A case study comparing a meta-analysis and a pooled analysis of studies of sinonasal cancer among wood workers.

A pooled analysis of raw data from 12 case-control studies of sinonasal cancer has recently been conducted by the International Agency for Research on Cancer; the summary odds ratio for all wood-related occupations was 2.0 (95% confidence interval = 1.6-2.5). We have conducted a meta-analysis of the published results of 14 studies on the same topic, including 11 studies from the pooled analysis, and compared the results for several categories of wood workers. Usable results were available for 12 studies: male wood workers had a summary odds ratio of sinonasal cancer of 2.6 (95% confidence interval = 2.1-3.3). The corresponding value based only on the studies that were also included in the pooled analysis was 2.4 (95% confidence interval = 1.9-3.0). When our meta-analysis was based on a relatively large number of studies, results tended to be higher than those of the pooled analysis. As compared with the meta-analysis, the pooled analysis controlled the effect of publication bias by using data from studies for which no specific result was reported, and it reduced misclassification of exposure; the resources used in the pooled analysis, however, were one order of magnitude greater than those needed for the meta-analysis. Whether a pooled analysis of raw data or a meta-analysis should be carried out depends on the research question; we propose some criteria for this decision.