Usual intakes of foods and nutrients

Usual intakes are an estimate of what people 'usually' eat, not just what they eat in one day. These estimates can be done for both foods and nutrients. 

National nutrition surveys usually collect a couple of days of dietary recalls (rather than covering longer periods) because of time, cost, and to reduce burden on respondents. Because these recalls are short‑term, they are adjusted to estimate long‑term or “usual” intake.

For the National Nutrition and Physical Activity Survey (NNPAS) 2023, the ABS collected two 24-hour dietary recalls and then modelled the results to create longer-term intakes. 

For the NNPAS 2023, nutrient and food consumption for the first (Day 1) and second 24-hour recall (Day 2) can be used to estimate usual (habitual) intakes for selected population groups.

Food consumption derived from a single 24-hour recall may not represent the usual consumption patterns of a person because there is often variation in foods consumed on a day-to-day basis. The second 24-hour recall is used to estimate and reduce within-person variation and produce a usual distribution for the population; this can be either for food consumption amounts by AUSNUT food group or serves of Australian Dietary Guidelines (ADG) foods and/or nutrient intakes. It is preferable to use usual food intakes or usual serves to assess dietary patterns and usual nutrient intakes to assess the nutritional status of a population, where possible.

The usual intakes are not just an average of the two days of data for an individual. An average of the two days usually leads to biased estimates (Herrick et al. 2018) as it does not:

• account for days when a person eats none of the food, or highly skewed amounts
• separate within-person (day-to-day) variation from the variation between people
• allow for the correlation between how often a food is eaten and how much is eaten
• use extra information (covariates). 

Estimates of usual intakes require two 24-hour recalls. Estimates of usual intakes for the Aboriginal and Torres Strait Islander population are not available from the National Aboriginal and Torres Strait Islander Nutrition and Physical Activity Survey (NATSINPAS) 2023 as the NATSINPAS only collected a single day of recall. 

Data presented in the Foods and Nutrients release for the NNPAS and NATSINPAS are calculated based on the Day 1 recall, that is the 24 hours (from midnight to midnight) prior to interview.

Usual Nutrient Intakes for the NNPAS are calculated from Day 1 and Day 2 recall data. 

Usual intakes are important because food and nutrient guidelines, such as the Nutrient Reference Values (NRVs) or Australian Dietary Guidelines (ADGs), are based on usual intakes. Without adjusting, the results could over- or under- estimate how many people meet these guidelines.

For example, suppose we want to estimate the number of people who consume at least two serves of vegetables per day. As shown in Figure 1, a single‑day intake measure may indicate a larger proportion of people consuming fewer than two serves. This is simply because their intake happened to be lower on that particular day. However, this may not reflect their typical eating pattern. The opposite could also be true, a single-day intake of serves could be higher than usual.  A usual‑intakes model can account for occasional decreases or increases in consumption. In Figure 1, the estimated proportion of people who usually consume two serves of vegetables per day is higher than what the single‑day data alone would suggest.

There are multiple ways to model usual intakes. The ABS used the National Cancer Institute (NCI) method for this study and for the previous Australian Health Survey (AHS) in 2011–13. The method models dietary intake from 24-hour recalls to:

• estimate the usual intake for groups of people
• account for the effects of different characteristics, such as age or day of week
• account, partially, for bias resulting from measurement error (NCI 2024).

The output from the NCI method is a simulated distribution of usual intakes for groups. The method relies on pooled group data for its calculations. It does not produce usual intake estimates for individual respondents.

The NCI method works best when estimating frequently consumed foods and nutrients. It may be applied by users to the different classifications output in the ABS DataLab, for example AUSNUT 2023. Caution is advised when using the method with infrequently consumed foods and dietary supplements.

The ABS used version 2 of the NCI Method SAS macros, available at Usual Dietary Intakes: SAS Macros for the NCI Method. These are the same versions used in 2011–13.

The NCI method models usual nutrient intake or usual food consumption as:

                                probability of consumption x amount consumed

There are three available models based on this premise.  The ABS used two:

1. One-part model: This model assumes everyone consumes the nutrient or food every day, so the probability of consumption is 1. It is used for nutrients and foods that almost everyone eats daily. Only the amount consumed is modelled, and any zero intakes are replaced with half of the lowest non-zero value in the dataset.
2. Correlated two-part model: Some foods and nutrients are not consumed every day, and the chance of eating them is correlated to how much is eaten. For example, people who drink tea tend to consume more than one cup a day. This model accounts for this correlation.

For the Usual Nutrient Intakes 2023 publication, the two-part model was used when more than 5% of people had zero amounts of the nutrient. This applied to folic acid, caffeine, alcohol, and percentage of energy from alcohol. 

All other nutrients use the one-part model, because almost all people consume them daily.

The data was split into three groups:

• children aged 2 to 11 years
• males aged 12 years and older
• females aged 12 years and older.

The method was run separately for each group. Males and females aged 12 years and over were separated to allow for differences in consumption by sex. Children aged 2–11 years were modelled together because males and females in this age range usually have similar intakes. Combining this group also increases the sample size. 

Alcohol and percentage of energy from alcohol were treated differently. For these, the model was run for both males and females aged 18 years and over and younger age groups were not included as they had too few non-zero alcohol intakes.

Covariates in the NCI method are independent variables or characteristics that describe individuals in a group. They may influence what was consumed, for example, respondents may eat different foods on different days of the week or be less inclined to report as much detail in a subsequent recall (Serban et al. 2022). 

The ABS used four covariates known to affect food and nutrient intake:

• age and sex (combined)
• day of week
• sequence (order) of recall.

The specific categories used are listed in the table below.

Data for age groups not listed above (e.g. total 2–17 year olds) were input to the distribution model (distrib macro) after the covariates have been considered and simulation run (i.e. processed with the mixtran macro and Monte Carlo component of the distrib macro) (NCI 2024).

Ratios compare one variable by another and can be compared to one another using a proportion (for example, the percentage of energy that was discretionary compared to non-discretionary). 

There are four nutrient ratios published for the NNPAS 2023. These are percentages of total energy intake from carbohydrates, proteins, fats, and pure alcohol (ethanol). 

Two approaches can be used when modelling usual intake of a ratio. These either apply the ratio to:

1. each person’s data before the model is run (create a “usual intake of the ratio”)
2. the group usual intake data after the model is run (create a “ratio of usual intake”).

The first approach was used for this survey, and the estimate of the distribution of the 'usual ratio of intakes'. Results should be interpreted based on this approach (Freedman et al. 2010). 

Ratios were calculated on an individual basis, as a percentage of the person’s total daily energy. Energy conversion factors were applied to calculate each macronutrient's contribution. For more information on the calculation of these ratios, see Macronutrient contribution to energy intake.

The Nutrient Reference Values for Australia (NRVs) are evidence‑based recommendations. They specify the amounts of essential nutrients needed to meet the requirements of healthy individuals and reduce the risk of adverse health outcomes. These recommendations are based on long-term or usual nutrient intakes. 

At the time of the Usual Nutrient Intakes 2023 release, the NRVs available were current as of 2017 (NHMRC 2017). Following a decision by the Department of Health, Disability and Ageing (DHDA) and the National Health and Medical Research Council (NHMRC), the age groups for reporting food and nutrient intakes from the NNPAS 2023 were changed. The ABS re-derived weighted NRVs for the new age groups to report survey estimates. For further information about how these NRVs were calculated, see Changes to Nutrient reference Values (NRV).

For each nutrient with an NRV, two steps were followed to estimate the proportion of the population below or above the reference value:

• Step 1. Estimate the population group’s usual nutrient intake distribution. The NCI method was used to estimate these for this survey.
• Step 2.  Compare usual intakes with the NRV to determine the proportion of people whose intake falls above or below the requirement for their age and sex group.

Specific information for the NRVs reported in the Usual Nutrients Intakes 2023 publication can be found below.

To stay healthy, our bodies need certain amounts of vitamins, minerals and other nutrients. One way to check if people are getting enough is by using the Estimated Average Requirement (EAR). 

The EAR is the daily amount of a nutrient that is expected to meet the needs of half the healthy individuals in a particular age and sex group (NHMRC 2006). Usual intake below the EAR indicates the people who are at higher risk of health problems from not getting enough of that nutrient.

To understand how many people might not be meeting their nutrient needs, we compare usual intakes with the EAR. The Usual Nutrients Intakes 2023 publication uses two main methods:

1. EAR cut‑point method: this method is used for most nutrients. It works well when nutrient requirements are even across the population. It simply counts how many people in a group have usual intakes below the EAR.
2. Full probability method: this method is used for iron. Iron needs vary more between people, so a more detailed method is needed to estimate the risk of inadequate intake.

More detail about how these methods were applied is listed below.

The UL is the highest average daily amount of a nutrient intake that most people can consume without a high risk of adverse health effects. As intake increases above the UL, the potential risk increases (NHMRC 2006).

The proportion of people with usual intake above the UL indicates how many may be at risk of health problems from getting too much of that nutrient (NHMRC 2017). Not everyone who consumes above the UL will have harmful effects, but their risks are higher than for people who stay at or below the UL (Health Canada 2017).

Suggested Dietary Targets (SDTs) are a daily average intake from food and beverages for certain nutrients that that may help in prevention of chronic disease. SDTs have been set for some nutrients where there is evidence that a daily average intake below or above the SDT could have benefits in reducing chronic disease risk (NHMRC 2017). Any comparisons of reported usual nutrient intakes to SDTs should be made considering the basis for the SDT used, and any new evidence available since it was set.

The Acceptable Macronutrient Distribution Ranges (AMDRs) define the recommended proportions of energy that should come from each macronutrient. These ranges aim to support good health while still allowing enough of other nutrients (NHMRC 2017). If usual intakes fall outside of these ranges, it suggests an imbalance that may raise long‑term chronic disease risk (Health Canada 2017). 

The Usual Nutrient Intakes 2023 reports the proportion of people with usual intakes below, within or above each AMDR. Macronutrients included are carbohydrates, protein, and fat. For details on how these are calculated, see Modelling ratios of nutrients, above.

There are no NRVs set for caffeine. There are recommended maximum daily levels of caffeine available to compare against (FSANZ 2023). The recommended levels are based on how many milligrams of caffeine a person consumes per kilogram of their body weight. Going above these levels increases the risk of negative health effects (FSANZ 2023). Comparisons of reported usual caffeine intakes to the maximum levels should consider how the limits are set. This includes any new evidence available since they were set.

Some nutrients have different NRVs for people who are pregnant or breastfeeding. This is because their bodies need additional nutrients to support the growth of the baby and produce breast milk (NHMRC 2017). 

The NNPAS 2023 did not report usual intakes for this population because there were too few people in the sample. 

Where usual nutrient intakes are compared with the NRVs, any results that sit outside of the reference values need to be interpreted carefully. It is important to understand how the NRVs were established, because this affects what the results mean. 

At the time of release of Usual Nutrient Intakes 2023, the current NRVs were those published by the NHMRC in 2006 and 2017 (NHMRC 2017). 

More information about how the NRVs were developed for each nutrient is available on the NHMRC Nutrient Reference Values website.  

Food and Nutrition Board: Institute of Medicine (FNB:IOM) (2000), DRI Dietary Reference Intakes: Applications in Dietary Assessment, National Library of Medicine website, accessed 02/03/2026.

Food Standards Australia New Zealand (FSANZ) (2023), Preventing foodborne illness: Caffeine, FSANZ website, accessed 10/03/2026.

Freedman LS, Guenther PM, Dodd KW, Krebs-Smith SM, Midthune D (2010), The Population Distribution of Ratios of Usual Intakes of Dietary Components That Are Consumed Every Day Can Be Estimated from Repeated 24-Hour Recalls, The Journal of Nutrition,140(1):111-6, accessed 02/03/2026.

Health Canada (2017), Reference guide to understanding and using the data: 2015 Canadian Community Health Survey – Nutrition, Government of Canada website, accessed 02/03/2026.

Herrick KA, Rossen LM, Parsons R, Dodd KW (2018), Estimating Usual Dietary Intake from National Health and Nutrition Examination Survey Data Using the National Cancer Institute Method, Vital and Health Statistics, 178(2):2-3, accessed 02/03/2026.

Institute of Medicine (IOM) (US) Committee on Nutrition Standards for National School Lunch and Breakfast Programs (2010), School Meals: Building Blocks for Healthy Children : Appendix I, Dietary Intake Data and Calculation of the Target Median Intake for Iron, National Library of Medicine website, accessed 02/03/2026.

Institute of Medicine (IOM) (US) Panel on Micronutrients (2001), Dietary Reference Intakes for Vitamin A, Vitamin K, Arsenic, Boron, Chromium, Copper, Iodine, Iron, Manganese, Molybdenum, Nickel, Silicon, Vanadium, and Zinc, National Library of Medicine website, accessed 02/03/2026.

National Cancer Institute (NCI) (2024), Usual Dietary Intakes: The NCI Method, NCI website, accessed 02/03/2026. 

National Research Council (1986), Nutrient Adequacy: Assessment Using Food Consumption Surveys, The National Academies website, accessed 02/03/2026.

National Health and Medical Research Council (NHMRC) (2006), Nutrient Reference Values: Macronutrient balance, Eat for Health website, accessed 02/03/2026.

National Health and Medical Research Council (NHMRC) (2017), Nutrient Reference Values for Australia and New Zealand Including Recommended Dietary Intakes, NHMRC website, accessed 02/03/2026.

Serban CL, Chirita-Emandi A, Perva IT, Sima A, Andreescu N, Putnoky S, Niculescu MD, Puiu M (2022), Intake Differences between Subsequent 24-h Dietary Recalls Create Significant Reporting Bias in Adults with Obesity, Applied Sciences, 12(5):2728, accessed 02/03/2026.