Australian Health Survey Louise Gates Director, ABS Health section louise.gates@abs.gov.au Paul Atyeo Assistant Director, Health section Susan Shaw Senior Analyst, Health Surveys Section July 2014
Structure of the Australian Health Survey General population sample size = 26,000 households NATIONAL HEALTH SURVEY (NHS) 15,500 Households 1 Adult + 1 child = 20,500 persons Detailed conditions Medications and supplements Health related actions Days of reduced activity Social & emotional wellbeing (18 yrs +) Physical activity (15 yrs +) Private health insurance status (18 yrs +) Breastfeeding (0-3 yrs) Disability status Alcohol consumption (15 yrs +) Family stressors (15 yrs +) Personal income (15 yrs +) Financial stress NATIONAL NUTRITION AND PHYSICAL ACTIVITY CORE CONTENT SURVEY (NNPAS) 25,000 Households 9,500 Households 1 Adult + 1 child (2 yrs +) = 32,000 persons 1 Adult + 1 child (2 yrs +) = 12,000 persons Household information Demographics Self-assessed health status (15 yrs +) Smoking (15 yrs +) Physical measures (height, weight, waist and body mass index) Dietary behaviours Blood pressure (5 yrs +) Female life stage (10 yrs +) Selected conditions NNPAS Telephone follow-up 2 nd dietary recall 8-day pedometer (5 yrs +) 24 hr dietary recall Food security Food avoidance Physical activity NATIONAL HEALTH MEASURES SURVEY (NHMS) All survey participants (aged 5 yrs +) invited to VOLUNTEER 11 000 persons Key blood tests (12yrs +) and urine tests (5yrs +) of nutritional status and chronic disease markers
AHS releases to-date Publication Release date First Results (cat. no. 4364.0.55.001) 29 th Oct 2012 Health Service Usage and Health Related Actions (cat. no. 4364.0.55.002) 26 th March 2013 Updated Results (cat. no. 4364.0.55.003) 7 th June 2013 Physical Activity (cat. no. 4364.0.55.004) 19 th July 2013 Biomedical Results for Chronic Diseases (cat. no. 4364.0.55.005) 5 th August 2013 Australian Aboriginal and Torres Strait Islander Health Survey: First Results (cat. no. 4727.0.55.001) 27 th November 2013 Biomedical Results for Nutrients (cat. no. 4364.0.55.006) 11 th December 2013 & 15 th April 2014 (Vitamin D) Nutrition First Results Foods and Nutrients (cat. no. 4364.0.55.007) 9 th May 2014 Australian Aboriginal and Torres Strait Islander Health Survey: Updated Results (cat. no. 4727.0.55.006) 6 th June 2014
AHS Nutrition THINGS TO COME Usual intake of nutrients Comparison with 1995 Comparison with biomedical results Results for Aboriginal and Torres Strait Islander peoples 2015 Comparison with Australian Dietary Guidelines 2015
AHS User s Guide AMPM: Automated Multiple Pass Method
Nutrition data collection 24 hr recall Dietary behaviours Discretionary salt Usual serves of fruit / veg Whether on a diet Food avoidance
Data collection AMPM (Automated Multiple Pass Method) USDA instrument (used in NHANES) 5 steps Automated sequencing Amount options linked to images
5 Steps of AMPM Collect list of foods & beverages consumed previous day Probe for any foods that were missed Collect time and eating occasion for each food Collect details: description, amount, additions Final probe for anything else
Phase 4 Detailed cycle Food-specific questions Specific ingredients Cooking method Oil type Some brands /product names asked for where relevant Amounts
Example question sequencing
Example food path
Coding Food database (AUSNUT 2011-13) 5,644 foods Unique food code, description, inclusions, exclusions Linked to Measures database & Nutrient database
AUSNUT 2011-13 Energy Moisture Macronutrients Protein Total Fat Saturated fat Monounsaturated fat Polyunsaturated fat Linoleic acid Alpha-Linolenic acid Omega 3 Trans fat Total Carbohydrate Total sugars Total starch Dietary Fibre Alcohol (kj) (g) (g) (g) (g) (g) (g) (g) (g) (mg) (mg) (g) (g) (g) (g) (g) Vitamins Preformed Vitamin A Pro Vitamin A Vitamin A retinol equivalent Thiamin Riboflavin Niacin Niacin equivalent Folate Folic acid Total Folates Folate equivalent Vitamin B6 Vitamin B12 Vitamin C Vitamin E Minerals Calcium Iodine Iron Magnesium Phosphorus Potassium Selenium Sodium Zinc Cholesterol Caffeine (µg) (µg) (µg) (mg) (mg) (mg) (mg) (µg) (µg) (µg) (µg) (mg) (mg) (mg) (mg) (mg) (µg) (mg) (mg) (mg) (mg) (µg) (mg) (mg) (mg) (mg)
2011-13 Food classification Similar to 1995 NNS Arranges foods in groups of like products Largely reflects overall product as consumed
Classification caution Concordance does not overcome some differences in coding Mixed dishes where cereal is the major ingredient Pizza Sandwich (includes hotdogs) Burgers Tacos Pasta dishes Savoury rice dishes Implications elsewhere (Bread, Pasta, Meat, Vegetables)
Supplements Collected at end of 24 hr recall Specific type of product, amount consumed Used AustL number to ID products and nutrient information
CURF
CURF
CURF CURF
CURF Data files (SAS, SPSS, STATA) Person Food Supplement Information files Formats Information Copyright Frequencies
CURF file structure Input Output Person records (12,153) Sum each nutrient for each person Person records (12,153) 2) Merge new dataset onto person level using ABSPID 24 hr recall Food records (341,197) Supplement records (25,168) Food records incl. nutrient values Supplement records incl. nutrient values Sum each nutrient for each person Food records incl. nutrient values Supplement records incl. nutrient values 1) Sum amount of g/kj/nutrient amt from defined food for each person (create dataset 12,153 records)
Citation using CURF Elements to include in referencing the CURF Source of data Name of Survey Type of CURF Method of access Statement of what findings based on In References Australian Bureau of Statistics, 2011-12, National Nutrition and Physical Activity Survey, 2011-12, Basic CURF, CD-ROM. Findings based on ABS CURF data. Example In text: ABS (2011-12)
Under-reporting Widely observed and documented that when people report on their food intakes in nutrition surveys, there is tendency to underestimate. This includes: actual changes in foods eaten because people know they will be asked about them, and misrepresentation (deliberate, unconscious or accidental), eg. to make their diets appear more healthy or be quicker to report.
Energy Intake (kj) Mean energy Intake for males and females by age group: 1995 NNS and 2011-12 NNPAS 14,000 12,000 10,000 8,000 6,000 Males, 1995 Males, 2011-12 Females, 1995 Females, 2011-12 4,000 2,000 0 2-3 4-8 9-13 14-18 19-30 31-50 51-70 71 + Age Group
Under-reporting Need to consider the effects of under-reporting (and other sources of error) when making comparisons between surveys We have presented a number of measures and analyses of under-reporting in the user s guide for this survey No measure is perfect Attempt to estimate the amount of energy that is missing, or proportion of people with implausibly low reported intakes Not necessarily equivalent to amount missing for other nutrients Based on comparing energy intakes to theoretical energy requirements for each individual (using Schofield equations and measured weight) Most analyses focus on comparing 1995 NNS and this survey, so apply the same test to both
Under-reporting Limited by needing to compare with the 1995 data set in our analysis of under-reporting, eg Did not include physical activity levels no measurement in 1995, note difficult to derive a PAL from 2011-12 Did not consider responses to diet questions not the same questions
Under-reporting Further planned analyses Comparison of intakes by food groups 1995 to 2011-12, by males and females may be able to draw some inferences about the likely effects of changes in under-reporting from observed patterns in the food intakes, comparing males and females Further information on the multiple regression modelling to be presented in September at the PHAA conference in Perth more on associations between EI:BMR and other variables, including responses to diet questions, in 1995 and 2011-12
Under-reporting Will now provide some more details on the methods used in the under-reporting analysis and the results All available in user s guide
Finding the under-reporting analysis!
Under-reporting Reported proportions of people with implausibly low reported energy intakes (low energy reporters, or LERs) Based on Goldberg cut-off of 0.9 for EI/BMR Individual basis Day one intakes Over 10 years of age With a measured body weight and height Not pregnant or lactating Drawbacks: not very sensitive, line in the sand Will only find the most extreme cases of under-reporting on an individual basis Similar to approach taken in 1995 and analyses of other international surveys, presented for consistency and comparison
Energy Intake (kj) Proportion of LERs for males and females by age group: 1995 NNS and 2011-12 NNPAS 30 25 20 15 10 Males, 1995 Males, 2011-12 Females, 1995 Females, 2011-12 5 0 10-13 14-18 19-30 31-50 51-70 71 and over All 10 and over Age Group (a) Respondents for whom a measured weight or height were not available, children under 10 years of age, and pregnant women were excluded from all analyses based on EI:BMR. All proportions are given as a weighted proportion of respondents for whom EI:BMR could be calculated.
Under-reporting Also presented several other analyses of underreporting Suggest that looking at LERs does not describe the problem well in this survey Still a substantial decline in energy intakes for males from 1995 to 2011-12 between plausible energy reporters from both surveys 8% or 12,100kJ compared to 11,080kJ That is, removing LERs would not remove underreporting bias Has other drawbacks as well lose a lot of sample
Distribution Outlines for Energy Intake over Basal Metabolic Rate for All Ages: NNS95 and NNPAS11-12 EI/BMR 1.55: minimum average energy requirement for normally active but sedentary population (not sick, disabled or frail elderly) 0.9: Goldberg cut-off for a plausible report, i.e. lower 95% confidence limit for a single day for a single individual, allowing for day-to-day variation in energy intakes, and errors in calculation of EI:BMR
How much energy is missing?
EI/BMR What effect has increasing BMI had? Differences by BMI 1.6 Persons 10 years and over: Mean reported energy intake over basal metabolic rate by BMI, 2011-12 1.5 1.4 1.3 1.2 1.54 1.42 1.30 1.28 1.1 1.0 0.9 1.12 1.08 0.8 0.7 0.6 Male Female 0.5 0.4 0.3 0.2 0.1 0.0 Normal weight Overweight Obese Body Mass Index
Multiple regression modelling Increasing proportion of men at a higher BMI has contributed to the apparent increase in under-reporting, but not the whole story Many other covariates significantly related to a higher likelihood of lower EI/BMR score in the model, eg dietary restriction perception of health as poor perception of self as overweight weekend vs weekday intake But after accounting for the effects of BMI, each of these only explained a very small additional amount of variation for men More on this at PHAA conference in Perth
EI/BMR Adjust for under-reporting? Taking out LERs does not remove under-reporting blunt approach, other drawbacks Results for each nutrient will be affected differently by under-reporting Persons 10 years and over: Mean reported energy intake over basal metabolic rate for plausible vs all reporters, 2011-12 1.6 1.5 1.4 1.3 1.2 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1.5 1.49 1.35 1.3 Full sample Plausible energy reporters Goldberg classification Males Females
Future releases on under-reporting Comparison of intakes by food groups 1995 to 2011-12, by males and females may be able to draw some inferences about the likely effects of changes in under-reporting from observed patterns in the food intakes, comparing males and females Further information on the multiple regression modelling to be presented in September at the PHAA conference in Perth more on associations between EI:BMR and other variables, including responses to diet questions, in 1995 and 2011-12
Under reporting (in summary) Present in both 1995 and 2011-12 Occurring in other countries Female under reporting is higher, but males had greater increase Energy deficit of 17% in males 21% females Unlikely to affect all foods equally
Usual intake of nutrients
Usual intakes of nutrients Publication planned for late 2014 Estimate usual intakes from 2 days of nutrient intakes Use new method (NCI method) Statistical model to estimate and remove within-person variation Therefore estimate distribution of usual intakes Publication will include Percentiles of intake by nutrient reference value (NRV) age and sex group Estimated proportions not meeting or exceeding NRVs where applicable Limited comparisons will made to AIs based on mean and median intakes only Note not valid to estimate proportions not meeting or exceeding NRVs based on distributions of day one intakes or day one day two average intakes
Australian Health Survey Contacts: louise.gates@abs.gov.au 02 62526415 paul.atyeo@abs.gov.au 02 65257612 susan.shaw@abs.gov.au 02 62526946