We performed a case-control study of neonatal polymicrobial infections in a tertiary hospital in North America, and identified polymicrobial bloodstream infections in nearly 14% of bloodstream infections. We also probed the electronic neonatal clinical database over a four year period from January 2009 to December 2012 for relevant clinical data including risk factors and outcomes. We observed that polymicrobial bloodstream infections were associated with an increased mortality (>3-fold) and increased duration of infection compared to monomicrobial bloodstream infections. Surgical intervention excluding NEC surgery was a significant risk factor for neonatal polymicrobial infections.
The annual frequency of polymicrobial bloodstream infections in our study was approximately 14% over a 16 year study period. The frequency of neonatal polymicrobial infections reported in literature is variable, ranging from 3.9 to 25% [10–12, 17, 19–21]. Contaminated multi-dose lipid emulsions were responsible for the high frequency (25%) of polymicrobial infections in one study . In most studies that report neonatal infections, the isolation of multiple organisms is often not discussed. The higher frequency of polymicrobial infections in our study compared to other reports in literature may be due to two reasons. First is the varying definition of polymicrobial infections in the different studies. We defined polymicrobial bloodstream infections as multiple organisms isolated during an infectious episode including those from a single blood specimen similar to Sutter et al., as we believe that this is a true reflection of a polymicrobial infection . Other studies defined polymicrobial infections as multiple pathogens isolated from a single blood specimen, which may account for the lower incidence of polymicrobial infections [11, 12, 17]. Using the latter definition the annual frequency of polymicrobial bloodstream infections in our study would be approximately 7.8%, comparable to the 10% frequency reported in North American by Bizzarro et al. . Secondly, increased incidence of polymicrobial bloodstream infections in our study may also relate to the large percentage of medically complex infants referred to our NICU including infants requiring surgical interventions, complex congenital heart disease, ECMO, abdominal wall defects, short gut syndrome and other congenital anomalies. This complex group of infants often requires indwelling vascular catheters for parenteral nutrition or medications for extended periods of time. The annual frequency of polymicrobial bacteremia in the neonatal intensive care unit in our study varied from 6.3 to 18.6% of all bacteremia during the 16-year study period. This annual variation may be in part due to expansion of the NICU, changes in the case-mix and referral patterns, nursing policy changes or implementation of catheter care bundles. There were significant decrease in the number of infections and polymicrobial infections between the time epochs, 1998-2009 and 2010-2012 (p < 0.01). We did not note any significant differences in the number of admissions of VLBW and ELBW infants. A dedicated vascular access team was formed in June 2009 followed by better implementation of catheter care bundles in June 2009, which along with increased compliance with hand hygiene and other infection control measures may have contributed to the decrease in infections.
In our study, cases and controls had similar demographics, a mean gestational age of approximately 30 weeks, birth weight of approximately 1500 g, similar male to female ratio and similar percentage being inborn. We focused on risk factors reported for polymicrobial infections in existing literature. The average age at the onset of the infections was 41 days in controls and 38 days in cases and hence all infections were in the late onset sepsis category and mostly occurred beyond the first 28 days of life. Almost all of the infections were late-onset infections and hence we did not collect data on risk factors for early onset sepsis such as maternal prolonged rupture of membranes, group B streptococcus colonization, urinary tract infection or chorioamnionitis. Subgroups of infections in neonates less than 28 days of age or those that were admitted from home were too small for meaningful comparisons. We noted a significant association of polymicrobial infections with surgery other than NEC and a trend towards association with direct jaundice. Cases had a higher incidence of central venous catheter compared to controls (97 vs. 91%, p = 0.20) but this association was not statistically significant.
In our case-control study, we observed more than 3-fold increase in associated mortality in neonates with polymicrobial bacteremia, but the retrospective and database oriented nature of our study precludes any conclusions of causality. Mortality due to polymicrobial infections has been reported to be at least 2-fold more than that of monomicrobial infections both in adults and children [1, 2]. Faix et al. reported that mortality due to polymicrobial infections in neonates is increased almost 3-fold in his study of 15 cases of polymicrobial infections in 1971-86 . More recent studies by Bizzarro et al. (study period 1989-2006), Tsai et al (study period 2004 to 2011) and Gupta et al. (1 year period in early 2000) did not report an increase in attributable mortality to polymicrobial infections in their studies [11, 12, 17]. The variability in mortality rates reported by different studies on polymicrobial infections may be multifactorial including differences in study design, variable definitions of polymicrobial infections, patient population, study periods, virulence of the organisms isolated or other unidentified factors . The mechanisms for increased mortality in polymicrobial infections are not clear. Similar increases in mortality are observed in animal models of systemic and local polymicrobial infections [22–25]. We have observed increased catheter infection and systemic dissemination in a polymicrobial biofilm catheter infection model compared to monomicrobial infection . The increased mortality may also arise from inherent host vulnerability that causes the infection in the first place (e.g. prematurity, short gut) or facilitation of one infection by the other (synergism) . Polymicrobial interactions in polymicrobial environments such as a mixed-species biofilms on catheters may induce a host of synergistic mechanisms including quorum sensing, induction of virulence among others that may contribute to enhanced mortality or morbidity of the host .
We also noted an increase in the duration of infection in cases compared to controls, which may be due to increased host susceptibility or a synergistic effect of the polymicrobial infection. We did not observe any differences in length of hospital stay or other clinically relevant neonatal outcomes such as BPD, IPPV or CPAP at 36 corrected weeks, HFOV, ECMO, ROP, NEC or IVH or PVL. We observed an increase in the use of INO therapy in cases compared to controls, which was not statistically significant.
The organisms isolated in our study were mostly similar in both monomicrobial and polymicrobial bloodstream infections, similar in antibiotic susceptibility patterns and comparable to those organisms from polymicrobial infections reported in literature. The most common organisms isolated in both monomicrobial and polymicrobial infections were CONS, Staphylococcus aureus, Escherichia coli, Enterococcus species, Klebsiella spp. and Candida spp. Infections due to Candida spp., Enterococcus species and Klebsiella spp. increased in frequency in cases compared to controls. The most frequent polymicrobial combinations were CONS with Candida spp. and other organisms with Enterococcus faecalis, similar to reported literature [1, 14, 27]. Studies from Asia, by Tsai et al. and Gupta et al. report a high incidence of Gram negative infections (approximately 60%) in the polymicrobial group [12, 17]. In the North American study by Bizarro et al., polymicrobial infections were preponderantly caused by Gram positive organisms (77%) similar to our study . Geographic variations in organisms isolated from neonatal polymicrobial infections and their antibiotic susceptibility patterns may partly explain the variation in mortality and morbidity across the world due to polymicrobial infections. In our study we did not discern any differences in the treatment regimens used to treat polymicrobial infections and monomicrobial infections that could explain the differences in mortality. Polymicrobial infections were treated adequately for all the organisms isolated and duration of therapy was consistent with our written neonatal guidelines.
Limitations of our study include being a retrospective, observational, case-control study. Bloodstream episodes were identified from the clinical microbiology database and hence the clinical data associated with the infection or their severity were not available. The clinical risk factors and outcomes were identified from the clinical database, which was available only for a 4 year period. Being a retrospective study, it is difficult to assign a causal relationship to the increased mortality associated with polymicrobial infections in our study. We did not assess long term developmental or growth outcomes as longitudinal assessment data were not available.
The human microbiome project and other microbiome studies emphasize the polymicrobial nature of organism communities in the human body [28, 29]. The availability of molecular culture-independent methods for detection of sepsis may increase the identification of polymicrobial infections . Guidelines or recommendations for therapeutic and preventive strategies against polymicrobial infections do not exist. Prolonged therapy with antibiotics targeting all organisms isolated in a polymicrobial infection and removing infected catheters remain the mainstay in therapy. Defining the epidemiology and clinical impact of polymicrobial infections may be the first step towards delineating optimal therapy and clinical outcomes. Preventive strategies should emphasize catheter care bundles that decrease central line associated bloodstream infections. Focused research is necessary to prevent and treat polymicrobial infections to improve clinical outcomes in the vulnerable neonate.