# Evaluation of energy density and levels of fat, sugar, and salt in processed and ultra-processed food with existing and novel nutrient profiling models to identify unhealthy food and drink: cross sectional analysis of a nationally representative database of food and drink items.

- **DOI:** [10.1136/bmjmed-2025-001750](https://doi.org/10.1136/bmjmed-2025-001750)
- **Rating:** Sound (3.1 out of 5)
- **Verdict:** A common-database comparison quantifies large differences between UK nutrient-profile rules, but its diet-derived thresholds do not independently validate individual-product healthfulness.
- **Framework version:** v5.13
- **Pipeline version:** v0.4
- **Evaluation date:** 2026-11-18
- **Evaluation page:** https://www.nabu.science/paper/10.1136%2Fbmjmed-2025-001750

## Key claims

1. Food and drink products that, according to UK government dietary guidance, contain excess amounts of harmful nutrients are not identified by current systems for regulating UK food and drink (FSA NPM and front-of-pack traffic lights labelling system).
   - Support: Tentative
   - Type: Primary finding
   - Basis: diet-level guidance is translated into unvalidated product-level thresholds without testing specificity
   - Location: Abstract, Conclusions
2. More than 90% of processed and ultraprocessed products are unhealthy according to current UK dietary guidance
   - Support: Tentative
   - Type: Secondary
   - Basis: whole-diet targets do not independently establish the healthfulness of individual products
   - Location: Discussion, Implications of this study
3. NPM1 identified 91.5% (95% CI 90.2% to 92.9%) of all ultraprocessed products as unhealthy, model No 2 identified 94.9% (93.9% to 96.0%), and model No 3 identified 98.7% (98.2% to 99.3%) of all ultraprocessed products.
   - Support: Supported
   - Type: Secondary
   - Basis: the database supports these specified-rule proportions rather than independently validated health classifications
   - Location: Abstract, Results
4. The most conservative new NPM, based on current UK government dietary guidance, identified significantly greater proportions of processed and ultraprocessed products as unhealthy than the current two systems.
   - Support: Supported
   - Type: Secondary
   - Basis: large classification differences hold within the common database despite uncertain market representativeness
   - Location: Abstract, Conclusions
5. The FSA NPM identified as unhealthy 57.6% of ultraprocessed products (95% confidence interval 55.2% to 60.0%), whereas the traffic lights labelling system identified fewer than half of all ultraprocessed products as unhealthy (48.2%, 45.8% to 50.6%).
   - Support: Supported
   - Type: Secondary
   - Basis: the Abstract and Table 4 consistently report these database classifications
   - Location: Abstract, Results

## Strengths

- **Common item set isolates classification differences** Current and revised FSA models, traffic lights, and three guidance-derived models are applied to the same 1,898 processed and ultraprocessed items. Table 4 directly quantifies their differing classifications.
  - Dimension: contribution
  - Location: Methods; Table 4
- **Explicit thresholds make comparisons readily inspectable** Tables 3 and 5 distinguish model definitions from nutrient-specific and combined-threshold results. Public analysis code provides an additional route for examining implementation.
  - Dimension: methodological rigour
  - Location: Tables 3 and 5; Data availability statement
- **Regulatory purposes and scope are distinguished** Table 1 distinguishes dietary advice, marketing regulation, and voluntary labelling. The Discussion explicitly acknowledges that the analysis can overestimate real-world FSA regulatory coverage.
  - Dimension: positioning
  - Location: Table 1; Discussion, Strengths and limitations

## Limitations

- **Dietary targets do not validate product healthfulness** Whole-diet recommendations become individual-product thresholds without an independent validity criterion. Excluding all 986 minimally processed foods also prevents assessment of specificity, despite claims that the new models are superior.
  - Dimension: methodological rigour
  - Location: Methods, New Eatwell Guide nutrient profiling models; Discussion, Implications; Strengths and limitations
- **Database coverage and classification uncertainty remain unquantified** An unweighted recall-derived item list is used as a proxy for marketed products, and legal FSA categories cannot be mapped directly. Extrapolated nutrient values and ambiguous Nova coding introduce additional uncertainty without a quantified sensitivity assessment.
  - Dimension: methodological rigour
  - Location: Methods, Food and drink items in scope; Identifying processed and ultraprocessed products by Nova classification
- **Headline interpretation and secondary reporting overreach** The headline attributes author-derived product classifications to government guidance. Secondary text also describes NPM1's three-parameter proportion as greater than traffic lights despite 5.1% versus 11.6%, transposes group labels, and mislabels percentage-point differences.
  - Dimension: reporting
  - Location: Abstract, Conclusions; Results introducing Table 5; Table 5; Discussion, Principal findings

## Reliability checks

Caveats (4 of 4 checks).

- **Results check:** Noted
  - "the front-of-pack traffic lights labelling system identified 48% and 46% of processed and ultraprocessed products as unhealthy, respectively." (Discussion, Principal Findings) vs "Ultraprocessed products (n=1636) | 48.2 (45.8 to 50.6)" (Table 4, ultraprocessed products). The Discussion assigns approximately 48% to processed products, whereas Table 4 assigns 48.2% to ultraprocessed products and 46.2% to processed products. The result itself is unaffected.
  - "identified 43-46% more (nearly twice as many) processed and ultraprocessed products as unhealthy than the traffic lights labelling system" (Discussion, Principal findings, paragraph 2) vs "NPM1 91.6%, 95% CI 88.2% to 95.0%" (Results, paragraph 1). The differences 91.5−48.2=43.3 and 91.6−46.2=45.4 are percentage points, while the relative increases are approximately 90% and 98%. The result itself is unaffected.
  - Minor reporting notes (1):
    - "significantly greater proportions of processed and ultraprocessed products as unhealthy based on containing two, three, or four parameters in excess" (Results, paragraph introducing Table 5) vs "≥3 parameters in excess (any nutrient or energy, or both) | 11.6 (10.0 to 13.1)† | 5.1 (4.0 to 6.1)‡" (Table 5, ultraprocessed products). For at least three excess parameters, NPM1 identifies 5.1% versus 11.6% under traffic lights, contradicting the stated greater proportion. The claim is not supported by the paper's own data.
- **Reporting and transparency:** Noted. Methods and Table 3 disagree about the sugar component of FSA NPM 2018, leaving implementation uncertain. Ethics exemption, competing interests, data-access routes, and analysis code are disclosed.
  - Conflict of interest: declared · Ethics approval: declared · Data availability: declared · Code availability: declared
- **Reference integrity:** Clean
  - 73 references in manuscript
  - 35 of 35 checkable references found in an index
  - 38 are books, websites or datasets — counted, but not index-checkable
  - References not verified automatically are resolved manually within 24 hours.

## Post-publication record

- **Status:** Clean
- **Summary:** No retraction or expression of concern found in Crossref (includes Retraction Watch).
- **Last checked:** 2026-11-18
