2.1. Study Population
We used questionnaire data from the NHS (n
= 19,257), which was conducted between June 2014 and July 2015 across all States and Territories in Australia [10
]. Specific methodology of the NHS can be found elsewhere [10
]. In brief, face-to-face interviews were conducted with a randomly selected adult of the household by trained Australian Bureau of Statistics (ABS) interviewers. For child participants, a parent or guardian answered the questions on behalf of children aged <15 years [11
]. Interviews were conducted in the participant’s private dwelling in metropolitan and rural areas of Australia [11
]. People were excluded from the survey if they were residents of non-private dwellings, such as hotels or boarding schools, or were visitors to a selected dwelling [11
]. The interview components of the NHS were conducted under the Census and Statistics Act 1905.
2.2. Identification of Supplement Users
In a face-to-face interview, participants were asked, “What are the names or brands of all the medications, vitamins, minerals or supplements you have taken in the last two weeks?” [10
]. Participants were encouraged to have the supplements in front of them, and the name and brand were recorded by the interviewer. For the purposes of the NHS, dietary supplements refer to products defined as Complementary Medicines under the Therapeutic Goods Regulations 1990 [12
]. Dietary supplements sold in Australia are regulated by the Therapeutic Goods Administration, which requires them to be listed but not registered (medicines are required to be registered) [12
]. Thus, demonstration of efficacy or safety of supplements is not required. It should be noted that products available on international websites are not regulated by the Therapeutic Goods Administration [12
]. The Therapeutic Goods Administration advises that consumers do not order dietary supplements over the internet unless the ingredients and legal requirements for importation into Australia are known. However, it is likely that some people obtain their dietary supplements online from international websites. The supplements recorded by the interviewer included those registered with the Therapeutic Goods Administration and those purchased overseas. The ABS categorised supplements into 28 groups (Supplementary Table S1
). For the purpose of this study, any participant who reported taking at least one dietary supplement in the previous two weeks was considered a “supplement user”.
2.3. Potential Predictors of Supplement Use
Age was provided as a categorical variable and we re-grouped age as follows: ≤9, 10–17, 18–29, 30–49, 50–69, and ≥70 years. We further categorised these groups as adults (≥18 years, n
= 14,560), adolescents (10–17 years, n
= 1964) and children (≤9 years, n
= 2733). Body mass index (BMI; measured weight in kilograms divided by measured height in metres squared) was categorized for adults according to the World Health Organization’s cut-off points for underweight, healthy weight, overweight, and obese [13
]. For adolescents, cut-off points for BMI categories were assigned using half-yearly sex-and-age specific thresholds as detailed by the International Obesity Task Force [14
]. We did not assess BMI in children, as BMI is not relevant for those aged <2 years, and our age group included children aged ≤9 years.
State/Territory was assigned for all participants as New South Wales, Victoria, Queensland, South Australia, Western Australia, Tasmania, Northern Territory, and Australian Capital Territory. Region of birth was assigned as Australia, Main English-speaking countries (Canada, Republic of Ireland, New Zealand, South Africa, United Kingdom, United States of America), and Other. As the majority of children were born in Australia and New Zealand, region of birth was assessed only in adults and adolescents.
Educational attainment for adults was defined as none after school, Certificate, Bachelor/Diploma, and postgraduate. Socioeconomic status was described by the Socio-Economic Indexes for Areas (SEIFA) 2011 Index of Relative Socio-Economic Disadvantage (IRSD). This is a general socioeconomic index that summarises a range of information about the economic and social conditions of people and households within an area with scores ranging from low (relatively greater disadvantage in general) to high (relative lack of disadvantage in general) [16
]. The SEIFA IRSD was categorised into quintiles.
Physical activity for adults was defined as low, moderate, or high based on the level of physical activity over the past week, incorporating recreation, sport, transport, and fitness [11
]. The data items that contributed to this variable were total minutes spent walking for transport in the last week; total minutes walked for fitness, recreation, or sport in last week; total minutes undertaken moderate exercise/physical activity in last week; total minutes undertaken vigorous exercise/physical activity in last week. Physical activity was divided into categories and each had an intensity factor score (e.g., walking for fitness = 3.5, walking for transport = 3.5, moderate exercise/physical activity = 5, and vigorous exercise/physical activity = 7.5). The intensity factor score was multiplied by the duration of physical activity. Varying levels of exercise/physical activity were defined as: low (no exercise to <800); moderate (800 to 1600, or more than 1600 but with less than 1-h vigorous physical activity); high (>1600 and with 1 h or more of vigorous physical activity). Although physical activity information was collected for all participants aged >15 years, we did not investigate physical activity in adolescents, as we did not have data for the entire adolescent group of 10–17 year olds.
Health condition was defined as whether a participant had ever had a long-term health condition, defined as a condition that had lasted, or was expected to last, for at least six months. Common long-term health conditions included asthma, arthritis, cancer, heart and circulatory conditions, diabetes mellitus, kidney disease, osteoporosis, mental or behavioural conditions, along with other less common health conditions [11
]. When a participant had a past or present health condition, their health condition was defined as “yes”. For adults, smoking was defined as current smoker, past smoker, or never smoked. Self-assessed health in adults was based on how participants felt about their health and was defined as excellent, very good, good, fair, or poor.
2.4. Statistical Analysis
We reported the survey-weighted prevalence of dietary supplement use by sex and age group. The characteristics of the participants were reported for supplement users and non-users among adults, adolescents, and children. All prevalence data were weighted to the Australian population in 2014/2015 [11
]. Survey-weighted logistic regression models were used to investigate the independent predictors of supplement use in adults (n
= 14,560), adolescents (n
= 1964), and children (n
= 2733). All models were mutually adjusted for all potential predictors. Potential predictors investigated for all participants were sex, State/Territory, and socioeconomic status. For adults and adolescents, region of birth, BMI category, and health condition were also assessed. We additionally investigated age group, education, physical activity, smoking, and self-assessed health as potential predictors of supplement use in adults. The NHS is based on a stratified, multistage area sample of private households. All households were assigned analytic weights to account for their sampling probability to be included in the survey, and the models accounted for the stratification and clustering of the complex sample design using the Taylor Series Linearization method [11
]. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA).