Evaluation and synthesis of health effects studies of communities surrounding arsenic producing industries.

Epidemiological studies designed to detect lung cancer risk and other health effects in communities surrounding arsenic-producing copper smelters were reviewed. The studies were about evenly divided in finding deleterious and 'beneficial' effects of arsenic. All of the studies had insufficient statistical power to detect the small increases in risk that may occur. Even the most powerful studies were not designed to detect relative risks less than about 1.2 and the majority of the studies had little power to detect risks under 2.0. Confidence intervals for the relative risks from these studies were not very useful in putting an upper bound on adverse effects of arsenic. Sources of bias and other difficulties with community health studies are also discussed. We argue that these studies may be a good and economical first investigation but, due to a lack of power, null findings do not rule out the possibility of excess risks that may be significant from a public health viewpoint.