Data Analysis

Descriptive epidemiological studies

An initial review of the data collected should be largely descriptive and not require detailed analysis. This is because a comparatively simple interpretation of the data might generate hypotheses for later investigation and so make sophisticated or expensive studies unecessary. A descriptive study should include robust epidemiological and microbiological information about the cases (with a clear case definition for outbreak), and microbiological and environmental data regarding the identified sources. It may become clear from a good descriptive study what the source of the outbreak organism is, and be sufficient to implement effective control measures.

Where important questions remain unanswered, then further, analytical studies may be required, where formal testing of a hypothesis can be performed.

Analytical studies

Epidemiological analytical studies analyse relationships between health status and other variables and allow measures of association to be calculated, such as odds ratios or risk ratios. Once hypotheses can be developed from descriptive analyses (based on epidemiological and/or microbiological information), previous knowledge of the disease and administration of the trawling questionnaire, analytical studies can be developed to test these. Formal testing of a hypothesis, however, may be unnecessary, provided a good descriptive epidemiological study is achieved, supported by strong epidemiological, environmental and laboratory evidence. But if this support is not there, or important questions remain unanswered, then further (analytical) studies may be required.

Further to the epidemiological studies, spatial statistical analysis might be helpful within or using a GIS, for further information please see here

Analytical studies to consider

Click here for a suggested epidemiological study protocol

The cohort study (prospective or retrospective) is the gold standard of analytic epidemiology as it allows calculation of indicators which have a very clear meaning and the results are immediately understandable. However, it is more realistic in an outbreak situation to use the case-control study; the use of this type of study has been repeatedly demonstrated in the investigation of Legionella outbreaks, most notably where cooling towers were found to be the source of transmission [1][2][3] (see also special types of case-control studies). Generally, in a case-control study, a variety of exposures can be studied, whereas in a cohort study, a variety of diseases can be studied. Various published outbreak reports and choice of analytical studies used are shown here. Although, this outbreak list is not exhaustive, case-control studies, rather than cohort studies, have predominantly been used. Some researchers have performed cohort studies with nested case-control design which can help mediate the cost of cohort studies. The advantages and disadvantages of each study design are outlined here. A guide to deciding which type of study is more appropriate under different circumstances can be found here.

Each Legionnaires' disease outbreak is unique, but there will be situations where information from descriptive epidemiology will be enough to identify an exposure source and begin implementation of public health control measures - the ultimate aim of outbreak investigation. However, in addition to the overall aim of stopping the outbreak and the emergence of new cases, there will be individual reasons to take into account when considering analytical investigation. Where analytical studies have been employed in (published) Legionnaires' Disease outbreaks, the most commonly used design has been the case-control study. Although the results of a cohort study allow measurement of the true risk of infection and are slightly more insightful, in a situation where the exposed population is large and ill-defined, the case-control design is the more practical. Moreover, the odds ratio achieved from a case-control study is a perfectly good measure for the association between exposure and illness.

Applied Analyses

Applied analyses, in the form of GIS or statistical and plume modelling, allow further visualisation and assessment of the data and can help to determine the source of infection and the population at risk. The use of these tools and the quality of the outputs is, however, reliant on the quality of the data collected and recorded during the descriptive analyses stage. Applied analysis using GIS, statistical methods [4] and/or plume modelling, allow further visualisation and assessment of the data and can help to determine the source of infection and the population at risk [5]. The use of these tools and the quality of the outputs is, however, reliant on the quality of the data collected and recorded during the descriptive analyses stage.