Strengths and limitations

A major strength of the current research was the large nationally representative sample. Another strength was the consistent methodology for the 24 h diet recall and the food group coding.

However, no food grouping scheme is optimal; basic foods are easy to classify but mixed dishes require complex decisions and the frequency of these increased in the second survey. Not being able to specify and quantify components of mixed dishes influenced the way they were reported within a diet recall, and hence their food group allocation. For example, in the 2008/09 NZANS, more mixed dishes for which participants did not know the recipe were reported within the meat food groups (Fig. 1). Therefore in the 2008/09 NZANS, the meat food groups (such as beef) included a greater proportion of non-meat food items (such as sauces, rice, potatoes) compared with the NNS97, so they could be described as being more dilute in terms of meat content.

Another factor complicating the comparison between the two surveys is that reported energy intake was lower in the 2008/09 NZANS across all age groups in men and for women aged 19–30 years (Supplementary Table 1). The decline in energy intake observed could be a combination of an actual decline and an increase in under-reporting. All dietary assessment methods are associated with some level of under-reporting. However, an increase in under-reporting over time could have implications for our findings. Secular changes have been documented with respect to increased under-reporting⁽ Reference Rennie, Jebb and Wright ^{12 )}, thought in part to be due to increased health consciousness and social desirability⁽ Reference Scagliusi, Ferriolli and Pfrimer ^{13 )} and an increased prevalence of overweight and obesity^{( 4 ,} Reference Gemming, Jiang and Swinburn ^{9 ,} Reference Garriguet ^{14 )}. There is also evidence for selective under-reporting of some foods considered less socially desirable, such as cakes and pastries, sugar and confectionery, and hot chips/French fries⁽ Reference Lafay, Mennen and Basdevant ^{15 )}. In order to take into account the decline in energy intake between the surveys, we also examined trends using percentage of total energy from each food group and conducted a sensitivity analysis with low energy reporters removed. For both analyses the results were consistent and in the same direction as our main analyses (Supplementary Tables 3 and 4). We are therefore confident that the trends we found are robust and reflect what has actually occurred within the NZ population.

Trends in food group consumption

The decline in the consumption of the staple carbohydrate foods, such as bread and potatoes, is congruent with a higher consumption of rice and pasta. Potato consumption has similarly been shown to have declined among adults in the UK⁽ Reference Whitton, Nicholson and Roberts ^{16 ,} Reference Wrieden, Armstrong and Sherriff ^{17 )}, France⁽ Reference Dubuisson, Lioret and Touvier ^{18 )}, Spain⁽ Reference Ribas-Barba, Serra-Majem and Salvador ^{19 )} and Sweden⁽ Reference Krachler, Eliasson and Johansson ^{20 )}. Suggested reasons for the displacement of potatoes in the UK include: (i) the longer time required to prepare and cook potatoes compared with pasta and rice; (ii) an increase in ethnic cuisine; and (iii) increased availability of rice and pasta⁽ Reference Riley ^{21 )}. These reasons are probably also applicable to NZ.

The large decrease we have shown in bread consumption should be closely monitored, in particular among young females. Our findings showed the compensatory increase in percentage of daily energy from bread-based dishes in males but not in females. Given the need to augment iodine intake and considering the reduced use of discretionary iodized salt, it is now mandatory to fortify bread with iodized salt in NZ and Australia⁽ Reference Edmonds and Ryan ^{22 )}. The effectiveness of this programme in NZ may be compromised if bread intake reduces. In addition, bread is a vehicle for folic acid – although unlike iodine, it is not mandatory that bread in NZ be fortified with folic acid⁽ Reference Bradbury, Williams and Mann ^{23 )}. Similar downward trends in bread consumption have occurred in the UK and the decline has been attributed to an increased perception of bread as ‘fattening’, the fad to reduce carbohydrate intake, avoidance due to perceived wheat or gluten intolerance or allergies, and increases in the relative price of bread⁽ Reference O’Connor ^{24 )}. In NZ, those of South Asian, East Asian and South-East Asian ethnicities are less likely than NZ Europeans to report consuming bread⁽ Reference Parackal, Smith and Parnell ^{25 )}. The proportion of these ethnicities is increasing in NZ. Between 1996 and 2006 the ‘Asian’ population approximately doubled to 9·2 % of the population (from 194 800 to 404 400)^{( 26 )}. Bread is still a major source of energy and nutrients in NZ, but the potential nutritional impact of ongoing declines in consumption should be monitored.

Despite an ongoing ‘5+A Day’ campaign in NZ, fruit intake has not changed since 1997. In contrast, fruit intake in the UK and France has increased⁽ Reference Whitton, Nicholson and Roberts ^{16 ,} Reference Dubuisson, Lioret and Touvier ^{18 )}. Furthermore, our findings indicate a decrease in the percentage reporting the food group vegetables, although this may be partly offset by a greater intake of vegetables from mixed dishes. Work is currently in progress to disaggregate mixed dishes to provide a complete estimate of vegetable intake as has been conducted in the UK⁽ Reference Fitt, Mak and Stephen ^{27 )} and the USA⁽ Reference Branum and Rossen ^{28 )}.

Between 1995 and 2002, the per capita red meat supply in NZ decreased by 36 %⁽ Reference Laugesen ^{29 )}. Descriptive results reported elsewhere showed Fe and Zn intakes declined for males between the NNS97 and 2008/09 ANS, and Zn intake declined among females^{( 4 )}. A reduction in beef and lamb consumption has also been shown in other countries, including the USA⁽ Reference Wang, Beydoun and Caballero ^{30 )} and in Australia along with an increase in poultry⁽ Reference Espinel and Innes-Hughes ^{31 )}. Our results revealed beef consumption decreased most among women aged 19–30 years, the age group most in need of haem Fe. However, a large increase in poultry consumption may to some extent have countered this effect. Young females may perceive chicken as light and fresh (‘feminine’ food), healthy (low in fat), versatile due to its blandness and convenient (portions that are easy to cook and prepare)⁽ Reference Kennedy, Stewart-Knox and Mitchell ^{32 )}. As in the UK⁽ Reference Whitton, Nicholson and Roberts ^{16 )}, poultry is now the most commonly eaten meat in NZ.

Our results show a reduction in reported butter consumption with concurrent small increases in fats and oils, and margarine. However, care should be taken in interpreting these results. The food group fats and oils included known ingredients in mixed dishes entered as a recipe. It also captures fats and oils possibly added before consumption, such as olive oil as a dip for bread. Similarly, butter and margarine consumption reflects known ingredients in mixed recipes (including sandwiches) and that added to food as a spread, but not baked goods and bread-based dishes (for which participants did not report recipes), and therefore we do not have a complete picture of consumption of various fats and oils. The percentage of daily energy from table spreads and margarine declined over the study period, possibly reflecting the greater availability of reduced-fat table spreads and the decline in bread consumption.

In 2008, food prices in NZ rose by 8 % from the previous year, the largest annual increase in food prices since 1990. There were substantial increases between 2001 and 2008 in the price of butter (89 %) and milk and cheese (22 %)^{( 33 )}. Prices for these foods have continued to rise and this may have had a lasting impact on food choices^{( 34 )}. It is interesting that although the percentage reporting milk declined, Ca intake was similar between the surveys for females and increased for males^{( 4 )}. This may be because of increased fortification of dairy products, including Ca-enriched milk which was not available in 1997.

We found a decline in the percentage reporting some ‘eat less’ foods (sugar and confectionery, cakes, biscuits and puddings). This change may explain the significant decrease in sucrose intake reported for males between the two surveys^{( 4 )}. There was a small increase in those reporting snack foods such as crisps and cereal bars. Within an Australian context, Branwell et al. argue that the disappearance of puddings and desserts is indicative of wider changes with respect to family eating habits which are driven by greater time pressures and increased availability of more convenient treats that can be eaten at any time of day⁽ Reference Banwell, Broom and Davies ^{35 )}.

Meat pies are considered an iconic national dish in NZ⁽ Reference Bell and Neill ^{36 )}. However, the combination of the pastry and poor-quality meat means they are often high in fat. Although still a popular takeaway food⁽ Reference Smith, Gray and Fleming ^{37 )}, their popularity is clearly waning.

Sex and age differences

The current analysis shows that there are differences in reported food choice between males and females across all age groups. This is congruent with other research which has shown a greater intake of foods such as fruit and vegetables among females⁽ Reference Wardle, Haase and Steptoe ^{38 –} Reference Roos, Lahelma and Virtanen ^{40 )}. Females have been described as having a more ‘virtuous’ approach to food compared with males⁽ Reference Beardsworth, Bryman and Keil ^{39 )}. They are more likely to report to adhering to dietary guidelines and more inclined to consider moral and ethical issues surrounding food choice⁽ Reference Beardsworth, Bryman and Keil ^{39 )}.

The greater contribution of meat to the diet of males compared with females has been shown in Europe⁽ Reference Ocke, Larrañaga and Grioni ^{41 ,} Reference Prättälä, Paalanen and Grinberga ^{42 )} and the USA⁽ Reference Wang, Beydoun and Caballero ^{30 )}. In the 2003–04 National Health and Nutrition Examination Survey males compared with females reported a higher intake of red meat but a lower intake of poultry⁽ Reference Wang, Beydoun and Caballero ^{30 )}. Research in the UK found more females reported they had decreased red meat intake in the past 10 years or had avoided red meat compared with males⁽ Reference Beardsworth, Bryman and Keil ^{39 )}. In our study we did not find evidence of a difference between males and females in the percentage who reported consuming beef, lamb or poultry or in the percentage of daily energy they contribute. We found, however, that males compared with females were more likely to report pies (mostly meat pies) and sausages and processed meats, all of which are higher in fat than leaner cuts of meat.

Nutrition education is frequently aimed at females as the gatekeepers of family health. Nevertheless, males carry a greater burden in terms of diet-related chronic disease. In 2006 CHD accounted for nearly double the number of deaths (12 %) in males compared with females (7 %) in NZ^{( 43 )}. Males are increasingly sharing the food preparation for a household⁽ Reference Gatley, Caraher and Lang ^{44 )} and may be a more important influence, and thus targets for future interventions. Health promotion campaigns targeting males should focus on decreasing consumption of processed meats, alcohol and sweetened beverages and on increasing both fruit and vegetable intakes.

Trends were in the same direction and of a similar magnitude by age group for nearly all food groups. This suggests that changes to the food supply and environment have impacted on different age groups equally. However, clear differences were found in food choices by age in the 2008/09 NZANS. Older adults were more likely to report fruit, vegetables, bread and potatoes compared with younger adults. They were also more likely to report cakes, puddings and biscuits compared with younger age groups, but were less likely to choose snacks, snack bars and sweetened beverages. Our results suggest older adults were more likely to follow a traditional NZ diet pattern with lower intakes of pasta and rice. To support this view, in previous research we found that young NZ adults were more likely to consume fast food compared with older adults⁽ Reference Smith, Gray and Fleming ^{37 )}.

Devine argues that a life-course approach should be taken to understand food preferences⁽ Reference Devine ^{45 )}. Earlier food preferences are maintained as we age and ‘past food experiences set the stage for current food choices’⁽ Reference Devine ^{45 )}. People born at different times have been exposed to different environments, nutritional information and dietary guidelines⁽ Reference Devine ^{45 )}. Some dietary changes occur in response to life events, such as pregnancy, but many dietary patterns are taken forward as we age. The dietary choices of the younger age groups do give some indication of future habits.