New study: Testing ZOE’s new 'Processed Food Risk Scale'

ZOE's scientists recently designed the Processed Food Risk Scale, which predicts how a processed food product is likely to impact health, simply by snapping a photo. We recently put it to the test in a scientific study.

We presented our results at the Nutrition Society Winter Meeting on January 21st, 2026. Below, we explain why we designed the tool, how it works, and what our study found.

In the UK and US, ultra-processed foods (UPFs) make up an average of almost 60% of our total energy intake. Study after study has confirmed that diets high in UPFs are associated with poorer health outcomes. 

Although this seems like an open-and-shut case, there is a fly in the ointment. Currently, the NOVA system, proposed as a research tool by Carlos Monteiro and his colleagues in 2010, is the most common way to determine whether a product is a UPF.

In other words, it defines some foods as UPFs when they’re actually neutral for health or even beneficial. As an example, NOVA lumps Weetabix (high in fibre with few additives) and Haribo (no nutritional benefits whatsoever) into the same NOVA category - UPF.

If you're interested, you can read more about the issues with NOVA here.

This is why ZOE’s scientists designed a novel tool that not only tells you whether something is highly processed, but also how it is likely to influence your health, especially when eaten frequently. As the abstract explains:

“There is an urgent need for a practical tool that characterises processing features relevant to health, to guide consumers in making healthier food choices.”

We don't buy the hype — and neither should you

Our new app reveals what the food labels won't, using data from the world's largest nutrition study run by ZOE.

Download

What we did

To build the ZPF Risk Scale, we focused on specific features of processed foods. These features have been scientifically or clinically linked to poor health outcomes, but are not currently accounted for by the NOVA system, as they are more recent discoveries:

• Additives: NOVA does take the number of additives into account. However, ZPF also takes into account the type of additives. This is important because some, like vitamin C, are not a problem for health, while others, like emulsifiers, might be.

• Hyperpalatability: This means a product is so delicious that it overrides your fullness signals, encouraging overeating. Hyperpalatibility is created by using specific amounts of fat, simple sugars, carbohydrates, and sodium. 

• Food matrix disruption: If the structure of the food is destroyed, it becomes easier to consume a large amount of energy in a short period (energy-intake rate).

By combining these features, we could categorise processed foods into four classes:

Putting it to the test

Once our scientists had designed the ZPF Risk Scale, they explored its links with heart and metabolic (cardiometabolic) health.

To investigate, they took data from 550 US participants in the ZOE PREDICT 2 study. These people provided weighed food diaries, which is the gold standard method of assessing dietary intake.  

We also collected blood samples and other information from each participant, including BMI, age, sex, education level, ethnicity, total energy intake, and diet quality.

This information enabled us to investigate associations between how much ZPF1–4 foods participants consumed and markers of good and poor health. 

What we found

These results were presented at the Nutrition Society Winter Meeting on January 21st, 2026. In all, 550 participants took part; 74% were female, and the average age was 43. 

• 23.1% of their energy came from ZPF-1 

• 17.4% of their energy came from ZPF-2

• 23.5% of their energy came from ZPF-3

• 17.5%  of their energy came from ZPF-4

Using the NOVA classification:

• 38.4% of their energy came from UPFs

Examining individual food groups revealed how the ZPF Risk Scale offers more nuanced insights. As an example, here’s how breakfast cereals fit into the scale:

• 18.8% of cereals were ZPF-1

• 45.2% of cereals were ZPF-2

• 19.9% of cereals were ZPF-3

• 16.1% of cereals were ZPF-4

In contrast, using the NOVA classification, 78.5% of cereals were defined as UPFs.

Does the ZPF Risk Scale work?

Finally, our scientists needed to check that the ZPF categories were associated with markers of health. Thankfully, they were.

For instance, they found that a higher intake of ZPF-4 foods (the highest risk foods) was significantly associated with: 

• Higher BMI and weight.

• Elevated glycA (a marker of inflammation). 

Interestingly, they found that a higher intake of ZPF-1 foods (no risk processed foods) was linked to 

• Lower BMI and weight.

• Lower levels of glycA.

Our new app gives you the power to see beyond the marketing

Make smarter, science-backed food choices in seconds. Scan. Score. Reveal the truth.

Download

Why it matters

UPFs are never out of the news. While it’s important that people understand which foods are likely to support their health, the NOVA classification system is unhelpful.

It wrongly demonises foods that are fine to eat (like those in ZPF-1), adding to consumers’ confusion.

Taking the breakfast cereal example above, you can see that according to NOVA, more than three-quarters of products are classified as UPFs. But when we use the ZPF Risk Score, almost three-quarters are low or no risk.

The ZPF Risk Scale is a huge step forward. Not just for nutrition science, but for public health. ZOE’s scientists write:

“The ZOE Processed Food Risk Scale captures specific processing features driving associations with health, providing a more precise and less stigmatising classification than NOVA. By distinguishing harmful from benign or potentially beneficial processed foods, the ZPF scale provides an evidence-based tool to inform consumer education, public health policy, and nutrition research.”

If you'd like to use the ZOE Processed Food Risk Scale yourself, download our app now.