Skip to main content

Table 3 Studies measuring exposure to outdoor food marketing

From: A scoping review of outdoor food marketing: exposure, power and impacts on eating behaviour and health

Author(s), year of publication

Study objectives

Study setting

Study Design/ participants

Methods

Advertising mediums included

Main results relating to outdoor advertising

Adams et al. (2010) [80]

To explore differences in the prevalence of outdoor food advertising, and the type and nutritional content of advertised foods, according to an area-based marker of socio-economic position (SEP)

Bus routes and areas around shops in Newcastle upon Tyne city centre, UK. (1371 ads)

Cross-sectional, observational, content analysis

All outdoor ads within city boundaries were identified, photographed, their size estimated (m2) and location determined using a GPS device. Ads were classified as food or non-food and food ads grouped into six categories based on the ‘eatwell plate’ plus an additional category: other foods. Nutritional information was obtained and classified as HFSS or not (using UKNPM). Data was compared using the English Index of Multiple Deprivation.

All outdoor advertisements for any product

15% (n = 211) of ads were for food (20% of ad space). “Advertised foods tended to be high in fat and low in protein, carbohydrate and fibre”. 33.9% of food ad space was devoted to HFSS products. 14.7% of food ads (n = 31) were for the ‘KFC Buffalo Toasted Twister’. All food ads featured specific products rather than brands generally.

PROGRESS: The proportion of ads for food was significantly smaller in the most (13%) vs. the least (18.4%) affluent tertile (p < 0.05) however foods advertised in the most affluent tertile were the poorest choice for health, whilst foods in the middle tertile were the best choice for health.

Total food advertising space was larger in the least affluent tertile, but proportion of advertising space for food was highest in the middle tertile(24.3%). HFSS advertising space was significantly higher in the middle than in the least affluent tertile (p < 0.001). Little evidence that nutritional content or types of food showed consistent socio-economic trends across tertiles.

Adjoian et al. (2019) [58]

To determine if outdoor advertising density for non-alcoholic drinks, food, tobacco products, and alcohol, is associated with neighbourhood poverty or other Census-level characteristics in New York City (NYC)

Random sample of retail-dense streets in New York City (1106 sampled street segments, 953 eligible) (16,305 ads for consumable products including alcohol and tobacco)

Cross-sectional, observational, content analysis

Street segments were stratified by poverty level (determined by census tract data) and borough. Photographs were taken of street-level ads for consumable products and coded for location type and content. For each ad, coders noted the presence of (1) non-alcoholic beverages, including sugary drinks, low-calorie drinks, water or seltzer, unsweetened coffee, other drinks and unknown drinks; (2) food, including fresh produce, sweets, and other, a notation was made if ads were for fast food.

Ads located on storefronts, awnings, building walls, construction walls, newsstands, bus shelters, subway entrances, etc. were counted as well as Ads located inside an establishment but intended to be seen from the street

41.8% of consumable ads were for food and 38.6% for non-alcoholic beverages.

Of all consumable products, most frequent categories were other food (29.2%), sugary drinks (27.6%), fresh produce (9.1%), sweets (8.8%) other drinks (5.3%)

Food made up 54% of all product images, non-alcoholic beverages 24%. The most prevalent product images were other food (30.8%), sugary drinks (16.2%) fresh produce (13.8%), sweets (9.2%) and other drinks (2.9%).

PROGRESS: For every 10% increase in proportion of Black residents there was a 6% increase in food images and 18% increase in non-alcoholic beverage images. Images for sugary drinks/water/seltzer and fruits and vegetables were more common in areas with a higher proportion of Black residents. Density of product images for non-alcoholic beverages and sugary drinks was higher in areas with a higher percentage of adults with less than a HS diploma.

Demographic: Less than 1% of ads were child focused.

Note: additional food data was not extracted due to being combined with other consumable products.

Barnes et al. (2016) [59]

1) To characterize the presence of ads for healthful and less healthful F&B

2) To describe the product placement of items for impulse buys and assess differences in marketing strategies among food stores

119 randomly selected small stores in Minneapolis, St Paul, Minnesota, US

Cross-sectional, observational, content analysis

Food retailer audits were completed on weekdays. Data collectors recorded whether images of healthful or less healthful foods were present. Healthful foods included fruit, vegetables, wholegrains. Less healthful foods defined as high calorie, low nutrient, including alcoholic beverages (all HFSS).

Images of healthful and less healthful foods on storefront doors or windows. Images defined to included brands, logos, or texts of specified food items and well-known products.

37% of stores (n = 44) had ads for healthful items on their exterior. 55% of food-gas marts had at least 1 healthful exterior ad, (17% for corner/small grocery stores). 11% of stores had only healthful exterior ads.

Note: additional food data was not extracted due to being combined with alcoholic beverages

Barquera et al. (2018) [81]

To analyze the characteristics of food advertising practices around 60 elementary schools in two cities and to evaluate compliance with the Pan American Health Organization (PAHO) recommendations and the local food industry self-regulatory marketing code.

100 m buffer zones around randomly selected schools in two cities in Mexico (29 schools in Curenavaca (16 private, 13 public), 31 schools in Guadalajara (11 private, 20 public)) (278 F&B ads)

Cross-sectional, observational, content analysis

Buffer zones inspected for any commercial source of F&B advertising. Photographs taken of all food ads and ads coded by food category (soda, juices, SSBs, sweet snacks, chocolates and candies, milk and dairy products, salty/fried snacks, water, ice cream and other) and noted brand name, type of ad and use of promotions. Compliance with PAHO and PABI (industry self-regulatory code) recommendations evaluated.

Poster, banner, sticker, painting on walls, flags billboards and walls not related to stores in the buffer areas.

278 F&B ads, 69.8% of which were outside (n = 194), only 3 billboards were identified (excluded from analysis).

PROGRESS: 73% of outdoor ads (n = 142) were identified around public schools, compared to 27% (n = 52) around private.

Basch et al. (2019) [60]

1) To determine the prevalence of SSB ads on LinkNYC kiosks and document the frequency containing sugar. 2) To examine whether the prevalence of SSBs varied by median income level of the census tract location.

Random selection of 100 LinkNYC kiosks (out of a possible 507) in New York city (2025 ads)

Cross-sectional, observational content analysis

Researchers stood at each kiosk for 10 minutes and coded beverage ads (type, size and flavour), zip code of each machine and corresponding median annual income for zip code using census tract data. Calories per serving of each beverage was ascertained as well as sugar content.

Digital LinkNYC kiosks - provide ability to make calls, get information and charge devices. With 55 in. screens on both sides

17.1% of ads were for non-alcoholic beverages, 59.4% (n = 206) of these were for SSBs including iced tea (n = 106) soda (n = 77) blended coffee drinks (n = 22) and flavoured water (n = 1). For the SSBs, the mean kilocalories and grams of sugar per serving was 149.90 and 35.04 respectively.

PROGRESS: No significant differences observed over income quartiles. (p = .38)

Cassady et al. (2015) [61]

To present patterns of advertising related to the two key obesity-related behaviors, diet and physical activity, in an economically and racially diverse urban area in Northern California, and investigate whether there are disparities in the distribution of these ads by neighbourhood income and race.

Urban area of Northern California – Sacramento, USA (171 health-related outdoor ads covering 23,971ft2)

Cross sectional, observational, content analysis

16 zip codes were categorised as high or low income and majority ethnic group. Trained researchers used an app (droidSURVEY) to record subject, language, GPS coordinates, and a photo of each ad. Area of each ad determined. Duplicates and electronic billboards removed (multiple messages). All ads coded as unhealthy F&B or physical activity. Ads were classed as healthy if they promoted F&B encouraged by the Dietary Guidelines for Americans and unhealthy if high-calorie, low-nutrition.

Billboards, bus shelters, bus benches and posters on storefronts (electronic billboards removed)

32% of all health-related ads were F&B, (52% of total advertising area). “National brand fast food and grocery stores dominated unhealthy food advertising, partly because most of the grocery store ad areas were devoted to images of pizza and ice cream”. Ad space for beverages was equally healthy and unhealthy (5ft2 vs 6ft2 per mile) but 4x the space was devoted to unhealthy vs healthy foods (24ft2 vs 6ft2 per mile), “about half of the healthy ad space was sponsored by the government or a non-profit agency campaign”.

PROGRESS: Low-income Latino and African American neighbourhoods had more F&B advertising (2-35x ad space) with 5x unhealthy food ad space compared to high-income white neighbourhoods and 6x that of low-income white neighbourhoods.

Low-income Latino and African American neighbourhoods had 50% more ad space for unhealthy beverages. All neighbourhoods except low-income white, had more space devoted to unhealthy than healthy food ads. Unhealthy beverage ads were more dense in low-income and multiracial neighbourhoods.

Chacon et al. (2015) [82]

To describe the type of snack foods advertised to children in stores in and around public schools and assess if there was an association between child-oriented snack food advertising and proximity to schools.

Outside 55 stores around two preschools and 2 primary schools in Mixco, Guatemala (321 snack food ads)

Cross-sectional, observational, content analysis

Using Google Earth, researchers located all stores within 200 m radius of school entrances and categorised them as less than 170 m away and more than 170 m away. Researchers coded store type, number of ads inside and outside stores and those that were child-oriented.

Posters, stickers, free-standing signs, banners, painting on walls or flags.

For exterior child-oriented ads – there was a median of 1 (0–2) outside all stores, 2 (1–2) < 170 m from schools and 1 (0–2) >/170 m from school.

Dowling et al. (2020) [62]

To estimate the density of street-level sugary drink ads across the 5 boroughs of NYC and describing variation by neighbourhood.

Random sample of 953 retail-dense street segments in New York City (4356 ads)

Cross-sectional, observational, content analysis

Surveyors photographed ads along both sides of segments and coded with nutrition information to determine if beverages were “sugary drinks” (Defined as featuring pictures or logos of beverages with added caloric sweetener and > 25 cal per 8 oz. serving). Poverty determined by census tract data.

Posters, digital signs, stickers on outdoor structures such as newsstands, bus shelters and payphones.

4356 ads were identified featuring sugary drinks, with 8197 sugary drink images. On average, there were 2 images per ad. Between 2.72 and 29.91 ads for sugary drinks were observed per 1000 ft. segment.

PROGRESS: Sugary drink ad density was 1.54 times (95% CI = 1.16, 2.04) as high for medium vs low poverty neighbourhoods and 1.66 times (95% CI = 1.26, 2.19) as great for high vs low poverty neighbourhoods. After adjusting for census tract characteristics, associations were still significant in two out of five boroughs.

Associations were observed between ad density and percentage of residents with less than a High school diploma (10-unit IRR = 1.15, 95% CI = 1.04, 1.27); percentage of asian or pacific islander residents (10-unit IRR = 1.10, 95% CI = 1.01, 1.19) and the percentage of black, non-latino residents (10-unit IRR = 1.20, 95% CI = 1.14, 1.26).

Egli et al. (2019) [83]

To add to the body of knowledge on obesogenic environments for children and the utility of Google Street View in geographical health research.

800 m buffers around 19 schools in Auckland, New Zealand (2472 ads).

Cross-sectional, observational, content analysis

Two independent research assistants ‘travelled’ around streets in the 800 m buffer areas on Google Street View between March and July 2017 and took screenshots of outdoor ads. Distance from school to ads was measured using Google Maps. Foods reflecting dietary patterns associated with increased risk of obesity and dental caries in childhood, processed, energy-dense, nutrient-poor were classed as “unhealthy”.

billboards, signs, flags, banners, balloons, neon signs, stickers, and bus shelter ads that were large enough to be seen on a 15 in. computer screen; with an identifiable logo or text; and located completely or partially on public land

2474 ads identified, 29.4% could not be identified (blurry), of identified ads, 15.5% were for F&B (calculated from available data).

Of total ads, the most frequent food categories were “food, other” (5.6%), “food, unhealthy” (5.4%), beverage other (2.1%), F&B other (0.9%), and SSBs (0.9%).

Demographic: 46.6% of F&B ads were marketed to children and adults, none to children only. Of ads marketed to both adults and children, there were significantly more unhealthy F&B than other F&B (P = 0.001).

PROGRESS: No significant difference in F&B advertising marketed to children by neighbourhood deprivation

Note: additional food data was not extracted due to being combined with alcoholic beverages

Fagerberg et al. (2019) [84]

To explore the proportion of ads directly related to ultra-processed foods (including sugary drinks and fast foods) in two areas of Stockholm, Sweden with low vs. high SES

Two districts of Stockholm, Sweden. Around/in subway stations and shopping malls (Östermalm (high SES) and Skärholmen (low SES)) –4092 ads

Cross-sectional, observational, content analysis

All ads were recorded by two researchers using smartphones. Pictures were taken, and annotated by a dietician according to difficulty to annotate, location, food or not, sugary drink or not, including food promotion or not. Following categorisation, another category was created (ultra-processed food) by adding the sum of “sugary drink” and “fast food” categories.

All ads in the subway station and 50 m to the left and right on the streets outside entrances to subway. All ads in shopping mall and on streets surrounding shopping mall. Excluded - flyers/brochures, ads inside/outside moving vehicles, inside stores. Subway escalator ads excluded as they change weekly.

32.8% of ads (1341 pictures) were for food products. 65.4% of food ads promoted ultra-processed products (877 pictures).

PROGRESS: No significant difference in proportion of food ads out of total ads by SES, however there was a significantly higher proportion of ultra-processed food ads in Skarholmen (low SES) compared to Ostermalm (high SES) (73% vs 59%, p < 0.001/p = 0.001).

The higher proportion of ultra-processed ads in Skärholmen can be explained by a higher proportion of fast food ads; 65.4% vs 48.8% (p = 0.000; researcher 1) and 61.1% vs 36.4% (p < 0.001; researchers 2 + 3).

There was no significant difference in the proportion of sugary drink ads: 30.7% vs 28.1% (p = 0.442; researcher 1) and 29.1% vs 30.6% (p = 0.700; researchers 2 + 3).

Feeley et al. (2016) [85]

To assess consumption of commercially produced foods including breast milk substitutes, commercially produced complementary foods and various snack food products consumed by children less than 2 years of age as reported by their mothers, and maternal recall of promotions of these products.

Outside sampled health facilities in Dakar, Senegal.

Cross-sectional survey

293 mothers of children utilizing public/faithbased/non-governmental health facilities

Mothers were interviewed about F&B consumed on the day/night prior to the interview. Data also gathered on weekly consumption, reasons for feeding, expenditure on snacks and promotional practices inside and outside health facilities for breast milk substitutes and commercially produced complementary foods and commercially produced snack foods.

Billboards

4.4% of participants reported seeing a billboard promoting breast milk substitute. 4.4% of participants reported seeing promotion of commercially produced complementary foods.

Huang et al. (2020) [86]

To determine whether there are associations between school decile, distance from school, walk and transit scores, and the prevalence of non-core F&B promotion at bus stops within walking distance from all Auckland schools.

500 m around 190 schools in Auckland, New Zealand (842 ads)

Cross-sectional, observational, Content analysis

Google Street View was used to screenshot images of bus shelter ads which were saved by school type. The shortest walking distance was calculated using the map tool in GSV. F&B were coded according to WHO Euro NPM, making them core or non-core.

Ads on each side of bus shelters large enough to be seen on a 15-in. computer screen and with an identifiable logo or text.

842 ads were identified, 25.5% promoted F&B. 50.2%) were non-core.

The food categories promoted were core foods (9.7% of all ads), non-core foods (8.1%), non-core beverage (3.6%), and core beverages (3%).

All ads promoting non-core F&B together were fast food, none promoted core F&B together.

The number of F&B ads per 100 m increased as the distance from schools increased.

PROGRESS: There were 702 ads around state or state integrated schools, 140 around private.

The greatest proportion of ads for non-core food (33.8%), core food (32.9%), and non-core F&B (50%) were found in low deciles, followed by high deciles then medium.

High decile schools had the greatest proportion of core beverage ads (40%).

When high-decile areas were combined with areas around private schools, the greatest number of all F&B ads (42.3%) and non-core ads (41.7%) were found in high SES areas.

Kelly et al. (2008) [87]

To quantify the volume of and factors associated with food ads in an environment where young children are commonly found (area surrounding primary schools). We also aimed to identify the specific food products that are advertised around primary schools.

20 Local government areas in Sydney and Wollongong, Australia with high/low population density and high/low SES. 250 and 500 m radius around 40 primary schools (9151 ads, 2286 for food)

Cross-sectional, Observational, content analysis

Two primary schools randomly selected from each government area. A coder navigated the 250 m and 500 m radius around schools to identify all ads. Ad characteristics captured: type (food/non-food), description of product, distance to school, size, location and foods further into core and non-core.

Standard commercial ads (billboards, posters) temporary advertising of events, outdoor furniture, signs on buildings (name and branded product information) Signage was excluded from analysis = all signs unaccompanied by branded product information (identification of sites etc.)

19.5% of ads were for F&B, 58% of which were non-core (without alcohol, calculated from available data), 5% of F&B were core, 15% tea or coffee.

Most frequently advertised foods were soft drinks (24%), coffee (15%), and ice cream and confection (14%).

< 250 m from schools 69.7 out of 114.9 food ads per km2 (60.7%) were non-core foods (not including alcohol, calculated from data) compared to 33.5 out of 58.8 food ads per km2 (56.8%) 250-500 m away from schools.

Note: additional food data was not extracted due to being combined with alcoholic beverages

Kelly et al. (2015) [88]

To identify, describe and quantify the volume of F&B ads around schools in two demographically and culturally disparate cities in Asia.

500 m around 30 primary schools in Ulaanbaatar (Mongolia; 1459 ads and Manila, The Philippines; 9687 ads).

Cross-sectional, observational, content analysis

Primary schools were randomly selected in each city. Researchers surveyed all streets within 250 and 500 m from schools. Noted distance from school, size of ad, setting, type and position, whether single or multiple foods, brand name and type. Foods were classed as core/healthy, non-core/unhealthy and miscellaneous.

billboards, posters, free standing signs, neon signs, stickers, electronic boards, banners, bus shelter signs and signs on outdoor furniture, bridge/awning signs- and painted buildings. Store signage with a product logo in addition to identification was included.

Mongolia – 1459 food ads were identified, 88% were non-core (not including alcohol, calculated from available data) 52% of non-core ads were SSBs.

Coca-cola (35%) and Pepsi (10%) were the most frequent food/drink brands advertised.

Philippines - 9687 food ads were identified, 78% were non-core (not including alcohol, calculated from available data). 56% of non-core ads were for SSBs.

The most frequently promoted brand was Coca-Cola (32%) followed by a local soft drink manufacturer (RC Cola – 8%)

Note: additional food data was not extracted due to being combined with alcoholic beverages

Liu et al. (2020) [89]

1) Measure and visualize children’s space and time exposures to unhealthy food advertising in public, outdoor spaces, using GPS and wearable camera technology; and 2) Test potential reduction of exposure by proposing banning unhealthy food advertising near schools, playgrounds, or in residential areas.

Wellington City, New Zealand

(59,150 ads)

Cross-sectional, observational, content analysis

138 students from 16 schools aged 11–14

Participants wore cameras and GPS devices for four consecutive days. Cameras captured a 136o image every 7 seconds. All images coded for food advertising exposure through content analysis. Foods classified as core or non-core based on WHO NPM. GPS data (recorded every 5 seconds) was linked to image codes

Ads in outdoor public areas

Children were exposed to a mean of 8.3 food ads every hour they spent in outdoor public areas (95% CI 7.9–8.7). 89.2% of these were unhealthy, 9.6% were healthy.

Shop fronts and main streets were the “most common settings” for unhealthy food products.

“Children were exposed most commonly to advertisements for fast food, sugary drinks, ice cream, cookies and confectionary.”

Weekday exposures “frequently occurred during times of travel to and from school”.

Public transport facilities (11 children), fresh food markets (6 children), and service station forecourts (n/a) were the least frequent exposure settings.

Lo et al. (2020) [90]

To examine the extent and characteristics of F&B promotion in Hong Kong mass transit railway (MTR) stations in districts with different SES and school density and to assess if children or lower SES audiences were specifically targeted by advertising.

8 stations of the Mass transit railways in Hong Kong (8064 ads)

Repeated cross-sectional, observational, content analysis

8 stations selected and classified by SES and school/non-school zone. Stations were visited on three occasions, photos and videos of all ads were taken - for digital promotion, an average of 3-minute videos were taken. F&Bs were categorised into three groups (core and healthy, non-core and unhealthy and seasonal and special) The number of F&B in each ad was documented and healthiness of each product reported.

All ads located in the MTR station area (escalator, platform, trackside, shops, paid and unpaid area and outside of stations)

10.6% of ads were food-related (calculated from available data). This was higher in seasonal periods (15.3 and 13.1% vs 7.3%).

53.5% of food ads were for non-core foods (excluding alcohol, calculated from available data), 43% were core and healthy, the remaining were seasonal and special (2.5%).

Most frequently advertised food products were: sweet breads, cakes and pastries (18.3%), meat and meat alternatives (17.3%), grains and bread (9%), SSBs (8.8%) and processed meat/alternatives (6.8%).

Stations in school zones had a significantly higher proportion of ads for processed grains (2.7% vs 7%), processed meat and meat alternatives (9.1% vs 4.9%), high-sugar snacks (1.5% vs 0.1%), sweet breads (23.2% vs 14.3%) and seasonal food/items for special purposes (1.7% vs 0.7%) and fewer ads for fruit and fruit products (1.5% vs 3.2%) and Chinese dried seafood and tonic (0% vs 1.9%).

PROGRESS: In high SES areas, 41.7% of food ads were core and 54.9 non-core (calculated without alcohol) compared to 47.1% core and 49.5% non-core (calculated without alcohol) in low SES.

High SES stations had a higher proportion of ads for processed meat/meat alternatives (7.9% vs 4%) and Chinese dried seafood and tonics (1.5% vs 0%), and fewer ads for fruit/fruit products (1.6% vs 4.6%), milk/dairy (0.5% vs 1.9%) and processed grains (0.9% vs 3.2%).

Note: additional food data was not extracted due to being combined with alcoholic beverages

Lowery & Sloane (2014) [63]

To examine associations between the content of outdoor advertising and neighbourhood ethnic/racial and socioeconomic composition to see whether particular communities disproportionately host harm.

Sign districts in Los Angeles (legislation limits outdoor ads to 21 commercially zoned sign districts)

Cross-sectional, observational, content analysis

7/21 sign districts were selected using census tracts within 500 ft. of each regional center. Data on the area of each sign district, street length and number of intersections obtained from City of Los Angeles Department of City Planning. Photographs of ads were taken monthly for 7 months. Harmful content coded into 5 categories (encouraging addictive behaviours (alcohol/tobacco), violence, unhealthy eating, unsafe environments for women and content inappropriate for children)

Billboards (digital and conventional)

Most frequent unhealthy food ads were for “fast food options (hamburgers, fried foods) and advertisements for soft drinks, flavoured beverages and candy”.

PROGRESS: Ads featuring unhealthy food options were most prevalent in the African American community in Baldwin Hills (18.6%) and the Latino community more densely populated with young people (12.7%)

Lucan et al. (2017) [64]

To assess all print advertising in all stations of a subway system for large urban county. The goal was to determine how placement of ads for foods and beverages related to subway ridership and to the demographics, dietary intake, and prevalence of diet-related conditions in the residential areas surrounding stations.

68 Subway stations in the Bronx, New York, USA (1586 ads, 284 distinct)

Cross sectional, observational, content analysis

Researchers rode all subway lines, assessed all print ads and decided whether they were targeting certain groups (minorities or youth).

F&B ads were recorded as “more healthful” or “less healthful”.

Census data was gathered, and ridership data was collected from the Metropolitan transport authority.

Self-reported fruit and vegetable consumption and presence of diet-related conditions was provided by the city health department

Wall posters, free-standing billboards, or other signage (on trash cans, receptacles, turnstiles, station clocks or benches)

39.7% (27/68) of stations had food ads. No ads promoted “more-healthful” foods or beverages, however 12 distinct ads (27.9%) pictured items classed as “more-healthful”.

Distinct ads: 10.9% (n = 31) were for F&B (calculated without alcohol), 39.3% were less healthful (calculated). There were 5 distinct ads for non-alcoholic beverages and 12 for food.

Including duplicates: 7.8% (n = 124) of ads were for F&B (calculated from available data), 39.3% were less healthful. There were 24 ads for non-alcoholic beverages and 40 for food.

Demographic: 26.1% of stations had less healthful food ads directed at youth.

Note: additional food data was not extracted due to being combined with alcoholic beverages

Maher et al. (2005) [91]

To examine the extent and content of outdoor food ads and food availability from outlets in the vicinity of secondary schools.

1 km radius around a convenience sample of 10 rural and urban high schools in New Zealand (1408 ads)

Cross-sectional, observational, content analysis

School neighbourhood SES obtained through Ministry of Education classification system. Researchers walked around the area, recorded ad info and took a photograph of all F&B ads. Food ads classified according to food type, products with desirable and undesirable characteristics labelled as healthy for a conservative approach

Billboards, neon signs, posters, stickers, free-standing signs, banners, painted buildings, bus shelter ads, flax and images in shop windows designed for viewing outside. Ads on buses/delivery vehicles excluded.

Approximately 57% of ads surrounding schools were for food products (excluding alcohol, calculated from available data). The most frequent food categories were soft drinks (21.6%), healthy foods (e.g. bread, milk; 18.8%), frozen confectionary (16.2%) and savoury snacks (11.4%).

The most frequent food brands advertised were Coca-Cola (17.6%), Tip Top ice cream (10.4%) and Meadow Fresh dairy foods (3.9%).

PROGRESS: There was a significantly greater proportion of ads for takeaway outlets (RR 1.54, p < .001) and fast food outlets (RR 1.67, p < .001) in higher SES areas.

Staple food ads were more frequent in low SES areas (RR 2.04, 95%CI 1.54 to 2.69, p < .00001). The proportion of frozen confectionary foods ads was higher in rural than urban neighbourhoods (p = 0.048)

Note: additional food data was not extracted due to being combined with alcoholic beverages

Moodley et al. (2015) [92]

To investigate the density of outdoor SSB advertising and the number of formal and informal vendors selling SSBs in a transforming, historically disadvantaged urban setting of South Africa.

5 areas of Soweto, Johannesburg (145 ads)

Cross-sectional, observational, spatial analysis

Researchers walked or drove through each street in the study area. Data was collected on location (GPS) and type of ads and food vendors. Photos were taken of all ads. Size of ads estimated (S/M/L). GPS used to create spatial point patterns and assess the association of points to the nearest school.

Billboards, bus stop ads, signs placed along the sidewalk, urban art on streets or buildings, large posters and signage for restaurants or food vendors (taxis and buses excluded)

145 SSB ads were identified. Ads were frequently located outside houses (53%), at small shopping centres (15.9%), on the street (13.8%), at schools (9%) or other buildings (4.8%).

Frequent formats were shop signs (54.5%), posters (16.6%), painted ads (10.3%), school signs (8.3%) and bus stops (4.1%).

62% of branded SSB ads were part of a display sign for a shop.

13/28 primary and secondary schools displayed SSBs on branded school signs.

There was an increase in SSB ads with decreasing proximity to nearest school (RI −2.7, 95%CI = 2.48–1.82).

Olsen et al. (2021) [93]

(1) Categorise the content of ads at bus stops across a large and varied geographical area. (2) Explore associations between the socio-spatial distribution of bus stop ads (3) Test for associations between specific categories of unhealthy commodity ads in the local area surrounding schools (4) Calculate children’s ‘real’ exposure to bus stop ads using individual mobility data of Scottish children.

1845 bus shelters in the central belt of Scotland, UK (including the two most populated cities - Glasgow and Edinburgh) (3123 ads)

Cross-sectional, observational, content analysis

229 children (aged 10–11) who took part in the SPACES study in the central belt of Scotland

CANVAS Google Street View software was used to identify bus stops and the most recent image captured of bus shelters was audited. Bus stops were assigned deprivation ranks using the Scottish Index of Multiple Deprivation. School locations were plotted on a map and 100 m, 200 m and 800 m road and path buffers were created around each school. Children were asked to wear accelerometer and GPS to measure “real exposure” over 8 consecutive days. Bus shelter ads were categorised as unhealthy food and/or drink, SSBs, alcohol, e-cigarette product, gambling

Bus shelters

Unhealthy commodity ads were frequently for fast food (15.3%), confectionary (6.8%), SSBs (3.8%), water (3.3%) and ice cream/frozen desserts (2.3%).

It was unlikely that unhealthy products were advertised within the school environment (100 m–800 m distance).

Children living in urban areas had greater exposure to unhealthy food (coef: 1.29, 95%CI 1.04–1.31) and unhealthy food and drink (coef: 1.29, 95%CI 1.04–1.31) than those in rural areas.

PROGRESS: There did not appear to be a social relationship between deprivation and ad location or type. Measures of real exposure found that children residing in more deprived areas had more contact with the transport network and so were significantly more exposed to unhealthy foods (coef: 1.18, 95%CI 1.06–1.31) and unhealthy food and drink (coef: 1.18, 95%CI 1.0–1.31).

Palmer et al. (2021) [94]

To develop a deep learning workflow to automatically extract and classify unhealthy ads from street view images.

Three areas of Liverpool, UK of differing levels of deprivation - City Centre, North Liverpool and South Liverpool (10,106 ads)

Cross-sectional, observational, content analysis

One researcher cycled on major roads in areas of interest with a camera attached recording images at 0.5 second intervals. Categorised areas (e.g. billboards) were extracted through seamless scene segmentation.

Billboards, company logos and store front signs.

10,106 ads identified in total, 13.2% (n = 1335) were for food. Following duplicate removal, 873 food ads remained.

PROGRESS: “Larger proportions” of food ads were found in deciles 1–6 (compared with 7–10). “Largest proportions” of food ads can be found within cosmopolitan groups, multicultural metropolitans and hard-pressed living groups.

Parnell et al. (2018) [95]

To assess the volume and type of unhealthy bus shelter ads near kindergartens, primary and secondary schools in five local government areas in Perth, Western Australia, and assess changes with seasonality.

37 Bus shelters 500 m around schools in 5 local governments (all high SES) in Perth, Western Australia – (293 ads over four audits)

Repeated cross-sectional, observational

Researchers visited each shelter, recording the company, product and location. Each shelter ad was photographed. This was completed four times over the course of a year. Ads were coded as healthy, moderate or unhealthy by a nutritionist using guidelines from the Australian government.

Bus shelters

Across the four audits, 21% of bus shelter ads were for unhealthy food/non-alcoholic beverage (62/293). This was 67.4% of all unhealthy ads (including alcohol and gambling).

Less than 1% of ads promoted a healthy product. There was no significant variation in the volume of unhealthy advertising by season.

Most frequent unhealthy food products were fast food meal deal (14.5% of unhealthy F&B ads) fast food burger (12.9%), chocolate (11.3%), ice cream (9.7%),

fast food chicken (6.5%) and fast food hash brown (6.5%).

Pasch & Poulos (2013) [65]

To document and describe F&B advertising around schools

1600 m around 4 middle schools in Austin, Texas, USA (563 ads)

Cross-sectional, observational, content analysis – pilot study

Data collectors documented the location and photographed all F&B ads (created the Outdoor Media Direct Observation Tool (DOT) on app FileMakerGo. Also documented ad type, category and description

All free standing, located on something other than a building, or billboards. Establishment advertising included ads directly attached to the building.

75.9% of F&B ads were directly attached to establishments, 24.2% free standing.

Ads were most frequently located on or at convenience stores/gas stations (median of 77 ads per school). There was a median of 8.5 street side ads per school.

There was a median of 22 ads per school associated with restaurants, particularly fast food (13 per school).

Pinto et al. (2007) [96]

To describe billboard advertising of F&B in Maputo, Mozambique.

9 busy downtown avenues in Maputo, Mozambique (707 billboards)

Cross-sectional, observational, content analysis

Researchers visited the chosen avenues and described billboards found. Items advertised were classified by 2 researchers as alcoholic beverages, soft drinks, fast food and non-fast food.

Billboards

10.7% (76/707) of ads advertised: soft drinks (n = 16), fast foods (n = 33) and non-fast foods (n = 27).

Rice was the most frequent non-fast food item advertised (60%). All fast foods were products of international companies.

PROGRESS: Soft drinks accounted for 15.3% of ads downtown and 16.1% in the suburbs. Non-fast food was more frequently advertised in the suburbs (48.4% vs 16.7%) and fast foods more frequently downtown (41.7% vs 9.7%), thought to reflect targeting of wealthier subjects - Pearson x(2) (3) = 15.2, p < .01).

Note: additional food data was not extracted due to being combined with alcoholic beverages

Poulos & Pasch (2015) [66]

To provide a detailed description of the development of a tool for documenting and describing primary data collected on outdoor F&B advertising and F&B outlets. Furthermore, inter-rater reliability of the tool for documenting F&B ads and outlets will be determined.

½ mile radius around 34 middle and 13 high schools in central Texas, USA (732 ads)

Cross-sectional, observational, content analysis

Researchers created maps and driving routes covering all streets. An app was used to record ad information, photographs and GPS. Products assessed included F&B, alcohol and tobacco.

Billboards, street side sign, Directory, a-frames, banners (e.g. on entrance to shopping centre) and any other sign or establishment advertising food or beverages. Signage directly placed on windows or walls of an establishment

499 establishment ads were identified, 233 other ads.

Urban schools had more ads in surrounding areas than suburban schools (229 vs 8).

Puspikawati et al. (2020) [97]

To survey areas around gathering places for children and adolescents (schools and other facilities) to identify outdoor F&B ads that may be risk factors for obesity, diabetes and CVD, in order to provide evidence for policy making.

Surabaya, the capital city of East Java (urban) and Banyuwangi district (rural), Indonesia. 0-500 m from gathering places for children and adolescents. (570 F&B ads in Banyuwangi, 960 in Surabaya)

Cross-sectional, observational, content analysis

Data collectors walked or rode a motorbike around all roads and alleys in chosen locations (0-100 m, 100-300 m, 300-500 m from gathering places) and recorded location and photographs via an app. All F&B ads were surveyed. Nutritional information of products was obtained from the packaging or internet and foods were classed as unhealthy (if containing high levels of fat, salt or sugar)

Banners, posters, billboards, video ads, outdoor ads on shops and stalls, stickers merchandise and any other items containing F&B ads

In Banyuwangi, 39.8% of F&B ads were for unhealthy foods and 1.8% healthy. SSBs were frequent (47.2%).

In Surabaya, 28.2% of F&B ads were for unhealthy foods and 1.8% healthy. SSBs were frequent (46.1%).

In both areas, density of unhealthy food, healthy food and healthy beverage ads increased closer to the facilities for children and adolescents (unhealthy beverages were combined with alcohol)

Note: additional food data was not extracted due to being combined with alcoholic beverages

Richmond et al. (2020) [98]

To examine the number and type of food ads to which children are exposed when using public transport or walking to school in Sydney, Australia.

21 schools in the greater Sydney area, Australia (762 ads)

Cross-sectional, observational, content analysis

Theoretical train, bus and walking routes were planned for a one-way trip to each school using Google Maps. Researchers travelled the routes and recorded all visible ads. Advertised foods were categorised as core, discretionary or miscellaneous

Ads greater than A4 size that appeared in outdoor spaces - in train stations, and on buses, bus shelters and telephone posts (those on shop fronts and sandwich boards were not included

Approximately 30% of ads were for F&B (calculated from available data). The majority of foods were discretionary (69%, calculated without alcohol).

Most frequently advertised food products were fast foods (23%), sugary drinks (17%) and snack foods (16%).

PROGRESS: There was no significant difference in the rate of core ads across schools in high and mid/low deciles.

Note: additional food data was not extracted due to being combined with alcoholic beverages

Robertson et al. (2017) [99]

To explore the socioeconomic patterning of food advertising at bus stops in Edinburgh, UK.

290 bus shelters in Edinburgh, UK (562 ads)

Cross-sectional, observational, content analysis

An app was used to record data on GPS coordinates, type of area, details about F&B product advertised, and price/special offers. Scottish Index of Multiple Deprivation was used to identify bus stop area level of socioeconomic deprivation.

Bus shelters

Advertised food categories include fast food outlets (39%), confectionary, coffee, food stores and fruit juice.

There were no ads for fruit or vegetables (besides juices), water or low sugar beverages.

PROGRESS: Across all food categories, there were no associations between increased prevalence of these ads and deprivation level. “for example, fast food outlet advertisements were no more likely to be present in lower versus higher socioeconomic areas (B = 0.248, 95% CI: −0.082, 0.578, p = 0.140).

Note: additional food data was not extracted due to being combined with alcoholic beverages

Sainsbury et al. (2017) [100]

To examine the extent of F&B advertising on the Sydney metropolitan train network, and to assess the nutritional quality of advertised products

178 train stations on the Sydney metropolitan train network, Australia – (6931 ads)

Repeated cross-sectional, observational, content analysis

All stations on the Sydney metropolitan network were surveyed for ads and grouped by SES according to postal area. Data was collected over one week in summer and one week in winter. All ads were recorded and photographs taken, information collected included on product name and description, brand/company, location and format. All were classed as food or non-food and food ads classed as core, discretionary (not for daily consumption) or miscellaneous (e.g. tea or brand only). These groups were further divided into 32 product subcategories. Ads were also coded into size categories.

Any commercial billboard, poster, flyers, branded furniture, vending machine and experiential displays promoting a product, service or brand - specifically the train station concourse, station platform, cross-track billboards or any external ads designed to be seen by commuters standing on the platform or entering/exiting the station. For rotating billboards, all ads shown within one complete rotation were recorded

Approximately 25.9% of ads (1795/6931) promoted food and non-alcoholic beverages. There was no significant difference between seasons.

The majority of F&B ads were for discretionary products (78.3% vs 8%, without alcohol, calculated from available data), with more discretionary foods (e.g. biscuits) in summer (p = 0.01) and more core foods (e.g. vegetable soup/water) in winter (p = 0.04).

The most frequent food categories advertised were Snack foods (25%) SSBs (23%), intense sweetened beverages (18.7%), bottled water (6.8%) and high fat savoury biscuits, sweet biscuits, cakes, muffins, pastries and pies (3.8%).

Just over 50% of food ads were on vending machines. Coca-Cola contributed 10.9% of F&B ads, PepsiCo 6.5%. 6.3% of F&B ads were brand only.

PROGRESS: There was no significant difference in the number of core ads between SES groups

Note: additional food data was not extracted due to being combined with alcoholic beverages

Settle et al. (2014) [101]

To examine the prevalence of outdoor food advertising at tram, bus and train public transit stops across the least and most socioeconomically disadvantaged areas of Melbourne, Australia

20 km radius of Deakin university, Geelong, Australia – 20 suburbs (10 from most and 10 from least disadvantaged areas

(233 ads over 558 transit stops)

Cross-sectional, observational, content analysis

Auditing was undertaken by a single person, who visited each of the transit stops. Data collectors classified foods ads into 9 categories and recorded whether they were brand or food ads as well as location of ads, if any ambiguity, photos were taken and additional information recorded. Population density obtained using census data.

All train stations and bus and tram stops where a shelter was present

Ads were identified for cold beverages (including alcohol), hot beverages, snack foods, fast food, dairy and other food. There were no ads for cereal or fruit/vegetable items.

PROGRESS: Broader product categories did not differ except for hot beverages (there were more in least-disadvantaged areas - 13% vs 2%, p = 0.004).

37% of ads were for convenience stores in least disadvantaged areas (compared to 15% in most disadvantaged, p = 0.026).

48% of ads for stores were for fast food restaurants in most-disadvantaged areas (compared to 19% in least-disadvantaged, p = 0.003). The proportion of ads for local restaurants was not significantly different by level of disadvantage.

Ads for diet soft drink were more frequently observed in the least-disadvantaged areas compared to the most-disadvantaged areas (57% vs. 25%, p = 0.002). Ads for flavoured milk (8% vs 25%, p = 0.028) and fruit juice (2% vs 20%, p = 0.004) were more frequent in most disadvantaged areas.

Note: additional food data was not extracted due to being combined with alcoholic beverages

Signal et al. (2017) [102]

To examine the frequency and nature of New Zealand children’s everyday exposure to F&B marketing across multiple media and settings. Marketing exposure was examined by SES and ethnicity.

Wellington region of New Zealand

Cross-sectional, observational, content analysis

168 year 8 children (aged 11–13)

Following four days of exposure, cameras were collected and images downloaded (images captured every 7 seconds). Children reviewed and deleted any photos they wished to before researchers viewed them. Age and BMI of children was obtained. Foods were classified as recommended or not recommended to be marketed to children (WHO NPM)

Street, shop front, shopping mall, private transport, public transport facility, onboard public transport and other retail. Images in convenience stores/supermarkets were excluded

In outdoor spaces, children were exposed to an average of 0.9 core F&B ads per day (7% of total) and 8.3 non-core F&B ads per day (30.4% of total).

The most frequent advertisements were shop front (3.6 per day (core and non-core)) followed by street (2.2 per day).

Note: additional data was not extracted due to being combined with non-outdoor ads

Timmermans et al. (2018) [103]

To explore and define socio-economic (SES) differences in urban school food environments in The Netherlands

400 m buffers around 21 secondary schools in Utrecht, the Netherlands (350 ads)

Cross-sectional, observational, content analysis

Researchers visited the buffer areas. For every ad, a picture was taken with a GPS tag. Food ads were categorised into ready to eat/vs foods that need to be prepared and then categorised into healthy/unhealthy. Neighbourhood SES was obtained from The Netherlands Institute for Social Research

Wall posters, banners, bus-stop ads, flags, free-standings signs. Stationary delivery vehicles promoting retail outlets, food brands, products or meals were included. Shop windows were excluded

40.95% of ads contained food products (143/350). There were on average 6.81 ads per school, however only 14/21 schools had ads in surrounding areas (with an average of 11.14 ads per school), 7 schools had none.

58% of ads were classed as unhealthy (13.7% healthy).

The most frequently advertised food categories were other - not ready to eat (26.6%), savoury pastries (13%), SSBs (11.2%), hamburgers/kebabs (9.94%) and sweet and savoury snacks (7.5%).

PROGRESS: SSBs (22.5% vs. 6%, p = 0.02), hamburgers and kebabs (22.5% vs. 7.7%, p = 0.04), diet soft drinks (7.5% vs 1.1%, p ≤ 0.05), vegetable snacks (10% vs 2.2%, p ≤ 0.05) and dairy with no added sugar (7.5% vs 1.1%, p ≤ 0.05) were more frequently advertised in lower SES school environments. No food groups were significantly more advertised in high SES school environments.

Trapp et al. (n.d.) [104]

Generate robust evidence describing, benchmarking and quantifying the volume and nature of all outdoor food advertising within a 500 m radius of Perth schools

500 m buffers around 64 schools in Perth, Australia, (5636 ads)

Cross-sectional, observational, content analysis

Buffer areas were scanned on foot for all outdoor ads. Tablets were used to enter data and photos of ads as well as geolocation to determine distance to school boundary. Ads were classed as small, medium or large and food or non-food. Food ads classed into core, non-core, miscellaneous and branding only.

Billboards, posters/banners, free-standing signs, painted walls, digital/LED billboards, Merchandising - outside food shops, non-food shops, road, building, bus shelter or train station

1993 food products were advertised across 1708 food adverts (total includes alcohol). Unhealthy foods were featured 1444 times, healthy foods 315 times and miscellaneous 234 times.

Most frequently advertised foods were fast food (pictured in 14% of food ads), SSBs (13%), tea and coffee (11%), high fat/salt meals (10%) and ice cream and iced confection (7%).

K-12 schools had a significantly higher proportion of healthy foods within 250 m (60%) compared to primary (33%) and secondary (30%).

PROGRESS: Schools located in low SES areas had a significantly higher proportion of unhealthy food (excluding alcohol) ads within 250 m (40%) compared to high SES areas (30%).

Note: additional food data was not extracted due to being combined with alcoholic beverages

Walton et al. (2009) [105]

To document the community food environment surrounding case study primary schools, and second, to consider whether aspects of the community food environment impact on the food environment within schools.

2 km buffers around 4 schools in the Wellington region of New Zealand (79 ads)

Cross sectional, case comparison, content analysis

Data collected by driving and walking the streets buffer areas. Locations of ads were captured and a photo taken at each location. Advertised products categorised as “everyday”, “sometimes” and “occasional”. Deprivation level defined by School’s decile rating (from Ministry of Education).

Attached to food outlets

63.3% (50/79) of ads were for “occasional” products, 22.8% (18) “sometimes” and 13.9% (11) “everyday”.

PROGRESS: School 4 (mid-low deprivation) had no ads.

School 1 (highest deprivation) had 28 total ads, 20 were “occasional”, 2 “everyday”. School 2 had 16 ads, 8 “occasional”, 1 “everyday”.

School 3 (mid-high deprivation) had 35 ads, 22 “occasional” and 8 “everyday”, 5 “sometimes”

Watson et al. (2021) [106]

To compare six nutrient profiling models for suitability in food marketing to children regulation

Public transport property in Sydney, Australia (946 ads)

Cross-sectional survey

Researchers visited six transport hubs, photographed and recorded F&B advertising. Ads were classified into identifiable F&B, master brand, generic F&B, incidental F&B (non-food ad). Foods were classified as eligible or ineligible to be advertised

Advertising on buses and at train stations - on platforms, concourses and around entry gates.

964 ads were photographed, 176 unique. 150 F&B ads with 220 unique products were analysed.

Of identified products (n = 210), 175 (83%) were ineligible to be advertised 35 (17%) were eligible (by COAG - Council of Australian Government Health Council guide).

The most frequent unhealthy food categories were unhealthy meals (38.6%), SSBs (19%), dessert/ice cream (8.6%), savoury snacks (6.2%) and confectionary (3.8%).

The proportion of foods eligible to be advertised ranged from 10 to 28% depending on model: Health star rating, WHO western pacific, WHO Europe and NOVA food classification system.

Yancey et al. (2009) [67]

To examine whether African Americans, Latinos, and people living in low-income neighbourhoods are disproportionately exposed to ads for high-calorie, low nutrient–dense F&Bs and for sedentary entertainment and transportation and are relatively underexposed to advertising for nutritious F&Bs and goods and services promoting physical activities.

Selected zip codes in Los Angeles, Austin, New York City and Philadelphia.

(2233 ads)

Cross-sectional prevalence study

Researchers used GPS and digital cameras to create a record of outdoor ads. They collected data on types, number of, and size of ads (only those in English and Spanish). Data was coded by product content, and targeting of ads based on photographs. A fast food ad that featured two young men playing soccer would have been coded as both “fast food” and “physical activity”.

Minimum of 8 × 12 inches, virtually all billboards, bus shelter ads, bus bench ads, sidewalk sandwich signs, murals painted on the side of buildings, some store window posters. Not for a product or service specific to the premises on which the ad was located or the name of an establishment

Sugary beverages were featured in 4.1% of ads across all areas, fast food 2.6% and other food 8.2%.

One ad promoted fresh fruit/veg. Salads were pictured on 8 (13%) of fast food ads in LA, fresh fruit and veg was shown in one ad for a grocery delivery service in New York.

Note: additional food data was not extracted due to being combined with alcoholic beverages

Zenk et al. (2021) [68]

To evaluate long-term changes in store marketing practices two-years (24-months) post-tax implementation

Stores in Oakland compared to Sacramento, California, US

Repeated cross-sectional, observational, content analysis

Researchers completed in-store audits pre- and post-tax implementation. Price promotions (on sale or discounted price) were measured for SSBs, artificially sweetened beverages and unsweetened beverages.

Signs, posters, flags, decals, stickers, marquees and sandwich boards on the building exterior and property

31.4% of stores in Oakland had exterior SSB advertising (29% at supermarkets, 32.4% at limited service stores).

Post-tax implementation, 24.5% of stores had exterior advertising (12.9% at supermarkets, 29.6% at limited service stores).

Sacramento (no tax implemented) baseline = 37.3%, post-tax = 33.1%

For exterior advertising, no significant changes were found in Oakland relative to Sacramento for SSBs, ASBs or USBs.

  1. Abbreviations: F&B Food and beverage, Ads Advertisements, SSBs Sugar-sweetened beverages, ASBs Artificially sweetened beverages USBs Unsweetened beverages, WHO World Health Organization, SES socioeconomic status, SEP Socioeconomic position