Meta-analysis: methods for combining data to improve quantitative risk assessment

Although individual studies that constitute the data base for a risk assessment are each evaluated quantitatively as well as qualitatively, assessment of the total data base frequently results in selection of data from particular studies, rather than an effort to combine data quantitatively. Meta-analysis, or the analysis of analyses, provides an approach for the joint evaluation of the results of several studies. In this report, two very similar cancer bioassays of trichloroethylene were used to illustrate some simple meta-analytic techniques and to evaluate the validity and value of these procedures. The results demonstrate that concepts such as the upper 95% confidence limit are highly dependent upon assumptions if several data sets are involved. Use of a Monte Carlo procedure resulted in an increase in the 95% upper confidence limit relative to the value determined by EPA, but by determining the variance of the maximum likelihood estimate of the linear coordinate of the estimated dose-response curve, the 95% upper confidence limit of the cancer potency factor was reduced.