The phenomenon of Protein Leverage has been demonstrated in previous research: strong regulation of protein intake causes overconsumption of non-protein energy in diets with a low proportion of energy from protein(Reference Raubenheimer and Simpson1–Reference Gosby, Conigrave and Raubenheimer3). A systematic review and meta-analysis of randomised controlled trials showed that when compared with carbohydrate and/or fat, protein acutely decreased the appetite-stimulating hormone ghrelin and increased the satiety hormones cholecystokinin and glucagon-like peptide-1(Reference Kohanmoo, Faghih and Akhlaghi4). Higher protein to non-protein ratios (P:NP) have been negatively associated with intakes of ultra-processed foods in adults and children(Reference Martínez Steele, Popkin and Swinburn5). Higher P:NP have also been positively associated with intakes of dairy and alternatives, meat and alternatives and nutrients found in those food groups including Ca and Zn and negatively associated with intakes of energy dense, nutrient poor discretionary foods in pregnant women(Reference Blumfield, Hure and MacDonald-Wicks6). However, the associations between P:NP and intakes of other dietary components in older adults who are susceptible to nutritional inadequacy have not been explored(Reference Ahmed and Haboubi7).
Nutritional inadequacy is a leading modifiable risk factor for morbidity, disability and mortality(Reference Ahmed and Haboubi7–Reference Schwingshackl, Knüppel and Michels9). Nutritional adequacy is defined by nutritional intake that meets the dietary requirements for the prevention of deficiency and chronic diseases(Reference Castro-Quezada, Román-Viñas and Serra-Majem10). Therefore, nutritional inadequacy encompasses dietary intake below, exceeding and/or outside of recommendations. Food groups, food subgroups and nutrients of concern with ageing include those at risk of deficiency, associated with reduced risks of chronic disease and associated with increased risks of chronic disease(Reference Ahmed and Haboubi7,11–Reference Bruins, Van Dael and Eggersdorfer13) . Thus, it is important to investigate whether dietary P:NP intakes are associated with intakes of other dietary components including food groups, food subgroups and nutrients in older adults, particularly in the transitional period of early old age between 65 and 75 years, where medical, physiological and social changes may influence diet(Reference Ahmed and Haboubi7).
The current study aims to explore the cross-sectional associations between dietary P:NP and intakes of other dietary components and assess the nutritional adequacy of the habitual diet (i.e. pre-intervention) of individuals aged 65–75 years from The Nutrition for Healthy Living (NHL) study.
Methods
Study design
The NHL study completed between April 2017 and June 2018 was designed to investigate the effects of randomised meal delivery diet interventions on health outcomes in generally healthy individuals aged 65–75 years living in Sydney, Australia(Reference Ribeiro, Simpson and Couteur14). The NHL study protocol has been described in detail elsewhere(Reference Ribeiro, Simpson and Couteur14). The NHL study is a single-blinded parallel randomised meal delivery diet intervention that involved 113 participants. Volunteers were recruited through paper media, electronic media, radio, television and social media advertisements. Participants were eligible to attend a screening interview if they were aged 65–75 years, both male or female (post-menopausal) volunteers, with a BMI between 20 and 35 kg/m2, individuals who were generally well and not taking medications known to affect weight or energy expenditure. Individuals diagnosed with type 1 diabetes mellitus, insulin-dependent diabetes mellitus, renal disease, liver disease, cancer or active neoplasms and hyperthyroidism (unless treated or under control), who experienced unintentional weight loss of > 10 % body weight over the past 5 years, who were current smokers, alcohol consumers of > 3 standard drinks per day, vegetarians, had food allergies and/or intolerances and had contraindications by treating doctor to changes in diet were excluded. Data were collected between April 2017 and June 2018. The NHL study received ethics approval from Sydney Local Health District Ethics Review Committee (Royal Prince Alfred Hospital Zone). The trial was registered with the Australian and New Zealand Clinical Trials Registry (ACTRN12616001606471) on 21 November 2016.
Dietary data
Of the 113 participants, 107 completed a pre-intervention baseline dietary assessment involving a seven-day weighed food record whom were included in the present study for cross-sectional analyses. Participants were asked to include as much detail as possible such as brand, preparation technique, leftovers, recipes and foods consumed outside of the home. An electronic scale (Ikea ORDNING scale) was provided in addition to photographic and written instructions. The recording procedure was demonstrated by the study coordinator.
Habitual nutritional intakes from baseline food records were converted to nutrient, food group and food subgroup intakes using FoodWorks 8 Professional for Windows (Xyris Software (Australia) Pty Ltd) with the Australian Food, Supplement and Nutrient Database 2013 (AUSNUT 2013)(15). Dietary P:NP was calculated on the basis of protein as a percentage of energy (%E), carbohydrate %E and fat %E summed to 100 % of energy intake. Nutrient intakes were compared with the nutrient reference values for Australia and New Zealand(16) according to age and sex. Estimated average requirement and adequate intake described amounts of nutrients required for adequate physiological function and prevention of deficiency diseases, whilst the suggested dietary target and acceptable macronutrient distribution range are related to chronic disease prevention(16). Consumption of alcohol was compared with the no more than 10 standard drinks a week recommendation of the Australian Guidelines to Reduce Health Risks from Drinking Alcohol(17). As free sugar (added sugar plus sugars in honey and fruit juice)(18) was not available in FoodWorks 8, added sugar intake was compared with the recommendation of the WHO guidelines(19).
Food group and food subgroup intakes were compared with the Australian Guide to Healthy Eating according to the age and sex(20). Consumption of wholegrain as a percentage (%) of total grain, red meat and seafood intake were compared with the recommendations of the Australian Dietary Guidelines(12). ‘Seafood’ and ‘fish’ are used interchangeably according to the Joint Food and Agriculture Organisation/WHO Expert Consultation on the Risks and Benefits of Fish Consumption(21). Due to the different classification of sugar, solid fat and alcohol as food groups in FoodWorks, a serve of discretionary food was defined as 4·2 g sugar (1 teaspoon), 4·8 g solid fat equivalents (1 teaspoon) or 10 g alcohol (1 standard drink)(22–Reference Ribeiro, Hirani and Senior24).
Ancillary data
Ancillary data were collected through self-reported questionnaires and physical assessment(Reference Ribeiro, Simpson and Couteur14). Weight was measured to the nearest 0·01 kg using an electronic scale and height was measured using a wall mounted stadiometer, which were used to calculate the BMI.
Physical activity was assessed through the use of the Physical Activity Scale for the Elderly(Reference Washburn, Smith and Jette25). ‘High’ and ‘low’ education levels were defined as with and without post-school qualifications, respectively. ‘Age Pension only’ referred to those who only received the Age Pension, whilst ‘other’ referred to those with other sources of income apart from the Age Pension including veteran pension, repatriation pension, superannuation, private income, business ownership, farm ownership, business partnership, wage, salary and/or other. ‘Married’ referred to those who were married, living with partner or in a de facto relationship, whilst ‘not married’ referred to those who never married, were divorced, separated, widowed or other. ‘Outright owner’ indicated the individual owned their own home, whilst ‘other’ meant they were paying off their home, paying rent to government for public housing, paying rent or board to a private landlord, or leasing and purchasing or other financial plan in a retirement village. ‘Australia/New Zealand’ indicated they were born in Australia or New Zealand and ‘other’ meant they were born in a country other than Australia or New Zealand. ‘Ex-smokers’ smoked > 100 cigarettes (5 packets) in their entire life, whilst those who ‘never smoked’ smoked < 100 cigarettes in their entire life. Self-rated health categories included ‘excellent’, ‘good’, ‘fair’, ‘poor’ and ‘very poor’(Reference Ware, Kosinski and Keller26). Living location was categorised as ‘metropolitan’ or ‘rural’ based on the reported postcode of residence(27). For meal preparation, ‘yes’ indicated participants can plan or cook full meals for themselves without help, and ‘no’ indicated participants can prepare some things but were unable to cook full meals for themselves and required some help.
Right-angled mixture triangle plot
The Nutritional Geometry Framework is an approach adapted from ecological studies to human nutrition that allows for the examination of the interacting effects of dietary components as opposed to single nutrient analysis(Reference Raubenheimer28,Reference Raubenheimer and Simpson29) . In proportion-based nutritional geometry, three component mixtures such as protein, carbohydrate and fat can be represented in a two-dimensional model(Reference Raubenheimer28,Reference Raubenheimer and Simpson29) . The right-angled mixture triangle has been used to plot the macronutrient composition of foods, meals, various animal and human diets(Reference Raubenheimer28,Reference Raubenheimer and Simpson29) . The right-angled mixture triangle was plotted, on the basis of protein as a percentage of energy (%E), carbohydrate %E and fat %E summed to 100 % of energy intake.
Statistical analysis
Normality tests (histogram, Q-Q plot and Shapiro–Wilk test) found that most data had a skewed distribution. Participant characteristics between those who reported and did not report habitual dietary data and across dietary P:NP tertiles were compared through χ 2, Fisher’s exact and median tests with Bonferroni correction for multiple tests where applicable.
Median (interquartile range) intakes of energy, nutrients, food groups and food subgroups were reported. Proportions of nutritional inadequacy for each nutrient, food group or food subgroup were determined by the number of participants who did not meet the recommendation divided by the total number of participants.
The associations between dietary P:NP and intakes of dietary components were evaluated through linear regression. Dietary P:NP was analysed as both continuous and categorical variables (i.e. categorised into tertiles with the lowest as the reference). Results are presented as β coefficients with 95 % CI. Models were adjusted for the following sociodemographic, lifestyle and health covariates: age, sex, BMI, country of birth, marital status, education level, source of income, housing arrangement, living location, smoking status, PASE, ability to prepare own meals and self-rated health. Collinearity diagnostics was conducted and there was no collinearity in the models with variance inflation factors < 2·5. P < 0·05 was considered statistically significant.
Data was analysed using Statistical Packages for Social Sciences (SPSS) version 25.0 software(30). Graphics were performed using R software(31).
Misreporting sensitivity analysis
Misreporting of energy intake was identified using the Goldberg cut-off method for seven-day dietary data(Reference Black32). The BMR for each individual was calculated using the Schofield equation based on sex, age and weight at baseline(Reference Schofield33). The appropriate values for coefficients of variation and physical activity level (PAL), according to age and sex, were used to calculate the energy intake to BMR ratio (EI:BMR) cut-offs(Reference Black32). Each individual’s EI:BMR was compared with the EI:BMR 95 % CI cut-offs for plausible reporters: 1·07–2·45 for females aged 65–74 years (PAL 1·62), 1·01–2·17 for females aged ≥ 75 years (PAL 1·48), 1·03–2·51 for males aged 65–74 years (PAL 1·61) and 1·02–2·34 for males aged ≥ 75 years (PAL 1·54). Non-plausible reporters had EI:BMR below and above the cut-offs. To assess bias introduced by misreporting in dietary assessment, sensitivity analyses including and excluding non-plausible reporters were conducted.
Results on associations of dietary P:NP as tertiles and continuous variables with intakes of other dietary components were different when non-plausible reporters were excluded. As such, all data are presented with non-plausible reporters excluded. Participant characteristics between those with plausible and non-plausible dietary data were compared through χ 2, Fisher’s exact and median tests.
Results
Of the 113 participants, 88 were included in the analyses (see online Supplementary Fig. S1, Online Resource). There were no differences in participant characteristics between individuals who reported dietary data (n 107) and those who did not report dietary data (n 6) (P > 0·05) (see online Supplementary Table S2, Online Resource). There were no differences in participant characteristics between individuals with plausible (n 88) and non-plausible dietary data (n 19) (P > 0·05), except that non-plausible reporters who all underreported their dietary intake had a higher BMI (P = 0·039) (see online Supplementary Table S3, Online Resource). Participant characteristics of individuals with plausible dietary data are reported in Table 1. Participants (59 female and 29 male) had a median (interquartile range) age of 69 years (67–71), BMI of 27·3 kg/m2 (25·0–29·5), high education level (86 %), other sources of income apart from the Age Pension (81 %), were outright owners of their home (77 %) and rated their self-reported health as excellent or good (82 %). There were no differences in participant characteristics between the dietary P:NP tertiles (P > 0·05).
Table 1 Participant characteristics (percentages and number of participants; median and interquartile range) across dietary protein to non-protein ratio tertiles
P:NP, protein to non-protein; PASE, Physical Activity Scale for the Elderly.
* P values were obtained using the median tests with Bonferroni correction for multiple tests, χ 2 and Fisher’s exact tests to compare all tertile groups for differences in participant characteristics.
Nutritional adequacy
Table 2 reports the intake of energy, nutrients, food groups and food subgroups and proportions of nutritional inadequacy of all participants and by tertiles of the dietary P:NP. The median intake of protein %E, carbohydrate %E and total fat %E for all participants was 17·9 %, 39·6 % and 35·9 % respectively. A small proportion of participants (19 %) consumed protein %E intakes outside of the recommendations, whereas the majority (85 %) had intakes below the recommendations for carbohydrate %E and more than half (55 %) exceeded the recommendations for total fat %E. Figure 1 plots the macronutrient composition of the habitual diet from the three components of all participants in a right-angled mixture triangle and the integrated region of the acceptable macronutrient distribution range recommendations. The range along the protein %E axis was less than along the axes for carbohydrate %E and fat %E intakes, indicating a narrower range of intakes for protein %E in comparison with carbohydrate %E and fat %E.
Table 2 Median daily intakes of energy, nutrients, food groups and food subgroups, recommended intake according to age and sex and the proportions of nutritional inadequacy of all participants and across dietary protein to non-protein ratio tertiles
P:NP, protein to non-protein; IQR, interquartile range; kJ, kilojoule; BW, body weight; EAR, estimated average requirement; F, female; yo, years old; M, male; %E, as a percentage of energy; AMDR, acceptable macronutrient distribution range; AI, adequate intake; std, standard; SDT, suggested dietary target.
* (1 discretionary food serve = 4·8 g solid fat equivalents, 4·2 g added sugars or 10 g alcohol).
Fig. 1 Right-angled mixture triangle (RMT) plot on baseline dietary macronutrient composition data of individuals (n 88) and the integrated region of the acceptable macronutrient distribution range (AMDR) from the nutrient reference values for Australian and New Zealand for fat (20–35 %), protein (15–25 %) and carbohydrate (45–65 %) is presented. Fat as a percentage of energy (%E) and protein %E increase along their respective axes and carbohydrate %E decreases with distance from the origin
A large proportion of participants had intakes below the recommendations for vegetables (82 %), fruits (77 %), dairy and alternatives (89 %), wholegrain % (89 %), meat and alternatives (44 %), seafood (50 %) and nutrients found in those food groups including dietary fibre (52 %), vitamin B6 (46 %), Ca (71 %) and K (51 %), while on the contrary exceeded the recommendations for discretionary foods (100 %) and related nutrients including saturated fat %E (92 %) and Na (55 %). Only a small proportion had intakes below the recommendations for protein per kg body weight (1 %), folate (1 %) and vitamin B12 (2 %). No participants had intakes below recommendations for Fe.
Associations between dietary P:NP and intakes of dietary components
Table 3 presents the adjusted analyses on the associations between dietary P:NP and intakes of energy, nutrients, food groups and food subgroups (for unadjusted analyses see Supplementary Table S4, Online Resource). As a continuous variable, dietary P:NP (per 1 % increment) was associated with lower intakes of energy (β –80·95 (95 % CI: –154·42, –7·48, P = 0·031)), discretionary foods (β –0·89 (95 % CI: –1·17, –0·61, P < 0·001)) and related nutrients including carbohydrate (β –3·86 (95 % CI: –5·90, –1·81, P < 0·001)), carbohydrate %E (β –0·42 (95 % CI: –0·72, –0·12, P = 0·007)), total fat (β –1·29 (95 % CI: –2·32, –0·26, P = 0·015)), saturated fat (β –0·76 (95 % CI: –1·18, –0·34, P < 0·001)), saturated fat %E (β –0·21 (95 % CI: –0·34, –0·08, P = 0·002)), free sugar (β –2·64 (95 % CI: –3·55, –1·73, P < 0·001)) and free sugar %E (β –0·50 [95 % CI: –0·69, –0·31, P < 0·001)) in adjusted analyses. Dietary P:NP (per 1 % increment) was also associated with higher intakes of meat and alternatives (β 0·09 (95 % CI: 0·05, 0·12, P < 0·001)), red meat (β 0·05 (95 % CI: 0·03, 0·07, P < 0·001), poultry (β 0·03 (95 % CI: 0·01, 0·05, P = 0·004)) and wholegrain % (β 1·17 (95 % CI: 0·10, 2·23, P = 0·032)) and nutrients found in those food groups including protein (β 1·97 (95 % CI: 1·19, 2·75, P < 0·001)), protein %E (β 0·60 (95 % CI: 0·55, 0·64, P < 0·001)), vitamin B12 (β 0·11 (95 % CI: 0·02, 0·19, P = 0·013)) and Zn (β 0·16 (95 % CI: 0·02, 0·31, P = 0·023)) in adjusted analyses.
Table 3 Associations between dietary protein to non-protein ratios with intakes of energy, nutrients, food groups and food subgroups in adjusted analyses, using linear regression presented as beta coefficients
P:NP, protein to non-protein; kJ, kilojoule; BW, body weight; %E, as a percentage of energy.
Adjusted by socio-demographic, lifestyle factors and health (age, sex, BMI, country of birth, marital status, education level, source of income, housing arrangement, living location, smoking status, PASE, ability to prepare own meals and self-rated health) (n 71).
* (1 discretionary food serve = 4·8 g solid fat equivalents, 4·2 g added sugars or 10 g alcohol).
Compared with the low dietary P:NP tertile, both the medium and high tertiles were associated with lower intakes of discretionary foods (medium tertile: β –6·38 (95 % CI: –9·57, –3·19, P < 0·001) and high tertile β –11·44 (95 % CI: –15·08, –7·79, P < 0·001)) and related nutrients including carbohydrate (medium tertile: β –29·14 (95 % CI: –52·17, –6·12, P = 0·014) and high tertile β –53·79 (95 % CI: –80·10, –27·49, P < 0·001)), free sugar (medium tertile: β –23·19 (95 % CI: –32·71, –13·67, P < 0·001) and high tertile β –38·09 (95 % CI: –48·97, –27·21, P < 0·001)) and free sugar %E (medium tertile: β –4·05 (95 % CI: –6·04, –2·05, P < 0·001) and high tertile β –7·38 (95 % CI: –9·66, –5·11, P < 0·001)) in adjusted analyses. Both the medium and high dietary P:NP tertiles were also associated with higher intakes of red meat (medium tertile: β 0·23 (95 % CI: 0·02, 0·43, P = 0·031) and high tertile β 0·64 (95 % CI: 0·41, 0·87, P < 0·001)) in adjusted analyses.
Compared with the low dietary P:NP tertile, the medium tertile was associated with lower intakes of vegetables (β –1·49 (95 % CI: –2·88, –0·09, P = 0·037)) and other vegetables (β –1·03 (95 % CI: –1·95, –0·10, P = 0·031)), total fat (β –13·80 (95 % CI: –25·75, –1·86, P = 0·024)), linoleic acid (β –2·48 (95 % CI: –4·47, –0·49, P = 0·016)) and vitamin E (β –3·42 (95 % CI: –6·74, –0·09, P = 0·044)) in adjusted analyses.
Compared with the low dietary P:NP tertile, the high tertile was associated with lower intakes of fruit juice % (β –9·82 (95 % CI: –18·91, –0·74, P = 0·035)), carbohydrate %E (β –7·03 (95 % CI: –10·77, –3·30, P < 0·001)), saturated fat (β –7·28 (95 % CI: –13·10, –1·47, P = 0·015)) and saturated fat %E (β –1·93 (95 % CI: –3·72, –0·13, P = 0·036)) and higher intakes of meat and alternatives (β 1·06 (95 % CI: 0·61, 1·52, P < 0·001)), milk alternatives (β 0·16 (95 % CI: 0·03, 0·30, P = 0·020)) and related nutrients including vitamin B12 (β 1·29 (95 % CI: 0·17, 2·42, P = 0·025)) and Zn (β 2·07 (95 % CI: 0·21, 3·94, P = 0·030)) in adjusted analyses.
Discussion
To our knowledge, this is the first study to examine the associations between dietary P:NP and intakes of other dietary components in older adults. Higher dietary P:NP was associated with lower intakes of energy, nutrients and a food group known to increase risk of chronic disease (saturated fat, saturated fat %E, free sugar, free sugar %E and discretionary foods) and higher intake of a dietary component known to reduce risk of chronic disease (wholegrain %)(11–Reference Bruins, Van Dael and Eggersdorfer13). Unsurprisingly, higher dietary P:NP was associated with a higher consumption of nutrients found in protein-rich foods and a protein-rich food group, which are also known to be at risk of deficiency in ageing (vitamin B12, Zn and meat and alternatives), and a protein-rich food subgroup associated with increased risks of chronic disease (red meat)(Reference Ahmed and Haboubi7,11,12) . Majority of participants consumed habitual diets that were nutritionally inadequate: many had intakes below recommendations for vegetables, fruits, dairy and alternatives, wholegrain %, meat and alternatives, seafood, dietary fibre, vitamin B6, Ca and K, but exceeded the recommendations for discretionary foods, saturated fat %E and Na. Compared with carbohydrate and fat, most participants had protein intakes within recommendations. We found that almost all participants consumed absolute protein intakes within recommendations. Furthermore, we observed that there was less variation in protein %E than carbohydrate %E and fat %E intakes, which is consistent with previous research demonstrating Protein Leverage(Reference Raubenheimer and Simpson1,Reference Lieberman, Fulgoni and Agarwal34) .
Nutritional adequacy
The substantial proportions of inadequate nutritional intakes in this cohort were surprising yet consistent with studies involving nutrition intervention participants interested in nutrition(Reference Cano-Ibáñez, Gea and Martínez-González35,Reference Livingstone, Celis-Morales and Navas-Carretero36) . Majority had intakes below recommendations for vegetables, fruits, dairy and alternatives, wholegrain %, meat and alternatives and seafood, and thus also had intakes below recommendations for nutrients found in these food groups including dietary fibre, vitamin B6, Ca and K. These food groups and nutrients are important for the prevention of chronic diseases such as CVD, diabetes and cancer(12,Reference Bruins, Van Dael and Eggersdorfer13) . Simultaneously, most exceeded the recommendations for discretionary foods, and thus also exceeded recommendations for saturated fat %E and Na, that are associated with increased risks of chronic disease(11,12) .
On the bright side, previous research has found the prevalence of nutritional inadequacies to be lower amongst older adults participating in nutrition intervention studies than other groups including those of the highest social class(Reference Jyväkorpi, Pitkälä and Puranen37). The proportions of nutritional inadequacy in the current study were lower for thiamin, riboflavin, vitamin C, vitamin E, folate, vitamin B12, iodine and Fe than those found in a systematic review and meta-analysis on micronutrient intakes in older adults aged 65+ years from thirty-seven observational studies, not based on nutrition intervention(Reference Ter Borg, Verlaan and Hemsworth38). Participants’ intakes of most food groups and food subgroups were also better matched to recommendations than the general population: the proportions below recommendations were lower for vegetables, dairy and alternatives, meat and alternatives and grains, whilst red meat intake was also lower and did not exceed recommendations compared with adults aged 51+ years in the Australian National Nutrition and Physical Activity Survey 2011–2012(39). However, the proportion below recommendations for fruits was higher, and there were lower intakes of fruits, seafood and wholegrain % and higher discretionary intake in the current study(40,41) . The proportions below recommendations were also lower for vegetables and higher for fruits compared with adults aged 65–74 years in the Australian National Health Survey 2017–2018(41).
Although majority had intakes below the recommendations for carbohydrate %E and more than half exceeded the recommendations for total fat %E, most consumed protein %E intake within recommendations. Thus, the median intakes of protein %E were similar, and carbohydrate %E was lower, whilst total fat %E including saturated fat %E was higher than adults aged 51+ years in the Australian National Nutrition and Physical Activity Survey 2011–2012(40). The macronutrient distribution consumed in the current study was also comparable to adults aged 51+ years in the Australian National Nutrition and Physical Activity Survey 2011–2012 in the ratio of one part protein to four parts non-protein (fat and carbohydrate), or one part protein to two parts carbohydrate(40). Research using model life tables and national macronutrient supplies for a subset of developed countries found that overnutrition from increasing total energy, carbohydrate %E and fat %E supplies increased mortality(Reference Senior, Nakagawa and Raubenheimer42). Furthermore, emerging research showed that macronutrient requirements may change across the lifespan and diets with equal parts protein to carbohydrate(Reference Senior, Solon-Biet and Cogger43) and moderate to high protein %E intakes(Reference Levine Morgan, Suarez Jorge and Brandhorst44) reduced mortality in later life.
It is notable that we observed such high rates of nutritional inadequacy even though participants in the NHL study were of high socio-economic background and rated their health as excellent/good. There are a number of factors influencing nutritional intakes that go beyond self-perceived health and income. For instance, factors such as social support, taste perception, health beliefs, willingness to change, nutrition knowledge, awareness of dietary guidelines and health systems incorporating nutrition care can all affect dietary intake(Reference Poggiogalle, Kiesswetter and Romano45,Reference Shlisky, Bloom and Beaudreault46) . Nutrition education programs can motivate older adults to improve nutritional intakes through changing nutrition knowledge, attitude and behaviours(Reference Shlisky, Bloom and Beaudreault46), as it was found that nutrition awareness and use of nutrition information were associated with adherence to dietary guidelines(Reference Vaudin, Wambogo and Moshfegh47,Reference Schwartz and Vernarelli48) . Future research should investigate strategies to address other determinants of nutritional intake in older adults.
Associations between dietary P:NP and intakes of dietary components
Dietary P:NP was inversely associated with intakes of energy, saturated fat, saturated fat %E, free sugar, free sugar %E and discretionary foods. Similar to the present study, it has been observed in the American population in analyses of National Health and Nutrition Examination Survey data that protein %E intake was inversely associated with ultra-processed food %E intake, whilst carbohydrate %E, added sugar %E, saturated fat %E and total energy intakes were positively associated with ultra-processed food %E intakes(Reference Martínez Steele, Raubenheimer and Simpson2,Reference Martínez Steele, Popkin and Swinburn5) . The Protein Leverage Hypothesis states that absolute protein intakes are regulated more strongly than fat and carbohydrate, resulting in excess energy consumption and predisposing to overweight and obesity on protein-dilute diets(Reference Raubenheimer and Simpson1). Hence, the protein diluting effects of discretionary foods could explain the inverse associations between dietary P:NP and energy intake.
Dietary P:NP was positively associated with intakes of vitamin B12, Zn, meat and alternatives, red meat, poultry and wholegrain %. The protein source, such as from plant or animal foods, could have different impacts on the associations of dietary P:NP with intakes of other dietary components. Future research should consider whether the associations between dietary P:NP and intakes of other dietary components are modified on semi-vegetarian and plant-based diets.
There are some study limitations. First, causation could not be assessed due to the cross-sectional design. Second, although the sample size was powered for the NHL study, the sample size was small for the current study. Third, Na and free sugar intake was likely underestimated; Na added in home prepared foods or at the table(Reference Nowson, Lim and Grimes49) and sugars from honey and fruit juice(18) were not accounted for, respectively. Fourth, the number serves of food groups and food subgroups were obtained using FoodWorks 8, which is based on the Australian Guide to Healthy Eating, but there are discrepancies on serve size equivalents, and we used an adapted definition for a serve of discretionary. Additionally, dietary P:NP was used to demonstrate the impact Protein Leverage could have on the intakes of other dietary components, and thus did not differentiate between the different types of carbohydrate and fat such as refined or unrefined carbohydrates and saturated or unsaturated fats. Lastly, participants were volunteers for a nutrition intervention study and majority were female, reflecting a highly motivated group and an imbalance in male representation. There were also strengths. Dietary assessment was conducted using a prospective seven-day weighed food record which is often used as a reference method. Sensitivity analyses were conducted and non-plausible reporters were excluded to better reflect true intakes. Furthermore, we accounted for confounders including socio-demographic, lifestyle and health factors in analyses between dietary P:NP and intakes of other dietary components.
In conclusion, we showed that diets with higher dietary P:NP were associated with lower intakes of energy, saturated fat, free sugar and discretionary foods and higher intakes of vitamin B12, Zn, meat and alternatives, red meat, poultry and wholegrain %. Vice versa for diets with lower dietary P:NP, indicating the potential protein diluting or protein-enriching effects of these dietary components. The current study also demonstrated that the majority of individuals aged 65–75 years from a high socio-economic background had inadequate nutritional intakes.
Acknowledgements
The authors would like to thank all participants for their time and support. The authors acknowledge the technical assistance of Dr. Gordon McDonald and Dr. Kathrin Schemann of the Sydney Informatics Hub, a Core Research Facility of the University of Sydney.
Financial Support
The Nutrition for Healthy Living study was funded by a philanthropic donation from Emeritus Professor George Palmer. RL was supported by an Australian Government Research Training Program Scholarship and an Australian Research Council Centre of Excellence in Population Aging Research Scholarship. RVR was supported by a Charles Perkins Centre Early Career Fellowship from Jennie Mackenzie. The funders were not involved in the conception, design, performance or approval of this work.
Authorship
The authors’ responsibilities were as follows: R.L., RV.R., V.H., S.J.S. and D.R.: conceptualised the analyses presented in this manuscript; R.L., R.V.R., V.H., S.J.S. and D.R.: designed the analyses; R.L.: conducted the analyses; R.L., S.J.S. and D.R.: interpreted the data; R.L.: wrote the manuscript; R.L., R.V.R. and V.H.: had primary responsibility for the final content; and all authors: were involved in editing, reading and approved the final manuscript.
Conflicts of Interest
The authors report no conflicts of interest.
Ethics of human subject participation
The study has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. All participants gave their informed consent prior to their inclusion in the study. The NHL study received ethics approval from Sydney Local Health District Ethics Review Committee (Royal Prince Alfred Hospital Zone).
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/S1368980023001726
Open access
