Legionella Case Studies
Cruise Ship Outbreak May 2003
More than 9.4 million passengers traveled on pleasure cruises departing from North American ports in 2004, an increase of 13% since 2003 and 41% since 2001.1 Cruise ships typically transport closed populations of thousands of persons, often from diverse parts of the world. Travelers are at risk for becoming ill while on board, most commonly from person-to-person spread of viral gastrointestinal illnesses. Certain environmental organisms, such as Legionella spp., pose a risk to vulnerable passengers. During November 2003–May 2004, eight cases of Legionnaires disease (LD) among persons who had recently traveled on cruise ships were reported to CDC. This report describes these cases to raise clinician awareness of the potential for cruise-ship–associated LD and to emphasize the need for identification and reporting of cases to facilitate investigation.
LD is a severe community– or health-care–associated pneumonia caused by Legionella spp., most commonly L. pneumonia. LD can result from inhalation or aspiration of warm (25°C–42°C), aerosolized water containing Legionella. Symptoms typically begin 2-10 days after exposure. Person-to-person transmission does not occur. Because symptoms of LD (e.g., fever, cough, or chest pain) are nonspecific, LD cannot be reliably distinguished from other forms of pneumonia on the basis of clinical presentation alone.
In the United States, LD can be reported to CDC through two surveillance systems. The National Electronic Telecommunications System for Surveillance collects information on all reportable diseases from state and territorial health departments but does not collect information on travel history. In contrast, the paper-based Legionnaires Disease Reporting System collects details of any recent travel from LD patients but receives data on only a fraction of the total cases estimated to occur. The cases described in this report were initially relayed to CDC by direct communication from state health departments, cruise lines, and the European Working Group for Legionella Infections (EWGLI), which operates a surveillance scheme (EWGLINET) for LD among European travelers (http://www.ewgli.org). Cases were defined as laboratory-confirmed LD in a person with cruise-ship travel during the 10 days before symptom onset. Exposure history was collected by the state and local health departments, and environmental samples, when obtained, were tested by contractors hired by the cruise lines.
The eight cases were among passengers who had been aboard five different cruise ships and associated with seven different voyages. Two of the eight cases occurred on the same voyage. The mean age of the patients was 55.8 years (range: 23-76 years). Five (63%) were male; seven (88%) were U.S. residents. The sole case in a foreign traveler occurred in a Dutch woman aged 23 years who had onset of fever and cough 4 days after returning from a cruise in the Caribbean. Two (25%) cases were fatal. Of the seven patients with known medical histories, six (86%) had comorbidities or risk behaviors known to be risk factors for LD (e.g., diabetes, history of heart disease, or smoking). The mean time from cruise-ship boarding to onset of symptoms was 10.4 days (range: 4-16 days). Although two passengers had symptoms before the end of their respective cruises, only one had LD diagnosed while still aboard the ship. Seven (88%) were diagnosed by urinary antigen testing for Legionella pneumophila serogroup 1 (Lp1). The only person with LD diagnosed by a fourfold increase in anti-Legionella spp. serology had a negative Legionella urinary antigen test. Only the Dutch traveler had a culture for Legionella obtained at the onset of illness. The culture was positive for Lp1; a urinary antigen test also was positive.
Two cases occurred on each of three cruise ships. Two patients were aboard the same ship during the same period but had been friends preceding the cruise and therefore had other exposures in common. A definite source of exposure could not be identified for any of the cases because of the limited number of cases. In addition, all but one patient lacked a clinical isolate, limiting the ability to link clinical and environmental isolates. For the Dutch passenger, the sole patient with a clinical isolate, environmental sampling was performed, but no matching environmental isolate was identified. Additional case-finding measures included review of infirmary records by cruise lines and CDC, passive surveillance by cruise lines, public health alerts via the Epidemic Information Exchange (Epi-X), and notifications to EWGLI in the event vacationing European travelers had become ill. Despite these activities, no other cases were identified.
_C Joseph, DrPH, EWGLINET, London, England. J van Wijngaarden, MD, Dutch Ministry of Health, Welfare, and Sport. P Mshar, MPH, Connecticut Dept of Health. C Oravetz, Flagler County Health Dept, Bunnell, Florida; AM Fix, MPH, Florida Dept of Health. CA Genese, MBA, New Jersey Dept of Health and Senior Svcs. GS Johnson, M Kacica, MD, New York State Dept of Health. B Weant, Guilford County Dept of Public Health, Greensboro, North Carolina; P Jenkins, EdD, North Carolina Dept of Health and Human Svcs. N Baker, MPH, Philadelphia Dept of Health. D Forney, J Ames, MPH, G Vaughan, MPH, Jon Schnoor, Vessel Sanitation Program, National Center for Environmental Health; D Kim, MD, M Guerra, DVM, Div of Global Migration and Quarantine; B Fields, PhD, M Moore, MD, Div of Bacterial and Mycotic Diseases, National Center for Infectious Diseases; C Newbern, PhD, M Thigpen, MD, EIS officers, CDC.
_CDC Editorial Note:
_During 1980-1998, CDC received an average of 360 paper-based reports of LD annually, primarily during summer months.2 However, previous research using population-based active surveillance estimated that 8,000-18,000 cases of Legionella spp. infection requiring hospitalization occur in the United States annually, suggesting that legionellosis is underdiagnosed and/or underreported.3 Since the first recognized outbreak of LD occurred in 1976 among persons attending the American Legion convention in Philadelphia, travel has been identified as a risk factor for both outbreak-associated4 and sporadic infection.5 However, for multiple reasons, outbreaks of travel-associated legionellosis are difficult to detect and investigate.6-7 First, trends toward empirical use of antimicrobial agents have led to declines in diagnostic testing for etiologic agents of community-acquired pneumonia.8 Second, the incubation period of 2-10 days allows travelers to return home before they have symptoms, making it unlikely for a medical provider to see more than a single case. Third, because LD can be diagnosed within hours of specimen collection by urine antigen testing, diagnosis by culture, which requires several days, has declined substantially in recent years.2
The lack of clinical isolates hinders epidemiologic investigations and prevention strategies. Legionella spp. can be identified by culture in up to 40% of freshwater environmental samples and in up to 80% of environmental samples by polymerase chain reaction.9 Although Lp1 causes approximately 70% of cases, at least 22 species of Legionella have been associated with disease in humans.9 To determine which of many potential environmental Legionella spp. is the causative organism, a clinical isolate from a respiratory culture must be matched to the environmental isolate by monoclonal antibody subtyping or by molecular methods. For these reasons, when evaluating a patient with suspected LD, clinicians should obtain a travel history and collect respiratory secretions for culture, in addition to collecting urine for antigen testing.
Reporting of LD is mandatory in every state. However, dispersion of travelers to multiple states after an exposure might result in a health department receiving only one report in association with a particular ship or hotel. Cruise-ship– associated travel poses additional difficulties for notification and investigation of LD cases. For cruise ships that sail in international waters, patients might be hospitalized in other countries, delaying or precluding reporting to authorities in the patients' home countries. Because travelers often stay in hotels before or after cruise-ship travel and often disembark at various international ports of call during a cruise, numerous potential sources exist for authorities to investigate. In certain instances, cruise-ship travel might be of insufficient duration (e.g., a single day or overnight trip) to be inclusive of the 2-10-day incubation period of LD. In addition, the limited number of reported cases associated with cruises limits the ability of traditional epidemiologic methods to identify a source. Thus, the task of identifying a source often relies on matching a clinical isolate to an environmental isolate. However, few cases have been reported for which an environmental isolate identified from a cruise ship (most often from a whirlpool spa) was identical to a clinical isolate from an ill passenger.6-7 Obtaining a clinical isolate from a patient with travel-associated LD is essential to identifying the source of infection.
Public health programs have focused on reducing the risk for LD among cruise-ship passengers. In 1994, CDC investigated an LD outbreak on board a cruise ship and subsequently issued recommendations to reduce transmission of Legionella spp. from shipboard whirlpool spas.10 In addition, CDC's Vessel Sanitation Program regularly conducts inspections of these spas and other environmental sources. Given the difficulties in confirming cases of LD, cooperation of clinicians and local, national, and international public health agencies is essential to foster diagnosis and prevention. Because a single case of LD in a traveler might indicate an outbreak, prompt recognition and direct reporting to local, state, and federal officials can prevent additional cases of travel-associated illness.
_1. US Department of Transportation. Maritime Administration. Data and statistics: cruise passenger statistics. Available at http://www.marad.dot.gov/marad_statistics.
2. Benin AL, Benson RF, Besser RE. Trends in Legionnaires disease, 1980-1998: declining mortality and new patterns of diagnosis. Clin Infect Dis. 2002;35:1039-1046. FULL TEXT | ISI | PUBMED
3. Marston BJ, Plouffe JF, File TM, et al. Incidence of community-acquired pneumonia requiring hospitalization: results of a population-based active surveillance study in Ohio. Arch Intern Med. 1997;157:1709-1718. ABSTRACT
4. Fraser DW, Tsai TR, Orenstein W, et al. Legionnaires' disease: description of an epidemic of pneumonia. N Engl J Med. 1977;297:1189-1197. ABSTRACT
5. Straus WL, Plouffe JF, File TM, et al. Risk factors for domestic acquisition of Legionnaires disease. Arch Intern Med. 1996;156:1685-1692. ABSTRACT
6. Jernigan DB, Hofmann J, Cetron MS, et al. Outbreak of Legionnaires' disease among cruise ship passengers exposed to a contaminated whirlpool spa. Lancet. 1996;347:494-499. FULL TEXT | ISI | PUBMED
7. Ricketts K, Joseph C, European Working Group for Legionella Infections. Travel-associated Legionnaires' disease in Europe: 2003. Euro Surveill. 2004;9:40-43. PUBMED
8. Bartlett JG. Decline in microbial studies for patients with pulmonary infections. Clin Infect Dis. 2004;39:170-172. FULL TEXT | ISI | PUBMED
9. Fields BS. Legionellae and Legionnaires' disease. In: Hurst CJ, Crawford RL, Knudsen GR, McInerney MJ, Stetzenbach LD, eds. Manual of environmental microbiology. 2nd Edition. Washington, DC: ASM Press; 2001:666.
10. CDC. Final recommendations to minimize transmission of Legionnaires' disease from whirlpool spas on cruise ships. Atlanta, GA: US Department of Health and Human Services, CDC; 1997.
Car plant workers in Ohio 2001
During March 12-15, 2001, four cases of Legionnaires' disease (LD) among workers at an automotive engine manufacturing plant (plant X) were reported to the Cuyahoga County Board of Health, Cleveland, Ohio; all four diagnoses were confirmed by Legionella urine antigen. Illness onset among the four workers occurred during March 2-4; two workers died. Beginning March 14, CDC assisted state and local health departments with an investigation to identify new cases and potential sources of Legionella transmission in the plant. This report summarizes the investigation; findings indicate an epidemiologic association with exposure to one of the plant finishing lines but did not identify a specific source.
Plant X manufactures cast iron engine components, is operated by approximately 2500 employees, and covers approximately 1.6 million square feet of floor space. The plant is divided into four areas: core making, mold production, iron melting, and finishing. A confirmed case of LD was defined as radiograph-confirmed pneumonia and laboratory evidence of Legionella infection, defined as a positive Legionella urine antigen or isolation of Legionella from respiratory secretions or lung tissue. Specimens from the four initial case-patients were sent to CDC for isolation of Legionella; available specimens included one sputum specimen, one broncho-alveolar lavage specimen, and lung tissue from the two decedents. Active LD surveillance was established in all hospitals in the greater Cleveland area. Hospital records and plant X employee absentee records were reviewed to identify additional cases. An environmental investigation was conducted to identify aerosol-producing water sources for Legionella transmission, including cooling towers, water hoses, and water heaters.
No additional confirmed LD cases were identified among the workers. Nine workers from plant X were hospitalized during February 14–March 28; four had pneumonia, and all nine had negative Legionella urine antigen tests. Legionella pneumophila, serogroup 1, was isolated from a worker's sputum sample, which was stored at 40°F (4°C) for greater than1 week before culture. Results are pending from lung tissue samples. Legionella was isolated from 18 (9%) of 197 environmental samples, and at least five species were identified. Three samples grew L. pneumophila, serogroup 1; none matched the clinical isolate by monoclonal antibody staining.
A case-control study was conducted to determine risk factors for exposure to Legionella among plant workers. A case-patient was defined as a worker at plant X during February 14–March 28 who had either a confirmed case of LD or a possible case of legionellosis. A possible case-patient of legionellosis was defined as a worker with a titer of anti-legionella IgG antibody 1:1024 and any two of the following symptoms: cough, shortness of breath, fever, headache, myalgia, or fatigue. Controls were randomly selected workers with fewer than two symptoms and IgG antibody 1:64. Serologic specimens were collected 4-5 weeks after the presumed exposure. Each study participant was asked detailed questions about time spent inside and outside of the plant and information about underlying medical conditions associated with LD.
Among 855 workers who were contacted, 484 (57%) agreed to participate in the case-control study; 11 met case criteria (four confirmed and seven possible cases), and 105 met criteria for controls. Visiting one of the finishing lines in the plant (odds ratio [OR] = 15.1; 95% confidence interval [CI] = 3.0-76.2) and working in the finishing region of the plant (OR = 3.8; CI = 1.0-13.8) were associated with disease.
Plant X was closed during March 14-19 to facilitate environmental sampling and decontamination. All water systems were decontaminated, and ongoing environmental surveillance for Legionella was implemented throughout the plant, including the finishing area. Sources of aerosolized water from the finishing area that had been sampled before decontamination did not yield cultures positive for Legionella. On the basis of the case-control study results, additional environmental samples were collected in the finishing area on April 14; all samples were negative for Legionella. County health officials are obtaining maintenance records from the implicated area of plant X to determine how transmission might have occurred.
T Allan, T Horgan, H Scaife, Cuyahoga County Board of Health, Cleveland; E Koch, S Nowicki, MK Parrish, E Salehi, Ohio Dept of Health. Respiratory Diseases Br, Div of Bacterial and Mycotic Diseases, National Center for Infectious Diseases; Hazard Evaluations and Technical Assistance Br, Div of Surveillance, Hazard Evaluations and Field Studies, National Institute for Occupational Safety and Health; and EIS officers, CDC.
_CDC Editorial Note:
Industrial plants can be a source for the propagation and transmission of Legionella. The identification of L. pneumophila in the environmental samples demonstrated that legionellae can survive in this work environment. The tightly clustered onset of illness, lack of other epidemiologic associations among the four confirmed patients besides working in plant X, and the results of the case-control study implicated a particular finishing line within the plant as the likely source of Legionella. The narrow period of illness onset and the failure to identify new cases among plant workers suggest that exposure to the infecting Legionella strain was short-lived and transient, which may explain the failure to find an environmental sample that matched the clinical isolate.
LD outbreaks have been reported in industrial settings, including an automotive plant where workers were exposed to contaminated metal-working fluids,1 factories that used water to cool molded plastics,2 and waste-water treatment facilities.3 In each setting, an aerosol-producing device was implicated. Guidelines to minimize the risk for Legionella transmission in these sites are available.4 In addition to LD, clinicians should consider hypersensitivity pneumonitis, metal fume fever, and humidifier fever as possible diagnoses of an acute febrile respiratory illness with systemic symptoms in persons who work in an industrial setting.5
Legionella species are estimated to account for 2%-15% of all community-acquired pneumonia; however, only 1200-1500 cases are reported annually.6,7 Appropriate diagnostic testing for LD includes Legionella urine antigen and culture of respiratory secretions. Legionella urine antigen tests provide rapid and accurate diagnosis of disease caused by L. pneumophila, serogroup 1; however, these tests do not identify less common species or serogroups and do not provide an isolate necessary to compare clinical with environmental isolates during outbreak investigations. LD also can be diagnosed by a four-fold rise in anti-legionella antibody titer or by direct fluorescent antibody on sputum samples, although the latter method lacks specificity and sensitivity. In addition to testing for Legionella urine antigen, the diagnosis and investigation of LD cases would be improved if clinicians obtained respiratory specimens for culture by a laboratory proficient in Legionella isolation. To facilitate appropriate investigation and improve understanding of disease associated with Legionella species, health-care providers should report legionellosis cases to county or state health departments, and state health departments should report legionellosis cases to CDC.
Murcia Spain outbreak
This LD outbreak is the largest to date in the world, with 449 confirmed cases and an estimated total number of cases of 650. The reported case-fatality rate (1%) is much lower than those observed in other community outbreaks. This rate can be attributed, at least partially, to the quick detection of the outbreak, early diagnosis of the disease, and appropriate treatment of patients. The explosive quality of the outbreak not only led patients to seek quick assistance at hospital emergency units but also helped clinicians to perform an accurate diagnosis and to immediately initiate adequate treatment, factors reported as linked to low case fatality.This explosive appearance could also be related to a lower presence of predisposing factors in case-patients in comparison with other community outbreaks,[which could also partially explain the low case-fatality rate.
The initial investigation encountered obstacles, such as a large number of potential sources of environmental contamination located in the northern part of the city and the absence of environmental Legionella isolates identical to those of patients. The case-control study showed a significant association, with a high consistency between the analyzed models and with a high magnitude of association, between passing through the zone around hospital H and being ill with LD. Results were similar even when the area radius was expanded to 400 m. However, large overlap of areas was observed within this radius, and multicollinearity among zones was a common finding.
The case-control study was designed to select patients residing outside the city of Murcia. We decided on this approach for two reasons. First, the incidence of LD was almost 1% in some neighborhoods, a rate within the 0.1% to 5% attack rate described for this disease. Therefore, all the persons living in these quarters could possibly have been exposed to Legionella, as has been described in outbreaks of other transmissible diseases. If everyone had been exposed, finding incidence differences between persons exposed and those not exposed would have been almost impossible. Second, persons residing outside the city would probably have a more accurate memory of the itineraries they followed in Murcia some weeks previously and would probably have a lesser number of routes than persons living within the city. Conducting 255 personal interviews with questions about itineraries within 2 weeks from the last case and 4 weeks from the outbreak onset may also have been important to our findings.
One concern in case-control studies is that participation rate is not reported consistently. Indeed, this information is usually omitted in case-control studies of outbreaks, especially when controls are selected from a population database, as was our situation. A further complication was that the study had to be conducted in July, when many people go on holiday. In spite of achieving the participation of one in two controls whom we initially selected, we evaluated possible selection bias. We determined that it was unlikely to have occurred since neither socioeconomic status nor predisposing risk factors for LD differed significantly among cases and controls. Information bias overestimating this outcome was ruled out since news media did not mention hospital H among the probable sources of the outbreak.
Meteorologic conditions were favorable for the emission of aerosols to be dispersed in a horizontal manner. Low wind speed together with atmospheric thermal inversion between June 29 and July 1 would have facilitated the presence of the aerosols in the environment.
The result of the epidemiologic study was subsequently confirmed by the isolation of a strain retrieved on October 30 from a sample from one of the cooling towers of the same hospital; that strain is identical to the strain isolated from the patients. The difficulties found previously in the isolation of this strain were not unexpected. The day after the outbreak was detected, when the first sample was taken, the cooling towers of hospital H were highly chlorinated, which could explain why these first samples gave negative results. Later samples retrieved on four different dates between July 28 and September 13 showed positive results to L. pneumophila but were characterized as different strains from those from patients. This strain was only isolated upon the restarting of one tower after it was shut down for more than 1 month, a condition that favors the reappearance of Legionella.The fact that the same clone of Legionella can be found in an installation for long periods is also documented.The possible contamination of the tower by new Legionella from the water supply was ruled out since the strain linked to the outbreak was not found in samples collected from many other installations during the same period, including July to November.
The coincidence of a nosocomial LD outbreak in hospital H reinforces the previous hypothesis. A nosocomial outbreak of LD as part of a wider community outbreak of the disease has been described, although in other outbreaks originating in the cooling towers of a hospital no cases of nosocomial LD were identified.The use of double HEPA filters on air-intake vents in some hospitals could justify, at least in part, these contradictory observations.
Our research indicates that the cooling towers of a hospital located in the northeastern part of the city of Murcia were the origin of this community outbreak. This study underlines important risk factors that must be taken into account to prevent a new LD outbreak. First, cooling towers had to be identified by aerial and direct inspection in the absence of any census of such installations. Second, the size, location, and state of maintenance of cooling towers are very important. In contrast with epidemics associated with relatively small systems, this outbreak was related to a large refrigeration system that seems to have infected patients up to 1.3 km downwind to the west from the cooling tower; this finding suggests that airborne infection with L. pneumophila may extend over a large distance from the dissemination source, as has been reported elsewhere. Although most of the installations in the area showed inadequate maintenance, the cooling towers from hospital H were poorly maintained and had a high-risk size and location. Once the outbreak was identified, urgent measures were undertaken to clean, disinfect, or close possibly contaminated sources. The cooling tower that was the source of the outbreak was subsequently replaced by an air-cooled system.
Before June 2001, no specific national legislation existed in Spain concerning LD, although a recommendation guide and legislation existed in several Spanish autonomous regions that had had community LD outbreaks. As an immediate consequence of this outbreak, a national law about prevention and control of LD was enacted in Spain 20 days after the outbreak began. The extent of this outbreak is useful to assess the relative role of cooling towers as a source of LD and highlights the importance of prioritizing control measures related to cooling towers among strategies to prevent LD in the community. Compliance with these measures would help to reduce not only community outbreaks but also, perhaps, sporadic cases that could be due to infected cooling towers.