|
Eleven cases of legionnaires´ disease, all patients living
in the same municipality, were admitted to a district hospital in northern
Portugal. Preliminary investigations suggested an association with the events
of the annual festivities in that municipality. Legionella pneumophila was
not isolated from any of the suspected sources, but evidence from a case
control study suggested that an aerosol produced by a decorative fountain
in the main square during the night of a rock concert was the likely vehicle
of infection. The prevalence of smoking was higher among cases than controls.
Introduction
Legionnaires’ disease has been a notifiable disease in Portugal since
1 January 1999. As far as we know, only sporadic cases have been reported
(1). During the first week of September 2000, a local health authority
was informed that a probable outbreak of legionnaires’ disease was occurring
in a small municipality in the north of Portugal and that several cases
had been admitted to the nearest district hospital. This event attracted
the attention of news media, whose coverage created some alarm in the
community.
On 6 September an outbreak control team (OCT) was formed, including regional,
district, and local public health professionals. The objectives of this
team were to confirm the existence of an outbreak, identify the sources
of infection, and implement immediate control measures if necessary.
Preliminary assessment
The OCT visited the hospital on 6 September. Thirteen cases of legionnaires’
disease had been admitted between 25 August and 4 September. They were
all interviewed in order to make a preliminary assessment of potential
risk factors. The only common characteristic was that 11 of them lived
in the same municipality; the other two had not visited that municipality
during 2000. Though we investigated these two patients in detail, the
results are not described in this report because we consider them as epidemiologically
distinct from the other 11 cases.
These 11 cases were men aged 33 to 73 years, 10 of whom were regular
smokers. One of the cases was seriously ill and was treated in the intensive
care unit. All 11 patients suffered fever, 5 had cough, 3 headache, 3
myalgia, 2 weakness, 2 shortness of breath, 2 vomiting, and 1 diarrhoea.
None of the patients died. According to the epidemic curve (figure) and
the reported incubation period of the disease (2 to 12 days) (2) we assumed
that exposure had occurred between 12 and 22 August. The annual festivities
of the municipality where the 11 patients lived had taken place between
10-14 August 2000. We were told that these festivities had been especially
crowded, hundreds of people had attended several popular events. On 6
September, we visited the municipality and its main square, where most
of the events had taken place. The main town square had two decorative
fountains, close to each other. Water samples from both fountains were
collected and sent to the National Institute of Health (Porto Delegation).
Distribution of cases by time and place confirmed the existence of an
outbreak. We conducted a case control study in order to investigate associations
between the presence at particular events of the festivities and the risk
of disease. We regarded the two decorative fountains in the main square
as potential vehicles of infection.
Methods
Case definition and case finding
According to the case definitions of the World Health Organization
(3) and the European Working Group on Legionella Infection (4), and considering
clinical and laboratory criteria, we classified cases as probable and
confirmed. A case was considered probable when an acute pulmonary infection
with focal signs of pneumonia, either by clinical or radiological evidence,
was associated with the presence of a single serum antibody titre of at
least 1:128 against L. pneumophila. A case was classified as confirmed
when besides the clinical criteria of pneumonia there was at least one
of the following laboratory criteria: isolation of L. pneumophila
or detection of L. pneumophila antigen in the urine or a fourfold
rise in the titre of L. pneumophila antibody in serum. Both probable
and confirmed cases were selected for the case control study.
Control definition and selection
The criteria for the definition of the controls were: male gender,
age 33 to 73 years, and residence in the same municipality as the cases.
We had no information to make a precise sample size estimate. Thus, we
used Epi Info (5) to estimate the sample size, considering the following
assumptions: a=0.05, b=0.20, 50% of the controls exposed, and a very high
odds ratio (OR >20). We then selected four controls for each case.
They were recruited amongst the population consulting the Primary Care
Health Centre (PCHC), for any reason, on 21 November and 7 December. We
used a systematic sampling method: all eligible citizens attending the
PCHC between 10am and midday were approached and invited to participate.
Data collection
We conducted interviews with all cases and controls, using a questionnaire
gathering information about smoking habits and presence at the events
of the festivities. The decorative fountains were never mentioned explicitly
in the questionnaire, in order to avoid leading participant's answers.
We listed all events of the festivities (transcribed from the official
programme) and for each of them, cases and controls were asked to answer
if they had or had not attended that event. We knew where the events had
taken place and were therefore able to find out if people had attended
events next to the decorative fountains. For the period of eight days
after the festivities, included in the period in which cases could have
been infected, we asked cases and controls whether or not they had been
in the town at particular times of the day (morning, afternoon, and night).
Data analysis
Epi-Info version 6.04 (5) was used for data entry and preliminary
analysis and Stata (6) for the remaining analysis. Crude ORs were estimated
for each festivity event and period of time in the days after the festivities.
To control for confounding we used a logistic regression model with all
variables statistically associated with the disease in the univariate
analysis.
Microbiological investigation
Urine specimens of the cases were examined for L. pneumophila
antigen, using an enzyme immunoassay technique (Binax®), at the district
hospital laboratory.
Sputum smears were sent to the National Institute of Health in Porto,
for conventional bacteriological investigation.
Specific serum antibody levels were measured, using indirect fluorescent
antibody (IFA) assay, at the National Institute of Health in Porto.
Environmental investigation
In order to identify the source of infection, we inquired about routine
and/or occasional exposure to aerosol-producing devices. Water samples
from the two decorative fountains in the town's main square and swabs
from an air conditioning device at the workplace of one case were analysed
and cultured for L. pneumophila at the National Institute of Health,
in Porto.
Results
Microbiology
Two patients had a positive legionella urinary antigen (L. pneumophila
serogroup 1, Binax ®), six patients had a single high serological
titre for legionella antibody equal or above 128, and five patients had
a fourfold or greater rise in the serological titre for legionella antibody.
None of the cultures performed was positive. Thus, five cases were classified
as probable and six as confirmed.
Analytical epidemiology
Forty of the 44 people initially approached were included as controls
in the analysis: one refused to take part and three were excluded because
they were not permanent residents in the considered municipality.
The age distribution of the 11 cases (mean 51 years; standard deviation
(SD) 13 years) and 40 controls (mean 51 years; SD 11 years) was similar
(p=0.89). Ten of the 11 cases (91%) were regular smokers, compared with
12 of the 40 controls (30%) (p<0.001). Using univariate analysis, we
found a significant association between the risk of developing the disease
and presence at five of the 53 official events of the annual festivities
(table). All five events took place in the main square, where the two
decorative fountains are located. Cases were more likely than controls
to have attended the rock concert on 11 August (OR=26.36), the country
music concert on 13 August (OR=9.35), fireworks on 14 (OR=6.22) and 15
August (OR=6.22), and the folk music concert on 10 August (OR=4.95).
Table1. Crude and adjusted (Mantel-Haenszel) Odds Ratio
estimates, according to the presence in the events of the festivities
|
Event / Date (Hour)/ Location of the event
|
Crude OR
(P)
|
Adjusted OR*
(P)
|
|
Concert de rock / Concert by a Rock Band
11-08 (23.00)
Place principale/ Main Square
|
26,63
(p=0,00023)
|
_
|
|
Concert du chanteur country / Concert by Country Singer
13-08 (22.00)
Place principale / Main Square
|
9,35
(p=0,0054)
|
7,65
(p=0,1098)
|
|
Feu d'artifice / Fireworks
14-08-00 (03.00)
Place principale / Main Square
|
6,22
(p=0,0155)
|
0,80
(p=0,6278)
|
|
Feu d'artifice / Fireworks
15-08-00 (01.00)
Place principale / Main Square
|
6,22
(p=0,0154)
|
1,28
(p=0,5930)
|
|
Concert du chanteur de folk / Concert by Folk Music Singer
10-08 (22.30)
Place principale / Main Square
|
4,95
(p=0,039)
|
0,35
(p=0,4127)
|
* Adjusted for the presence at the Rock Band concert, on the
11th of August.
After adjusting for confounding, the only event that remained significantly
associated (OR=43.32) with disease was the rock concert on 11 August (table).
Ten of the 11 cases (91%) attended the rock concert. The aetiological
fraction was 97.7% (95% confidence interval (CI): 13.0% - 99.9%) when
an adjusted OR was used for its calculation (7).
Environmental investigation
All cultures for L. pneumophila were negative. Besides the
fountains and one working place air conditioning device, no other suspected
devices were identified.
Discussion
There was a strong association between attendance at the rock concert
on the night of 11 to 12 August at the main square and disease. That association
was not confounded by attendance at other events, which were not associated
with disease.
We assessed the possibility that selection bias had influenced the validity
of the study. The age and residence distribution of cases and controls
was such that no relevant selection bias was likely to have occurred.
The use of an exhaustive list of events in the questionnaire, avoiding
explicit mention of the decorative fountains, minimised the possibility
that participants would be led to suggest the suspected source of infection.
The apparent association found in the univariate analysis was due to
the fact that townspeople who attended one concert at night were more
likely to attend more than one concert and other night events.
This explains the potential confounding, which was controlled by multivariate
analysis.
Only one case could not be explained by attendance at the rock concert,
and the computed value of the aetiological fraction was very high (though
with a wide confidence interval). The higher prevalence of smoking habits
among cases is consistent with previous observations (2).
It is likely that L. pneumophila was not isolated from the water
samples from the decorative fountains because, by the time samples were
collected, the fountains had already been cleaned and disinfected and
fresh treated water had been added to the system. According to the municipal
records, only one of the two decorative fountains was working during the
rock concert. Curiously, it was not the fountain closest to the public
and the stage. Direct observation and analysis of night photographs have
shown that a huge amount of aerosol was produced while the fountain played.
Thus, biological plausibility adds to the epidemiological evidence. We
concluded that it is very likely that aerosol particles produced by the
decorative fountain in the main square during the night of the rock concert
were the vehicle for spread of L. pneumophila, and that these particles
were inhaled by some people who were present at the concert. Other outbreaks
of legionellosis spread by decorative fountains have been described (8,9).
It is possible that the same source of L. pneumophila caused other
infections that were not detected.
Control measures
Although the decorative fountains had already been cleaned and disinfected
by the time the health authority was informed of the outbreak, some technical
guidelines (10) for control measures for decorative fountains were elaborated
and sent to the municipal authorities.
Acknowledgements
We wish to thank all staff working in the municipality, the district
hospital, the local health centre, the National Institute of Health, and
the Regional Public Health Centre who helped us during the outbreak investigation.
|