Search strategy

The search strategy was structured to identify any studies that compared the types and amounts of food and drinks habitually consumed by low SEG with those consumed by higher SEG. It was anticipated that any identified studies would be heterogeneous⁽²⁾. Therefore, the study design was more consistent with a scoping review than a comprehensive systematic review, and the search strategy was informed by the methodology for Joanna Briggs Institute Scoping Reviews Methodology⁽²²⁾. The research question was considered in population, intervention, comparator, outcome and time format and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses⁽²³⁾ statement was used to guide review processes.

The search population included Australians of any age and gender categorised as belonging to a low SEG by any method.

The intervention was defined as a study which captured the dietary intake of the population of interest. The comparator was the dietary intake of Australians of any age or gender categorised as belonging to a higher SEG than the search population.

Outcomes were defined as types and amounts of dietary intake, either as an assessment of the full diet, or selected food groups, or as foods or drinks.

The search timeframe was restricted to documents published from January 1999 to September 2019, as it was considered that earlier documents may lack relevance due to changes in the social and food industry landscape and dietary patterns of Australians over the last 20–25 years⁽⁴⁾.

Databases of peer-reviewed literature and targeted websites were searched, and all included references were also hand-searched for any missing relevant documents. All stages of the search and data extraction process were conducted by M.L., with 10 % of abstracts and data extractions cross checked by A.L. to control for inter-observer bias.

The peer-reviewed literature databases searched were The Cochrane Library, PubMed, MEDLINE, EMBASE, CINAHL, Informit Health Collection and Web of Science (Science Citation Index and Conference Proceedings Citation Index).

The search terms used were (nutrition OR diet OR diets OR food OR foods OR drinks) AND (consum* OR purchas* OR buy OR intake) AND (low-income OR low income OR low socioeconomic) AND Australia. Low income was used as a proxy for a low SEG as it is used frequently in research relating to the affordability of healthy diets^(24–26). Other potential search terms as proxies for low SEG, such as ‘education level’ and ‘deprivation’, were tested but found to lack specificity to SEG and did not lead to the identification of any additional studies. Location was restricted to Australia, as dietary intake and socio-economic structures are influenced by many country-specific social and economic issues; thus, overseas findings would potentially be less applicable to the Australian situation. Food purchasing terms were included in the search as a potential proxy for dietary intake.

The listed databases were searched and resulting citations were downloaded into EndNote X8⁽²⁷⁾. Duplicates were removed, and the following inclusion and exclusion criteria were systemically applied to screen remaining citations, based on title, then abstract, then full text.

The websites searched were The Australian Prevention Partnership Centre/The Sax Institute, Australian Health Policy Collaboration, Public Health Association of Australia, National Health and Medical Research Council, National Preventive Health Agency, Commonwealth Health Department, Australian Institute of Health and Welfare and the Grattan Institute. Search terms ((Diet OR nutrition) AND (socio-economic OR income)) were systematically entered into each website-specific search engine. The first five page returns or the first ten items listed (when sorted by relevance) from each search were scrutinised.