Cardiovascular

Ultra-Processed Food-Rich Dietary Proxies, Dietary-Pattern Context, and Cardiometabolic Outcomes in Spain: Weighted Evidence Consistent with Age Confounding and Possible Reverse-Causality Bias from the 2023 Spanish Health Survey.

TL;DR

Weekly fast-food consumption was the most consistent adverse dietary marker associated with all prevalent cardiometabolic outcomes (excess weight, obesity, hypertension, and diabetes) in the 2023 Spanish Health Survey, while patterns consistent with age confounding and possible reverse-causality bias were identified for other dietary proxies.

Key Findings

Prevalence of excess weight, hypertension, and diabetes in the 2023 Spanish Health Survey sample was high.

  • Sample size was n = 21,032 from the 2023 Spanish Health Survey.
  • Prevalence of excess weight reached 53.6%.
  • Prevalence of hypertension reached 23.9%.
  • Prevalence of diabetes reached 7.4%.
  • BMI was derived from self-reported weight and height; hypertension and diabetes were defined as self-reported medical diagnoses.

Weekly fast-food consumption was consistently associated with excess weight and obesity.

  • Fast-food consumption was associated with excess weight: OR 1.20 (95% CI 1.07–1.35).
  • Fast-food consumption was associated with obesity: OR 1.28 (95% CI 1.09–1.50).
  • These associations were observed in weighted logistic models adjusted for sociodemographic and lifestyle factors.
  • Fast food was described as 'the most consistent adverse dietary marker co-occurring across all prevalent outcomes.'

Weekly fast-food consumption was associated with prevalent hypertension and prevalent diabetes.

  • Fast-food consumption was associated with prevalent hypertension: OR 1.18 (95% CI 1.02–1.37).
  • Fast-food consumption was associated with prevalent diabetes: OR 1.33 (95% CI 1.10–1.61).
  • These associations persisted after adjustment for sociodemographic and lifestyle factors.
  • Fast food was the only dietary marker consistently associated with all four cardiometabolic outcomes examined.

Excluding participants with severe chronic comorbidities strengthened the association between fast-food consumption and prevalent diabetes.

  • 4,107 participants with severe chronic comorbidities were excluded in this sensitivity analysis.
  • After exclusion, the OR for fast-food consumption and prevalent diabetes increased to 1.46 (95% CI 1.17–1.81), compared to 1.33 (95% CI 1.10–1.61) in the full sample.
  • This pattern was interpreted as empirical evidence against reverse-causality fully explaining the fast-food–diabetes association.
  • Exclusion of severe comorbidities was one of three bias-probing strategies used, alongside stratification by self-rated health and model-based beverage contrasts.

Processed meat was specifically associated with prevalent excess weight and obesity but not with prevalent cardiometabolic diagnoses such as hypertension or diabetes.

  • Processed meat showed associations with excess weight and obesity outcomes.
  • Processed meat was not associated with prevalent hypertension or prevalent diabetes in adjusted models.
  • This pattern distinguished processed meat from fast food, which was associated with all four outcomes.
  • The authors described processed meat as 'specifically associated with prevalent excess weight and obesity rather than prevalent cardiometabolic diagnoses.'

Stratified and sensitivity analyses revealed empirical patterns consistent with age confounding and possible post-diagnosis dietary changes (reverse causality) influencing cross-sectional associations.

  • Bias-probing analyses included stratification by self-rated health, exclusion of n = 4,107 participants with severe chronic comorbidities, and model-based beverage contrasts.
  • The authors noted that cross-sectional surveys 'often underestimate or invert these associations due to age structure, diagnosis-related dietary changes, and reporting bias.'
  • The study highlighted that individuals with existing diagnoses may have changed their diets post-diagnosis, potentially masking or inverting true dietary associations.
  • The authors stated that 'stratified and sensitivity analyses provided empirical patterns consistent with age confounding and possible post-diagnosis dietary changes influencing cross-sectional associations.'

The study used dietary proxy mapping rather than direct NOVA classification to characterize ultra-processed food exposure.

  • Dietary items from the survey were mapped as UPF-rich proxies, processed-meat exposure, protective dietary markers, and beverage comparators.
  • The survey data did not allow direct NOVA classification of individual foods.
  • Weighted logistic regression models were used and adjusted for sociodemographic and lifestyle factors.
  • The data source was the most recent (2023) national health survey from Spain with a total sample of n = 21,032.

What This Means

This research used data from 21,032 people surveyed in the 2023 Spanish Health Survey to examine whether eating ultra-processed foods — particularly fast food and processed meat — was linked to common health problems like obesity, high blood pressure, and diabetes. The study found that people who ate fast food at least weekly were more likely to have all four of these health conditions compared to those who ate it less often. Processed meat was linked to being overweight or obese, but not specifically to high blood pressure or diabetes diagnoses. The researchers also investigated whether the study's design could distort the results. Because this was a snapshot survey (cross-sectional), people who had already been diagnosed with conditions like diabetes or hypertension may have changed their diets after diagnosis, potentially making their current diets look healthier than before — a problem called reverse causality. The researchers tested this by removing participants with severe chronic illnesses from the analysis and found that the link between fast food and diabetes actually became stronger, suggesting that reverse causality was not fully explaining the fast-food findings. This research suggests that fast food consumption is a meaningful marker of cardiometabolic risk even in national survey data, but it also highlights important limitations of cross-sectional health surveys. Age differences between groups and dietary changes after diagnosis can distort associations, making some harmful foods appear safer or even protective. The authors caution that results from this type of study require careful interpretation and should not be used uncritically in public health policy without accounting for these potential biases.

Have a question about this study?

Citation

Cuadrado-Corrales N, Zamorano-León J, Carabantes-Alarcón D, Mérida D, Bodas-Pinedo A, Chico-Rodríguez M, et al.. (2026). Ultra-Processed Food-Rich Dietary Proxies, Dietary-Pattern Context, and Cardiometabolic Outcomes in Spain: Weighted Evidence Consistent with Age Confounding and Possible Reverse-Causality Bias from the 2023 Spanish Health Survey.. Nutrients. https://doi.org/10.3390/nu18162615