The unequal distribution of health and disease within (and across) populations is often seen as unfair. Hence, the goal of public health is not only to improve the overall health of the population but also to reduce the unequal distribution of health within the population. In fact, the relationship between social factors and health was intensely studied over the past 70 years since the emergence of the field. Yet, health inequalities persist and even widened in some countries [1], [2]. While we have learned a great deal about the magnitude of health inequalities and how they arise, we struggle to identify ways on how to tackle these inequalities. This may be partly because studying health inequalities is complex, with many contributing factors – ranging from the overall context to individual behavior – that are interrelated. This is also reflected in the CSDH framework that was created by the Commission on Social Determinants of Health (CSDH) [3]. This framework illustrates the components of societies that affect health and health inequalities. They distinguish between determinants of health inequity which define the societal position, and determinants of health which ultimately link societal position to health.

The CSDH framework is partially based on the Diderichsen model [4]. This model [4] offers theoretical explanations for the pathways underlying health inequalities. The authors identify four main pathways: social stratification, differential exposure, differential susceptibility (or differential impact), and differential consequences. Social stratification describes the process that determines a group’s position in society, shaped by the overall socioeconomic and political context. This position creates systemic advantages or disadvantages that, in turn, influence health. This may be driven by some groups being more exposed to certain determinants of health and disease than others, a pathway known as differential exposure. In addition, the experienced systemic advantages and disadvantages of some groups may affect how strongly these determinants affect health, which is a pathway known as differential susceptibility or differential impact. Finally, the pathway of differential consequences describes how health advantages themselves can feed back into and further worsen a group’s societal position, thereby reinforcing inequality. Hence, differential exposure and differential impact describe how societal position affects health and health inequalities, while social stratification and differential consequences refer to the source and reinforcement of societal position itself which can be driven by selection processes or reverse causality based on the overall context, or one’s health [4]. We present an adaptation of the CSDH framework in Fig. 1 which combines the components of the Diderichsen model with the CSDH framework to indicate the pathways that underlie health inequalities.

The CSDH framework and the Diderichsen model show that we have a good understanding of how health inequalities are produced (and potentially reinforced). This is largely driven by contributions from associational research that assesses the relationship between indicators of societal position and determinants of health. However, in order to truly understand the pathways which Diderichsen et al. defined for the purpose of identifying entry points for policy interventions [4], associational research is not enough. This type of research cannot effectively identify targets for interventions that reduce heterogeneities in the population.

To illustrate this, we provide an example. Associational research that could inform the CSDH framework would be to quantify the association between socioeconomic position (SEP) and smoking. Previous research showed that the prevalence of smoking is higher among low SEP groups than high SEP groups [5]. It was also reported that people who smoke have a higher risk of esophageal cancer [6]. Based on these two associational research findings, we can infer that people from low SEPs smoke more often and have a higher risk of esophageal cancer than people from high SEPs. This information is helpful for building e.g. the framework in Fig. 1 and quantify what groups in society are at risk [7]. However, what we cannot infer is whether people from low SEP would get less esophageal cancer if they would be moved to high SEP or if they would smoke the same amount as people from high SEP [8]. Yet, this type of knowledge is needed when we are interested in quantifying the potential impact of policies as Diderichsen et al [4]. described in their theoretical model. Hence, in order to effectively identify targets for interventions that potentially reduce health inequalities in the population, we need to understand how the world would change under some hypothetical intervention rather than describing the world as it exists [9]. While this can be achieved through randomized controlled trials (RCT), they are not always feasible or ethical to perform in the case of health inequalities. Instead, we can leverage population-based cohort studies to quantify the effect of hypothetical interventions through the use of the counterfactual outcomes framework [10].

In this commentary, we argue that the evidence needed to begin tackling health inequalities already exists. In many contexts, further insight from descriptive or associational research adds little to what is already captured by the established theoretical frameworks, such as the CSDH framework or Diderichsen model. We propose building on this foundation by integrating these existing theoretical models with the counterfactual outcomes framework. This integration clarifies the specific questions being addressed, moves beyond identifying groups-at-risk, and instead identifies points for intervention. This allows us to gain an even better understanding of the underlying pathways that link societal position to health inequalities, namely differential exposure and differential susceptibility, while considering that health inequalities do not exist in isolation. Whereas existing commentaries and evidence focused on either providing guiding questions for health inequalities research (e.g [11], [12]) or the statistical implementation of it (e.g [13], [14]), our contribution is to bridge these stages, and show how to let one inform the other to generate more informative evidence and understanding of how health inequalities can be tackled. In the next sections, we provide a step-by-step guide from research question to operationalization. For a visual representation see Fig. 2.