La valutazione dell’esposizione professionale nelle grandi banche dati

Background: Big data sets seldom include occupational information. To extend their use to the study of occupational health hazards, big data sets need to include a limited number of variables best predictive of the level of exposure to workplace risk factors. Objective: To calculate the predictive value of the occupational variables used in a case-control study, and to explore whether and to what extent a limited set of such variables could effectively replace the whole set in assessing exposure level. Methods: We conducted a principal component categorical analysis on 8 occupational exposure variables meant to define level of occupational exposure to solvents in a case-control study. For each variable, we calculated the predictive value. We used the Cronbach α test to calculate the agreement between the assessment based upon the selected variables, and that based upon the whole set of occupational variables. Results: The combination of direct exposure to the risk factor, lack of compliance with use of the prescribed personal protective equipment, lack of adequate ventilation, and possibility of skin contact (in case of chemical agents) accounts for 54% of the between-individuals variance of the exposure level estimates. The Cronbach α test value (0.878) indicates good agreement between the assessment resulting from these variables and that resulting from using the whole set. Conclusions: Including a limited number of occupational variables in big data sets is of paramount importance to explore the aetiological relevance of occupational hazards, particularly in the case of health outcomes associated with the recent technological changes.