 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
,
* Department of Pediatrics, Baylor College of Medicine, Houston, Texas
The Woman's Hospital of Texas, Houston, Texas
Department of Molecular Virology & Microbiology, Baylor College of Medicine, Houston, Texas
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods. A retrospective cohort study was conducted of infants who had ISD and were in the neonatal intensive care unit of the Woman's Hospital of Texas, Houston, from January 2000 to June 2002. Confirmed ISD was defined as clinical sepsis and Staphylococcus aureus (SA) isolated from 
1 blood culture (BC) or a sterile body site excluding urine or coagulase-negative staphylococci (CoNS) isolated from 2 BC or from 1 BC and a sterile body site. Probable ISD was defined as CoNS isolated from 1 BC or a sterile body site for which clinical and laboratory data review by 3 infectious disease specialists indicated that antimicrobial treatment was appropriate. Confirmed and combined confirmed plus probable cases were analyzed.
Results. A total of 149 episodes (83 confirmed [39 SA, 44 CoNS], 66 probable) in 137 infants (mean gestational age [GA]: 27.6 weeks [22.4–36.4]; mean birth weight: 981 g [350–2995]) were reviewed. Four (3%) infants had early-onset infection (2 SA, 2 CoNS). Median age at infection onset was similar (17 days SA; 18 days CoNS). Intravascular catheters (IVC) were in situ in a minority of infants with ISD episodes (38% SA, 43% CoNS). CoNS more than SA infections were associated with very low birth weight (<1500 g), lower GA, a history of more IVCs and concurrent total parenteral nutrition, but IVC and parenteral nutrition days were similar. By multivariate analysis correcting for birth weight and complications of prematurity, hypoxia at the time of sepsis evaluation was significantly associated with CoNS and hypotension with SA infections; other clinical features were similar. Methicillin-resistant SA caused 8% of SA infections. Among bacteremic infants, SA more frequently than CoNS involved 2 sites. Overall, SA had more focal complications (primarily bone and joint) than CoNS, resulting in a 2- to 3-fold higher SA-associated morbidity rate. Mortality directly attributable to either organism was similar (5% SA; 5% confirmed, 3% confirmed/probable CoNS).
Conclusion. CoNS ISD occurred in smaller, more premature infants than SA and was IVC associated in a minority of cases. Hypoxia and hypotension were the only presenting features that differentiated CoNS and SA. SA-associated morbidity was substantial, but SA infection carried no greater risk of death (5%) than CoNS.
Key Words: invasive staphylococcal disease • neonate • sepsis • Staphylococcus aureus • coagulase-negative staphylococci
Abbreviations: SA, Staphylococcus aureus • CoNS, coagulase-negative staphylococcus • VLBW, very low birth weight • ISD, invasive staphylococcal disease • IVC, intravascular catheter • TPN, total parenteral nutrition • BC, blood culture • NICU, neonatal intensive care unit • WHT, Woman's Hospital of Texas • C, confirmed • C/P, confirmed and probable • MSSA, methicillin-susceptible Staphylococcus aureus • MRSA, methicillin-resistant Staphylococcus aureus • GA, gestational age • CSF, cerebrospinal fluid • CA, community associated
It has been 100 years since the first report of a nursery outbreak caused by Staphylococcus.1 In the years since 1904, staphylococci have been a prominent cause of serious late-onset infections in neonates. Waves of staphylococcal epidemics have occurred in 20-year cycles amid a constant background of endemic and sporadic disease activity. For much of the 20th century, Staphylococcus aureus (SA) was the causal organism. In recent years, coagulase-negative staphylococci (CoNS) have emerged as important pathogens, especially in very low birth weight (VLBW; <1500 g) neonates.2–9 Staphylococci presumably initiate colonization in the first few days or weeks of life.10,11 Historically, invasive disease rates vary from 3% to 70%,12 and long incubation periods are characteristic of some infection manifestations.
The preterm and VLBW infant is at particular risk of developing invasive staphylococcal disease (ISD), not only because of immature host defense mechanisms but also as a consequence of the manipulations attendant to the intensive care required in the first few weeks of life. The need for procedures that cause interruption in skin integrity, such as placement of intravascular catheters (IVC) as well as endotracheal and upper gastrointestinal tract tubes, and the necessity for prolonged total parenteral nutrition (TPN) or steroid use each increase colonization and infection rates. Contemporary data regarding the clinical presentation, course, complications, and outcome of SA and CoNS infections in this vulnerable population are lacking. This is especially true for CoNS infections because isolation of this organism from a single blood culture (BC) can reflect contamination rather than true infection. The purpose of this report is to describe the contemporary incidence, demographics, clinical manifestations, management, and outcome of infants who had ISD and were admitted to a large level II and III neonatal intensive care unit (NICU) in Houston, Texas, during a 30-month period.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
All infants who were admitted to the NICU and had staphylococci isolated from any site during the study period were identified through microbiology laboratory records. Infants who did not have clinical signs of sepsis13,14 and in whom SA or CoNS was isolated from a nonsterile site or who had isolation of CoNS from a single culture where this isolate was considered a contaminant (eg, antimicrobial therapy was discontinued once culture results were available) were excluded. Infants who met the definition for confirmed or probable ISD (Table 1) were included. Information from the medical record was entered onto a standardized data form by 1 of 2 investigators (C.M.H., D.L.P.). Data gathered included demographics, antenatal history, mode of delivery, complications of prematurity, parenteral hyperalimentation, IVC type and use, history of anti-infective use, indication and duration, clinical presentation of episodes of ISD, laboratory test results, history of staphylococcal colonization, diagnostic imaging test results, therapeutic interventions, and clinical outcome.
|
2 and Fisher exact tests. Normally distributed data were assessed by means and t test; when positive or negative skewing of data occurred, statistical significance was assessed by medians and the Mann-Whitney U test. When appropriate, multiple logistic regression analysis assessed the significance of variables. |
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Among C episodes, CoNS accounted for 44 (53%), methicillin-susceptible SA (MSSA) for 35 (42%), and methicillin-resistant SA (MRSA) for 3 (4%); in 1(1%) episode, both MSSA (1 BC) and CoNS (2 BCs) were isolated. When probable ISD episodes were included in the analysis, there were 63 additional episodes of CoNS (44% of all C/P episodes, 59% of all CoNS episodes), and 3 in which MSSA and CoNS both were isolated during the same episode. Thus, SA caused ISD in 40 infants who were admitted to NICU (1.5% of all infants; 7% of VLBW infants) and CoNS in 97 (4% of all infants; 17% of VLBW infants).
Maternal and Neonatal Characteristics
Sixty-one (45%) infants were white, 46 (34%) were black, 21 (15%) were Hispanic, 5 (4%) were Asian, and 4 did not have ethnicity recorded. Of 50 infants who were products of multiple gestations, 33 were twins, 13 were triplets, and 4 were quadruplets. Baseline antenatal and neonatal characteristics of infants with SA and CoNS are summarized in Table 2. Infants with CoNS were more likely to be VLBW (C: 93% vs 76% [P = .06]; C/P: 92% vs 75% [P = .01]) and had lower gestational ages (GAs) and birth weights than those with SA (Table 2). Complications of prematurity were more common in infants with CoNS infections, possibly reflecting their significantly earlier GA.
|
72 hours of age), with no difference in distribution between SA and CoNS. Four infants (3 with confirmed infections) had early onset of ISD (Table 3).
|
.02). Eighteen (17%) of 103 infants who had chest radiographs performed at sepsis evaluation had radiographic findings consistent with pneumonia (3 SA; 14 CoNS; Table 4). None of the infants with the diagnosis of SA pneumonia had pneumatocele formation or pleural empyema.
|
|
Risk Factors for Infection
Fewer than half of all C and C/P ISD episodes (38% SA, 43% CoNS) occurred while an IVC (umbilical arterial catheter, umbilical venous catheter, central or peripherally inserted central venous catheter, or peripheral arterial catheter) was in situ. Infants with CoNS infection more often than SA had a history of umbilical arterial catheter placement (C: 95% vs 74% [P = .01]; C/P: 93% vs 73% [P = .004]), a higher number of IVCs before developing ISD (median: 3 vs 2; C: P = .1; C/P: P = .03) and were more often receiving TPN (C: 82% vs 54% [P = .009]; C/P: 76% vs 54% [P < .001]) at the onset of infection. There was no significant difference in previous duration of IVC or TPN or in previous colonization rates between the infecting staphylococcal species.
Clinical Manifestations of Infection
Clinical manifestations of SA and CoNS infections are summarized in Table 4. Bacteremia without a focus was the most common presentation for both organisms and was associated with an IVC in situ in 20% of SA infections (both C and C/P): 36% of C and 25% of C/P CoNS cases. CoNS bacteremia in C cases persisted longer than SA (P = .02) regardless of concurrent IVC use. The occurrence of abscess was increased 1- to 2-fold in SA infection and involved multiple sites in 28% of SA episodes. Multiple sites of infection also were affected in 4 of 5 SA and 1 of 2 CoNS episodes involving bone and joint. SA infection more frequently disseminated (involved 2 or more sites) than CoNS, but the difference was not significant (C: P = .26; C/P: P = .1) for the entire cohort or for VLBW infants. However, when this analysis was confined to infants with bacteremia, the association became highly significant (C: P = .008; C/P: P = .001).
Three infants developed endocarditis (1 MSSA; 2 CoNS) proved by echocardiography. Another infant (CoNS) had endocarditis equivalent with portal vein thrombosis. The infant with MSSA endocarditis died on day 9 of therapy from multiorgan failure. The remaining 3 patients received a mean of 36.5 days of antimicrobial therapy (range: 32–42), and all recovered.
A cerebrospinal fluid (CSF) culture that yielded CoNS or SA was rare (Table 4). One infant with disseminated SA infection had mild CSF pleocytosis (65 cells/mm3) with a negative CSF Gram stain and sterile culture. Hematologic indices were similar in episodes of CoNS or SA; white blood cell differential counts showed neutrophil predominance more commonly in CoNS infections (mean: 51% vs 37%; P = .001 for both C and C/P) without a difference in immature neutrophil forms.
More than 1 pathogen was isolated during an ISD episode in 13 confirmed (6 SA, 7 CoNS) and 24 (6 SA, 18 CoNS) C/P infections. There was more morbidity in infants with multiple pathogens isolated than those with SA or CoNS alone (C: P = .003; C/P: P = .025), but overall mortality rates and specifically ISD-related mortality or morbidity rates were not different.
MRSA Infections
Three infants developed confirmed ISD caused by MRSA (Table 5). One infant grew Klebsiella pneumoniae from a peripheral BC and Candida parapsilosis from CSF on the same day that MRSA was isolated from a BC. No infant had disseminated MRSA disease. Each MRSA isolate was susceptible to vancomycin, gentamicin, trimethoprim-sulfamethoxazole, and clindamycin.
|
Other Therapies
Despite the association of hypotension with SA infections by multivariate analysis, there was no difference in the number of infants who received volume (fluid bolus) or pressor support between SA and CoNS episodes in both C and C/P cases. Ventilatory support was more frequently increased with CoNS infections. During episodes of bacteremia when an IVC was in situ, the IVC was removed more often with CoNS bacteremia than with SA (C: 68% vs 25% [P = .03]; C/P: 40% vs 25% [P = .5]). There was no difference in the occurrence of complications or in outcome between infants who were treated with an IVC in situ and those in whom it was removed, regardless of staphylococcal species.
Outcome
Among the infants with confirmed ISD, the overall mortality rate was 10% (8 of 79) and morbidity was 30% (24 of 79); 7 additional infants with probable infection died, and 25 had significant morbidity. When these rates were adjusted as being attributable to an ISD episode (as a result of either clinical events during presentation or complications of the infection or treatment), SA was more likely to result in a poor outcome but not death (Fig 2). Each infant with SA- or CoNS-related mortality or morbidity had a history of bacteremia, although not necessarily disseminated infection. One infant with SA-related morbidity weighed 1610 g at birth; all other infants with ISD-related mortality or morbidity were VLBW.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The presenting features of neonatal SA disease were well documented >4 decades ago.14–17 The most dramatic presentation is septic shock with respiratory distress, temperature instability (more commonly hypothermia than fever), and evidence of poor perfusion, which often is rapidly fatal.5,8,18 Although clinical signs generally are nonspecific, the frequency and rapidity with which SA causes focal pyogenic complications in almost all organ systems including lungs, skin and soft tissues, bones, and joints can be helpful clinically in distinguishing the etiology, even before BC results are available. Prolonged SA bacteremia despite adequate antimicrobial therapy should prompt a search for pyogenic foci, endocarditis, or spread to the central nervous system. Focal manifestations of SA infection without concurrent bacteremia are less likely to cause fulminant disease but nevertheless cause significant morbidity, including pneumonia or pleural empyema, bullous impetigo, recurrent furunculosis, cervical adenitis, and breast abscesses.19,20 Specific syndromes, such as toxic epidermal necrolysis21 and toxic shock syndrome,22 also have been described in neonates.
On the basis of these historical reports, we anticipated that our data would verify that SA caused more severe disease than CoNS, with a greater incidence of focal complications and poor outcome. Our findings only partially confirmed these assumptions. SA was more likely than CoNS to cause disseminated disease and was associated with more pyogenic complications, such as abscess or bone and joint infection. Morbidity rates were 4-fold higher in survivors of SA disease, predominantly from the sequelae of bone and joint infection. "Classical" presentations, such as pneumonia with pneumatoceles or pleural empyema,15,16 did not occur. There were few differentiating clinical signs at the time of sepsis onset to suggest SA rather than CoNS. Hypotension was associated with SA infection by multivariate analysis, but infants with CoNS infections more often required additional respiratory support even when corrected for birth weight, GA, and complications of prematurity. Although cutaneous manifestations were more common with SA infection, no characteristic skin lesion to aid in diagnosis was observed in infants with either organism. Pyogenic bone and joint infection, considered a rare manifestation of CoNS infection,23 was diagnosed in 2 infants. Potentially serious complications of ISD, such as endocarditis and pneumonia, were equally or more common among infants with CoNS than with SA.
Eight percent of SA infections that were acquired while the infant was in the NICU were MRSA and demonstrated an antimicrobial susceptibility typical of strains reported to be community associated (CA).24–28 CA MRSA is an increasingly frequent isolate from patients in many areas of the United States.24 The newborn infant, even when hospitalized from birth, can become colonized by organisms from the community,29 whether by vertical transmission from the mother29,30 (as probably occurred in infant B, Table 5) or by horizontal transmission during visits from siblings or relatives31 or even from colonized health care workers. Because ISD is preceded by colonization, CA MRSA may assume a greater importance in the NICU as it becomes widespread in communities.
The Centers for Disease Control and Prevention recommend restricting vancomycin use to prevent the emergence of vancomycin-resistant organisms.32 Some investigators maintain that use of vancomycin is unnecessary for the empirical therapy of suspected late-onset sepsis in VLBW infants,18,33–35 citing as reasons the low rate of fulminant sepsis associated with CoNS,18,33 favorable outcome from invasive CoNS infections in general, and a low rate of MRSA infections. However, in our cohort, CoNS sepsis occurred in the absence of established risk factors except VLBW in most infants, there were few clinical clues to distinguish CoNS from SA (the 2 most common pathogens causing late-onset sepsis in VLBW infants), and the outcome from CoNS ISD was not reliably benign. MRSA caused a substantial number of SA infections in our NICU population, and delaying appropriate therapy until culture results were available in these infants could have had harmful consequences. In our NICU, initiating vancomycin for the empirical therapy of suspected late-onset sepsis in VLBW infants until culture results are available at 48 hours seems prudent. Efforts to promote judicious vancomycin use must, of necessity, accompany this practice. Improving the definition of confirmed CoNS infection from culture contamination by routinely obtaining 2 BCs from separate sites in the evaluation of late-onset sepsis, whenever possible,36 should be encouraged. Vancomycin should be discontinued after 48 hours unless culture data support its ongoing use. Even so, 25% of our infants with MSSA infections received vancomycin to complete therapy, despite the fact that anti-staphylococcal penicillins are more rapidly bactericidal than vancomycin, have a narrower antimicrobial spectrum, and are the recommended drugs of choice for MSSA infections.37
The most noteworthy finding in our study was that the mortality rate attributable to SA was similar to that for CoNS (5% for each in C cases; 5% SA and 3% CoNS in C/P cases combined). This SA-related mortality rate is similar to or lower than that described in the literature.5,8,9,30,38 Our cohort differs from others in which CoNS-related mortality was rare.8,9,18,33,39 Undoubtedly, innate differences between infants who are affected by each pathogen, such as the proportion of VLBW infants at risk and the definition of invasive CoNS infection, influences discrepancies between our data and those of others.8,9,18,33,39 Infants with CoNS had lower birth weight and GA than infants with SA, rendering them more susceptible to poor outcome even with a less virulent organism. We believe that the assumption that CoNS is a "benign" pathogen is exaggerated by misclassification of episodes in which isolation of CoNS from a single BC is defined as "sepsis" instead of as a contaminant that cannot lead to a poor infant outcome.
There are some limitations to our study. First, we lacked definitive microbiologic evidence (2 sterile site cultures yielding CoNS) for the diagnosis of invasive CoNS infections in 59% of episodes (probable infections). For these episodes, we relied on the opinions of infectious diseases specialists who evaluated data collected retrospectively. Each specialist was blinded so that a decision on the validity of assigning an episode as "probable" was based on strict clinical and microbiologic criteria that would prompt a full course of therapy. Our analyses of C cases were very similar to C/P cases, suggesting that our probable cases most likely represented true CoNS infections. However, it would be important to verify our observations by a prospective observational study of ISD in neonates. Second, our study was limited to a single NICU, where there is an emphasis on uniformity of clinical practice among the neonatologists. Although there were no clusters or outbreaks of ISD during the study period, it is possible that practice differences in our NICU, such as emphasis on strict cohorting of admissions, routine surveillance for MRSA colonization to promote early isolation of culture-positive infants, and others, influenced infection rates or prevalence of methicillin-resistant isolates. However, it is unlikely that demographic characteristics, clinical features, and outcome would differ if similar case definitions and empirical vancomycin were used by others.
In summary, staphylococci cause a significant disease burden among neonates who are admitted to the NICU. CoNS and SA are the most frequently isolated late-onset pathogens in most reported studies,5,8 but the incidence of CoNS infection likely is overestimated. Using a strict definition, CoNS and SA caused ISD in 17% and 7% of VLBW infants, respectively, and in 4% and 1.5% of all NICU admissions. CoNS more than SA was associated with VLBW, earlier gestation, noninfectious complications of prematurity, increased number of IVCs before ISD but not at onset of ISD, and concurrent use of TPN. There were few signs differentiating SA from CoNS infections at the time of evaluation for sepsis. SA infection more often was associated with disseminated disease and caused significantly more skin, soft-tissue, bone, and joint infections, leading to a greater number of sequelae related to SA than CoNS infections. Mortality rates were identical (5%), despite the more indolent nature of CoNS infection, and most likely resulted from the more fragile nature of infants who developed invasive CoNS disease. A small but important proportion of SA infections in our NICU were caused by MRSA. These data should be considered carefully when initiating empirical therapy for suspected late-onset sepsis for neonates in the NICU.
|
ACKNOWLEDGMENTS |
|---|
We thank Jayne Finkowski-Rivera, MD, and the physicians of Medical Center Neonatology Associates; Elena Zaccaria, MASCP, CIC, Director of Infection Control, Sally K. Chabavizadeh, MASCP, MS, Department of Microbiology, and the staff of the Medical Records Department, The Woman's Hospital of Texas (Houston, TX) for assistance in performing this study. We thank Marcia A. Rench, BSN (Baylor College of Medicine, Houston, TX) for helpful advice on study design, data collection, and analysis; E. O'Brian Smith (Baylor College of Medicine) for advice on statistical analyses; and Robin Schroeder (Baylor College of Medicine) for assistance in preparing the manuscript.
|
FOOTNOTES |
|---|
Reprint requests to (C.M.H.) Department of Pediatrics, Section of Infectious Diseases, Baylor College of Medicine, One Baylor Plaza, Rm 302A, Houston, TX 77030 chealy{at}bcm.tmc.edu
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  E. V. Capparelli, B. T. Bloom, T. J. Kueser, D. G. Oelberg, E. M. Bifano, R. D. White, R. L. Schelonka, S. A. Pearlman, J. Patti, and S. V. Hetherington Multicenter Study To Determine Antibody Concentrations and Assess the Safety of Administration of INH-A21, a Donor-Selected Human Staphylococcal Immune Globulin, in Low-Birth-Weight Infants Antimicrob. Agents Chemother., October 1, 2005; 49(10): 4121 - 4127. [Abstract] [Full Text] [PDF]
|
|
Read all P3Rs
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2004 American Academy of Pediatrics. All rights reserved. |
||
|
||
) and CoNS (
) infections in neonates.