This secondary analysis provides nationally representative data on meat/poultry/fish consumption patterns using disaggregated data for over 1500 recipes and meat products. The use of disaggregated data reflect a more precise consumption profile of this food group and demonstrate the effect of disaggregation on survey data using broad food groupings. Our analysis revealed that the use of broad food groupings to estimate meat consumption tended to overestimate intakes of red meat by 9%, poultry (25%), and fish (18%) but underestimated intakes of processed meat (−17%). The overestimation of red meat, poultry and fish was likely due to the broad food grouping method capturing all food components (such as vegetables and grains) within mixed dishes, (for example curries and stir-fries), where meat/poultry/fish was a major component. The underestimation of processed meats is likely due to the broad food grouping method not capturing meats from mixed dishes where meat is only a minor component, such as pizzas, pies and pasta dishes. Consequently, nutrient contributions from meat/poultry/fish were also underestimated in non-disaggregated survey data, with estimates showing differences in key nutrient contributions up to 50%. These findings confirm results from previous studies [21, 25, 26] that overall meat/poultry/fish intake is overestimated in national dietary surveys, when disaggregation is not taken into account. Studying the effect of disaggregation on consumption data is thus a significant issue that should be considered in epidemiological studies examining associations between food/nutrient intake and health outcomes , for monitoring changes over time , and for the development of food based dietary guidelines. Hence, this study provides the detailed meat/poultry/fish classifications and stratifications for gender, age, and socio-economic subgroups.
Based on the disaggregated data, more than 90% of participants reported consuming meat/poultry/fish on the day prior to the interview, with red meat (e.g. beef, lamb, pork) consumed by approximately half, poultry and processed meat by two fifths, and fish/seafood by one fifth of the population. Per-capita data revealed total meat/poultry/fish intake was 152 g per day (118 g for children and 162 g for adults). Red meat (beef, lamb and pork) was the highest contributing meat category (38%), followed by poultry (30%), processed meat (17%) and fish/seafood (15%). Within the red meat category, beef was the most popular meat type, followed by lamb and pork while kangaroo and game meat were consumed in minimal amounts. In the poultry category, chicken was the major meat type, with other poultry meats such as duck, turkey and quail only contributing less than 2 %. In the fish/seafood category, finfish and canned fish were the highest contributors. Organ and offal meat consumption was negligible. In the processed meat category, sausage followed by ham and bacon contributed most to per-capita intake. Overall, the most popular meat type for children and adults of all ages, was chicken.
Comparison of per-capita intakes to other surveys is problematic due to methodological differences. For example, previous analyses of Australian national nutrition surveys have not disaggregated all meat components from mixed dishes as was done in the present analysis [2, 28]. Our data can be compared to similarly disaggregated data from the United Kingdom (UK) and the United States (US). In the UK, average daily intakes of meat/poultry/fish in 2006–11 were 144–173 g for males and 100–117 g for females aged between 36 and 64 years , compared to our findings of 193 g for adult males, and 136 g for adult females. Higher intakes were reported in the US at 255–281 g per day for adults in 2004 .
Red meat consumption for adult males and females was estimated at 75 g and 62 g per day, respectively. Assuming these mean intakes are representative of daily intakes, the total red meat intake on a weekly basis can be estimated to be 525 g for males and 430 g for females. The recommendation for red meat consumption for Australian adults is set at 455 g per week based on both meeting nutrient requirements and as a limit to avoid excessive consumption associated with increased risk of colorectal cancer risk (> 700 g cooked red meat per week) . Although current intakes are close to the recommended intake for women and somewhat higher for men, the proportion regularly consuming excessive red meat intakes cannot be accurately determined from one or two days of recall data [30, 31].
Consumption of fish/seafood was 22 g per-capita per day, translating to 154 g per week, or approximately 1.5 serves per week. This is below the recommended dietary guidelines (two serves per week)  but comparable to previous Australian studies by Meyer et al.  and Rahmawaty et al. . In Europe, the per-capita total fish/seafood intake using disaggregated data was similar for adults, a 27 g for men and 29 g for women  compared to our analysis (27 g for men and 24 g for women).
Consumption of processed meat was relatively common with 38% of the population reporting consuming some type of processed meat on the day prior to the interview, although per-consumer intake was lower than that for non-processed meats (45 g versus 80–100 g). Previous findings that processed pork (including ham and bacon) is the most frequently consumed type of pork in the 2007 Australian National Children’s Nutrition and Physical Activity Survey are reflected in our analysis. Comparisons with other surveys are limited due to the differences in definition of processed meats and reporting methods [34, 35].
Per-consumer data revealed that some meat types were reported in larger amounts per day than others, for example the quantities of lamb, sausages and finfish were larger than those of beef and pork. This is likely to reflect the different portion sizes of various meat cuts consumed, for example two sausages versus a small amount of minced pork in a stir-fry. Most processed meats (except sausage) and seafood were consumed in small quantities (20–50 g) per day.
Socio-economic status was positively associated with fish/seafood consumption but not with any other meat/poultry category. Previous analysis in Australia and the US showed only very modest and inconsistent differences of total meat consumption across socio-economic categories [1, 29]. Data from the U.S. national survey showed slightly larger total meat/poultry/fish consumption in higher-income men, as well as a smaller red meat consumption among high-income women . This may be due to the difference in the cost of meat/poultry/fish types and people’s perceptions and response to nutrition education [5, 36,37,38].
The contribution of meat/poultry/fish to intakes of key nutrients was highlighted in this secondary analysis, particularly for protein, omega-3 fatty acids, iron, and zinc. Beef, chicken, and fish contributed higher amounts of these nutrients to overall intakes when compared with processed meat. Similar findings have been indicated in the U.S. for iron and zinc intake contribution from meat/poultry/fish consumption . In contrast, processed meat contributed significant amounts of saturated fat intake, a consistent observation in previous national surveys [25, 28, 34]. Evidence from observational studies in Ireland and Britain reported that consumption of processed meat was associated with poorer diet quality, lower socio-economic characteristics, and other health related risk factors when compared with other categories of meat/poultry/fish [26, 40].
The significant discrepancy between the contribution of meat/poultry/fish to intakes of key nutrients, in combination with the high proportion of participants consuming meat/poultry/fish, highlights the importance of recipe disaggregation. Another area that would benefit from the use of disaggregated data is the risk assessment for chemical food contaminants that may occur in meat.
Of particular importance to nutrition surveys is a widely observed tendency for people to misreport their food intake. The prevalence of potential under-reporting behaviour in the NNPAS was calculated to be between 19 and 23% using the Goldberg cut-off (energy intake vs. basal metabolic rate < 0.9) by the ABS . The impact of misreporting in the assessment of meat consumption is unknown. Some studies suggest that unhealthy foods with high fat and sugar contents are more likely to be under-reported than core foods such as meats and alternatives [41, 42] although this is not a consistent finding . Further work into the impact of under-reporting on the consumption patterns of different foods from the survey results is required.
The strength of this study included the use of a large representative sample of Australian children and adults. All meat categories and meat types were well described, and stratifications by gender, age, and socio-economic subgroups were undertaken to enable comparison to other studies. A potential limitation was the use of a single 24-h recall to estimate food consumption, although this is a valid method to estimate mean group intakes. The second day was not used as the individual meat types were infrequently consumed (e.g. lamb and pork) and the statistical methods analysing usual intake of such foods requires further dietary assessment on usual frequency of consumption .