HEALTH IMPACT OF THE EMISSIONS FROM A REFINERY: CASE-CONTROL STUDY ON THE ADULT POPULATION LIVING IN TWO MUNICIPALITIES IN LOMELLINA, ITALY

In the municipalities of Sannazzaro de’ Burgondi and Ferrera Erbognone (Pavia, Italy), an oil refinery is operating since 1963; the private company running the plant was authorised to set up a new facility on the condition of sponsoring an epidemiological study. For this reason, the University of Pavia developed the CONSAL Project (Conoscenza e Salute). The present thesis focuses on the ante-operam phase of Study 1. Its specific aim was to investigate the health impacts of the emissions from the refinery on the adult population, mutually adjusting the estimates for additional information collected through a survey. The study was designed as a case-control. Cases were the subjects admitted to hospital between 2002 and 2014 due to acute conditions of respiratory, cardiovascular or gastrointestinal systems; controls were selected among the subjects that were not hospitalised. Cases and controls, with a ratio of 3:1, had to be alive, aged 20-64 years in the reference timespan, and were balanced for age, gender and municipality. Data were extracted from the databases of the local Health Protection Agency; personal information were also checked with the Municipal Registries. 1046 subjects (257 cases, 789 controls) were enrolled and received a mailed survey; 15 were later excluded because of death or unavailability. Respondents were 563 (54.6%), with a significant difference by municipality and no difference by age, gender or case/control status. 22 subjects declared to live elsewhere, so 541 were included in the analyses. The fallout of refinery’s emissions was predicted by the AERMOD model; PM10 was chosen as a tracer pollutant. Individual exposure was assigned on the basis of the home address and then recoded in 2, 3 or 4 clusters by using K-means models. The 2-clustered PM10 exposure came out to coincide with the two municipalities. An excess of “risk” with the increase in clustered PM10 level was found; however, none of the effect estimates was statistically significant. The Odds Ratios (ORs), between 1.40 and 1.60, were similar with municipality (2-clustered exposure) and 3- or 4-clustered exposures. Logistic regression analyses disclosed similar (and still non-significant) estimates while adjusting for age, gender, lifetime cigarette smoking and for being diagnosed or treated for other diseases that could be ascribed to the definition of case. A comparative analysis of informativity showed that the best model is the one using municipality instead of 3- or 4-clustered exposures. A secondary analysis, evaluating the influence of several factors on self-perceived health, disclosed that living further from the refinery was reducing the “risk” of a negative self-perception of health (OR=0.859), albeit the effect was not statistically significant. The results described above are consistent with toxicological and epidemiological evidence. However, they might have been affected by various biases. First, the definition of cases and controls might have been influenced by the use of administrative databases. Moreover, restricting to hospitalised cases might have excluded those with a less severe condition while, at the same time, restricting to those who were alive at the time of assessment might have excluded the most-severe ones. In addition, the unexpectedly strong association between predicted PM10 concentrations and municipality, and the fact that the enrolment of cases and controls was balanced by municipality, may have led to an underestimation of the effects of environmental exposure. It should also be considered that those factors are measured in the present, while the outcome occurred in the past. Finally, the response rate was lower than expected, so that study power was reduced. Concluding, the results indicate a non-conclusive evidence supporting a possible excess of hospitalisation risk among people living in Sannazzaro de’ Burgondi, in comparison with those from Ferrera Erbognone.