Outbreak of Bloodstream Infection With the Mold Phialemonium Among Patients Receiving Dialysis at a Hemodialysis Unit
Background Molds are a rare cause of disseminated infection among dialysis patients.
Objective. We evaluated a cluster of intravascular infections with the mold Phialemonium among patients receiving hemodialysis at the same facility in order to identify possible environmental sources and prevent further infection.
Design. Environmental assessment and case‐control study.
Setting. A hemodialysis center affiliated with a tertiary care hospital.
Methods. We reviewed surveillance and clinical microbiology records and performed a blood culture survey for all patients. The following data for case patients were compared with those for control patients: underlying illness, dialysis characteristics, medications, and other possible exposure for 120 days prior to infection. Environmental assessment of water treatment, dialysis facilities, and heating, ventilation, and air‐conditioning (HVAC) systems of the current and previous locations of the dialysis center was performed. Samples were cultured for fungus; Phialemonium isolates were confirmed by sequencing of DNA. Investigators observed dialysis access site disinfection technique.
Results. Four patients were confirmed as case patients, defined as a patient having intravascular infection with Phialemonium species; 3 presented with fungemia, and 1 presented with an intravascular graft infection. All case patients used a fistula or graft for dialysis access, as did 12 (75%) of 16 of control patients (
). Case and control patients did not differ in other dialysis characteristics, medications received, physiologic findings, or demographic factors. Phialemonium species were not recovered from samples of water or dialysis machines, but were recovered from the condensation drip pans under the blowers of the HVAC system that supplied air to the dialysis center. Observational study of 21 patients detected suboptimal contact time with antiseptic agents used to prepare dialysis access sites.
Conclusion. The report of this outbreak adds to previous published reports of Phialemonium infection occurring in immunocompromised patients who likely acquired infection in the healthcare setting. Recovery of this mold from blood culture should be considered indicative of infection until proven otherwise. Furthermore, an investigation into possible healthcare‐related environmental reservoirs should be considered.
Received September 20, 2005; accepted January 27, 2006; electronically published October 20, 2006.
Members of the genus Phialemonium, so named for its similarity to both Phialophora and Acremonium organisms, are a rare cause of invasive mold infections and, at this time, comprise the species Phialemonium curvatum and Phialemonium obovatum.1 Although these molds have been recovered from air, soil, water, and sewage, they rarely cause disease in humans. Disease has been reported mostly as sporadic infections among persons with impaired host defenses.2‐7 A previous reported cluster of 2 infections occurred in immunosuppressed, hospitalized persons.8 Although the 2 fungal isolates in this cluster were related by DNA sequencing, the infections were thought to be hospital acquired, and no environmental source of exposure was found. A recent cluster involving 3 otherwise healthy male patients with endocarditis was linked to contaminated prefilled syringes for injection.9
This report describes the epidemiology and control of a cluster of 4 cases of intravascular Phialemonium infection among hemodialysis patients. The cases described in this study were reported elsewhere.10 Briefly, in August 2001, a rapidly fatal case of disseminated Phialemonium infection occurred in a 35‐year‐old patient (patient B) at hemodialysis center A (HCA) in Illinois. In February 2001, Phialemonium isolates were recovered from a culture of blood obtained from another patient (patient A) at the same dialysis center. This isolate had been interpreted as a contaminant, because patient A had no symptoms and 2 sets of blood cultures were sterile in the months after the initial isolate was recovered. However, in December 2001 patient A required surgical revision of a nonfunctioning arteriovenous (AV) graft. Cultures of tissue from the removed graft grew Phialemonium species, and patient A ultimately died from disseminated Phialemonium infection.
Before the 2 deaths, HCA was relocated in March 2001 to a newly constructed unit housed on a different floor within the same medical office building. Though the infections were initially thought to be associated with the old unit, 2 additional Phialemonium infections occurred among HCA patients in July and August 2002, raising concerns that exposure to the source of infection was ongoing.
The microbiologic characteristics of this mold and the clinical manifestations of the 4 infections were reported separately.10 All isolates were of a distinct morphological type but were shown by partial ribosomal sequencing to be closely related to reference isolates of P. curvatum. Two patients from this outbreak died of the infection; both developed overwhelming infection associated with fungemia and endocarditis. Endocarditis also appeared to develop in a third patient, who was successfully treated. The fourth patient died from noninfectious causes after successful treatment of an AV graft infected with P. curvatum.
Methods
Case Definition and Ascertainment
We defined a case as isolation of Phialemonium species in a culture of a specimen obtained from a patient of HCA between September 2000 and December 2002 (the outbreak period). We conducted case finding by reviewing HCA surveillance data and clinical microbiology laboratory records for molds isolated during the outbreak period. Surveillance data maintained by HCA included catheter‐related bloodstream infection (BSI) rates, prevalence of various dialysis access types (eg, grafts, fistulas, and catheters), unit census, and use of inpatient dialysis. In addition, prior to this investigation, HCA had instituted enhanced surveillance for Phialemonium BSIs, which consisted of performing 1 set of fungal blood cultures for all patients. For all case patients, clinical and demographic data were obtained by detailed medical record review.
Environmental and Infection Control Assessment
We conducted an environmental assessment consisting of 2 parts. First, we performed a visual inspection of the previous and current locations of HCA and the inpatient dialysis unit in the affiliated hospital (hospital A). This visual inspection included assessment of the dialysis water treatment and heating, ventilation, and air‐conditioning (HVAC) systems. During the inspection, we obtained specimens for culture from several environmental surfaces and tap water. The Illinois Department of Public Health cultured samples obtained from the new unit between August and October 2002, including environmental surfaces, water at all points of the water‐treatment process, and tap water. Samples obtained from surfaces and water in the old unit and the hospital dialysis unit during this period were also cultured. Finally, cultures of water specimens obtained from different parts of the HVAC system at both unit locations were performed, including standing water in the condensation pan.
Second, we reviewed dialysis and infection control policies and practices. To measure compliance with policies relating to cleaning and accessing AV grafts, fistulas, and catheters, we conducted an observational study by means of a data collection tool, in accordance with the relevant policies. Investigators observed the preparation of dialysis access sites by patients and healthcare workers, the disinfecting agent(s) used, and the time the agent was in contact with the site.
Laboratory Evaluation
Environmental samples collected in August were processed at the Illinois Department of Public Health microbiology laboratory in Chicago. Samples were incubated on blood agar or malt extract agar. Recovered molds were evaluated by standard methods at the Great Lakes Center of Excellence in Environmental Health at the University of Illinois at Chicago. Most samples were collected in October, and were sent to the Centers for Disease Control and Prevention (CDC; Atlanta GA) for processing. Surface swab samples were plated on Sabouraud dextrose agar with chloramphenicol. Water samples were cultured by the membrane filtration method; membrane filters were then placed on Rose‐Bengal and Sabouraud dextrose agar plates.11 Suspected Phialemonium isolates were subcultured on potato dextrose agar and incubated at 25°C. Evaluation of macroscopic and microscopic morphology was used to confirm the isolates as Phialemonium species.12 Internal transcribed spacer (ITS) sequencing of 2 environmental isolates was performed at CDC using ITS5 and ITS4 primers and the BigDye Terminator Cycle Sequencing kit (Applied Biosystems). Sequences were analyzed with an ABI 310 instrument (Applied Biosystems).
Isolates from all 4 case patients were originally evaluated at the Fungus Testing Laboratory at the University of Texas Health Science Center at San Antonio. Fungal isolates from 2 of the 4 case patients were also examined at the CDC. All isolates were sent to the Centraalbureau voor Schimmelcultures (Utrecht, The Netherlands), where ITS sequencing of case patient isolates confirmed that they were P. curvatum.10
Case‐Control Study
A case‐control study was conducted to assess potential risk factors for infection. Four unmatched controls per case were randomly selected from among HCA patients who received dialysis, regardless of the type of dialysis access or the history of recent BSIs. Because the incubation period (the time from infection to the development of symptoms) for molds such as Phialemonium is unknown, we chose to define the risk period as the 120 days preceding the date of symptom onset or the first isolation of Phialemonium in culture, whichever was first. For patient A, who had isolates recovered in both February and December 2001, the risk period for the December isolate was used, because the patient remained asymptomatic without treatment after the February isolate was recovered. Data were abstracted from medical records and dialysis run sheets for underlying illnesses, symptoms or signs of infection, medications administered, dialysis access type, dialysis session and day, dialyzer characteristics, and receipt of in‐hospital dialysis.
Univariate statistical analysis was performed using SAS, version 8.2 (SAS Institute). Continuous data were compared by means of t tests and categorical variables by means of χ2 tests (
). The Wilcoxon rank sum test was used for analysis of nonparametric data.
Results
Case Ascertainment and Description
Four patients met the case definition (Table 1). The median age was 68 years (range, 35‐77 years), 3 of 4 case patients were men, and all were black. The median duration of hemodialysis was 4.5 years (range, 3‐11 years). All case patients had AV grafts or fistulas, rather than catheters; 3 of 4 case patients underwent inpatient dialysis at hospital A prior to infection, whereas 1 did not. The clinical course of each case patient has been described elsewhere.10
Hemodialysis Center Surveillance Data
During the outbreak period, the mean census for HCA was 214 patients, with the monthly census increasing consistently over the outbreak period, from 186 to 248 patients. During the outbreak period, 65% of patients had AV grafts or fistulas, and 35% had permanent catheters, whereas all case patients had AV grafts or fistulas. The pooled mean BSI rate at HCA during the outbreak period was 2.6 BSIs per 100 person‐months (range, 0‐5.9 BSIs per 100 person‐months). Months in which Phialemonium infections occurred were not months in which the patient census or monthly BSI rate were high (ie, greater than the 75th percentile). Expanded blood culture surveillance identified no additional cases of Phialemonium infection among current HCA patients.
Environmental and Infection Control Assessment
Before March 12, 2001, HCA was located on the third floor of a 7‐story medical office building that is physically contiguous with hospital A. The old unit location currently houses a patient‐holding area and storage area for the hospital's gastroenterology laboratory. HCA staff reported little renovation activity while patients were treated at either unit. Visual inspection of the old unit site revealed mold contamination of dialysis water drains at some of the dialysis station locations. Samples obtained from these drains had been cultured previously, and Phialemonium species were not recovered (Table 2). Because 3 of 4 cases occurred after HCA was moved to the newly constructed facility on the fifth floor of the office building, we focused our environmental assessment on that location.
HCA is on the entire fifth floor of the medical office building. The unit is divided into 2 rooms, one with 24 and the other with 22 dialysis stations. This location also houses the water treatment system for the unit, comprising 2 carbon filters, a reverse osmosis unit, and a microfilter. After processing, treated water and dialysate are further filtered in the individual dialysis machines, passing through an internal ultrafilter, as well as through the dialyzer itself.
No evidence was found in dialysis maintenance records that the water treatment system had failed, nor was any single dialysis machine linked to every case of Phialemonium infection. Phialemonium species were not recovered from cultures of 10 water samples obtained from locations before and 7 water samples obtained from locations in the water‐processing circuit, or from 2 cultures of dialysate from each of 3 dialysis machines (Table 2). In addition, by report of the municipal water district office, municipal water chlorination and coliform levels were stable and appropriate throughout the outbreak period.
Before July 2002, HCA processed and reused dialyzers, but this procedure was suspended in late July 2002; the dialyzer reuse–processing loop and reuse tank were not being used at the time of this investigation. During previous visits by Illinois Department of Public Health staff, it was noted that the large storage tank in the reuse circuit had no lid and was open to the air rather than being covered with an air vent containing a bacteriologic filter.
We also inspected the HVAC system that supplies treated air to HCA. Multiple large HVAC units are in place on the seventh floor of hospital A, from which the air used in HCA is drawn. The building engineering staff estimates that 85% of the distributed air is recirculated and 15% is fresh. One of these units supplies the old HCA location, and 2 supply the new location. We cultured water collected from condensation drip pans under these 3 units; Phialemonium species were recovered from 2 of 3 cultures. The ITS sequences of these isolates were identical to each other and to the ITS sequences of the 4 patient isolates previously identified as P. curvatum.
Free‐standing disposable pump soap dispensers were used at nurse and patient handwashing sinks as the preferred method of hand hygiene. All medications used at HCA were purchased as single‐dose vials. No medications are compounded or prepared separately, although some medications, such as antibiotics, require reconstitution with sterile saline or water on site at HCA. Unused medications from opened vials are not pooled. All medications were obtained directly from manufacturers or third‐party distributors. No medications were obtained from compounding pharmacies.
Hospital A houses a small inpatient dialysis unit managed by HCA, where HCA patients undergo dialysis when they are admitted to hospital A. Three of 4 case patients had been dialyzed in this inpatient unit prior to onset of infection, whereas 1 (patient B) had not. Cultures of samples obtained from several sources in this unit were performed, including treated and untreated water (2 samples each), environmental surfaces (8 samples), and HVAC system ducts (3 samples); Phialemonium species were not recovered from these cultures.
Overall, we found poor compliance with soap‐and‐water hand hygiene among patients and suboptimal agent contact time when povidone‐iodine was used by healthcare workers as the disinfectant (Table 3).
Case‐Control Study
The 4 case patients were compared with 16 control patients in our case‐control study. All case patients had a fistula or graft for dialysis access, and 12 (75%) of 16 control patients also had these types of access (
by the Fisher exact test). Case and control patients did not differ significantly with respect to other dialysis characteristics, such as dialyzer reuse, dialysis session, or number of sessions and total hours dialyzed during the risk period. Case and control patients also did not differ significantly with respect to medications received, physiologic parameters (eg, blood hemoglobin or ferritin level), or demographic characteristics (eg, race, sex, and ethnicity).
Discussion
This investigation of an outbreak of hemodialysis‐associated endovascular infections with P. curvatum identified no common source of exposure, although some evidence suggests the HVAC system may have served as a transient reservoir. Although case patients were as likely as control patients to use AV grafts or fistulas for hemodialysis, all case patients had AV grafts or fistulas rather than catheters. Patients with grafts and fistulas differ from those with catheters in several ways. First, implanted synthetic grafts or surgically created fistulas may act as a nidus for subclinical Phialemonium infection. Small inocula may produce overt infection more readily in a graft or fistula than in a catheter. Second, grafts and fistulas require excellent skin site preparation prior to dialysis access. We observed suboptimal performance of skin antisepsis. We hypothesize that suboptimal skin preparation may allow viable molds on the skin to be inoculated into the bloodstream or the AV graft. We hypothesize that airborne and waterborne transmission are 2 possible causes of skin contamination.
Spores of some molds are spread efficiently via the waterborne route.13‐16 To explore the hypothesis that municipal water and/or HCA water might contain Phialemonium species, we obtained several tap water samples and swab specimens of faucets and faucet aerators for culture. Phialemonium species were not recovered from any of these sources. We found no evidence for disruptions in municipal water chlorination, which might have led to contamination of HCA water with Phialemonium species. This does not preclude, however, the past or intermittent presence of mold in hospital water. Of note, Phialemonium infection was reported in a newborn nursery in a different hospital within the same metropolitan area.7
Mold spores may also be spread via the airborne route.13 The hospital HVAC system treats and distributes air throughout the hospital and clinic buildings. The Phialemonium species isolated from 2 condensation drip pans under air‐conditioning units in the hospital may be conidia that have settled into the pans from the ambient air near the HVAC unit. They may also be flora that were present on the cooling coils of the HVAC units and were therefore in the flow of air circulation downstream from the units. In either case, turbulent airflow in this area could be associated with dissemination of contaminated air downstream from the existing filters. These HVAC units supply all old and half of the new unit locations, although the units themselves are physically distant from HCA. None of the 10 cultures of samples from downstream surfaces within the airflow system, including ceiling intake and exhaust covers in both units, yielded Phialemonium species. We did not attempt to obtain large‐volume air samples because the cooling season had passed. However, re‐assessment of the system in February did not yield Phialemonium species. In contrast, we found no evidence that an intrinsically or extrinsically contaminated medication was the source of these infections. Medications from single‐use vials were not pooled, and HCA receives no medications from compounding pharmacies, eliminating these possible sources of contamination.17‐19
Other possible scenarios are less likely. Although the reuse tank in HCA was not fitted with the proper biologic filter cover, for dialyzer reprocessing to have caused this outbreak would require several systematic failures, including failure to fill the dialyzer completely with germicide or filling it with germicide of the wrong concentration. If this had occurred, we would have expected to find increases in the episodes of pyrogenic reactions and cases of bacteremia, neither of which were evident on review of surveillance records from HCA.18,20,21
There are several limitations to this investigation. With only 4 case patients, we were unlikely to have enough statistical power to identify factors associated with case patient status. However, we do think we detected all case patients, as this mold was able to grow in routine blood culture bottles, and the IDPH performed a blood culture survey to detect any asymptomatic infections among patients receiving dialysis at HCA. Second, we were not able to limit the period of risk to only the old or new HCA location. The first Phialemonium isolate from patient A was cultured prior to the relocation of HCA to the new unit on March 12, 2001. All subsequent infections occurred in patients who had received dialysis in both the old and new unit locations, including patients C and D, who had onset of infection more than 1 year after the unit was relocated. Although the incubation period for Phialemonium species is unknown, it may be long. In prior mold outbreaks with defined exposure periods, the reported incubation period has been as long as 152 days.19 It is possible that exposure to the source of infection was associated with the old unit location only, but we cannot rule out an ongoing environmental exposure occurring in both units.
On the basis of observational and descriptive data, recommendations for improving aseptic technique have been made by the National Kidney Foundation Clinical Practice Guidelines for Vascular Access.22,23 On the basis of our findings, we recommend washing the access sites with an antibacterial soap or scrub and water to decrease the amount of skin microflora that may be introduced into the patient's bloodstream during cannulation and cleansing the skin by applying 70% alcohol and/or 10% povidone‐iodine in a circular rubbing motion. Staff may need to be educated that alcohol has a short bacteriostatic action time and can be applied in a rubbing motion for just 1 minute immediately prior to needle cannulation, whereas povidone iodine needs to be applied for 2‐3 minutes for its full bacteriostatic action to take effect and must be allowed to dry prior to needle cannulation. Furthermore, the HVAC unit coils and condensation drip pans were cleaned with an Environmental Protection Agency–registered disinfectant with a fungicidal label claim, and specific recommendations for improving drainage and disinfection of condensation drip pans were provided by the IDPH.
This outbreak adds to previous published reports of Phialemonium infection occurring in immunocompromised patients who likely acquired infection in the healthcare setting.5,7‐9 Recovery of this mold from culture of blood from patients with underlying illness, especially those with suspected endovascular infection, should be considered indicative of infection until proven otherwise. Furthermore, an investigation into possible healthcare‐related environmental reservoirs, especially contaminated air, should be considered.
Acknowledgments
We thank Richard C. Summerbell, Centraalbureau voor Schimmelcultures Fungal Biodiversity Center, Utrecht, The Netherlands, for sharing the internal transcribed spacer sequences of the patient isolates, permitting comparison with the environmental isolates.
This project was supported in part by the Association of Schools of Public Health, the Centers for Disease Control and Prevention, and the Agency for Toxic Substances and Disease Registry (cooperative agreement S1944‐21/23).
References
- 1. Gams W, McGinnis MR. Phialemonium, a new anamorph genus intermediate between Phialophora and Acremonium. Mycologia 1983; 75:977‐987.
- 2. McGinnis MR, Gams W, Goodwin MN Jr. Phialemonium obovatum infection in a burned child. J Med Vet Mycol 1986; 24:51‐55.
- 3. Magnon KC, Jalbert M, Padhye AA. Osteolytic phaeohyphomycosis caused by Phialemonium obovatum. Arch Pathol Lab Med 1993; 117:841‐843.
- 4. King D, Pasarell L, Dixon DM, McGinnis MR, Merz WG. A phaeohyphomycotic cyst and peritonitis caused by Phialemonium species and a reevaluation of its taxonomy. J Clin Microbiol 1993; 31:1804‐1810.
- 5. Heins‐Vaccari EM, Machado CM, Saboya RS, et al. Phialemonium curvatum infection after bone marrow transplantation. Rev Inst Med Trop Sao Paulo 2001; 43:163‐166.
- 6. Schonheyder HC, Jensen HE, Gams W, et al. Late bioprosthetic valve endocarditis caused by Phialemonium aff. curvatum and Streptococcus sanguis: a case report. J Med Vet Mycol 1996; 34:209‐214.
- 7. Gavin PJ, Sutton DA, Katz BZ. Fatal endocarditis in a neonate caused by the dematiaceous fungus Phialemonium obovatum: case report and review of the literature. J Clin Microbiol 2002; 40:2207‐2212.
- 8. Guarro J, Nucci M, Akiti T, et al. Phialemonium fungemia: two documented nosocomial cases. J Clin Microbiol 1999; 37:2493‐2497.
- 9. Strahilevitz J, Rahav G, Schroers HJ, et al. An outbreak of Phialemonium infective endocarditis linked to intracavernous penile injections for the treatment of impotence. Clin Infect Dis 2005; 40:781‐786.
- 10. Proia LA, Hayden MK, Kammeyer PL, et al. Phialemonium: an emerging mold pathogen that caused 4 cases of hemodialysis‐associated endovascular infection. Clin Infect Dis 2004; 39:373‐379.
- 11. American Public Health Association. Standard Methods for the Analysis and Water and Wastewater. 19th ed. 1998.
- 12. de Hoog GS, Guarro J, Gene J, Figueras MJ. Atlas of Clinical Fungi. 2nd ed. Utrecht, The Netherlands: Centraalbureau voor Schimmelkultures; 2000.
- 13. VandenBergh MF, Verweij PE, Voss A. Epidemiology of nosocomial fungal infections: invasive aspergillosis and the environment. Diagn Microbiol Infect Dis 1999; 34:221‐227.
- 14. Warris A, Voss A, Abrahamsen TG, Verweij PE. Contamination of hospital water with Aspergillus fumigatus and other molds. Clin Infect Dis 2002; 34:1159‐1160.
- 15. Warris A, Voss A, Verweij PE. Hospital sources of Aspergillus: new routes of transmission? Rev Iberoam Micol 2001; 18:156‐162.
- 16. Anaissie EJ, Stratton SL, Dignani MC, et al. Pathogenic Aspergillus species recovered from a hospital water system: a 3‐year prospective study. Clin Infect Dis 2002; 34:780‐789.
- 17. Grohskopf LA, Roth VR, Feikin DR, et al. Serratia liquefaciens bloodstream infections from contamination of epoetin alfa at a hemodialysis center. N Engl J Med 2001; 344:1491‐1497.
- 18. Arduino MJ. CDC investigations of noninfectious outbreaks of adverse events in hemodialysis facilities, 1979‐1999. Semin Dial 2000; 13:86‐91.
- 19. Center for Disease Control and Prevention. Exophiala infection from contaminated injectable steroids prepared by a compounding pharmacy—United States, July–November 2002. MMWR Morb Mortal Wkly Rep 2002; 51:1109‐1112.
- 20. Rudnick JR, Arduino MJ, Bland LA, et al. An outbreak of pyrogenic reactions in chronic hemodialysis patients associated with hemodialyzer reuse. Artif Organs 1995; 19:289‐294.
- 21. Welbel SF, Schoendorf K, Bland LA, et al. An outbreak of gram‐negative bloodstream infections in chronic hemodialysis patients. Am J Nephrol 1995; 15:1‐4.
- 22. NKF‐K/DOQI Clinical Practice Guidelines for Vascular Access: update 2000. Am J Kidney Dis 2001; 37(Suppl 1):S137‐S181.
- 23. Tokars JI, Arduino MJ, Alter MJ. Infection control in hemodialysis units. Infect Dis Clin North Am 2001; 15:797‐812, viii.