Big frogs in a little pond: individual studies help to fill gaps in Australia's small pond of dietary data.

The importance of monitoring the food and nutrient consumption of Australia's population and key subgroups has been highlighted as one of the four priority objectives of Australia's National Food and Nutrition Policy (1) and is a priority action area in Eat Well Australia (2).

The 1995 National Nutrition Survey (3) provided, for the first time, reliable estimates of mean food and nutrient intakes for the population as a whole and for major population subgroups of the Australian population. With appropriate adjustments, the 1995 survey results could also be compared with the 1983 and 1985 National Dietary Surveys of adults and of children, to give the first indication we have of recent trends in food and nutrient consumption (4,5).

From 1998 to 2001 the work of the Australian Food and Nutrition Monitoring Unit (AFNMU) contributed to filling data gaps, refining new measurement methods and creating a framework for an ongoing system for food and nutrition monitoring. One of AFNMU's final recommendations called for the 'key next steps' of another national nutrition survey, together with interim use of standardised short survey questions to measure selected food habits (6). However, to date there has been little progress in obtaining funding for another national survey, or in coordinating the states to collect similar interim dietary information, or in creating a systematic method of collecting and disseminating food and nutrition related data.

Until a system for routine food and nutrition monitoring is in place, selected studies such as that reported by Webb et al. in this issue (7) can help to fill some gaps in information about special groups, as well as supply valuable methodological information useful in designing shortcut dietary assessment methods.

In the article, Meat consumption among 18-month-old children participating in the Childhood Asthma Prevention Study, Webb et al. raise a number of considerations which are important in obtaining useful information from population dietary studies.

Detailed methods of dietary data collection enable multiple uses

Webb et al. stress the importance of, and need to use detailed methods of dietary data collection when studying groups for whom there is little dietary information. Apart from providing detailed quantitative data on food intake, the data can be used to inform the development of purpose- and age-specific measuring tools. In this article the investigators used a weighed food record (WFR) to obtain the best estimate of food and nutrient intake of 18-monthold children. The WFR method involved considerable training of each family individually to keep the record, and home visits by the study nutritionist to check details, such as accuracy of weights, brand names, etc. While it is important that this level of detail be clearly documented, it is actually unusual to find a paper that reports the full details of dietary data collection and analysis, an issue that plagues nutrition journals. Recommendations from Nelson and others (8) have stressed the need for such detail in interpreting the results of studies.

Assessment and reporting of food as well as nutrient intake

The CAPS paper illustrates the importance of investigating food intake data along with nutrient intake. For example, as discussed in it, both meat and milk intake can be a subject for concern among toddlers, in part because of their importance in iron and zinc nutrition. The continued reliance on milk as the main protein source during toddlerhood has been identified as perhaps the main problem with low intakes of meat and hence suboptimal zinc and iron status among young children (9-11). However, Webb et al.'s findings suggest that the types of meat that toddlers are given are also likely to be important to the zinc and iron status of toddlers. In this study, the most popular meats were beef mince, chicken nuggets, grilled beef sausage and other processed meats. Beef mince and beef sausage were the only meats consumed regularly that contain significant amounts of iron and zinc. There are many urban myths about feeding toddlers, and one of them is likely to be that toddlers will only eat chicken and processed meats at that age. The authors point out that children consumed red meats, including beef mince, in the form of mixed dishes such as stews and casseroles, and so it is certainly probable that children of this age will eat red meat when suitably prepared. While the mean intakes for iron met the Recommended Dietary Intake (RDI) for this age group, the mean intake of zinc was marginally lower. This concurs with findings of many select studies of preschool children, where zinc is increasingly of concern. While dairy foods also make a significant contribution to zinc intake in young children (see Table 4 in Webb et al.), the reduction in milk intake over time can compromise zinc intake if not replaced with meats high in zinc.

Respondent burden and response rate

Survey planners generally try to strike a balance between the level of accuracy and detail in data collection methods and the effects on participant burden and response rates. Yet, in the Webb et al. study, response rates and participation were quite respectable, considering that weighed food records require considerable time and effort on the part of respondents. It is notable that the US nutrition monitoring system has successfully used WFR on thousands of respondents in its Continuing Survey of Food Intake of Individuals (12) and in several studies of large groups, response rates to WFR are higher than might be expected (13). Compared with other dietary assessment methods, the WFR data are often so valuable as to warrant consideration of this method with some populations, even on large groups, provided the purposes and uses for collecting the dietary data merits the effort, detail and expense.

Need for biomedical data

Webb et al. acknowledge that this study would have benefited from the inclusion of biochemical measures, e.g. of iron and zinc status on these children, to assess the associations with intakes and types of meats. Iron deficiency anaemia is the most common nutritional deficiency worldwide. In Australia, we lack good data on its prevalence, particularly among preschool age children, and selected studies in Australia and New Zealand suggest that it is a problem among a significant minority of children, particularly those who are socioeconomically disadvantaged.

Biomedical indicators collected along with dietary data offer numerous opportunities to assess interrelationships, understand mechanisms, and gauge the validity of dietary data. Interrelationships are of key interest for issues such as carefully measured dietary intakes of meat and vegetables, as they may relate to blood iron and zinc levels, as well as serum carotenoids, blood folate, homocysteine, and vascular disease.

Similar to the need to weigh up the respondent burden/response rates in selecting dietary assessment methods as described above, there is a similar need to weigh up the costs and benefits of biomedical measures (particularly blood assays) in planning population surveys and studies for particular purposes. The collection of biomedical indicators may have a negative influence on response rate. For example, the AusDiab study, a national study to determine the prevalence of diabetes and cardiovascular risk factors, had a low response rate; nationally approximately 28% (14). The AusDiab two-hour physical examination included two blood draws for glucose and extensive medical interviews and lifestyle questionnaires. Although the AusDiab study collected important data on overweight and obesity as well as diet by food frequency questionnaire, there is concern that the data may be biased towards a more health conscious population and therefore may not be representative of the general Australian population (15). The response rates obtained in the pilot study for the Australia Health Measurement Survey were also low, which may be been at least partly due to the need for participants to provide a blood sample (16).

In this regard, smaller, purpose-specific studies which include biomedical measures may have better success in collecting such information, particularly if clinicians and those who collect the data take steps to establish good relationships with subjects.

Need to collect information on special population subgroups

The study reported by Webb et al. highlights the lack of food and nutrition data on one- to two-year-olds, only one of many data gaps resulting from the lack of a system for monitoring nutrition of the Australian population, especially among vulnerable groups.

Webb et al.'s paper contributes new knowledge in the area of food habits of Australian one-to two-year-olds. Recent and ongoing surveys miss this age group entirely. The 1995 NNS sampled those aged two years and above, while the National Health Surveys collect data regarding breastfeeding and related feeding practices in those aged up to one year old. Most computer assisted telephone interview (CATI) surveys used by states around the country have similar gaps, with the one-to two-year-olds often not included. Although this study, as the authors acknowledge, is not a randomly selected population-based study, it is a good start in looking at food habits among this age group, especially because of the sample size of the study and the kind of dietary data it has generated.

Need for clear objectives of how these data will be used and what the information needs are for these data

Another important use of nutrition survey data, is to fill gaps in methods of monitoring food and nutrition in the population. As the Webb et al. article points out there is currently considerable interest in developing methods that can be used in large population surveys in a variety of groups, and in children in particular, especially in relation to food habits related to obesity. The study gives us, for the first time, portion sizes consumed by this group of small eaters. Such information could be used, along with information about other foods, to generate age-specific diet questionnaires.

The Australian Guide to Healthy Eating provides no quantitative direction to parents and carers for feeding children aged one to four (17). Perhaps this study could provide the basis for further population-based surveys, and some more specific draft dietary guidance for a neglected age group.

Further debate, planning and advocacy are needed to address and resolve issues related to monitoring food and nutrient intake, not only in the general population but particularly among nutritionally vulnerable groups. Meanwhile, purpose-specific studies such as CAPS can generate valuable data when investigators plan for, and exploit the full potential of the data they collect.

(See paper by Webb et al., pages 12-20)

References

1. Commonwealth Department of Health, Housing and Community Services. Food and Nutrition Policy. Canberra: Australia Government Publishing Service; 1994.

2. SIGNAL. Eat Well Australia: an agenda for action for public health nutrition 2001-2010. Canberra: National Public Health Partnership; 2001.

3. Australian Bureau of Statistics. National Nutrition Survey: nutrient intakes and physical measurements: Australia, 1995. Canberra: ABS; 1998.

4. Cook T, Rutishauser IHE, Allsop R. The bridging study--comparing results from the 1983, 1985 and 1995 Australian national nutrition surveys. Canberra: Commonwealth Department of Health and Aged Care; 2001.

5. Cook T, Rutishauser IHE, Seelig M. Comparable Data. Canberra: Commonwealth Department of Health and Aged Care; 2001.

6. Marks G, Rutishauser IHE, Webb K, Picton, P. Key food and nutrition data of Australia 1990-1999. Canberra: Commonwealth Department of Health and Aged Care; 2001.

7. Webb K, Rutishauser I, Katz T, Knezevic N, Lahti-Kosk M, Pear J, et al. Meat consumption among 18-month-old children participating in the Childhood Asthma Prevention Study. Nutr Diet 2005;62:12-20.

8. Nelson M, Margetts BM, Black AE. Checklist for the methods section of dietary investigations. Metabolism 1993;42:258-66.

9. Soh P, Ferguson EL, McKenzie JE, Skeaff S, Parnell W, Gibson RS. Dietary intakes of 6-24 month-old urban South Island New Zealand children in relation to biochemical iron status. Public Health Nutr 2002;5:339-46.

10. Cowin I, Emond A, Emmett P, and the ALSPAC study team. Association between composition of the diet and haemoglobin and ferritin levels in 18-month old children. Eur J Clin Nutr 2001;55:278-86.

11. Thane CW, Walmsley CM, Bates CJ, Prentice A, Cole TJ. Risk factors for poor iron status in British toddlers: further analysis of data from the National Diet and Nutrition Survey of children aged 1.5-4.5 years. Public Health Nutr 2000;3:433-40.

12. USDA. Food Surveys Research Group. Food and Nutrient Intakes by Children 1994-96, 1998, Table set 17, USDA Continuing Survey of Food Intakes by Individuals. Beltsville: Maryland. http:\\ww.barc.usda.gov/bhnrc/foodsurvey/home.htm.2002 Accessed 7 January 2005.

13. Bingham S, Nelson M, Paul A, Haraldsdottir J, Loken EB, Van Staveran W. Methods for data collection at an individual level. In: Cameron M, van Staveren W. Manual on methodology of food consumption studies. Oxford: Oxford University Press; 1988.

14. Dunstan DW, Zimmet PZ, Welborn TA, Cameron AJ, Shaw J, de Courten M, et al. Australian Diabetes, Obesity and Lifestyle Study (AusDiab). Methods and response rates. Diabetes Res Clin Pract 2002;57:119-29.

15. Coyne T, Findlay M, Ibiebele T, Firman D. Overweight and obesity in Australia: an underestimate of the true prevalence? MJA 2004;180:93.

16. Australian Institute of Health and Welfare. Australian Health Measurement Survey: report on the pilot test. Canberra: AIHW; 2004.

17. Smith A, Kellet E, Schmerliab Y. The Australian Guide to Healthy Eating. Canberra: Commonwealth Department of Health and Family Services; 1998.

Terry Coyne, BSc, MSc, PhD, RD

Senior Lecturer 

School of Population Health

University of Queensland