Poster | 6th Internet World Congress for Biomedical Sciences |
Mariarosa Re(1)
(1)Dipart. Medicina Preventiva, Occupazionale e di Comunitą. University of Pavia - Pavia. Italy
[Hygiene, Public Health & Preventive Medicine] |
[Occupational Health] |
Software Statistical calculations for correlation-based procedures may be performed by SPSS or any other easily available statistical package (e.g. SYSTAT etc.)
Procedures The strategy for applying correlation methods to multiple exposure evaluation consists of the following steps:
(a) Use of simple correlation techniques for testing correlation of measurements of individual substances with each other and with total quantity of all contaminants. Data required at this step need be collected through a preliminary survey. As concentrations of contaminants in air follow a lognormal statistical distribution (7), the parametric (Pearson) correlation test (8) is applied to logarithms of concentrations. If a significant correlation is found at this stage, an index or predictor substance can be singled out: for instance the one that is most abundant in the mixture, or the one that most correlates with total quantity of contaminants, or the one that is most easily analysed, or the one that is most toxic. The sum of concentrations of all components can itself be a predictor for exposure evaluation. This choice is particularly convenient when the correlation with individual substances is high and the total quantity can be determined by a simpler or less expensive analytical technique than required for specific determination of single substances.
(b) Routine monitoring of multiple exposure can be performed by direct measurement of the index substance and prediction of concentrations for each unmeasured component of the mixture and for total quantity of contaminants by statistical regression.
(c) An alternative technique to step (b), that is more reliable but less practical as it requires not only measuring the index component, but also a small random sample of concentration values for each component to be estimated, involves using the concentration of the index component as auxiliary variable and ratio or regression estimators (9).
Examples of working activities where it is expected that correlation-based methods may be advantageously applied for multiple exposure monitoring are reported in the following section.
[Hygiene, Public Health & Preventive Medicine] |
[Occupational Health] |