A model of underlying socioeconomic vulnerability in human populations: evidence from variability in population health and implications for public health

Abstract

Drawing from insights into the variability of complex biologic systems we propose that the health of human populations reflects the interrelationship between underlying vulnerabilities (determined by population-level social and economic factors; e.g., income distribution) and capacities (determined by population-level salutary resources, e.g., social capital) and how populations, shaped by these vulnerabilities and capacities, respond to intermittent stressors (e.g., economic downturns) and protective events (e.g., introduction of a school). Monitoring this dynamic at the population-level can be accomplished by examining not only rates of illness and mortality, but variability in rates, either between populations or within populations over time. We used mortality data from New York City neighborhoods between 1990 and 2001 to test two related hypotheses consistent with this model of population health: (a) There is greater variability in mortality rates at a point in time between neighborhoods that are characterized by socioeconomic vulnerability; and (b) there is greater variability in mortality rates over time within neighborhoods that are characterized by socioeconomic vulnerability. We found that neighborhoods characterized by social and economic vulnerability displayed substantial variability in particular mortality rates. Mortality rates displaying the greatest variability were from causes that may be sensitive to social conditions (e.g., homicide or HIV/AIDS rates). Variability in population health existed both between neighborhoods with underlying vulnerability at one point in time and within vulnerable neighborhoods over time. The results of this analysis are consistent with a theory of underlying socioeconomic vulnerabilities of human populations and suggest that variability in population health may be an important consideration in population health assessment.

Introduction

In the field of ecology, the dynamics of groups of different species are studied as a means of allowing prediction of group behaviors and outcomes, both at equilibrium and in response to specific interventions (Levins, 1975). Although health-related empiric studies have traditionally focused on identifying individual-level characteristics that determine health, more recent work has shown that group characteristics may also importantly affect human health (O’Campo, 2003). For example, population-level socioeconomic status is associated with health-related behaviors independent of individual socioeconomic status (Subramanian, Kim, & Kawachi, 2002). Also, it is increasingly recognized that simple linear cause and effect paradigms that assume that individuals are independent of one another are over-simplifications that fail to take into account patterns of connectedness among individuals (Koopman & Lynch, 1999). For example, social network dynamics are associated with risk behavior and transmission of infectious diseases (Koopman & Longini, 1994; Latkin, Forman, Knowlton, & Sherman, 2003). Therefore, groups of individuals can be seen as population systems and the dynamics of human populations may determine health in their own right (Robert, 1999). Early theoretical and empiric work assessing the ecology of human health showed that characteristics of human populations, defined at the county level in two US states, are associated with population health in a manner similar to biological populations (Karpati, Galea, Awerbuch, & Levins, 2002).

Complex systems are systems that are inadequately described by unidirectional causal relationships and that may require the consideration of multidirectional causal relationships (e.g., feedback) in order to permit both accurate description and prediction (Levins, 1974). Although it has long been accepted in ecology that population system dynamics are complex, there has been relatively less attention paid to the complexity of human population systems and how this complexity may shape population health. Several observations suggest that human populations behave as complex systems. First, there are multiple examples of discontinuous changes in health in relation to monotonic changes in exposures facing human populations (Philippe & Mansi, 1998). For example, the relation between population health and several environmental exposures encompass threshold and sigmoid curves, both hallmarks of nonlinear dynamics (Maynard et al., 2003). Second, the effects of particular exposures on human populations can linger well beyond removal of the exposure. For example, the population mental health consequences of disasters are well known to persist beyond the disaster itself (Galea et al., 2003b,). Third, multiple diseases, including infectious diseases and neoplastic diseases, frequently share determinants that are affected by common environmental exposures (Koopman & Lynch, 1999; Koopman & Longini, 1994). Although none of these observations in and of themselves define complex systems, they provide empiric evidence that human populations exhibit complex system behaviors and that the application of ecologic principles may assist in describing, and understanding, population health.

Positing that human populations are complex systems whose properties are of empiric and potentially practical interest we hypothesize that the health of human populations reflects the interrelationship between underlying vulnerabilities and capacities and how populations, shaped by these vulnerabilities and capacities, respond to intermittent stressors and protective events. A wide range of factors may be considered to be underlying vulnerabilities, including a paucity of material resources (e.g., low income) available in a given human population, or the presence of a natural tectonic fault line that predisposes a population to earthquakes. Conversely, examples of underlying capacities may include social capital and abundant availability of natural resources. Intermittent stressors include the closure of a large employer, or a natural disaster. Conversely, protective events, such as the opening of a new school or an increase in group cohesiveness due to the success of a local sports team, also occur intermittently. Importantly, the intermittent influences interact with the underlying conditions to shape health at any particular moment. We note that intermittent stressors can be considered destabilizing phenomena while intermittent positive events may be stabilizing.

This model is a heuristic to explain how underlying and intermittent conditions may affect population health and of necessity represents a simplification. Therefore, while the model suggests that vulnerabilities and capacities are distinct constructs, the “absence of vulnerability” can be considered a capacity if we are comparing health indicators across different human populations. Similarly, the boundaries between intermittent “stressors” and “positive events” are simplifying devices intended to help explain the potential system dynamics that can affect population health. In addition, as noted in the examples given above, the range of factors that may shape population health may be “social and economic” factors, but also potentially “geographic”, “climatologic” or a range of other categories that generally fall outside the realm of social epidemiology. Although we highlight here the role of underlying social and economic vulnerabilities we posit that this model of population health may be relevant to a range of potential vulnerabilities and capacities that extend beyond the focus of this paper.

Multiple academic disciplines have considered vulnerability as an important characteristic of both individuals and of populations (Bankoff, 2003; Turner et al., 2003; Cohen & Hamrick, 2003). Although definitions of vulnerability vary in the scientific literature, it is generally considered to be the capacity for harm in an individual or system in response to a stimulus. It has been postulated that different elements of vulnerability exist, including genetic and biologic vulnerability at the individual level (Cohen & Hamrick, 2003; Heath & Nelson, 2002) and social vulnerability at the group level (McKeehan, 2000). Individuals who possess specific characteristics are frequently termed “vulnerable”; for example children, homeless persons, and minority inner-city populations have been termed “vulnerable” in recent scientific publications suggesting they are more likely to be harmed by external stressors than are others in the general population (Stergiopoulos & Herrmann, 2003; Shi, 2000). In the field of disaster preparedness, it has long been recognized that certain groups are more vulnerable to the effects of disasters than others. For example, wealthier communities are more likely to rebound from the consequences of natural disasters than less wealthy communities (Nelson, 1990). There is, in turn, complementary evidence that certain characteristics may confer protections on individuals or populations and may be considered capacities. For example, social capital has been shown to be associated with lower population mortality and potentially protects populations from the effects of income maldistribution (Kawachi, Kennedy, Lochner, & Prothrow–Stith, 1997).

Having acknowledged human populations to be units of scientific interest and proposed a model that incorporates both underlying vulnerabilities/capacities and intermittent stressors/protective events, we can now consider how the interrelationship of these factors is reflected in the health of populations. Of particular interest to population health assessment is variability, a component of complex systems that reflects the impact of external stressors and the complex system's attempt at maintaining homeostasis (Levins & Lopez, 1999). Variability of health indicators in human populations may be particularly informative in the study of underlying population vulnerability. In considering variability exhibited by ecologic systems, I.I. Schmalhausen observed that systems at the boundary of their tolerance are more vulnerable to small differences in circumstance and display more variability than systems not similarly stressed (Schmalhausen, 1949). Schmalhausen argued that through the process of biological evolution a species’ phenotype is stable within the normal range of environmental variation. However, in extreme environmental conditions greater phenotypic variation manifests between organisms as characteristics of species that had not previously been a basis of selection are expressed. Extending Schmalhausen's observation, Levins and Lopez (1999) suggest that the impact of intermittent stressors will result in greater variability in outcome among vulnerable human populations than among populations that are not characterized by underlying vulnerability. Summarizing this argument, in populations with low levels of vulnerability (e.g., high income) the rates of disease and mortality would be expected to be stable as the population is resilient to changes that may occur in other conditions. However, in populations with high levels of vulnerability (e.g., low income) there would be greater variability in rates of disease and morality as characteristics of populations that are untested at low levels of vulnerability are expressed in the vulnerable state. For example, in wealthy populations, intermittent stressors may not affect health as the resources conferred by wealth keep disease and mortality rates constant. Conversely, in poor populations intermittent stressors or protective events may be critical in determining disease and mortality rates as there are fewer material protections available to the population. Variability is produced both by the random or uneven distribution of these intermittent events and by differences in underlying vulnerabilities and capacities. Therefore, returning to our model, intermittent stressors, or destabilizing events, affect a homeostatic system and produce varying degrees of change in the population system's properties; the variability in the change in system properties is a function of the extent to which the system is characterized by underlying vulnerabilities or capacities. For example, a geographically isolated community characterized by limited employment opportunities may not cope as well with the sudden departure of a major employer as a community where employment opportunities are abundant. However, the range of responses among populations with limited employment opportunities to the departure of a major employer may be broad and predicated on the distribution of population capacities, such as social capital, to help those newly unemployed. Conversely, the responses of populations characterized by abundant employment opportunities in the face of a departing employer are not dependent on the distribution of other capacities or vulnerabilities. As such, these populations’ responses may be less variable than the responses of vulnerable populations. When examining human populations as population systems of interest, the variability exhibited by specific health indicators may provide evidence for underlying population vulnerability.

We used data from New York City (NYC) neighborhoods to test two hypotheses consistent with the model of population health proposed here: (a) There is greater variability in mortality rates at a point in time between neighborhoods that are characterized by socioeconomic vulnerability; and (b) there is greater variability in mortality rates over time within neighborhoods that are characterized by socioeconomic vulnerability. We note that the analysis shown here focuses on population vulnerability and that we do not formally assess the other elements (i.e., capacities, intermittent stressors, or protective events) of the proposed model. In addition we focus on social and economic factors as underlying vulnerabilities of interest (Marmot, Kogevinas, & Elston, 1987; Geronimus, Bound, & Waidmann, 1999; Adler et al., 1994; Sampson, Morenoff, & Gannon-Rowley, 2002; Robert, 1999). We intend this analysis to illustrate how the behavior of human populations may be suggestive of complex system dynamics and to set the stage for further work that explicitly considers the relationships among underlying socioeconomic vulnerabilities and capacities and how these characteristics of human populations may assist in public health prediction.