Outbreak of Multidrug‐Resistant Serratia marcescens Infection in a Neonatal Intensive Care Unit
Background. Serratia marcescens causes healthcare‐associated infections and significant morbidity and mortality in neonatal intensive care units (NICUs). We report the investigation and control of an outbreak of multidrug‐resistant (MDR) S. marcescens infection at an NICU.
Methods. An outbreak investigation and a case‐control study were undertaken at a 36‐bed NICU in a tertiary care hospital in Baltimore, Maryland, for the period from October 2004 through February 2005. The outbreak investigation included case identification, review of medical records, environmental cultures, patient surveillance cultures, personnel hand cultures, and pulsed‐field gel electrophoresis (PFGE). The case‐control study included case identification and review of medical records. Infection control measures were implemented. Eighteen NICU neonates had cultures that grew MDR S. marcescens during the study period. The case‐control study included 16 patients with the outbreak strain or an unidentified strain of MDR S. marcescens and 32 control patients not infected and/or colonized with MDR S. marcescens, treated in the NICU for at least 48 hours during the study period.
Results. PFGE analysis identified a single strain of MDR S. marcescens that infected or colonized 15 patients. Two patients had unique strains, and 1 patient’s isolate could not be subtyped. An unrelated MDR S. marcescens isolate was recovered from a sink drain. Exposure to inhalational therapy was an independent risk factor for MDR S. marcescens acquisition after adjusting for birth weight. Extensive investigation failed to reveal a point source for the outbreak.
Conclusion. A single epidemic strain of MDR S. marcescens spread rapidly and threatened to become endemic in this NICU. Transient carriage on the hands of healthcare personnel or on respiratory care equipment was the likely mode of transmission. Cohorting patients and staff, at the cost of bed closures and additional personnel, interrupted transmission and halted the outbreak.
Received November 7, 2007; accepted March 3, 2008; electronically published March 28, 2008.
Serratia marcescens is a gram‐negative bacillus that causes pneumonia, bloodstream infections, central nervous system infections, urinary tract infections, and conjunctivitis. The organism has been responsible for outbreaks in neonatal intensive care units (NICUs), where it is associated with significant morbidity and mortality, especially in neonates of low birth weight.1‐12 Reported sources of these outbreaks include contaminated breast pumps, breast milk, soap, disinfectant, laryngoscope blades, and air conditioning ducts.1,2,4,13,14 The organism can survive on the skin for extended periods of time, and transient hand carriage is thought to be a mode of transmission when no source is identified.10,15 Multidrug‐resistant (MDR) strains of S. marcescens are reported to cause more invasive infections and to demonstrate a greater propensity for transmission.15,16
In November 2004, the Department of Hospital Epidemiology and Infection Control at the Johns Hopkins Hospital (JHH) detected a cluster of 6 neonates in the NICU whose cultures grew MDR S. marcescens. This MDR phenotype was not detected in the NICU prior to October 2004. Additional cases were subsequently detected, and the number of neonates infected or colonized with S. marcescens during the outbreak increased markedly, compared with the preceding 21 months (Figure 1). We describe an investigation and case‐control study to determine the cause of the outbreak of MDR S. marcescens infection and the implementation of infection control measures to interrupt transmission.
Figure 1. Bar graph showing the incidence of Serratia marcescens infection in the neonatal intensive care unit at the Johns Hopkins Hospital from January 2003 through August 2005. The outbreak period was from October 2004 through February 2005, when a new multidrug‐resistant phenotype and a marked increase in the number of cases were noted. The patterns of the bars indicate various strains of S. marcescens, as defined by pulsed‐field gel electrophoresis.
Methods
Setting
The JHH is a 926‐bed tertiary care hospital, including a level III, 36‐bed NICU, in Baltimore, Maryland. The NICU consists of 3 rooms (pods 1, 2, and 3) that surround a common storage and medication dispensing area (Figure 2). Each pod accommodates 10 adjacent patient beds (with approximately 1 meter of space between beds) arranged concentrically around a central charting area. Three smaller isolation rooms accommodate 2 patient beds each (Figure 2).
Figure 2. Floor plan of the neonatal intensive care unit at the Johns Hopkins Hospital. Three 10‐bed rooms, designated pods 1, 2, and 3, surround a central medication area. Also shown are 3 rooms used for isolation precautions and an adjacent area for charting and office space.
At the time of the outbreak, Pseudomonas had contaminated the sink drains in the NICU, and so there was a restriction on water usage. For hand hygiene, the NICU staff used an alcohol‐based hand sanitizer, which was located in the central charting area of each pod. If the hands of staff members were visibly soiled, they washed their hands with soap and water and then applied an alcohol‐based surgical scrub. Bottled water was used to bathe the patients.
Epidemiologic Investigation
Microbiology records were reviewed to identify cases of infection and to determine the baseline rate of S. marcescens infection in the NICU prior to the outbreak. Case patients were defined as any neonate hospitalized in the NICU from October 2004 through February 2005 who had 1 or more clinical or surveillance cultures that yielded MDR S. marcescens. Any isolate that was susceptible to no more than 3 classes of antimicrobial agents was defined as MDR S. marcescens. Medical records were reviewed to assess potential risk factors, including younger gestational age; low birth weight; mode of delivery (vaginal vs cesarean); location of delivery (born at JHH vs born at another hospital and then transferred to JHH); low Apgar score; previous surgery; underlying diseases; previous mechanical ventilation, central venous catheterization, arterial catheterization, nasogastric intubation, or Foley catheterization; greater length of stay in the NICU; exposure to parenteral nutrition, inhalational medication therapy, head ultrasound, echocardiography, or eye examination; and ingestion of breast milk or infant formula.
NICU cultures were performed of the environment, including counters, sinks, bottle warmers, diaper scales, personnel work areas, breast milk, equipment in the breast milk room, infant formula, inhalational medications, saline ampules, incubators, eye examination equipment, radiology equipment, air ducts, soap, lotion, ventilators, oxygenation and humidification devices, and other respiratory care equipment. Each member of the NICU staff was examined for the presence of artificial fingernails, dermatitis, or other skin lesions on the hands, and hand cultures were obtained.
Microbiologic Methods
Clinical cultures were processed using standard practices on routine media, and susceptibility testing was performed with the agar dilution method. Hand cultures from healthcare personnel were obtained by rinsing the hands in brain‐heart infusion liquid media and plating it on 5% sheep blood and Macconkey agars. Environmental samples were cultured on brain‐heart infusion plates supplemented with 5% sheep blood and gentamicin at a concentration of 10 μg/mL. Pulsed‐field gel electrophoresis (PFGE) was performed by using the GenePath system (Bio‐Rad) on all available isolates. Bacterial DNA was digested with Spe1 using run parameters of 5‐35 seconds for 24 hours. The gels were analyzed with Molecular Analyst Fingerprinting Plus software (Bio‐Rad). Isolates were considered genetically related if their PFGE patterns differed by 3 bands or fewer.
Case‐Control Study
All neonates hospitalized in the NICU who had at least 1 culture that grew the outbreak strain or the unidentified strain of MDR S. marcescens between October 15, 2004, and February 28, 2005, were included as case patients (
) in the case‐control study. Two neonates infected or colonized with unique strains of MDR S. marcescens, as determined using PFGE analysis, were excluded from the study. Control patients (
) were randomly selected from a list of neonates not infected or colonized with MDR S. marcescens who were treated in the NICU for at least 48 hours during the study period. The ratio of case patients to control patients was 1:2, and the patients were not matched on any variables. Risk factor data were collected for the period from October 15, 2004, to February 28, 2005, for control patients and for the period from October 15, 2004, to the date of first recovered MDR S. marcescens isolate for case patients. The study protocol was reviewed and approved by the institutional review board of Johns Hopkins University.
Statistical Analysis
Categorical variables were analyzed with a 2‐sided Fisher exact test. Continuous variables were analyzed with the Student t test. All tests were 2‐sided. A multivariate logistic regression model was constructed using forward selection. Variables with a P value of less than .1 in the univariate analysis were candidates for the multivariable analysis, and a significance level of .05 was the criterion for remaining in the model. Data analyses were performed using Stata, version 8 (Stata).
Results
Epidemiologic Investigation
A total of 18 patients were identified as infected and/or colonized with MDR S. marcescens between October 2004 and February 2005 (Figure 1). The baseline rate of S. marcescens infection for the 21 months prior to the outbreak was 0.8 cases/month, compared with 3.8 cases/month during the outbreak (Figure 1). The organism grew in the endotracheal tube aspirate cultures of 8 patients and in the urine cultures of 6 patients. In 7 infected patients, MDR S. marcescens was detected in the blood, conjunctiva, peritoneal fluid, and/or periumbilical abscess fluid. In 4 colonized patients, MDR S. marcescens was detected in only the perianal swab samples for surveillance culture. Eight patients’ cultures from more than 1 body site grew MDR S. marcescens. Two case patients died during NICU hospitalization. These deaths were not attributed to MDR S. marcescens infection.
PFGE analysis revealed that 15 (83%) of the 18 patients were infected or colonized with a single strain of MDR S. marcescens (Figure 3). Two patients had unique, unrelated strains, and 1 patient’s isolate was not available for analysis (ie, the unidentified strain). The outbreak strain and the unidentified strain had similar antimicrobial susceptibility profiles; both were resistant to ticarcillin, piperacillin, ceftriaxone, cefepime, gentamicin, and tobramycin and susceptible to amikacin and gatifloxacin. Most of the isolates recovered were susceptible to piperacillin‐tazobactam and cefotetan, although isolates recovered from the samples of 4 patients from the later part of the outbreak were resistant to these agents. The 2 unique strains of MDR S. marcescens had distinct antimicrobial susceptibility profiles and were susceptible to more antimicrobial agents than the outbreak strain.
Figure 3. Pulsed‐field gel electrophoresis patterns of 8 isolates of multidrug‐resistant Serratia marcescens recovered from 8 patients during the neonatal intensive care unit outbreak. An identical strain was isolated from blood, urine, sputum, conjunctiva, abscess fluid, or perianal surveillance culture samples from 15 patients, 8 of which are represented in the figure. ETS, endotracheal suction.
The majority of case patients were hospitalized in pod 1 of the NICU at the time of MDR S. marcescens acquisition, and none of the case patients spent more than 24 hours in pod 3 before acquiring the organism (Figure 2).
Examination of the hands of NICU personnel did not reveal any dermatitis, lesions, or artificial fingernails. Hand cultures obtained from NICU personnel did not grow S. marcescens. One environmental culture sample from a sink drain in pod 1 yielded a strain of S. marcescens that was unrelated to the patient isolates analyzed with PFGE. Two environmental culture samples from a NICU oxygenation and humidification system yielded Ralstonia pickettii.
Case‐Control Study
Sixteen patients met the case definition for the case‐control study. Two patients whose cultures grew unique strains of MDR S. marcescens were excluded. The group of control patients was similar to the group of case patients with respect to sex, race, length of NICU exposure, and gestational age (Table). The mean birth weight was lower for case patient than for control patients (1,276 g vs 1,848 g; 
), although this trend did not reach statistical significance (Table).
In univariate analysis, exposure to inhalation therapy and the presence of an arterial line were significantly associated with MDR S. marcescens infection and/or colonization (Table). In a multivariate model, exposure to inhalational medication therapy was an independent risk factor for acquisition of MDR S. marcescens after adjusting for birth weight and the presence of an arterial line (odds ratio [OR], 4.8 [95% confidence interval, 1.2‐20]; 
).
Termination of the Outbreak
Education about Serratia and standard infection control procedures were reviewed with NICU personnel. Additional dispensers of alcohol‐based hand sanitizer were installed throughout the NICU, and individual bottles were distributed to NICU staff. Cleaning and disinfection procedures were reviewed and reinforced with environmental services personnel. The fact that culture samples from the oxygenation and humidification system yielded R. pickettii was reported to the Food and Drug Administration, and the system's use was discontinued.
Contact isolation precautions were implemented for all neonates with MDR S. marcescens infection. NICU personnel were instructed to gather the neonates infected with MDR S. marcescens into one area of the unit and to dedicate staff to care for only these patients. Initially, this was not implemented because of the severity of some patients' illness, which raised concerns about the staff‐to‐patient ratio and the space available to accommodate the necessary medical equipment, such as extracorporal membrane oxygenation machines, in the cohort area. When incidents of MDR S. marcescens infection continued in January 2005, the Department of Hospital Epidemiology and Infection Control mandated that all infected or colonized patients be cohorted, and compliance with infection control procedures was reinforced.
Changes to staff assignments, addition of more staff, and closure of beds were all required to accomplish cohorting without compromising patient care. The NICU staff assignments were adjusted so that nursing staff and respiratory therapists were cohorted to care solely for either affected or unaffected neonates. Prior to cohorting, NICU nurses routinely cared for 2 patients or more with various levels of acuity. During cohorting, many neonates infected or colonized with MDR S. marcescens required a nurse‐to‐patient ratio of 1:1 because these patients tended to be severely ill, and it was often deemed too much for 1 nurse to care for 2 case patients. More respiratory therapists were added to accommodate the cohorting. NICU beds were closed to new admissions because of these staffing issues, to avoid placing patients without MDR S. marcescens infection in the cohort area and to accommodate necessary large equipment such as extracorporal membrane oxygenation machines in the cohort area. Physician and nurse practitioner teams limited the number of people entering the cohort area and, whenever possible, examined neonates who were not affected with MDR S. marcescens before examining those who were affected.
Weekly surveillance cultures of stool and endotracheal tube aspirate were performed to detect whether there was any asymptomatic colonization of neonates with MDR S. marcescens. Neonates who were colonized were placed in contact isolation and cohorted with case patients. These infection control measures successfully halted the outbreak, and the final case of MDR S. marcescens infection with the outbreak strain occurred on February 5, 2005. Isolation precautions, cohorting of the patients and staff, and surveillance cultures continued until the last case patient was discharged from the hospital, on April 22, 2005.
Discussion
We report the investigation and termination of a clonal outbreak of MDR S. marcescens infection in an NICU. A single epidemic strain of MDR S. marcescens spread rapidly and threatened to become endemic in this NICU. At the height of the outbreak, one‐third of the neonates hospitalized in the NICU were infected or colonized with MDR S. marcescens. The outbreak strain was resistant to most commonly prescribed antimicrobial agents. Treatment options for infected patients were limited by both the organism’s antimicrobial resistance profile and the contraindication for the use of certain agents such as fluoroquinolones in pediatric patients. Furthermore, antimicrobial resistance appeared to increase during the course of the outbreak, with later MDR S. marcescens isolates showing resistance to pipercillin‐tazobactam and cefotetan. Epidemic spread, such as that seen in this outbreak, is reported to occur more often with antimicrobial‐resistant strains of S. marcescens than with antimicrobial‐susceptible strains of the organism.15,16
An extensive epidemiologic investigation, including environmental cultures and cultures of healthcare worker hands, did not reveal the source of the organism. The results of the case‐control study showed that low birth weight, the presence of an arterial line, and receipt of inhalation medication therapy were associated with acquisition of MDR S. marcescens. Other investigators have previously found that low birth weight is a risk factor for S. marcescens infection among neonates.6 The prescence of an arterial line in a patient was probably not clinically related to this outbreak, because MDR S. marcescens grew primarily in endotracheal tube aspirate cultures and grew in blood samples from only 3 case patients. In this study, after controlling for low birth weight and the presence of an arterial line in a multivariable model, receipt of inhalation medication therapy remained an independent risk factor for acquiring MDR S. marcescens. Given this finding and the fact that the organism grew in endotracheal tube aspirate cultures of many of the neonates, we were concerned that contaminated nebulized medications or respiratory care equipment were sources or modes of transmission. However, each patient had his or her own inhalant dispenser, and culture samples of inhaled medications and of respiratory care equipment did not yield S. marcescens. It is possible that the nebulized medications, the respiratory care equipment, or both were the source of S. marcescens and that our cultures failed to detect the organism. During the investigation, R. pickettii was recovered from culture samples of a Vapotherm gas humidification system. Use of this type of system was discontinued, and the product was later recalled by the manufacturer and the Food and Drug Administration because of additional reports of Ralstonia contamination.17 Because Serratia was not recovered from the culture samples, there was no evidence that this system contributed to this outbreak.
Although the presence of an arterial line and the receipt of inhalational therapy were statistically significant risk factors, the 95% confidence intervals were wide, and the case‐control study and the multivariable analysis were limited by a small sample size. It is possible that both of these risk factors were in fact markers for more severely ill and more susceptible neonates. Other studies have found contamination of breast milk responsible for NICU outbreaks of S. marcescens infection. Our environmental cultures and our case‐control study did not implicate breast milk, formula, or nutritional supplements. These variables were difficult to assess in the risk factor analysis because most neonates received multiple sources of nutrition.
The location of the patients within the NICU apparently played a role in the transmission of MDR S. marcescens, because the majority of case patients acquired the organism in pod 1, and none spent significant time in pod 3. This implies that there may have been an undetected environmental source or vector of transmission in pod 1. It is also possible that, in pod 1, transmission from source patients was due to crowding, insufficient space between beds, and lapses in hand hygiene. The literature shows that overcrowding in NICUs is associated with a higher risk of transmission of healthcare‐associated infections.18,19
Although in our outbreak—as in many reported NICU outbreaks of S. marcescens infection—no point source was identified, implementation of standard infection control interventions and cohorting interrupted transmission.6 We believe that transient carriage on the hands of healthcare personnel or on respiratory care equipment was the most likely mode of transmission. Initially, cohorting of patients and staff was not completely implemented because it was difficult to overcome logistical problems, balance patient acuity with staff assignments, and accommodate large equipment in the cohort area. Once cohorting of patients and staff was completely implemented, at the cost of bed closure and additional staffing requirements, transmission was interrupted and the outbreak was halted.
Acknowledgments
Financial support. L.L.M. received grant support from the Research Scientist Development Award, grant number 5 K01 CI000300, from the Centers for Disease Control and Prevention. Additional support for the outbreak investigation and interventions was provided by the JHH.
Potential conflict of interests. T.M.P. serves on the advisory boards for Pfizer, Replidyne, and 3M and received recent research funding from 3M and Sage, as well as a speaker honorarium from Ortho McNeil.
References
- 1. Uduman SA, Farrukh AS, Nath KN, et al. An outbreak of Serratia marcescens infection in a special‐care baby unit of a community hospital in United Arab Emirates: the importance of the air conditioner duct as a nosocomial reservoir. J Hosp Infect 2002; 52:175‐180.
- 2. Cullen MM, Trail A, Robinson M, Keaney M, Chadwick PR. Serratia marcescens outbreak in a neonatal intensive care unit prompting review of decontamination of laryngoscopes. J Hosp Infect 2005; 59:68‐70.
- 3. Jang TN, Fung CP, Yang TL, Shen SH, Huang CS, Lee SH. Use of pulsed‐field gel electrophoresis to investigate an outbreak of Serratia marcescens infection in a neonatal intensive care unit. J Hosp Infect 2001; 48:13‐19.
- 4. Jones BL, Gorman LJ, Simpson J, et al. An outbreak of Serratia marcescens in two neonatal intensive care units. J Hosp Infect 2000; 46:314‐319.
- 5. Lai KK, Baker SP, Fontecchio SA. Rapid eradication of a cluster of Serratia marcescens in a neonatal intensive care unit: use of epidemiologic chromosome profiling by pulsed‐field gel electrophoresis. Infect Control Hosp Epidemiol 2004; 25:730‐734.
- 6. Sarvikivi E, Lyytikainen O, Salmenlinna S, et al. Clustering of Serratia marcescens infections in a neonatal intensive care unit. Infect Control Hosp Epidemiol 2004; 25:723‐729.
- 7. Steppberger K, Walter S, Claros MC, et al. Nosocomial neonatal outbreak of Serratia marcescens—analysis of pathogens by pulsed field gel electrophoresis and polymerase chain reaction. Infection 2002; 30:277‐281.
- 8. Villari P, Crispino M, Salvadori A, Scarcella A. Molecular epidemiology of an outbreak of Serratia marcescens in a neonatal intensive care unit. Infect Control Hosp Epidemiol 2001; 22:630‐634.
- 9. Prasad GA, Jones PG, Michaels J, Garland JS, Shivpuri CR. Outbreak of Serratia marcescens infection in a neonatal intensive care unit. Infect Control Hosp Epidemiol 2001; 22:303‐305.
- 10. Milisavljevic V, Wu F, Larson E, et al. Molecular epidemiology of Serratia marcescens outbreaks in two neonatal intensive care units. Infect Control Hosp Epidemiol 2004; 25:719‐721.
- 11. Fleisch F, Zimmermann‐Baer U, Zbinden R, et al. Three consecutive outbreaks of Serratia marcescens in a neonatal intensive care unit. Clin Infect Dis 2002; 34:767‐773.
- 12. Stoll BJ, Hansen N, Fanaroff AA, et al. Late‐onset sepsis in very low birth weight neonates: the experience of the NICHD Neonatal Research Network. Pediatrics 2002; 110:285‐291.
- 13. Gransden WR, Webster M, French GL, Phillips I. An outbreak of Serratia marcescens transmitted by contaminated breast pumps in a special care baby unit. J Hosp Infect 1986; 7:149‐154.
- 14. Sartor C, Jacomo V, Duvivier C, Tissot‐Dupont H, Sambuc R, Drancourt M. Nosocomial Serratia marcescens infections associated with extrinsic contamination of a liquid nonmedicated soap. Infect Control Hosp Epidemiol 2000; 21:196‐199.
- 15. Waters V, Larson E, Wu F, et al. Molecular epidemiology of gram‐negative bacilli from infected neonates and health care workers’ hands in neonatal intensive care units. Clin Infect Dis 2004; 38:1682‐1687.
- 16. Almuneef MA, Baltimore RS, Farrel PA, Reagan‐Cirincione P, Dembry LM. Molecular typing demonstrating transmission of gram‐negative rods in a neonatal intensive care unit in the absence of a recognized epidemic. Clin Infect Dis 2001; 32:220‐227.
- 17. FDA Public Health Notification: Precautions in Using the Reintroduced Vapotherm 2000i Respiratory Gas Humidification System. FDA Web site. 2007. http://www.fda.gov/cdrh/safety/020107‐vapotherm.html. Accessed October 19, 2007.
- 18. Haley RW, Bregman DA. The role of understaffing and overcrowding in recurrent outbreaks of staphylococcal infection in a neonatal special‐care unit. J Infect Dis 1982; 145:875‐885.
- 19. Harbarth S, Sudre P, Dharan S, Cadenas M, Pittet D. Outbreak of Enterobacter cloacae related to understaffing, overcrowding, and poor hygiene practices. Infect Control Hosp Epidemiol 1999; 20:598‐603.
-
Presented in part: 16th Annual Meeting of the Society for Healthcare Epidemiology of America; Chicago, Illinois; March 18‐21, 2006 (Abstract 118).