In our study, we observed that the average dietary diversity score was 5 food groups, and about two thirds of the older adults reached the MMD (consumption of at least five of the ten food groups). Women consumed significantly less legumes, and consumed more dark green leafy vegetables, other fruits and vegetables rich in vitamin A, and other fruits. This same trend of difference between the sexes is already observed in national studies with adults – according to the Surveillance System for Risk and Protective Factors for Chronic Diseases by Telephone Survey (Vigitel), for example, in the set of 27 cities surveyed in 2021 (capitals of the 26 Brazilian states and the Federal District), the frequency of consumption of beans on five or more days of the week was 60.4%, being higher among men (65.9%) than among women (55.8%). The frequency of consumption of five or more daily servings of fruits and vegetables was 22.1%, being higher among women (26.4%) than among men (16.9%)11.
Oliveira et al. (2015), in a study that evaluated the inadequacy of fruit and vegetable consumption by 1,255 adults and older adults using primary care facilities in Belo Horizonte (MG) in 2014, found an inadequacy of 76.0% among women, prevalence statistically lower than that observed among men (83.8%). The authors discuss that men as less concerned with their health conditions, and it is also assumed that they have less knowledge about current dietary recommendations, considering the consumption of vegetables may be less important for health12. A study with English older adults also described that gender differences in fruit and vegetable intake were substantially attenuated by controlling for nutritional knowledge, that is, the lower nutritional knowledge of men explains a significant part of their lower consumption 13.
We also observed that older individuals had a higher consumption of dairy products and a lower consumption of eggs. The Brazilian National Dietary Survey, conducted in a sub-sample of the Household Budget Survey (NDS-HBS, in Portuguese Inquérito Nacional de Alimentação-Pesquisa de Orçamento Familiar—INA/POF) already showed that older persons had a higher prevalence of consumption of whole milk and lower consumption of eggs compared to adults 6, but studies that compare the intake of specific food groups between different age groups in older groups are rare. A study that compared the dietary intake of two cohorts of participants of SABE Study, in Sao Paulo, showed that those born between 1946 and 1950 (younger cohort) had significantly lower consumption of dairy products than the older cohort (born between 1936 and 1940); when the prevalence was evaluated separately by sex, this difference was significant only among men 14. It is not possible to know the reasons involved in these differences, but it is possible to raise the hypothesis that the older participants in the present study have already received guidance from health professionals about the importance of dairy products in bone health, since they are the main sources of dietary calcium. Previous studies already descried that older adults are more likely to make positive dietary changes due to greater understanding of the benefits of a proper diet, or due to chronic diseases requiring the adoption of healthier habits 15.
Nevertheless, a more recent study evaluated the evolution of food consumption of the Brazilian population using data from the INA/POF of 2008–2009 and 2017–2018, and described a decrease in the consumption of rice, beans, meat, bread, fruit, milk and dairy, processed meats and carbonated drinks, as well as an increase in the consumption of sandwiches, regardless of sex and age; in both sexes, the decrease in rice consumption was more pronounced in older adults, as well the increase in consumption of sandwiches, which showed an estimate of more than 4 times more chances among older adults16. These data show that, even in older cohorts, that seem to be more conscious about diet quality, the substitution of traditional meals with snacks and convenience foods is also present in this population, which have high calorie content and excess sodium or sugar, which can have a negative health impact 16.
A study that analyzed the trends in dietary intake among older Americans from 1977–2010 showed that milk was the third major source of energy among older Americans in 1989-91 and the fourth in NHANES 2005-10, showing that maybe younger cohorts are indeed decreasing their dairy consumption; on the other hand, dairy desserts were in sixteenth in the food rank, and jumped to the fifth place in the same period 15 – this may reinforce a shift from unprocessed or minimally processed food to ultraprocessed products also in older adults.
Considering the differences described here in the group of eggs, no consistent data were found to compare our results. In a relatively recent past, the restrictive consumption of eggs was recommended, due to their cholesterol content, mainly in people with dyslipidemia or risk of cardiovascular diseases; however, this recommendation is outdated and no longer used in dietary guides in most countries, including Brazil, since the literature has already shown that moderate egg consumption has little impact on the lipid profile17. However, it is possible to hypothesize that older participants continue to restrict egg consumption considering these older recommendations, but it is not possible to discuss this issue in depth based on the design and variables of the present study.
Regarding the mean DDS and prevalence of MDD described here, it was observed that a considerably high proportion of the participants does not reach the recommended minimum diversity. There are few diversity studies with older populations, most of them quite recent, using different indicators, which makes it difficult to compare these data.
Study by Rodríguez-Ramírez and collaborators (2022) evaluated data from more than 10,000 Mexicans of both sexes and all age groups (≥ 1 year old) using the same diversity indicator (10 groups proposed in the MDD-W), and described mean DDS of 4.75 for older adults, lower than that found in the present study; about 68.4% of them did not reach the MDD, considering, however, that adults and older populations should have a minimum consumption of 6 groups to achieve greater probability of adequacy9. A study conducted with older adults in Thailand found a mean DDS of 18.4, using an adaptation of the version adopted by FAO, which ranges from 0 to 32 points, that is, on average, the score was slightly higher than 16, value that would correspond to half the score in this version 18.
Regarding the factors associated with the DDS score, the present study found that higher diversity was positively associated to income, previous diagnosis of cancer and sporadic alcohol intake; on the other hand, the presence of cognitive decline, sedentary lifestyle and anorexia of aging were associated with lower diversity.
A population study in Spain found that DDS was significantly higher in women, non-smokers, and with lower educational level 19. In the present study, there was no difference regarding sex, and higher income was associated with higher diversity. Some studies have already shown that higher income and education are associated with better diet diversity, both in the older adults and in other population groups 10,20−22; however, Otsuka et al. (2017) did not observe an association between diversity and educational level in Japanese 23, in line with the result presented here, because despite schooling being associated with DDS in the crude analyses, it did not remain significant in the adjusted model. Possibly, this occurred because both variables are important indicators of socioeconomic status and are closely related in developing countries, and when placed together in the regression model, they can attenuate each other's effect.
The relationship between socioeconomic status and the quality of the diet is still controversial in the literature. In general, it is considered that more educated individuals possibly value healthy eating more, since they understand the importance in the prevention or control of NCDs, also tending to have healthier behaviors such as the practice of physical activity 24,25. In our study, physical activity was also associated with higher DDS, indicating behaviors towards healthy practices in general. However, higher income and education may also be associated with greater access to lower quality foods, such as ultraprocessed foods 26,27. In the case of older populations, possibly income has a greater effect than education in obtaining higher dietary diversity; moreover, the degree of processing is not analyzed in the diversity indicator, so it is possible that this effect exists and was not captured in our results.
In our study, social support was not significant in the final model, despite being widely described in the literature as an important factor associated with better dietary quality in older people and lower likelihood of food insecurity28–32. A qualitative study that analyzed food choices and access to food among low-income older adults pointed out that social interactions, especially with family and friends, can positively affect the eating behavior, increasing the consumption of healthier meals and decreasing the consumption of fast meals, such as toast and cereal33.
Indeed, the importance of social interactions for older adult's diet has been highlighted by the Brazilian Ministry of Health - in the "Protocol for the Use of the Food Guide for the Brazilian Population in Dietary Guidance for the Older Adults", developed as an instrument to support clinical practice in Primary Care, there is an incentive to have meals in the company of family or friends to make this moment more pleasurable, which gives a sense of belonging and integrity, stimulating food intake 34. Possibly, in the present study, social support lost significance because it was also associated with several other model variables, such as income and physical activity.
Reporting a previous diagnosis of cancer, on the contrary, was associated with a higher score on the EDD. These results could be explained by the hypothesis that changes in eating habits and lifestyle may occur after the diagnosis of the disease, and as our study is cross-sectional, these changes may have already occurred in participants with a previous diagnosis. In 2022, the American Cancer Society published a new guideline on nutrition and physical activity for cancer survivors, emphasizing the importance of lifestyle changes from the moment of diagnosis, for better tolerance of treatment, prevention of recurrence and possibly delay in mortality35. Liu and colleagues (2021) conducted a cohort study of nearly 18,000 elderly people to assess the impact of changes in dietary diversity on mortality in China, and found that those who had a drastic improvement in DDS over the course of follow-up also had an increase in mortality, and the authors report that this group had more underlying diseases, which may explain this change in DDS36.
Anorexia of aging showed a consistent negative association in the results presented here. Anorexia of aging is known as a multicausal syndrome, related to diseases, polypharmacy and physiological factors intrinsic to aging itself, in addition to social, psychological, environmental and lifestyle factors that can further affect eating habits and nutritional status, leading to weight loss and malnutrition 37. A previous study of our research group showed that the intake of most nutrients is significantly lower in older adults with anorexia, except for carbohydrates, which may point to poorer quality diets of these persons, preferring palatable and easy to eat items 38. Anorexia screening is essential to prevent malnutrition, sarcopenia, frailty, cognitive decline, and mortality in older adults and should be considered in any geriatric assessment 4,39.
We also found a significant negative association between DDS and cognitive decline, which agrees with other results already described in the literature, such as cross-sectional studies in China and Taiwan 40,41. Longitudinal studies have also shown that less dietary diversity is associated with a higher incidence of cognitive decline over time 23,42−44. It is important to mention that almost all these studies were conducted in eastern countries, such as China, Japan and Taiwan, so it is not known whether diversity also has an impact on cognition in other populations.
However, in the results presented here, it is not possible to identify whether cognitive decline would be a determinant or a consequence of lower dietary diversity, since both directions of association would be plausible. In addition to the fact that less dietary diversity would be responsible for the lack of several essential nutrients for mental health, older adults who already present cognitive deficit may have difficulty in various activities of daily living related to food intake, such as shopping and preparation of items, difficulty in bringing food to the mouth, difficulty swallowing, among others, which can reduce food variety and compromise nutritional status 2.
Lifestyle characteristics also showed an important association with dietary diversity in the present study; sedentary individuals had lower DDS and MDD prevalence than active ones; those who consumed alcoholic beverages sporadically (1 to 4 times a month) also had a higher diversity than those who did not drink or had weekly consumption; and smokers also had less diversity in the crude analyses, but this association lost significance in the adjusted model. These results were, in part, expected, as it is described in the literature that individuals with better lifestyle practices generally also have better diet quality 24,45,46.
There are some studies that seek to understand the interrelation of lifestyle characteristics more broadly. Noble et al. (2015) performed a systematic review describing how the literature evaluated clustering of health risk behaviors called "SNAP" (smoking, nutrition, alcohol and physical activity). They basically found two types of clusters, that had already been suggested by De Vries et al. (2008): addictive behaviors (such as smoking and alcohol) that require restriction or abstinence, and health-promoting behaviors (physical activity and good diet) that require active involvement. However, the most reported grouping was the absence of any of the analyzed SNAPs, that is, most of the studies analyzed in the review found that many individuals who do not present one of the risk factors tend not to present the others 47,48.
There are few studies that have evaluated the association of diet diversity as an indicator of nutrition with other lifestyle markers, therefore, comparison with our results is limited. The results found so far in the literature seem to indicate that smoking is a factor more strongly associated with lower diversity (although this variable did not remain significant in our final model); the results described in relation to alcohol consumption and physical activity are null or conflicting. A study with elderly Japanese people found a negative association between diet diversity and smoking and alcohol consumption (classified as current, previous or never, without considering frequency), but not with physical activity, as well as another study in Spain, which also described a negative association with smoking and a non-significant association with the level of physical activity; alcohol consumption was not analyzed 19,22. Another Japanese study also described a negative association with smoking and a non-significant association with alcohol consumption (in volume/day), and did not assess physical activity 23 .
The positive association found in the present study between dietary diversity and sporadic alcohol intake can be explained by the various existing recommendations that relate moderate consumption with lower risks of diseases and cardiovascular mortality 49. These recommendations are generally based on Mediterranean-style diets, in which moderate wine consumption is one of the most marked characteristics 50. It has also been related to better cognitive performance in older adults 51. Thus, it is possible that the older adults with more attention to their diet also consider the recommendation to associate a varied diet with moderate alcohol consumption.
However, it is important to mention that more recent studies have shown that perhaps there is no safe alcohol consumption limit, especially for older adults. A recently published study performed a combined analysis of data from nearly 600,000 participants in 83 prospective studies, and discussed complex and diverse potential mechanisms by which alcohol consumption can exert cardiovascular effects, highlighting that higher alcohol consumption was approximately linearly associated with increased risk of all stroke subtypes, coronary heart disease excluding myocardial infarction, heart failure, and several less common subtypes of cardiovascular disease. The authors discussed that, although the threshold for the lowest risk of all-cause mortality was around 100g per week, for cardiovascular disease subtypes other than myocardial infarction, there were no clear thresholds, therefore, their data support the adoption of minimum limits of alcohol consumption, considerably below what is recommended in most current guidelines 49.
This study has limitations that must be considered when interpreting its results. First, the method used to assess food consumption (24h-FR) is subject to memory bias, which may be compromised among older adults, although all interviewers have been trained and used photographic manuals in order to minimize the possible errors. It is also important to mention that the 24h-FR is a method that describes the interviewee's current food consumption, and may not be suitable for estimating usual consumption. However, it is believed that this bias was minimized considering that: 1) this is the method used by the indicator according to FAO (2016)7; 2) older adults tend to have a monotonous diet most of the time, varying little between days 3; 3) it is well described that if the sample is large enough, as the one studied here, even a single day of consumption per individual can be used to estimate the average habitual consumption of the population, eliminating extreme values due to population distribution, and, to estimate intra-individual daily variation, it is usually statistically more efficient to increase the number of individuals in the sample than to increase the number of days above 2 days per individual 52,53.
Moreover, some important variables that could be related to a higher dietary diversity were not available in the questionnaire, and therefore were not analyzed, such as access to markets and places where food is purchased, food environment, or functional status. It is possible that, if they were included, they would bring different results to the study. However, the sample had good functionality, considering that all respondents went to the research site (primary care units), and the high prevalence of physical activity.
Furthermore, the diversity indicator used here (MDD-W) has not been validated in older populations. However, it has been the method indicated by FAO to assess dietary diversity in different populations, not only in women of reproductive age, and has been used with other groups, such as children and adults, including latinos 7. Rodríguez-Ramírez and collaborators (2022), for example, used this indicator to analyze data from more than 10,000 Mexicans of both sexes and all age groups (≥ 1 year of age)9. The Estudio Latino Americano de Nutrición y Salud (Latin American Study of Nutrition and Health - ELANS), conducted in 8 countries in South America (Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, Peru and Venezuela) also used this indicator to assess the dietary diversity of individuals aged between 15 and 65 years of both sexes 10.
Finally, it is important to discuss that LMICs, such as Brazil, are undergoing an accelerated and heterogeneous nutritional transition, characterized by increased intake of unhealthy fats, refined carbohydrates and added sugar, i.e., the use of dietary diversity scores can be a major limitation in these contexts, as they do not capture the 3 important dimensions of diet quality (adequacy, variety, and moderation); in particular, the moderation dimension is often absent or inadequately evaluated 54,55. However, its objective is to evaluate the minimum presence of the food group, being considered it as a reliable proxy of adequacy, in addition to diversity.