Application of a geospatial query tool to characterize retail food environment and examine associations with dietary quality: evidence from three Chilean cities from the SALURBAL project

doi:10.21203/rs.3.rs-3962222/v1

Download PDF

Research Article

Application of a geospatial query tool to characterize retail food environment and examine associations with dietary quality: evidence from three Chilean cities from the SALURBAL project

https://doi.org/10.21203/rs.3.rs-3962222/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

Little work has been done in Latin American contexts to characterize food retail environments and assess its influence on dietary behaviours. In particular, food retail environment data do not exist for Chilean contexts. The aim of this study is to characterize the retail food environment across neighbourhoods in three major Chilean cities and explore possible associations between the retail food environment and fruit and vegetable consumption among its residents.

Methods

A geospatial query tool was used to identify, georeference, and classify food establishments (FE, n = 48400) across 3 large Chilean cities (Santiago, Valparaíso, Concepción). Neighbourhood characteristics came from the 2017 Census (n = 2442 neighbourhoods) and person level data came from the 2017 National Health Survey (Encuesta Nacional de Salud, n = 1275 adults). Lower dietary quality was proxied by non-daily consumption of a fruit or vegetable (ND-FV, prevalence 17.3%). Random intercept logistic regression models adjusted for survey participant age, gender, educational attainment, neighbourhood population density and social environment.

Results

Median density of FE was 41.36 per km^2 and the density of FE increased with neighbourhood population density and more favourable social environment. Ready-to-consume FE and small food retail were the most prevalent types (68% and 11%, respectively); and the adjusted odds of ND-FV consumption was approximately 20% higher with increases in the z-score of those FE types. No associations were found between other types of FE and ND-FV consumption.

Conclusion

These findings underscore the need for targeted public policies aimed at promoting healthier food environments, reducing social inequalities in food access within Chilean cities, and improving dietary quality.

Community food environment

Food retail

Urban health

Chile

Latin America

Dietary quality

There has been fairly strong evidence in high-income countries in the northern hemisphere that aspects of the food environment and food availability influence dietary behaviours (1–6). However, much less work on this topic has been done in Latin American contexts leaving gaps in understanding the community food environment itself, as well as associations between the food-environment and dietary practices (3).

The reasons why information from wealthy northern contexts may not be transportable to the southern hemisphere are related to the difference in types and space distribution of food retail, the target population reached by each establishment, the cultural relevance of the diverse food retails, and the categorization of establishments based on the healthiness of the products offered (7–12). For instance in northern countries, supermarkets, grocery stores and convenience stores are common, and the former is considered as a proxy for healthy food availability and the principal point of purchase of households (7). In contrast, in Latin American countries, supermarkets are characterized by the offer of affordable ultra-processed products (8), and household food purchase is split between retail outlets such as supermarkets, traditional markets, specialized stores (i.e. fruit shops), and small local shops (i.e. “tiendas de abarrotes” or “bodegas”) (9). In addition, traditional markets, small local shops, and street vendors are culturally relevant in Latin American countries, as they are part of the gastronomic culture of these countries (10, 11).

However in recent decades, significant changes have occurred in the community retail food environment globally (12, 13) and in Latin American contexts (12–15). Smaller food establishments are losing ground to larger international food companies (12, 14), leading to more ultra-processed foods (UPF) being commercialised (14). This has made unhealthy foods more available in local communities, characterizing a transition to less healthy diets with wide variations across regions and countries (14). This shift has health implications, as higher consumption of UPF is associated with a higher risk of overweight and obesity, all-cause mortality, metabolic syndrome, and depression in adults (16).

A major barrier to studying the community retail food environment is the difficulty of gathering relevant data. Collecting direct field data over large areas is costly, and in most Latin American contexts, a comprehensive data source of food establishments does not exist (3). Research on this topic in the region has mostly been small-scale, focused on limited sections of cities (2, 17) and typically requires observation data tools. Collecting direct field data over large areas is costly (3).

To address this limitation, we developed an innovative tool designed to automatically capture food retail data from Google Earth (GE). The tool demonstrated moderate to excellent validity and was previously reported in the Journal of Urban Health (18). The aims of this current study are: (1) apply the tool to extract information about the community food retail environment in three large Chilean cities; (2) characterize the retail food environment across neighbourhoods; (3) link the food environment data to a population based surveillance survey that collected data on fruit and vegetable consumption; and (4) explore possible associations between the retail food environment and fruit and vegetable consumption.

Study spatial units

The present study used data from three Chilean cities: Santiago, Valparaíso, and Concepción. The three cities were chosen because they have a high population, high population density, and high tourism and/or retail activity – characteristics under which the web data mining tool we used to extract retail food information performed well in prior studies (18). Within the three cities, there were a total of 2442 “zona censales”. “Zona censal” are analogous to a US census tract, on average include approximately 700 households, and were used to proxy ‘neighbourhood’ units.

This cross-sectional study used secondary data and was conducted as part of the SALURBAL project (SALud URBana en América Latina - Urban Health in Latin America) (19). The SALURBAL study protocol was approved by the Drexel University Institutional Review Board (IRB) with ID #1612005035 and by appropriate site-specific IRBs.

Community retail food environment data

We obtained information about the community retail food environment data from Google’s map data. We wrote a query tool that utilized Google Cloud services to search within a designated area for food retail points of interest (POI), which has been described in detail elsewhere (18). In brief, our POI search terms within Google’s food retailer type classification and establishment’s names. We determined the search terms using an inventory of types of food retail common in Latin America (20). Our query tool returned a list of 48400 food retailers, including their names, addresses, geographic coordinates, food retail type (the primary registered food retail type and up to seven additional retail types).

We then used a programmatic procedure to clean, de-duplicate and classify the retail establishments into seven categories (fruit and vegetable stores, fresh food retail, small food retail, supermarkets, ready-for-consumption food retail, convenience stores, fast food chains and candy and ice cream shops) and three health-related categories (most healthy, mixed, less healthy). Additional file 1 lists the seven categories and the rationale behind the definition of each food group. Less than 3% of the food establishments that we compiled were unable to be successfully classified (n = 1450). These establishments were retained in the total but are not represented when data are stratified by food establishment type (see Additional file 2).

Finally, food establishments densities were calculated at the neighbourhood level by dividing the number of food establishments per square kilometre area within each neighbourhood.

Neighbourhood census characteristics: population and social environment data

We utilized data from the 2017 Chilean census to establish neighbourhood population density and evaluate the social environment. Considering these two variables is crucial when studying the connections between food environment characteristics and unhealthy eating patterns.

Previous research has noted that neighbourhood s with higher population density tend to have a denser availability of food stores in Latin American countries (3, 21, 22). Similarly, the availability of total food stores, along with the distribution of specific types, varies based on the neighbourhood 's socioeconomic status (2, 23, 24).

Population density was calculated by dividing the population by the neighbourhood area (km²). To characterize the neighbourhood 's social environment, we used a summary index developed for the SALURBAL study, as detailed elsewhere (25). This index integrated variables related to sanitation, housing, employment, and education dimensions available for the neighbourhood s in our study. Variables were converted to z-scores and subsequently averaged. Higher values indicate a more favourable social environment within a neighbourhood .

Individual data

Individual level survey data came from year 2017 Encuesta Nacional de Salud (ENS, National Health Survey), a cross-sectional survey that is nationally representative of households in Chile. The survey is administered primarily in-person and collects information on a wide range of health behaviours and outcomes for surveillance and planning purposes. There were 1292 adults (aged > = 20) who completed the survey in the 3 cities. We used survey variables for age, gender, educational level (incomplete high school, completed high school or above), and fruit/vegetable consumption. We excluded 17 individuals missing one or more of these variables, leaving 1275 in the individual-level sample (see Additional file 2).

Dietary measures

Fruit/vegetable consumption was asked via the following questions: “In a typical week, how many days do you eat fruits (response options 0–7)?” and “In a typical week, how many days do you eat vegetables or vegetable salads (response options 0–7)? Don’t include potatoes or legumes.”. These variables were summed together and capped at 7 days to create a new variable corresponding to the number of days per week that fruits or vegetables are consumed. Additionally, we created a binary variable that reflected non-daily consumption of a fruit or vegetable (AKA 'FV') vs. daily, serving as a proxy for lower dietary quality. Participants were assigned to FV 'non-daily' if they reported consuming FV fewer than 7 days a week (vs 'daily,' which was 7 days a week)."

Statistical analyses

Descriptive statistics are presented for neighbourhood census characteristics and food establishments densities. We present the distribution of the neighbourhood data within each city (Tables 1 and 2), mean/median food establishment densities by census characteristics (Table 3) and the proportion of the different types of establishments investigated (see Additional file 3). Survey data on non-daily vs. daily consumption of FV was presented by demographic characteristics and food establishment’s densities and census characteristics (Table 4).

Table 1

Neighbourhood census characteristics, total and by city, Chile, 2017 (n = 2442 neighbourhoods).
	Overall	Santiago	Valparaíso	Concepción
Number of neighbourhood s ^a	2442	1,700	386	356
Population density (number of people / km^2)
Median	10365	12494	6802	5079
25th-75th percentile	5791–15252	8100–17535	3814–9759	3814–9759
Social environment index (z-score)
Median	0.31	0.46	0.13	-0.23
25th-75th percentile	-0.17–0.85	0.01–0.96	-0.47–0.62	-0.81–0.53
^a Neighbourhood s refer to Chilean 'zona censales.' The social environment index is a summary measure based on variables related to four dimensions (sanitation, housing, employment, education) assessed at the 'zona censal' level. Higher values indicate a better social environment within neighbourhood s.

Table 2

Retail food environment density at the neighbourhood-level (n = 2442) by city, Chile, 2017.
			Total	Santiago	Valparaíso	Concepción
Number of food establishments			48400	38255	4962	5193
Food establishments density (count / km^2)
	Total ^a
		Median	41.36	56.00	11.50	14.80
		25th-75th percentile	15.30–73.55	32.20–87.20	0–30.40	4.02–34.00
	Most Healthy
	Fruit and vegetables stores
		median	0.00	0.00	0.00	0.00
		25th-75th percentile	0–0	0–0	0–0	0–0
	Fresh food retail
		median	0.00	0.00	0.00	0.00
		25th-75th percentile	0–1.36	0–2.17	0–0	0–0
	Mixed
	Small food retail
		median	4.10	5.08	0.49	2.23
		25th-75th percentile	0–8.91	0.89–9.86	0–6.05	0–5.91
	Supermarkets
		median	2.41	3.48	0.00	0.61
		25th-75th percentile	0–6.11	0–7.28	0–2.39	0–3.67
	Ready-for-consumption food retail
		median	23.19	32.30	5.60	7.85
		25th-75th percentile	7.92–47.61	16.40–56.00	0–17.30	1.84–17.50
	Less Healthy
	Candy and ice cream shops
		median	0.00	0.00	0.00	0.00
		25th-75th percentile	0–0	0–0.74	0–0	0–0
	Fast-food chains
		median	0.00	0.00	0.00	0.00
		25th-75th percentile	0–0	0–0	0–0	0–0
	Chain convenience stores
		median	2.37	4.51	0.00	0.00
		25th-75th percentile	0–8.04	0–10.30	0–1.23	0–1.28
^a The row for ‘total’ includes all classifiable establishments n = 46950 + unclassifiable establishments n = 1450.

Table 3

Food establishments density at the neighboorhood-level (n = 2442) by population census characteristics, Chile, 2017.
		Population density/urbanicity density (per area km^2)				Intraurban social environment index (z-score)
		Q 1 (low)	Q 2	Q 3	Q 4 (high)	Q 1 (low)	Q 2	Q 3	Q 4 (high)
Food establishments density (establishments/km^2)
Total ^a
	mean	8.64	11.10	13.20	15.20	26.7	52.2	64.9	134.00
	median	4.32	15.80	15.80	15.80	16.2	45.3	53.7	61.6
Most Healthy
Fruit and vegetable stores
	mean	0.27	0.50	0.77	0.85	0.32	0.64	0.74	0.69
	median	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Fresh food retail
	mean	0.59	1.10	1.57	1.90	0.72	1.25	1.61	1.58
	median	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Mixed
Small food retail
	mean	2.77	6.30	6.90	10.00	3.83	6.28	7.10	8.74
	median	0.98	4.60	5.39	7.20	1.89	4.76	4.89	4.41
Supermarkets chains
	mean	1.53	3.49	4.97	8.66	2.59	4.32	4.94	6.80
	median	0.30	2.51	3.52	4.98	0.00	2.58	3.36	2.85
Ready-for-consumption food retail
	mean	19.50	37.10	52.00	81.00	15.10	30.70	40.90	103.00
	median	4.94	23.20	30.40	39.00	8.59	23.00	30.20	42.30
Less Healthy
Candy and ice cream shops
	mean	0.49	0.83	1.22	1.99	0.45	0.68	1.02	2.37
	median	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Fast-food chains
	mean	0.41	0.46	0.59	1.22	0.04	0.20	0.46	1.97
	median	0.00	0.00	0.00	0.00	0.00	0.00	0.00	0.00
Chain convenience stores
	mean	1.24	3.73	6.14	11.70	2.73	6.29	6.23	7.56
	median	0.00	2.31	4.40	8.49	0.00	3.95	4.19	2.30
^a The row for ‘total’ includes all classifiable establishments n = 46950 + unclassifiable establishments n = 1450.

Table 4

Individual (n = 1275) and neighbourhood-level (n = 319) characteristics by fruit and vegetable consumption frequency, Chile, 2017^a.
	Fruit and vegetable consumption frequency
	Total sample	Daily	Non-daily	p-value
Individual-level characteristics
Number of participants	1275	1055 (82.7)	220 (17.3)
Gender, n (%)^b
Female	785 (61.6)	903 (64.7)	102 (46.4)	< 0.001
Male	490 (38.4)	372 (35.3)	118 (53.6)
Age in years, median (IQR) ^c	50 (31–65)	52 (33–66)	38 (27–54)	< 0.001
Education, n (%)
< Primary	135 (10.7)	116 (11.0)	20 (9.1)	0.36
Primary	580 (45.5)	468 (44.4)	112 (50.9)
Secondary	393 (30.8)	331 (31.4)	62 (28.2)
University	166 (13.0)	140 (13.3)	26 (11.8)
Neighbourhood-level characteristics ^c
Population density (number of people / km^2)	9809.5 (6003.6–14443.3)	9751.5 (5987.6–14373.3)	10403.2 (6345.2–14484.6)	0.31
Intraurban social environment index (z-score)	0.2 (-0.2,0.6)	0.2 (-0.2,0.6)	0.1 (-0.3,0.6)	0.031
Total ^d	39.1 (18,63)	38.9 (18,63)	39.6 (18,62.6)	0.95
Most Healthy
Fruit and vegetable stores	0 (0,0)	0 (0,0)	0 (0,0)	0.86
Fresh food retail	0 (0,1.5)	0 (0,1.6)	0 (0,1.5)	0.65
Mixed
Supermarkets chains	2.8 (0,5.4)	2.8 (0,5.4)	2.9 (0,5.4)	0.74
Small food retail	4.8 (1.8,8.9)	4.6 (1.7,8.6)	5.3 (2.6,9.3)	0.08
Ready-for-consumption food retail	19.8 (8.4,40.1)	20.2 (8.4,40.4)	18.3 (8.5,36.7)	0.88
Less Healthy
Candy and ice cream shops	0 (0,0)	0 (0,0.1)	0 (0,0.5)	1.00
Fast-food chains	0 (0,0)	0 (0,0)	0 (0,0)	0.98
Chain convenience stores	2.6 (0,7.9)	2.7 (0,7.9)	2.3 (0,7.9)	0.60
^a 3 cities combined.
^b Chi-square test; ^c Mann-Whitney U Test.
^d The row for ‘total’ includes all classifiable establishments n = 46950 + unclassifiable establishments n = 1450.

To estimate the odds of non-daily FV consumption (outcome) according to food establishment’s densities (exposures), we used two-level mixed effect logistic regression models of individuals nested within neighbourhoods (random intercept for neighbourhoods). Model Akaike Information Criterion (AIC) guided our decisions regarding the best structure for modelling the nested structure of the data. We used a progressive modelling strategy to better understand how adjustment affects the association between food retail types and daily FV consumption (Table 5). We fitted three models for each food establishment exposure: (1) unadjusted single exposure model; (2) adjusted by person-level confounders (age, gender, and individual educational level); and (3) model 2 variables plus census characteristics (population density and intra urban social environment index). To aide interpretation, all independent variables were standardized to a mean of zero and a standard deviation of 1 (z-score).

Table 5

Association between non-daily fruit and vegetable consumption and neighbourhood-level retail food environment, Chile, 2017 (n = 1275)^a.
Food establishments density per area (count / km^2)		Model 1	Model 2	Model 3
Food establishments density per area (count / km^2)		OR (CI)
Total ^b		1.08 (0.94, 1.23)	1.09 (0.94, 1.26)	1.15 (0.97, 1.37)
Most healthy	Fruit and vegetable stores	0.94 (0.79, 1.10)	0.93 (0.78, 1.10)	0.94 (0.79, 1.12)
	Fresh food retail	1.07 (0.92, 1.23)	1.07 (0.92, 1.24)	1.10 (0.94, 1.28)
Mixed	Small food retail	1.15 (1.00, 1.31)	1.16 (1.01, 1.34)	1.20 (1.03, 1.40) ^c
	Supermarkets chains	1.01 (0.87, 1.17)	1.01 (0.86, 1.18)	1.03 (0.88, 1.20)
	Ready-for-consumption food retail	1.08 (0.94, 1.23)	1.10 (0.95, 1.27)	1.17 (1.1, 1.35) ^c
Less healthy	Candy and ice cream shops	0.98 (0.84, 1.15)	0.99 (0.84, 1.17)	1.02 (0.87, 1.20)
	Chain convenience stores	0.98 (0.84, 1.14)	0.96 (0.82, 1.13)	0.97 (0.82, 1.14)
	Fast-food chains	0.99 (0.85, 1.16)	1.0 (0.85, 1.16)	0.96 (0.80, 1.13)
^a3 cities combined.

Supplementary analyses assessed the proportion of total variation in daily non-FV consumption attributed to differences between neighbourhoods (see Additional file 4). We describe this variability via the Median Odds Ratios (MOR) which is an appropriate statistic for binary outcomes such as ours (non-daily FV consumption) (26). The MOR can be roughly interpreted as an Intraclass Correlation Coefficient used in a linear model.

All analyses were conducted in R programming language, version 4.2.2.

Neighbourhood characteristics

Among the three cities, Santiago had the highest population density (12494 people/km²), in contrast to Valparaíso and Concepción (6802 and 5079 people/km2, respectively, Table 1). Santiago also had a more favourable neighbourhood social environment score (0.46, compared to vs. 0.13 and − 0.23 for Valparaíso and Concepción, respectively).

Food establishments densities

We identified and georeferenced 48400 establishments and were able to classify 46950 of them (97% of all establishments). The majority (66.01%) of the classified establishments were ready-for-consumption food establishments, followed by small food establishments (10.76%), convenience establishments (7.47%), supermarkets (6.97%), fresh food (2.13%), candy (1.68%), FV (1.01%), and fast food chains (0.96%). Most of the establishments were in the health-related category ‘mixed’ (83.74%) with only 3.14% in the category ‘most healthy’ (a category that included FV establishments and fresh food retail) (see Additional file 3).

Patterns in neighbourhood food establishment density aligned with population density, with Santiago being the highest (56 establishments/km²) compared to densities less than 15 establishments km² in the other two cities (Table 2 and Table 3). Density of the ‘most healthy’ establishments was very low, with most neighbourhoods lacking a FV/fresh food establishment. Compared to Santiago, Valparaíso and Concepción had notably lower densities of “less healthy” food establishments - a category that included chain convenience establishments, which was the third most common type of establishment in Santiago (Table 2).

Food establishment densities also tended to be higher in areas with more favourable social environment index. For example, the median density per km² of all food establishments was 61.6 and 16.2 in highest and lowest quartiles of social environment, respectively (Table 3). Within the food establishment categories, the pattern was most evident for ready-for-consumption establishments: median density per km² was 42.3 and 8.59 for highest and lowest quartiles of social environment.

Survey participant residential environment and fruit and vegetable consumption

Table 4 shows results for individual-level analyses using the survey data (n = 1275). On average, the sample was 61.6% female and median age 50 years. The majority (56.2%) had less than secondary school education. Respondents lived in cities with a median population density of approximately 9810 persons per square kilometre and a median social environment z-score 0.2.

Among participants, 17.3% did not consume FV daily. Non-daily FV consumption was higher for males, younger people, and lower education. The availability of food establishments, categorized by healthiness, showed no significant variations based on consumption frequency (Table 4).

Exploratory associations between fruit and vegetable consumption and retail food environment

Most of the adjusted associations tested between food establishment density and odds of non-daily FV consumption were null (Table 5). The exception to this was for two food establishment categories that were the most prevalent in neighbourhoods: 'small food retail' (11% of the food establishments) and 'ready for consumption' (68% of the food establishments). Additionally, the association was strengthened after additional adjustment for potential confounders (comparing unadjusted model 1 to adjusted models 2 and 3). In the fully adjusted model, the odds of non-daily FV consumption was 20% and 17% higher with increases in the z-score of small food retail and ready-for-consumption food establishments, respectively (OR 1.20, 95% CI: 1.02, 1.40; and OR 1.17, 95% CI: 1.10, 1.37, results from model 3).

The MOR from the fully adjusted model was approximately 1.30, which in our context can very roughly be interpreted as 30% of the variability in non-daily FV consumption is between neighbourhoods within cities (see Additional file 4).

This study used a novel web data mining tool, utilizing Google Cloud services, to identify and characterize the retail food environment of three cities in Chile. It also investigated the association between the retail food environment and the non-daily consumption of fruits and vegetables among 1275 residents. We successfully identified, classified and georeferenced 48400 food establishments across the three cities. Neighbourhoods with higher population density and social environment had higher density of food establishments. Residing in neighbourhoods with higher densities of small food retail and ready-to-consume foods was associated with higher odds of non-daily FV consumption.

Our study found that most food establishments were in the health-related category "mixed" (signifying they sell healthier and unhealthier foods), a finding that aligns with patterns previously observed in other Latin American countries (21, 27). In particular, our work found the food establishment categories ‘small food establishments’ and ‘ready-for-consumption establishments’ were the most prevalent. The high prevalence of these types of establishments that sell UPF and high calorie ready-for-consumption meals, as opposed to establishments primarily selling healthy foods, may underlie insufficient fruit and vegetables intake among many Chileans (28) as well as escalating rates of overweight and obesity in Chilean urbanized areas over the past decade (29). Notably, Chile leads the region in ultra-processed product sales and ranks second only to Peru in the increase in daily per capita retail sales of ultra-processed products also in the last decade (30). In this context our findings provide valuable insights for the development of public health policies designed to prevent the displacement of fresh food markets by supermarkets (15).

The positive correlation we found between virtually all types of food retail and more favourable neighbourhood social environment also aligns with work from to other Latin American countries (3, 21). The high availability of all types of foods/beverages may be contribute to obesity patterns observed among those with higher socioeconomic status in Chile, where individuals of higher socioeconomic status are experiencing a faster increase in obesity rates compared to those in lower socioeconomic strata (29).

This paper presents a proof of concept for utilizing Google Cloud Services, specifically Google Earth, as a valuable tool for food environment surveillance in Latin America where usable food establishment data are scarce across large geographic areas (18). Traditional methods for food environment data collection through direct field observations can pose significant challenges, especially in low-middle-income countries such as Chile, given their labour-intensive and expensive nature (31). Additionally, relying solely on secondary data sources often necessitates validation studies (32), which can be a limiting factor in countries where comprehensive administrative databases are not readily available at the national level, as is the case in most Latin American countries (3). However, it's essential to acknowledge that even with the web tool, this effort can be time-consuming and expensive in part due to Google's query fee structure (18). Nevertheless, it represents a promising avenue for researchers and policymakers seeking to gather valuable insights into food access and distribution in areas where such data are lacking, ultimately facilitating evidence-based strategies for improving community food environments and public health.

Our study has several significant strengths. (1) We compiled a census of almost 50,000 food establishments in cities where no primary or secondary data on retail community food environment exist. (2) We described the neighbourhood context where the food establishments are located (population density and social environment). (3) We linked food environment data to survey participants in Chile's national health survey and examined associations between food environment and fruit and vegetable consumption.

However, our study also has noteworthy limitations. (1) We may have under- or over-identified food establishments; however measurement error may be fairly random. A prior validation study conducted in Brazilian cities compared records queried from Google Earth to a field audit of food establishments, and determined that under- or over-reporting did not differ by neighbourhood socioeconomic status in an urban Brazilian context (2). Google Earth is unable to detect street markets/street food vendors, which are an important food supply for FV and fish in Chile (33). It is possible that the lack of an association reported in this study between the healthiest food establishment category and FV consumption, was due to the absence data on informal food establishments. (3) There was a 4-year difference between the Google Earth data collection (August and November 2021) and dietary survey data (2017) and the retail environment could have changed. (4) Participants self-reported fruit/vegetable consumption based only on a couple of questions in the survey. Nevertheless, non-daily FV consumption aligned with expected patterns found in population based dietary surveillance (for example higher among males, younger people, lower educated) (34).

In conclusion, our study highlights significant disparities in food establishment densities across Chilean neighbourhoods, with higher densities of unhealthy food establishments and a scarcity of healthier food establishments. In our study, nearly 20% of adults did not consume a fruit or vegetable every day. We found that residing in neighbourhoods with higher densities of small food establishments and ready-for-consumption food establishments was associated to a higher likelihood of not consuming fruits and vegetables daily. These findings underscore the need for targeted public policies aimed at promoting healthier food environments, reducing social inequalities in food access within Chilean cities, and increasing fruit and vegetable consumption. Policymakers in Chile can use our characterization of food environments to enhance local food commerce, promote healthier food access, and address social inequalities in food access and fruit/vegetable consumption in urban communities.

Ethics approval and consent to participate

This study was conducted in accordance with the guidelines in the Declaration of Helsinki, and all procedures involving human subject data were approved by the Drexel University Institutional Review Board (protocol code 1612005035, approval date February 6, 2017). We didn’t obtain informed consent because we used secondary data obtained as public dataset.

Consent for publication

Not applicable.

Availability of data and materials

This study used data from SALURBAL project. These data are not publicly available. The SALURBAL project welcomes queries from anyone interested in learning more about its dataset and potential access to data. To learn more about SALURBAL’s dataset, visit https://drexel.edu/lac/ or contact the project at [email protected]. The query script written used by this paper can be made available for academic research and non-commercial purposes and is provided ‘as-is’ upon request; please contact [email protected].

Competing interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper

Funding

The Salud Urbana en América Latina (SALURBAL)/ Urban Health in Latin America project is funded by the Wellcome Trust [205177/Z/16/Z].

LVB is supported by the 'Grade 10 Scholarship Program' of the Rio de Janeiro State Research Support Foundation (FAPERJ). ICSJ is supported through a doctoral scholarship from the Academic Excellence Program of the Coordination for the Improvement of Higher Education Personnel (CAPES). LVB also held a fellowship from CAPES for part of the duration of the project. LOC received funding for data collection from The Brazilian National Council for Scientific and Technological Development (grant number: CNPq – 401787/2015–0).

The funding sources had no involvement in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.

Authors' contributions

LOC and MCM conceived and contributed to the design of the study. MFP, DAS, and VPM developed the query tool for data acquisition from Google Maps and implemented a programmatic procedure to process and classify food establishments. LVB designed the study, curated, analyzed and interpreted the data, and drafted the manuscript. DCRSO and AHA interpreted the data and contributed to the manuscript drafting. All authors critically revised the manuscript and approved the final version.

Acknowledgements

The authors acknowledge the contribution of all SALURBAL project team members. For more information on SALURBAL and to see a full list of investigators see, https://drexel.edu/lac/salurbal/team/ .

SALURBAL acknowledges the contributions of many different agencies in generating, processing, facilitating access to data or assisting with other aspects of the project. Please visit https://drexel.edu/lac/data-evidence for a complete list of data sources. The findings of this study and their interpretation are the responsibility of the authors and do not represent the views or interpretations of the institutions or groups that compiled, collected, or provided the data. The use of data from these institutions does not claim or imply that they have participated in, approved, endorsed, or otherwise supported the development of this publication. They are not liable for any errors, omissions or other defect or for any actions taken in reliance thereon.

The authors also acknowledge Google Earth for use of its content. Google expressly disclaims the accuracy, adequacy or completeness of any data and shall not be liable for any errors, omissions or other defects in, delays or interruptions in such data, or for any actions taken in reliance thereon. Mention of trade names, commercial practices or organizations does not imply endorsement by the authors, the institutions where the authors work or funding entities.

The authors would like to thank Uriel Moreira Silva for their assistance with the social environment data.

Chor D, Cardoso LO, Nobre AA, Griep RH, Fonseca MDJM, Giatti L, et al. Association between perceived neighbourhood characteristics, physical activity and diet quality: results of the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil). BMC Public Health. 2016;16(1):751. 10.1186/s12889-016-3447-5.
Duran AC, Diez Roux AV, Latorre MDRDO, Jaime PC. Neighbourhood socioeconomic characteristics and differences in the availability of healthy food stores and restaurants in Sao Paulo, Brazil. Health Place. 2013;23:39–47. 10.1016/j.healthplace.2013.05.001.
Pérez-Ferrer C, Auchincloss AH, De Menezes MC, Kroker-Lobos MF, Cardoso LDO, Barrientos-Gutierrez T. The food environment in Latin America: a systematic review with a focus on environments relevant to obesity and related chronic diseases. Public Health Nutr. 2019;22(18):3447–64. 10.1017/S1368980019002891.
Pinheiro AC, Quintiliano-Scarpelli D, Flores JA, Álvarez C, Suárez-Reyes M, Palacios JL, et al. Food Availability in Different Food Environments Surrounding Schools in a Vulnerable Urban Area of Santiago, Chile: Exploring Socioeconomic Determinants. Foods. 2022;11(7):901. 10.3390/foods11070901.
Vedovato GM, Trude ACB, Kharmats AY, Martins PA. Degree of food processing of household acquisition patterns in a Brazilian urban area is related to food buying preferences and perceived food environment. Appetite. 2015;87:296–302. 10.1016/j.appet.2014.12.229.
Westbury S, Ghosh I, Jones HM, Mensah D, Samuel F, Irache A, et al. The influence of the urban food environment on diet, nutrition and health outcomes in low-income and middle-income countries: a systematic review. BMJ Glob Health. 2021;6(10):e006358. 10.1136/bmjgh-2021-006358.
Cantor J, Beckman R, Collins RL, Dastidar MG, Richardson AS, Dubowitz T. SNAP Participants Improved Food Security And Diet After A Full-Service Supermarket Opened In An Urban Food Desert: Study examines impact grocery store opening had on food security and diet of Supplemental Nutrition Assistance Program participants living in an urban food desert. Health Aff. 2020;39(8):1386–94. 10.1377/hlthaff.2019.01309.
Mendes C, Miranda L, Claro R, Horta P. Food marketing in supermarket circulars in Brazil: An obstacle to healthy eating. Prev Med Rep. 2021;21:101304. 10.1016/j.pmedr.2020.101304.
Ortiz-Hernández L, Romo-Avilés M, Rosales Chavez JB. Main Retailers In Which Mexican Households Acquire Their Food Supply. JNBE. 2022;54(8):718–27. 10.1016/j.jneb.2022.04.213.
Meza-Hernández M, Durán-Galdo R, Torres-Schiaffino D, Saavedra-Garcia L. Contenido de sodio en preparaciones de consumo frecuente fuera del hogar en tres zonas de Lima Metropolitana, Perú. Rev Peru Med Exp Salud Publica. 2023;340–7. 10.17843/rpmesp.2023.403.12939.
Rosales Chavez JB, Bruening M, Ohri-Vachaspati P, Lee RE, Jehn M. Street Food Stand Availability, Density, and Distribution Across Income Levels in Mexico City. IJERPH. 2021;18(8):3953. 10.3390/ijerph18083953.
Alexander E, Yach D, Mensah GA. Major multinational food and beverage companies and informal sector contributions to global food consumption: implications for nutrition policy. Glob Health. 2011;7(1):26. 10.1186/1744-8603-7-26.
Baker P, Machado P, Santos T, Sievert K, Backholer K, Hadjikakou M, et al. Ultra-processed foods and the nutrition transition: Global, regional and national trends, food systems transformations and political economy drivers. Obes Rev. 2020;21(12):e13126. 10.1111/obr.13126.
Monteiro CA, Moubarac J-C, Cannon G, Ng SW, Popkin B. Ultra‐processed products are becoming dominant in the global food system. Obes Rev. 2013;14(S2):21–8. 10.1111/obr.12107.
Popkin BM, Adair LS, Ng SW. Global nutrition transition and the pandemic of obesity in developing countries. Nutr Rev. 2012;70(1):3–21. 10.1111/j.1753-4887.2011.00456.x.
Lane MM, Davis JA, Beattie S, Gómez-Donoso C, Loughman A, O’Neil A, et al. Ultraprocessed food and chronic noncommunicable diseases: A systematic review and meta‐analysis of 43 observational studies. Obes Rev. 2021;22(3):e13146. 10.1111/obr.13146.
Pessoa MC, Mendes LL, Gomes CS, Martins PA, Velasquez-Melendez G. Food environment and fruit and vegetable intake in a urban population: A multilevel analysis. BMC Public Health. 2015;15(1):1012. 10.1186/s12889-015-2277-1.
De Menezes MC, De Matos VP, De Pina MDF, De Lima Costa BV, Mendes LL, Pessoa MC, et al. J Urban Health. 2021;98(2):285–95. 10.1007/s11524-020-00495-x. Web Data Mining: Validity of Data from Google Earth for Food Retail Evaluation.
Diez Roux AV, Slesinski SC, Alazraqui M, Caiaffa WT, Frenz P, Jordán Fuchs R, et al. A Novel International Partnership for Actionable Evidence on Urban Health in Latin America: LAC-Urban Health and SALURBAL. Global Chall. 2019;3(4):1800013. 10.1002/gch2.201800013.
De Menezes MC, Pereira De Castro Junior PC, Kroker-Lobos MF, Ferrer CP, Tumas N, De Castro IRR, et al. Is it appropriate to import existing food retail environment definitions for the Latin American context? A systematic search and expert knowledge. Cities Health. 2023;7(1):46–58. 10.1080/23748834.2022.2112502.
Cardoso L, de O, Oliveira DCRS, de Menezes MC, Guimarães JMN, Meireles AL, Ramírez-Toscano Y et al. The role of social and built inequities on the community food environment in Mexican cities. Health Place. in press.
Justiniano ICS, De Menezes MC, Mendes LL, Pessoa MC. Retail food environment in a Brazilian metropolis over the course of a decade: evidence of restricted availability of healthy foods. Public Health Nutr. 2022;25(9):2584–92. 10.1017/S1368980022000787.
Macdonald L, Olsen JR, Shortt NK, Ellaway A. Do ‘environmental bads’ such as alcohol, fast food, tobacco, and gambling outlets cluster and co-locate in more deprived areas in Glasgow City. Scotland? Health Place. 2018;51:224–31. 10.1016/j.healthplace.2018.04.008.
Leite MA, Assis MMD, Carmo ASD, Costa BVDL, Claro RM, Castro IRD, et al. Is neighbourhood social deprivation in a Brazilian city associated with the availability, variety, quality and price of food in supermarkets? Public Health Nutr. 2019;22(18):3395–404. 10.1017/S1368980019002386.
Vaz CT, Coelho DM, Silva UM, Andrade ACDS, López FG, Dueñas OLS, et al. Social environment characteristics are related to self-rated health in four Latin America countries: Evidence from the SALURBAL Project. Health Place. 2023;83:103110. 10.1016/j.healthplace.2023.103110.
Merlo J. A brief conceptual tutorial of multilevel analysis in social epidemiology: using measures of clustering in multilevel logistic regression to investigate contextual phenomena. J Epidemiol Community Health. 2006;60(4):290–7. 10.1136/jech.2004.029454.
Pérez-Ferrer C, Auchincloss AH, Barrientos-Gutierrez T, Colchero MA, De Oliveira Cardoso L, De Carvalho M, et al. Longitudinal changes in the retail food environment in Mexico and their association with diabetes. Health Place. 2020;66:102461. 10.1016/j.healthplace.2020.102461.
Vega-Salas MJ, Caro P, Johnson L, Papadaki A. Socio-economic inequalities in dietary intake in Chile: a systematic review. Public Health Nutr. 2022;25(7):1819–34. 10.1017/S1368980021002937.
Sepúlveda-Peñaloza A, Cumsille F, Garrido M, Matus P, Vera-Concha G, Urquidi C. Geographical disparities in obesity prevalence: small-area analysis of the Chilean National Health Surveys. BMC Public Health. 2022;22(1):1443. 10.1186/s12889-022-13841-2.
Pan American Health Organization. Ultra-processed food and drink products in Latin America: trends, impact on obesity, policy implications. PAHO. 2015. https://iris.paho.org/bitstream/handle/10665.2/7699/9789275118641_eng.pdf. Accessed 15 Oct 2023.
Lucan SC. Concerning Limitations of Food-Environment Research: A Narrative Review and Commentary Framed around Obesity and Diet-Related Diseases in Youth. J Acad Nutr Diet. 2015;115(2):205–12. 10.1016/j.jand.2014.08.019.
Fleischhacker SE, Evenson KR, Sharkey J, Pitts SBJ, Rodriguez DA. Validity of Secondary Retail Food Outlet Data. Am J Prev Med. 2013;45(4):462–73. 10.1016/j.amepre.2013.06.009.
Food and Agriculture Organization – FAO; Oficina de Estudios y Políticas Agrarias – ODEPA; Confederación Gremial Nacional de Organizaciones de ferias libres, persas y afines - ASOF. Características Económicas y Sociales de Ferias Libres de Chile. 2013. https://www.asof.cl/wordpress/wp-content/uploads/2013/04/INFORME-ESTRUCTURAL-ENFEL-2013-final-1.pdf. Accessed 15 Oct 2023.
Dehghan M, Akhtar-Danesh N, Merchant AT. Factors associated with fruit and vegetable consumption among adults: Fruit and vegetable consumption. J Hum Nutr Diet. 2011;24(2):128–34. 10.1111/j.1365-277X.2010.01142.x.

No competing interests reported.

Download PDF

Reviews received at journal
09 Oct, 2024
Reviews received at journal
03 Oct, 2024
Reviews received at journal
30 Sep, 2024
Reviewers agreed at journal
24 Sep, 2024
Reviewers agreed at journal
23 Sep, 2024
Reviewers agreed at journal
20 Sep, 2024
Reviewers agreed at journal
09 Sep, 2024
Reviewers agreed at journal
06 Sep, 2024
Reviewers invited by journal
05 Sep, 2024
Editor invited by journal
29 Feb, 2024
Editor assigned by journal
29 Feb, 2024
Submission checks completed at journal
26 Feb, 2024
First submitted to journal
16 Feb, 2024

You are reading this latest preprint version

Application of a geospatial query tool to characterize retail food environment and examine associations with dietary quality: evidence from three Chilean cities from the SALURBAL project

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Background

Methods

Study spatial units

Community retail food environment data

Neighbourhood census characteristics: population and social environment data

Individual data

Dietary measures

Statistical analyses

RESULTS

Neighbourhood characteristics

Food establishments densities

Survey participant residential environment and fruit and vegetable consumption

Exploratory associations between fruit and vegetable consumption and retail food environment

Discussion

Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1