Each year, hospital-acquired legionellosis cases result in prolonged hospitalization with elevated mortality rates.Reference Collier, Stockman, Hicks, Garrison, Zhou and Beach 1 These cases are predominantly associated with Legionella pneumophila serogroup 1 (sg1) strains present in the hospital hot water systems.Reference Demirjian, Lucas and Garrison 2 – Reference Durando, Orsi and Alicino 4 A few outbreaks and isolated cases have been related to L. pneumophila (Lp) serogroup 5 (sg5).Reference Perola, Kauppinen, Kusnetsov, Karkkainen, Luck and Katila 5 – Reference Mineshita, Nakamori, Seida and Hiwatashi 7 Several factors can contribute to Legionella growth and persistence within hospital water systems: temperature, stagnation, biofilm, material, disinfectant, and water quality.Reference Buse, Schoen and Ashbolt 8 – Reference Moritz, Flemming and Wingender 12 Key measures to control Lp in hot water systems are to maintain elevated water temperatures throughout the system and to minimize stagnation through optimal water circulation. 13 – 16 Although an infectious dose has not been determined, several countries have established action levels between 1,000 and 10,000 colony-forming units (CFU)/L, and a concentration higher than 10,000 CFU/L requires immediate corrective actions. 13 , 15 , 17 – 19
At the same time, healthcare facilities are encouraged to implement energy and water conservation devices to meet accreditation requirements, such as Leadership in Energy and Environmental Design certification. The use of waste heat recovery systems to preheat hospital hot water prior to the water heater is an option offered to hospitals to increase energy efficiency. However, operational practices and ideal growth conditions associated with these devices can promote the development and persistence of Lp unless a thorough risk assessment is performed. In this study, we report a nosocomial outbreak of Lp sg5 in the year following the installation of 2 heat exchangers in 2 distinct hospital wings as part of an energy conservation upgrade. The objectives of our study were to determine the source of the outbreak and to understand the role of the heat recovery systems.
METHODS
Two nosocomial cases of Lp were reported in August 2014 within a wing (wing A) of a 400-bed tertiary care university hospital in Sherbrooke, Canada. Clinical samples were cultured on buffered charcoal-yeast extract (BCYE) medium, and isolates were sent to the Laboratoire de Santé Publique du Québec for identification and serogroup confirmation as described previously.Reference Levesque, Plante and Mendis 20 Following the reported cases, 250 mL of first flush hot water were collected from 25 taps (baths and sinks) within wing A (300 beds) and from 9 taps within wing B (100 beds, supplied by a separate hot water system). The heat exchangers from wings A and B were investigated in June and July 2015, respectively. Biofilm samples and water (2 L) were collected from 3 locations: at the water inflow pipe, inside the heat exchanger, and at the water outflow pipe. Two additional biofilm samples were recovered within the heat exchanger. Environmental samples were cultured according to the Association Française de Normalisation NF T90-431 method 21 with the addition of a 1-mL filtration. Briefly, different volumes were filtered through sterile 47 mm diameter 0.45 µm mixed ester cellulose membranes (Millipore) and an untreated sample volume of 0.2 mL was plated on glycine-vancomycin-polymyxin-cycloheximide selective agar (Biokar Diagnostics) and incubated at 36°C for 10 days. Before plating, acid treatment was applied to filtered samples (pH, 2; 5 min). Typical colonies that developed after 4 to 10 days were subcultured on confirmation plates (BCYE agar without and with cysteine) for 2 to 4 days, at 36°C. Resulting colonies that developed on BCYE agar, but not on BCYE without cysteine, were considered as Legionella spp. The Legionella latex test (DR0800; Oxoid) was used for Lp confirmation. The calculated detection limit for the culture method was 10 CFU/L for both Legionella spp. and Lp. Pulsed-field gel electrophoresis (PFGE) and sequence-based typing (SBT) were performed as described previously.Reference Levesque, Plante and Mendis 20 In some cases, nested-SBT protocol was performed on DNA extracts obtained from water sampling.Reference Ginevra, Lopez and Forey 22 PFGE patterns were analyzed according to the criteria of Tenover et alReference Tenover, Arbeit and Goering 23 and with BioNumerics software, version 6.5 (Applied Maths), by the unweighted pair-group method with arithmetic average clustering method using the Dice coefficient with both position tolerance and optimization of 1%. Biofilm and water samples from the heat exchanger were analyzed for the presence of protists through direct microscopy and 18s polymerase chain reaction amplification using the following primers: Euk1A (5′-CTGGTTGATCCTGCCAG-3′); Euk516r (5′-ACCAGACTTGCCCTCC-3′).
Statistical analysis (Z-test and Kruskal-Wallis) was performed with Statistica, version 10 (Dell), to compare the percent positivity and level of Lp in wing A and wing B. The significance level was set at P=.05.
RESULTS
Nosocomial legionellosis was diagnosed in 2 patients hospitalized in wing A (August 2014), and clinical specimens were positive for Lp sg5. The first case occurred in the oncology ward, on the seventh floor of wing A. The patient was admitted on July 12 with acute myeloid leukemia. The first signs of pneumonia appeared on August 1. A sample was collected through bronchoalveolar lavage and the presence of Lp sg 2–14 was confirmed. The second case occurred in a patient admitted on July 25 to the medical intensive care unit (ninth floor, wing A) for a third-degree atrioventricular block. His main comorbidity was a chronic obstructive pulmonary disease for which he took prednisone. He had a permanent pacemaker placement on July 26 and was transferred to the surgical ward (eighth floor, wing A), where he stayed from July 27 to August 5. Significant coronary artery disease was discovered and he underwent coronary artery bypass graft surgery on August 5. Following this intervention, he was transferred to the surgical intensive care unit (third floor, wing B), where he remained for the rest of his stay. The first signs of pneumonia appeared on August 8 and clinical samples were recovered through bronchoscopy. The presence of Lp sg 2–14 was also confirmed. Provided the incubation time of 2 to 10 days, the patient was possibly exposed during the 3-day stay in wing B (intensive care unit) but most probably during the 10-day stay in the surgical ward in wing A.
Environmental investigation from the hot water systems revealed a large proportion of Lp sg5 positive taps with high levels of contamination (88% in wing A and 56% in wing B; Table 1), whereas there was no Lp detected in the cold water feeding into the hospital.Reference Bédard, Boppe and Kouamé 9 The percentage of positive taps and the level of contamination by Lp were significantly higher in wing A compared with wing B. A copper-silver ionization treatment was present on both hot water systems at the time of the outbreak. Disinfection by heat shock (≥60°C for ≥7 minutes at each tap) was conducted as previously describedReference Bédard, Boppe and Kouamé 9 in wing A (August 2014) and in wing B (September 2014). The disinfection was followed by the implementation of a higher temperature set point at the water heater outlet (≥60°C) in both wings. Following temperature corrective measures, no cases of legionellosis were reported despite written directives asking hospital physicians to obtain Legionella cultures of respiratory specimens for all cases of nosocomial pneumonia. No Lp were detected in any of the water and biofilm heat exchanger samples from wing B (including the feed and outflow water samples), whereas high positivity for Lp sg5 was detected in wing A in water samples (3/3) and swabs from the inner surface of the heat exchanger (4/5). A gradient of Lp was observed in water samples from wing A heat exchanger: 510 CFU/L in the feed water, 5,000 CFU/L in the heat exchanger water, and 88,000 CFU/L in the outflow water, prior to the water heater. Protists were not readily observed by microscopy in any of the collected samples. Polymerase chain reaction also failed to detect protists in the biofilm swabs and in water samples. The heat exchangers were stopped at the time of the sampling and have not been back in service since.Reference Bédard, Boppe and Kouamé 9
NOTE. CFU, colony-forming units.
The hospital hot water was preheated with residual energy from the building heating system loop with single-pass heat exchangers (34 plates in wing A and 21 plates in wing B). The available surfaces were estimated at 15 m2 (wing A) and 5.5 m2 (wing B), with water volumes of 11.4 L and 3.8 L, respectively, resulting in a very high surface-to-volume ratio of 14 cm−1. Temperatures within the heat exchangers ranged from 9°C to 46°C, and prolonged stagnation was observed during the night, resulting in no flow for 48% and 51% of the time in wing A and wing B, respectively. The average flow rates during daily operation were estimated to be 18 L/min in wing A and 43.8 L/min in wing B, well below the maximum designed flow rate of 230 L/min. Figure 1 illustrates the flow diagram of the heat exchangers for both wings. The hot water temperature feeding into both distribution systems was less than 55°C before the outbreak.
In total, 34 clinical and environmental isolates (wing A), 7 environmental isolates (wing B), and 4 environmental isolates collected in 1995 (wing A) were typed by PFGE and SBT (Figure 2). Environmental isolates dating from 1995 were collected from the hot water system in wing A as part of a previous case investigation, where the source of infection was found to be unrelated to this hospital. Genotyping results revealed that all isolated environmental strains (1995 and 2014) harbored the same related PFGE pattern as the outbreak strain (Figure 2). All typed isolates were also from the same SBT type (ST-1427). Partial SBT profiles were obtained from 12 additional hot water sample DNA extracts using the nested-SBT protocol (Table 2), collected a year after the outbreak.Reference Bédard, Boppe and Kouamé 9 The obtained alleles were similar to ST-1427 for 11 of the 12 DNA extracts, suggesting the presence of the same strain in those water samples.
NOTE. F, polymerase chain reaction amplification failed; NEW, allele number pending; ST, sequence type.
DISCUSSION
The environmental investigation indicated the hot water system as the most probable source of the outbreak. Although Lp was detected in both systems investigated, the level of contamination was significantly higher in wing A. Despite the observed reduction in hot water contamination levels after the heat shock disinfection,Reference Bédard, Boppe and Kouamé 9 Lp was still detected by culture in more than 45% of faucets and in the recirculation water for both wings in February 2015, 6 months after the implementation of corrective measures. The high level of Lp outbreak strain detected in all water samples from the heat exchanger in wing A at the time of sampling suggests its potential colonization and role at the time of the outbreak. Although no protists could be isolated, their presence in heat exchangers should also be monitored in light of the manufacturer’s warning of biofouling risk due to organisms such as protozoa, a natural reservoir for Lp.Reference Lau and Ashbolt 24
The relatedness of the environmental strains (n=4 in 1995; n=39 in 2014–2015) and patient strains (n=2) confirms a system-wide contamination with the established Lp sg5 strain and suggests its persistence over a period of 20 years in the hot water system. The heat exchanger in wing A promoted the Lp sg5 genotype present in the system, whereas no colonization of the wing B heat exchanger was identified. Detailed investigation of the flow diagrams and onsite validation showed important differences between the design and operation of the 2 hot water systems, including the heat exchanger configuration (Fig. 1). The heat exchanger from wing A was fed by a combination of cold makeup water and recirculated hot water depending on demand, and up to 48% of the recirculated water did not transit through the flash water heater. The risk of Lp proliferation in heat exchangers is exacerbated by (1) the prevailing environmental conditions (eg, temperature, surface area, surface-to-volume ratio, materials); (2) operational conditions (eg, low flow, stagnation); and (3) the microbial load and presence of Lp in the feed water, which was the case in wing A for the recirculated water feed. The presence of the outbreak strain in the system for the past years combined with feeding contaminated recirculated water into a heat exchanger providing ideal growth conditions likely culminated in the higher Lp loads measured in wing A and a higher risk of infection. The contamination observed in wing B was associated with areas having recirculation deficiencies, preventing the hot water temperature from being maintained in these areas. Resolution of the identified deficiencies contributed to the reduction in Lp contamination observed.Reference Bédard, Boppe and Kouamé 9
The physical characteristics and operating conditions of heat exchangers provide ideal conditions for biofilm formation and the development of opportunistic pathogens. In the present study, although physical characteristics of the heat exchangers were similar in the 2 wings, the piping diagram and operating conditions were different. Design and operation of hot water system should prevent Lp proliferation and prevent the conditions in which amoebae-hosting biofilms develop. Furthermore, hot water system operators should not rely on the passage of water through the water heater to inactivate Lp from the recirculating stream and makeup preheated water. Short exposure to elevated temperatures may not be sufficient to inactivate certain strains of Lp shown to resist high temperatures (70°C) for prolonged periods (60 minutes) and to develop heat resistance after shock treatment.Reference Allegra, Berger, Berthelot, Grattard, Pozzetto and Riffard 25 , Reference Allegra, Grattard, Girardot, Riffard, Pozzetto and Berthelot 26
Our study was subjected to a number of limitations. First, although the same strain was isolated in the system in 1995, we could only hypothesize that it was present continuously in the system over the 20 years preceding the outbreak in 2014 since there was no monitoring of Lp performed in the hot water system during those years. Second, PFGE could not be performed on strains isolated from the hot water system samples 1 year after the outbreak. Third, owing to the urgency to apply corrective measures, the observed decrease in system contamination cannot be attributed to a specific corrective measure but rather to the sum of actions that were taken to eradicate the contamination. Finally, sampling of the heat exchangers was not performed at the time of the outbreak because they were not suspected initially.
In healthcare facilities serving patients more vulnerable to legionellosis, the risk associated with the installation of such devices needs to be carefully evaluated with respect to the important costs associated with legionellosis hospitalization (US $34,000/episode) and the elevated mortality rate.Reference Collier, Stockman, Hicks, Garrison, Zhou and Beach 1 Estimated energy savings in the present case study ranged from US $700 to US $1,700 per month. The addition of energy conservation devices and operational procedures should be evaluated by the water safety committee together with the infection prevention and control team, and weighed against the risk of exposing patients and the burden of preventive monitoring.
ACKNOWLEDGMENTS
We thank chair staff, especially Yves Fontaine and Jacinthe Mailly, and the participating healthcare facility personnel, especially Serge Kouamé and Jules Racine.
Financial support. Natural Sciences and Engineering Research Council of Canada through the Industrial Chair on Drinking Water and a Fonds de recherche du Québec–Nature et technologies Team Research Project grant (to S.P.F. and M.P.).
Potential conflicts of interest. All authors declare no conflicts of interest relevant to this article.