitan area. The NICU is a 30-bed referral unit with approximately 350 admissions yearly; the majority of patients are transferred from primary or secondary NICUs after various lengths of stay. All patients admitted to the NICU between December 1999 and June 2000 were included in the study. Demographic information was collected through chart review and included the following: sex, presence of a central venous catheter, exposure to antibiotics, duration of vancomycin exposure, occurrence of CoNS infection during hospitalization, and length of stay.

As part of the NICU admission protocol, swab specimens of the nasopharynx, rectum, and skin are obtained from all patients to document the colonizing flora, including CoNS, and to aid infection control procedures and the individual infant's medical care. Subsequently, swab specimens of the same sites are obtained from all patients residing in the NICU on routine surveillance days on a quarterly basis. The purpose of scheduled surveillance is to identify the presence of antibiotic-resistant pathogens and to determine the pattern of aminoglycoside resistance among gram-negative bacilli so that recommendations can be made for isolation and empirical antibiotic therapy; the presence of CoNS is not routinely evaluated during this quarterly surveillance. During the study period (December 1999 to June 2000), specimens for culture were obtained on admission and were processed according to the NICU protocol. On subsequent scheduled surveillance days, swab specimens of the same sites were obtained from all infants residing in the NICU and were processed as usual; in addition, the specimens were evaluated for the presence of CoNS. If the day of admission was the same day as a scheduled screening day, only one set of swabs was obtained. Skin and stool specimens were collected by using the BBL CultureSwab Plus Collection & Transport system (Becton Dickinson, Basel, Switzerland). Nasopharyngeal swab specimens were collected by using Copan Venturi Transystem transport swabs and media (Copan, Corona, Calif.).

CoNS isolation and species identification.

All specimens were inoculated onto mannitol-salt agar (Becton Dickinson, Cockeysville, Md.) and incubated for 48 h at 37°C. Morphologically distinct colonies were subcultured individually onto BBL Trypticase soy agar containing 5% sheep blood (Becton Dickinson, Cockeysville, Md.), incubated for 24 h, and identified as CoNS by a conventional methodology as described in the Manual of Clinical Microbiology (23). Coagulase production was determined by the Staphaurex rapid slide coagulase test (Murex Biotech, Kent, England) by mixing one to two colonies from a 24-h pure culture with 1 drop of reagent and observing the mixture for coagulation. S. aureus ATCC 29213 and S. epidermidis ATCC 12228 were used as positive and negative controls, respectively. All presumptive CoNS isolates were frozen at −70°C in cryopreservation vials (Protect Bacterial Preservers; Key Scientific Products, Round Rock, Tex.). Identification of the species of all isolates was performed by using the API STAPH identification system (BioMerieux, Marcy l'Etoile, France) according to the instructions of the manufacturer. Results yielding a quality of identification of 85% or higher were accepted. S. lentus ATCC 700403, S. xylosus ATCC 700404, S. capitis ATCC 35661, and Micrococcus sp. strain ATCC 700405 were used as quality controls. For isolates that yielded a quality of identification of <85%, additional tests were performed as indicated by the API STAPH Analytical Profile Index, 4th ed. (Reference 20-590, September 1997; BioMerieux,). S. saprophyticus ATCC 49453, S. aureus ATCC 29213, and S. epidermidis ATCC 12228 were used as quality control organisms for the additional tests.

Antimicrobial susceptibility testing. (i) Broth microdilution technique.

The broth microdilution technique was used for initial antimicrobial susceptibility testing, for which customized broth microdilution panels manufactured by DADE Microscan (Dade International Inc., West Sacramento, Calif.) were used. A suspension with a density equivalent to that of a 0.5 McFarland standard inoculum was prepared in saline to achieve a final density of approximately 5 × 10⁵ CFU/ml. The panels were incubated at 37°C for 20 to 24 h and were read visually. MICs were read as the lowest concentration at which growth was inhibited. All isolates for which vancomycin MICs were ≥2.0 μg/ml were retested by the same technique in duplicate for confirmation of vancomycin susceptibility. To validate the results for isolates for which vancomycin MICs were 0.25, 0.5, and 1.0 μg/ml, a random sampling of isolates was retested. S. aureus ATCC 29213 was used as a quality control organism.

(ii) E-test for vancomycin susceptibility.

All isolates for which vancomycin MICs of >2.0 μg/ml were repeatedly demonstrated by the broth microdilution method were tested by E-test (AB Biodisk, Solna, Sweden). The standard E-test procedure was performed with these isolates by using Mueller-Hinton II agar (Becton Dickinson, Sparks, Md.) with an inoculum density equivalent to that of a 0.5 McFarland standard. In addition, an investigational E-test protocol of AB Biodisk for detection of glycopeptide resistance (E-test clinical manual, AB Biodisk, 2000) was performed with an enhanced medium (brain heart infusion [BHI] agar; Becton Dickinson, Cockeysville, Md.) and a high inoculum density (equivalent to that of a 2.0 McFarland standard). After 24 h pure cultures of CoNS on blood agar were used to make a bacterial suspension with densities equivalent to 0.5 and 2.0 McFarland standard inoculum densities in BBL Trypticase soy broth (Becton Dickinson, Sparks, Md.). By using sterile transfer pipettes, 0.2 ml of the suspension with a density equivalent to that of a 0.5 McFarland standard was dispensed onto Mueller-Hinton II agar and BHI agar and 0.2 ml of the suspension with a density equivalent to that of a 2.0 McFarland standard was dispensed onto BHI agar. Vancomycin E-test strips were placed onto the agar with sterile forceps. The cultures were incubated for 48 h at 37°C. The inhibition ellipse was read at 24 and 48 h, in accordance with the recommendations of the E-test reading guide (AB Biodisk). This procedure was performed in duplicate, and the results were read independently by two observers to ensure interobserver consistency. S. aureus ATCC 29213 was used as the quality control organism.

Molecular analysis by PFGE.

The protocol used for the preparation of chromosomal DNA was modified from that of Bannerman et al. (1). A culture of a single isolated colony was incubated overnight in 10 ml of BHI broth, and 1.5 ml was then harvested by centrifugation (Eppendorf model 5415C; Brinkmann Instruments, Inc., Westbury, N.Y.) at 10,000 × g for 10 min in a microcentrifuge tube. The cells were washed in 1.5 ml of TE (Tris-EDTA) buffer and centrifuged again. The washed cells were resuspended in 0.5 ml of TE buffer, and 5 μl of a 10-mg/ml solution of lysostaphin was added to the culture suspension. A total of 500 μl of 2% pulsed-field gel electrophoresis (PFGE)-grade agarose dissolved in TE buffer was added to the cell suspension, and the components were rapidly mixed and pipetted into plug molds (Bio-Rad, Richmond, Calif.). The plugs were allowed to solidify at room temperature and were then cooled at 4°C for about 15 min. After the plugs were cooled, each plug was placed in a 50-ml tube containing 5 ml of EC buffer (6 mM Tris-Cl, 1 M NaCl, 0.1 M EDTA, 0.5% Brij 58, 0.2% deoxycholate, 0.5% Sarkosyl) with 10 μl of a 1-mg/ml solution of proteinase K, and the embedded cells were lysed for 2 h at 37°C without agitation. After this lysis step, the EC buffer was removed. The plugs were washed once with 30 ml of TE buffer and were then covered with 30 ml of fresh TE buffer with 10 μl of 1 mg of proteinase K per ml and incubated for 1 h at 56°C without agitation. The plugs were cooled at 4°C for 15 min and were then washed twice with 30 ml of TE buffer on an orbital shaker for 15 min. The plugs were then transferred to 10 ml of fresh TE buffer for storage at 4°C.

The plug was placed in 500 μl of restriction enzyme buffer for 15 min at 25°C. The buffer was removed, and the plug was covered with a total volume of 400 μl of restriction enzyme buffer containing 20 U of SmaI. The plugs were incubated overnight at 25°C without shaking. Following overnight incubation, the DNA fragments were analyzed by loading trimmed slices of the plug into wells of a 1% PFGE-grade agarose gel prepared with 0.5× TBE (Tris-borate-EDTA) buffer. The wells were sealed with 1% agarose. Electrophoresis was performed with the CHEF-DR III electrophoresis cell (Bio-Rad). PFGE size standards were used in each run. The running conditions were as follows: initial pulse, 5 s; final pulse, 40 s; voltage, 6 V/cm; run time 23 h; and temperature, 12 to 14°C. The gels were stained with 0.05% ethidium bromide and photographed. Interpretation of the PFGE banding patterns was done visually according to the criteria of Tenover et al. (39).

RESULTS

A total of 122 patients were hospitalized in the NICU during the study period and were eligible for the study. A total of 111 infants were admitted to the NICU during the study period; 11 were already hospitalized at the start of the study. Screening was inadvertently omitted for four infants at admission. Cultures of samples obtained at admission were available for 107 patients; for 91 patients, this was the only time that samples for culture were obtained. Samples for culture were obtained from 14 patients only on a scheduled screening day. Fourteen patients had two culture episodes (for 13 patients, at admission plus at a scheduled screening; for 1 patient, at two scheduled screenings); 3 patients had three culture episodes (at admission plus two scheduled screenings). A total of 408 specimens were collected during the study period; for 18 patients, less than the 3 specimens that were to be collected according to the protocol were obtained during an episode of specimen retrieval for culture. Ninety-seven of 122 (79.5%) patients hospitalized in the NICU were colonized with at least one CoNS isolate during the study period, with the total yield being 321 CoNS isolates. Eighty-one patients (75.5%) were colonized with a CoNS on admission, whereas 91 to 100% were colonized subsequently (P < 0.001); 26 patients were not colonized on admission. There was no difference between patients with CoNS colonization and those without CoNS colonization by sex, presence of a central venous catheter, or exposure to an antibiotic. The average age at NICU admission was 13.7 days (range, 0 to 250 days). The average age of infants not colonized with CoNS at admission was 3.6 days (median, 1 day; range, 0 to 43 days), whereas it was 8.7 days (median, 4 days; range, 0 to 121 days) for colonized infants (P was not significant); all patients for whom admission screening was inadvertently omitted were admitted on the day of birth. The average duration of the NICU stay for all patients was 26.9 days (range, 1 to 292 days). The durations of NICU stay for patients ever colonized with CoNS and patients not colonized were 31.2 days (range, 1 to 292 days) and 8.5 days (range, 1 to 39 days), respectively (P < 0.001).

Table 1 displays the numbers of CoNS isolates recovered by screening episode and specimen source. The average numbers of CoNS isolates per colonized infant on admission and for each of the three subsequent screening episodes were 2.8, 2.9, 2.8, and 2.3, respectively.

TABLE 1.

Sources of CoNS isolates

Screening episode	No. of patients colonized/no. of patients tested (%)	Total no. of isolates	No. of isolates from the following sourcea:
			Skin	Rectum	Nasopharynx
Admission screening	80/107 (74.7)	222	98	76	48
NICU screening 1	12/12 (100)	39	14	10	15
NICU screening 2	11/12 (92)	35	7	13	15
NICU screening 3	10/11 (91)	25	10	7	8
Total		321	129	106	86

^aPatients may have had more than one species present per episode and/or more than one site colonized.

A total of 305 of 321 isolates were identified to the species level with a ≥85% confidence of identification, and 16 isolates required additional biochemical testing for identification, as determined by the API STAPH Analytical Profile Index results. A total of 12 species were identified. S. epidermidis was predominant, accounting for 69% of isolates. S. warneri (11.8%), S. haemolyticus (9.7%), and S. hominis (5.6%) were the next most commonly isolated species; these four species accounted for 97% of the CoNS isolates. Other species identified were S. saprophyticus (n = 3); S. sciuri (n = 3); S. xylosus (n = 2); and S. chromogenes, S. caprae, S. capitis, S. cohnii, and S. lugdunensis (n = 1 isolate each). S. warneri was isolated more often from the skin and stool (32 of 39 isolates) than from the nasopharynx (7 of 39); otherwise, species and isolation sites were not associated.

Of 97 colonized patients, 53 (55%) were colonized with more than one species of CoNS during the study period; all of these patients were colonized concurrently at separate sites or the same site. Patients colonized with more than one CoNS species were colonized with an average of 2.4 species (range, 2 to 4 species).

Antimicrobial susceptibility testing by the broth microdilution method was performed with 309 of 321 CoNS isolates; 12 isolates were not viable for susceptibility testing. The vancomycin MICs at which 50 and 90% of isolates were inhibited were 1.0 and 2.0 μg/ml, respectively. The vancomycin susceptibilities of the four predominant species (S. epidermidis, S. warneri, S. haemolyticus, and S. hominis) are shown in Table 2 and Fig. 1. Vancomycin MICs were >2.0 μg/ml for 12 isolates, and these isolates were assigned to a group with decreased vancomycin susceptibility (DVS); the MIC did not exceed 4.0 μg/ml for any of the isolates. Of these 12 isolates, 9 (75%) were S. warneri, 2 were S. epidermidis, and 1 was S. haemolyticus. E-tests were performed under standard conditions with isolates for which vancomycin MICs were >2.0 μg/ml, and the vancomycin MICs by the E-test ranged from 2.0 to 3.0 μg/ml. The investigational E-test with enhanced culture conditions (BHI agar, high inoculum density) led to elevations of the vancomycin MICs for the same isolates to 6.0 to 32 μg/ml.

TABLE 2.

Distribution of vancomycin MICs for the four predominant CoNS species^a

Species	No. of isolates for which MIC (μg/ml) was:
	0.25	0.5	1.0	2.0	4.0
S. epidermidis	0	10	124	84	2
S. warneri	0	1	12	17	9
S. haemolyticus	0	2	21	7	1
S. hominis	1	6	7	4	0
Total	1	19	164	112	12

^aAs determined by the broth microdilution test method.

FIG. 1.

Distribution of vancomycin MICs for the predominant CoNS species. The bars represent concentrations of 0.25, 0.5, 1.0, 2.0, and 4.0 μg/ml from left to right, respectively.

The clinical risk factors associated with CoNS isolates for which vancomycin MICs were >2.0 μg/ml by broth microdilution testing are illustrated in Table 3. Exposure to vancomycin for any duration and, especially, exposure to vancomycin for >10 days were associated with colonization with CoNS with DVS (P = 0.021 and 0.002, respectively), as was prolonged NICU stay (P = 0.039). Isolation of a CoNS with DVS was most significantly associated with the species S. warneri (P < 0.0001).

TABLE 3.

Patient risk factors associated with colonization with CoNS with DVS

Patient factor	No. (%) of patients:			P valuea
	Colonized with CoNS (n = 97)	Colonized with CoNS with DVS (n = 9)	Not colonized with CoNS with DVS (n = 88)
Male	66 (68)	8 (88)	58 (66)	0.264
Central venous catheter	26 (27)	4 (44)	22 (25)	0.243
Exposure to any antibiotic	56 (58)	7 (77)	49 (56)	0.294
Exposure to vancomycin	21 (22)	5 (55)	16 (18)	0.021
Vancomycin exposure for >10 days (range, 11-49 days)	13 (13)	5 (55)	8 (9)	0.002
NICU stay for >28 days (range, 29-292 days)	33 (34)	6 (67)	27 (31)	0.039

^aDetermined by the Fisher exact test.

Thirty-eight of 39 S. warneri isolates were available for analysis. Seventeen patients were colonized with S. warneri at admission, and one of these patients was colonized with a strain with DVS; 11 patients acquired S. warneri after admission, and 6 of these patients were colonized with strains with DVS. Vancomycin MICs were >2.0 μg/ml for nine isolates. A single S. warneri isolate with DVS was recovered from a culture of a sample obtained on admission in December 1999, which predated the recovery of eight S. warneri isolates with DVS from six patients during a scheduled nursery screening in March 2000. PFGE analysis revealed the presence of six distinct clonal groups among the S. warneri population. All nine S. warneri isolates with DVS (MICs > 2.0 μg/ml) appeared to be either indistinguishable or closely related by PFGE (Fig. 2). No fully susceptible S. warneri isolate appeared to share this relatively unique PFGE pattern.

FIG. 2.

PFGE analysis of S. warneri isolates with DVS. Lanes 1 and 2, fully vancomycin-susceptible isolates recovered in December 1999; lanes 3 to 11, S. warneri isolates with DVS (the isolate in lane 3 was recovered in December 1999; the isolates in lanes 4 to 11 were recovered in March 2000); lanes 12 to 14, fully vancomycin-susceptible isolates recovered in March 2000; lane 15, molecular weight marker. The patterns in lanes 6 to 9 and 11 are indistinguishable and represent the outbreak strain. The patterns in lanes 3, 4, and 10 are closely related to the modal pattern for the population; the isolate in lane 4 appears to have gained a restriction site, leading to a pattern with a three-fragment difference from the modal pattern. Isolates in lanes 1, 2, and 12 to 14 differ from the outbreak strain and from each other.

DISCUSSION

CoNS can cause significant clinical disease, generally following a breach of normal host defenses (including the presence of a catheter or prosthesis), or less commonly, in individuals with deficient defenses (3, 21, 35, 36, 41). CoNS are the leading cause of bacteremia in the NICU setting, where immunologically immature infants rely on invasive devices for their care. Venous catheters have been implicated in more than one-half of the cases of CoNS bacteremia in NICUs (2, 11, 15, 26). Infections at metastatic sites, such as the central nervous system, heart, bones, and joints, are especially common; and such infections in this vulnerable population are especially difficult to treat (14, 28-30).

Multidrug resistance is a common characteristic of the CoNS, especially among those recovered during hospitalization. Despite the widespread use of vancomycin since its release in 1956, virtually all CoNS have remained susceptible to the concentrations of vancomycin expected to be achieved in serum; and although resistance was first reported in clinical isolates of S. epidermidis (40) and S. haemolyticus (33, 41) in the 1980s, this phenomenon has not become widespread to date. However, recently, we and others have noted increasing vancomycin MICs for clinically important staphylococcal species, both S. aureus and CoNS (5, 6, 13, 17, 38). The mechanism of reduced vancomycin susceptibility in CoNS is unclear but may be related to the selection of resistant subpopulations under pressure of antimicrobial exposure (9, 33, 34, 42). In our study, colonization with CoNS with DVS was significantly associated with vancomycin exposure, especially of long duration (>10 days). The vancomycin MICs at which 50 and 90% of isolates were inhibited for the total population of CoNS were 1.0 and 2.0 μg/ml, respectively, and none of the predominant species were exquisitely susceptible to vancomycin (MICs ≤ 0.25 μg/ml).

S. epidermidis was the most prevalent colonizing species in our study, accounting for 69% of all CoNS isolates. S. warneri was the next most commonly isolated species, accounting for 11.8% of all CoNS isolates. There is limited information on the prevalence of colonizing species of CoNS; however, S. warneri generally has not been identified commonly (8, 15, 25). S. epidermidis has been reported to be the CoNS species most often responsible for clinical disease (8, 11, 15, 25); however, present data or data from closed settings are limited, and species determination is not performed routinely in clinical laboratories. Certain CoNS species have been implicated in significant disease, including S. lugdunensis in endocarditis (19); S. haemolyticus in peritonitis (32, 33), meningitis (18), and bacteremia (24, 41); and S. warneri in bacteremia (20), endocarditis (4, 10, 20, 43), and vertebral osteomyelitis (22, 43).

In our study, although S. warneri accounted for only 12% of the CoNS, it was the second most commonly isolated species and accounted for 75% of all CoNS for which vancomycin MICs were >2.0 μg/ml (P < 0.0001). The vancomycin MICs for S. warneri clustered at 1.0 to 4.0 μg/ml, with the vancomycin MICs for a significant proportion of the S. warneri isolates (23%) being 4.0 μg/ml, whereas the MICs for all other species clustered at 0.5 to 2.0 μg/ml. Using PFGE, we were able to show that all S. warneri isolates with DVS represented an outbreak strain that was uniquely different from vancomycin-susceptible S. warneri. This finding, together with the epidemiologic data, suggest that an infant (transferred from another nursery) who was colonized with an S. warneri isolate with DVS at the time of admission to our NICU in December 1999 was the source of nosocomial spread over the next 3 months, especially to infants highly exposed to vancomycin.

The nosocomial spread of CoNS in the NICU setting has been documented previously. Low et al. (24) reported the ongoing presence of an endemic strain of S. haemolyticus colonizing and causing disease in a NICU over a 5-year period; Huebner et al. (16) documented the persistence of an endemic S. epidermidis strain among bloodstream isolates of NICU patients over a 10-year period. It has been postulated that the infants themselves are the reservoirs of endemic strains of staphylococci, with organism transmission occurring by the hands of health care workers (7). Although Low et al. (24) did not investigate the reservoir or mode of transmission of the S. haemolyticus strain, Patrick et al. (30) and Huebner et al. (16) isolated from the nasopharynges and hands of health care workers, respectively, CoNS that were genetically related to the CoNS colonizing or causing disease in patients during the same time period.

Under the guidelines of the National Committee for Clinical Laboratory Standards, CoNS for which the vancomycin MIC is 4.0 μg/ml are reported as susceptible. Under enhanced culture conditions with enriched media and a high inoculum density, the vancomycin MICs for our isolates of CoNS with DVS rose as high as 32 μg/ml. The discrepancy in susceptibility to vancomycin under routine and enhanced conditions is perhaps related to the selection of resistant subpopulations, or heteroresistance (9, 32, 34, 42); however, there is no clinical evidence at present to support this concept in vivo. In other reports of CoNS with reduced susceptibilities to vancomycin, resistance appears to develop in a stepwise fashion and is related to the duration of vancomycin administration (13, 33). Given the high prevalence of multidrug-resistant CoNS and the emergence of reduced vancomycin susceptibilities among other gram-positive organisms, the detection of vancomycin heteroresistance among CoNS may be of clinical relevance. Treatment with vancomycin at the routine dosage would not be expected to be effective at sequestered sites of infection, such as the central nervous system, for organisms for which the MIC is 4.0 μg/ml or at any site for organisms for which the MIC is 32 μg/ml.

We documented rising MICs of vancomycin for CoNS in our NICU, primarily due to the unsuspected nosocomial transmission of a single S. warneri strain with decreased vancomycin susceptibility. Noting that the vancomycin MICs for S. warneri were overall higher than those for the other CoNS species in our population, we speculate that certain species or strains of CoNS may be more capable of glycopeptide resistance than others. Additionally, demonstration of enhanceable resistance to vancomycin in vitro is a cause for concern about the potential, substantial clinical impact of these species or strains of CoNS.