This study quantitatively compared the nutrient intake, health, and environmental impacts of hypothetical alternative dietary scenarios replacing animal foods with plant-based analogues, utilising the consumption data from the last Portuguese National Dietary Survey (2015–2016) as the baseline.

In the isoenergetic alternative scenarios, minimal discrepancies were observed in the total grams of food consumed within a day when PBAs substituted animal-based foods. This result hints at the scenarios’ practicality as possible real-life alternatives without disrupting the current dietary pattern, as described previously for PBAs [6, 7]. While the deviations in the quantity of food consumed were slight, the prevalence of inadequate protein, vitamins A, B6, riboflavin, and calcium intake (calcium only for the vegan scenario), and the probability of inadequacy for vitamins B12 and phosphorus increased compared to the reference, particularly in the Vegan scenario. These findings align with a previous study [33] and are expected as animal foods are important sources of these nutrients [12].

On the contrary, fibre, folate, magnesium, and iron intake increased in the alternative scenarios. If the fibre, folate, and magnesium results were expected, since plant-based foods typically contain high levels of these nutrients, the iron result was unexpected since iron inadequacy is described as a frequent drawback of vegetarian dietary patterns [12, 34]. This result was already described in a previous review of nutrient adequacy for plant-based diets [35]. It might emerge because most PBAs in our database are based on pulses and nuts and are probably iron-fortified. However, the potential lower bioavailability of iron in these products was not considered in the analyses, hampering a steady conclusion on this aspect, according to what has been suggested in a recent study [36].

Regarding DALYs, the health impact of the alternative scenarios depended on whether PBAs were assigned the epidemiological risk estimates associated with UPF consumption, that is, whether PBAs are considered UPF for the model. If PBAs were treated as UPF, the ovolactovegetarian and vegan scenarios would not result in net health benefits, with the vegan scenario even suggesting a potential public health risk at the highest replacement proportion studied, particularly among females. Notably, the classification of PBAs as UPF or non-UPF does not change their intrinsic nutritional composition or the actual effects on human health; rather, it affects the predicted health impact in the model by applying different epidemiological risk estimates. Arguments exist for both standpoints (i.e., considering PBAs as UPF or non-UPF). On the one hand, as mentioned in the introduction section, most epidemiological evidence linking UPF consumption to health outcomes primarily stems from studies involving foods with nutritional characteristics distinct from PBAs [15]. Accordingly, other authors disagree with classifying all PBAs as UPFs since it implies that all these products are nutritionally unbalanced, which has been evidenced otherwise [37].

On the other hand, apart from the nutritional profile, further mechanisms are proposed to explain the impact of UPF consumption on health outcomes that could be applied to PBAs. Notably, the so-called “matrix effect”, defined as the “food qualitative and holistic fraction”, is considered a central aspect influencing the health effects of foods. The importance of the matrix effect is underscored by the argument that two foods with the same composition but different matrices or structures may have diverse health effects (e.g., fruit vs. fruit juice) [38]. PBAs predominantly comprise a blend of unstructured, purified, fractionated, and highly processed ingredients, aligning with the classic attributes of UPF [39]. Several mechanisms are proposed to explain the role of the food matrix effect on health outcomes, namely its impact on organoleptic characteristics and palatability that could contribute to an accelerated eating rate and delayed satiety signalling, ultimately leading to increased overall food intake and other aspects such as impact on hormonal secretions, transit time, degree of chewing, and particle size after chewing. Furthermore, other considerations include the presence of contaminants introduced during processing and packaging (such as acrylamide, acrolein, BPA, and phthalates) [14, 38].

The NOVA classification, applied in this study to classify foods as UPF, is a practical, systematic approach to classify foods according to their nature, extent, and purposes of the submitted industrial processes. Current nutritional epidemiological research on the associations between the consumption of highly processed foods and health has mainly used the NOVA classification [40], and public health policymaking and guidelines are starting to cover food processing through this classification system [41, 42]. Nevertheless, there is criticism within the scientific community around the NOVA classification. Misclassification and inconsistencies have been described for NOVA, as well as considerable heterogeneity in the nutrient composition of foods from the same NOVA group [43]. In addition, the NOVA classification system is primarily descriptive, with a greater emphasis on sociocultural aspects than on the physical–chemical properties of foods. As a result, this focus may not fully capture the actual impact of food processing and the potential variations in the nutritional profiles of foods [44,45,46]. PBAs’ processing degree might even benefit PBAs over other non-processed plant-based protein sources. Unprocessed plant proteins may have lower nutritional quality than animal proteins due to the limited amounts of essential amino acids and lower digestibility [47]. It has been described, however, that food processing enhances the quality of plant-based proteins by degrading the cell wall, removing non-protein components, and inactivating or reducing antinutritional factors, such as protease inhibitors, phytates, and polyphenols [37].

These concerns around the NOVA functionality, reliability, and robustness build upon the uncertainties of considering the UPF approach in the current study. Moreover, assuming that the effect of all UPF on health outcomes is similar without considering possible differences in the diverse foods classified as UPF by NOVA may be a flawed approach. Still, until now, few longitudinal studies have focused on the specific effect of PBAs on health outcomes. Notwithstanding, a recent large multinational cohort study found an increased risk of multimorbidity of cancer and cardiometabolic diseases for total UPF consumption. Still, plant-based alternatives were not associated with risk in the subgroup analyses [48]. Accordingly, if PBAs yield different effects compared to conventional UPF, the findings of this study indicate that transitioning to any alternative dietary scenario (pescatarian, ovolactovegetarian, and vegan) would benefit the Portuguese population. The most considerable benefit would be observed quantitatively in the pescatarian scenario, where only meat is replaced by PBAs, with around 89 k DALYs saved (100% scenario), which could be explained in a simple manner that each year, almost 89 k individuals were expected to live one more healthy year if PBAs replaced meat. The comparatively lower quantitative benefit in the ovolactovegetarian and vegan scenarios suggests that the advantages of increasing fibre solely through plant-based substitutes of fish and dairy may not sufficiently counterbalance the simultaneous risks associated with reducing the original animal-based versions. These findings emphasise meat as the pivotal factor for the necessary dietary transition, particularly concerning the health impact, while minimising the required changes in the population’s dietary pattern and the potential inadequacy of some relevant nutrients like protein, vitamins A, B6, B12, riboflavin, and phosphorus. The results slightly differed by sex, with males, in general, benefiting further from the substitution of meat with a plant-based counterpart. This result may be explained by the current dietary pattern differences between males and females, as males generally have a higher animal-based food consumption than females, as shown in previous studies from this population [49]. Also, the incidence of the health outcomes considered, along with the respective burden of disease, estimated in DALYs, differs between sexes [26] and helps explain the differences found.

Regarding the environmental impact, both indicators observe the most substantial changes in the vegan scenario. However, upon comparing the vegan and pescatarian scenarios, the magnitude of difference may not justify a complete transition, given that the substitution of meat alone already leads to a significant reduction in both GHGE and LU. There are still some considerations regarding this assessment. Namely, the information on the impacts of PBAs is limited in the dataset used (SHARP-ID). Consequently, the Tofu and Textured-Soy protein values from the SHARP-ID were attributed to all PBAs. A recent study that reviewed the environmental impact of meat substitutes has reported varying environmental impacts in plant-based meat products depending on product formulation, especially for egg-containing PBAs [8]. However, the effect of this limitation in the results most likely does not affect our conclusions, as they are aligned with the ones from the same study when comparing the environmental impact of PBAs with meat.

To our knowledge, this is the first quantitative risk–benefit assessment evaluating the impact of substituting animal-based foods with plant-based analogues using a health composite metric, the DALYs. By minimising the disruption of the current dietary patterns, the transition from animal-based foods to PBAs reflects a possible, feasible and more realistic approach for the first stage of the necessary dietary transition to a more sustainable dietary pattern. This assessment used a second-order Monte Carlo simulation to address both exposure variability and uncertainty in dose–response and DALYs data. However, additional sources of uncertainty could change the quantitative health impact estimates (∆DALY) and, thus, should be discussed. Namely, there are other constituents of PBAs, other than fibre, that can impact health outcomes and are not being considered, such as sodium. Many PBAs have been reported to contain substantial sodium levels [50, 51], which can introduce potential unquantified health risks of these foods, as sodium is associated with increased risk of cardiovascular diseases [52]. Quantifying sodium using food consumption data is challenging mainly due to the variability in discretionary salt use in recipes and at the table during meals [53]. In the food consumption data from IAN-AF 2015–2016, an average amount of added salt is considered in meat, fish, and egg recipes to avoid underestimating sodium intake. The estimated increase in sodium intake in the alternative scenarios was not very pronounced, as depicted in Figure S4 (Online Resource) and indirectly shown in Tables 3 and S2 (Online Resource). Still, it remains unclear whether substituting these animal-based foods (typically with low sodium levels) with the respective PBA would decrease discretionary salt use, further smoothing the slight differences in sodium intake. For these reasons, it was decided not to include sodium in the quantitative RBA.

Furthermore, the results should be carefully interpreted, as other health effects could have been missed in the quantitative assessment by changing exposures in the scenarios. Still, the selection of health outcomes was based on a thorough methodology, selecting only the effects where high-quality dose–response data were available. The degree of evidence was agreed upon by international agencies or umbrella reviews [22,23,24,25] as high, strong, convincing, or probable, strengthening the confidence that these results reflect the best available evidence thus far. Furthermore, nutrient adequacy was assessed to complement the quantitative ∆DALY results and give further insight into the scenarios’ diet quality by considering additional possible trade-offs. Also, the results are highly dependent on the definition of alternative scenarios. In this study, three alternative scenarios (Vegan, Ovolactovegetarian, and Pescatarian) with different proportions of substitution (proportions: 33%, 50%, 67%, 100%) were considered. Although the definition of these proportions was ad hoc, they allowed us to capture a wide range of substitution levels, providing a representative gradient of substitution effects. Still, other cut-offs (e.g., 25%, 50%, 75%, 100%) could have been used as well, and would yield slightly different results. Nevertheless, we do not expect such a change to affect the overall interpretation and conclusions of the study materially.

Moreover, this study focused specifically on Portugal, its general population’s current consumption of animal-based foods [16, 17], and used country-specific data on DALYs for the selected outcomes [26]. Thus, the estimated health and environmental impact of these scenarios may differ in different settings, with varying consumption patterns of animal-based foods and other disease burden values concerning the selected health outcomes, as described in a previous modelling study [2].

An additional remark must be made regarding the terminology used to define the scenario levels in this study, namely the definitions of “vegan”, “ovo-lacto-vegetarian”, and “pescatarian”. These terms were used to facilitate the description of results and the interpretation of the view of the foods replaced by PBAs in each alternative scenario. Still, they should not be regarded as a faithful representation of the traditional definition of these dietary patterns. These dietary patterns are frequently described as nutritionally balanced, in line with available evidence on healthy eating [2]. This may not necessarily be true for the alternative scenarios modelled in the current study.

In conclusion, this study’s findings support PBAs as a viable alternative to meat products, resulting in public health benefits for the Portuguese population and reduced environmental impact if PBAs replace at least one-third of meat products. This substitution would significantly decrease the Portuguese population’s DALYs, regardless of whether the model included the epidemiological risk estimates associated with UPF consumption to PBAs, and would result in the lowest decrease in protein intake of all scenarios tested. Despite the clear environmental benefits, the health impact of substituting other animal-based foods, such as fish or dairy, with PBAs is still unclear. Future epidemiological research, specifically focusing on the consumption of PBAs, would be essential to clarify the effects of these foods on health outcomes.