Health care-associated infections in neonatal intensive care units (NICUs) cause significant morbidity and mortality. 1 The epidemiology of nosocomial infections among neonates in NICUs from 99 hospitals during a 6-year period from October 1986 to September 1994 has been described from the National Nosocomial Infection Surveillance (NNIS) system of the CDC. 2 Bloodstream infection (BSI) and pneumonia were the most common nosocomial infections, and Gram-positive cocci (GPC) were the most common pathogens identified in that report.
There have been recent reports of increased Gram-negative infections in NICUs. 3–5 The National Institute of Child Health and Human Development Neonatal Research network reported that although Gram-positive organisms caused the majority of late onset (>3 days) sepsis in very low birth weight (VLBW) neonates (<1500 g), mortality was highest with Gram-negative sepsis. 6
We investigated the epidemiology of health care-associated infections in a Level III NICU at a free-standing children’s hospital.
MATERIALS AND METHODS
Description of the NICU
The NICU at Children’s National Medical Center is a Level III, 40-bed unit with an average of 450 annual admissions. Children’s National Medical Center is an urban, university-affiliated, multidisciplinary, regional referral center. All neonates are transferred from other facilities, because the hospital has no obstetric service. Hospital epidemiology personnel make daily rounds with the clinical team and collect data prospectively on all neonates. The hospital epidemiologist (NS) reviews all health care-associated infections on a regular basis.
All health care-associated infections from August 1996 to July 2001 were analyzed retrospectively. Neonates are monitored for nosocomial infections at all body sites, and data are collected prospectively according to standard surveillance protocols and nosocomial infection site definitions from the NNIS system. 7
Patient days, average length of stay, device days and device utilization rates were calculated with NNIS methods. 8 Infections were analyzed by year, by site, by pathogen and by birth weight category Neonates were classified into four birth weight categories: 1 to 1000 g; 1001 to 1500 g; 1501 to 2500 g; and >2500 g.
Health care-associated (nosocomial) infections were defined in accordance with the NNIS criteria. 8
NNIS nosocomial infection. A localized or systemic condition resulting from an adverse effect to the presence of an infectious agent(s) or its toxin(s) that meets the following criteria. It occurs in an NNIS patient admission, there is no evidence that it was present or incubating at the time of hospital admission unless the infection was related to a previous NNIS patient admission, and it meets the criteria for a specific site infection site as defined. 8
Infection rates. Infection rates were calculated per 1000 patient days or device days. For patients with BSI or pneumonia, device days for central venous catheters or ventilator days, respectively, were calculated.
Statistical analyses were performed using the chi square test for categoric variables and the z test for proportions to compare infection rates in different years.
During the study period (1996 to 2001), 665 pathogens caused 640 health care-associated infections in the NICU. Isolated organisms were considered pathogens only if in addition to a positive culture other site-specific criteria for a nosocomial infection were also satisfied. Overall Gram-negative rods (GNR) were more common than GPC. Ceftazidime resistance was present in 56 of 81 (69%) Enterobacter isolates. The distribution of pathogens is shown in Table 1.
The bloodstream and lower respiratory tract were the two most common sites of health care-associated infections. Forty-two percent (279 of 665) of the isolates were identified from the bloodstream, and 22.5% (150 of 665) were identified from the lower respiratory tract. Coagulase-negative staphylococci were the most common cause of BSI (104 of 279, 37%), followed by Candida spp. (63 of 279, 23%) and Enterobacteriaceae (52 of 279, 19%). GNR were the most common pathogens isolated from respiratory tract infections (84 of 150, 56%). Other body sites cultured included the urine, surgical site, skin, eye and central nervous system. The distribution of pathogens by body site is shown in Table 2.
The total numbers of pathogens isolated from health care-associated infections per year during the 5-year period were comparable except for an increase in the year 1998 to 1999 (Fig. 1). No epidemics were identified during this time period. The infection rate per 1000 patient days for all birth weight categories combined was 15.42 in 1998 to 1999. This was significantly higher compared with 11.44 in 1996 to 1997 (P = 0.01). In 2000 to 2001 the infection rate was significantly lower at 9.97 (P < 0.01).
About 60% of all pathogens (388 of 665) and GNR (173 of 284) were identified from neonates weighing ≤1000 g (P = 0.2). A bimodal distribution of infections was seen with a peak in neonates weighing ≤1000 g and a second peak in those weighing >2500 g.
The important findings in our study are that GNR were the predominant pathogens causing health care-associated infections in a Level III NICU. Only during the first year (1996 to 1997) of the study were GPC more common than GNR.
In the NNIS report (1986 to 1994), ∼18% of all isolates were GNR and >60% were GPC. 2 The National Institute of Child Health and Human Development Neonatal Research network also reported that late onset sepsis in very low birth weight neonates was predominantly caused by GPC. 6 Seventy-five percent of pathogens were Gram-positive cocci and 18% were GNR in this study conducted from 1991 to 1993, overlapping the time period in which the NNIS study was done. The predominance of GNR in our study could be a reflection of the different time periods in which the studies were conducted.
A national point prevalence survey of nosocomial infections in 29 Pediatric Prevention Network NICUs was performed in August 1999. 9 Thirty-two percent of all pathogens isolated in this prevalence survey were Gram-negative, which is higher than the earlier NNIS report. This survey represents the epidemiology of health care-associated infections at a point in time and will not be able to assess a change in pathogen prevalence over time. Our report provides a comprehensive picture during a 5-year period. The point prevalence survey included only 117 health care-associated infections compared with 640 infections in our longitudinal study.
Semiannual reports published by the CDC provide information on pooled means and percentiles of the distribution of device-associated infection rates by birth weight category, pooled means and percentiles of the distribution of device utilization ratios and prevalence of antimicrobial resistance among selected nosocomial isolates in the NNIS system. 10 Summary data published recently by NNIS also provide information on risk-adjusted infection rates but not on pathogen distribution. 11
A cyclic pattern of pathogen distribution in hospitals has been described. 12 GPC were the predominant pathogens in the earlier part of the last century followed by Gram-negative organisms in the 1970s. In the late 1980s and early 1990s, GPC, especially methicillin-resistant Staphylococcus aureus and vancomycin-resistant enterococci, reemerged. The European Organization for Research and Treatment of Cancer trials performed between 1973 and 1994 demonstrated a change in the type of pathogens isolated from oncology patients with febrile neutropenia. 13 Gram-negative pathogens were the dominant bloodstream isolates from 1973 to 1986, followed by Gram-positive pathogens in subsequent years. This was possibly related to the widespread use of ceftazidime and ceftriaxone, which were introduced in 1985. 14 Increasing antimicrobial resistance among Gram-negative bacteria may thus be contributing to our observed reemergence of GNR as a dominant pathogen.
Group B streptococci comprised only ∼5% of isolates in the NNIS report, so it is unlikely that the reduction in GPC seen in our NICU is secondary to a reduction in the incidence of group B streptococcal sepsis. Association between maternal ampicillin use and neonatal sepsis caused by ampicillin-resistant Escherichia coli sepsis is unclear. 15–19 We did not assess maternal antibiotic use; thus its contribution to the increase in GNR isolates cannot be delineated.
Bloodstream and lower respiratory tract were the two most common sites of infections in our NICU, similar to prior national reports. 2, 9 However, the proportion of GNR bacteremia in our NICU was higher. The increasing incidence of GNR bacteremia in a NICU has also been reported by others. 3, 4 A recent study highlighted the role of GNR in late onset fulminant sepsis in a NICU. 20 Thus the increasing incidence of GNR bacteremia can have a bearing on the empiric treatment of clinical sepsis in high risk neonates, which should include adequate antimicrobial coverage for GNR.
The patient population in NICUs has changed during the last few years, with a significant improvement in the survival of preterm and VLBW infants. In the NNIS report 44.5% (5868 of 13 179) of infections occurred in infants ≤1500 g, and in our report 69% (444 of 640) of infections were in infants ≤1500 g, of which 82% (364 of 444) were in neonates ≤1000 g. In the NNIS report, the 1- to 1000-g birth weight category was included only in 1992; therefore the exact distribution before that is unavailable. Distribution by birth weight category in our NICU is similar to that in the point prevalence survey. 9 It is thus possible that the higher number of GNR infections in our NICU could be a reflection of a change in the patient population in recent years.
Preterm neonates and VLBW neonates are more likely to have invasive therapeutic interventions and to be exposed to broad spectrum antimicrobials. Both these factors are associated with a higher incidence of GNR infections. 21
We conclude that there has been a change in the epidemiology of health care-associated infections in the NICU at our institution. Given other recent reports of increase in GNR infections, this likely reflects a widespread change rather than being an institution-specific phenomenon. This observation can have an impact on the choice of antimicrobials in the empiric treatment of health care-associated infections in neonates.
We thank Kantilal M. Patel, Ph.D., for statistical assistance and Marie-Michèle Léger, PA-C., M.P.H. and Dorleen Brown for surveillance data collection and entry.
1. Jarvis WR. Epidemiology of nosocomial infections in pediatric patients. Pediatr Infect Dis J 1987; 6: 344–51.
2. Gaynes RP, Edwards JR, Jarvis WR, et al. Nosocomial infections among neonates in high-risk nurseries in the United States. Pediatrics 1996; 98: 357–61.
3. Hervas JA, Ballesteros F, Alomar A, et al. Increase of Enterobacter
in neonatal sepsis: a twenty-two-year study. Pediatr Infect Dis J 2001; 20: 134–40.
4. Shah SS, Ehrenkranz RA, Gallagher PG. Increasing incidence of Gram-negative rod bacteremia in a newborn intensive care unit. Pediatr Infect Dis J 1999; 18: 591–5.
5. Yu JL, Wu SX, Jia HQ. Study on antimicrobial susceptibility of bacteria causing neonatal infections: a 12 year study(19:87–1998). Singapore Med J 2001; 42: 107–10.
6. Stoll BJ, Gordon T, Korones SB, et al. Late-onset sepsis in very low birth weight neonates: a report from the National Institute of Child Health and Human Development Neonatal Research Network. J Pediatr 1996; 129: 63–71.
7. Emori TJ, Culver DH, Horan TC, et al. National Nosocomial Infections Surveillance System (NNIS): description of surveillance methods. Am J Infect Control 1991; 19: 19–35.
8. NNIS Manual. US Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention. May 1999;V-1.
9. Sohn AH, Garrett DO, Sinkowitz-Cochran RL, et al. Prevalence of nosocomial infections in neonatal intensive care unit patients: results from the first national point-prevalence survey. J Pediatr 2001; 139: 821–7.
10. Semiannual Report Aggregated Data from the National Nosocomial Infections Surveillance (NNIS) System December 2000 (accessed January 2002). http://www.cdc.gov/ncidod/hip/surveill/nnis.htm.
11. National Nosocomial Infections Surveillance (NNIS)System Report, Data Summary from January 1992 to June 2001 (accessed January 2002). http://www.cdc.gov/ncidod/hip/NNIS/members/2001NNIS_report.pdf.
12. Weinstein RA. Nosocomial infection update. Emerg Infect Dis 1998; 4: 416–20.
13. Klastersky J. Science and pragmatism in the treatment and prevention of neutropenic infection. J Antimicrob Chemother 1998; 41 (Suppl D): 13–24.
14. Jones RN. Contemporary antimicrobial susceptibility patterns of bacterial pathogens commonly associated with febrile patients with neutropenia. Clin Infect Dis 1999; 29: 495–502.
15. Joseph TA, Pyati SP, Jacobs N. Neonatal early-onset Escherichia coli
disease. Arch Pediatr Adolesc Med 1998; 152: 35–40.
16. Towers CV, Carr MH, Padilla G, Asrat T. Potential consequences of widespread antepartal use of ampicillin. Am J Obstet Gynecol 1998; 179: 879–83.
17. Mercer BM, Carr TL, Beazley DD, et al. Antibiotic use in pregnancy and drug-resistant infant sepsis. Am J Obstet Gynecol 1999; 181: 816–21.
18. Baltimore RS, Huie SM, Meek JI, et al. Early-onset neonatal sepsis in the era of group B streptococcal prevention. Pediatrics 2001; 108: 1094–8.
19. Chen KT, Tuomala RE, Cohen AP, et al. No increase in rates of early-onset neonatal sepsis by non-group B streptococcus or ampicillin-resistant organisms. Am J Obstet Gynecol 2001; 185: 854–8.
20. Karlowicz GM, Buescher ES, Surka AE. Fulminant late-onset sepsis in a neonatal intensive care unit, 1988–1997, and the impact of avoiding empiric vancomycin therapy. Pediatrics 2000; 106: 1387–90.
21. Waterer GW, Wunderink RG. Increasing threat of Gram-negative bacteria. Crit Care Med 2001;29(Suppl.):N75–81.