About 10,000 chemicals are added to Americans' food to make it taste better or look better, to thicken it, preserve it, or otherwise improve it.
That's an awful lot of chemicals for the federal Food and Drug Administration to monitor. How do they do it?
It turns out that, often, they don't. According to the authors of a study published in JAMA Internal Medicine this month, the agency doesn't even know what all the food additive chemicals are.
Laws that date to 1958 led to an additive designation called GRAS: Generally Regarded as Safe. GRAS applies to about 43 percent of those 10,000 chemicals.
It's the equivalent of a free pass to use the additive without advance permission. Originally, the list included common substances such as vinegar or oil, but it has grown in both size and complexity over the years. It now includes things like potato protein isolates and sodium ferrous citrate and beta glucans from oat bran.
In short, a manufacturer can come up with a food additive, determine on its own that the additive qualifies as GRAS, begin using it, and never notify the FDA - as has happened in about 1,000 cases, the researchers estimate.
Some companies do notify the FDA - often for marketing reasons. For the recent study, researchers examined all 451 notifications from 1997 to 2012 that were published on the agency's website.
They found that in every case, determinations were made by people with strong ties to the manufacturers - employees, or experts hired by a manufacturer or its agents.
"These are real conflicts of interest," said lead author Thomas G. Neltner, director of a food additives project of the Pew Charitable Trusts, which funded the study.
What researchers don't know is whether there were bad decisions. Often, it's impossible for them - or the FDA - to tell, Neltner said.
Even when notices are submitted, "there is no disclosure to the FDA of those conflicts of interest," said coauthor Maricel Maffini, a Washington-based expert on food additives for Pew. "That's what the FDA cannot evaluate."
Neltner suspects that companies don't blatantly declare an unsafe additive safe, that it's more likely a matter of "a thumb on the scale," a consultant being less than rigorous, or ignoring a twinge of doubt.
During the years covered by the study, the researchers found only one case in which the agency rapped some knuckles. In 2010, after reports of serious injuries and deaths among young people consuming caffeinated alcoholic beverages, the agency sent letters to four companies, asking whether they had made GRAS determinations about the use of caffeine in that combination.
None had, and all four "appear to have removed caffeine from such products," the study's authors wrote.
The research paper was accompanied by scathing commentary from Marion Nestle, a New York University nutrition professor.
"This state of affairs might not matter if all food additives were safe at current levels of intake. But some are not," she wrote. "A few additives once assumed to be safe, such as cyclamate salts and sulfites, are now banned or no longer considered GRAS."
Things may change - eventually. In 1997, the FDA proposed new GRAS rules. Thirteen years later - after a 2010 report by the Government Accountability Office criticized the program as too weak - the agency reopened the comment period for the rules.
Now, an FDA spokeswoman said in an e-mail that she could not predict a timeline for final regulations, but that the agency "plans to issue guidance to the industry" on GRAS criteria.
Leon Bruner, a senior vice president with the Grocery Manufacturers Association, defended the GRAS program as a "thorough and comprehensive process," although he agreed it could be improved.
"Ensuring the safety of our products - and maintaining the confidence of consumers - is the single most important goal of our industry," he said in an e-mailed statement.
Meanwhile, another study by Neltner and Maffini came out last week in another journal, Reproductive Toxicology.
They searched the FDA's database and found that, for about 80 percent of food additives, the agency lacked relevant information, including toxicity data, about the safe amount to eat.