Particulate Matter Exposure and Lung Cancer: A Review of two Meta-Analysis Studies
share
The current regulatory paradigm is that PM2.5, over time causes lung cancer. This claim is based on cohort studies and meta-analysis that use cohort studies as their base studies. There is a need to evaluate the reliability of this causal claim. Our idea is to examine the base studies with respect to multiple testing and multiple modeling and to look closer at the meta-analysis using p-value plots. For two meta-analysis we investigated, some extremely small p-values were observed in some of the base studies, which we think are due to a combination of bias and small standard errors. The p-value plot for one meta-analysis indicates no effect. For the other meta-analysis, we note the p-value plot is consistent with a two-component mixture. Small p-values might be real or due to some combination of p-hacking, publication bias, covariate problems, etc. The large p-values could indicate no real effect, or be wrong due to low power, missing covariates, etc. We conclude that the results are ambiguous at best. These meta-analyses do not establish that PM2.5 is causal of lung tumors.
READ FULL TEXT