| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
School of Health Sciences (C.E.C.)
Research Centre for Gender and Health (A.F.Y.)
The University of Newcastle, Callaghan, New South Wales, Cancer Epidemiology Centre, Cancer Council of Victoria, Carlton, Victoria (A.H.), AUSTRALIA
Address correspondence to: Clare E. Collins, PhD, BSc, Dip Nutr Diet, Dip Clin Epi, APD, Associate Professor in Nutrition and Dietetics, School of Health Sciences, University of Newcastle, NSW, Australia, HA12 Hunter Building, University Drive, Callaghan, NSW, 2308. AUSTRALIA. E-mail: Clare.Collins{at}newcastle.edu.au
 |
ABSTRACT |
|---|
 TOPFOOTNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Design: Cross-sectional survey in a nationally representative sample of mid-aged women participating in a cohort study.
Subjects: Data from 9,895 women aged 50–55 who participated in the 2001 survey and had four or less missing values on their food frequency questionnaires were used to calculate the Australian Recommended Food Score (ARFS) based on adherence to Australian Dietary Guidelines.
Measure of Outcome: Correlates of ARFS were investigated including, mean nutrient intakes and indices of self-rated health and health service use. Associations were examined using ANOVA for continuous variables and Chi-squared tests for categorical variables. Area of residence and educational attainment were used as covariates in all modeling, to adjust for sampling frame and socioeconomic status.
Results: The maximum ARFS was 74, with a mean of 33.0 ± 9.0 and 21% achieving a score > 40. Higher ARFS was associated with indicators of higher socio-economic status, better self-rated health and lower health service use, p<0.0001, higher intakes of micronutrients and lower percentage of energy as total or saturated fat, p<0.0001.
Conclusions: The Australian Recommended Food Score can be used to rank mid-aged women in terms of diet quality and nutrient intake and is associated with indices of self-rated health and health service use. The ARFS can be used to measure future associations with health outcomes and mortality.
Abbreviations: AHEI = Alternate Healthy Eating Index • ALSWH = Australian longitudinal Study of Women's Health • ARFS = Australian Recommended Food Score • DQES = Dietary Questionnaire for Epidemiological Studies • BMI = body mass index • BMR = basal metabolic rate • EI/BMR = ratio of energy intake to basal metabolic rate • EI = energy intake • ES = energy standardized • FFQ = food frequency questionnaire • RFS = Recommended Food Score
|
INTRODUCTION |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Other researchers have used measures of dietary quality in longitudinal studies of women. Kant developed the Recommended Food Score (RFS) [2, 3] based on frequency of consumption of foods recommended in national dietary guidelines, independent of amounts, from a 62-item food frequency questionnaire (FFQ). The RFS has been shown to be inversely associated with mortality after a median of 9.5 years of follow-up in a prospective cohort of 42000 women participating in the Breast Cancer Detection Demonstration project [4]. McCullough et al developed the Alternate Healthy Eating Index (AHEI) from a 130-item semi-quantitative FFQ [5] as a modification of the Healthy Eating Index developed at the US Department of Agriculture to track the quality of diet in the US over time [6]. Component food groups of the AHEI are weighted to a maximum score of 10 based on achieving intakes consistent with national guidelines in each component. Women in the top quintile of the AHEI had an 11% reduction in risk of major chronic disease compared to the bottom quintile [7]. In 1995, over 40 000 women were recruited to the Australian Longitudinal Study on Women's Health (ALSWH) which was established to investigate multiple factors affecting the health and well being of women, over a 20-year period. The mid-aged cohort were the first group to complete a food frequency questionnaire in 2001 as part of ALSWH. In 2004, a comparison was made of the reported dietary intakes with National Dietary Guidelines [8] but the researchers considered it impossible to quantitatively measure dietary variety in the context of their analysis [9]. We believe it is appropriate to calculate a diet quality score from the FFQ data, with the intention of developing a simplified dietary questionnaire that could be used to replace the FFQ in subsequent follow-ups in the ALSWH to provide a dietary predictor of morbidity and mortality, given recent reports indicating that poor diet quality increases the risk of morbidity and mortality [10].
A diet that follows dietary guidelines should be low in fat and high in fruits and vegetables and the nutrients they contain, with generally high nutrient density. The FFQ data from the mid-age women in 2001 can be used to assess how well a new dietary index predicts diet quality. In the long-term, the value of the diet quality score as a predictor of health outcomes can be determined. However, with the cross-sectional data associations between a diet quality score with self-reported measures of health status can be assessed.
Aim
The aim of this study was to calculate an index of diet variety and quality, reflecting adherence to the Australian Dietary Guidelines, for women in the mid-age cohort of the ALSWH. A second aim was to examine whether better scores using this index also reflected higher macronutrient and micronutrient intakes and whether higher scores are related to indices of better self-rated health.
|
MATERIALS AND METHODS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
The response rate for Survey 3 of the mid-age cohort was 83% of women who had completed Survey 1 and had not died or become too ill to complete further surveys. The non-respondents included those who did not complete Survey 3 (7.4%), withdrew from the study completely (2.8%) or could not be contacted (6.8%) [12]. Of the women who completed Survey 3 (then aged 50–55 years), 11196 completed the FFQ but after excluding those with greater than four missing items, only 88.4% FFQs (n=9895) were considered usable.
Assessment of Dietary Intake
Dietary intake was assessed using an FFQ known as the Dietary Questionnaire for Epidemiological Studies (DQES) Version 2. The questionnaire was developed by the Cancer Council Victoria as an update of an FFQ used in a cohort of Australian-, Greek- and Italian-born volunteers aged 40–69 when the FFQ was completed. The items included in the FFQ were based on those found in a pilot study to make important contributions to nutrient intakes of volunteers from the same ethnic backgrounds as cohort members. Both the development of the questionnaire [13] and its validation in mid-aged Australian women have been previously reported [14]. This questionnaire asks respondents to report their usual consumption of 74 foods and six alcoholic beverages over the preceding 12 months using a 10-point frequency scale. The categories are never, less than once per month, once per week, twice per week, three to four times per week, five to six times per week, once per day, twice per day, 
three times per day. Questions on the total intakes of fruit and vegetables are used to adjust the intakes of individual fruits and vegetables, which tend to be over-estimated. Portion photographs of vegetables, potatoes, meat and casserole dishes are used to calculate a portion factor that is applied to scale up or down the standard portions of foods that showed variation by gender or ethnicity in the weighed food records from which the FFQ was derived. Additional questions are asked about the number of serves or type of fruit, vegetables, bread, dairy products, eggs, fat spreads and sugar and further details are provided in Appendix 1. Nutrient intakes were computed from NUTTAB 1995, the most recent national government food composition database of Australian foods [15], using software developed by the Cancer Council of Victoria.
|
Measures of Socio-demographic and Health-Related Variables
Socio-demographic factors potentially associated with quality of food intake included area of residence (classified as urban or non-urban), highest level of education (up to 12 years of education only or tertiary qualifications), country of birth (Australia or other country), marital status (married, including living as married, or not) and ability to manage on available income (easy or not too bad versus having at least some difficulty). Health behaviors included being a current smoker (yes/no) and having adequate levels of physical activity (yes/no) [18]. The women were asked about frequency and total duration of physical activity in the previous week. For the purpose of categorizing this as "adequate", a threshold for health benefit was defined as moderate activity equivalent to 30 minutes, five times or more per week (150 minutes), which could be achieved by a combination of walking and moderate or vigorous activities [18].
Self-rated health was categorized as excellent/very good/good or fair/poor. This single item has proved to be a parsimonious measure of general health perception and an independent predictor of survival [19]. Women were classified as having depressive symptoms if their score on the CESD-10 was 10 or more [20, 21]. Health service use was assessed by the number of visits to a general medical practitioner in the previous year (categorised as 0–6 visits, 7+ visits) and whether the woman had consulted a specialist medical practitioner in the previous year (yes/no). Women were asked to report their weight either to the nearest kilogram (kg) or in stones and pounds. If reported in imperial measures, a conversion factor of 2.203 was used to convert pounds to kilograms, with one stone equal to 6.35 kilograms. Body Mass Index (BMI) was calculated as weight (kg) divided by height (m) squared. BMI scores were assigned to one of the four categories according to the recommendations of the World Health Organization as underweight (BMI <18.5), acceptable weight (BMI 18.5 to <25), overweight (BMI 25 to <30) or obese (BMI 30) [22].
Analysis
Basal metabolic rate (BMR) for each woman was calculated using the Schofield equations [23]. It is generally recognized that mis-reporting is a widespread problem where dietary data is collected by self-report [24]. To try and identify the group least likely to under- or over-report one can examine the ratio of energy intake (EI) to basal metabolic rate. If a ratio of 1.55 represents a moderate level of physical activity for this group, then FFQs giving an energy intake of 1.27–2.1 times BMR can be considered plausible [24, 25]. A sub-sample of women whose ratio of EI to BMR fell in the range of 1.27 to 2.1 [24] was considered least likely to have mis-reported their dietary intake and the analyses using ARFS were repeated using this sub-sample (referred to as the Valid EI/BMR ratio group).
The socio-demographic characteristics of women in the Valid EI/BMR subset were compared to those where an ARFS could be calculated but the EI/BMR was not valid and women who did not have an ARFS score. Tests of association were conducted using Chi-squared analysis. Descriptive statistics for the contribution of each of the core food component scores to the total ARFS were calculated. ARFS results were divided into quintiles and the association between quintiles and macronutrient and micronutrient intake levels was examined using analysis of variance models. Nutrient intakes were also standardized per 1000 Calories by dividing reported intakes by total daily kilojoules and multiplying by 4184.
Area of residence and educational attainment were used as covariates in all statistical modeling, to adjust for the sampling frame and for socioeconomic status, using ANOVA. Given the large sample size and multiple comparisons undertaken only p-values < 0.001 for statistical significance in models were considered [26].
Data manipulation and statistical analyses were performed used SAS (SAS Institute. SAS/STAT User's guide, Version 8. Cary, NC: SAS Institute Inc.; 1999.)
|
RESULTS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
The mean age (n=9882) was 52.7 ± 1.5 years (range 50–55 years) and 52.7 ± 1.5 years (n=2355) for the EI/BMR sub-sample. Several socio-demographic characteristics were associated with whether the women had an ARFS or not and for those who did, whether they had a valid EI/BMR ratio. Women with both an ARFS and a valid EI/BMR ratio were more likely to be married or living as married (p=0.006) and to have adequate physical activity (p=0.002) than other women. Women with an ARFS, but not a valid EI/BMR ratio were more likely to be Australian-born (p=0.003) and live in a non-urban area (p=0.002). Women without an ARFS had less education (p< 0.0001), were more likely to have fair or poor self-rated health, (p <0.0001), to have had more visits to a general medical practitioner in the past year (p=0.006) and were more likely to have depressive symptoms (p=0.0007) than women with an ARFS. There were no associations between having an ARFS with or without a valid EI/BMR ratio, and reported level of difficulty in managing on income, current smoking or having consulted a medical specialist in the previous year.
Table 2 reports the demographic characteristics of women by quintile of ARFS. Women with higher ARFS score tended to have more education, be married or living as married, not have difficulty managing on their income, not currently smoke, report their health as good to excellent, not report depressive symptoms, had less than six consultations with a general medical practitioner and have adequate levels of physical activity. There was no statistically significant association with area of residence, country of birth, or having consulted a medical specialist.
|
Table 3a reports the mean macronutrient and selected micronutrient intakes by quintiles of ARFS unadjusted for total energy intake and Table 3b reports them standardised per 1000 Calories for all women and for the subgroup with a valid EI/BMR ratio. There was a weak, though highly significant, correlation between total energy intake and ARFS (r = 0.21, P<0.0001) which disappeared when we examined the subgroup least likely to have misreported their intakes on the FFQ (valid EI/BMR), (r = 0.02, P>0.05). In the unadjusted data, a higher ARFS was associated with higher absolute intakes of total fat, fibre, beta-carotene, calcium, folate, iron, zinc, sodium, thiamin, niacin, riboflavin, vitamin C, retinol equivalents, alcohol and vitamin E. Higher ARFS was also associated with a reduced percentage of energy intake as fat, saturated fat and monounsaturated fat, and increased percentage of energy from carbohydrates and protein, p<0.001. Most of the increases in absolute nutrient intakes across quintiles of ARFS were much smaller when nutrients were standardised for energy, while the total grams of fat decreased as energy standardised ARFS increased (Table 3). Similar trends were demonstrated in the valid EI/BMR ratio sub-group except that the associations with total energy and percentage of energy from polyunsaturated fat and beta-carotene were not significant, p>0.05.
|
|
|
|
DISCUSSION |
|---|
A questionnaire to measure ARFS would be simpler, carry a lower respondent burden to complete and have a lower cost of analysis compared to the full FFQ, due to inclusion of only those foods consistent with dietary guidelines and the reduction in the number of frequency categories in reporting. The ARFS is associated, as expected with indices of better self-reported health and a more nutrient dense, lower fat diet, hence it may be an appropriate alternative to an FFQ when specific dietary factors are not the focus of the research.
To achieve the maximum score of 74 for the ARFS requires not only consuming the recommended two fruit and five vegetables or more per day on average, it also means that each of the 14 different fruits (including juice) and 22 different vegetables are consumed at least once a week. This also applied to the other food group subscales given in Table 1, grains (14 items), meat and protein foods (12 items) and dairy (seven items). Thus people may be achieving the recommended number of serves for all food groups, but unless they encompass this wide variety of fruit and vegetable types, they would not achieve the maximum ARFS. However, with the large number of vegetable types to select from, if each were eaten once a week, this would equate to an average of less than four types per day (i.e. 22/7 = 3.1), and for fruit 14 types once a week is equivalent to two per day.
Dietary diversity scores that count each "healthy’ item consumed are likely to be correlated with intake of energy and other nutrients, because more food is consumed [27]. However, while there was a weak correlation between energy intake and ARFS in this study, this disappeared when we examined the sub-group, least likely to have mis-reported their dietary intake. When nutrient intakes were expressed per 1000 Calories, the positive associations between ARFS and nutrient intakes remained, except for beta-carotene. This indicated that a higher score reflected a better quality diet and nutrient profile for a standardized quantity of energy.
In the sub group with EI/BMR in the range from 1.27 to 2.1, there was little difference in energy intake across quintiles of ARFS, but the range of ARFS was similar to that in the full analysis, indicating that it was possible to vary the ARFS without consuming more energy. This also reflects the possibility that women with higher ARFS had lower intakes of other "unhealthy’ foods that contribute to energy intake, but not ARFS.
The fact that BMI was not higher in women with a high ARFS suggests that either their higher physical activity compensated for higher energy consumption, or they have over-estimated their intake of different foods. These results are, however, similar to those found by Kennedy [28] who, using the Healthy Eating Index to measure diet quality and variety found that lower BMI was associated with higher intakes of vegetables and fruits.
As a group these women did not achieve the national recommendations for fat consumption less than 30% of total energy intake and saturated fat less than 8%. As the ARFS score increased, the percentage of energy contributed by protein and carbohydrate increased and the percentage from fat decreased. While higher ARFS were consistently associated with higher micronutrient intakes, for many micronutrients and fibre, only those in the highest ARFS quintiles achieved the national recommended dietary intakes. The biggest contributors to variation in ARFS were the fruit and vegetable components. This was expected given the large number of items in the FFQ, with possible maximum component scores of 14 and 22. Clearly, there is room to improve the diversity within core food groups, while balancing energy intake through a reduction in energy dense, nutrient poor foods.
These results are important given that Kant has demonstrated an association between dietary variety and quality and mortality in women, using the RFS [3, 4]. In 42 254 women with a mean age of 61 years enrolled in the Breast Cancer Detection Demonstration project, the RFS was inversely associated with total mortality (RR = 0.8; p < 0.001) over 9.5 years. In addition, there was a reduction in risk for those in the highest quartile of RFS versus the lowest for cancer mortality (RR = 0.74; p < 0.001), breast cancer mortality (RR = 0.75; p < 0.06), colon/rectum cancer mortality (RR = 0.49; p < 0.01) and lung cancer mortality (RR = 0.54; p < 0.001) [4]. There is an important difference between the RFS and the ARFS in that we have included component scores for moderate alcohol intake and fat type. Therefore, it will be important to assess whether the relationship between the ARFS and mortality in the ALSWH differs from these findings.
McCullough and Willett [7] have recently demonstrated in 67 271 women from the Nurses’ Health Study, that those in the top versus the bottom quintile of AHEI score had a significant reduction in major chronic disease risk (RR = 0.89, 95% CI 0.82–0.96, p < 0.009), and a lower relative risk for CVD (RR = 0.72, 95% CI 0.60–0.86, p <0.001), although the AHEI score did not predict cancer risk. The AHEI uses a combination of components that reflect food group and nutrient intakes. It maximizes each component score at 10 (vegetables, fruit, nuts and soy, ratio of white to red meat, fiber, trans fat, ratio of poly to saturated fat, alcohol) or 7.5 (duration of multivitamin use). Like the ARFS it does reflect fat type and moderate alcohol consumption. Consequently, future comparisons of morbidity and mortality in ALSWH with the results from studies such as the Nurses’ Health Study will be important.
To date in Australia there have been limited studies reporting indices of overall diet quality useful for evaluating associations with indicators of health or disease. In particular no studies have been undertaken in nationally representative samples of adults, as opposed to patient populations. Wahlqvist [29] calculated a food variety index from 7-day food records in people with type 2 diabetes and healthy controls to measure associations with indices of arterial wall compliance. In 1987 Kune et al [30] calculated dietary risk scores in 715 subject with colorectal cancer and 727 matched controls based on food groups and nutrient intakes assessed from a quantitative diet history covering the previous 20 years. As the ALSWH will continue for 20 years the opportunity will arise to assess the ARFS as a predictor of morbidity and mortality in this population.
Limitations
As with all instruments used to measure to dietary intake, the ARFS has its limitations. There is a low sensitivity for scoring given that having a recommended food once per week adds one point to the total score in the same way as does having the same food three or more times per week does. However, the direction of this bias is likely to weaken any associations with outcome variables. A study is currently underway to examine the relationship between ARFS and indicators of health service utilization and expenditure in this population.
Under and over-reporting may also have occurred in the FFQ and we have attempted to address this by repeating the analyses on the sub-group for whom ratios of energy intake to basal metabolic rate are in a range consistent with physical activity levels from sedentary to highly active women in this age group. Another limitation is the heavy weighting given to vegetables and fruit, which can be attributed to existing interest in different types of vegetables and fruit in relation to cancer at the time the FFQ was developed by the Cancer Council of Victoria. However, this issue is common to other dietary variety tools [2, 5, 6]. In addition, Australian dietary guidelines recommend high targets of five daily serves of vegetables and two serves of fruit.
|
CONCLUSION |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
The ARFS is associated with increased density of nutrients other than fat and saturated fat and is largely independent of energy intake, hence appears to appropriately reflect diet quality and offers the potential for the development of a brief tool to measure diet quality and variety. This would reduce respondent, analytic and cost burdens if used as an alternative to the FFQ possibly allowing more regular assessment of diet quality and variety in women participating in the ALSWH. While other researchers have shown that low diet variety and quality are associated with increased morbidity and mortality, time will tell if this is also the case in the ALSWH. Finally, this study highlights the gap between present national dietary recommendations and self-reported dietary habits and may provide a basis on which to refine key dietary messages for this population of mid-aged Australian women.
|
ACKNOWLEDGMENTS |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
Received June 29, 2006. Accepted December 8, 2006.
|
REFERENCES |
|---|
|
TOPFOOTNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSCONCLUSIONACKNOWLEDGMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. J. Wengreen, C. Neilson, R. Munger, and C. Corcoran Diet Quality Is Associated with Better Cognitive Test Performance among Aging Men and Women J. Nutr., October 1, 2009; 139(10): 1944 - 1949. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
