Food Additives Worth Reading About

E 160c in a sausage is paprika extract. E 300 is ascorbic acid, which is vitamin C. Both look equally alarming on a label, and both mean the same thing every other such number means: the substance sits in a register of permitted food additives and has an index there.

E 171, titanium dioxide, sat in that same register alongside them until 2021. It is gone now. That is the shortest answer to the question of whether E numbers are safe: the number records that a substance was once assessed and cleared, and says nothing about whether the clearance will survive the next review.

What the letter E actually means

EFSA puts it plainly: in the EU all food additives are identified by an E number, and their safety is evaluated before they can be permitted for use in food. EFSA advises, the Commission and the member states decide. Ukraine runs a double frame on top of that. Under article 31-1 of the food safety law, an operator may use an additive only if it has passed state registration in Ukraine or has been recognized as safe in the European Union. The conditions come from a health ministry order in force since 2024, and two of its principles matter to a shopper: the level of use is set at the lowest value needed for the intended effect, and where no maximum exists the additive falls under quantum satis, no more than is necessary for the purpose. In unprocessed food, additives are banned outright apart from cases listed explicitly.

The same order sorts additives into 27 functional classes, from antioxidants and colors to emulsifiers, preservatives and sweeteners. A class describes the technological job a substance does, and there is no scale of harm anywhere in that classification.

Registration has no expiry date and is not final either: new scientific information showing a risk to health obliges the authority to suspend, revoke or amend it, and obliges the manufacturer to forward such information within three working days.

Titanium dioxide: how a number leaves the list

E 171 whitened icing, chewing gum, sauces and confectionery. In May 2021 an EFSA panel reassessed it. General toxicity showed nothing: no adverse effects even at 1,000 mg per kilogram of body weight per day. The problem lay elsewhere. Titanium dioxide particles “have the potential to induce DNA strand breaks and chromosomal damage, but not gene mutations”, and no threshold below which this stops could be identified. Hence the conclusion that E 171 “can no longer be considered as safe”.

Commission Regulation (EU) 2022/63 took effect on February 7, 2022, with the wording that titanium dioxide “may no longer be used in foods”. Stock made under the old rules could be sold until August 7, 2022. As a color for medicines it stayed provisionally.

Emulsifiers: the loudest topic of recent years

The story started in mice, where emulsifiers thinned the gut mucus layer and shifted the microbiota. In 2022 it reached people. At the University of Pennsylvania nine healthy adults ate an emulsifier-free diet and seven ate the same diet plus 15 grams of carboxymethylcellulose a day for 11 days. The difference showed: more abdominal discomfort after meals, less microbiota diversity, lower short chain fatty acids. In two of the seven, bacteria pushed into the inner mucus layer that is normally sterile.

Fifteen grams a day is a lot. That is why the decisive study is a different one, published in 2026 from Leuven. Sixty healthy people spent two weeks on an emulsifier-free diet, then four weeks receiving one of five substances inside brownies (carboxymethylcellulose, polysorbate-80, carrageenan, soy lecithin or native rice starch) against placebo. Realistic doses, blinded allocation.

The result came out more restrained than either camp expected. Short chain fatty acids dropped in the carboxymethylcellulose group, with the same tendency elsewhere though not always significant. Fecal calprotectin, C-reactive protein, lipopolysaccharide-binding protein, cholesterol and the other metabolic markers showed no difference against placebo. At realistic doses, inflammation did not appear.

What the large cohorts found, and how to read their numbers

The French NutriNet-Santé cohort produced three large papers on emulsifiers. These are observations: people record what they ate, and nobody adds anything to anyone’s diet.

An HR of 1.03 means three percent on top of the risk already there, not three times it, and it is tied to a dose step: 100 mg of carrageenan a day is a couple of spoons of ice cream. The authors name the limits themselves, exposure measurement error and residual confounding. People who eat many emulsifiers usually eat much processed food in general, and no questionnaire separates the two.

Carrageenan: the only one with two human trials

Carrageenan E 407 comes from red seaweed and goes into dairy, plant drinks and meat products. In Tübingen in 2024, twenty men took 250 mg of carrageenan twice a day or placebo, two weeks each, in a crossover design.

Across the group there was no difference. Among the overweight participants, whole body and hepatic insulin sensitivity fell, C-reactive protein and IL-6 rose, and intestinal permeability increased. That is a subgroup analysis inside twenty people, so it is a signal rather than a finding. The Leuven trial independently saw transcellular permeability rise in its carrageenan group against baseline. Two separate teams saw the same thing in the same substance, which makes carrageenan the one emulsifier here worth watching.

Nitrites: the additive against the natural source

The same French cohort counted nitrates and nitrites from three sources at once: food additives, water and food itself. Among 101,056 participants, 3,311 cancers were diagnosed over 6.7 years.

High intake of additive nitrates went with higher breast cancer risk (HR 1.24), additive nitrites with higher prostate cancer risk (HR 1.58). For natural sources there was no association at all. That matters for anyone worried about nitrate in vegetables: here the problem sat in curing salt, not in salad. What is known about vegetable nitrate on its own is covered in our piece on nitrates.

The six colors behind the label warning

Six colors carry a mandatory warning: E 102 tartrazine, E 104 quinoline yellow, E 110 sunset yellow, E 122 carmoisine, E 124 ponceau 4R, E 129 allura red. The wording is that they “may have an adverse effect on activity and attention in children”.

Behind that sentence sits one trial, the Southampton study of 2007: 153 three-year-olds and 144 eight and nine-year-olds, double-blind crossover, a drink with a color mix and sodium benzoate. The effect found is small, 0.20 in the younger children and 0.12 to 0.17 in the older ones. The authors phrase it carefully, “artificial colours or a sodium benzoate preservative (or both)”, because the drink made it impossible to separate them.

It gets more interesting. A 2017 systematic review of meta-analyses showed the effect size depends on who rates the child: parents 0.44 and 0.21, teachers 0.08, independent observers 0.11. Among raters blind to the drink, it nearly vanishes.

The same material was reworked by genotype in 2010. The response tracked with polymorphisms in the histamine degradation genes HNMT T939C and HNMT Thr105Ile, and in older children with DAT1 as well. The authors argue that this genetic variation explains why earlier studies contradicted each other. In practice, some children react to these colors and some do not, and you cannot tell in advance which you have.

Glutamate: the scare did not hold, and the file is still open

E 621 tops the popular ranking of harm and barely deserves it for the reasons people fear it. Blind placebo-controlled challenges rarely confirm consistent symptoms at realistic dietary doses. Transient intolerance shows up at high doses on an empty stomach, and IgE-mediated allergy is very rare. A 2026 review also names expectation: someone who knows they ate glutamate reports symptoms more often.

Calling glutamate entirely safe does not work either, and the reason is an EFSA document from 2017. The panel derived the first group acceptable daily intake for glutamic acid and glutamates E 620 to E 625, 30 mg per kilogram of body weight per day, and noted immediately that intake in some population groups exceeded both that value and doses linked to adverse effects in humans. Genotoxicity is not an issue.

Aspartame: what group 2B means

In July 2023 two WHO bodies spoke about aspartame almost simultaneously and differently. JECFA reaffirmed its safety and called the epidemiology “not convincing”. IARC placed aspartame in group 2B, possibly carcinogenic to humans, on the strength of three liver cancer studies, while stating that the evidence is limited because “chance, bias or confounding could not be ruled out as an explanation for the positive findings”.

The most detailed analysis of that disagreement comes from authors who include an American Beverage Association officer and two consultants from the firm Gradient, and their position is that the classification should have been group 3. Worth keeping the affiliation in mind. The largest independent check is older: the US National Cancer Institute followed 285,079 men and 188,905 women aged 50 to 71 and found no association between aspartame and hematopoietic cancers (RR 0.98 at 600 mg a day or more) or glioma.

What to do at the shelf

Reading ingredients by number is pointless: E 300 is vitamin C, E 407 has two human trials with a warning signal, and they look identical on a label. Working out your own dose is impossible too, because the quantity is never printed. There are also categories where looking for numbers makes no sense at all: additives are banned in unprocessed food, and what that means for a bag of cut leaves is in our piece on preservatives in bagged salads. Something else works: the length of an ingredient list tracks the degree of industrial processing, and processing rather than any single number carries most of the associations the cohorts found. The Leuven trial showed it sideways, since two weeks of emulsifier-free eating lowered cholesterol before anything was added to anyone.

If you want something more specific, three positions have grounds: carrageenan E 407, because two independent trials saw intestinal permeability rise; nitrites in cured meat, because the association was found for the additive rather than natural sources; and the six warning-labeled colors, if the house has a child who reacts to them. Only watching your own child can establish that last one.

Key points

• An E number is an index in a register of permitted additives, sorted into 27 classes by technological job. No scale of harm is built into it.
• Registration has no expiry but is revoked on new scientific information. E 171 left the EU list in 2022 exactly that way.
• At realistic doses, emulsifiers produced neither inflammation nor a metabolic shift in humans. One finding held: lower short chain fatty acids.
• Cohort hazard ratios of 1.03 to 1.15 mean three to fifteen percent on top of existing risk, tied to a dose step, from self-reported food.
• Carrageenan E 407 is the only one where two independent human trials saw intestinal permeability increase.
• Nitrates and nitrites showed a cancer association as additives. For natural sources there was none.
• The color effect on children’s attention is small and depends on who rates the child and on histamine degradation genes.
• Glutamate is not an allergen in blind challenges, yet EFSA set an ADI of 30 mg/kg and noted that some groups exceed it. Aspartame sits in group 2B on three studies with limited evidence.

How a food additive differs from a pesticide residue comes through in our piece on glyphosate in grains: a residue arrives from the field with no permitted purpose, an additive is put in deliberately and to a limit. How a list of numbers turns into an instrument of fear is covered in our look at the Dirty Dozen. How a mycotoxin differs from an additive is shown in our piece on mold on food: nobody puts it in, it appears on its own. The wider picture sits in our food safety hub.