Dietary patterns in the healthy oldest old in the healthy aging study and the Canadian longitudinal study of aging: a cohort study

Broadly, healthy aging refers to development and maintenance of the functional ability to enable wellbeing [1]. The emphasis of healthy aging is on healthspan not lifespan with some definitions also emphasizing aging in the absence of disease [2]. The oldest old, those who are eighty-five years and older [3] are among the fastest growing segment of the population in some parts of the world [4]. Few people however, live to this age without developing chronic diseases. It is thus important to understand the influence of modifiable lifestyle factors such as diet on the achievement of exceptional longevity and the role, if any, of these factors on healthspan.

Several studies among long-lived populations have yielded important insight into dietary patterns and nutrients associated with morbidity or life-span. Long-term caloric restriction and traditional dietary patterns such as the Okinawan diet and Mediterranean diet have been found to be potentially important for reducing age-related diseases and supporting longevity [5, 6]. Some studies have compared dietary intake and diet patterns of centenarians to those of ‘younger’ adult cohorts from the same geographical region [7‐12]. Such studies generally report lower energy and/or fat intake for the centenarians as compared to community-dwelling older adults in their 60s or 80s. Conversely, centenarians have also been observed to consume more high-fat foods, and were less likely to avoid dietary cholesterol [10, 13].

There are in relative terms, few studies of the oldest old and centenarians in gerontology research, and even fewer studies of populations that have survived to that age in the absence of a major age-related chronic disease. Studies that characterize dietary intake in such populations will add to our knowledge base and provide critical information for hypothesis testing. Our aim was thus to assess the dietary intake of a population of men and women 85 years and older who were free of chronic diseases and compare dietary patterns to adults aged 65 to 86 years using data from a Canada -wide population-based study of aging. We hypothesized that compared to the younger cohort, older healthy adults would have dietary patterns that more closely follow guidelines for chronic disease risk reduction such as more frequent consumption of fruits and vegetables, whole grains and lean protein.

Table 1 presents the descriptive statistics of 122 SS and 12,626 CLSA participants. A smaller proportion of SS participants lived with partners (37% versus 62%, p < 0.01) and had a Bachelor’s degree or higher (37% versus 49%) compared to CLSA participants. Never smokers were more common among SS (53% versus 43%) and current smokers were less common (0% versus 5.6%) compared to CLSA participants. Differences were also observed for alcohol consumption and BMI, whereby regular drinking and habitual drinking were less and more prevalent in SS than CLSA, respectively. The prevalence of normal BMI (63% versus 28%) was higher among SS participants while the prevalence of overweight and obese BMI were lower among SS (overweight: 28% versus 43% and obese: 4.1% versus 28%) compared to CLSA participants.

Principal component analysis identified two factors that explained 23% of the total variance. Foods contained in each factor and factor loadings are shown in Table 2. The first factor was characterized by western foods. The strongest associations for factor 1 were processed meat (factor loading 0.62), red meat (0.59), sauces and gravies (0.58), fried potatoes (0.54) and non-fried potatoes (0.54). Associations between factor 1 and high-sugar snacks, high-fat dairy, butter and margarine and salty snacks also exceeding factor loading thresholds (> 0.30). The second factor was characterized by nutrient-rich foods including fruits (0.61), non-starchy vegetables (0.69), whole grains (0.46), legumes (0.45), nuts and seeds (0.46), low-fat dairy products (0.31), salad dressing (0.37) and fish (0.49). The factors are hereafter referred to as ‘western dietary factor’ and ‘nutrient-rich dietary factor’.

Table 3 shows the unadjusted and adjusted ORs and 95% CIs for being a SS based on logistic regression. The highest quartile of the western dietary factor was associated with greater odds of being a SS (Model 1, Q4 OR = 2.06, 95% CI 1.29–3.38). After adjustment for confounders, associations were strengthened (Model 2, Q4 adjusted OR = 3.21, 95% CI 1.91–5.51). No associations were observed for quartiles 2 or 3. The highest quartile of the nutrient-rich dietary factor was associated with greater odds of being a SS in the unadjusted model (Model 1, Q4 OR = 1.75, 95% CI 1.10–2.85). Associations were attenuated in Model 2 after adjustment for covariates (Q4 OR = 1.57, 95% 0.95–2.66). No associations were observed for quartiles 2 or 3 in either model. Variables that were associated with group (SS or CLSA) and dietary pattern included sex, ethnicity, marital status, household income, education, smoking status, alcohol consumption and BMI). Of these, smoking, BMI and alcohol consumption had the largest impact on effect estimates. Covariate estimates are shown in Additional file 1: Table 1.

This study contributes new evidence on dietary patterns and healthy aging in a Canadian population. The nutrient-rich dietary pattern was not associated with being a SS after adjustment for confounders, but the highest quartile of the western dietary pattern was associated with greater odds of being a SS. This was in contrast to our hypothesis as the western dietary pattern was predominately characterized by consumption of foods considered to be less healthy including processed meat, red meat, sauces and gravies, fried and non-fried potatoes, high sugar snacks, salty snacks and high fat dairy products. Nonetheless, it is important to note that the western dietary component contained other foods that did not meet our threshold for component loading (e.g. poultry and eggs) but may have still contributed to associations.

Consistent with findings from other studies of older adults including the Georgia Centenarian Study [21, 22], the Alameda County Study [23] and the Harvard College Alumni Study [23], SS engaged in health behaviors known to promote healthy aging [17]. Compared to CLSA participants, SS had a higher prevalence of never smokers. None were current smokers and few were obese, although habitual alcohol drinking was more prevalent among SS.

Our finding of an apparent disconnect whereby the less healthy western dietary pattern and the nutrient-rich dietary pattern were both associated with greater odds of being a SS (in unadjusted models) is similar to findings from the Georgia Centenarian Study [13, 24]. They reported centenarians consumed a more varied diet including higher consumption of milk and grains compared to adults in their 60’s. Conversely, centenarians were also more likely to consume whole milk and biscuits and less likely to follow nutrition guidelines for chronic disease risk reduction. The authors suggested that the differences in diet may reflect functional problems in the very old, for example oral health and chewing functions. We did not specifically query difficulty eating but the prevalence was likely low among SS as the inclusion criteria of no chronic diseases correlates with a generally healthy population, and the prevalence of meal replacements was low among SS (data not shown).

It is unclear whether the tendency of SS to follow a western-dietary pattern reflects generational and/or cohort differences with CLSA participants, longevity-related differences or possibly fatalism (e.g. approaching the end of one’s life) which has a negative impact on health behaviors [25, 26]. For example, higher frequency of high fat foods may reflect ingrained generational dietary patterns. National dietary guidelines were not significantly modified until the 1980’s in Canada to emphasize energy balance and moderation (limiting fat, sugar, salt and alcohol) [27]. A prior study by Takeda et al. [8] reported that dietary trends among two cohorts of centenarians studied in 1981 and 1995 differed with respect to cereal, egg, algae products and legume intake that were attributed to generational differences. There is also a potential influence of sociodemographic factors. Education was lower among SS and a higher proportion of SS lived alone, both of which are associated with poor diet [28]. Associations between the western-dietary pattern and SS remained with adjustment for both variables suggesting that marital status and education were not driving the observation.

There are several limitations to our study. While it is not practical to monitor food intake over the lifespan, current diets may not reflect past dietary habits that may be more likely to contribute to healthy aging. Appropriate comparisons for the oldest old are limited. The vast majority of individuals in the same birth cohort as the oldest old generally do not survive to an advanced age and those that do, have chronic disease may be unable to participate in research studies. Our use of the CLSA cohort from the same geographic areas as SS’s is the same approach used by centenarian studies [13, 24] but it also introduces potential generational and cultural influences. To minimize generational differences we included CLSA participants up to the oldest age range at the time of dietary assessment (age 86), however the majority of CLSA participants were from younger generations than SS; 188 (1.5%) were aged 85 and older. The SDQ was chosen as the dietary tool in our study to facilitate comparison to the CLSA. However, the SDQ was not designed to quantify the amount of foods consumed or nutrient intake, and is likely prone to recall bias inherent in self-reported assessment of dietary intake. Our results and inferences are thus limited to frequency of consumption rather than amount. This may be an important limitation as total food intake decreases with age [8]. Our sample size of 122 SS although large with respect to the rarity of this population is modest for characterizing dietary patterns, and generalizability to other populations is limited. Lastly, there are numerous factors related to dietary intake and healthy aging that were not measured including social factors, physical activity and genetic variation that would be informative to assess in future studies [29].

In conclusion, our results suggest a western dietary pattern is associated with increased odds of healthy aging in this sample of older Canadian adults. This finding adds to the limited evidence base of dietary intake among the healthiest oldest old, but its implications should be considered in the context of possible generational differences in diet and limitations of the assessment of food and nutrient frequency but not amount of food.