Letter to the editor: “Review of the epidemiologic literature on residential exposure to perchloroethylene” by John A. Bukowski

Response to letter from J.E. Storm and colleagues concerning the paper “Review of the epidemiologic literature on residential exposure to perchloroethylene”

“…To support their reliance on VCS, Storm et al (2012) suggest that color vision is not considered a sensitive indicator of solvent-induced effects, citing one paper by Lomax et al (2004). While I agree with Lomax's overall conclusion that deficits in color vision have not been clearly or consistently demonstrated within the PERC occupational literature, it is incorrect to imply that color vision is not recognized as a sensitive indicator of solvent toxicity.…”

“…This is essentially the definition of publication bias in situ (Phillips, 2004). Storm et al (2012) highlight the importance of their adjusted logistic regression analyses, suggesting that these provide "consistent and coherent" evidence of a negative effect from PERC exposure, and citing selected tables from the report to support this statement. Yet, the NYSDOH (2010) authors provide no discussion of regression diagnostics that would give readers some indication of the reliability of the logistic estimates.…”

“…It is interesting that Storm et al (2012) also cite results from their recent published paper, which was not available when I was writing my 2011 review. In the published version (Storm et al, 2011), their color-vision testing is not even mentioned.…”

“…Dear Sir, I thank Storm et al (2012) for their lengthy critique of this one small aspect of my wider review. However, the level of effort and ardor of their reply could make one conclude that, to paraphrase Shakespeare, they doth protest too much.…”

“…The authors raise four rebuttal points, and I will address the second (selective reporting) first, as it underlies all the others. Storm et al (2012) seem to feel that selective reporting is not possible in the New York State Department of Health (NYSDOH) study because the interested reader can dig out more detail from their report (NYSDOH, 2010). However, publication bias in situ (as within-study selective reporting is termed by Phillips, 2004) concerns primarily what is presented to the reader, not that which is potentially available.…”