Nutrition plays a central role in health, with poor dietary habits and nutritional intake being associated with a range of chronic diseases(1). Intake of specific nutrients into the body reflects the diet that a person eats, which involves the food choices made by an individual usually as part of food purchasing behaviour.
Poor dietary habits have been linked to a number of factors, including socio-economic status (SES). SES may influence nutrition directly as a resource limitation associated with price as a barrier to food choice. SES may also influence nutrition through the impact of dietary knowledge on food choice(Reference Turrell and Kavanagh2). Another pathway that SES may influence nutrition through limiting food choice is by the impact of tooth loss on chewing ability. Lower intakes of specific nutrients (e.g. β-carotene, vitamin C, folate and dietary fibre) have been found for the edentulous and people with few natural teeth(Reference Joshipura, Willett and Douglass3). The relationship between dentition status and nutritional intake is important because of the well-known health sequelae of poor dietary habits(Reference Krall, Hayes and Garcia4).
The literature on tooth loss and nutrition has identified a plausible link between chewing disability and restricted diets(Reference Joshipura, Willett and Douglass3, Reference Krall, Hayes and Garcia4), possibly reflecting a general restriction in diet or selective avoidance of some healthy foods on the basis of their being difficult to chew. However, the observed links between tooth loss and SES, and dietary knowledge and SES(Reference Turrell and Kavanagh2), could mean that the association between tooth loss and diet may be confounded. Hence, it is necessary to demonstrate that tooth loss and related chewing disability is an independent predictor of diet.
The generally higher levels of tooth loss among the elderly places them at risk of nutrition-related problems compared with younger persons. These diet-related health problems may be further exacerbated by lower SES, and may be influenced by variation in dietary knowledge. The aim of the present study was to examine grocery purchasing behaviour in relation to chewing ability, dietary knowledge and SES among elderly persons.
Methods
A random sample of adults aged 60–70 years living in Adelaide, South Australia, was drawn from the Electoral Roll. Data were collected by mailed self-completed questionnaires in 2008, with multiple follow-up mailings to non-respondents.
Outcome variable
Grocery purchasing was assessed by sixteen grocery items (see Table 1) using multiple responses for each item in relation to what type of grocery item they usually buy, based on previous reports using the Grocery Purchasing Index(Reference Turrell and Kavanagh2). For example, the food-type item of ‘bread’ had response options such as ‘white’, ‘wholemeal’, ‘multigrain’, etc. Participants were advised to report on themselves (i.e. ‘your usual food shopping’) or ‘the person who shops for you’ to cover such cases as a person shopping on behalf of their spouse. The responses to the grocery items were classified into ‘recommended’ and ‘regular’ categories, with items in the recommended category being those suggested by health authorities as preferable to minimise the risk of diet-related disease. These were based on national guidelines produced by the Australian National Health and Medical Research Council(5). For example, a response of ‘white’ to the ‘bread’ item would be classed as ‘regular’, while a response such as ‘wholemeal’ would be classed as ‘recommended’. Responses to each food-type item were coded as 0 if never purchased, 1 if the regular option was purchased exclusively, 2 if both the regular and recommended options were purchased and 3 if the recommended option was purchased exclusively. These coded responses were summed to form the purchasing index. Following the reported method for the index(Reference Turrell and Kavanagh2), the index scores were adjusted for the number of food-type items usually purchased and re-scored on a range from 0 to 100, with high scores indicating greater compliance with dietary guideline recommendations.
Median responses to grocery purchasing items and responses grouped into categories of purchasing
†Recommended included(2): bread (wholemeal, multi-grain, white high in fibre, rye, soy and linseed), rice (wholemeal or brown), pasta (wholemeal or brown), baked beans (salt-reduced or unsalted), fruit juice (unsweetened), tinned fruit (in natural juice), milk (reduced-fat, low-fat, high-calcium, high-iron, high-protein, reduced-lactose, no cholesterol, soy or soy and linseed), cheese (reduced-fat, low-salt), yoghurt (low-fat), beef mince (lean), chicken (without skin), tinned fish (in spring water), vegetable oil (salt-reduced, fat-reduced), butter (salt-reduced, unsalted), solid cooking fat (margarine, solidified oil).
Explanatory variables
Self-reported number of teeth was classified as inadequate dentition if less than twenty-one teeth were present, consistent with the case definition used elsewhere, such as in the UK adult dental health survey(Reference Kelly, Steele and Nuttall6) and the National Survey of Adult Oral Health, Australia 2004–2006(Reference Slade, Spencer and Roberts-Thomson7). Chewing ability was assessed using five items based on previous reports of chewing index scores (see Table 2)(Reference Leake8). Responses of ‘yes’ to each chewing item were coded as 1 and summed to produce a chewing index score. Scores of 0–4 were classed as ‘chewing deficient’, whereas a score of 5 was classed as ‘chewing competence’(Reference Leake8).
Percentage of correct responses to dietary knowledge items
Dietary knowledge was collected using twenty true/false items (see Table 3)(Reference Turrell and Kavanagh2). Correct answers were coded as 1, and summed to produce a dietary knowledge score where higher scores indicated better dietary knowledge. Scores were dichotomised into high dietary knowledge if fifteen or more items were correct.
Distributions of explanatory variables and associations of explanatory variables with chewing ability
**P < 0·01, χ 2 test.
SES was assessed using the measure of subjective social status whereby persons rate themselves on a ladder ranging from 0 to 10 representing where they stand in society with higher scores representing those best off in terms of education, money and jobs(Reference Adler, Epel and Castellazzo9). This measure was dichotomised with scores less than 5 coded as lower SES.
In addition, annual income (up to $AU20 000 and >$AU20 001), age (60–64 and 65–70 years) and sex (male and female) were included as explanatory variables to control for their potential effects.
Analysis
The analysis was restricted to dentate persons (with some of their own natural teeth). Associations of chewing ability with dentition status, dietary knowledge, SES, income, age and sex were tested using the χ 2 statistics. Differences in mean grocery purchasing scores were tested using Mann–Whitney U tests(10). The multivariate model was fitted using linear regression for all main effects and two-way interactions were tested, but the interactions were only retained if statistically significant at the P < 0·05 level.
The research was approved by the Human Research Ethics Committee of the University of Adelaide.
Results
Response
Responses were collected from 444 persons (response rate = 68·8 %). The majority of the respondents were dentate (88·6 %). Among the dentate respondents, 48 % were men and 67 % were born in Australia.
Distributions
For the grocery purchasing items (Table 1), generally a small percentage of persons reported purchasing the regular option only, except for rice (67·2 %) and pasta (73·3 %). In addition, small percentages reported that they ‘do not buy’ each food-type item except for cooking fat (83·0 %). The grocery purchasing index score ranged from 33 to 100, with a median of 78 and a mean of 76 (sd 13).
For the chewing index, a high percentage of persons reported being able to chew the items in the index, ranging between 91·1 % for ‘bite off and chew whole fresh apple’, 96·1 % for ‘chew fresh carrot’, 96·3 % for ‘chew firm foods such as steak or dried apricots’ and 98·7 % for ‘chew fresh lettuce salad’ and ‘chew boiled vegetables’. The summed chewing index score ranged from 0 to 5, with a median of 5·00 and a mean of 4·82 (sd 0·67). When classified by chewing ability, 10·3 % were defined as chewing deficient.
The majority of persons gave correct answers to the dietary knowledge items, ranging between 81·9 % and 98·9 % (Table 2). The summed dietary knowledge score ranged from 9 to 20, with a median of 19·0 and a mean of 18·3 (sd 1·9). When dichotomised into dietary knowledge categories, 90·8 % were classified as having high dietary knowledge.
The responses to the subjective social status item ranged between 0 and 10, with a median of 6·0 and a mean score of 5·9 (sd 1·9). When classified into SES groups, 21·3 % of persons were in the lower SES group.
Associations
Chewing ability was related to dentition status and SES, but not to dietary knowledge (Table 3). Chewing deficiency was more prevalent among persons with inadequate dentition (26·4 %) compared with those with adequate dentition (4·9 %), among the lower SES group (18·2 %) compared with the higher SES group (8·6 %) and in the lower income group (20·9 %) compared with the higher income group (7·4 %).
Unadjusted mean grocery purchasing scores were lower for persons with chewing deficiency, lower dietary knowledge, lower SES, lower income and for men (Table 4). Multivariate regression coefficients showing chewing deficiency (−5·0) and low SES (−4·0) were both associated (P < 0·05) with lower grocery purchasing scores, as were men (−3·7) compared with women. However, dietary knowledge, income and age were not statistically significant. No interactions were statistically significant.
Grocery purchasing scores: unadjusted means by chewing ability, dietary knowledge, social status, income, age and sex and coefficients from multivariate regression analysis
Ref., reference category.
*P < 0·05; **P < 0·01.
Discussion
The findings show that grocery purchasing behaviour among older adults in Australia varied by chewing ability and SES, but not by the level of dietary knowledge. This indicates that persons with chewing deficiencies and lower social status are less likely to comply with the recommended dietary guidelines. The link between food choice and dietary quality means that attention is needed to understand factors that influence food purchasing behaviour(Reference Turrell, Hewitt and Patterson11). The findings of the present study are important because of their relevance to the issues of tooth loss, chewing and diet in the elderly. The accumulation of tooth loss over the life course places older adults at potential risk of nutrition-related problems due to their generally higher levels of tooth loss compared with younger persons(Reference Walls, Steele and Sheiham12). There is also evidence of socio-economic disparities in dietary patterns and nutrition(Reference Giskes, Turrell and Patterson13, Reference Giskes, Turrell and Patterson14). The impact of dietary knowledge on food choice has been linked as one pathway through which SES may influence nutrition(Reference Turrell and Kavanagh2). However, in this group of older adults, dietary knowledge was not related to food choice.
Chewing ability
People with a compromised dentition have been found to have significantly impaired masticatory function compared with those with an intact dentition(Reference Chauncey, Muench and Kapur15). Among the elderly dentate adults in the present study, while only a minority were classified as having a chewing deficiency, chewing deficiency was associated with inadequate dentition and lower compliance with dietary guidelines. Edentulous and people with few natural teeth have been reported to have lower intakes of specific nutrients (e.g. β-carotene, vitamin C, folate and dietary fibre)(Reference Joshipura, Willett and Douglass3–Reference Krall, Hayes and Garcia4, Reference Nowjack-Raymer and Sheiham16–Reference Lowe, Woodward and Rumley17). The present study shows that the link between tooth loss and chewing ability with diet persists after controlling for the effects of both dietary knowledge and SES.
Dietary knowledge
Previous studies have shown links between dietary knowledge and eating behaviour(Reference Sharma, Gernand and Day18, Reference Shaikh, Yaroch and Nebeling19), with variation in nutrition knowledge being associated with characteristics such as education, age, sex and work status(Reference Hendrie, Coveney and Cox20, Reference De Vriendt, Matthys and Verbeke21). However, dietary knowledge was not associated with food purchasing behaviour in the present study of older adults when adjusted for SES and chewing ability. One consideration was that the levels of dietary knowledge were quite high among the respondents, perhaps indicating the success of health promotion campaigns in improving nutrition knowledge. It should also be considered that knowledge does not necessarily translate into behaviour. However, socio-economic variation in dietary knowledge could provide a link between socio-economic position and food purchasing behaviour(Reference Turrell and Kavanagh2), and diet-related health promotion may be more effective if tailored to the needs of disadvantaged groups(Reference Turrell, Hewitt and Patterson11).
Socio-economic status
Subjective social status was used as a measure of SES for older adults in the present study as the age range included a cross-section where some may still be employed, others may not be working in formal employment, and yet others may be retired from work. Subjective social status has been related to a range of health-related factors such as self-rated health, heart rate and body fat distribution(Reference Adler, Epel and Castellazzo9), and reflects the cognitive averaging of standard markers of socio-economic position rather than psychological biases(Reference Singh-Manoux, Adler and Marmot22). Previous studies have established a link between SES and diet. For example, SES has been related to fruit and vegetable intake among Australian adults(Reference Brennan, Singh and Liu23), while data on US adults found that better SES improved the likelihood of adequate fruit and vegetable intake as well as overall diet quality(Reference Beydoun and Wang24). The present study confirms the link between SES and food purchasing, and further shows the independent effects of both SES and chewing ability.
Strengths and limitations
The sampling is likely to generate a representative sample of the general population of adults in the age range since it was based on the Electoral Roll and voting is compulsory. Comparison with census data for similarly aged Adelaide residents from 2006(25) showed a similar percentage by sex (48 % men in the sample compared to 46 % in the census), but there was a higher percentage of Australian-born persons in the sample (67 %) compared to the census (58 %). It is also possible that despite the adequate response rate, elderly persons in high care facilities may be under-represented, and the findings may be generalised to community-dwelling older adults. While the findings are based on self-reports of usual purchasing behaviour, it is considered important to understand dietary behaviour, which is distinct from the downstream consequences of behaviour (i.e. food and nutrient intake)(Reference Turrell, Hewitt and Patterson11). The cross-sectional nature of the study limits the ability to comment on the observed associations in terms of causal relationships.
Implications and conclusions
The non-significant result for dietary knowledge is consistent with findings that suggest that extended nutrition education would have only limited effects on food intake(Reference Korinth, Schiess and Westenhoefer26). The persistence of chewing ability and SES as predictors of food purchasing does not support the hypothesis of confounding by dietary knowledge. While chewing ability was related to food purchasing behaviour, further research is required to document the types of foods that are restricted and to establish the public health significance of such differences in terms of nutrition and health outcomes. In addition, further research may also shed more light on the extent and pattern of tooth loss associated with restricted dietary intakes. For example, studies have shown that tooth loss may need to be severe in order to influence diet(Reference Savoca, Arcury and Leng27). Another consideration for future research could be the investigation of whether factors such as both chewing disability related to tooth loss and dietary patterns cluster together along with other common health behaviours. There are suggestions that while tooth loss is related to diet, other psycho-social factors may be more influential(Reference Bradbury, Thomason and Jepson28). In summary, the present study showed that among older adults, chewing deficiency and lower social status were associated with lower compliance with dietary guidelines, independent of dietary knowledge.
Acknowledgements
The collection of data was funded by the Australian Dental Research Foundation (ADRF). D.S.B. was supported by a Career Development Award (627037) from NHMRC. The authors have no conflict of interest to declare. The contribution of each author to the manuscript was as follows: D.S.B. and K.A.S. collaborated in the conception and design of the study, and were the chief investigators on the funding application to ADRF. D.S.B. performed the final analyses and drafting of the manuscript. All authors contributed to completing the final version of the manuscript.
