This study examined the evolution of recipes in popular cookbooks from 2008 to 2018 and assessed how the ingredients in these recipes correspond with food purchases among Flemish households. These cookbooks revealed a noticeable trend towards fewer meat recipes and a rise in fish and dairy or plant-based recipes. Moreover, there is an increasing preference for plant-based ingredients like vegetables, accompanied by a decline in meat and dairy use. On the macronutrient front, recipes reflected reduced total fat content while showcasing an increase in carbohydrates, sugar, and fiber. Notably, our findings highlight a correlation between food groups in cookbook recipes and the purchasing habits of Flemish households.

A traditional Flemish main course typically includes meat or fish, vegetables, potatoes and often a sauce [5, 29]. Classic Flemish cookbooks and dietary habits traditionally emphasized meat as a central constituent of the Flemish diet [5]. The recent shift towards fewer meat-based recipes in modern cookbooks is noteworthy but aligns with international and Flemish government dietary guidelines that promote plant-based diets and sustainable food consumption [30, 31]. Reynolds (2022) also observed a global trend where omnivorous cookbook recipes are leaning more plant-based [32]. Consistent with this gradual pivot towards plant-centric recipes, a growing segment of the population in Belgium and other Western countries is diminishing their meat consumption, gravitating towards vegetarian or flexitarian diets [33, 34].

In terms of macronutrient composition, our study observed a decrease in total fat content in recipes between 2010 and 2018, with an upward trajectory for total carbohydrates, sugar, and fiber. However, total protein and saturated fat contents remained consistent. Considering the Flemish population’s average intake leans towards excessive fat and insufficient complex carbohydrates, the trends in cookbook content seem beneficial [35]. A plausible rationale for the spike in carbohydrate and fiber content relative to fat is the broader acceptance of plant-based diets. Incorporating more vegetables typically augments the carbohydrate and fiber content, while minimizing meat and dairy reduces fat content [36]. Interestingly, the sugar content rose despite a curbed use of added sugars and sweeteners. The extra sugar, therefore, is likely not added sugars, but natural sugar from food groups like fruit or vegetables, which besides sugar also contain vitamins, minerals and antioxidants [37, 38]. However, our current methodology doesn’t allow us to verify this definitively.

The recipes from 2018 remain notably high in fat and protein, reflecting the influence of influencers who dominated the Flemish cookbook charts between 2013 and 2018, championing low-carbohydrate diets [39]. The Belgian Superior Health Council advises a daily energy intake comprised of 50–55% carbohydrates, 30–35% fat, and 15% protein. However, cookbook recipes from 2018 had 27% carbohydrates, 41% fat, and 32% protein [35]. Despite a sugar increase, these recipes conform to dietary guidelines, with sugar constituting 10% of total calories. Saturated fat accounted for 12% of total calories, slightly above the recommended 10%.

These findings echo previous studies that found recipes from celebrity chefs, online platforms, and healthy eating blogs to be high in fat, saturated fat, and protein [18, 40,41,42]. The influence of globalization on cookbook discourses and food media phenomena like the rise of celebrity chef or influencers, also seems to manifest in cookbook recipes [13]. The macronutrient trends in our study resonate with those from the Belgian National Food Consumption Survey (NFCS) conducted in 2004 and 2014 [43]. Similar to our study, the NFCS reported a reduced total fat intake and increased carbohydrate intake from vegetables by the Flemish population, with consistent total protein consumption. However, unlike our findings, the NFCS did not record significant shifts in total carbohydrate consumption.

The shift in ingredients featured in cookbook recipes further emphasizes the global trend towards plant-based diets. The notable rise in the use of vegetables, nuts & seeds, and cheese, contrasted by a decrease in meat, dairy, sugar, sweeteners, and alcohol, aligns well with international dietary recommendations which advocate for increased consumption of vegetables and nuts & seeds, and a reduction in meat intake [44, 45]. While the Belgian Superior Health Council recognizes dairy as a dietary staple, the evident rise in cheese recipes and the simultaneous decline in other dairy product usage make it ambiguous if the overall dairy consumption aligns with the council’s guidelines [44]. It is therefore, unclear whether the usage of total dairy have become more in line with the guidelines of the Superior Health council. Some studies do suggest replacing dairy with nuts or other plant-based sources would decrease cardiovascular diseases and all-cause mortality. This observed negative correlation between the usage of nuts & seeds and dairy in recipes might then, indeed have beneficial implication [46].

The observed decline in meat and dairy use in recipes has an inverse correlation with the use of vegetables and cheese. Conversely, there is a positive association between vegetable use and both fish and cheese use. This can be attributed to the rising popularity of pescatarian, flexitarian, and vegetarian diets that often utilize fish and cheese as substitutes for meat [47, 48]. While total protein content remains unchanged, the general decrease in meat and dairy usage compared to nuts & seeds and vegetables usage implicates a move towards plant-based protein sources. As it is harder to fulfill the protein requirements with plant-based sources, enough variety remains important [33]. Yet, past research has underscored the health benefits of a higher intake of plant-derived proteins over their animal-based counterparts, highlighting a reduced risk of cardiovascular diseases and overall mortality [49]. Comparing these trends to the Belgian food consumption, the FNCS recorded no significant alterations in meat and vegetable intake from 2004 to 2014. However, the non-profit EVA points to a downward trajectory in meat consumption in subsequent years [50]. The reduction in meat purchases resonates with the prevailing skepticism in the scientific community towards red and processed meats, driven by environmental and health apprehensions [51, 52]. This sentiment seemingly reverberates in both consumer purchasing behaviors and the narratives of mainstream media .

The observed correlations between ingredients featured in cookbook recipes and household purchases illuminate a significant insight into the relationship between media representation and consumer behavior. Notably, there is a pronounced alignment between the rising prominence of plant-based food groups and cheese in recipes and an uptick in household purchases of these items. This concurrence underscores a decrease in meat-centric recipes, leading to households increasing their purchases of alternatives like cheese, oils, eggs, and nuts & seeds. Meanwhile, the amplified vegetable presence in recipes is accompanied by increased purchases of cheese, grains, and nuts & seeds, indicative of a pivot towards dietary guidelines [53]. Fish usage in cookbooks was the only food group that showed a negative correlation with meat usage. However, it displayed a positive relationship with meat in household purchases. This indicates that an increase in fish recipes in cookbooks, as part of a move to reduce meat consumption, does not necessarily lead to more fish purchases. Moreover, as the rise in fish recipes coincides with a trend towards plant-based recipes, individuals might become more aware of their consumption of animal-based products, leading them to decrease their fish purchases. Supporting this idea, fish usage in recipes has a negative correlation with fish household purchases, though it is not statistically significant.

When examining specific food groups, no correlation was found between their usage in cookbooks and the purchases of vegetables, potatoes, and eggs. This lack of correlation is likely because few recipes primarily focus on these ingredients. With oils, the discrepancy might arise because authors do not consistently report the quantity used, potentially introducing bias to the results. A notable negative correlation was observed for the food groups grains and legumes.

As a typical Flemish breakfast comprises cereals or sandwiches with a variety of toppings, including both sweet and savory options, and a typical Flemish lunch consists of sandwiches with cheese or ham, the prevalent use of grain products like bread and cereals in these meals might explain the weaker correlation between the usage of grain products in main course recipes in cookbooks and its occurrence in household purchases [29]. Legumes, on the other hand, are not typically used as a primary protein source in traditional Flemish diets but rather as a vegetable [54]. This might account for their lack of correlation.

Overall, the observed correlations and consistent trends between cookbook recipes and household purchases underscore an association between media consumption and dietary patterns. Boylan et al. (2012) highlighted that media serve as the primary source of food information for most people. Similarly, Proesmans et al. (2022) identified celebrity chefs as the foremost source of food information, with family and influencers following closely behind. These correlations emphasize the viability of cookbook recipe content as a barometer for dietary trends within the general populace, a notion previously posited by Buisman & Jonkman in 2019 [1].

This study presents several limitations that warrant discussion. Firstly, there are subtle discrepancies between the nutritional data procured from the NUBEL and the USDA nutritional databases. This is because ingredients in American food products can vary from those in Belgian products. For instance, cereal grain products in the US are mandated to be enriched with synthetic folic acid since 1997, which may lead to slight differences when compared to Belgian products [55]. Secondly, in our analysis of cookbook recipes, we consistently chose the lower quantity when given a range for an ingredient. Similarly, when presented with multiple ingredient options, we always opted for the first. Such choices might introduce a minor bias in the results. Thirdly, as the literature suggests, some degree of caution is warranted when interpreting food purchase data as a reflection of dietary intake [56]. For instance, unlike food balance sheets, food purchase data does not account for out of home consumption. Given the increasing importance of eating out of home, this omission is a significant limitation in our data [57]. Furthermore, there are differences between what households buy, what they actually consume, and what they use as a main ingredient in a dish. We also do not know how people prepare their meals or whether they follow cookbook recipes exactly. Lastly, a time lag might exist between the acquisition of a cookbook and subsequent household purchases. Given that our data is annual and comprises only ten time points, we were unable to conduct time-series analysis to establish possible lagged effects, as such analyses require a minimum of forty observations [28].

Prior research exploring the relationship between dietary intake and cookbook content has predominantly focused on cross-sectional data or on the content of a singular cookbook [1, 7]. Such approaches are likely inadequate to fully capture evolving trends. Additionally, our annual data on household purchases sheds new light on the highly discussed link between recipe content in mainstream media and dietary patterns, as gauged by household purchases in our study. Given the increasing reliance on media as the primary source of food information, comprehending media’s role in disseminating health and food knowledge becomes imperative for effective health communication and promotion [11]. To ascertain the presence of causal effects, future studies adopting an experimental design are necessary.