Hostname: page-component-848d4c4894-ttngx Total loading time: 0 Render date: 2024-05-30T07:17:43.832Z Has data issue: false hasContentIssue false

Food environment trajectories: a sequence analysis from the CARTaGENE cohort

Published online by Cambridge University Press:  22 January 2024

Habila Adamou
Affiliation:
Center for Research in Regional Planning and Development (CRAD), Laval University, Quebec, Canada Evaluation Platform on Obesity Prevention, Quebec Heart and Lung Institute Research Center, Quebec, Canada
Éric Robitaille
Affiliation:
Institut national de santé publique du Québec, 190, boulevard Crémazie Est, Montréal, Québec, Canada ESPUM, Université de Montréal, Montréal, Canada
Marie-Claude Paquette*
Affiliation:
Institut national de santé publique du Québec, 190, boulevard Crémazie Est, Montréal, Québec, Canada Département de nutrition, Université de Montréal, Montréal, Canada
Alexandre Lebel
Affiliation:
Center for Research in Regional Planning and Development (CRAD), Laval University, Quebec, Canada Evaluation Platform on Obesity Prevention, Quebec Heart and Lung Institute Research Center, Quebec, Canada
*
*Corresponding author: Email marie-claude.paquette@inspq.qc.ca
Rights & Permissions [Opens in a new window]

Abstract

Objective:

The purpose of this study was to create a typology of longitudinal exposure to food environment based on socio-economic context.

Design:

Food environment trajectories were modelled using a sequence analysis method, followed by a logistic regression to describe those trajectories.

Setting:

The study took place in Quebec, Canada, using food environment data from 2009, 2011 and 2018 merged with participants’ demographic and socio-economic characteristics.

Participant:

At recruitment, 38 627 participants between the ages of 40 and 69 years from six urban areas in Quebec were included in the CARTaGENE cohort study. The cohort was representative of the Quebec urban population within this age range.

Results:

Our study revealed five trajectories of food access over time: (1) limited access to food stores throughout the study period, (2) limited access improving, (3) good access diminishing, (4) good access throughout the period and (5) low access throughout the period. Logistic regression analysis showed that participants who were unable to work (OR = 1·42, CI = 1·08–1·86), lived in households with five or more persons (OR = 1·69, CI = 1·17–2·42) and those living in low-income households (OR = 1·32, CI = 1·03–1·71) had higher odds of experiencing a disadvantaged food environment trajectory. Additionally, the level of education and age of participants were associated with the odds of experiencing a disadvantaged food environment trajectory.

Conclusions:

The study demonstrates that people facing socio-economic disadvantage are more likely to experience a disadvantaged food environment trajectory over time.

Type
Research Paper
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2024. Published by Cambridge University Press on behalf of The Nutrition Society

Diet plays a major role in determining the health status of a population(1). In 2019, the most important risk factors associated with mortality and morbidity in Canada were smoking, high BMI, high blood pressure, high fasting glucose and poor diet(2). The adoption of a healthy diet depends not only on individual determinants (e.g. food preferences, nutritional knowledge and psychological factors) but also on environmental determinants such as the characteristics of the physical, economic, political and sociocultural environments (e.g. family context, physical and economic access to and availability of food, social status, and income)(Reference Walker, Keane and Burke3,Reference Raine4) .

The effects of the characteristics of the food environment (e.g. accessibility to food stores and food policies) on population health have received increased attention in recent years. Most studies have focused on associations between food environment characteristics and chronic disease(Reference Osei-Assibey, Dick and Macdiarmid5Reference Pérez-Ferrer, Auchincloss and de Menezes8), diet quality(Reference Needham, Orellana and Allender9), the quality of the food supply(Reference Boone-Heinonen, Gordon-Larsen and Kiefe10Reference Durette and Paquette13), and fruit and vegetable consumption(Reference Boone-Heinonen, Gordon-Larsen and Kiefe10,Reference Durette and Paquette13) . Of the chronic diseases, obesity is most often used to measure the impact of the food environment on population health(Reference Chennakesavalu and Gangemi7,Reference Needham, Orellana and Allender9,Reference Holsten14Reference Cobb, Appel and Franco16) . Other studies have explored the links between neighbourhood socio-economic status and the food environment to which individuals are exposed(Reference Filomena, Scanlin and Morland11,Reference Berger, Kaufman and Bader12,Reference Richardson, Meyer and Howard17) . Much of this research focuses on the food environment around the schools or homes of young people(Reference Osei-Assibey, Dick and Macdiarmid5,Reference Engler-Stringer, Le and Gerrard18Reference Yang, Zhang and Feng22) to explore its links with childhood obesity, diet quality and food supply quality.

The food environment is complex, and developing indicators to characterise it reliably is particularly challenging(Reference Seijo, Spira and Chaparro23Reference Widener25). Most measurements that exist can be grouped into three broad categories: availability, accessibility and quality of food offerings(Reference Lebel, Daepp and Block24,Reference Widener25) . The most common types of food sources used to develop these measurements are grocery stores, supermarkets and fast-food restaurants(Reference Lebel, Daepp and Block24). Food availability is usually measured using the number of food sources, the density per area or the ratio of food sources to people(Reference Filomena, Scanlin and Morland11,Reference Holsten14Reference Cobb, Appel and Franco16,Reference Engler-Stringer, Le and Gerrard18) . Accessibility is measured using Euclidean or network distances between food sources and the nearby residential locations, schools, or workplaces(Reference Boone-Heinonen, Gordon-Larsen and Kiefe10,Reference Durette and Paquette13,Reference Richardson, Meyer and Howard17,Reference James, Seward and James O’Malley26,Reference Mahendra, Polsky and Robitaille27) . Finally, studies most often measure the quality of food offered by analysing the food supply offered or by calculating a food quality index(Reference Osei-Assibey, Dick and Macdiarmid5,Reference Boone-Heinonen, Gordon-Larsen and Kiefe10,Reference Engler-Stringer, Le and Gerrard18) . Food sources are often identified using an existing classification of commercial stores or lists of specific store names(Reference Yang, Zhang and Feng22).

Most of these studies of food environment report weak associations with health indicators or diet quality(Reference Osei-Assibey, Dick and Macdiarmid5,Reference Cobb, Appel and Franco16,Reference Engler-Stringer, Le and Gerrard18) . Studies showing significant associations between the food environment and weight status or eating behaviours primarily investigated urban and low socio-economic environments(Reference Chennakesavalu and Gangemi7,Reference Holsten14) . Most used cross-sectional designs(Reference Chennakesavalu and Gangemi7) and did not control for the duration of exposure to a given food environment, which could explain the weak associations.

Understanding the relationship between food environments and population health may hinge on considering both the duration and the trajectory of exposure to these environments(Reference Swinburn, Egger and Raza28Reference Letarte, Pomerleau and Tchernof30). More recent studies elaborated new strategies to measure the exposition of individuals’ food environment trajectories in time. Many studies used longitudinal analysis for identifying trajectories of access to healthy food store type(Reference Richardson, Meyer and Howard17,Reference Needham, Strugnell and Allender31) , unhealthy food outlets(Reference Berger, Kaufman and Bader12,Reference Maguire, Burgoine and Monsivais32) or both(Reference Needham, Strugnell and Allender31,Reference Maguire, Burgoine and Monsivais32) . Other strategies aimed at understanding changes in the food environment over time by simultaneously measuring exposure to the home food environments, the workplace food environments and the food environment along home–work commutes(Reference James, Seward and James O’Malley26). The relationship between exposure and utilisation was also recognised to be influenced by the temporal and spatial context within which individuals encounter food retailers(Reference Liu, Widener and Smith29).

Results from these studies reported various findings for several health issues such as obesity or food consumption. However, all studies concluded to some extent that individuals residing in socio-economically disadvantaged conditions showed higher exposure to long-term unhealthy food environment trajectories and had a diminished supply of health-promoting foods in comparison to more affluent communities(Reference Rummo, Guilkey and Ng33Reference Pinho, Mackenbach and Den Braver35). Some further observed that the weekly consumption of fast food among individuals was linked to an unhealthy food environment and elevated fast-food restaurant density, especially within disadvantaged communities(Reference Boone-Heinonen, Gordon-Larsen and Kiefe10,Reference Rummo, Guilkey and Ng33) , but that this disparity may dissipate over time due to larger increases in proximity to fast food in wealthier neighbourhoods(Reference James, Seward and James O’Malley26).

However, all studies acknowledged that measuring exposure to the food environment is challenging and has limitations in comprehending the intricate relationship between food store availability and healthy eating. They often have shortcomings such as the reliance on inaccurate commercial databases for food establishment data, heterogeneity of geographic measurements or indexes used, and the availability of representative longitudinal individual-level data along with precise geographic information. Furthermore, despite many studies, associations between food environments and health are often inconsistent since results vary importantly according to political and socio-economic context. These limitations prevent policymakers from a clear description to address public health challenges related to the food environment. Longitudinal approaches adapted to political context are thus needed to orient policymakers to better address issues related to the food environment under their jurisdiction.

The objective of this study was to create a typology of longitudinal exposure to urban food environments by socio-economic context in Quebec.

Materials and methods

We utilise sequence analysis to 1) create food environment trajectories and a 2) typology of these trajectories. Additionally, logistic regression models were used to characterisze disadvantaged food environment trajectories based on demographic and socio-economic characteristics. In our study, logistic regression allowed us to estimate the likelihood of a participant being in a disadvantaged food environment trajectory according to demographic and socio-economic variables. All analyses were done in 2022.

Individual data: sample and variables

CARTaGENE is a publicly funded research platform that was developed in 2003 to facilitate health research and support decision-making in Quebec, Canada. The platform comprises a population-based cohort of 43 000 participants from six metropolitan areas in the province. The CARTaGENE cohort is an ongoing study that includes participants from both Phase A (2009–2010) and Phase B (2013–2014) recruitment periods. The study focuses on participants aged 40–69 years, who were representative of the urban Quebec population in this age group at the time of recruitment. CARTaGENE is the largest prospective study of adult health in Quebec and includes both biological samples and individual data. The platform aims to reduce healthcare costs and promote public health by providing a valuable tool for researchers and decision-makers. Data were collected on demographics and socio-economic characteristics, physical and mental health, nutrition, and living environments. Participants’ administrative data from Quebec’s health insurance plan (Régie de l’assurance maladie du Québec (RAMQ)) was combined with the CARTaGENE data. More information on the recruitment, development and data management of the CARTaGENE cohort is available in this reference(Reference Awadalla, Boileau and Payette36) and on the platform’s website https://cartagene.qc.ca/. Participants’ sex, age, the highest level of education, occupational status, marital status, annual household income and the number of dependents in the household were obtained from the CARTaGENE cohort data.

Participant selection

Food environment data

Data on the food environment were collected by Quebec’s public health institute (INSPQ, Institut national de santé publique du Québec). The data on the food environment in Quebec include information about the availability and proximity of food sources. The information is gathered based on the 2016 census dissemination area (DA), which is the smallest spatial unit in the Canadian census that provides socio-economic data(37). Our definition of food environment is based on a food store access index created by Quebec’s public health institute (INSPQ). This index is available for 2009, 2011 and 2018, and it measures the accessibility of food stores such as grocery stores, supermarkets, farmer’s markets, and fruit and vegetable shops. In our study, the food stores access index was categorised as: 1. food desert, 2. limited access to food stores and 3. favourable access to food stores. Accessibility of food stores was calculated using an area where the centre is geographically weighted according to residential distribution and the nearest food store. One kilometre or more is used to define low access to food stores in urban areas. A food desert is defined as a DA with low access to food stores which is also in the most materially disadvantaged quintile(Reference Robitaille and Bergeron38). For more detailed information on the food store access index, see Robitaille and Bergeron(Reference Robitaille and Bergeron38).

Our study involved matching the food store access index with the respective DA where participants lived in 2009, 2011 and 2018. This allowed us to obtain the food store access index for each participant in the CARTaGENE cohort for those years. The procedure is shown in Fig. 1.

Fig. 1 Sample selection process

Construction of food environment trajectories

To classify and differentiate imperceptible subgroups of sequences based on their reactions to a collection of detectable indicators (food access stores), we employ sequence analysis followed by optimal matching to determine the requisite conversions among the various modalities of food access stores. Ultimately, we employ inertia jumps to determine the number of classes to be chosen.

Sequence analysis is an exploratory classification methodology designed to unveil patterns in data(Reference Billari39Reference Robette41), providing a condensed representation of the sample. Sequence analysis, employed to discern sequence patterns, transitions and temporal trends, facilitates the identification of latent subgroups of individuals based on their responses to a set of observable indicators. This leads to trajectory construction. In our case, each trajectory is described by a sequence, that is, by a chronologically ordered sequence of elementary ‘access to food stores’. We use optimal matching to compare dissimilarities between sequence pairs. Then, hierarchical ascending classification to group sequences into several classes based on their proximity.

Optimal matching, a key approach, involves determining, for each pair of sequences, the minimum number of substitutions (where one element is replaced with another), deletions (where one element is removed) and insertions (where one element is added) needed to align them. In this study, optimal matching sequence analysis computed dissimilarity between sequence pairs in the sample(Reference Billari39Reference Robette41). Subsequently, a sequence typology was constructed, grouping similar sequences through hierarchical ascending classification, where costs were computed based on application-specific criteria.

Although attributing costs to social distance in the social sciences is challenging, a matrix of substitution costs was employed, where all costs were constant and set at 2 (Table 1). The calculated distance between sequences incorporated an insertion/deletion (indel) cost equal to 1(Reference Robette41,Reference Letarte, Gagnon and McKay42) . The primary goal was to ascertain whether the sequence order within trajectories justifies an indel value of 1.

R (R Core Team, 2019) TraMineR package(Reference Gabadinho, Ritschard and Müller43) facilitated sequence analyses. The classification iteratively grouped individuals with similar experiences in successive food environments from 2009 to 2018. The resulting information was presented as a dendrogram – a classification tree – where each level represented a subset of individuals. This dendrogram, based on inertia jumps, aided in determining the number of classes. Hierarchical ascending classification associated with Ward’s criterion was utilised for trajectory typologies, seeking to minimise heterogeneity within classes while maximising differences between classes. This approach identified five classes of food store access trajectories for CARTaGENE cohort participants between 2009 and 2018, denoted as food environment trajectories.

Results

Sample characteristics

Table 2 provides information on the characteristics of the participants based on the variables included in the analyses. Many of the participants were female (56 %) and married (67 %). Additionally, 47 % of the participants held a university degree, 70 % were employed and 56 % had favourable access to food stores in 2009, 52 % in 2011 and 49 % in 2018.

Table 1 Matrix of insertion, suppression and substitution costs between the three modalities of the classification variable

Table 2 Distribution of participants by variable included in the analyses

Table 3 Demographic and socio-economic characteristics distribution of participants by food environment trajectories

Description of the food environment trajectories

Results from the sequence class analysis revealed five types of access to food stores (Fig. 2). All participants fell into one of these five trajectories of access to food stores. Each food environment trajectory has its characteristics which are presented below.

Fig. 2 Food environment trajectory typologies. This figure is used for a better visualisation of the ten typical sequences of each class. In other words, the ten most frequent sequences in each class that are representative of the class

Fig. 3 Factors of exposure within trajectories characterised by food desert environments

Trajectory 1 – Limited access throughout: This food environment trajectory includes participants who experienced a stable food environment trajectory between 2009 and 2018 characterised by low access to food stores throughout the studied period. Nearly, all these participants in this trajectory lived in food environments with low access to food stores even if/when they moved.

Trajectory 2 – Limited access improving: Trajectory 2 includes participants who had low access to food stores initially, but for some participants, food access improved over time. Between 2011 and 2018, food store access of individuals in this trajectory oscillated between low access and favourable access. In 2009, nearly 65 % of people in this trajectory lived in areas with low access to food stores, while 5 % were in food deserts and 30 % were in areas with favourable access to food stores. In 2018, at the end of the observation period, 58 % lived in areas with low access to food stores, 6 % lived in an area considered a food desert and 36 % had favourable access to food stores.

Trajectory 3 – Good access diminishing: This food environment trajectory encompasses participants who initially experienced favourable access to food stores, only to witness a subsequent deterioration. Specifically, this category includes individuals residing in areas where access to food stores was initially favourable but underwent a decline after 2011 (Fig. 2). At the beginning of the observation period in 2009, 80 % of participants in this trajectory resided in DA with favourable access to food stores, reaching 100 % in 2011. However, by the conclusion of the observation period in 2018, nearly 60 % found themselves in areas characterised by low access to food stores.

Trajectory 4 – Good access throughout: This food environment trajectory is characterised by favourable access to food stores throughout the study period. It includes participants who experienced a stable food environment trajectory between 2009 and 2018, with nearly all of them living in environments with favourable access to food stores.

Trajectory 5 – Low access (food desert): Participants in this food environment trajectory remained in a food desert throughout the study period. This trajectory includes participants that had largely stable food environment trajectories between the three types of food environments. By 2011, almost all the participants that started in a food desert had better access to food stores in this trajectory. Conditions improved both for those who were in food deserts and those who started with low access to food stores (Table 2). This class includes a high proportion of women and people from households that earned less than 100 000 CAD$ per year (Table 3).

Demographic and socio-economic factors associated with the disadvantaged food environment trajectory

Out of the five food environment trajectories, two can be classified as disadvantaged. These are the food environment trajectory with low access to food stores (trajectory 5) and the food environment trajectory with limited access to food stores (trajectory 1). The only difference between the participants in these two trajectories was based on the material deprivation index. The participants in both trajectories lived in areas with low food access, whereas trajectory 5 participants were also living in materially deprived environments. To analyse the determinants of belonging to the disadvantaged environment trajectory, we created a trajectory that includes participants from the low access to food stores (food desert) group compared with the rest of the sample.

Disadvantaged food environment trajectory

Participants’ sex and marital status were not found to be associated with experiencing a disadvantaged food environment trajectory (Fig. 3). However, individuals who were divorced had slightly higher odds (OR = 1·01, adjusted CI = 0·80–1·97) of experiencing a disadvantaged food environment trajectory compared with those who were married. Participants who were 60 years of age and above had lower odds of experiencing a disadvantaged food environment trajectory when compared with those who were aged 40–44 years. The OR was 0·40 with a CI of 0·26–0·61. This suggests that increased age is associated with a decreased likelihood of having a disadvantaged food environment trajectory. The association between the current employment situation and the disadvantaged food environment trajectory was weak. The study found that unemployed participants (OR = 1·31 CI = 0·90–1·84) or participants who were unable to work (OR = 1·42 CI = 1·08–1·86) were more likely to experience a disadvantaged food environment trajectory than those who were employed. Inactive participants (OR = 1·24 CI = 0·80–1·72) were also more likely to experience a disadvantaged food environment trajectory compared with those who were employed.

There is a significant association between the disadvantaged food environment trajectory and the level of education, annual household income, and number of dependents in the household (P < 0·001). Having an annual household income of CAD 100 000 or more was found to decrease the likelihood of experiencing a disadvantaged food environment trajectory. For instance, individuals from households earning less than CAD 100 000 per year (OR = 1·42 CI = 1·08–1·86) had higher chances of experiencing a disadvantaged food environment trajectory compared with those from households earning more than CAD 150 000 per year. Additionally, the odds of experiencing a disadvantaged food environment trajectory increased when there were more than five persons in the household. Households with five or more people (OR = 1·69 CI = 1·17–2·46) had higher odds of experiencing a disadvantaged food environment trajectory compared with households with only one person. The level of education a person attains is related to their access to healthy food options. People with only elementary education had a higher odd (OR = 1·30 CI = 0·69–2·23) to experience a disadvantaged food environment trajectory compared with those with a high school education. However, those with a college-level education, graduate degrees or university certificates had a lower odd to experience a disadvantaged food environment trajectory compared with those with only a high school education.

Discussion

This study aimed to create a socio-economic-based typology of longitudinal exposure to the food environment from 2009 to 2018. We used sequence analysis to create five different food environment trajectories with varying levels of food access over a period of 9 years. While some participants had constant access to food stores, others experienced fluctuations between 2009 and 2018. We found that age, employment status, education level, number of dependents in the household, and household income and age of participants were the most significant determinants of a disadvantaged food environment trajectory.

Demographic factors related to the disadvantaged food environment trajectory

Between 2009 and 2018, the percentage of people living in food deserts with limited access to food increased from 43·45 % to 50·62 %, according to the CARTaGENE population-based cohort. This trend is also evident in many North American and European countries(Reference Richardson, Meyer and Howard17,Reference James, Seward and James O’Malley26,Reference Needham, Strugnell and Allender31,Reference Kelman, Pool and Gordon-Larsen44) . It is important to investigate the deterioration of food environments to better understand it and guide decision-makers in developing strong public policies to ensure food access for everyone. In this study, the use of sequence analysis allowed us to create longitudinal food environment indicators, which helped us understand the different trajectories of food store access in Quebec over time.

Demographic factors have been very rarely used to explain disparities in unhealthy food environment trajectories or food environment longitudinal indicators, making it difficult to compare our results with other studies. The CARTaGENE data provide us with this opportunity to assess the odds that participants have of belonging to the five types of food environment trajectories we created, based on their demographic characteristics rather than on community characteristics.

Our findings indicate that women in our cohort are more likely to reside in an unfavourable food environment trajectory. This implies that, through their residential trajectories, women have greater exposure to areas with limited access to food stores compared with men. However, this disparity is not statistically significant. A physical environment study conducted on the Multi-Ethnic Study of Atherosclerosis (MESA) cohort did not observe any discrepancy in access to healthy food environments between sexes, as seen through descriptive analyses(Reference Christine, Auchincloss and Bertoni45). Nevertheless, another study of the same cohort reports a weak correlation between the participants’ sex and the local food environment(Reference Moore, Diez Roux and Nettleton46). Our study found that age was the most significant factor associated with a disadvantaged food environment trajectory. As young people grew older, they were more likely to experience this trajectory. In fact, from the age of 40 years onwards, the likelihood of experiencing a disadvantaged food environment trajectory increased significantly. These findings are contrary to those reported in the 40-year food environment study of the Framingham Heart Study cohort. Researchers did not find a consistent relation between the sex and age of participants and access to a supermarket or fast-food outlets(Reference James, Seward and James O’Malley26).Finally, we found that married individuals generally had better access to healthy food options when compared with those who were divorced, separated or widowed. This trend was observed to be consistent with the number of dependents in the household, as households with more than five dependents had a higher probability of experiencing a disadvantaged food environment.

Socio-economic factors are related to the disadvantaged food environment trajectory

There have been several studies that indicate significant differences in the trajectories of unhealthy food environments between disadvantaged and affluent socio-economic communities(Reference Richardson, Meyer and Howard17,Reference James, Seward and James O’Malley26,Reference Swinburn, Egger and Raza28,Reference Needham, Strugnell and Allender31) . A previous study conducted in Australia revealed that irrespective of the area’s level of food access or dietary status, the food supply in poorer communities was less health-promoting as compared with that of their affluent counterparts in the long run(Reference Swinburn, Egger and Raza28). Our results further highlight the socio-economic inequalities based on the annual household income and the current employment status of participants. Participants who were unemployed, inactive or unable had higher chances of experiencing a disadvantaged food environment trajectory as compared with those who were employed. Furthermore, individuals belonging to households earning less than 100 000 $CAD annually were found to have a greater likelihood of experiencing a disadvantaged food environment trajectory as compared with those in higher-income households. It has been established that socio-economic disadvantage is linked with an unhealthy food environment over a long period(Reference Richardson, Meyer and Howard17,Reference Kolak, Bradley and Block34,Reference Pinho, Mackenbach and Den Braver35,Reference Hirsch, Zhao and Melly47) . Additionally, other studies have found that low median household income is associated with a higher concentration of fast-food outlets in the neighbourhood over a long period(Reference Berger, Kaufman and Bader12,Reference Rummo, Guilkey and Ng33,Reference Hirsch, Zhao and Melly47) .

The results of our study indicate that the level of education of participants is linked to socio-economic inequality and has an impact on the food environment trajectories. While a few studies have analysed individual socio-economic characteristics of food environment trajectories, most studies focus on median household income per DA. Two American studies found that individuals with low education levels are more likely to experience a persistent low-access trajectory to supermarkets. Our study also suggests a strong relationship between the level of education and disadvantaged food trajectory. However, the direction of this relationship is not entirely clear. On the one hand, participants with a high school level education are more exposed to a disadvantaged food environment trajectory than those with elementary education levels. On the other hand, participants with college-level education are more exposed to a disadvantaged food environment trajectory than participants with technical education levels. Finally, some high-income residential neighbourhoods also lack access to food stores(Reference Robitaille and Bergeron38). However, one study on obesogenic environments found that disadvantaged communities have fewer supermarkets than advantaged communities(Reference Giskes, van Lenthe and Avendano-Pabon48).

This study has limitations due to its methodology, dataset and food environment indicators. The study relied on data from the CARTaGENE cohort for only three time points: 2009, 2011 and 2018. This limited data availability between 2009 and 2018 reduced the accuracy of the food environment trajectories. Additionally, the food environment indicators used in this study only accounted for categories of food stores like grocery stores, supermarkets, farmer’s markets, and fruit and vegetable shops(Reference Luiten, Steenhuis and Eyles49). These categories were characterised as contributing healthy foods to the food environment. However, research shows that supermarkets and grocery stores also offer a variety of unhealthy, highly processed foods(Reference James, Seward and James O’Malley26). Therefore, adding in-store indicators to the physical access indicators would improve the food indicators and the food environment trajectories.

Another limitation is that the CARTaGENE cohort was designed to recruit adults between 40 and 69 years from metropolitan areas in the province of Quebec. Therefore, the results cannot be generalised to the general population. While this study has its limitations, it boasts several notable strengths. The methods employed to create food environment trajectories were reliable, and a large sample size was included. The observed socio-economic disparities in food environments may help to shed light on the varying rates of obesity among different socio-economic groups. The food environment trajectories established in this study hold promise for analysing chronic diseases in CARTaGENE cohort participants. Ongoing research will explore the potential correlation between these trajectories and the weight status of participants in the CARTaGENE cohort.

Conclusions

The literature on food environments generally focuses on socio-economic disparities between affluent and disadvantaged communities when it comes to access to healthy food stores. However, our results lead to further analyses of the trajectories of individuals’ food environments, which reveal socio-economic inequalities related to individual demographic and socio-economic characteristics. Age, level of education, current employment situation, annual household income and the number of dependents in the household appear to affect the food environment trajectories among the CARTaGENE cohort.

This is one of the few studies that create longitudinal food environment indicators, identify food environment trajectories, and their individual demographic and socio-economic determinants. This study shows that socio-economic disadvantage was associated with a disadvantaged food environment trajectory. Thus, the promotion of healthy food environments requires both the zoning of food outlets in territorial planning and the reduction of socio-economic inequalities. Our results provide insights into the promotion of healthy eating environments in Québec and help to better identify the disadvantaged groups that are most exposed to unhealthy food environments.

Public policies should aim to improve food environments, especially in neighbourhoods with vulnerable communities.

Acknowledgements

The authors would like to thank Véronic Fortin at Institut national de santé publique du Québec for her help with document research and Marianne Dubé at Institut national de santé publique du Québec for the database mergers.

Informed consent statement

Informed consent was obtained from all subjects involved in the study.

Data availability statement

Restrictions apply to the availability of these data. Data were obtained from CARTaGENE, CHU Ste-Justine, and data requests should be directed to them (https://www.cartagene.qc.ca, accessed on 4 January 2022).

Financial support

This research was conducted as part of an internship at the Institut national de santé publique du Québec in 2022. The internship was funded by the Quebec Population Health Research Network (Réseau de recherche en santé des populations du Québec).

Conflicts of interest

The authors declare no conflict of interest. The funders had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, or in the decision to publish the results.

Authorship

Conceptualisation: H.A., E.R. and M.C.P.; Methodology: H.A., E.R. and M.C.P.; Analysis: H.A.; Writing – Original Draft Preparation: H.A.; Writing – Review and Editing: H.A., E.R., M.C.P. and A.L.; Funding Acquisition: E.R. and M.C.P. All authors have read and agreed to the published version of the manuscript.

Ethics of human subject participation

This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving research study participants were approved by the Comité d'éthique de la recherche en sciences et en santé (CERSES-19-063-D) and from the ethic review committee of CARTaGENE (ethic # 549966). Written informed consent was obtained from all subjects.

References

Institute for Health and Metrics and Evaluation (2020) GBD 2019 cause and risk summary: [Dietary risk – Level 2 risk] [Internet]. University of Washington: Institute for Health and Metrics and Evaluation; (Global Health Metrics); available at https://www.healthdata.org/results/gbd_summaries/2019/dietary-risks-level-2-risk (accessed September 2022).Google Scholar
Institute for Health and Metrics and Evaluation (2021) Canada profile [Internet]. University of Washington; (Country profile); available at https://www.healthdata.org/canada (accessed September 2022).Google Scholar
Walker, RE, Keane, CR, Burke, JG (2010) Disparities and access to healthy food in the United States: a review of food deserts literature. Health Place 16, 876884.CrossRefGoogle ScholarPubMed
Raine, KD (2005) Les déterminants de la saine alimentation au Canada: aperçu et synthèse. Can J Public Health 96, S815.CrossRefGoogle Scholar
Osei-Assibey, G, Dick, S, Macdiarmid, J et al. (2012) The influence of the food environment on overweight and obesity in young children: a systematic review. BMJ Open 2, e001538.CrossRefGoogle ScholarPubMed
Gamba, RJ, Schuchter, J, Rutt, C et al. (2015) Measuring the food environment and its effects on obesity in the United States: a systematic review of methods and results. J Community Health 40, 464475.CrossRefGoogle ScholarPubMed
Chennakesavalu, M & Gangemi, A (2018) Exploring the relationship between the fast food environment and obesity rates in the US vs. abroad: a systematic review. J Obes Weight Loss Ther 8, 117.Google Scholar
Pérez-Ferrer, C, Auchincloss, AH, de Menezes, MC et al. (2019) The food environment in Latin America: a systematic review with a focus on environments relevant to obesity and related chronic diseases. Public Health Nutr 22, 34473464.CrossRefGoogle ScholarPubMed
Needham, C, Orellana, L, Allender, S et al. (2020) Food retail environments in greater Melbourne 2008–2016: longitudinal analysis of intra-city variation in density and healthiness of food outlets. Int J Environ Res Public Health 17, 1321.CrossRefGoogle ScholarPubMed
Boone-Heinonen, J, Gordon-Larsen, P, Kiefe, CI et al. (2011) Fast food restaurants and food stores: longitudinal associations with diet in young to middle-aged adults: the CARDIA study. Arch Intern Med 171, 11621170.CrossRefGoogle Scholar
Filomena, S, Scanlin, K, Morland, KB (2013) Brooklyn, New York foodscape 2007–2011: a five-year analysis of stability in food retail environments. Int J Behav Nutr Phys Activ 10, 17.CrossRefGoogle ScholarPubMed
Berger, N, Kaufman, TK, Bader, MDM et al. (2019) Disparities in trajectories of changes in the unhealthy food environment in New York City: A latent class growth analysis, 1990–2010. Social Sci Med 234, 112362.CrossRefGoogle ScholarPubMed
Durette, G & Paquette, MC (2021) Liens entre l’environnement alimentaire communautaire et l’alimentation : synthèse des connaissances, Institut national de santé publique du Québec. Canada; available at https://policycommons.net/artifacts/2066428/liens-entre-lenvironnement-alimentaire-communautaire-et-lalimentation/2820457/ (accessed Nov 2023).Google Scholar
Holsten, JE (2009) Obesity and the community food environment: a systematic review. Public Health Nutr 12, 397405.Google ScholarPubMed
Gibson, DM (2011) The neighborhood food environment and adult weight status: estimates from longitudinal data. Am J Public Health 101, 71.CrossRefGoogle ScholarPubMed
Cobb, LK, Appel, LJ, Franco, M et al. (2015) The relationship of the local food environment with obesity: a systematic review of methods, study quality, and results: the local food environment and obesity. Obesity 23, 13311344.CrossRefGoogle Scholar
Richardson, AS, Meyer, KA, Howard, AG et al. (2014) Neighborhood socioeconomic status and food environment: a 20-year longitudinal latent class analysis among CARDIA participants. Health Place 30, 145153.CrossRefGoogle ScholarPubMed
Engler-Stringer, R, Le, H, Gerrard, A et al. (2014) The community and consumer food environment and children’s diet: a systematic review. BMC Public Health 14, 522.CrossRefGoogle ScholarPubMed
Cutumisu, N, Traoré, I, Paquette, MC et al. (2017) Association between junk food consumption and fast-food outlet access near school among Quebec secondary-school children: findings from the Quebec Health Survey of High School Students (QHSHSS) 2010–11. Public Health Nutr 20, 927937.CrossRefGoogle ScholarPubMed
Lebel, A, Morin, P, Robitaille, É et al. (2016) Sugar sweetened beverage consumption among primary school students: influence of the schools’ Vicinity. J Environ Public Health 2016, 110.CrossRefGoogle ScholarPubMed
Morin, P, Demers, K, Robitaille, É et al. (2015) Do schools in Quebec foster healthy eating? An overview of associations between school food environment and socio-economic characteristics. Public Health Nutr 18, 16351646.CrossRefGoogle ScholarPubMed
Yang, S, Zhang, X, Feng, P et al. (2021) Access to fruit and vegetable markets and childhood obesity: a systematic review. Obesity Rev 22, e12980.CrossRefGoogle ScholarPubMed
Seijo, M, Spira, C, Chaparro, M et al. (2021) Development of physical activity and food built environment quality indicators for chronic diseases in Argentina. Health Promotion Int 36, 15541565.CrossRefGoogle ScholarPubMed
Lebel, A, Daepp, MIG, Block, JP et al. (2017) Quantifying the foodscape: a systematic review and meta-analysis of the validity of commercially available business data. Krukowski RA. PLoS One 12, e0174417.CrossRefGoogle Scholar
Widener, MJ (2018) Spatial access to food: retiring the food desert metaphor. Physiol Behavior 193, 257260.CrossRefGoogle ScholarPubMed
James, P, Seward, MW, James O’Malley, A (2017) Changes in the food environment over time: examining 40 years of data in the Framingham Heart Study. Int J Behav Nutr Physical Activity 14, 84.CrossRefGoogle ScholarPubMed
Mahendra, A, Polsky, JY, Robitaille, É et al. (2017) Geographic retail food environment measures for use in public health. Health Promotion Chronic Dis Prev Can 37, 6.Google ScholarPubMed
Swinburn, B, Egger, G, Raza, F (1999) Dissecting obesogenic environments: the development and application of a framework for identifying and prioritizing environmental interventions for obesity. Preventative Med 29, 563570.CrossRefGoogle ScholarPubMed
Liu, B, Widener, MJ, Smith, LG et al. (2023) Integrating coordination of food purchasing into activity space-based food environment research: toward a household perspective. Health Place 82, 103046.CrossRefGoogle Scholar
Letarte, L, Pomerleau, S, Tchernof, A et al. (2020) Neighbourhood effects on obesity: scoping review of time-varying outcomes and exposures in longitudinal designs. BMJ Open 10, e034690.CrossRefGoogle ScholarPubMed
Needham, C, Strugnell, C, Allender, S et al. (2022) Beyond food swamps and food deserts: exploring urban Australian food retail environment typologies. Public Health Nutr 25, 11401152.CrossRefGoogle ScholarPubMed
Maguire, ER, Burgoine, T, Monsivais, P (2015) Area deprivation and the food environment over time: a repeated cross-sectional study on takeaway outlet density and supermarket presence in Norfolk, UK, 1990–2008. Health Place 33, 142147.CrossRefGoogle Scholar
Rummo, PE, Guilkey, DK, Ng, SW et al. (2017) Beyond supermarkets: food outlet location selection in four U.S. cities over time. Am J Prev Med 52, 300310.CrossRefGoogle ScholarPubMed
Kolak, M, Bradley, M, Block, DR et al. (2018) Urban foodscape trends: disparities in healthy food access in Chicago, 2007–2014. Health Place 52, 231239.CrossRefGoogle ScholarPubMed
Pinho, MGM, Mackenbach, JD, Den Braver, NR et al. (2020) Recent changes in the Dutch foodscape: socioeconomic and urban-rural differences. Int J Behav Nutr Physical Activity 17, 43.CrossRefGoogle ScholarPubMed
Awadalla, P, Boileau, C, Payette, Y et al. (2013) Cohort profile of the CARTaGENE study: quebec’s population-based biobank for public health and personalized genomics. Int J Epidemiol 42, 12851299.CrossRefGoogle ScholarPubMed
Statistique Canada (2021) Aire de diffusion. Statistique Canada. Ottawa. (Dictionnaire: Recensement de la population); available at https://www12.statcan.gc.ca/censusrecensement/2021/ref/dict/az/Definition-fra.cfm?ID=geo021 (accessed August 2022).Google Scholar
Robitaille, E & Bergeron, P (2013) Accessibilité géographique aux commerces alimentaires au Québec : analyse de situation et perspectives d’interventions. In Institut national de santé publique du Québec. Canada.Google Scholar
Billari, F (2001) Sequence Analysis in Demographic Research. Special Issue on Longitudinal Methodology. Can Stud Popul 28, 439458.CrossRefGoogle Scholar
Robette, N & Thibault, N (2009) Analyse harmonique qualitative ou méthodes d’appariement optimal. Une analyse exploratoire de trajectoires professionnelles. Popul 63, 621646.Google Scholar
Robette, N (2011) Explorer et décrire les parcours de vie: les typologies de trajectoires. Paris: CEPED. Les collections du CEPED.Google Scholar
Letarte, L, Gagnon, P, McKay, R et al. (2021) Examining longitudinal patterns of individual neighborhood deprivation trajectories in the province of Quebec: a sequence analysis application. Social Sci Med 288, 113695.CrossRefGoogle ScholarPubMed
Gabadinho, A, Ritschard, G, Müller, N et al. (2011) Analyzing and visualizing state sequences in R with TraMineR. J Stat Softw 40, 137.CrossRefGoogle Scholar
Kelman, J, Pool, LR, Gordon-Larsen, P et al. (2019) Associations of unhealthy food environment with the development of coronary artery calcification: the CARDIA study. J Am Heart Assoc 8, e010586.CrossRefGoogle ScholarPubMed
Christine, PJ, Auchincloss, AH, Bertoni, AG et al. (2015) Longitudinal associations between neighborhood physical and social environments and incident type 2 diabetes mellitus: the multi-ethnic study of atherosclerosis (MESA). JAMA Intern Med 175, 13111320.CrossRefGoogle ScholarPubMed
Moore, LV, Diez Roux, AV, Nettleton, JA et al. (2008) Associations of the local food environment with diet quality—a comparison of assessments based on surveys and geographic information systems: the multi-ethnic study of atherosclerosis. Am J Epidemiol 167, 917924.CrossRefGoogle ScholarPubMed
Hirsch, JA, Zhao, Y, Melly, S et al. (2023) National trends and disparities in retail food environments in the USA between 1990 and 2014. Public Health Nutr 26, 10521062.CrossRefGoogle ScholarPubMed
Giskes, K, van Lenthe, F, Avendano-Pabon, M et al. (2011) A systematic review of environmental factors and obesogenic dietary intakes among adults: are we getting closer to understanding obesogenic environments? Environmental factors and obesogenic dietary behaviours. Obes Rev 12, e95106.CrossRefGoogle Scholar
Luiten, CM, Steenhuis, IH, Eyles, H et al. (2015) Ultra-processed foods have the worst nutrient profile, yet they are the most available packaged products in a sample of New Zealand supermarkets. Public Health Nutr 19, 530538.CrossRefGoogle Scholar
Figure 0

Fig. 1 Sample selection process

Figure 1

Table 1 Matrix of insertion, suppression and substitution costs between the three modalities of the classification variable

Figure 2

Table 2 Distribution of participants by variable included in the analyses

Figure 3

Table 3 Demographic and socio-economic characteristics distribution of participants by food environment trajectories

Figure 4

Fig. 2 Food environment trajectory typologies. This figure is used for a better visualisation of the ten typical sequences of each class. In other words, the ten most frequent sequences in each class that are representative of the class

Figure 5

Fig. 3 Factors of exposure within trajectories characterised by food desert environments