Coagulase-negative staphylococcal infections in a neonatal intensive care unit: In vivo response to cloxacillin

Abstract

BACKGROUND

Following the introduction of cloxacillin and gentamicin as the first line of treatment for possible late-onset sepsis (LOS) in the authors’ neonatal intensive care unit (NICU), it was subsequently noted that very low birth weight (VLBW) infants improved clinically, despite subsequently positive blood cultures for oxacillin-resistant, coagulase-negative Staphylococcus (CONS). The results of the management of VLBW infants with CONS sepsis during one calendar year, based on clinical rather than laboratory findings, are presented.

METHODS

VLBW infants with LOS were identified through the neonatal database, and the charts of those with CONS were reviewed for antibiotic usage, antibiotic resistance pattern, clearance of CONS from the blood and NICU discharge status. Oxacillin sensitivity was determined by the presence of the mecA gene.

RESULTS

From January 1 to December 31, 2002, 27 VLBW infants, treated in the authors’ NICU for LOS due to CONS, were identified. The mean age of LOS infants with CONS was 15 days (median 12 days; range three to 54 days), the mean birth weight (± SD) was 904±247 g, and the mean gestational age at birth (± SD) was 27±2 weeks. All infants were started on cloxacillin and gentamicin, and improved clinically over the first 48 h. Six isolates were sensitive to cloxacillin. Twenty-three infants grew oxacillin-resistant CONS, eight of whom had persistence of CONS on repeat culture secondary to central lines. Two infants grew two strains of CONS. Eighteen of 22 infants (82%) with in vitro oxacillin-resistant CONS had clearance of bacteremia with cloxacillin and gentamicin. Ten infants (37%) received vancomycin, based on the authors’ guidelines. There were no cases of prolonged bacteremia requiring rifampicin. Three infants died, but none of the deaths could be attributed to CONS.

DISCUSSION

The authors describe clinical improvement with clearance of CONS using cloxacillin and gentamicin, despite laboratory results suggesting oxacillin resistance. The authors’ unit policy was based on clinical response and permitted the continuation of cloxacillin, provided that a repeat blood culture was negative. Vancomycin use was suggested for clinical deterioration or persistence of CONS. These results question the in vitro tests of resistance. Clearance of oxacillin-resistant CONS from the blood points to in vivo sensitivity, while the laboratory testing suggests in vitro resistance. The absence of subsequent positive blood cultures for CONS confirms clearance of this organism.

CONCLUSION

It was demonstrated that cloxacillin (150 mg/kg/day dose), along with gentamicin, can clear CONS from the blood within 48 h. The relationship between in vivo and in vitro sensitivities also needs to be further studied both in the laboratory and in a prospective trial.

In January 2000, following three cases of Staphylococcus aureus septicemia in the Ottawa Hospital – General Campus (Ottawa, Ontario) neonatal intensive care unit (NICU) (including one related death), the antibiotic policy for the early management of suspected late-onset sepsis (LOS) in very low birth weight (VLBW) infants was changed from vancomycin to cloxacillin, in combination with gentamicin. Treatment of positive cultures, received at 24 h to 48 h, was guided by the antibiotic sensitivity pattern.

Over the next six months, it was noted that most infants improved on cloxacillin and gentamicin. However, the antibiotic sensitivity patterns for coagulase-negative Staphylococcus (CONS) showed that more than 80% of the strains were resistant to oxacillin and gentamicin. Repeat cultures often showed clearance of bacteremia, before starting vancomycin. Those that remained positive had central lines in situ, and once removed, cultures became negative, suggesting line contamination rather than true septicemia.

An audit of the clinical management during the first 12 months of this policy showed that there was a significant reduction in the use of vancomycin with no serious adverse effects compared with the prior 12-month period. Oxacillin resistance during these two periods was 68% and 79% for 51 and 41 episodes of CONS sepsis, respectively. The reduction in vancomycin use was of the order of 38%, mainly due to those infants with suspected LOS whose cultures were negative.

Following this report, the antibiotic policy was further reviewed and was amended to permit the continuation of cloxacillin beyond 48 h in the presence of clinical improvement, provided that a repeat blood culture was sent. Vancomycin was permitted at any time for clinical deterioration or in the presence of persistence of bacteremia. Gentamicin was stopped after 48 h in the absence of Gram-negative organisms.

We report the findings for VLBW infants with late-onset CONS infection born in 2002.

METHODS

The following data were abstracted from the neonatal database for VLBW infants with LOS: patient name, date of birth, birth weight, gestational age at birth, LOS pathogen and day of onset, and NICU outcome.

LOS was defined as any positive blood or cerebrospinal fluid culture on or after day 5 of life. LOS was further subdivided into three categories according to the pathogen: bacterial, CONS and fungal (Vermont-Oxford Network). Repeat blood cultures were not routinely sent to confirm bacteremia but were suggested before a change in antibiotics.

The charts of infants with LOS due to CONS were reviewed for the following: dates of all blood cultures, blood culture results, antibiotics prescribed, duration of antibiotic treatment, dates of removal of central lines, CONS sensitivity patterns and discharge status. The days of antibiotic usage were calculated from the charts.

Sepsis workup consisted of a single blood culture, cultured in the laboratory using the BacT/Alert (Organon Teknika Corporation, USA) microbial detection system. Antibiotic sensitivity was performed on all positive cultures using Vitek automated system (bioMerieux Vitek Inc, USA). CONS was not further speciated.

‘Contaminant’ CONS was defined as either a repeat negative blood culture within 24 h, or a clinical decision made and recorded by the attending neonatologist with simultaneous cessation of antibiotics.

RESULTS

From January 1 to December 31, 2002, 128 infants weighing less than 1500 g were born alive in the Ottawa Hospital –General Campus. Ten infants died in the delivery room or within five days of birth, and a further seven infants died after five days of life in the NICU, for a mortality rate of 13%.

Thirty-three infants had 36 episodes of LOS: CONS (n=27; one possible contaminant, two mixed CONS, three infants died); Escherichia coli (n=3); S aureus (n=3; one infant died of a brain stem anomaly); Klebsiella pneumoniae (n=1); Candida species (n=1; died); and Neisseria species (n=1).

The overall rate of CONS infection was 21% of liveborn infants (23% of infants alive on day 5). While there were three deaths (11%), none could be assigned primarily to CONS.

S aureus remained present within the NICU, but no episodes of invasive S aureus were recorded, and there were no deaths related to this infection. Candida species continue to be seen rarely in the authors’ NICU, with a prevalence of one to two cases per year (less than 1%).

Three infants had separate episodes of bacteremia or septicemia with a single pathogen: one had CONS, K pneumoniae and S aureus); another CONS and E coli; and the third CONS and S aureus (Table 1).

TABLE 1.

Characteristics and treatment outcomes of infants with coagulase-negative Staphylococcus (CONS)

Case	Sex	Birth weight, g	Gestational age, weeks	Age- CONS, days	Cloxacillin treatment, days	Gentamicin treatment, days	Vancomycin treatment, days	Oxacillin, sensitivity	Central line	Outcome	Other infections
1	M	640	25	8	10	3	0	False	No	d/c day 112	Klebsiella pneumoniae (day 18); S aureus (day 84)
2	M	966	26	6	5	1	7	False	UVC, UAC	d/c day 104
3	F	595	30	12	9	2	0	False	No	d/c day 58
4	M	970	26	19*	2	2	14	False	No	d/c day 85	Possible meningitis
5	F	609	24	9	10	2	0	False	No	Died day 99	S aureus (day 50)
6	F	942	25	7	10	1	2	False	UVC, UAC	d/c day 119
7	F	896	27	16	7	2	0	True/false†	No	d/c day 61
8	F	1201	30	15	10	2	0	False	No	d/c day 47
9	F	1205	30	9	10	1	0	False	No	d/c day 52
10	M	944	29	14	10	2	0	True	PICC	d/c day 48
11	F	689	26	29	10	3	0	False	No	d/c day 95
12	M	640	26	10	10	4	2	False	UVC	d/c day 105
13	F	924	26	15	10	2	0	False	PICC	d/c day 69
14	M	800	27	54	10	5	0	True	No	d/c day 100
15	F	771	25	12	8	1	0	False	No	d/c day 80
16	F	738	24	7	7	2	0	True	UAC	d/c day 131
17	F	1486	30	21	8	3	0	True	No	d/c day 31	Escherichia coli (day 8)
18	F	976	26	9	6	3	4	False/false†	No	d/c day 58
19	F	1429	30	36*	4	4	0	False	No	d/c day 57
20	M	680	23	9	2	2	10	False	UVC, UAC	d/c day 118
21	M	865	26	12	4	4	6	False	No	d/c day 103
22	M	471	25	9	2	8	6	False	UVC	Died day 17
23	M	1047	32	3	10	3	0	True	No	d/c day 32
24	F	1121	29	17	10	2	0	False	No	d/c day 53
25	M	1076	26	11	5	2	5	False	No	d/c day 94
26	F	921	28	10	8	2	0	False	UVC, PICC	d/c day 57
27	M	800	25	14	2	4	8	False	PICC	Died day 46

^*Considered contaminant;

^†Mixed growth with two CONS organisms. Age-CONS Day of life of first positive blood culture; d/c Discharged; F Female; M Male; PICC Peripherally inserted central catheter; UAC Umbilical artery catheter; UVC Umbilical venous catheter

Of the 27 infants with CONS, 12 were male (44%). The mean birth weight was 904±247 g, and mean gestational age was 27±2 weeks. The mean age of onset of CONS LOS was 15 days (range three to 54 days, median 12 days).

Table 2 displays the antibiotic sensitivity patterns and the rate of clearance with cloxacillin. Twenty-four of 28 isolates (86%) were cleared with cloxacillin and gentamicin alone. Seventeen infants received only cloxacillin after 48 h of therapy, while 10 were prescribed vancomycin, according to the predetermined NICU guidelines for CONS. Thirteen of 23 infants (56%) with oxacillin-resistant CONS received no vancomycin at any stage.

TABLE 2.

Coagulase-negative Staphylococcus (CONS) antibiotic sensitivity patterns and response to antibiotics

CONS antibiotic sensitivity	Number*	Cleared by cloxicillin plus gentamicin (culture proven)	Persistent CONS	Unknown†
Oxacillin, n (%)	6 (2)	6 (100)	0 (0)	0 (0)
Vancomycin, n (%)	22 (79)	18 (82)	3 (13.6)	1 (5)

^*28 CONS isolates, two contaminants excluded;

^†No blood culture drawn before vancomycin

CONS isolates were not further speciated, but Staphylococcus epidermidis has been the most common species in the recent past.

DISCUSSION

CONS is the organism most commonly isolated in LOS in VLBW infants. Almost all infants in our NICU (data not shown) are subjected to one septic workup for LOS and have antibiotic therapy commenced. We modified our LOS antibiotic policy so as to restrict the use of vancomycin in our infants. In this review of our practice, we have demonstrated that in infants with CONS bacteremia or septicemia, clinical improvement, along with clearance of CONS from the blood, occurred using a combination of cloxacillin and gentamicin, despite the presence of CONS, which is resistant to oxacillin in vitro. This result may have major implications in the management of LOS, leading to a major reduction in the use of vancomycin.

The scientific basis for our results is unclear. Suggestions include a high false-positive ‘contamination’ rate with CONS, in vivo synergy between cloxacillin and gentamicin, and poor sensitivity of in vitro testing for oxacillin resistance, which is based on the presence of the mecA gene.

In early 2000, following an outbreak of several cases of S aureus septicemia in our NICU, there was discussion regarding the choice of empirical antibiotic treatment for VLBW infants with possible LOS. A decision was made to change from vancomycin to cloxacillin in combination with gentamicin, to both optimize coverage against S aureus and to reduce vancomycin use, in accordance with reports in the literature (1,2). Vancomycin use was to be restricted to CONS culture-positive infants. The recent report (3) of a case of vancomycin-resistant S aureus further highlights the need for the judicious use of antibiotics.

Following the change, there were no further deaths from S aureus, despite several new cases. We reported our findings in abstract form (4), showing a reduction in the use of vancomycin of 38% over the previous 12-month period, mainly in infants with culture-negative suspected LOS. Our policy was to stop antibiotics at 48 h, as proposed by Isaacs and Moxon (5), and more recently by Kaiser et al (6). At the same time, we also noted that infants improved on the new regimen, even those whose blood cultures subsequently grew CONS, most of which showed laboratory resistance to oxacillin. In view of the clinical improvement, repeat blood cultures were taken before changing to vancomycin, and most of these were negative. Our antibiotic policy for LOS was thus revised in 2001 to permit the continuation of cloxacillin beyond 48 h, provided that a repeat blood culture was taken and was negative. Vancomycin use was permitted at the discretion of the attending neonatologist for any clinical deterioration or for persistent bacteremia.

Blood culture technique and contamination

All blood cultures in our NICU are taken by the nursing staff, using a standard technique, which consists of cleansing the area with chlorhexidine and waiting 60 s before venous sampling. Central line cultures – umbilical artery catheters or umbilical venous catheters – were not taken, nor were peripherally inserted central catheter (PICC) line cultures because the PICC used do not permit to removal of blood. Eight infants (30%) had positive repeat cultures, most of whom had indwelling central lines; once these were removed, the cultures became negative with cloxacillin alone in all but three infants. These three infants were prescribed vancomycin, as per protocol, but continued to have CONS; one developed ecthyma and survived (case 20), and a second infant (case 22) subsequently died with persistent CONS, hypotension and renal failure, thought to be related to his severe intrauterine growth restriction status. The third infant (case 27) had persistent bacteremia until removal of a PICC line.

Our overall rate of CONS infection – 21% of liveborn infants (23% of infants alive on day 5) – is in line with other reports (7,8). While there were three deaths, none could be assigned primarily to CONS, and this is in accordance with Karlowicz et al (9), who showed that CONS does not appear to be a frequent cause of fulminant sepsis.

We are unable to calculate the mean time to clearance of CONS because repeat cultures were not routinely performed. Three infants had repeat cultures within 24 h, all of which were negative and thus considered ‘contaminants’. One infant (case 22) died with a persistently positive culture. Three infants with oxacillin-sensitive CONS did not have repeat cultures sent. In the remaining 20 infants who had repeat cultures sent, the mean number of days to a negative blood culture was 6±4 days, but the timing of the repeat cultures had a very wide range (two to 19 days).

The true contamination rate for CONS sepsis is unknown. Struthers et al (10) reported a rate of 5% when comparing dual cultures taken at different sites. Kadilkar et al (11) reported a rate of close to 60% in a prospective study of VLBW infants. Our practice of taking a single blood cultures is in line with that of the majority of neonatologists surveyed by Rubin et al (12). Furthermore, a sepsis workup was only carried out if an infant showed any of the abnormal clinical signs that are generally regarded as signs of possible sepsis.

Three infants (11%) satisfied our definition for ‘contaminant’ CONS. All had negative repeat blood cultures within 24 h (cases 4, 19 and 23). One infant had posthemorrhagic hydrocephalus (case 4), which could have caused the abnormal clinical signs. Coincidentally, this infant also had a lumbar puncture to control his hydrocephalus, and although this did show a very light growth of CONS, the other cerebrospinal fluid indexes did not suggest meningitis. Treatment with vancomycin was recommended and continued for 14 days, despite the negative repeat blood culture. In only one infant (case 19) were antibiotics stopped.

Time to positivity of blood culture was not used as a criterion for contamination, although this has been proposed by others (6,13).

Antibiotic management

Our decision to use cloxacillin and gentamicin as empirical therapy for suspected LOS is not unique. Karlowitz et al (9) reported their experience over a four-year period and concluded that this combination was reasonable due to the low frequency of fulminant sepsis caused by CONS. Our experience shows that this regimen can actually clear the CONS bacteremia, as it did in 24 of 27 infants (89%), despite in vitro resistance in most isolates. Allen and da Silva (14) recently reported that ampicillin and netilmicin were a safe combination for suspected LOS.

Our method of continuing cloxacillin alone while awaiting repeat blood cultures is similar to that reported by Matrai-Kovalskis et al (15), who prescribed vancomycin to just 9% of infants with CONS. In our NICU, over one-half of the infants with oxacillin-resistant CONS (13 of 23 [56%]) received no vancomycin at all. In the 10 infants who did receive vancomycin, all but three had a negative blood culture before commencing vancomycin (cases 20, 22 and 27 as previously described, who had persistent bacteremia despite line removal).

Krediet et al (16) reported the continued use of cephalothin in the presence of clinical improvement, despite an 85% rate of methicillin resistance on mecA gene testing. They reduced their vancomycin use from 62% to 21% over a two-year period.

It appears difficult to explain the clearance of CONS in 85% of the infants based on in vitro sensitivities. There may be synergy between cloxacillin and gentamicin in the dosages used. However, on further testing, all the oxacillin-resistant CONS samples were also resistant to gentamicin (data not shown). We used a cloxacillin dose of 150 mg/kg divided and administered every 8 h – a reasonably large dose – but one which we thought was appropriate because of the rise in the prevalence rate of S aureus, a much more serious pathogen than CONS.

Another explanation may lie in the in vitro testing for resistance to oxacillin, which is based on the presence of the mecA gene in CONS. Related to this, the American Academy of Pediatrics Red Book (17) states that “the MIC breakpoints for CONS for methicillin are based on the presence of the mecA gene, and therefore, the potential for the organism to develop resistance. The presence of the gene however, does not predict an accurate measure of gene expression (ie susceptibility or resistance).” While a recent report from our hospital laboratory (18) showed good correlations between oxacillin resistance testing by agar dilution, disk diffusion using the Vitek (GPS-105 card) (bioMerieux Vitek Inc, USA) and the presence of the mecA gene by multiple polymerase chain reaction, they were not 100%. However, Gradelski et al (19) found that 50% of mecA-negative strains of CONS were falsely identified as oxacillin resistant. Recent changes in the breakpoint for in vitro sensitivity, published by the National Committee for Clinical Laboratory Standards (20), have increased the number of isolates that are classified as resistant to methicillin in the laboratory, with a sensitivity reaching 98%, but evidence for clinical correlation is very sparse. Krediet et al (16), using cephalothin, showed clinical recovery in 22 of 25 infants whose CONS was mecA positive, when there was concordance of mecA gene carriage and oxacillin resistance in 85% of isolates.

A recent report (21) suggests that the simultaneous presence of both the mecA and the vanA genes in methicillin-resistant S aureus may lead to inactivation of the mecA beta-lactam resistance mechanism but not of the vancocomycin resistance. Further work on CONS is warranted to see whether this mechanism is present and may explain what we have noted clinically.

CONCLUSIONS

We have shown that the combination of cloxacillin and gentamicin can lead to the clearance of CONS, even in the presence of oxacillin resistance in vitro. Furthermore, continuation of cloxacillin alone led to clinical and laboratory resolution of CONS infection in most infants, and completely avoided the use of vancomycin in more than one-half of the infants with oxacillin-resistant CONS. These findings question the interpretation of in vitro testing of antibiotic sensitivity in relation to CONS based on the presence of the mecA gene.

We believe that it may be safe to continue cloxacillin in infants who are improving clinically and in whom a repeat blood culture is taken to confirm clearance of CONS, and that a randomized trial may help to confirm (or refute) this approach. Infants who do not improve should have central catheters removed promptly, and vancomycin should be considered. We also believe that the relationship between in vitro and in vivo sensitivities of CONS needs to be further studied both in the laboratory and in a prospective trial.