This study provides a cross-national assessment of dietary patterns in relation to the PHD across nine European countries. Our findings show large differences between countries and significant sociodemographic variations, but also point to a convergence toward overall low adherence to PHD targets. While some countries had relatively high intakes of vegetables, dairy, and fish, these were counterbalanced by excessive consumption of red meat, saturated fats, and added sugars, as well as insufficient intake of whole grains, legumes, and nuts. In absolute terms, European mean intakes standardised to 2500 kcal underscore the magnitude of these gaps. Average consumption of whole grains (~ 30 g/d; PHD target = 232 g/d), legumes (~ 30 g/d; PHD target = 75 g/d), nuts (~ 5 g/d; PHD target = 50 g/d), and vegetables (~ 240 g/d; PHD target = 300 g/d) were well below of the PHD recommended average targets. In contrast, intakes of foods to balance and to limit exceeded PHD targets markedly, including red meat (~ 100 g/d; PHD target = 28 g/d), added sugars (~ 55 g/d; PHD target = 31 g/d), and saturated fats (~ 35 g/d; PHD target = 11.8 g/d). Notably, most of the countries failed to meet the PHD targets for both encouraged and limited foods , which is consistent with prior evidence from high-income countries [45, 46].

Marked heterogeneity in European dietary patterns was observed, in line with cultural traditions and established nutritional trends. Northern European countries, such as Finland, reported high whole grain consumption, consistent with traditional rye- and oat-based foods previously associated with improved health outcomes [47, 48], whereas intake was low in France, highlighting persistent gaps despite national efforts [49,50,51]. Legumes and nuts were underconsumed across all countries, in line with previous studies showing that no European country meets recommended targets [23, 52,53,54]. Mediterranean countries, including Spain and Portugal, demonstrated higher intakes of fish and unsaturated oils, consistent with traditional dietary patterns [55, 56]. Northern and Central European countries relied more on animal fats and red meat, a pattern that has been attributed in previous studies to entrenched dietary habits and socioeconomic factors [55, 57,58,59,60]. Although fruit and vegetable intake exceeded global averages [46], most countries still fell short of PHD targets [61, 62], suggesting that public health efforts should continue to prioritize these food groups. Similarly, added sugars and saturated fats were above recommended levels across Europe [63,64,65], underscoring the need for integrated policy approaches that combine regulatory measures, education, and agricultural strategies [66,67,68]. Overall, our results align with previous European evidence and highlight persistent challenges in achieving adherence to healthy and sustainable dietary patterns [69,70,71,72]. A detailed country- and food-group-specific discussion is provided in Supplementary Material. Composite indices enhanced the dietary assessment by integrating the PHD thresholds and encompassing the multidimensional structure of the diet. Furthermore, prior research indicates that these indices are differentially associated with nutritional and environmental indicators, underscoring their utility in evaluating complementary aspects of PHD [35]. Specifically, proportional scoring-based indices (e.g., WISH) better capture dietary variability and nutritional quality, whereas indices using graded or binary scoring systems (e.g., ELI) tend to correlate more strongly with environmental impact measures [35]. PHD indices that are scored using an unbounded proportional metric, such as ELD‑I, capture both dimensions [35]. Moreover, the dominance analysis helped to interpret how different PHD indices operationalise dietary components. The predominance of plant-based foods in WISH, the stronger contribution of foods to limit in ELD-I, and the mixed pattern observed for ELI reflect differences in index structure and weighting. However, these dominance patterns represent statistical contributions under correlated dietary components and should not be interpreted as causal influence. Taken together, these methodological differences underscore the value of applying multiple indices to evaluate complementary aspects of adherence to the PHD and provide a more robust assessment of dietary patterns.

In this study, composite indices confirmed that overall adherence to the PHD was low. Spain, the Netherlands, and Portugal exhibited the highest scores, while the UK, Hungary, and Estonia had the lowest scores. These findings are consistent with previous country-specific studies. In Spain, the ENRICA study (n = 13,105) reported a mean PHD score of 87 (out of 140) [73], whilst the EPIC-NL study in the Netherlands (n = 35,496) found an average score of 73 (out of 140) [74]. By contrast, data from the UK Biobank (n = 125,372) showed a median PHD score of 59 (out of 110) [75], and a nutrient-based EAT-Lancet score in a study from Hungary (n = 359) yielded a median of 2 (out of 12) [76]. Collectively, despite differences in scoring metrics and study designs, these studies reveal patterns consistent with our findings, reinforcing the evidence that adherence to the PHD is generally low in high-income countries [70].

Low alignment with the PHD was characterised by excess red meat, saturated fats, and added sugars, alongside insufficient whole grains, legumes, nuts, and vegetables, in line with previous research [45, 70]. Eggs, poultry, and dairy intakes were relatively consistent across countries. The Netherlands stood out in terms of whole grain and nut consumption, while Spain led in fish, legume, and unsaturated fat consumption, which may reflect the closer alignment of the Mediterranean diet with the PHD [70, 77]. Previous research, using a different PHD index from those applied in the present study, has shown that modest adherence to PHD dietary patterns in high-income countries was largely driven by excess red meat and added sugars [45]. In Central and Eastern European countries, such as Hungary and Estonia, tuber overconsumption was an additional contributing factor. Over the period 1990–2018, the greatest improvements in this alternative PHD index were observed in high-income countries (+ 6.5 points), largely due to higher component scores for tubers and nuts, with the Netherlands showing the largest gains [45].

Across the nine European countries studied, age was the most consistent predictor of higher adherence to the PHD, with older adults scoring significantly higher than younger adults, particularly in France, Spain, and Portugal. These findings are consistent with previous research showing healthier dietary patterns and lower consumption of ultra-processed food and meat among older adults in Europe, including Spain [78, 79], Switzerland [80, 81], the UK [82], Portugal [83], and Finland [84, 85]. Older age has also been linked to greater vegetable and fruit intakes [79, 86] and lower fast food and (processed) meat consumption [78, 87]. Previous studies have linked these differences to greater cooking skills, more time for food preparation, adherence to traditional diets, generational norms, increased health consciousness, and greater disposable income to spend on healthier foods [79].

Sex differences were more variable across countries, although women tended to score higher on the PHD indices. This aligns with the literature indicating that women report healthier and more plant-forward diets, including lower red and processed meat intake and higher fruit and vegetable consumption [79, 88,89,90]. Similar sex-related differences have been observed for sustainable dietary indices [88,89,90,91,92] and for specific components, such as vegetable and seafood [85, 93]. This pattern is especially pronounced in Baltic and Nordic regions, such as Estonia, where men’s higher red and processed meat intakes drive the gap [84, 94, 95]. Health consciousness, environmental awareness, and willingness to change dietary habits may explain women’s higher scores, whereas cultural norms and men’s preference for energy-dense foods may contribute to lower scores among men [74, 76, 79, 88,89,90].

Educational disparities appear to be an important correlate of diet in the European countries where education data were available. Regarding this, higher education was associated with better dietary scores, particularly in France, Portugal, the Netherlands, and Estonia, consistent with previous findings linking education to healthier and more sustainable diets, including higher intakes of fruit, vegetable, whole grain, and fish, and lower intakes of red and processed meat [88,89,90, 96, 97]. In France, higher education correlates with plant-forward diets and higher multidimensional sustainability of diets (i.e., environmental, nutritional, economic and sociocultural dimensions) [88]. Across Europe, lower education levels are often linked to higher processed meat consumption, lower fruit and vegetable intake, and higher ultra-processed food consumption [79, 80, 88, 98]. However, in Spain, Switzerland, and Hungary, educational differences were smaller, reflecting patterns similar to those in previous research [81, 99]. These variations in dietary patterns by education level have been linked to higher income levels, better nutritional awareness, food literacy, health- and environment-focused attitudes, and enhanced access to quality foods among individuals with higher education levels [79, 98].

This study drew on harmonised data from a large and diverse sample of adults across nine European countries, offering a multidimensional perspective on diet quality through three complementary indices. However, some limitations should be acknowledged. The cross-sectional design limits causal inference, and self-reported intake may introduce recall and social desirability bias, especially for foods such as red meat and added sugars; however, 24-hour recalls are generally more accurate than other dietary assessment methods [100]. Another potential limitation relates to the lack of usual intake modelling. Although such methods can reduce within-person variability in 24-hour recall data, the heterogeneity in the number of recall days and the structure of dietary information across surveys did not allow a harmonised application [101, 102]. Given that several food groups were consistently consumed far below or far above recommended levels, the overall impact on our adherence estimates is likely limited [101]. To further assess the robustness of our findings to differences in survey design, we conducted a sensitivity analysis excluding the UK and France, the two countries with a higher number of recall days (Supplementary Material: Figure S15). Results were very similar, and the overall interpretation of shortfalls and excesses relative to PHD targets remained unchanged. Nevertheless, future research should incorporate harmonised usual intake modelling to improve the estimation of intake distributions and better quantify the proportion of populations meeting or exceeding dietary recommendations.

Another methodological consideration relates to the use of foods “as consumed” in the harmonised datasets. While this approach reflects real-world intake more accurately, it also introduces heterogeneity because the use of conversion factors may differ across original national surveys [103]. This may affect comparability for food groups whose weight change markedly with cooking, particularly legumes, whole grains, or meat. In addition, the PHD specifies reference values for legumes in their dry form, whereas intake in our datasets was captured only as consumed. Thus, although we applied standard harmonisation procedures, these transformations may introduce uncertainty in the estimation of adherence. Future research should aim to harmonize conversion factors across surveys and, where possible, collect both raw and cooked weights to align more precisely with PHD recommendations.

On the other hand, national surveys varied in terms of demographic composition and collection periods (e.g., Estonia 2013–2015 vs. UK 2020), affecting comparability. As a result, the country rankings presented here should be interpreted as time-stamped snapshots rather than stable national characteristics. Stronger temporal harmonisation is required to assess the persistence of these patterns over time. Some datasets lacked key variables, such as education in the UK and Finland, limiting the analyses. In addition, only two education levels were used, which was the most feasible approach to harmonize data due to the heterogeneity of education classifications across surveys, which may restrict the ability to draw strong conclusions. Future research should consider additional factors such as income, employment, and urbanisation [76]. Standardised recipes ensured comparability but may not capture regional differences in preparation, portion size, and ingredient quality, potentially misestimating foods that are difficult to quantify, such as added sugars or fats [104]. We encourage future research to explore more detailed assessments of nutrient composition, for instance fat quality (e.g., the ratio of unsaturated to saturated fats), as more comprehe