Welcome to the Pathways of Effect project

Welcome to the Pathways of Effect project website—a Health Canada sponsored research project on data associations between social determinants, human health, and well-being indicators, using publicly available data from the Alberta Foothills region. On this webpage, you will find information about the project, research results drawing from real-world data that model the pathways of effect among social determinants of health on a variety of health outcomes, and relevant resources to support health impact assessments of federally designated projects under Canada’s Impact Assessment Act.

This research project explored new modeling techniques that show associations along the ‘pathways of effect’ between social, economic and environmental conditions and health outcomes at a population level. The purpose of the project was to illustrate the potential use of structural equation modeling (SEM) to highlight where regional pressures on determinants of health are observed in available datasets.

SEM is a combination of statistical research procedures that were used to measure broad concepts of interest that are relevant to health impact assessment, and environmental impact assessment more broadly. Concepts within a social determinants of health framework are difficult to be reduced down to a single measure of human health and/or well-being. SEM provides a way of combining data into high-level categories and then measuring how strongly these categories are associated with each other. Each category describes an aspect of the social determinants of health along the ‘pathways of effect’ continuum. The ultimate goal of SEM in this analysis is to highlight which relationships among determinants show up in real-world data at a regional scale.

For clarity around definitions and key concepts used in this website, readers are encouraged to consult the Frequently Asked Questions section. For more information on the technical analysis, indicators, and research process, readers should consult the full technical report available here

About This Project

Canada’s Impact Assessment Act aims not only to protect the physical environment but also people’s lives, when companies propose to carry out major natural resource and infrastructure projects (also called designated projects). It is meant to lessen negative project impacts, as well as boost positive project impacts, including on human health. Many factors in people’s lives contribute to their social, physical and mental well-being, one way or another, and some of these factors can be affected by a new development project. Health Canada has adapted the WHO framework on the social determinants of health to develop a generic diagram of effect pathways (above, the “adapted” framework) that shows how projects can increase or reduce health risks for nearby communities. This can help fit the project effects into the local human context and identify ways of mitigating potential bad effects and enhancing good effects.

The adapted framework illustrates the building blocks that link the project with conditions of daily life that determine health (i.e., health determinants). A project’s various components and activities include, for example, physical construction and operations, recruiting workers, and setting up living quarters for some of these workers (work camps). These activities would first have positive or negative effects on the surrounding environmental, economic, and social (including cultural) conditions, before having any human health effects. Project influences on the surrounding environmental conditions may involve the quality of the air, water, and soil as well as wildlife, plants, and landscape (e.g., linear density of roads, railways, transmission lines, pipelines). The physical environment provides the very basic necessities for life. Economic and social conditions are more complex and reflect people’s material circumstances. Under economic conditions, we have employment opportunities, work situations (e.g., shiftwork), and means for living (e.g., income, home ownership). The social conditions are how people live (e.g., education level, housing quality), feed themselves, and make use of their place of residence (e.g., green space, community infrastructure, community safety).

The material aspects of these resources, opportunities, and life situations contribute positively or negatively to community well-being (i.e., quality of life), as well as influence what health-related behaviours are likely to be taken up (e.g., eating habits, physical activity, substance use) and what biological exposures are likely to occur. All factors along the “material” pathway may ultimately have biological effects (e.g., hypertension, quality of biological processes) on physical well-being, including brain functions, and this has implications for mental well-being. In the other direction mental well-being, has implications for physical well-being through the biological stress response (e.g., hypertension, stress hormones) and associated coping mechanisms. Along the “psychosocial” pathway, some economic or social conditions found on the material pathway may have a psychological aspect to them that is either comforting (e.g., financial security, the enjoyment of the right to adequate housing) or that is stressful (e.g., mental toll of physically gruelling work conditions). Interpersonal sources of comfort or stress (e.g., community belonging and family violence) are also psychosocial factors linked to mental well-being. As well, the beneficial psychosocial effects of cultural practices (including use of traditional lands and waters) are critical for the mental well-being of Indigenous Peoples, which encompasses emotional and spiritual wellness.

Cumulative effects on the physical and living environments (at the community level) arising from a variety of past, present and future resource-development activities of varying scale may have net positive or adverse health impacts. That being said, it is difficult to make a definitive link between any single determinant of health and health outcomes, at the individual level. However, best practices can be used to manage project-specific influences on host communities (e.g., environmental safeguards, communication plans, road safety measures, community investments, wellness programs) and minimize health risks concerning physical and mental health (including infections), as well as injuries and physical harms.

Understanding Effects Pathways in Health Impact Assessments

Self-rated Health

Self-rated health is a combined measure of human health that takes into account physical and mental health status. It has consistently been shown to be a highly effective and reliable measure of overall health of individuals. The pathways model tested several hypotheses, including: [1] social determinants of health are directly associated with self-rated health; [2] self-rated health is indirectly affected by multiple determinants (e.g. social, economic and environmental conditions) through biological and behavioural pathways. In the images below, modeling software called SPSS was used to generate the model structure. It uses rectangles to represent real-world measures from data collected from each Local Geographic Area of Alberta, the small circles (labeled with an ‘e’) represent measurement error to estimate potential uncertainties with collected data, and the larger ovals represent what are referred to as “latent concepts” (i.e. unobserved variables that typically cannot be measured by any one single indicator). Curved lines with double-headed arrows represent covariances between the social, environmental and economic conditions (i.e. the place-based contextual factors labeled as ‘intermediary factors’ in the Draft Framework), and straight arrows represent pathway of effect relationships between latent factors. The numbers on each line are called “standardized regression weights”. They are basically the measure of association between two given variables in the model, where a change in one variable will lead to a corresponding change in another. The stars signify whether or not those relationships are statistically significant, which is to say we would expect that relationship to exist beyond just chance alone.

Explore Pathways of Effect

For the first hypothesis, we find that as social conditions (measured by the proportion of populations living in inadequate housing and who do not have a high-school diploma) decrease, self-rated health described by the population as “good”, “very good” or “excellent” increases. This relationship is statistically significant (p<0.05), meaning we would expect this relationship to exist greater than 95% of the time in the broader population beyond chance alone. This demonstrates the so-called “direct” pathway of effect between social conditions and self-rated health.  Moreover, the model shows that 83.7% of the variability in self-rated health was attributable to the entire suite of variables included in the model.

For the second set of hypotheses, we can analyze the indirect influence of social conditions (or other economic and/or environmental determinants) on biological and behavioural pathways, and simultaneously determine the influence of biology and behaviour on self-rated health. For example, looking at our social conditions, we can see that there is a direct and statistically significant association with both behaviour and hypertension. In other words, as social conditions increase (i.e. rates of housing inadequacy and population proportions of people without a high-school degree increase), we expect health behaviours to be negatively impacted (i.e. rates of people meeting Canadian guidelines for healthy diets and physical activity to decrease).  We also find similar results for the influence of social conditions on hypertension, which is to say worse social conditions lead to correspondingly high rates of hypertension.

We can then look at the associations between both health behaviours and biological conditions to determine their impact on self-rated health. Here, we see that as health behaviours improve (i.e. more people meet Canadian guidelines for physical activity and a healthy diet), we see a lower proportion of the population with hypertension and that those same health behaviours are positively associated with good, very good or excellent self-rated health. This provides evidence for the relevance and importance of indirect influences among social determinants on health behaviours in terms of how they influence self-rated health. You can now go back to the overall findings image and interpret other pathways of interest based on statistical associations found in this particular data set.

Chronic Obstructive Pulmonary Disorder Prevalence (COPD)

Now that we have explored the model in some detail, we can take a look at the model in relation to other health outcomes of interest. Consider COPD, a chronic inflammatory lung disease that causes obstructed airflow in the lungs and which is associated with difficulty breathing, shortness of breath and coughing and wheezing. COPD is commonly associated with chronic exposure to poor air quality and smoking, but what about the social determinants of health?

Looking at the pathway model, we can see that social conditions are directly associated with COPD prevalence (i.e. higher proportions of inadequately housed and undereducated people are directly associated with higher rates of COPD prevalence), but that social conditions are also statistically associated with both health behaviours and hypertension, that health behaviours are also a significant predictor of hypertension, and finally, that hypertension is also associated with higher population prevalence of COPD.  We also know that the included variables in the model predict about 60% of the variance in COPD prevalence. Therefore, there are opportunities to consider other variables that may additionally explain the outcome, and it should be noted that smoking status was not included but is a likely predictor that accounts for some of the remaining variability. The model also shows that economic conditions are positively associated with hypertension.

Mental Health Emergency Department Visit Incidence

Similar patterns emerge when looking at the model structure in relation to emergency department visits attributable to mental health. Here, the model predicts 83.8% of the variance in the health outcome. We find evidence of a direct and positive impact of social conditions on mental health emergency department visits, where higher rates of housing inadequacy and low education lead to higher incidence of mental health emergency department visits. Further, economic conditions directly influence hypertension prevalence, hypertension prevalence is negatively associated with the health outcomes of interest. Similar storylines for the above two health outcomes are present in terms of social conditions being associated with both behaviour and hypertension, and that health behaviours are a driver of hypertension.

This website and the accompanying report are intended to be exploratory. As a standalone prototype of the method, the goal is to determine its effectiveness and generate dialogue on the Social Determinants of Health and their relevance to impact assessment. As a result, there are several limitations of this application that are worth identifying.

First, the pilot was conducted at a regional scale with a relatively small sample size which may limit the explanatory potential of the analysis. Because of the regional scale, it may also not be as useful for detecting the influences of specific projects. We provide guidance and recommendations on how to apply the method to project-specific impacts in the technical report.

Second, the indicators utilized reflect often-considered determinants of health identified through impact assessment documentation relevant to Alberta. They are not intended to represent the full suite of determinants of health. Because data were sourced only for indicators with applicability across the entire province, broader suites of more specific indicators directly relevant to a project of interest should be considered in the future, and not all data may be available for other parts of Canada.

Per the discussion of Health Canada’s Draft Framework (above), future consideration of other variables of interest in future analysis could include exposure to: physical, chemical and biological hazards as part of environmental conditions; the consumption (in terms of both quality and quantity) of food and water; levels of physical activity; the use of drugs, alcohol and cigarettes; sleep quality; and associated measures of mental well-being. This could also include asset-based measures to consider health promoting factors that may mitigate these identified risk factors (e.g. healthy eating, physical activity, avoidance of substance use, restful sleep, good mental well-being).

Limitations

Third, the indicators were not necessarily identified through in-depth consultation with Indigenous communities, and therefore should not be thought to reflect Indigenous government and/or community priorities. Future processes may want to prioritize engagement with impacted communities for project-specific assessments to determine the right ‘mix’ of indicators and data to be collected and monitored to support pathways of effect analysis.

Fourth, while structural equation modeling is a powerful analysis tool, it should be noted that there is no single method for measuring causal associations. Structural equation modeling’s usefulness is in capturing the interrelationships and ‘pathways of effect’, but care should be taken in making the case that a pathway is in fact a causal driver of a given association between variables or indicators.

Despite these limitations, the results suggest that the approach has high applicability to monitoring and measuring ‘pathways of effect’ between broad categories of determinants of health relevant to impact assessment. The approach can be supportive of land-use planning with a “health-in-all-policies/healthy public policy” approach, encourage the identification and measurement of upstream and downstream indicators that can influence changes to health status, and to support cumulative effects assessments at both the project-specific and regional level of analysis by considering past, present and potential future stressors.

Research that replicates this approach with other datasets, but especially those at the project-specific level of analysis will be useful to further demonstrate the usefulness of this approach in supporting impact assessment. For more information on technical limitations, visitors are encouraged to read the full technical report.

Technical Report

This project website reports on key findings from the application of a pathways of effect analysis between multiple determinants of health across seven health outcomes. To read the technical report which outlines the methodology and in-depth findings, please download it here.

Frequently
Asked
Questions

  • The Government of Canada defines the determinants of health as “the broad range of personal, social, economic and environmental factors that determine individual and population health.”

  • According to the World Health Organization, health impact assessment (often referred to as HIA) is “a practical approach used to judge the potential health effects of a policy, programme or project on a population, particularly on vulnerable or disadvantaged groups. Recommendations are produced for decision-makers and stakeholders, with the aim of maximising the proposal’s positive health effects and minimising its negative health effects. The approach can be applied in diverse economic sectors and uses quantitative, qualitative and participatory techniques.”

  • According to the Impact Assessment Agency of Canada, a ‘pathway of effect’ refers to the expected link between a designated project and a valued component of interest. Pathways of effect analysis refers to a “systemic way of breaking down a series of proposed cause-and-effect relationships and interactions into steps. The purpose is to understand the route by which health, social and/or economic effects and their interactions occur.”

  • Per the Impact Assessment Agency of Canada, “the cause-and-effect relationship does not need to be proven. Instead, the cause-and-effect relationship merely needs to be plausible within the context of the project.” Moreover, while some methods like structural equation modeling get us closer to understanding causal linkages, it should be noted that no methodology will be able to assess causal relationships with absolute certainty, and relationships under analysis can be highly dependent on the multiple contexts in which a designated project is embedded (e.g. local and/or regional environmental, economic, social and political conditions).

  • Structural equation modeling is a combination of statistical techniques that aim to: [1] reduce large amounts of data and indicators into meaningful representations of concepts that are immeasurable by any single measure alone (e.g. it is not possible to simply measure the ‘social determinants of health’ through any single measure because it represents multiple concepts associated with the contexts in which people live, work and play); [2] confirm the ‘structure’ of those broader concepts through measures of statistical validity; and, [3] model relationships between and among variables included in the model to measure pathways of effect.

  • Modeling relationships among individual variables is entirely possible to test for associations and whether or not certain ‘pathways’ are mediated or in some way explained by interactions with other indicators. These types of linear associations can be helpful for highly specific indicators, but the value of structural equation modeling is its ability to scale up multiple indicators of interest into a conceptualization of broader determinants of health, and then test associations to human health and well-being.

  • This pilot was conducted at the level of Local Geographic Areas—a regional geography—in the province of Alberta. While this level of analysis is perhaps more useful for demonstrating the utility of structural equation modeling in supporting regional and strategic assessments, it can be easily tailored to project-specific impacts through appropriate data collection techniques. For more information, readers can consult the technical report.

  • A project’s influences may differ between various groups within the workforce and local community, leading to health inequalities. These differential outcomes are assessed using an analytical process called gender-based analysis plus. For instance, exposure to environmental contaminants will be greater for populations that spend more time in or consume more from the environment (i.e., traditional or country foods). Indirect effects may also have differential impacts. Some project workers’ stressful occupation may not only affect their mental well-being, but may lead to the adoption of coping mechanisms, like the use of drugs and alcohol, which may then bring about community safety issues or family conflicts. This is how a project’s effects differ between groups of individuals (i.e., identities of project workers, neighboring community members, workers’ families).