Legionella species have been recognized as human pathogens since 1977. Since then, more than 50 species have been identified, at least half of which have been associated with human disease. Despite this, surveillance programmes in Europe  and the USA  have consistently shown Legionella pneumophila to be the predominant cause of Legionnaire’s disease, accounting for 90% or more of cases. Cases caused by Legionella longbeachae come a distant second or third, accounting for 5% or less of cases. In contrast, surveillance data from Australia and New Zealand [3,4] have consistently shown L. longbeachae to be either the leading cause of Legionnaires’ disease or a close second to L. pneumophila. A recent survey of rural Thailand also found L. longbeachae to be the predominant cause of Legionnaires’ disease there .The clinical features and outcomes of Legionnaires’ disease caused by L. pneumophila have been extensively studied, but data on disease caused by L. longbeachae are sparse. A review by Muder and Yu in 2002  suggested that disease caused by non-L. pneumophila species was clinically similar to that caused by L. pneumophila, but occurred mainly in immunosuppressed patients. This suggestion was based mainly on data from disease caused by Legionella micdadei and other non-L. longbeachae species. The Canterbury region of New Zealand, in common with the rest of Australasia, experiences a high rate of Legionnaires’ disease caused by L. longbeachae. We sought to investigate whether the clinical features and outcomes associated with this were different from those associated with disease due to L. pneumophila.
From the Canterbury Health Laboratories database, we identified all clinical specimens with growth of Legionella species from 1998 to 2008 for patients from the Canterbury region. The case notes were retrospectively reviewed. All specimens had originated from patients admitted to Christchurch hospital with a clinical illness consistent with Legionnaires’ disease. Mortality data were obtained from the New Zealand National Registry of Births and Deaths. Data prospectively collected by the local Public Health Unit were reviewed. Dichotomous and continuous variables were tested by Fisher’s exact test and t-test, respectively.
Definitions of hospital-acquired and healthcare-associated disease were as used in the American Thoracic Society/Infectious Diseases Society of America guidelines . For CURB-65  and systemic inflammatory response score  criteria, confusion was scored if any mention was made of new confusion in the admission notes. Other data were derived from observations and blood tests performed on admission.
We identified 50 episodes of culture-positive Legionnaires’ disease in 48 unique patients (two patients had two separate episodes of disease caused by L. longbeachae). L. longbeachae was identified in 24 (48%) episodes, L. pneumophila in 20 (40%), and other Legionella species in six (12%). Patient characteristics, including risk factors for immunosuppression and underlying lung disease, are shown in Table 1. In episodes where criteria as defined by Cunha  could be applied (adults, no concomitant heart rate-lowering medications, or arrhythmia and temperature >38.9°C), relative bradycardia was present in 10 of 12 (83%) episodes caused by L. longbeachae and in eight of 14 (57%) episodes caused by L. pneumophila (p 0.21). In both groups, effective antibiotics (any of macrolide, quinolone or rifampicin) were started a median of 1.0 days after admission. The other admission details are shown in Table 2. No patients had other potential pathogens identified on admission. Among episodes caused by L. longbeachae, subsequent respiratory specimens from four patients during the admission yielded the following organisms: herpes simplex virus, Pseudomonas aeruginosa, Klebsiella pneumoniae, and Acinetobacter species. Among episodes caused by L. pneumophila, subsequent specimens from five patients yielded the following organisms: Aspergillus fumigatus, K. pneumoniae, Escherichia coli, Serratia species and herpes simplex virus. None of these isolates resulted in a change of treatment.
Table 1. Patient characteristics
| || Legionella longbeachae || Legionella pneumophila |
|Median age, years (range)||73.2 (51–86)||59.3 (12–89)*|
|Males, no. (%)||20 (83)||16 (80)|
|Acquisition, no. (%)|
| Community||24 (100)||13 (65)|
| Healthcare-associated||0||5 (25)*|
| Hospital-acquired||0||2 (10)|
| International travela||2 (8)||3 (15)|
|Predisposing factors, no. (%)|
| Smokerb||5 (21)||6 (30)|
| Diabetes||3 (13)||1 (5)|
| Chronic renal impairment||3 (13)||0|
| COPD||1 (4)||2 (10)|
| Liver failure||0||0|
| Haematological malignancy||4 (17)||2 (10)|
| Solid organ transplant||1 (4)||0|
| HIV infection||0||1 (5)|
| Immunosuppresive medicationsc||5 (21)||6 (30)|
Table 2. Admission details and outcomes
| || Legionella longbeachae || Legionella pneumophila |
|Season of admission, no. (%)|
| Summer (December–February)||9 (38)||2 (10)*|
| Autumn (March–May)||6 (25)||6 (30)|
| Winter (June–August)||0||4 (20)*|
| Spring (September–November)||9 (38)||8 (40)|
|Observations on admission (median)|
| Heart rate||95/min||102/min|
| Respiratory rate||20/min||24/min|
|Severity score at admission|
| CURB-65 (mean)||2.3||1.6|
| SIRS (mean)||2.7||2.8|
|Reported symptoms at admission|
| Days of symptoms (median)||4||5|
| Respiratory, no. (%)a||17 (71)||16 (80)|
| Gastrointestinal, no. (%)b||9 (38)||8 (40)|
| Confusion, no. (%)||4 (17)||4 (20)|
| Headache, no. (%)||7 (29)||3 (15)|
| Myalgia, no. (%)||4 (17)||3 (15)|
|Laboratory findings on admission|
| Anaemia, no. (%)||9 (38)||8 (40)|
| Thrombocytopenia, no. (%)||2 (8)||7 (35)|
| Acute renal failure, no. (%)c||10 (42)||6 (30)|
| Hyponatraemia, no. (%)||13 (54)||13 (65)|
| Abnormal liver function tests, no. (%)d||19 (88)||17 (89)|
| Total leukocyte count (mean)||14.1 × 109/L||16.8 × 109/L|
|Radiology on admission, no. (%)|
| Lobar consolidation||13 (54)||12 (60)|
| Multilobar consolidation||9 (38)||3 (15)|
|Pulmonary infiltrates||0||1 (5)|
| Pleural effusion||2 (8)||2 (10)|
| Not done||2 (8)||4 (20)|
|Interventions in ICU|
| Episodes requiring admission, no. (%)||10 (42)||8 (40)|
| Days in ICU (mean)||13.0||13.5|
| Invasive ventilatione||9.6||11|
| Inotrope supporte||2.0||3.6|
| Renal replacement therapye||2.3||2.5|
| Inpatient mortality, no. (%)||5 (21)||5 (25)|
| Follow-up film obtained, no. (%)||13 (54)||8 (40)|
| Weeks from discharge (median)||6.0||7.9|
| Persisting changes, no. (%)||4 (31)||2 (25)|
In the 12 months after discharge, there was one death following an episode caused by L. pneumophila. A 69-year-old woman with a background of rheumatoid arthritis, interstitial lung disease and chronic, active hepatitis on long-term azathioprine and prednisone died following an episode of respiratory failure of unknown cause 3 months after discharge. There were four re-admissions following episodes caused by L. pneumophila (one each of: pneumonia with no identified pathogen, cholelilithiasis, atrial fibrillation, and possible seizures) and two following episodes caused by L. longbeachae (one each of: spontaneously resolving supraventricular tachycardia and left ventricular failure).
To our knowledge, this is the first study to compare clinical features, risk factors and outcomes of patients from the same population hospitalized with Legionnaires’ disease caused by L. longbeachae and L. pneumophila. It also represents one of the largest case-series of Legionnaires’ disease caused by L. longbeachae reported to date.
Canterbury Health Laboratories is the only facility to offer Legionella culture within the region, and therefore we will have captured all culture-positive cases. Only culture-proven cases were included, as species identification by serology may be unreliable, and an equivalent antigen test for L. longbeachae is unavailable. Previous studies have shown that 4–11% of 300–350 admissions with pneumonia to Christchurch Hospital are due to Legionella species [11,12]. We therefore estimate that our series represents 11–36% of all cases of Legionnaires’ disease requiring hospitalization over this period.
We found that the clinical features, including markers of disease severity (CURB-65, systemic inflammatory response score criteria, and intensive-care unit treatment), and outcomes are broadly similar for disease caused by the two organisms. The only significant differences were in age at onset, season of admission, and place of acquisition. One possible explanation is the known association between exposure to potting soils and L. longbeachae infection. Canterbury has a temperate climate, and gardening activity peaks in the warmer months. The proportion of the population reporting participation in gardening also increases with age, being particularly high in those over 65 years of age . L. pneumophila has been intermittently isolated from the hospital water systems over this time, increasing its association with healthcare-related infections.
The prevalence of predisposing factors for immunosuppression or respiratory disease was not significantly different between the two groups. This contrasts with the situation in Europe and the USA, where disease caused by L. longbeachae occurs predominantly or exclusively in immunosuppressed patients [14–17]. Case-series from Australia [18,19] suggest that immunosuppression is less apparent in cases seen there, but none have directly compared with cases of Legionnaires' disease caused by L. longbeachae disease caused by L. pneumophila. Whether disease caused by L. longbeachae is different in Australasia than in Europe and the USA is unclear. A study by Doyle et al.  suggested that strains of L. longbeachae originating from Australian clinical samples were more virulent than those from the northern hemisphere in an animal model and in vitro.
Both L. pneumophila and L. longbeachae are important causes of Legionnaires’ disease in Canterbury, and the clinical presentation and outcomes associated with these organisms are broadly similar. This has important implications for the reliance on urinary antigen testing in the exclusion of Legionnaires’ disease.