Towards an epidemiological understanding of the effects of long-term institutional changes on population health: A case study of Canada versus the USA

Abstract

This paper uses a comparative case study of Canada and the USA to argue that, in order to fully understand the associations between population health and the socioeconomic environment we must begin to place importance on the dynamic aspect of these factors—examining them as they evolve over time. In particular, for institutional and policy shifts that often unfold over decades, population health must attend to these big, slow moving processes by adopting a historical perspective to the knowledge base. We compare Canada and the USA on basic health outcomes and a range of determinants of health for which routine data have been collected for all or most of the period between 1950 and the present. During the analysis that follows, we are able to establish that, at the level of society (i) greater economic well being and spending on health care does not yield better health outcomes, that (ii) public provision and income redistribution trump economic success where population health is concerned, and (iii) that the gradual development of public provision represents the buildup of social infrastructure that has long-lasting effects on health status. Our case study shows what can be gleaned from a comparative perspective and a long-term view. The long view allows us to detect the gradual divergence in health status between these two societies and to trace potential institutional causes that would otherwise go unnoticed. The perspective introduced here, and in particular the comparison of Canada and the USA, provides strong support for the use of cross-national comparative work, and a historical perspective on the investigation of societies that successfully support population health.

Introduction

Population health is determined by the circumstances and contexts of life, from the most intimate to the broadest socioeconomic, such as the institutions and policies of a society. We argue here that, in order to fully understand the associations between population health and the socioeconomic environment we must begin to place importance on the dynamic aspect of these factors—examining them as they evolve over time. In particular, for institutional and policy shifts which often unfold over decades, population health must attend to these big, slow-moving processes by adopting a historical perspective to the knowledge base. Without such a longitudinal perspective, many effects of socioeconomic conditions may remain otherwise undetected. We illustrate this historical approach using a comparative case study of population health in Canada versus the USA.

From cross-sectional and life course studies, the mechanisms connecting society and health can be framed at three levels of societal aggregation. At the macro level are such society-wide influences as levels and fluctuations of national income, patterns of distribution, and policies intended to affect these (e.g. income support, education, health care or employment policies). Of note, a number of recent studies based on the OECD nations have begun to examine health using a political economy or socio-political perspective, inherently macro in nature. Navarro et al. (2004) found that infant mortality was inversely associated with the proportion of the population voting for left-leaning (pro-redistribution) parties. The association of distribution (and redistribution) with infant mortality was also cited in a study using the Theil measure of wage inequality (Macinko, Shi, & Starfield, 2004). In addition to infant mortality, Muntaner et al. (2002) found that a host of variables associated with the ‘welfare state’ were associated with a variety of infants’ and children's health outcomes.

At an intermediate, or ‘meso’ level, are the characteristics of one's immediate community or workplace. Influences here include, inter alia, how people interact with each other and the levels of local trust and civility, in the community and the workplace. These will be reflected, in part, in the nature and availability of schools, libraries, newspapers, policing and parks, and also in the nature of work characteristics and environments (Kaplan, Pamuk, Lynch, Cohen, & Balfour, 1996; Kawachi et al., 1997; Putnam, 1993; Rose, 1995). At the most ‘micro’ level, there are the influences on health associated with private life, such as the ability to purchase goods and services, the nature and quality of personal social support: intimate relationships, friendships, and the availability of personal help when needed (Berkman, 1995). Not all relevant influences fit neatly into one level of social aggregation. For example, job insecurity and sense of control are perhaps best understood as resulting from the interaction between macro, meso and micro influences on the individual at a particular stage in the life course (Hertzman, 2001).

We have substantial evidence about the relationships between macro, meso, and micro determinants and population health from cross-sectional and life course perspectives (particularly meso and micro determinants). In fact, Coburn (2004) notes that a sociological account of population health necessitates the inclusion of historical change in health and in institutional features of society. But we have few studies on the association between socioeconomic influence and population health from an historical perspective that can capture big, slow-moving processes. This study attempts to fill some of this gap by focusing on a longitudinal perspective of macro determinants and population health trajectories using a comparative case study approach. A comparison of Canada and the USA illustrates how both institutional/policy regimes and population health trends can change slowly over years and decades, yet result in large differences between societies. Further, this case study helps to specify which features of institutions and policies (e.g. for education, employment, redistribution, public spending) have the greatest returns to population health.

The ‘gradient effect’ of socioeconomic status (SES) is apparent for an inordinate number of health outcomes (Adler et al., 1994). The observation that SES gradients in health are ubiquitous suggests that institutional and policy determinants are also fundamental for population health. Here, there are two main lines of reasoning. First, if socioeconomic inequalities are critical for health, then so too are the societal conditions that create the inequalities themselves. The extent of socioeconomic stratification that exists in society is not innate. Rather, it is strongly influenced by actions and inactions taken by societies that, cumulated over time, become embedded in institutions. This notion is powerfully illustrated by contrasting poverty rates before and after taxes and transfers are accounted for. Using data from the Luxembourg Income Study, Smeeding and Ross (1999) found that, based on market income (i.e. prior to taxes and transfers), at 31%, poverty rates¹ in the USA were up to 5–6% lower than in several OECD nations, including France and Sweden, and on par with others such as Australia, Canada, Spain, and Germany. However, after taxes and transfers, the USA had the highest poverty rate among the OECD nations at 18%, between 6% and 11% higher than all other OECD nations with the exception of Australia which has a post tax and transfer poverty rate of 16%.

The difference in poverty rates is even more marked with respect to children. Prior to transfers, poverty rates across OECD nations for lone parents are consistently high, from a low of 32% in Italy, to a high of 80% in the Netherlands. However, after government redistributive measures are taken into account, the rate for lone mothers was reduced to approximately 10% in many OECD nations, with a low of 4% in Germany. By contrast, the poverty rate for lone mothers in the USA remains at 60% (Beaujot & Liu, 2002). In epidemiological speak, the function of societal policies can be thought of as necessarily causally prior to the effects of SES on health outcomes.

The second line of reasoning stems from the observation that SES gradients vary in their ‘steepness’ across societies, such that there is no necessary, or predictable, level of health associated with any given position on the socioeconomic spectrum. Instead, health status at any given position is highly dependent on the extent to which different societies tie SES to one's ability to procure health-promoting resources and, conversely, the extent to which SES serves as a sorting mechanism for ‘exposures’ that are harmful to health. Institutional arrangements, then, play a role in distributing health promoting resources and buffering individuals from ‘exposures’ adverse for health more successfully than others. In other words, in some societies, factors fundamental to health are provided as rights of citizenship, rather than according to socioeconomic privilege.

These two types of roles of institutional arrangements are difficult to separate, since they may function in a reciprocal manner. That is, reductions in socioeconomic inequality provide public support for increases in systems that distribute resources in an egalitarian manner, and the egalitarian distribution of resources in turn may reduce levels of socioeconomic inequality (Kawachi, 2000).

The effect of income inequality on health outcomes has been investigated at length in the epidemiological literature. However, rarely has this form of inequality been framed by this literature as a benchmark of society's institutional orientation. In other social science domains, however, distribution of income is considered the sin qua non of the welfare state; more egalitarian distribution is equated with social policies that demonstrate egalitarian tendencies with respect to income and other forms of resource distribution (Esping-Andersen, 1990). Accordingly, we present here a brief review of the literature on this topic, viewing income inequality as a fundamental aspect of policy and institutional orientations in a society.

Early cross-national research showing a significant association between income inequality and average health status seemed consistent (Wilkinson (1990), Wilkinson (1992), Wilkinson (1996)). Soon, however, the results of these studies were contested, with charges of poor quality data, lack of control for potential ‘confounders’ such as transfer payments and social spending (Judge, 1995). The characterization of transfers and social spending as confounders is highly debatable, since these may instead be part of the causal mechanism linking inequality to health (Kawachi, 2000; Wilkinson, 1998). In other words, policies that provide transfers and facilitate social spending lead to greater equality of income, rather than these factors being independently tied to health status.

As well, there has also been suggestion that results were driven primarily by the USA, which would thus mean that the finding of high-income inequality leading to poor health was as based on exception and could not be considered a general rule (Judge, 1995; Lynch, Harper, & Davey Smith, 2003; Mackenbach, 2002). The notion that the USA is an outlier has gained momentum with recent null results found for within-nation studies conducted in Japan (Shibuya, Hashimoto, & Yano, 2002), Denmark (Osler et al., 2002), and New Zealand (Blakely, Atkinson, & O’Dea, 2003). In addition, Ross and colleagues have examined this association at several levels of geographic aggregation, including metropolitan area and state/province. Their findings suggest that, in Canada (and several European nations), regional differences in income inequality are not significantly associated with differences in population health. However, in the USA (and Britain) significant associations do exist (Ross et al. (2005), Ross et al. (2000)).

Though collectively these results lend themselves to an ‘outlier’ interpretation, there are also other compelling explanations. First, the USA and Britain have the highest levels of income inequality among the wealthy nations and much wider ranges in inequality among regions, which may be one explanation for an association within those countries and not others (Ross et al. (2000), Ross et al. (2005)). But that is not the whole story. The difference in the strength of the association does not appear to be solely related to the degree of income inequality. Accompanying social policies also seem to matter. In a study of US metropolitan areas, Dunn, Burgess, and Ross (2005) found that the overall level of public expenditure partially accounted for the effect of income inequality on mortality.

Since these analyses are cross-sectional in nature, it is difficult to assess causal direction (or reciprocity) between distribution of income and public provision. However, the results do suggest that these factors operate in tandem. This would explain null results in countries that, in addition to (and likely due to) low levels of income inequality, also have more generous welfare states, and offer more public provision to more of their populations.