Multilevel random intercept poisson model for estimation for association between cancer incidence and arsenic concentration in drinking water wells in Córdoba, Argentina.

ABALLAY, LR; FRANCISCA, FM; DÍAZ, MP; MUÑOZ, SE

Florianópolis

Congreso; 55ª Reunião da RBRAS e 15ª RARG - XXV International Biometric Conference. International Biometric Society.; 2010

In Argentina, high content of arsenic (As) in underground water is widely known. There is no evidence yet of an association between the pattern of cancer incidence and As distribution in groundwater supplies. To assess the association between the incidence pattern of main cancers, and As in groundwater of the hydro-geological regions (aquifers) of Córdoba province, a multilevel Poisson model was applied. The response variable were the Age Standardized Incidence Rates (ASIRs) for breast, prostate, lung, colon and bladder cancers. Mean As concentration of 518 monitoring wells, latitude and longitude were included as fixed effects at level 1. Aquifers, at level 2. The results indicated than the random intercept model was more suitable respect to the single Poisson model, having better Akaike criterium. As promoted lung and bladder cancers in both sexes, and female colon cancer. Breast cancer had no association with As.