Journal of Perinatology (2006) 26, 111–114 r 2006 Nature Publishing Group All rights reserved. 0743-8346/06 $30 www.nature.com/jp ORIGINAL ARTICLE A comparison of neonatal Gram-negative rod and Gram-positive cocci meningitis

PB Smith1,2, CM Cotten1, HP Garges1, KF Tiffany1, RW Lenfestey1, MA Moody1,JSLi1,2 and DK Benjamin Jr1,2 1Department of Pediatrics, Duke University, Durham, NC, USA and 2Duke University Clinical Research Institute, Durham, NC, USA

intrapartum antibiotics to prevent disease from group B Objective: Neonatal meningitis is an illness with potentially devastating streptococci (GBS) has been shown to select for more resistant consequences. Early identification of potential risk factors for Gram-negative Gram-negative rod (GNR) organisms.3,4 The importance of GNRs rod (GNR) infections versus Gram-positive cocci (GPC) infection prior to as a pathogen in the neonatal period, therefore, may be obtaining final culture results is of value in order to appropriately guide increasing.3,5–7 expirical therapy. We sought to compare laboratory and clinical parameters Practitioners often rely on cerebrospinal fluid (CSF) parameters of GNR and GPC meningitis in a cohort of term and premature infants. rather than CSF culture to define meningitis because of the use Study Design: We evaluated lumbar punctures from neonates cared for of intrapartum antibiotics and the practice of delaying lumbar at 150 neonatal intensive care units managed by the Pediatrix Medical puncture (LP) until after blood culture results are available. Group Inc. We compared cerebrospinal fluid (CSF) parameters (white Moreover, CSF cultures may be negative in the face of true 8–11 blood cell count, red blood cell count, glucose, and protein), meningitis. Although normal values for CSF parameters in demographics, and outcomes between infants with GNR and GPC healthy full-term infants and for full-term infants with meningitis 1,12,13 meningitis. CSF cultures positive with coagulase-negative staphylococci have been described, the differentiation between Gram- were excluded. negative and Gram-positive infections based on cell-count, glucose, and protein levels in the CSF is less well defined. Most studies Results: We identified 77 infants with GNR and 86 with GPC meningitis. describing CSF findings in GNR meningitis are from single centers There were no differences in gestational age, birth weight, infant sex, and often include older children and adults.12,14–16 The purpose of race, or rate of Caesarean section. GNR meningitis was more often this report was to compare CSF parameters observed in culture- diagnosed after the third postnatal day and was associated with higher proven neonatal meningitis, based on the assumption that if CSF white blood cell and red blood cell counts. GNR meningitis diagnosed in parameters are identified that distinguish GNR from Gram-positive the first 3 days of life was associated with antepartum antibiotic exposure. cocci (GPC) meningitis, practitioners could make better initial No difference was noted in either CSF protein or glucose levels. After antimicrobial therapy choices. correcting for gestational age, there was no observed difference in mortality between infants infected with GNR or GPC.

Conclusion: Compared to GPC meningitis, GNR meningitis was associated with several aspects of the clinical history and laboratory Methods findings including older age of presentation, antepartum exposure to A cohort of neonates discharged from 150 neonatal intensive care antibiotics, and elevated CSF white blood cell and red blood cell counts. units (NICUs) managed by the Pediatrix Medical Group Inc. from Journal of Perinatology (2006) 26, 111–114. doi:10.1038/sj.jp.7211438 1997 to 2004 that had a LP performed was evaluated. The data were obtained from an administrative database with collection Keywords: cerebrospinal fluid; mortality; antibiotics method described previously.10,17 LP data from neonates with CSF reservoirs and shunts were excluded from the study. Meningitis was diagnosed by either a positive CSF culture or a positive antigen test Introduction from the CSF with a corresponding positive blood culture with the The incidence of bacterial meningitis is higher in the first month same organism. CSF was processed at each site according to the of life than at any other period.1,2 The widespread use of individual hospital’s laboratory standards. Positive CSF cultures for coagulase-negative staphylococci were excluded. Only the first Correspondence: Dr DK Benjamin Jr, PO Box 17969, Duke Clinical Research Institute, positive CSF culture from each patient was used in the analysis. Durham, NC 27705, USA. E-mail: [email protected] The Duke University Institutional Review Board provided Received 8 September 2005; revised 31 October 2005; accepted 12 November 2005 permission to conduct this analysis. A comparison of neonatal GNR and GPC meningitis PB Smith et al 112

Outcomes We compared CSF parameters (white blood cell (WBC) and red blood cell (RBC) counts, glucose, protein) and demographic data from infants with meningitis caused by GNR and GPC. For the analyses, mortality was defined as death of infant prior to discharge. P-values were calculated using nonparametric testing, either Wilcoxon rank sum or Fisher’s exact test; except when logistic regression was used. CSF parameter values were described by their medians rather than their means as each CSF parameter dataset contained several extreme outliers. Analysis was conducted using SAS 8.02, (SAS Institute, Cary, NC, USA). All reported P-values are two-tailed. Figure 1 Cell count (per mm3).

Results Table 1 Demographics of the cohort We evaluated CSF results from 14 017 lumbar punctures; 86 infants GNR (n ¼ 77) GPC (n ¼ 86) P-value were infected with GPC and 77 with GNR. The demographics of the cohort are outlined in Table 1: 38% of the infants were very-low- Gestational age 0.28 birth-weight (<1500 g) and 56% were born preterm. There were no 25% tile 28 27 significant differences between the two groups in birth weight, Median 33 37 gestational age, gender, race, or mode of delivery. Infants 75% tile 38 39 undergoing LP after the third day of life LP were more likely to be infected with GNR organisms. Birth weight (g) 0.18 Of the 60 neonates diagnosed with meningitis in the first 3 days 25% tile 1075 917 Median 1718 2808 of life, 69% (11/16) of those exposed to antepartum antibiotics were 75% tile 2805 3457 infected with GNRs while only 30% (13/44) of those with no previous antibiotic exposure were infected with GNRs. After Sex 0.75 controlling for preterm delivery (as a dichotomous variable) using Male 39 (51%) 46 (53%) logistic regression analysis, GNR infection in the first 3 days of life Female 38 (49%) 40 (47%) was associated with previous antepartum antibiotic administration (OR ¼ 4.6 (1.2, 16.8), P ¼ 0.023). Preterm delivery itself was Day of life tapped 0.039 associated with a higher risk of GNR infection in the first 3 days 0–3 days 24 (40%) 46 (59%) of life with an OR of 3.3 (1.04, 10.6). >3 days 36 (60%) 32 (41%) There were significant differences between WBC and RBC counts between infants infected with GNR and GPC (Figure 1). The Amiotic fluid 0.76 median WBC count was 1217/mm3 (25th%, 75th%: 43, 3263) for Meconium 6 (9%) 6 (8%) No meconium 50 (91%) 66 (92%) neonates with GNR meningitis and 187 (38, 980) for neonates with GPC meningitis (P ¼ 0.043). The median RBC count was Delivery method 0.59 3 465/mm (100, 3275) and 160 (18, 1145) for GNR and GPC, Vaginal 33 (58%) 47 (63%) respectively (P ¼ 0.024). No difference was observed in CSF glucose C-section 24 (42%) 28 (37%) levels; 22 mg/dl (9, 52) for GNR infections and 36 (3, 55) for GPC infections (P ¼ 0.89) or in CSF protein levels; 296 mg/dl (171, Race 0.31 588) and 279 (146, 550) for neonates with GNR and GPC, White 44 (57%) 46 (53%) respectively (P ¼ 0.79). Median WBC counts for individual species Black 9 (12%) 12 (14%) are given in Table 2. Hispanic 17(22%) 23 (27%) The mortality associated with selected species of GNRs and GPC Other 7 (9%) 5 (6%) are shown in Figure 2. The mortality of GNR meningitis was 13% (8/64) compared with 4% (3/75) in GPC meningitis. Using logistic regression to control for gestational age, the mortality observed in eight infants with nonenteric GNR infections died, including the the GNR cohort was not statistically different than the mortality five infected with Pseudomonas aeruginosa. The mortality rate observed in the GPC cohort, OR ¼ 2.7 (0.6, 11.3). None of the was 16% (4/25) in neonates with Escherichia coli.

Journal of Perinatology A comparison of neonatal GNR and GPC meningitis PB Smith et al 113

In this study, we found significantly higher WBC and RBC counts in the infants with GNR meningitis. One explanation for the increased RBC count observed in the GNR cohort might be from higher rates of traumatic LPs in this lower birth weight cohort. Although the previously observed differences in CSF glucose and protein levels were not observed, this study was compatible with the previous observations that CSF parameters in GNR infections are more likely to be abnormal.1,12 Higher CSF WBC and RBC counts in the proper context should increase the clinician’s suspicion for GNR meningitis. Neonatal GNR meningitis has been associated with increased mortality (30–41%) when compared with GPC meningitis (15%).2,11,16 The mortality rate for GNR meningitis in our sample, Figure 2 Mortality by organism. 13%, is lower than most previously published findings. One reason for this finding may be our inclusion of all neonates with meningitis regardless of birth weight. Another recent study that Table 2 CSF WBC by species included neonates of all birth weights also found a low mortality rate, 7% (3/43), in neonates with meningitis, including only 8% Species N Median Range (1/13) in GNR meningitis.21 Stoll et al.11 examined only VLBW All GNRs 46 1217 (1–38 000) infants in which the mortality rate following GNR meningitis was Escherichia coli 22 2525 (1–15 900) found to be 41%. However, only 3/22 (14%) VLBW neonates with Pseudomonas 4 757 (7–7480) GNR meningitis in our study died. The small number of cases of Enterobacter 4 657 (4–5850) GNR meningitis in the prior studies as well as our study may contribute to the variability observed in the mortality rates. All GPC 41 187 (0–47 400) A recent multicenter study demonstrated that mortality for GBS 28 271 (2–13 000) Pseudomonas bacteremia (33%) is higher than mortality from Staph aureus 9 35 (0–980) other GNRs, GPC and even Candida.22 However, all five of the Enterococcus 2 23 701 (3–47 400) neonates with Pseudomonas meningitis in this study survived. This difference may be related to the fact that in order to be included in this study, an infant had to have the clinical stability Discussion to undergo a LP. This potential selection bias applies to each of The presentation of meningitis in infants, especially preterm the observations regarding mortality. neonates, is notoriously subtle and nonspecific so recognizing In this cohort of neonates, diagnosis after the third day of life which infants are at increased risk for GNR infections is was associated with GNR meningitis. These findings were consistent important.1,2,15,18,19 GNR meningitis has been associated with GNR with previous reports of the emergence of GNR organisms as a bacteremia, preterm birth, male sex, neural tube defects, and common etiology of late-onset-sepsis in VLBW infants.23 Delaying urinary tract anomalies.14,15,20 Diagnosis by culture is often the spinal tap because of instability of the lower birth weight complicated by empirical use of antibiotics and delay in LP. infants may contribute to the later diagnosis of GNR compared The purpose of this study was to determine if there are features to GPC meningitis. of the CSF analysis that might be predictive of GNR meningitis. An important finding from this study is the observation that Prompt identification of GNR meningitis in order to institute of infants diagnosed with meningitis in the first 3 days of life, proper antimicrobial treatment is crucial in improving outcomes.16 presence of GNR infection was associated with previous antepartum CSF parameters are often used to determine if empirical therapy antibiotic exposure. Several centers have observed an increase in should be given or if patients should be treated for meningitis in GNR bacteremia in newborns with the advent of maternal the presence of negative CSF cultures.10 prophylaxis for GBS.24 In addition, administration of antepartum Previous studies examining GPC and GNR CSF parameters are ampicillin has been associated with ampicillin resistance in GNR limited. The findings from one previous study demonstrated isolates from infected infants.3,5,25 In some hospitals, 70% of E. coli normal CSF WBC in only 4% of neonates with GNR meningitis isolates are resistant to ampicillin which is commonly used compared to 30% with GBS.12 In addition, GNR meningitis was empirically for neonatal sepsis.25 Also, consistent with previous associated with a lower CSF/serum glucose ratio and higher levels studies, E. coli was found to be the most commonly isolated of protein. GNR.2,11,14,15 This study provides contemporary data evaluating

Journal of Perinatology A comparison of neonatal GNR and GPC meningitis PB Smith et al 114 mortality rates of meningitis caused by several types of bacteria, 7 Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA et al. including GBS. This information has implications in the Changes in pathogens causing early-onset sepsis in very-low-birth-weight continuing evaluation of the importance of maternal prophylaxis infants. N Engl J Med 2002; 347: 240–247. for GBS colonization. It is important to note, however, that of 8 Kanegaye JT, Soliemanzadeh P, Bradley JS. Lumbar puncture in pediatric those infants with meningitis diagnosed in the first 3 days of bacterial meningitis: defining the time interval for recovery of cerebrospinal life, only 27% had received antepartum antibiotics. This fluid pathogens after parenteral antibiotic pretreatment. Pediatrics 2001; suggests the need for further evaluation of current strategies 108: 1169–1174. 9 Riordan FA, Cant AJ. When to do a lumbar puncture. Arch Dis Child 2002; using antepartum treatment for prevention of early onset 87: 235–237. neonatal infection. 10 Garges HP, Moody MA, Cotten CM, Smith PB, Tiffany K, Lenfesty R et al. There are several limitations of this study. The data were Diagnosis of neonatal meningitis: the lumbar puncture, the blood culture collected for clinical documentation and were not collected with and CSF parameters. Pediatrics (in publication) 2005. a prospectively identified study question. Therefore, although 11 Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA et al. collected prospectively, the data contain many of the limitations To tap or not to tap: high likelihood of meningitis without sepsis among very of a retrospective study. In addition, due to privacy concerns low birth weight infants. Pediatrics 2004; 113: 1181–1186. and lack of informed consent, individual center information 12 Sarff LD, Platt LH, McCracken Jr GH. Cerebrospinal fluid evaluation in could not be analyzed. Care should be taken in interpreting the neonates: comparison of high-risk infants with and without meningitis. results demonstrating the association of antepartum antibiotics J Pediatr 1976; 88: 473–477. and GNR meningitis diagnosed in the first 3 days of life. This 13 Bonadio WA, Stanco L, Bruce R, Barry D, Smith D. Reference values of group of infants consisted of the cohort that had received LPs, not normal cerebrospinal fluid composition in infants ages 0 to 8 weeks. Pediatr Infect Dis J 1992; 11: 589–591. the entire cohort of infants exposed to antepartum antibiotics. 14 Mangi RJ, Quintiliani R, Andriole VT. Gram-negative bacillary meningitis. A number of infants with true meningitis were likely missed as only Am J Med 1975; 59: 829–836. infants with positive cultures were included in the analysis. 15 Unhanand M, Mustafa MM, McCracken Jr GH, Nelson JD. Gram-negative Strengths of the study include the large number of sample subjects enteric bacillary meningitis: a twenty-one-year experience. J Pediatr 1993; as well as the large number and diversity of NICUs. In addition, 122: 15–21. these data represent information obtained under recent and current 16 Sarman G, Moise AA, Edwards MS. Meningeal inflammation in neonatal care practices. Gram-negative bacteremia. Pediatr Infect Dis J 1995; 14: 701–704. In summary, compared to GPC meningitis, GNR meningitis was 17 Thorp JA, Jones PG, Peabody JL, Knox E, Clark RH. Effect of antenatal and associated with several differences in clinical presentation postnatal corticosteroid therapy on weight gain and head circumference (antepartum antibiotic exposure and later onset), and laboratory growth in the nursery. Obstet Gynecol 2002; 99: 109–115. parameters (higher CSF WBC and RBC counts). 18 Feigin RD, Cherry JD, Demmler GJ, Kaplan SL. Textbook of Pediatric Infectious Diseases. 5th edition. Philadelphia, PA: WB Saunder Co; 2004, pp 945–948. 19 Schwersenski J, McIntyre L, Bauer CR. Lumbar puncture frequency and References cerebrospinal fluid analysis in the neonate. Am J Dis Child 1991; 145: 54–58. 1 Klein JO, Feigin RD, McCracken Jr GH. Report of the task force on diagnosis 20 Mulder CJ, van Alphen L, Zanen HC. Neonatal meningitis caused by and management of meningitis. Pediatrics 1986; 78: 959–982. Escherichia coli in The Netherlands. J Infect Dis 1984; 150: 935–940. 2 Overall Jr JC. Neonatal bacterial meningitis. Analysis of predisposing factors 21 Wiswell TE, Baumgart S, Gannon CM, Spitzer AR. No lumbar puncture in and outcome compared with matched control subjects. J Pediatr 1970; 76: the evaluation for early neonatal sepsis: will meningitis be missed? 499–511. Pediatrics 1995; 95: 803–806. 3 Towers CV, Carr MH, Padilla G, Asrat T. Potential consequences of 22 Benjamin DK, DeLong E, Cotten CM, Garges HP, Steinbach WJ, Clark RH. widespread antepartal use of ampicillin. Am J Obstet Gynecol 1998; 179: Mortality following blood culture in premature infants: increased with 879–883. Gram-negative bacteremia and candidemia, but not Gram-positive 4 Byington CL, Rittichier KK, Bassett KE, Castillo H, Glasgow TS, Daly J et al. bacteremia. J Perinatol 2004; 24: 175–180. Serious bacterial infections in febrile infants younger than 90 days of age: 23 Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA et al. the importance of ampicillin-resistant pathogens. Pediatrics 2003; 111: Late-onset sepsis in very low birth weight neonates: the experience of the 964–968. NICHD Neonatal Research Network. Pediatrics 2002; 110: 285–291. 5 McDuffie Jr RS, McGregor JA, Gibbs RS. Adverse perinatal outcome and 24 Shah SS, Ehrenkranz RA, Gallagher PG. Increasing incidence of Gram- resistant Enterobacteriaceae after antibiotic usage for premature rupture of negative rod bacteremia in a newborn intensive care unit. Pediatr Infect Dis the membranes and group B streptococcus carriage. Obstet Gynecol 1993; J 1999; 18: 591–595. 82: 487–489. 25 Terrone DA, Rinehart BK, Einstein MH, Britt LB, Martin Jr JN, Perry KG. 6 Hervas JA, Ballesteros F, Alomar A, Gil J, Benedi VJ, Alberti S. Increase of Neonatal sepsis and death caused by resistant Escherichia coli: possible Enterobacter in neonatal sepsis: a twenty-two-year study. Pediatr Infect Dis consequences of extended maternal ampicillin administration. Am J Obstet J 2001; 20: 134–140. Gynecol 1999; 180: 1345–1348.

Journal of Perinatology