Mapping the extent of unhealthy food advertising around schools in Tāmaki Makaurau/Auckland

ABSTRACT

Advertisements promoting energy-dense, nutrient-poor food and drinks influence children’s dietary choice and preference and are pervasive in children’s environments across Aotearoa, New Zealand. Using a combination of datasets from separate studies, this research seeks to describe the extent of children’s unhealthy food/drink advertising exposure in all primary and intermediate school food zones (n = 381) in Tāmaki Makaurau, Auckland, and assess whether this exposure correlates with indicators of neighbourhood deprivation. To achieve this, we developed a simple advertising exposure score using images of advertising collected from Google Street View, and investigated its spatial correlation with indicators of neighbourhood deprivation. Outdoor advertising was present in proximity to 61.9% of the sampled schools (n = 236); of which 83.1% of adverts promoted unhealthy food and/or drinks (n = 2669). Advertising exposure increased statistically significantly with neighbourhood deprivation (p < 0.05). The results show a clear need for policy intervention that regulates and limits unhealthy food/drink advertising around schools in New Zealand. The advertising score proved a useful, easy to understand tool to measure and visualise children’s potential exposure to unhealthy food/drink advertising. The score may be used to increase community awareness of the ubiquity of unhealthy advertising in children’s neighbourhoods and advocate for policy change.

Practitioner points

• Advertising present in the vicinity of 62% of Auckland schools; 83% of adverts promoted unhealthy food and/or drinks

• Novel advertising score assesses the spatial distribution of unhealthy advertising exposure.

• Advertising exposure score is significantly related to neighbourhood deprivation.

• Findings show the need for policy action to limit children’s advertising exposure around their schools in NZ.

KEYWORDS:

• Unhealthy food and beverages
• marketing to children
• outdoor advertising
• school zones
• area deprivation

Introduction

Advertisements promoting energy-dense, nutrient-poor food and drinks influence children’s dietary choice and preference (Cairns et al. 2013). Unhealthy food and beverage marketing has become almost ubiquitous in Aotearoa New Zealand (NZ) with children exposed to advertising online, on TV, in stores, at school and outdoors, as well as being exposed to unhealthy brand marketing through channels like sports sponsorship (Vandevijvere and Swinburn 2015; Garton et al. 2022). Children are especially attractive targets for food companies, because they are both current and future customers, younger children have limited ability to discern marketing from fact, and children’s dietary preferences shape their household’s purchases (Boyland and Whalen 2015; Tatlow-Golden et al. 2021). Children’s exposure to unhealthy food and beverage marketing, including but not limited to child-persuasive techniques, is also recognised as a breach of children’s rights (World Health Organization (WHO) 2016; Garde et al. 2018; Keeley et al. 2019; Tatlow-Golden and Garde 2020, 2021).

The urgent need to limit children’s exposure to unhealthy food and drink marketing is widely recognised among both researchers and policymakers. The WHO Set of Recommendations on the Marketing of Foods and non-alcoholic beverages (WHO 2010, 2016) received unanimous support from the World Health Assembly, though global implementation has been limited (Boyland and Whalen 2015; Tatlow-Golden et al. 2021). Several national restrictions or codes on harmful product marketing to children are in place; however, these are usually voluntary and/or self-regulatory, and not linked to an enforcement mechanism. In NZ for example, the Children and Young People’s Advertising Code is self-regulated by the Advertising Standards Authority (ASA) and critical analysis has shown it to be ineffective in protecting children from exposure to unhealthy marketing (Sing et al. 2020).

This study focuses on outdoor advertising around schools, as a key setting for children’s exposure. Internationally, unhealthy food outlets and unhealthy food/drink advertising are often present in close proximity to schools, particularly in urban areas, and unhealthy food and drinks are advertised more frequently than healthy options (Finlay et al. 2022; França et al. 2022). The food environment around schools is of particular concern since this is where children spend much of their time (França et al. 2022). In particular, transit to/from school is an important space for social interactions and children report that this is when they buy food (usually unhealthy snacks) with their friends (Egli et al. 2020). Thus, the availability of unhealthy food/drink alongside the presence of unhealthy food/drink advertising around schools have both a direct impact on children’s dietary behaviour by influencing their consumption patterns, and an indirect impact by shaping their dietary norms and preferences (Cairns et al. 2013).

Previously, Vandevijvere et al. (2018) examined the presence of unhealthy food/drink advertising around NZ schools and found almost two thirds (63%) of the products advertised were classified as ‘non-permitted’ to be marketed to children according to the WHO nutrient profile model for Europe. Similarly, the KidsCam study, which analysed photographs taken every seven seconds from cameras worn by children as they walked or played in their neighbourhoods in Wellington NZ, found that the exposure to marketing for ‘non-core’ (i.e. ‘unhealthy’) foods was twice as high as for ‘core’ (‘healthy’) foods (27.3 vs 12.3 times/day) (Signal et al. 2017). On average, children were exposed to 7.4 unhealthy food/drink advertisements for every hour spent in outdoor public spaces (Liu et al. 2020). Furthermore, half of the food/drink marketing exposures occurred in public spaces, with most exposures on the school journey (Signal et al. 2017; Liu et al. 2020).

In recent years, several studies have examined outdoor food advertising around schools in Auckland specifically (Egli et al. 2019; Huang et al. 2020; Brien et al. 2023). Being NZ's largest city and home to almost a third of the country’s population, Auckland provides an interesting case study. The city is characterised by rich ethnic diversity (Statistics NZ 2020), as well as large socioeconomic disparity, with some of the richest and poorest residents of NZ living in close proximity (Exeter et al. 2020).

In NZ, children living in the most socioeconomically deprived areas are 2.5 times as likely to have obesity as children living in the least deprived areas (Ministry of Health NZ 2021), and ‘food swamps’ have been one proposed explanation for this disparity (Swinburn et al. 2011). A positive relation between higher densities of food retailers and an area’s level of socioeconomic deprivation has been consistently recorded in high-income countries (Jebeile et al. 2022), and for NZ specifically (Pearce et al. 2007; Sushil et al. 2017), however this included all types of retailers – not just unhealthy outlets. One NZ study did observe that the density of convenience stores was significantly higher around urban schools with the highest proportion of students from socioeconomically deprived areas (Vandevijvere et al. 2016). Another possible contributor is that children’s exposure to unhealthy outdoor food/drink advertising is not equally distributed. Previous scholarship has consistently found unhealthy outdoor advertising to be more present around schools with the highest number of students from a low socioeconomic background (Vandevijvere et al. 2018; Huang et al. 2020; Trapp et al. 2022; Brien et al. 2023).

A first study using Google Street View (GSV) to capture outdoor advertising in school areas in NZ (including billboards, signs, flags, banners, neon signs, stickers, and bus shelter advertisements) identified more than three times more unhealthy than healthy outdoor food and drink advertisements around Auckland schools (Egli et al. 2019). Subsequent studies using GSV for data collection have focused on bus shelters and the fronts of convenience stores as key advertising spaces. Huang et al. (2020) observed that 12.8% of advertised products on bus stops within 500 m walking distance from primary and intermediate schools were ‘non-core’ (i.e. unhealthy) foods or drinks. Similarly, Brien et al. (2023) found that half of all advertisements on convenience stores within walking distance of primary schools were for unhealthy products (50.5%, only 9.4% for healthy food/drink products). The majority (78.2%) of these advertisements were targeted towards children (Brien et al. 2023).

However, such NZ data has never been spatially visualised using Geographic Information Systems, with the power to identify spatial patterns with other neighbourhood characteristics. Furthermore, the data has not been translated into a format that can be easily understood by lay persons.

Aims and objectives

In this study, we combined and supplemented existing data on unhealthy outdoor advertising around Auckland’s schools collected with GSV (Huang et al. 2020; Brien et al. 2023), to answer the following research questions:

• What is the extent of children’s unhealthy food/drink advertising exposure in their school food zone in Tāmaki Makaurau, Auckland?

• Does children’s exposure to unhealthy outdoor food/drink advertising correlate with indicators of neighbourhood deprivation?

To follow this line of inquiry, we used spatial and statistical analytical approaches. A secondary objective was to develop a simple advertising exposure score, which could also be applied in other settings. The purpose of the advertising exposure score was to condense the data from various sources of advertising exposure around schools into an aggregate measure that was easy to understand and interpret, and also to allow the team to map its spatial distribution across the study area.

Materials & methods

We used the Quantum Geographic Information System (QGIS) software in this study to visualise and investigate the spatial distribution of children’s exposure to unhealthy outdoor food/drink advertising within close proximity (500 m) to schools, which this study refers to as ‘school food zone’ (see Box 1), in the Auckland Region. This enabled a more comprehensive assessment of the outdoor food/drink marketing environment by combining various data sources. In a second step, this exposure score was correlated with the level of deprivation of each school’s neighbourhood environment.

Study area

The study area includes the combined areas of the Auckland, Waitematā, and Counties Manukau District Health Boards (Figure 1). This covers an area of 6556 km² and is home to 1,695,200 inhabitants, or 33% of the total NZ population (Statistics NZ 2022). There are large differences in population density across Auckland, with over 90% of the population located in the region’s main urban area, a classification for the area around a city or major urban area with a minimum of 30,000 inhabitants (Exeter et al. 2016).

Figure 1. Location and school roll (population) of primary and intermediate schools within the entire study area (A) and central Auckland (B). The category ‘Rural Area’ includes schools located in rural area, rural centres, minor urban, and secondary urban areas (Ministry of Education n.d.). Base Map: Statistics NZ (2017, 2021a).

The NZ Ministry of Education uses an analogue classification for the schools included in this study according to their level of urbanicity and differentiates between major urban, secondary urban, minor urban, rural centres, and rural area (Ministry of Education n.d.). To account for structural differences between the densely populated central Auckland and the more rural peripheral region, we combined the four area types with a population of less than 30,000 inhabitants into the single category ‘rural area’. 84.5% of the sampled schools are located in a main urban area (see Figure 1).

We used the address of each school as an estimate of its main entrance and drew a 500 m road network buffer around it to measure advertising catchments. These buffers correspond to the area around each school which can be reached by walking for 500 m. The total area covered by the 500 m road network buffers around the sampled schools is 107 km² (of which 94.5 km² are in the main urban area) (Supplementary material, Figure C3).

Box 1. Definition of key concepts used in this study.

Table

Key Concepts In the context of this study, children are defined as those attending primary education at one of Auckland’s primary (elementary school in the US) (‘full primary’: years 1–8; ‘contributing': years 1–6) or intermediate schools (middle school in the US) (years 7–8) (Ministry of Education 2022a). All 381 primary (full or contributing) and intermediate schools for which both Brien et al. (2023) and Huang et al. (2020) collected data were included in the present study. These are all classified as either state or state-integrated. The school food zone refers to the area around each school that is within 500 m walking distance of the school’s main entrance (Huang et al. 2020; Brien et al. 2023). This is the buffer zone size used most frequently in research on the presence of outdoor advertising around schools (Finlay et al. 2022). The type of food and drink advertising considered is limited to outdoor advertising on the fronts and premises of convenience stores, and on bus shelters. Unhealthy food and drink advertising is defined as those promoting products which are not recommended for marketing to children according to the WHO Regional Office for Europe Nutrient Profile Model (WHO 2015), in line with definitions applied by previous studies (Huang et al. 2020; Brien et al. 2023), however other previous NZ studies have used the terms ‘non-core’ and ‘core’ to define ‘unhealthy’ and ‘healthy’ foods. In the context of this study, the term ‘food/drink advertisement’ refers to advertisements of food items and/or non-alcoholic beverages. Exposure to unhealthy outdoor food/drink advertising is defined as the presence of a convenience store or bus stop displaying unhealthy food/drink marketing within the school food zone (Huang et al. 2020; Brien et al. 2023). Neighbourhood deprivation is multidimensional, comprising both social and material aspects; its level and experience vary across areas and population groups (Exeter et al. 2017). In NZ, deprivation at the school level was measured by school deciles until 2023. The Ministry of Education provided this measure for each school, based on the percentage of students living in an area of low socioeconomic status. The higher the decile of a school, the lower the proportion of students from a socioeconomically deprived area (Ministry of Education 2022b). School deciles for 2019 for the sampled schools are available from the school statistics of the New Zealand Ministry of Education (Ministry of Education 2022c). In January 2023, the newly introduced Equity Index (EQI) replaced the school decile as the preferred measure of socioeconomic disadvantage at school level (Ministry of Education 2023). Each school is assigned a numeric value between 344 and 569 points, based on 37 variables selected to measure the socio-economic circumstances of its student body (as measured in the previous three years). The higher the EQI, the more barriers to educational achievement the students of a school are expected to face (Ministry of Education n.d.). Given its relevance to policymakers, the EQI serves as the main indicator of deprivation in the present study. Alternatively, the NZ Index of Multiple Deprivation 2018 (IMD18) combines several domains of deprivation beyond education into an index which returns the level of deprivation at the neighbourhood level (Exeter et al. 2017). The IMD18 decile of each school’s neighbourhood was included as an alternative indicator for neighbourhood deprivation (Exeter et al. 2020).

Advertisement data

Original advertisement data

This study is based on outdoor food/drink advertising data collected on bus stops (Huang et al. 2020) and convenience stores (Brien et al. 2023) in the Auckland region, and supplemented with additional data collection to fill any gaps (e.g. convenience stores around intermediate schools not included in Brien et al. 2023). Huang et al. (2020) and Brien et al. (2023) used Google Street View (GSV) to record the number and type of outdoor food and drink advertisements within 500 m walking distance of schools in Auckland (Table 1). Each advertisement was classified as either food/drink or non-food/drink, with food/drink advertisements further classified as healthy or unhealthy according to specified criteria (see Box 1).

Table 1. Overview of the original data sets.

Dataset; Publication	Study period (GSV image capture dates)	Data structure
Bus Stops; Huang et al. (2020)	August 2019 to February 2020 (February 2012 to May 2019)	573 schools; 512 bus stops; 842 advertisements
Convenience Stores - Primary Schools; Brien et al. (2023)	May 2020 to February 2021 (February 2012 to September 2020)	371 schools; 296 stores; 3693 advertisements
Convenience Stores - Intermediate and Additional Primary Schools; unpublished	September 2021 to March 2023 (February 2012 to January 2023)	62 schools; 114 stores; 1423 advertisements

Combination of data sets

We used the R Statistical Software (R Core Team 2019) for data cleaning and manipulation. The combined advertisement data were aggregated by school, and each school was assigned the total number of advertisements, food/drink advertisements, unhealthy food/drink advertisements, convenience stores and bus stops located within 500 m walking distance (Supplementary material, Appendix A.3). Appendix A1 and A2 (Supplementary material) describe several issues and discrepancies in the original records during the data aggregation process.

Advertising score

To investigate spatial differences in the exposure to unhealthy outdoor food/drink advertising of school children across Auckland, we constructed a composite index using information about the occurrence of unhealthy food/drink advertisements and the number of all advertisements as two indicators:

(i)	Proportion of unhealthy food/drink advertisements (A)

The proportion of unhealthy food/drink advertisements around each school was defined relative to the number of food/drink advertisements: $\mathit{A}={\displaystyle \frac{\mathrm{number}\mathrm{of}\mathrm{unhealthy}\mathrm{food}/\mathrm{drink}\mathrm{advertisements}}{\mathrm{number}\mathrm{of}\mathrm{food}/\mathrm{drink}\mathrm{advertisements}}}$This relative measure of the presence of unhealthy outdoor food/drink advertising was used rather than the total number of unhealthy food/drink advertisements, because it contains additional information on the dietary choices that are encouraged in the area through the display and promotion of unhealthy food/drinks which encourages and normalises their consumption (Cairns et al. 2013).

Defining the presence of unhealthy food/drink advertisements as proportion of food/drink advertisements rather than total advertisements, is in line with other relative measures used in the assessment of food environments. The Retail Food Environment Index (RFEI), for example, is the relative density of unhealthy to healthy food outlets (Zhang and Huang 2018).

(ii)	Number of total advertisements (B)

This indicator is a measure of the exposure to advertisements in general. It is important to know where children are most exposed to advertising, since all advertisement, regardless of the promoted product, is potentially harmful to children (Powell 2020).

Both the proportion of unhealthy food/drink advertising (A) and the number of total advertisements (B) were min–max normalised and combined into an equally-weighted, composite index for each school: $\mathrm{Advertising}\mathrm{Score}={\displaystyle \frac{A+B}{2}}$The Advertising Score returns a value between 0 and 1 for each school, with higher values indicating a higher level of exposure to unhealthy food/drink advertisements and outdoor advertising in general.

We considered including school roll as a weight to estimate the potential impact of the presence of unhealthy food/drink advertising for each school neighbourhood environment. However, school roll appears to be significantly higher in the main urban area (see Figure 1, also section ‘School characteristics’). Therefore, school roll is not part of the advertising score but is referred to when assessing the extent and spatial distribution of unhealthy food/drink advertising across Auckland, and the impact of potential exposure.

Spatial analysis

The mapping and analysis of the data was completed using the open-source software QGIS Desktop 3.16.4 (QGIS Development Team 2021). Spatial data layers were projected to the coordinate system NZGD2000 / New Zealand Transverse Mercator 2000 (EPSG: 2193) and clipped to the extent of the study area.

Geocoding

The address of each school was geocoded using the MMQGIS plugin (v2021.9.10; Minn 2021), and a Google Maps Geocoding API key. The information entered included school name, street number, street name, postcode, and city. The resulting point shapefile was checked for accuracy and manually corrected by referring to an OpenStreetMap basemap (OpenStreetMap 2023).

Neighbourhood deprivation data

Three indicators for the level of neighbourhood deprivation were explored in this study (see Box 1). While school deciles and the Equity Index (EQI) are defined at the school level, IMD18 data is only available at the neighbourhood level. Therefore, each school was assigned the IMD18 overall deprivation decile of its neighbourhood using a spatial join. Given that the EQI is the most current measure of socioeconomic deprivation at the school level and most relevant to policymakers in Auckland, it served as the main indicator for the analysis (Box 1).

Spatial distribution of advertising exposure and deprivation

Initially, we visually assessed the spatial distribution of advertising exposure according to school location and investigated potential variations by school size. In a second step, kernel density estimation maps were created to detect local peaks and patterns in the extent of advertising exposure. Kernel density estimation results were displayed as raster maps where each cell (10 m cell size) was assigned the cumulative advertising score or EQI of all schools within a 1250 m radius. The same was done for IMD18 decile scores, based on a centroid layer of the 2018 data zones (Exeter et al. 2020). The resulting heat maps allowed a comparison of the spatial distribution of advertising exposure to that of socioeconomic status and neighbourhood deprivation in Auckland.

Statistical analysis

All statistical analyses were performed using R Statistical Software (v3.6.1; R Core Team 2019). Predicted probabilities were calculated using the – arm – R package (v1.12.2; Gelman and Su 2021).

Bivariate and multivariate linear regression models were used to assess the relation between the advertising score and the level of deprivation at the school level. A categorical variable to differentiate between schools in Auckland’s main urban and more rural areas was included as covariate (see section ‘Study area’), as well as the number of advertising spaces (convenience stores and bus stops) around each school.

Additionally, a two-part model was run to account for the positive skew and high number of zero values included in the data (due to the absence of convenience stores and bus stops in several school zones) (Boulton and Williford 2018). A binary variable differentiates the sample into schools with an advertising score of either zero or above zero. First, a logistic regression model with this binary variable as the outcome variable was used to assess which factors impact the presence or absence of unhealthy outdoor food/drink advertising. Then, linear regression models were run, including only those schools with an advertising score larger than zero. Comparing these to the linear regression model that includes all sampled schools allowed an assessment of how the abundance of zero values impacted the regression results.

Results

School characteristics

Overall, advertising data from 381 school food zones was collected, of which 53.3% (n = 203) are contributing, 36.0% (n = 137) full primary, and 10.8% (n = 41) intermediate schools. School roll ranged between 2 and 1238 students, with a mean of 402 students. Eighty-four percent (84.5%, n = 322) of the sampled schools were located in a main urban area. Most rural schools were small, with 400 or fewer students enrolled (Figure 1). On average, the number of students enrolled in schools within the main urban area was double that of students enrolled in the more rural areas (mean roll 438 vs 202 students, p < 0.001). The mean EQI of the sampled schools is 438.3, with values covering almost the entire range of possible scores (ranging from 344 to 559 index points) (Appendix B, Figure B1).

Advertising exposure

A total number of 5947 advertisements were recorded around the sampled schools. This included 3212 food/drink advertisements, of which 83.1% (n = 2669) were for unhealthy food and/or drinks. The mean number of advertisements within 500 m of a school was 15.6 for total advertisements and 7.0 for unhealthy food/drink advertisements, with a maximum number of 236 total and 95 unhealthy food/drink advertisements counted for a single school. There were no advertisements in 145 school zones (38.1%).

The advertising score for the sampled schools ranged between 0 and 0.91 index points. The sample included a high number of zero values. However, the scores above zero were approximately normally distributed (Supplementary material, Figure B2). Most schools in Auckland’s more rural areas received low advertising scores (mainly < 0.4). Higher values were found almost exclusively in central Auckland (Supplementary material, Figure C1).

Kernel density maps helped to visualise the general distribution of the advertising score across central Auckland. Darker areas indicate a higher concentration of schools with high advertising scores, while lightly coloured areas indicate the presence of schools with generally lower advertising scores. Schools with high advertising scores were concentrated in areas within the centre of Auckland (in the suburbs Mount Eden, Mount Roskill, Avondale/New Lynn), north of the Waitematā Harbour (Belmont, Birkdale), west of the Waitematā Harbour (Henderson), southwest of the Manukau Harbour (Māngere), east of the Manukau Harbour (Ōtāhuhu), and east of the Tāmaki River (Howick) (Figure 2A).

Figure 2. Kernel density maps of the advertising score (A), school Equity Index (B), and Index of Multiple Deprivation (C) in central Auckland. Base Map: Statistics NZ (2017, 2021b).

There were two large areas in South Auckland (Papatoetoe, Manurewa) where advertising scores peaked as well, although to a lesser degree. However, these areas of higher advertising exposure coincided with a concentration of high EQI scores (Figure 2B). Furthermore, these schools were located in areas with more deprived neighbourhoods, as indicated by high IMD18 deciles (Figure 2C). This was also the case for the area with high advertising scores in Māngere.

In contrast, two peaks of high advertising scores in the centre of Auckland (Mount Eden, Mount Roskill) coincided with peaks in the EQI, but not of neighbourhood deprivation. The advertising score peaks in the north, east, and western centre did not coincide with visible peaks of either EQI scores or IMD18 deciles.

Regression analysis

The advertising score increased with the level of deprivation, as measured by the school EQI. On average, a school with the highest possible EQI is expected to have an advertising score that is 0.074 index points higher than a school with the lowest possible EQI located in the same type of area (urban/rural) and with the same number of advertising spaces around it (Table 2). Thus, socioeconomic disadvantage was a significant predictor of advertising exposure. This difference equals 8.1% of the range of advertising scores present in the school sample. The scale of the increase was consistent across models and alternative measures for deprivation (i.e. school decile and IMD18 decile) (Appendix B.2, Table B1).

Table 2. Regression coefficients and intercepts for linear regression of the advertising score with Equity Index and covariates.

	Model 1 (all)	Model 2 (all)	Model 3 (all)	Model 4 (Adv Score > 0)	Model 5 (Adv Score > 0)
Intercept	0.181 (0.099)	−0.04 (0.102)	−0.096 (0.078)	0.297 (0.086) ***	0.226 (0.094) *
EQI Score	0.0001 (0.0002)	0.0003 (0.0002)	0.00033 (0.00016) *	0.0003 (0.0002).	0.0004 (0.0002) *
Urbanity	–	0.192 (0.034) ***	0.086 (0.027) **	–	−0.042 (0.048)
Adv. Spaces	–	–	0.074 (0.004) ***	–	0.028 (0.004) ***
Adjusted R²	−0.002	0.073	0.465	0.0091	0.1619

Notes: Standard errors given in brackets. Statistical significance coded: < 0.001 ‘***’, < 0.01 ‘**’, < 0.05 ‘*’, < 0.1 ‘.’

The level of urbanicity and the number of advertising spaces were statistically significant predictors of the advertising score (Table 2). Schools located in the main urban area of Auckland generally received higher advertising scores than schools located in more rural areas. For each additional advertising space (e.g. store or bus stop) within 500 m of a school, the advertising score increased by 0.074 index points (Table 2). Including urbanity and the number of advertising spaces as covariates improved model fit significantly (Appendix B.4, Figures B3 and B4).

The model demonstrated that a school’s location within or outside a main urban area had a large impact on the presence or absence of (unhealthy) outdoor food/drink advertising in its vicinity. The probability for a school in Auckland’s main urban area to have an advertising score of zero was 39.8%, which was significantly lower than for a school in a more rural area (83.1%; p < 0.001) (Appendix B.3, Table B2).

When including only those schools with an advertising score above zero, the model still predicted a statistically significant difference of 0.09 index points between schools with the highest and lowest possible EQI scores (Table 2). The explanatory power of the models decreased (Adjusted R² 0.162 vs 0.465), however, the residuals appeared smaller and less scattered compared to those of the models including the entire sample (Appendix B.4, Figure B5).

The location of a school in a main urban area (urbanity = 1) was no longer a statistically significant predictor of the advertising score when the sample was reduced to advertising scores above zero. The direction of the estimate was reversed compared to the original model. Notably, of the 204 schools with an advertising score above zero, 95% (n = 194) were located in a main urban area.

Discussion

Summary

This study assessed the extent of children’s potential exposure to unhealthy outdoor food/drink advertising around primary and intermediate schools in Auckland, and determined if there was a spatial correlation with indicators of neighbourhood deprivation. For this purpose, an advertising exposure score was designed, based on the proportion of unhealthy food/drink advertisements and the number of total advertisements visible near schools.

The results of the spatial and statistical analysis showed that (1) outdoor advertising was present around 61.9% of the sampled schools (n = 236), (2) most food/drink advertisements promoted unhealthy food and/or drinks (n = 2699, 83.1%), and (3) there was a statistically significant relationship between the advertising exposure score and socioeconomic disadvantage.

We developed a simple and replicable advertising score as a tool to compare the extent of unhealthy outdoor food/drink advertising exposure across school food zones. In this study, the score proved sensitive enough to indicate differences in exposures between different areas of the city (Figure 2). The advertising exposure score showed a statistically significant increase with the EQI, school deciles and IMD18 (see Table 2 and Appendix B). Future outreach, community engagement and research will determine the utility of this tool for increasing community awareness about children’s exposure to unhealthy food/drink marketing in children’s neighbourhoods.

Interpretations

This study demonstrates that the majority of children attending a primary or intermediate school in Auckland are at risk of regular exposure to unhealthy food/drink advertising around their school. This risk is not distributed equally across Auckland. Students attending a school with a high EQI (or low school decile) are more likely to be exposed to unhealthy food/drink advertising on their way to and from school than children attending schools with a lower level of student socioeconomic disadvantage. There are large areas of overlap between high advertising scores and high levels of school EQI and neighbourhood deprivation, especially in South Auckland. These areas are known for struggling particularly with the consequences of socioeconomic disparities (Parahi and Shepherd 2018; ATEED 2020). Thus, the presented findings show the potential for unhealthy food/drink advertising to add to existing health disparities and issues of environmental and social injustice within the region.

Our findings are consistent with the earlier study of Vandevijvere et al. (2018), in terms of unhealthy outdoor advertising being more present around schools with the highest number of students from low socioeconomic backgrounds. However, we have observed a higher proportion of food/drink advertisements being unhealthy, as their figure was only about two thirds of the food/drinks advertised (63%). Our results are closer to those from the KidsCam study, where 89.2% of the recorded food/drink advertisements were classified as unhealthy (Liu et al. 2020).

The presented results can be compared to studies from Australia that have found outdoor advertising to be present around most primary and secondary schools, with the likelihood of exposure to unhealthy (non-core) outdoor food/drink advertisements increasing with proximity to schools (Kelly et al. 2008; Parnell et al. 2019; Trapp et al. 2022), as well as similar correlations of unhealthy outdoor advertising exposure and socioeconomic circumstances of the student body (Trapp et al. 2022).

Implications for policy and practice

There is a clear need for policy intervention that regulates and limits unhealthy food/drink advertising around schools. In NZ, existing restrictions on unhealthy food/drink advertising are voluntary and/or self-regulatory, and research has shown this approach to be insufficient to protect children from the harmful effects of marketing (Sing et al. 2020; Garton et al. 2022). The most straightforward measure would be to ban all unhealthy food/drink advertising within a certain radius around schools. Liu et al. (2020) estimated that banning all unhealthy food advertising within 400 m of schools in Wellington would reduce children’s total exposure to unhealthy food advertisements by approximately 25%. To restrict unhealthy food/drink advertisements on and around convenience stores and public transport is within the remit of local councils (e.g. through expansion and enforcement of signage bylaws) and transport authorities.

For example, Auckland Transport’s Advertising Policy contains a standard that advertising will ‘support health and healthy lifestyle choices’; however, it adds that Auckland Transport continues to support and endorse industry self-regulation ‘such as no advertisement of high saturated fat, salt or sugar products within 300 metres of a primary or intermediate school’ (Auckland Transport 2021). This sits in contrast to clearer language that advertisements promoting gambling, alcohol brands or products, or tobacco and vaping products will not be approved (Auckland Transport 2021). The policy therefore requires an update putting unhealthy food and drinks on par with other known harmful products that should not be marketed around children. Furthermore, a 500 m road network radius, as applied in this study, would be consistent with the recommendations for food marketing policy issued recently by public health experts in NZ (Garton et al. 2022).

Likewise, Auckland Council’s Signs Bylaw regulates the location, number, size, and content of signs on private and public land to ‘provide for signs that protect people and environment;’ yet it refers to the voluntary and self-regulated Advertising Standards Authority (ASA) in terms of what advertising is permitted (Auckland Council 2023). The ASA’s rules and procedures on food and beverage advertisements in its Children and Young People’s Advertising Code have been heavily criticised by academics and civil society as being weak and ineffective in protecting children from harmful marketing (Sing et al. 2020; Castles 2021; Garton et al. 2022; Egli et al. 2023). Critically, it does not condone unhealthy food/drink advertising targeting children, which is defined as being in locations where children gather (examples being in schools, school grounds, pre-school centres and playgrounds), but does not include their immediately surrounding areas (Advertising Standards Authority n.d.). The Signs Bylaw should contain mandatory and enforceable rules restricting all unhealthy product advertising that is visible from roads and footpaths in school zones. Given this suggested ban would apply to outdoor advertisement only (and not retail presence), and that the 500 m road network buffers cover less than 15% of Auckland's urban core, (see Appendix C, Figure C3), we believe establishing such a boundary is reasonable. As Brien and colleagues (2023) have pointed out, policy action to encourage the promotion of healthy food and drinks within and outside convenience stores in school zones could help to reinforce these stores as an important part of NZ communities, with the potential to be a positive part of the food environment for children.

There is the opportunity to use this advertising score as a tool to increase community awareness of the ubiquity of unhealthy advertising in children’s neighbourhoods. In the next phase of this research the team will work with schools and community groups to disseminate the information on school scores, alongside what health promotion, education and advocacy opportunities there are to advocate for policy change.

Ultimately, a suite of measures must be taken to improve the food environments in and around NZ schools. There are already efforts to improve the food and drinks available within schools, for example through the Ka Ora, Ka Ako | Healthy School Lunches Programme and through encouraging schools to adopt water-only policy. However, the school food environment does not end at the school gate. If efforts to improve children’s diets and overall nutritional health are to be successful, a more comprehensive approach which limits the presence of outdoor unhealthy food/drink advertising and other marketing around schools and in other key children’s settings is needed.

Strengths and limitations

The simplicity of the advertising exposure score makes it easy to understand and interpret. This is especially advantageous when communicating findings to a general audience and to political decision-makers. It is also possible to apply the same score in different settings, since it does not include any location-specific variables. The proportion of unhealthy food advertisements and the total number of advertisements are basic measures which could be included in virtually all research assessing the presence of unhealthy outdoor food advertising. Most of these studies are designed similarly: they count and categorise advertisements in a pre-defined area, either in person or through virtual audits on GSV (Finlay et al. 2022).

Another strength is the use of GSV to capture advertising, which facilitated timely and cost-effective data collection. In recent years, GSV has become increasingly recognised as an effective tool for data collection in health research generally (Rzotkiewicz et al. 2018) and outdoor food advertising research specifically (Finlay et al. 2022). The advertising exposure score, developed and presented here, demonstrates the potential for the synthesis and visualisation of such spatial data. To the author’s best knowledge, this study is the first to develop such a score and use it to assess the spatial distribution of unhealthy advertising exposure.

One study limitation is that the advertising exposure score is a measure of potential exposure, not actual exposure. The score indicates where in Auckland the presence of unhealthy food advertisements and outdoor advertising in general is high, relative to other areas in the region. Few studies measure the actual exposure of children to food advertising as this requires considerable amounts of time and resources, as well as the active participation of children in data collection (Finlay et al. 2022).

The advertising exposure score is simple in its design. Notably, the score does not include a measure of the ‘population at risk’, i.e. the number of children attending each school. Including school roll as a weight in the advertising exposure index was considered, however, this was ultimately deemed inappropriate. However, it is important to consider the number of students who are potentially exposed to the unhealthy food advertising in each school neighbourhood when considering the impact of advertising exposure.

There are other important variables when considering the exposure to outdoor advertising that are not captured by the advertising exposure score, for example the size of an advertisement. In its present state, the advertising exposure score gives the same weight to an A4-sized poster as it does to a large billboard on the side of a store. Another possibility would be to include a measure of the degree of ‘unhealthiness’ of an advertisement, for example by assessing the portion sizes of the food on display (Porter et al. 2023). Nevertheless, the score we present serves as a basic model which could be refined in future research and to which additional data could be added.

Conclusions

This study developed a simple advertising exposure score and applied existing data to measure the extent of children’s potential exposure to unhealthy food/drink advertising in their school food zones in Auckland, NZ. It also investigated the spatial correlation of this score with indicators of neighbourhood deprivation. The results of the spatial and statistical analysis showed that (1) outdoor advertising was present around 61.9% of the sampled schools (n = 236), most of which were in urban areas; (2) most food advertisements promoted unhealthy food and/or beverages (n = 2699, 83.1%); and (3) there was a statistically significant relation between the advertising exposure score and socioeconomic disadvantage as measured by both EQI, school deciles, and IMD18. The advertising exposure score proved to be a simple and easily interpreted summary measure of children’s potential exposure to unhealthy food/drink advertising, which may easily be replicated in other settings.

Acknowledgements

The Senior Author (VE) conceived the original study design. KG and VE funded and oversaw data collection. KK completed data analysis and interpretation, with supervision from KG and VE, and DE provided input in data analysis, interpretation and visualisation. KK led the generation of figures, and writing of the manuscript. All authors were involved in writing the paper, reviewing changes based on peer-review feedback and had final approval of the submitted versions. The authors wish to thank Amanda Brien for the training of research assistants and supervising the original data collection, as well as Grace Shaw, Chelsea Riddell, and Selda Meneses for collecting the Google Street View data used in this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data reported in this study is available upon reasonable request.