Assessing the Cost of Healthy and Unhealthy Diets: A Systematic Review of Methods

Purpose of Review Poor diets are a leading risk factor for chronic disease globally. Research suggests healthy foods are often harder to access, more expensive, and of a lower quality in rural/remote or low-income/high minority areas. Food pricing studies are frequently undertaken to explore food affordability. We aimed to capture and summarise food environment costing methodologies used in both urban and rural settings. Recent Findings Our systematic review of high-income countries between 2006 and 2021 found 100 relevant food pricing studies. Most were conducted in the USA (n = 47) and Australia (n = 24), predominantly in urban areas (n = 74) and cross-sectional in design (n = 76). All described a data collection methodology, with just over half (n = 57) using a named instrument. The main purpose for studies was to monitor food pricing, predominantly using the ‘food basket’, followed by the Nutrition Environment Measures Survey for Stores (NEMS-S). Comparatively, the Healthy Diets Australian Standardised Affordability and Price (ASAP) instrument supplied data on relative affordability to household incomes. Summary Future research would benefit from a universal instrument reflecting geographic and socio-cultural context and collecting longitudinal data to inform and evaluate initiatives targeting food affordability, availability, and accessibility. Supplementary Information The online version contains supplementary material available at 10.1007/s13668-022-00428-x.


Introduction
Poor diets, described as those low in fruits, vegetables, and whole grains, and high in red and processed meats and ultra-processed foods, are a leading risk factor for chronic disease globally [1]. In most high-income countries (HIC), poor diets disproportionally affect lower socioeconomic populations, Indigenous Peoples, and those living in rural and/or remote areas [2][3][4][5]. Rather than solely a consequence of individual behaviours, poor diets are critically informed by broad contextual factors, including social, commercial, environmental, and cultural influences [6,7]. Crucially, the consumption of a healthy diet is constrained by the range, affordability, and acceptability of foods available for sale [8]. Research suggests that healthy foods are often harder to access, more expensive, and often of a lower quality in rural, remote, or low-income/high minority areas, than in metropolitan or high-income areas [9][10][11][12]. Such food environments contribute to higher rates of diet-related non-communicable diseases and food insecurity [13,14]. In order to improve population diets, all aspects of the food environment must be addressed to ensure healthy foods are affordable, available, and of adequate nutritional quality [15].
Price is a primary factor impacting food choice, diet quality, and food security, therefore having affordable, acceptable, healthy food should be a political and social priority [8,15,16]. Some research suggests that healthy diets are associated with greater total spending [17][18][19], while other studies report that adherence to a healthy diet is less expensive than current or 'unhealthy' diets [9,20,21]. Regardless, the cost of a healthy diet is a proportionately large household expense (> 30% of household income) and may therefore be This article is part of the Topical Collection on Public Health Nutrition * Cherie Russell c.russell@deakin.edu.au 1

3
considered 'unaffordable' [22]. Additionally, public perception that healthy diets are expensive is high, which itself may be a barrier to the purchase of healthy foods [23]. Therefore, improving the affordability of healthy food could improve population diets, regardless of context [24].
To address the issue of food affordability and inform appropriate attenuating policy and intervention strategies, food pricing studies are frequently undertaken. Food pricing, however, is not a universal construct and is highly influenced by country and context. Numerous methods have been developed to measure food pricing, with data therefore not always comparable or replicable, and of limited value to inform appropriate policy [25]. Most studies that collect food pricing data conclude that food prices are rising, making healthy eating unaffordable for many populations. However, few studies to date have used this data to suggest strategies to improve affordability. Our systematic review aims to capture and summarise food environment costing methodologies used in HIC, in both urban and rural settings, between 2006 and 2021. In addressing this aim, we answer the following questions: (i) What is the stated purpose of collecting data on food prices, including whether the data is used to inform or advocate for interventions? (ii) Which instruments are being used to measure food pricing? (iii) What are the strengths and limitations of each instrument as reported by study authors?

Methods
To address the research aim, we undertook a systematic review of the literature, following the Preferred Reported Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [26]. We followed four steps: (i) systematic search for relevant literature; (ii) selection of studies, (iii) data extraction, and (iv) analysis and synthesis of results.

Systematic Search Strategy
After consultation with a research liaison librarian, databases used included EBSCOHOST (Academic Search Complete, CINAHL Complete, GlobalHealth, Medline Complete, and PsychINFO) and Informit. We chose these databases for their comprehensiveness and conventional use in the public health nutrition discipline. We identified search terms using a scoping review and key words used in previous food pricing reviews [15,23,27,28]. We searched both article abstracts and titles using the following search string: 'food affordability' OR 'food cost' OR 'food price*' OR 'food promotion*'. We completed an initial search for studies published 2016-2021 in October 2021, followed by a search for studies published 2006-2015 in December 2021.

Selection of Studies
Studies were included if they were English, peer-reviewed journal articles presenting original research, monitored food prices in a high-income country/s, and were published between 2006 and 2021. The article by Glanz (2006) [15] is considered a seminal paper in food pricing research and was therefore chosen as the starting date for our search. Studies prior to this date were considered unlikely to be relevant to the research question and were thus excluded. Review articles, opinion pieces, posters, perspectives, study protocols, viewpoints, editorials, and commentaries were excluded, as well as those assessing middle-or low-income countries.
Study screening involved an initial review of all titles and removal of duplicates by A1 using online Covidence software [29], followed by abstract screening (A1), and then full text screening of remaining studies (A1). A second reviewer independently screened all articles by abstract and full text to minimise bias (A2 and A3). Disagreements were resolved through discussion between researchers; where no agreement was reached, a third party acted as an arbiter (A2 and A3). Limited hand searching was conducted given the volume of papers identified. Online Resource 1 presents a PRISMA flow chart of the study selection process.

Data Extraction
Included studies were uploaded to an Endnote (V. X9) [30] library. We systematically extracted details of each study to Microsoft Excel (V. 2112), including the author/s, year published, article title, aim, pricing instrument used (if specified), country and geographical context (e.g. urban or rural), type of data collected, number and type of locations assessed, number and type of food items captured, population (if the study used sales receipts to estimate food prices), time period of study, strengths, limitations, and conclusions.

Data Analysis and Synthesis
The coded data were used to identify major themes that were then synthesised in the results. We used an inductive thematic approach for our analysis, with the results discussed between the research team to limit researcher subjectivity [31]. We used Microsoft Excel to calculate descriptive statistics and graphical outputs.

Overview of Studies
Database searching identified 2737 studies, with 1882 studies remaining after removal of duplicates. After abstract screening, a total of 287 were identified for full-text screening, with 187 excluded, and a total of 100 studies included in this systematic review (Online Resource 1).
We observed an increasing number of studies each year, with peaks in 2013, 2014, and 2018 ( Fig. 1).
Most studies measured food prices in the USA (n = 47), followed by Australia (n = 25). Urban food environments were assessed more frequently (n = 74) than rural (n = 33). Most studies were cross-sectional (n = 77). Most studies included instore price audits (n = 59), followed by online price audits (supermarket websites, n = 13), or electronic point of sale data (consumer receipts, register sales, or electronic scanning of food prices in the home, n = 12), and a combination of these (n = 17). Most studies collected food price data from more than 20 food retail outlets (n = 34) ( Table 1).
Details of all included studies, grouped according to data source used (instore price audits, online price audits, electronic point of sale, and combinations of these), are shown in Tables 2, 3, and 4. Details include instrument used (if applicable), purpose of data collection, country, context, study type (e.g. cross-sectional, longitudinal), healthiness comparisons (between healthy and unhealthy products or diets), study author, and year. The use of a named instrument was captured to identify commonalities in usage of instruments, and not as an indication of study quality. When assessing differentials in 'healthiness', studies either presented a comparison of a 'healthy diet' with an 'unhealthy or currently consumed diet' or a comparison of the cost of 'healthy' and 'unhealthy' foods or product categories.

Overview of Instruments Used to Measure Food Prices
Of the 100 included studies, 57 used a named instrument to measure food prices, as described below. The remaining 43 studies did not name a pre-existing data collection instrument; instead, the authors described the data collection methodology used, for example, in store, online, or via electronic sales data.

Healthy Diets Australian Standardised Affordability and Price (ASAP) Instrument
Following critiques of existing food baskets, the previously described INFORMAS instrument was refined to  [121] assess and compare the price and affordability of healthy and current diets in Australia, leading to the development of the Healthy Diets Australian Standardised Affordability and Price (ASAP). This instrument assesses the cost of a 'recommended' Australian diet (defined by the Australian Dietary Guidelines and Australian Guide to Healthy Eating) and the cost of the 'current' Australian diet (as reported in the 2011-12 Australian Health Survey) using the reference household of two parents and two children (boy aged 14 years; girl aged 8 years) [128]. Thus, all studies using this instrument present a comparison of the cost of a 'healthy' and 'unhealthy' diet in their analysis. Intrinsic to the instrument, the relative affordability of a healthy diet is measured against household incomes. The ASAP instrument was used by four studies to collect food price data in physical instore locations [9,85,86] or from online supermarkets -Compare prices of healthy and unhealthy foods [62] -Measure price of healthy food over time [63] -Assess food costs in an area [64] USA -3 Urban -3  [94]. Studies were conducted in both rural (n = 2) [9,85,94] and urban (n = 2) [85,86,94] contexts.

Other Instruments
Several other named instruments were identified, used in single studies. These included the Diet and Nutrition Tool for Evaluation (DANTE) [101], the Flint Store Food Assessment Instrument [60], the Food Label Trial registry tool [76], the New Zealand Food Price Index [111], the USDA Food Store Survey Instrument [73], USDA Low-cost food plan [55] and audit forms developed by the Yale Rudd Center [39], the Hartford Advisory Commission on Food Policy [59], and the USDA Authorized Food Retailers' Characteristics and Access Study [43]. Only three instruments compared healthy and unhealthy products [43,76,111] and none analysed the relative affordability of food.

Instrument Strengths and Limitations
The strengths and limitations of instruments commonly used across studies, as identified by study authors, are presented in Online Resource 2. Commonly cited limitations, regardless of instrument used, included that actual purchasing behaviours were not captured (unless electronic point of sales data was utilised); culturally important and region-specific products were often not captured; tools were cross-sectional in nature, thus seasonality or changes overtime were not considered; and out-shopping, described as food purchases undertaken outside the local residential geography, including internet orders or foods purchased during travel to other communities, could not be accounted for. While some food basket studies and those using the ASAP instrument did contextualise the relative affordability of healthy foods and/or diets, this was not a part of the methodology for NEMS-S. Other limitations specific to NEMS-S included the length of the survey, and a low convergence between NEMS-S results and consumer perceptions of affordability. Specific limitations for food basket studies included results being constrained by the reference family used and the assumption that food is shared equally among household members. Additionally, most instruments did not capture geographical information regarding access to food retail outlets or availability of foods within food retail outlets. Authors less commonly described instrument strengths. For NEMS-S, cited strengths included the ability to compare food prices between healthy and unhealthy options, that it has strong inter-rater and test-re-test reliability, and that it has been validated in multiple countries. ASAP studies, and some food basket studies, included a comparison between healthy and current ('unhealthy') diets (based on actual consumption) and included alcohol in the survey.

Discussion
Our systematic review details the key purposes, and methodologies used, for measuring food prices in HIC between 2006 and 2021. While most studies were conducted solely to monitor food prices in specific locations, some sought to report price changes over time, and others collected data to assess comparability of food costs to healthier alternatives, average earnings, welfare payments, rurality, and socioeconomic position. Most studies measured food prices in urban areas, using instore food price audits, with an emerging use of online data collection evident. The most frequently used instruments were 'food baskets', used predominantly to monitor food prices; the NEMS-S instrument, used to provide data on relative cost and availability; and the ASAP instrument, use to provide data on relative affordability.
Our review differs from previous reviews of food price and affordability instruments [23,28] by taking a broadened focus on food pricing measures used in HIC globally and including new technology that is affording opportunities for electronic food pricing data collection. While a previous review critiqued food pricing measures for relevance specific to a rural context, our review includes both rural and urban contexts [28]. Another review [23] also describes the components of individual instruments, such as the identification of differently sized 'food baskets', ranging between 30 and 200 food items. Such critique was beyond the scope of our research questions.
Despite emerging options for electronic methodologies, the predominance of in person, instore data collection continues, notwithstanding the time-consuming and resourceintensive nature of this method. Studies indicate that these instore instruments can be targeted and applied within multiple contexts, such as rural [9][10][11][12], Indigenous [129,130], and low socioeconomic areas [85]. Perhaps researchers consider instore data collection as providing real-world insights at a community and population health level. Our review identified that food pricing instruments were mostly used to monitor food prices at a single point in time (crosssectional) rather than changes at different time points (longitudinal). Instruments that enable the comparison of food prices in terms of a healthy diet (as recommended by dietary guidelines) compared with current dietary patterns (as reported through population health surveys) [128], and relative affordability for families, appear to provide data of greater practice and policy relevance with regard to community strategies, taxes, and subsidies that have potential to enhance food affordability, availability, and accessibility.
Technological innovations are an emerging alternative to in person data collection, facilitating the acquisition of online supermarket prices, a less labour-intensive method for capturing food prices [131]. To date, this method has been used within major chain-supermarkets, with a recent study reporting similar results when comparing pricing data obtained instore versus online [94]. This method therefore holds potential where an online supermarket presence exists, which was increasingly the case during the COVID-19 pandemic [53], providing rapid feedback to inform price promotions. However, for smaller and/or independent food retail outlets, frequently located in rural areas, online data collection does not appear to capture the contextual nuances of instore price promotions.
Our review found an over-representation of food pricing studies within urban areas. This is consistent with multiple studies that reflect inequities experienced within rural environments [132], and rural food environments are no exception [133]. The predominance of research within urban areas may also reflect a pragmatic researcher response to the physical proximity of stores (ease of measurement) and larger population reach (potential for greater population impact). Previous research shows significant differences in incomebased variables, food environments, and the affordability of healthy food between urban and rural settings [134]. There is therefore a need for rural-specific food pricing studies, using appropriate instruments, to evaluate and inform ruralspecific food environment initiatives [28].
During the period covered by this review, high level experts from the World Health Organization [135], the Lancet Commission [136], and the Food and Agricultural Organisation of the United Nations [137] have identified the potential benefits that initiatives located within food retail environments can provide in nudging dietary choices towards healthier options through instore food pricing and promotion, with the overall aim of improving population level diets [14]. Measures of food pricing, and the relative affordability of a healthy diet, are important to both inform and measure the effectiveness of such initiatives. However, few studies in our review explicitly aimed to inform initiatives or strategies, either at the community or policy level. Assessment of author-reported strengths and limitations of food pricing instruments and methodologies also identified a need for a universal instrument that reflects contextual geographic and socio-cultural information; is intended to be used repeatedly over time; and is adaptable to different country/cultural/contextual settings [17,23]. Future research would benefit from linking the purpose of undertaking food pricing data collection more explicitly to potential initiatives. Our review supports this call and suggests that the instrument selected should suit the context and collect longitudinal data to provide greater insights into the design and effectiveness of initiatives that make healthy food not only affordable but also available and accessible.

Strengths and Limitations
This systematic review provides a current and comprehensive overview of international food pricing studies across HIC. We acknowledge that while food prices are an important factor influencing food choice, it is only one component of the food environment; however, analysing instruments that assess food acceptability, availability, and accessibility was beyond the scope of this review. This review focused on HIC and a similar review on food pricing studies in low-and middle-income countries would be informative. This review may have missed additional relevant data as it only included English language studies and did not include grey literature or hand searching of reference lists.

Conclusion
Food security has come under heightened scrutiny given the food supply interruptions experienced worldwide during the COVID-19 pandemic. While studies providing a snapshot of food prices can be useful to identify areas impacted by rising food prices, much of this cross-sectional data is known. This review raises questions regarding the purpose of collecting food price data, and how this data can best be used to inform change through practice and policy strategies. We suggest that longitudinal studies using a consistent methodology, which acknowledges contextual nuances and demonstrates temporal changes in food pricing, are needed to inform and to evaluate community-based or legislative strategies to improve the relative affordability of a healthy diet.

Competing Interests
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.