Making food labels easier to read didn’t stop unhealthy eating

By Tarun Sai LomteReviewed by Lauren HardakerJan 13 2026

A large UK survey reveals that making food warning labels easier to read reduced sandwich consumption at first, but as consumers and manufacturers adapted, the effect faded and eventually reversed, raising questions about one-size-fits-all labeling policies.

Study: Food Label Readability and Consumption Frequency: Isolating Content-Specific Effects via a Non-Equivalent Dependent Variable Design. Image credit: Drazen Zigic/Shutterstock

In a recent study published in Nutrients, researchers investigated associations between food label readability and intake frequency in the UK.

Why label visibility alone may fail to change diets

Diet-related health issues represent a critical challenge in the UK. Ultra-processed foods (UPFs) account for more than half of the dietary energy intake. Among UPFs, ready-to-eat meals (REMs) are rapidly gaining popularity, and unlike snacks, they are consumed as complete meal replacements. Consequently, the nutritional profile of REMs has a significant impact on the overall quality of the diet.

In 2013, the UK introduced the Multiple Traffic Light (MTL) front-of-pack nutritional labeling policy. While self-reported understanding is high, data indicate a modest shift in REM purchasing behavior. The plausible explanation could be the distinction between label presence and readability. A colorful label may be visually salient, but it would fail as a communicative tool if the text is below the perceptual threshold of consumers. The study explicitly distinguishes visual salience from readability, positioning print size as a structural “gatekeeper” for effective label communication.

Tracking consumption shifts across six years of labeling

In the present study, researchers examined the associations between perceived MTL readability and self-reported frequency of food intake in the UK. They analyzed data from four waves of the Food and You Survey (2012–18). Participants reported intake frequency for 14 food items. The researchers focused on unhealthy labeled target REMs, such as precooked meats and prepackaged sandwiches, and controls, including fresh meat (unlabeled) and dairy products (labeled). These control categories were selected as non-equivalent dependent variables (NEDVs) to help isolate the effect of label readability from broader dietary or market trends.

The original frequency scale was transformed into a three-point variable: 1) never, 2) monthly, aggregating frequencies from once a fortnight or month to less than once per month, and 3) weekly, aggregating frequencies from once or twice per week to at least once daily. Perceived MTL print size was the independent variable. Across survey waves, participants reported how easy it was to read food product labeling in terms of the print size on a five-point Likert scale.

Sociodemographic, geographical, and behavioral variables included age, sex, household size, religious affiliation, shopping responsibility, household income, country of residence, urbanicity, and other relevant covariates available consistently across survey waves. The researchers used ordinal logistic regressions to assess the associations between MTL print size perceptions and intake frequency of target REMs and controls. The survey year was modeled categorically to capture non-linear temporal dynamics following the policy’s introduction.

Early deterrence gives way to increased intake over time

The study included 8,948 participants. High-frequency sandwich intake was more prevalent in females, and frequent consumers were significantly younger, with high representations from those aged 18–24 (20 %) and 25–34 (24 %). Individuals aged 65 and above were predominantly non-consumers. In contrast, consumers of precooked meats had a distinct demographic profile. Most weekly consumers of precooked meats were female (53 %), whereas most non-consumers were male (55 %).

Moreover, older cohorts had high rates of consumption, with those aged 65 and above constituting a significant proportion of weekly (21 %) and monthly (25 %) consumers. Christians represented a substantial majority of weekly consumers, while individuals of other religions were disproportionately prevalent among non-consumers. Furthermore, the consumption patterns of fresh meat and dairy products exhibited distinct distributions.

Dairy product intake was virtually universal, with a majority of participants being weekly consumers. A balanced gender representation was observed in this segment, with a substantial proportion of older adults aged 65 years or older (21 %). For fresh meat, most non-consumers were male (60 %), whereas most consumers were female (53 %), with a high incidence of older adults (22 %) in the consumer cohort.

Across food categories, households with a lower income consistently constituted a larger proportion of weekly and monthly consumer segments. Notably, individuals with reduced shopping responsibility were a majority within the high-frequency sandwich intake segment (56 %), compared to non-consumers. However, the distribution of shopping responsibility for other foods was uniform across intake frequencies.

Individuals aged 16–24 consistently had the greatest ease of label comprehension. By contrast, middle-aged individuals faced significant challenges in readability, whereas older adults aged 65 and above had moderate readability. People from high-income households consistently reported greater ease of reading than those from low-income households. Geographically, Welsh residents had the greatest ease of reading, although these differences narrowed over time.

Participants with less shopping responsibility reported greater ease of reading than those with primary responsibility. This pattern is consistent with the study’s broader evidence of heterogeneity in label engagement across consumer subgroups.

Furthermore, the team observed a significant inverse relationship between readability and the frequency of sandwich intake in 2012. Specifically, each unit increment in readability correlated with 9 % lower odds of more frequent consumption. This pattern is consistent with a short-term warning effect of readable red labels on discretionary, impulse-driven food choices.

Notably, this association evolved over the study period, and by 2018, a one-unit increase in readability was associated with a 4 % increased odds of frequent intake. The authors interpret this reversal as evidence of habituation and market adaptation, rather than as a simple failure of labeling. In particular, they note that widespread manufacturer reformulation following the introduction of MTL labels may have altered the informational content of readable labels, weakening or repurposing their original warning signal.

This dynamic association was not found across other food items. The analyses for dairy products, fresh meat, and precooked meat revealed no temporal dynamics, and the association with readability or its interactions with the survey year did not reach statistical significance.

Label effects depend on product context and consumer habits

Taken together, a significant and exclusive dynamic association was observed with prepackaged sandwiches; this relationship reversed from 9 % lower odds in 2012 to 4 % higher odds in frequent consumption per unit increment in readability by 2018. This finding challenges the durability of deterrent effects from increased label readability over time, rather than suggesting that readable warnings are inherently ineffective.

Null associations were found for precooked meat, dairy, and fresh meat. The divergence between sandwiches and precooked meat highlights the importance of consumption context, with label salience appearing most influential for discretionary, time-pressured purchases rather than habitual or planned food components.

Overall, the findings suggest that the efficacy of label salience is contingent not only on the consumer but also on the product’s context and the label’s content. The study also emphasizes that observed associations are potentially bidirectional, as frequent consumption may itself reduce perceived label readability through habituation.

Although the analyses use a strong quasi-experimental design, the results remain associational, and causal direction cannot be definitively established. This underscores the need for adaptive policy standards, moving beyond uniform label designs to consider category-specific strategies, including clarity for impulse food categories and complementary interventions where print size alone is unlikely to influence behavior.

Download your PDF copy now!