Serious hazard classifications cover 63 per cent of the food-contact chemicals with any data

Subtract the 13,211 chemicals with no harmonized hazard data from the 15,159 in the inventory and 1,948 are left, which is the set anyone has assessed ^[14]. The 1,222 classified as carcinogenic, mutagenic, toxic to reproduction, harmful to specific organs after repeated exposure or endocrine disrupting come out of that set ^[2][4]. Among assessed chemicals, then, the hazard rate is about 63 per cent ^[15]. The 8 per cent divides by the whole inventory, and 87 per cent of that inventory has no harmonized data ^[2][16].

Sixty-three per cent is not an estimate of the true rate. Testing is not random: substances get assessed because someone had a reason to suspect them, so the 1,948 are enriched for hazard and the rate across the untested 13,211 is almost certainly lower. How much lower, nobody can say from harmonized data, because 13,211 chemicals do not have any ^[3]. Dr Helene Wiesinger, scientific communication officer at the Food Packaging Forum Foundation and one of the researchers, told Fortune that "there were huge data lags" ^[19].

Substitution is the cheap response to a hazard finding. The study undercuts it. "So take bisphenol A, and we say we don't want to use bisphenol A anymore, so we use bisphenol S instead, because it's functionally similar," Wiesinger said ^[6]. Wiesinger said researchers call the pattern "regrettable substitution" ^[18]. Bisphenol S was later found to cause negative hormonal effects "comparable to or worse" than its predecessor, based on a study from the National Library of Medicine ^[8]. Roughly 93 per cent of people over the age of six have detectable levels of bisphenol A, according to the National Institute of Environmental Health Sciences ^[7].

Harmonized testing arriving on a material share of the 13,211 and returning hazard rates in the single digits would mean the 63 per cent was selection bias and little else ^[3][15]. Wiesinger warned that the findings are most effective in the hands of policymakers ^[12]. Regulators continuing to work chemical by chemical would mean the priority groups never become an enforceable category: about 70 per cent of the more than 4,000 chemicals fell into clusters structurally similar to known hazards, including ortho-phthalates and PFASs, which is roughly 2,800 substances, though the study as reported does not define what the 4,000 counts ^[10][9][17].