Metaanalysis is a technique for combining the results of similar studies to arrive at a single estimate of effect or risk. It is of proven value in the field of clinical trials but its use as an epidemiological tool is less well established. The principal obstacles to the valid use of metaanalysis are lack of consistency in the design, conduct and analysis of studies, publication bias, and reporting bias. Metaanalysis has been used to question whether an excess of lung cancer is associated with postmodification production of chromates and whether tumors of any site other than the lung are associated with any level of chromium exposure. The metaanalysis of lung cancer postmodification did not show a statistically significant excess, but the combined studies lacked the statistical power to exclude the possibility of a continuing risk of moderate size. The data in the literature for tumors at sites other than the lung were too impaired by heterogeneity of reporting and reporting bias for metaanalysis to be applied with any validity. A more uniform approach to the planning and reporting of studies is essential if metaanalysis is to gain credibility as an epidemiological tool.
Copyright 1997 Academic Press.