- Email: [email protected]
A Flavobacterium meningosepticum outbreak among intensive care patients
Marian Pokrywka, MS Kathryn Viazenko, BSN, CIC Jeanne Medvick, MT (ASCP) Stephanie Knsbe, RN Sheila McCwl, RN, MPH A. Williem Pasculle, ScD John N. Dowling, MD Pittsburgh, Pennsylvania
A Flawbacterium meningosepticum outbreak, involving 12 infected and 47 colonized intensive care patients during the months of February through July 1990, was investigated. F. meningosepticum was isolated from tap water and ice, but these environmental strains eventually proved to be distinct from those colonizing patients. A review of newly colonized patients’ charts revealed that a common factor among the patients was daily changes of ventilator tubing pasteurized in the hospitals central sterile department. More than 90% of patients in the outbreak had been on ventilators that used the pasteurized tubing. An investigation of the pasteurization process found that two pasteurizer tanks had been operating at suboptimal temperatures ( c 62” C). Cultures of water from the tanks and droplets of water found in the pasteurized tubing grew species of Acinetobacter, Morazella, and Pseudomonas but did not grow F. meningosepticum. After deficiencies in the pasteurization process were corrected, the outbreak terminated. Despite the failure to culture F. meningosepticum, an analysis of gram-negative bacillaty isolates showed that the deficiency in the pasteurization process was a major contributor to colonization of ventilated patients by bacteria ubiquitous in tap water. (AJIC AM J INFECT CONTROL1993;21:139-45)
Before the outbreak described here, Fhvobacterium meningosepticum was an uncommon cause of colonization and infection in patients at Presbyterian University Hospital. During the S-month period preceding this outbreak, only four Flavobacterium isolates were recovered from the intensive care unit (ICU) patient population of the hospital. From February 25, 1990, to July 23, 1990, a total of 59 ICU patients became colonized or infected with this bacterium. Because of the ubiquitous nature of F. meningosepticum and its From the Departments of Infection Control and Clinical ogy, Presbyterian University Hospital and the University burgh School of Medicine, Pittsburgh, Pennsylvania. Reprint requests: Marian Pokrywka, Diseases, 971 Scaife Hall, University 15261. 0 1993 by the Association 0196-6553/93
MS, Division of Infectious of Pittsburgh, Pittsburgh, PA
for Practitioners
$01 .OO + 0.10
Microbiolof Pitts-
17/47/42920
in Infection
Control,
Inc.
documented recovery from tap-water sources,’ efforts to control the spread of this microorganism were initially directed at eliminating the use of potable tap water, other than for handwashing and sponge baths, from the clinical care of ICU patients. Patient contacts with tap water that were identified included patient ice and ice-water ingestion, oral suspensions mixed with tap water, mouth care, tube feedings, and radiologic contrast fluids mixed with water. All hospital departments involved with the routine care of ICU patients were instructed to use sterile or pasteurized water for procedures involving this patient group. Only pasteurized water was used for making ice and ice chips for consumption by patients. In addition, because direct contact was also a potential route of transmission, in-service training on aggressive handwashing and good aseptic technique was carried out on each patient unit. Despite these control measures, cases continued to occur. 139
140
Pokrywka et al.
In July 1990, a retrospective study of the common factors among three ICU patients who became colonized with F. meningosepticum within the same 48-hour period led to an investigation of ventilator tubing as a possible source of the microorganism. Ventilator tubing, a reusable item in our institution, is processed in washerpasteurizers that use potable tap water. Deficient pasteurization of ventilator-y tubing apparently served as the source of F. meningosepticum and other gram-negative bacilli colonizing the respiratory tracts of intubated patients. METHODS Surveillance
and case deflnltion
Presbyterian University Hospital is a 636-bed tertiary care facility affiliated with the University of Pittsburgh Medical Center. The hospital provides 74 critical care beds in seven ICUs comprising one trauma, one cardiothoracic, one medical, one medical-surgical, and three liver transplantation units. On a daily average, 60% to 70% of the ICU patients are maintained on mechanical ventilation. No patients outside the ICUs are on ventilators. Infection surveillance in the ICUs is conducted by a staff of five infection control practitioners. Centers for Disease Control National Nosocomial Infections Surveillance Systems’ definitions for all infections are strictly followed. In our study, a “case” was defined as an ICU patient whose culture at any body site was positive for F. meningosepticum during an ICU admission or as long as 48 hours after transfer from an ICU. No patients had F. meningosepticum isolated before admission to an ICU unless they were transferred from another ICU. Cases were identified by daily laboratory report review performed with the hospital’s computerized data system and by chart review conducted in the ICUs by the infection control practitioners. Surveillance cultures for F. meningosepticum were not done because the ICU population is frequently cultured because of fever and respiratory problems. Environmental
culturkrg
Potable water sources in the hospital’s ICUs were cultured for the presence of F. meningosepticum. Sink spigots in private ICU cubicles were sampled by swabbing the inside of the water spigot with a sterile cotton swab and plating the specimen on 5% sheep blood agar (SBA) and eosinmethylene blue agar with 10 pg/rnl gentamicin
June
AJIC 1993
added (EMBG). Ice machines located on the patient units were also sampled. The spigot on the water dispenser and the lip on the ice dispenser were swabbed with a sterile cotton applicator and plated. Ice from the machines was collected in clean Styrofoam cups, allowed to melt, and plated in 1 ml and 0.1 ml aliquots on SBA and EMBG. All plates were incubated at 37” C and examined at 24, 48, and 72 hours. Antibiotic susceptibility studies were done for all environmental and patient F. meningosepticum isolates with the Kirby-Bauer method. In addition, patient isolates were submitted to the Centers for Disease Control for serotyping studies, which were performed by Dr. Robert Weaver. Chart
review
In July 1990, the charts and computerized billing records of three patients in one ICU who all had cultures positive for F. meningosepticum within a 48-hour period were examined retrospec-. tively for procedures or tests that could involve the use of tap water or a common exposure to a water source. The three patients were liver transplant recipients requiring intensive care and mechanical ventilation. Patient activities, tests, medications, and treatments were reviewed for 7 days before the first positive culture. Specific factors reviewed included operations, other invasive procedures, anesthesia, dialysis, respiratory therapy treatments, drug aerosol treatments, oral intake, radiologic procedures, transfusions, bronchial lavages, feeding-tube placement, irrigations, ultrasonographic studies, blood gas capnograph tests, enteral feedings, and nursing assignments. Central
sterile
Inrpectlon
After the chart review analysis, an inspection of the central sterile department (CSD) was made. The CSD operates two washer-pasteurizers for the reprocessing of ventilator tubing. Information was obtained on operating temperatures, maintenance records, log books, and equipment drying chambers. Culturlng
of pasteurkers
and tublng
Water from the CSD’s two washer-pasteurizer tanks was collected in a sterile specimen container. One ml and 0.1 ml aliquots were plated on SBA and EMBG. Cultures were incubated at 37” C and examined at 24,48 and 72 hours. Growth was recorded quantitatively in colony-forming units per milliliter. Pasteurized silicone ventilator tub-
AJIC Pohywka et al.
Volume 21, Number 3
Table
1. Distribution
of Flavobacferiurn
141
cases by location
csses Unit
Typs of ICU
SICU MICU PICU 46 64 95 105
hlMclha
Cardiothoracic Medical Medical/surgical Trauma/neurosurgery Liver transplant Liver transplant Liver transplant
3 1 1 2 1 2 2 12
TOTAL
Attack rate
(%) =
Cases Discharges + transfers + deaths
cola-a 7 4 0 11 4 12 6 47
Total
Altack mta (%)
11 5 1 13 5 14 10 59
2.0 2.0 0.3 6.0 2.0 11.0 11.0 3.0
x IOQ.
ing with obvious residual moisture was retrieved from respiratory care storage areas located on the ICUs. Fluid for culture was obtained by clamping off the ends of the tube with binder clips and swinging the tube in a circular pattern to force all moisture to one end. The retrieved fluid (usually less than 5 ml) was plated in 1 ml and 0.1 ml aliquots on SBA and EMBG. Plates were incubated as above and growth estimated quantitatively.
tract. Ninety-three percent of the 59 case patients were on ventilatory support before culture positivity. The distribution of cases and infections across the seven ICUs and the attack rate for each ICU are depicted in Table 1. ICUs housing patients with extended lengths of stay tended to have higher attack rates. The occurrence of new cases over time is illustrated in Fig. 1.
Computer-assisted review of all aerobic gramnegative bacillus isolates from ICU patients was completed for the outbreak period of approximately five months and for five months before and five months after the designated period. A tabulation of all Enterobacteriaceae and nonEnterobacteriaceae isolates, both community acquired and nosocomially acquired, was compiled and examined for changes in the number of isolations during the three periods. It was assumed that community-acquired isolates would remain relatively stable in number and therefore would not affect the analyses of interest. Multiple isolations of an organism from the same patient were only counted once. Differences in the number of isolations during the three periods were compared with the x2 test with Yates’ correction.
Samples of ice or swabs taken from the ice machines in five of the hospital’s ICUs were positive for F. meningosepticum. Flavobacterium could be grown from the melted ice samples from each of the five ICUs and from either the ice dispenser lip or the water dispenser of the ice machines on three of the five units. In addition, spigots from individual patient cubicle sinks also grew Flavobacterium on two of five units. Antibiotic susceptibility studies by disk-diffusion testing showed the environmental isolates to be extremely sensitive to antibiotics. In contrast, patient isolates were noted to be resistant to clindamycin, vancomycin, trimethoprim/sulfamethoxazole, or ciprofloxacin. Although Fhwbacterium could be isolated from potable water sources in the ICUs, the environmental strains did not have the same antibiograms as the patient isolates. F. meningosepticum isolates submitted to the Centers for Disease Control for serotyping were found to be non--able.
Fifty-nine cases were identified. Twelve of the 59 cases were infections: one intraabdominal abscess, two bacteremias, eight pneumonias, and one case of bronchitis. The remaining 47 cases were colonizations, primarily of the respiratory
Ch*rI mubw Analysis of the patients’ charts showed mechanical ventilation to be the only common risk of possible exposure to tap water before culture positivity. Patients on ventilators receive daily
AJIC 142
Pokrywka et al.
June
I 4/l
515
-I
1993
711
613
Date Fig. 1. Occurrence of F. meningosepficum colonizations and infections in ICUs during the outbreak period, February 25, 1990, through July 23, 1990.
tubing changes, which appear as patient charges’ in the hospital’s computerized billing system. Central
sterile
inspection
Used ventilator tubing is gas sterilized and then washed and pasteurized for reuse. Washerpasteurizers in the hospital’s CSD were filled with city tap water and used to pasteurize hundreds of silicone tubes and fittings each week for use in ICUs. An inspection of the CSD revealed a number of deficiencies in the pasteurization process. Primarily, it was noted that the pasteurization tanks were operating at a temperature of 48” C rather than the manufacturer’s recommended temperature of 77” C. Other deficiencies included lack of record keeping regarding the tank temperatures during operation, lack of a regular maintenance schedule for shutdown and cleaning, and incomplete drying of the ventilator-y tubing after pasteurization because of inadequate air flow in a drying chamber. Recommendations were made to remedy all deficiencies in the system. Culturing
of pasteurizers
and tublng
Cultures were obtained from the tanks of the two operating washer-pasteurizers. Samples of the water from one pasteurizer tank grew Pseudomonas vesicularis and Bacillus sp (Table
2). Fluid from pasteurized tubing taken from four ICU workrooms grew a variety of organisms, including Bacillus sp, Acinetobacter sp, Pseudomonas stutzeri, Pseudomonas vesicularis, and Moraxe2Za sp (Table 2). The recovery of large numbers of P. vesicularis and Bacillus sp from both the pasteurization tank water and pasteurized tubing indicating indicated that the pasteurizer water was the source of these bacteria. However, none of the pasteurized tubing samples or the water from the pasteurizer tanks grew Flavobacterium. Analysis
of gram=negative
baclllary
isolates
Because F. meningosepticum was not grown from the pasteurizer tanks, a review of all gramnegative bacilli isolated from ICU patients-for 5 months before the outbreak, during the S-month outbreak, and 5 months after the outbreak- was undertaken to determine whether intervention in the pasteurization process had affected the gramnegative flora of ICU patients. Table 3 illustrates the gram-negative bacilli surveyed. The isolation of bacterial species belonging to the family Enterobacteriaceae did not change from the preoutbreak period to either the outbreak or postoutbreak periods (Table 3). Overall, the isolation of non-Enterobacteriaceae increased during the outbreak period and decreased after reinstitution of
AJIC Volume
21,
Table
Number
Pokrywka
3
et al.
f43
2. Cultures of pasteurized water and ventilator tubing Gram poaittive
Central sterile Pasteurizer Pasteurizer Ventilator SICU
1 2
No growth TNTC
Gmm negathfe
Bacillus sp
No growth TNTC
P. vesicularis
tubing
500
Unit 46
2000
Bacillus sp S epidermidis
Unit 95
TNTC
Bacillus sp
>lOO
P. vesicularis Moraxella sp Acinetobacter sp
>lOOO
Unidentified
Unit
500
S. epidermidis
TNTC > 1000
P. stutzeri
105
TA!TC. Too numerous
T8ble
to count:
CFU/m/, colony-forming
No growth 750 >lOOO
Unidentified
gram-negative
bacillus
gram-negative
bacillus
units per mililiter.
3. Isolates of gram-negative rods from ICU patients Bacterium
Pro outbreak
lntmoutbreak
Poatoutbmek
Enterobacteriaceae
Citrobacter diversus Citrobacter freundii Enterobacter aerogenes Enterobacter cloacae Escherichia coli Klebsiella sp Morganella morganii Proteus mirabilis Serratia marcescens SUBTOTAL
5 14 30 86 68 70 7 24 16 320
2 20 36 98 64 80 4 22 18 344
7 21 41 93 72 74 6 16 19 349
122 0 4 8 97 37 268 588
111 1 58 6 86 54 316 660
70 1 14 12 89 62 249 598
NortEnterobacteriaceae
A. anitratis* Aeromonas hydrophila F. meningosepticum * Flavobacterium sp Pseudomonas aeruginosa Pseudomonas sp SUBTOTAL TOTAL
*Significant
difference
(JI c 0.001) between
periods
in number
of isolates.
appropriate pasteurization (Table 3). However, the ratio of non-Enterobacteriaceae to Enterobacteriaceae isolations was not significantly different during the three intervals (x’ = 5.00,3 degrees of freedom, p c 0.1). Further examination of the non-Enterobacteriaceae revealed that F. meningosepticum isolates increased during the outbreak and decreased after the outbreak, relative to all other gram-negative bacillary isolates (x2 = 36.50, 3 degrees of freedom, p c 0.001). It is interesting that Acinetobacter isolates, which
were recovered from the pasteurized tubing, did not increase during the outbreak but did decline significantly after the outbreak (x2 = 17.79, 3 degrees of freedom, p < 0.001). DIDWSSIDN Flavobacterium
meningosepticum is a saprophytic soil bacterium and an uncommon agent of human infection. Accounts of nosocomial disease from F. meningosepticum include cases of neonatal meningitiq3 bacteremia,4-6 and pneumonia.’ In
144
Pokrywka et al.
1970, Olsen’ suggested that F. meningosepticum is widely distributed in the hospital environment. The mechanism of entry into the hospital by this saprophyte was put forward by Mot&n’ in a description of airway colonization of hospitalized patients by Flawbacterium sp. During a 5-month survey, Moulin’ recovered Flavobacterium from open water reservoirs in the Boston metropolitan area and related them to patient isolates from the Beth Israel Hospital. This suggested that the bacteria enter the hospital by way of the city water supply. Flavobacteria have been shown to survive the chlorination process, resisting levels of chlorine as high as 100 mg/kg.’ Moulin further suggested that transmission of Flawbacterium from the environment to patients could occur from tap water or ice. Colonized patients could then serve as secondary reservoirs of the bacterium and cross-contamination by hand carriage might amplify the numbers of affected patients. Nosocomial cases of FZavobacterium airway colonization and infection in intubated ICU patients have been documented.‘, ’ Not all studies, however, have been successful in identifying the source of the bacterium and recovering it from the hospital environment.” In our investigation, we proceeded with the premise that the source of the bacterium was the city water supply. Initial efforts to stop the outbreak centered around eliminating tap-water contact with patients’ mouths and airways. This approach seemed to be substantiated when cultures of potable water, ice machine spigots, and ice from the ICU areas showed growth of Flavobacterium. However, antibiograms of the environmental Flawbacterium isolates showed them to be highly susceptible, whereas the patient isolates were more resistant, suggesting that bacteria from the potable water may not have been the direct source for patient colonization. Our ICU patients are generally maintained on multiple antibiotic regimens for infection or prophylaxis, however, and this could have been responsible for the selection of the more resistant strains that became airway colonizers. Attempts to type the patient isolates were unsuccessful; a definitive marker to study crosscontamination between patients was thus not available. Because most of our ICU patients, both colonized and uncolonized, required mechanical ventilation, it was not immediately clear that this could be a source of Flawbacterium. It was not until the retrospective chart review of three colonized patients had been completed and the various
June
AJH: 1993
control measures instituted had failed to affect the outbreak that the pasteurized ventilator tubing was implicated as the environmental source. The likelihood of FZavobacterium-colonized patients passing their bacteria to susceptible patients through their secretions by transfer from one ICU to another or by hand transmission by physicians, nurses, and respiratory therapists had been entertained. One possibility that had not been considered was that the reusable tubing was being contaminated during the pasteurization process itself and that strains of FZavobacterium from tap water could survive the inadequate heating process and remain viable in the moisture droplets allowed to collect in the processed tubing. It is likely that the pasteurizer tanks, which were filled with city tap water and left to sit at suboptimal temperatures for days at a time, were themselves “incubators” for the Flavobacteria and other non-Enterobacteriaceae known to be present in tap water. The inability to recover Flawbacterium from the pasteurizer tanks and ventilator tubing may have been caused by an initial delay in sampling the tanks after recommendations were made to increase the operating temperature and to the small number of moisturecontaining tubes sampled from each ICU. The prompt termination of the outbreak after correction of the pasteurization process is convincing evidence that this mechanism played a significant role in the airway colonization of intubated patients. Because of the lack of microbiologic evidence of Flawbacterium in the pasteurizer tanks, the numbers of gram-negative bacillary isolates recovered from specimens in the ICUs before, during, and after the outbreak period were examined. The hypothesis was that there would be a decrease in the recovery of non-Enterobacteriaceae as a result of the intervention in the pasteurization process, because nonfermenters can be isolated from city water supplies, whereas the numbers of Enterobacteriaceae would not be affected. Specifically, it was expected that this decrease would be seen most dramatically in the recovery of F. meningosepticum. The analysis showed a rise in numbers of non-Enterobacteriaceae isolates during the outbreak period and a concomitant decrease in their numbers after the pasteurizer intervention to a level lower than that of the preoutbreak period. This decrease was ascribable to a statistically significant decrease in the recovery of F. meningosepticum and Acinetobacter anitratus. It seems likely that contaminated ventilator tubing
AJIC Volume 21, Number 3
Pokrywka et al.
contributed to the high numbers of Acinetobacter recovered before and during the outbreak, because isolates of this bacterium also declined after proper pasteurization was instituted. It was the isolation of Flavobacterium, however, rather than Acinetobacter, that precipitated the investigation because the background isolation rate of the former is low compared to that of the latter in our hospital. Our experience was similar to that described in a report published by the Hospital Infection Society of London” regarding an outbreak of Acinetobacter sp respiratory infections among ventilator-dependent patients associated with an equipment washer operating at lower than optimal temperatures. This may be a shared problem in institutions that reprocess items by washing and pasteurization. We gratefully acknowledge the technical support Sharon Krystofiak and the Microbiology Laboratory the secretarial assistance of Ms. Renita Prince and Edwards.
of Ms. staff and Ms. Betty
References 1. Moulin intensive 2. Centers
3.
4.
5.
6.
7.
8.
9. 10.
11. GC. Airway care unit. for Disease
colonization by J Clin Microbial Control. National
Flawbacterium
in an
1979; 10: 155-60. Nosocomial Infec-
145
tion Surveillance System Manual. Atlanta: Centers for Disease Control, 1988. King EO. Studies on a group of previously unclassiiied bacteria associated with meningitis in an infant. Am J Clin Path01 1959;31:241-7. Harrington SP, Perlino CA. Flawbacterium meningosepticum sepsis: disease due to bacteria with unusual antibiotic susceptibility. South Med J 1981;74:764-6. Bumakis TG, Mioduch HJ, LeBar WD, Yalamanchi RR. Sepsis from Flavobacterium meningosepticum, an uncommon pathogen with unusual susceptibility patterns. South Med J 1986;79:518-9. Coyle-Gilchrist MM, Crewe P, Roberts G. Flavobacterium meningosepticum in the hospital environment. J Clin Path01 1976;19:824-6. Teres D. ICU-acquired pneumonia due to Flavobacterium meningosepticum. JAMA 1974;228:732. Olsen H. Flawbacterium meningosepticum: bacteriological, epidemiological and clinical study. Danish Med Bull 1970;17:171-2. Herman LG, Himmelsbach CK. Detection and control of hospital sources of Flavobacteria. Hospitals 1965;39:72-6. Brown RB, Phillips D, Barker MJ, Pleczarka R, Sands M, Teres D. Outbreak of nosocomial Flawbacterium meningosepticum respiratory infections associated with use of aerosolized polymyxin B. AM J INFECT CONTROL 1989;17: 121-5. Cefai C, Richards J, Gould FK, McPeake P. An outbreak of Acinetobacter respiratory tract infections resulting from incomplete disinfection of ventilatory equipment. J Hosp Infect 1990:15:177-82.
A Flawbacterium meningosepticum outbreak, involving 12 infected and 47 colonized intensive care patients during the months of February through July 1990, was investigated. F. meningosepticum was isolated from tap water and ice, but these environmental strains eventually proved to be distinct from those colonizing patients. A review of newly colonized patients’ charts revealed that a common factor among the patients was daily changes of ventilator tubing pasteurized in the hospitals central sterile department. More than 90% of patients in the outbreak had been on ventilators that used the pasteurized tubing. An investigation of the pasteurization process found that two pasteurizer tanks had been operating at suboptimal temperatures ( c 62” C). Cultures of water from the tanks and droplets of water found in the pasteurized tubing grew species of Acinetobacter, Morazella, and Pseudomonas but did not grow F. meningosepticum. After deficiencies in the pasteurization process were corrected, the outbreak terminated. Despite the failure to culture F. meningosepticum, an analysis of gram-negative bacillaty isolates showed that the deficiency in the pasteurization process was a major contributor to colonization of ventilated patients by bacteria ubiquitous in tap water. (AJIC AM J INFECT CONTROL1993;21:139-45)
Before the outbreak described here, Fhvobacterium meningosepticum was an uncommon cause of colonization and infection in patients at Presbyterian University Hospital. During the S-month period preceding this outbreak, only four Flavobacterium isolates were recovered from the intensive care unit (ICU) patient population of the hospital. From February 25, 1990, to July 23, 1990, a total of 59 ICU patients became colonized or infected with this bacterium. Because of the ubiquitous nature of F. meningosepticum and its From the Departments of Infection Control and Clinical ogy, Presbyterian University Hospital and the University burgh School of Medicine, Pittsburgh, Pennsylvania. Reprint requests: Marian Pokrywka, Diseases, 971 Scaife Hall, University 15261. 0 1993 by the Association 0196-6553/93
MS, Division of Infectious of Pittsburgh, Pittsburgh, PA
for Practitioners
$01 .OO + 0.10
Microbiolof Pitts-
17/47/42920
in Infection
Control,
Inc.
documented recovery from tap-water sources,’ efforts to control the spread of this microorganism were initially directed at eliminating the use of potable tap water, other than for handwashing and sponge baths, from the clinical care of ICU patients. Patient contacts with tap water that were identified included patient ice and ice-water ingestion, oral suspensions mixed with tap water, mouth care, tube feedings, and radiologic contrast fluids mixed with water. All hospital departments involved with the routine care of ICU patients were instructed to use sterile or pasteurized water for procedures involving this patient group. Only pasteurized water was used for making ice and ice chips for consumption by patients. In addition, because direct contact was also a potential route of transmission, in-service training on aggressive handwashing and good aseptic technique was carried out on each patient unit. Despite these control measures, cases continued to occur. 139
140
Pokrywka et al.
In July 1990, a retrospective study of the common factors among three ICU patients who became colonized with F. meningosepticum within the same 48-hour period led to an investigation of ventilator tubing as a possible source of the microorganism. Ventilator tubing, a reusable item in our institution, is processed in washerpasteurizers that use potable tap water. Deficient pasteurization of ventilator-y tubing apparently served as the source of F. meningosepticum and other gram-negative bacilli colonizing the respiratory tracts of intubated patients. METHODS Surveillance
and case deflnltion
Presbyterian University Hospital is a 636-bed tertiary care facility affiliated with the University of Pittsburgh Medical Center. The hospital provides 74 critical care beds in seven ICUs comprising one trauma, one cardiothoracic, one medical, one medical-surgical, and three liver transplantation units. On a daily average, 60% to 70% of the ICU patients are maintained on mechanical ventilation. No patients outside the ICUs are on ventilators. Infection surveillance in the ICUs is conducted by a staff of five infection control practitioners. Centers for Disease Control National Nosocomial Infections Surveillance Systems’ definitions for all infections are strictly followed. In our study, a “case” was defined as an ICU patient whose culture at any body site was positive for F. meningosepticum during an ICU admission or as long as 48 hours after transfer from an ICU. No patients had F. meningosepticum isolated before admission to an ICU unless they were transferred from another ICU. Cases were identified by daily laboratory report review performed with the hospital’s computerized data system and by chart review conducted in the ICUs by the infection control practitioners. Surveillance cultures for F. meningosepticum were not done because the ICU population is frequently cultured because of fever and respiratory problems. Environmental
culturkrg
Potable water sources in the hospital’s ICUs were cultured for the presence of F. meningosepticum. Sink spigots in private ICU cubicles were sampled by swabbing the inside of the water spigot with a sterile cotton swab and plating the specimen on 5% sheep blood agar (SBA) and eosinmethylene blue agar with 10 pg/rnl gentamicin
June
AJIC 1993
added (EMBG). Ice machines located on the patient units were also sampled. The spigot on the water dispenser and the lip on the ice dispenser were swabbed with a sterile cotton applicator and plated. Ice from the machines was collected in clean Styrofoam cups, allowed to melt, and plated in 1 ml and 0.1 ml aliquots on SBA and EMBG. All plates were incubated at 37” C and examined at 24, 48, and 72 hours. Antibiotic susceptibility studies were done for all environmental and patient F. meningosepticum isolates with the Kirby-Bauer method. In addition, patient isolates were submitted to the Centers for Disease Control for serotyping studies, which were performed by Dr. Robert Weaver. Chart
review
In July 1990, the charts and computerized billing records of three patients in one ICU who all had cultures positive for F. meningosepticum within a 48-hour period were examined retrospec-. tively for procedures or tests that could involve the use of tap water or a common exposure to a water source. The three patients were liver transplant recipients requiring intensive care and mechanical ventilation. Patient activities, tests, medications, and treatments were reviewed for 7 days before the first positive culture. Specific factors reviewed included operations, other invasive procedures, anesthesia, dialysis, respiratory therapy treatments, drug aerosol treatments, oral intake, radiologic procedures, transfusions, bronchial lavages, feeding-tube placement, irrigations, ultrasonographic studies, blood gas capnograph tests, enteral feedings, and nursing assignments. Central
sterile
Inrpectlon
After the chart review analysis, an inspection of the central sterile department (CSD) was made. The CSD operates two washer-pasteurizers for the reprocessing of ventilator tubing. Information was obtained on operating temperatures, maintenance records, log books, and equipment drying chambers. Culturlng
of pasteurkers
and tublng
Water from the CSD’s two washer-pasteurizer tanks was collected in a sterile specimen container. One ml and 0.1 ml aliquots were plated on SBA and EMBG. Cultures were incubated at 37” C and examined at 24,48 and 72 hours. Growth was recorded quantitatively in colony-forming units per milliliter. Pasteurized silicone ventilator tub-
AJIC Pohywka et al.
Volume 21, Number 3
Table
1. Distribution
of Flavobacferiurn
141
cases by location
csses Unit
Typs of ICU
SICU MICU PICU 46 64 95 105
hlMclha
Cardiothoracic Medical Medical/surgical Trauma/neurosurgery Liver transplant Liver transplant Liver transplant
3 1 1 2 1 2 2 12
TOTAL
Attack rate
(%) =
Cases Discharges + transfers + deaths
cola-a 7 4 0 11 4 12 6 47
Total
Altack mta (%)
11 5 1 13 5 14 10 59
2.0 2.0 0.3 6.0 2.0 11.0 11.0 3.0
x IOQ.
ing with obvious residual moisture was retrieved from respiratory care storage areas located on the ICUs. Fluid for culture was obtained by clamping off the ends of the tube with binder clips and swinging the tube in a circular pattern to force all moisture to one end. The retrieved fluid (usually less than 5 ml) was plated in 1 ml and 0.1 ml aliquots on SBA and EMBG. Plates were incubated as above and growth estimated quantitatively.
tract. Ninety-three percent of the 59 case patients were on ventilatory support before culture positivity. The distribution of cases and infections across the seven ICUs and the attack rate for each ICU are depicted in Table 1. ICUs housing patients with extended lengths of stay tended to have higher attack rates. The occurrence of new cases over time is illustrated in Fig. 1.
Computer-assisted review of all aerobic gramnegative bacillus isolates from ICU patients was completed for the outbreak period of approximately five months and for five months before and five months after the designated period. A tabulation of all Enterobacteriaceae and nonEnterobacteriaceae isolates, both community acquired and nosocomially acquired, was compiled and examined for changes in the number of isolations during the three periods. It was assumed that community-acquired isolates would remain relatively stable in number and therefore would not affect the analyses of interest. Multiple isolations of an organism from the same patient were only counted once. Differences in the number of isolations during the three periods were compared with the x2 test with Yates’ correction.
Samples of ice or swabs taken from the ice machines in five of the hospital’s ICUs were positive for F. meningosepticum. Flavobacterium could be grown from the melted ice samples from each of the five ICUs and from either the ice dispenser lip or the water dispenser of the ice machines on three of the five units. In addition, spigots from individual patient cubicle sinks also grew Flavobacterium on two of five units. Antibiotic susceptibility studies by disk-diffusion testing showed the environmental isolates to be extremely sensitive to antibiotics. In contrast, patient isolates were noted to be resistant to clindamycin, vancomycin, trimethoprim/sulfamethoxazole, or ciprofloxacin. Although Fhwbacterium could be isolated from potable water sources in the ICUs, the environmental strains did not have the same antibiograms as the patient isolates. F. meningosepticum isolates submitted to the Centers for Disease Control for serotyping were found to be non--able.
Fifty-nine cases were identified. Twelve of the 59 cases were infections: one intraabdominal abscess, two bacteremias, eight pneumonias, and one case of bronchitis. The remaining 47 cases were colonizations, primarily of the respiratory
Ch*rI mubw Analysis of the patients’ charts showed mechanical ventilation to be the only common risk of possible exposure to tap water before culture positivity. Patients on ventilators receive daily
AJIC 142
Pokrywka et al.
June
I 4/l
515
-I
1993
711
613
Date Fig. 1. Occurrence of F. meningosepficum colonizations and infections in ICUs during the outbreak period, February 25, 1990, through July 23, 1990.
tubing changes, which appear as patient charges’ in the hospital’s computerized billing system. Central
sterile
inspection
Used ventilator tubing is gas sterilized and then washed and pasteurized for reuse. Washerpasteurizers in the hospital’s CSD were filled with city tap water and used to pasteurize hundreds of silicone tubes and fittings each week for use in ICUs. An inspection of the CSD revealed a number of deficiencies in the pasteurization process. Primarily, it was noted that the pasteurization tanks were operating at a temperature of 48” C rather than the manufacturer’s recommended temperature of 77” C. Other deficiencies included lack of record keeping regarding the tank temperatures during operation, lack of a regular maintenance schedule for shutdown and cleaning, and incomplete drying of the ventilator-y tubing after pasteurization because of inadequate air flow in a drying chamber. Recommendations were made to remedy all deficiencies in the system. Culturing
of pasteurizers
and tublng
Cultures were obtained from the tanks of the two operating washer-pasteurizers. Samples of the water from one pasteurizer tank grew Pseudomonas vesicularis and Bacillus sp (Table
2). Fluid from pasteurized tubing taken from four ICU workrooms grew a variety of organisms, including Bacillus sp, Acinetobacter sp, Pseudomonas stutzeri, Pseudomonas vesicularis, and Moraxe2Za sp (Table 2). The recovery of large numbers of P. vesicularis and Bacillus sp from both the pasteurization tank water and pasteurized tubing indicating indicated that the pasteurizer water was the source of these bacteria. However, none of the pasteurized tubing samples or the water from the pasteurizer tanks grew Flavobacterium. Analysis
of gram=negative
baclllary
isolates
Because F. meningosepticum was not grown from the pasteurizer tanks, a review of all gramnegative bacilli isolated from ICU patients-for 5 months before the outbreak, during the S-month outbreak, and 5 months after the outbreak- was undertaken to determine whether intervention in the pasteurization process had affected the gramnegative flora of ICU patients. Table 3 illustrates the gram-negative bacilli surveyed. The isolation of bacterial species belonging to the family Enterobacteriaceae did not change from the preoutbreak period to either the outbreak or postoutbreak periods (Table 3). Overall, the isolation of non-Enterobacteriaceae increased during the outbreak period and decreased after reinstitution of
AJIC Volume
21,
Table
Number
Pokrywka
3
et al.
f43
2. Cultures of pasteurized water and ventilator tubing Gram poaittive
Central sterile Pasteurizer Pasteurizer Ventilator SICU
1 2
No growth TNTC
Gmm negathfe
Bacillus sp
No growth TNTC
P. vesicularis
tubing
500
Unit 46
2000
Bacillus sp S epidermidis
Unit 95
TNTC
Bacillus sp
>lOO
P. vesicularis Moraxella sp Acinetobacter sp
>lOOO
Unidentified
Unit
500
S. epidermidis
TNTC > 1000
P. stutzeri
105
TA!TC. Too numerous
T8ble
to count:
CFU/m/, colony-forming
No growth 750 >lOOO
Unidentified
gram-negative
bacillus
gram-negative
bacillus
units per mililiter.
3. Isolates of gram-negative rods from ICU patients Bacterium
Pro outbreak
lntmoutbreak
Poatoutbmek
Enterobacteriaceae
Citrobacter diversus Citrobacter freundii Enterobacter aerogenes Enterobacter cloacae Escherichia coli Klebsiella sp Morganella morganii Proteus mirabilis Serratia marcescens SUBTOTAL
5 14 30 86 68 70 7 24 16 320
2 20 36 98 64 80 4 22 18 344
7 21 41 93 72 74 6 16 19 349
122 0 4 8 97 37 268 588
111 1 58 6 86 54 316 660
70 1 14 12 89 62 249 598
NortEnterobacteriaceae
A. anitratis* Aeromonas hydrophila F. meningosepticum * Flavobacterium sp Pseudomonas aeruginosa Pseudomonas sp SUBTOTAL TOTAL
*Significant
difference
(JI c 0.001) between
periods
in number
of isolates.
appropriate pasteurization (Table 3). However, the ratio of non-Enterobacteriaceae to Enterobacteriaceae isolations was not significantly different during the three intervals (x’ = 5.00,3 degrees of freedom, p c 0.1). Further examination of the non-Enterobacteriaceae revealed that F. meningosepticum isolates increased during the outbreak and decreased after the outbreak, relative to all other gram-negative bacillary isolates (x2 = 36.50, 3 degrees of freedom, p c 0.001). It is interesting that Acinetobacter isolates, which
were recovered from the pasteurized tubing, did not increase during the outbreak but did decline significantly after the outbreak (x2 = 17.79, 3 degrees of freedom, p < 0.001). DIDWSSIDN Flavobacterium
meningosepticum is a saprophytic soil bacterium and an uncommon agent of human infection. Accounts of nosocomial disease from F. meningosepticum include cases of neonatal meningitiq3 bacteremia,4-6 and pneumonia.’ In
144
Pokrywka et al.
1970, Olsen’ suggested that F. meningosepticum is widely distributed in the hospital environment. The mechanism of entry into the hospital by this saprophyte was put forward by Mot&n’ in a description of airway colonization of hospitalized patients by Flawbacterium sp. During a 5-month survey, Moulin’ recovered Flavobacterium from open water reservoirs in the Boston metropolitan area and related them to patient isolates from the Beth Israel Hospital. This suggested that the bacteria enter the hospital by way of the city water supply. Flavobacteria have been shown to survive the chlorination process, resisting levels of chlorine as high as 100 mg/kg.’ Moulin further suggested that transmission of Flawbacterium from the environment to patients could occur from tap water or ice. Colonized patients could then serve as secondary reservoirs of the bacterium and cross-contamination by hand carriage might amplify the numbers of affected patients. Nosocomial cases of FZavobacterium airway colonization and infection in intubated ICU patients have been documented.‘, ’ Not all studies, however, have been successful in identifying the source of the bacterium and recovering it from the hospital environment.” In our investigation, we proceeded with the premise that the source of the bacterium was the city water supply. Initial efforts to stop the outbreak centered around eliminating tap-water contact with patients’ mouths and airways. This approach seemed to be substantiated when cultures of potable water, ice machine spigots, and ice from the ICU areas showed growth of Flavobacterium. However, antibiograms of the environmental Flawbacterium isolates showed them to be highly susceptible, whereas the patient isolates were more resistant, suggesting that bacteria from the potable water may not have been the direct source for patient colonization. Our ICU patients are generally maintained on multiple antibiotic regimens for infection or prophylaxis, however, and this could have been responsible for the selection of the more resistant strains that became airway colonizers. Attempts to type the patient isolates were unsuccessful; a definitive marker to study crosscontamination between patients was thus not available. Because most of our ICU patients, both colonized and uncolonized, required mechanical ventilation, it was not immediately clear that this could be a source of Flawbacterium. It was not until the retrospective chart review of three colonized patients had been completed and the various
June
AJH: 1993
control measures instituted had failed to affect the outbreak that the pasteurized ventilator tubing was implicated as the environmental source. The likelihood of FZavobacterium-colonized patients passing their bacteria to susceptible patients through their secretions by transfer from one ICU to another or by hand transmission by physicians, nurses, and respiratory therapists had been entertained. One possibility that had not been considered was that the reusable tubing was being contaminated during the pasteurization process itself and that strains of FZavobacterium from tap water could survive the inadequate heating process and remain viable in the moisture droplets allowed to collect in the processed tubing. It is likely that the pasteurizer tanks, which were filled with city tap water and left to sit at suboptimal temperatures for days at a time, were themselves “incubators” for the Flavobacteria and other non-Enterobacteriaceae known to be present in tap water. The inability to recover Flawbacterium from the pasteurizer tanks and ventilator tubing may have been caused by an initial delay in sampling the tanks after recommendations were made to increase the operating temperature and to the small number of moisturecontaining tubes sampled from each ICU. The prompt termination of the outbreak after correction of the pasteurization process is convincing evidence that this mechanism played a significant role in the airway colonization of intubated patients. Because of the lack of microbiologic evidence of Flawbacterium in the pasteurizer tanks, the numbers of gram-negative bacillary isolates recovered from specimens in the ICUs before, during, and after the outbreak period were examined. The hypothesis was that there would be a decrease in the recovery of non-Enterobacteriaceae as a result of the intervention in the pasteurization process, because nonfermenters can be isolated from city water supplies, whereas the numbers of Enterobacteriaceae would not be affected. Specifically, it was expected that this decrease would be seen most dramatically in the recovery of F. meningosepticum. The analysis showed a rise in numbers of non-Enterobacteriaceae isolates during the outbreak period and a concomitant decrease in their numbers after the pasteurizer intervention to a level lower than that of the preoutbreak period. This decrease was ascribable to a statistically significant decrease in the recovery of F. meningosepticum and Acinetobacter anitratus. It seems likely that contaminated ventilator tubing
AJIC Volume 21, Number 3
Pokrywka et al.
contributed to the high numbers of Acinetobacter recovered before and during the outbreak, because isolates of this bacterium also declined after proper pasteurization was instituted. It was the isolation of Flavobacterium, however, rather than Acinetobacter, that precipitated the investigation because the background isolation rate of the former is low compared to that of the latter in our hospital. Our experience was similar to that described in a report published by the Hospital Infection Society of London” regarding an outbreak of Acinetobacter sp respiratory infections among ventilator-dependent patients associated with an equipment washer operating at lower than optimal temperatures. This may be a shared problem in institutions that reprocess items by washing and pasteurization. We gratefully acknowledge the technical support Sharon Krystofiak and the Microbiology Laboratory the secretarial assistance of Ms. Renita Prince and Edwards.
of Ms. staff and Ms. Betty
References 1. Moulin intensive 2. Centers
3.
4.
5.
6.
7.
8.
9. 10.
11. GC. Airway care unit. for Disease
colonization by J Clin Microbial Control. National
Flawbacterium
in an
1979; 10: 155-60. Nosocomial Infec-
145
tion Surveillance System Manual. Atlanta: Centers for Disease Control, 1988. King EO. Studies on a group of previously unclassiiied bacteria associated with meningitis in an infant. Am J Clin Path01 1959;31:241-7. Harrington SP, Perlino CA. Flawbacterium meningosepticum sepsis: disease due to bacteria with unusual antibiotic susceptibility. South Med J 1981;74:764-6. Bumakis TG, Mioduch HJ, LeBar WD, Yalamanchi RR. Sepsis from Flavobacterium meningosepticum, an uncommon pathogen with unusual susceptibility patterns. South Med J 1986;79:518-9. Coyle-Gilchrist MM, Crewe P, Roberts G. Flavobacterium meningosepticum in the hospital environment. J Clin Path01 1976;19:824-6. Teres D. ICU-acquired pneumonia due to Flavobacterium meningosepticum. JAMA 1974;228:732. Olsen H. Flawbacterium meningosepticum: bacteriological, epidemiological and clinical study. Danish Med Bull 1970;17:171-2. Herman LG, Himmelsbach CK. Detection and control of hospital sources of Flavobacteria. Hospitals 1965;39:72-6. Brown RB, Phillips D, Barker MJ, Pleczarka R, Sands M, Teres D. Outbreak of nosocomial Flawbacterium meningosepticum respiratory infections associated with use of aerosolized polymyxin B. AM J INFECT CONTROL 1989;17: 121-5. Cefai C, Richards J, Gould FK, McPeake P. An outbreak of Acinetobacter respiratory tract infections resulting from incomplete disinfection of ventilatory equipment. J Hosp Infect 1990:15:177-82.