3.1 Literature review: Legionnaires' disease outbreak response

Fredrikstad and Sarpsborg, Norway, May 2005 (Nygard et. al, 2008)[1]

On the 21 ^st May 2005, health authorities were alerted to an outbreak of Legionnaires' disease (Legionnaires' disease) in the neighbouring municipalities of Fredrikstad and Sarpsborg, Norway. The resulting outbreak investigation utilised GIS to identify a commercial air scrubber as the source of the outbreak. The information collected as part of the investigation included mapping each surviving patient's home location and movements over the 12 days prior to the onset of symptoms. The locations of the potential sources of infection (including 23 businesses with cooling towers) were also mapped and attack rates and risk ratios were calculated for the populations living within various radii of each potential source. Although those infected lived up to 20 km apart, results of the analyses revealed that those people living within 1 km of a particular air scrubber were most at risk, and only for this source did that risk decrease as the radius increased. Climatic data were also incorporated into a Gaussian puff model to describe the dispersal of aerosols from a number of potential sources. Analyses revealed that the plume modelled for the responsible air scrubber had the best fit with the distribution of cases.

Murcia, Spain, July 2001 (Garcia-Fulgueiras et. al, 2003)[2]

In July 2001 the world's largest Legionnaires' disease outbreak to date occurred in Murcia, Spain. More than 800 suspected cases were reported with 449 confirmed. In conjunction with traditional statistical analyses, GIS-based analyses formed part of the outbreak investigation that identified cooling towers at a city hospital as the source. A standardised questionnaire with a strong focus on 'urban mobility' was administered to 662 suspected patients collecting information for the 14 days prior to the patients' onset of symptoms. From this questionnaire a variety of data-types were mapped including home and work addresses, travel movements and method of transport. Thirty 'zones' were also mapped around potential sources for the dispersal of contaminated aerosols (such as cooling towers). The case-control study conducted by the authors' analysed movements through each of these zones and revealed a strong association between passing through the zone surrounding the hospital cooling tower and being ill with Legionnaires' disease (in all 8 multivariate analyses described in the paper). In fact, Legionnaires' disease was 4.8-11.4 times more likely to be developed in those who passed through the zone surrounding the hospital during the identified 'risk period' than persons who did not travel through this zone.

Pas-de-Calais, France, November 2003 (Nguyen et. al, 2006)[3]

Between November 2003 and January 2004 Pas-de-Calais, France experienced a community wide Legionnaires' disease outbreak with 86 laboratory confirmed cases. A cooling tower was identified as the likely source of the outbreak with the authors suggesting from their investigation that the distance of airborne transmission of L. pneumophila could be greater than previous studies had reported. A standardised questionnaire was completed as part of a matched case-control study collecting information about medical history and personal characteristics; housing and living conditions; and daily outdoor activities and exposures for the 10 days before the onset of illness. Data were gathered on each location visited, the time spent at each location and also the type of transportation used to travel to each location. The data collected revealed that all cases lived or visited an area within a 12 km radius of the responsible plant but did not frequent any places in common. Although some cases were identified within a 12 km radius the majority of cases resided within 6 km of the responsible outbreak source with the highest attack rates identified among residents of the commune in which the responsible source was located. A Gaussian dispersion model (which took into consideration hourly meteorological data on temperature, humidity, nebulosity and wind speed and direction) was used to simulate the dispersion of aerosols from cooling towers. Analyses revealed a good fit between the dispersal of the plumes simulated by the model and the spatial distribution of cases. Temporal alignments were also identified between cooling tower operations and the various 'waves' of the outbreak.

Hereford, England 2003 (Kirrage et. al, 2007)[4]

In October and November 2003 Hereford, England experienced a Legionnaires' disease outbreak with 28 identified cases. The investigation involved interviewing all cases (or a proxy) using an adaption of the Health Protection Agency's standard Legionella Questionnaire. Information was collected on predisposing risk factors, demographic factors, recent movements within and outside Hereford, and specific visits or proximity to an aerosol-generating system, such as a cooling tower, within a 14 day period prior to the onset of symptoms. Information gathered on the place of residence and the daily movements of each case were entered into a GIS were the locations of cooling towers in the area, with 250 m, 500 m, and 1000 m radii plotted around each of them. Wind direction and velocity data for half hour periods were also obtained from a local schools weather station and incorporated into the analyses. A 'composite score' methodology was employed to quantify the risk of exposure and therefore identify the likely source of contamination amongst a large number of potential sources in and around the city centre. The composite score was calculated for each potential source by allocating one point to each source for each of the cases that had either been (a) within 500m of the source or (b) within 1 km downwind of the source at any time during the incubation period. The maximum score was therefore 56 points from the 28 cases. When reviewing the composite scores 3 potential sources were identified as having significantly more cases residing or travelling within 500 m, and who had been downwind of them, than all other sources. Additional epidemiological and microbiological investigation then enabled the rapid identification of a single cooling tower as the source of the outbreak.

North Shropshire, England 2006 (Carr et. al , 2010)[5]

Between August and September 2006, in North Shropshire 6 cases of Legionnaires' disease were reported over a 19 day period (3 times as many cases as had been reported for the same time of the previous 3 years). The initial hypothesis was that this 'cluster' had arisen due to exposure to a contaminated aerosol producing device in the community. GIS was used as part of the outbreak response and helped target the environmental investigation. A GIS database was created consisting of potential sources of aerosol exposure, residential locations of cases, workplace locations, daily travel movements and wind direction. Visualisation of laboratory confirmed Legionnaires' disease cases and their movements revealed that they were scattered over a wide area with no significant clustering. Two mile radii were established around each cases place of residence, their workplace and any other locations visited socially. Two sites with four registered cooling towers that were within a 2 mile radius of the residences or workplaces of four of the cases were identified and visited. Environmental health officers, who also identified other potential sources of contamination such as unregistered cooling towers and car washes within that area, took samples from the identified facilities - they were all culture negative for Legionella species. Despite a detailed investigation into the occurrence of the Legionnaires' disease cluster it was not possible to identify a common source as being responsible for the outbreak. From the available epidemiological, environmental and microbiological information it was concluded that the cluster of Legionnaires' disease cases identified constituted a pseudo-outbreak. Indeed the author makes comment that at the same time other parts of England and parts of Europe were experiencing large increases of cases of Legionnaires' disease.

Barcelona, Spain 2000 ( Jansà et.al , 2002)[6]

During October and November 2000 an outbreak of Legionnaires' disease was investigated in an inner city district of Barcelona. 54 patients related to the outbreak were identified with a case fatality rate of 5.5%. In association with epidemiological and environmental investigation, spatial analysis was employed to assist in the identification of the outbreak source. Case data was collected from each patient using a structured case questionnaire and in addition each case was presented with a map of the district and asked to identify the locations in space they had occupied during the 48 hours prior to their admission to hospital. In the absence of a register of cooling tower locations an identification exercise was carried out through visual reconnaissance and also a number of inquiries being made at various public buildings in the area. Although the map-based element of the case questionnaire did not identify any patterns that helped to establish the source of the outbreak an examination of incidence rates proved more revealing. Incidence rates were calculated using census tracts - each containing approximately 400 people. The incidence rate for the affected area as a whole was calculated as 4.61 cases per 1000 population, however, breaking down the incidence rates by census tract revealed that the northern part of the district was more heavily affected (6.40/1000) than the southern area (2.23/1000). This area was subsequently shown to be in the closest proximity to the cooling towers responsible for the outbreak, as identified through further environmental and microbiological investigation.

South Wales, 2010 (Unpublished and Keramarou & Evans, 2010)[7]

Between August and September 2010, an outbreak of Legionnaires' disease occurred in South East Wales with 22 cases identified. In conjunction with an epidemiological investigation carried out by Public Health Wales (PHW) the Health Protection Agency (HPA) utilised a number of mathematical, statistical and GIS based analytical methods to assist with the identification of a potential outbreak source(s). Although no definitive source(s) was established the epidemiological investigation conducted by PHW (see Keramarou & Evans , 2010) identified 2 space-time clusters centred on the upper Rhymney Valley and the lower Cynon Valley, with both areas in close proximity to cooling towers. GIS-based analyses conducted by the HPA corroborated the findings of the PHW epidemiological investigation with regard to the Rhymney Valley area, identifying 2 cooling towers (within close spatial proximity to each other) as the most likely outbreak sources. Although some of the statistical techniques employed by the HPA provided supporting evidence for a second cluster in the Cynon Valley the GIS-based techniques did not identify this area as a likely source of the outbreak.

As part of the GIS-based outbreak investigation home locations and other visited locations were mapped for each case for the 14 days prior to the onset of symptoms. Travel routes between those locations were inferred using a GIS-based road network analysis and cooling tower locations were obtained from PHW Communicable Disease Surveillance Centre (CDSC) and mapped. Using the residential address of each case and population data from the HSE National Population Database, attack rates were calculated within 1 km of each cooling tower at 200 m intervals, under the assumption that for the responsible cooling tower the attack rate would decrease with distance. The results, however, revealed that no cooling tower displayed the expected radial attack rate profile. By altering the methodology to incorporate each point along a cases travel history as a potential location of infection and thereby artificially boosting the number of 'cases' within the analysis, the results yielded two cooling tower locations showing a decreasing attack rate with distance - both in the Rhymney Valley area. A raster based kernel density analysis was also utilised that incorporated each location visited by each case and the inferred travel routes taken between them. The purpose of the kernel density analysis was to identify those areas in space within close proximity to the locations visited or the routes travelled, that are common between cases. The analysis again revealed an area in the Rhymney Valley that reflected a high incidence of cases coming within relatively close spatial proximity, and therefore suggested an area where a common source of infection could potentially be located. Within the area identified two cooling towers were in operation. Weighting locations and travel routes based on the expected incubation period and re-running the kernel density analysis revealed similarly high values in the Rhymney Valley area.

Alcoi, Spain 1999-2000 (Martínez-Beneito et al., 2006)[8]

Three consecutive outbreaks of Legionnaires' disease occurred in the industrial city of Alcoi in east Spain between September 1999 and December 2000. 36 cases were identified in the first outbreak; 11 in the second outbreak; and 97 in the third outbreak. Industrial cooling towers belonging to Alcoi's textile factories discharge aerosols into the city atmosphere and it was suspected that the source of the outbreaks could have been one or more of these towers.

Martínez-Beneito et al. (2006) identified a group of controls who had an admission to hospital in the same period as cases in the first outbreak and who were of the same gender and roughly the same age. Residential postcodes were obtained and a spatial point process model was constructed with the aim of identifying whether the geographical distribution of the cases could be considered to be random within space.

Ripley's K function was estimated for cases and controls (Ripley, 1981). A difference between these measures was then calculated and tested for statistical significance. Results of the significance tests confirmed higher aggregation of cases than of controls in all outbreaks. Risk surface maps were also estimated for each outbreak. These were given by the difference between the observed probability of being a case at a particular location and the expected probability of being a case within the city. Thus areas of high risk were highlighted and suggested areas to focus on in attempts to find a source for each outbreak.

One of the strengths of this approach is that it uses neither administrative geography nor a network of points; the only locations are the (x,y) coordinates of the cases and controls. This minimises computing overhead and also gives results in a continuous domain. However only home locations were considered and it would be desirable to incorporate workplace locations in future.

Delaware, United States of America 1994 (Brown et al., 1999)[9]

29 cases of community-acquired Legionnaires' disease were reported in Delaware, United States, between July and September 1994. A hospital cooling tower was implicated as a likely source of the outbreak and Brown et al. (1999) describe a case-control study designed to investigate the importance of distance from the cooling tower and duration of exposure. They interviewed participants about potential areas of exposure by using maps divided into a grid. Concentric square blocks were marked on the grid at distances of 0.125, 0.25, 0.5, 0.75 and 1 mile around the hospital. Each case patient and matched control was then asked about possible sites of exposure in the two weeks before the date of onset of illness of the case, and about the number of visits made and the length of time spent in each block. Separate regression models were used to determine the change in risk associated with a change in frequency and duration of exposure in each block. A further regression model assessed the change in risk associated with a change in distance from the cooling tower.

Risk of illness was found to decrease by 20% for each 0.1 mile increase in distance from the hospital, up to one mile from the hospital, but increased by 8% for each hour spent in blocks within 0.125 miles of the hospital. The median dose of modelled exposure was higher for cases than controls.

1. NYGARD K., WERNER-JOHANSEN O., RONSEN S., CAUGANT D. A., SIMONSEN O., KANESTROM A., ASK E., RINGSTAD J., ODEGARD R., JENSEN T., KROGH T., HOIBY E. A., RAGNHILDSTVEIT E., AABERGE I. S. & AAVITSLAND P. (2008) An outbreak of Legionnaires' disease caused by long-distance spread from an industrial air scrubber in Sarpsborg, Norway Clinical Infectious Diseases 46(1), pp.61-69 http
2. GARCIA-FULGUEIRAS A., NAVARRO C., FENOLL D., GARCIA J., GONZALEZ-DIEGO P., JIMENEZ-BUNUALES T., RODRIGUEZ M., LOPEZ R., PACHECO F., RUIZ J., SEGOVIA M., BALADRON B. & PELAZ C. (2003) Legionnaires' disease outbreak in Murcia, Spain Emerging Infectious Diseases 9(8), pp.915 - 921 http pdf
3. NGUYEN T. M. N., ILEF D., JARRAUD S., ROUIL L., CAMPESE C., CHE D., HAEGHEBAERT S., GANIAYRE F., MARCEL F., ETIENNE J. & DESENCLOS J.C. (2006) A community-wide outbreak of Legionnaires' disease linked to industrial cooling towers: How far can contaminated aerosols spread? Journal of Infectious Diseases 193, pp.102-111 http
4. KIRRAGE D., REYNOLDS G., SMITH G.E. & OLOWOKURE B; HEREFORD LEGIONNAIRES' OUTBREAK CONTROL TEAM (2007) Investigation of an outbreak of Legionnaires' disease: Hereford, UK 2003 Respiratory Medicine 101(8), pp.1639-1644 http
5. CARR R., WARREN R., TOWERS L., BARTHOLOMEW A., DUGGAL H.V., REHMAN Y., HARRISON T.G. & OLOWOKURE B (2010) Investigating a cluster of Legionnaires' cases: public health implications Public Health 124 (6), pp. 326-331. http
6. JANSÀ, J.M., CAYLÀ, J.A., FERRER D., GRACIA J., PELAZ C., SALVADOR M., BENAVIDES A., PELLICER T., RODRIGUEZ P., GARCÉS, J.MA., SEGURA A., GUIX J., PLASENCIA A., ARMENGOU J.MA., BOTIA A., MORENO B., FERRER A., GARCÍA DE OLALLA P., GUIX J., PAÑELLA H., PEDRO R., SANZ M., VILLALBÍ RAMÓN, J., DÍEZ A., GARCÍA-CARASUSAN M., MANZANERA R., BENEYTO V., MASDÉU, J., GARRIGA R., DOMÍNGUEZ A., SALLERAS L., ALVAREZ J., BRAVO C., FLORES J., GUDIOL F., GRAU R., MATIA L., MONTERDE R., MORELL F., OROMÍ, J., SABRIÀ, M., TRILLA A. & VAQUÉ, J (2002) An outbreak of Legionnaires' disease in an inner city district: Importance of the first 24 hours in the investigation International Journal of Tuberculosis and Lung Disease 6(9), pp.831-838 http pdf
7. KERAMAROU M. & EVANS M.R (2010) A community outbreak of Legionnaires' disease in South Wales, August-September 2010 Eurosurveillance 15 (42), pp.1-4. http pdf
8. MARTÍNEZ-BENEITO M. A., ABELLÁN, J. J., LÓPEZ-QUÍLEZ A., VANACLOCHA H., ZURRIAGA O., JORQUES G. & FENOLLAR J. (2006) Source detection in an outbreak of Legionnaires' disease Lecture Notes in Statistics 185(3), pp.162-182 http
9. BROWN C. M., NUORTI P. J., BREIMAN R. F., HATHCOCK A. L., FIELDS B. S., LIPMAN H. B., LLEWELLYN G. C., HOFMANN J. & CETRON M. (1999) A community outbreak of Legionnaires' disease linked to hospital cooling towers: An epidemiological method to calculate dose of exposure International Journal of Epidemiology 28, pp.353-359 http pdf