Is globalization healthy: a statistical indicator analysis of the impacts of globalization on health

In the past, globalization has often been seen as a more or less economic process characterized by increased deregulated trade, electronic communication, and capital mobility. However, globalization is becoming increasingly perceived as a more comprehensive phenomenon that is shaped by a multitude of factors and events, and that is reshaping our society rapidly; it encompasses not only economic, political, and technological forces, but also social-cultural and environmental aspects. This increased global economic integration, global forms of governance, and globally inter-linked social and environmental developments are often referred to as globalization. However, depending on the researcher or commentator, globalization is interpreted as growing integration of markets and nation-states and the spread of technological advancements [1]; receding geographical constraints on social and cultural arrangements [2]; the increased dissemination of ideas and technologies [3]; the threat to national sovereignty by trans-national actors [4]; or the transformation of the economic, political and cultural foundations of societies [5]. In our view, globalization is an overarching process encompassing many different processes that take place simultaneously in a variety of domains (e.g., governance structures, markets, communication, mobility, cultural interactions, and environmental change). The pluralistic definition of globalization by Rennen and Martens [6] offers a conceptualization capturing the complexity of different dimensions;, processes; scale-levels; and linkages and pathways; characterizing the relationship between globalization and health. Hence, contemporary globalization is defined as the intensification of cross-national interactions that promote the establishment of trans-national structures and the global integration of cultural, economic, ecological, political, technological and social processes on global, supra-national, national, regional and local levels [6].

Looking at the health of populations, Martens [7] and Huynen [8], amongst others, argue that changes in drivers of disease are brought about not only by economic changes, but also by changes in the social, political, and environmental domains at local, regional, and global levels. Health improvements experienced in developed countries over the past centuries are mainly vested in social and environmental changes, whereas more recent health improvements in developing countries can be broadly related to knowledge transfer and socio-cultural determinants. Nowadays, global processes influence all these important health determinants. Hence, globalization and its underlying processes have brought about vast changes in both health determinants and related health outcomes. As a result, the geographical scale of important health issues is significantly increasing [9]. The link between global mobility and the spread of infectious diseases is perhaps the best-known health effect of globalization. However, it is only one of the many possible health implications of globalization. Many scholars have tried to conceptualize the possible linkages between globalization and health. Woodward et al. [10], for example, propose a framework based on three component circular processes of globalization: openness; cross-border flows; and rules and institutions. However, their conceptualization mainly focused on the health effects of economic globalization. Labonte and Torgerson [11] review different conceptualizations of the globalization-health relationship, resulting in a diagrammatical synthesis that mainly focuses on governmental policy changes as well as economic determinants of health, but with the inclusion of an environmental pathway. Hence, many of these approaches primarily emphasize the economic and institutional side of globalization, defining globalization in a rather narrow way. Labonte and Schrecker [12, 13] took a somewhat different approach in their framework for the Commission of Social Determinants of Health, conceptualizing how globalization affects disparities in access to social determinants of health.

Because of the multitude of underlying processes shaping the globalization-health link, ideas about globalization, health determinants and possible outcomes should be broadened. The causality of human health is multi-factorial and many population health problems are invariably embedded in a global context [8]. Taking this broader view on globalization and global health, Huynen et al. [9] developed an integrated conceptual framework for the health implications of globalization. We can conclude that a variety of both negative and positive effects are expected to influence our health in the (near) future [8, 9] (see Table 1 for examples), but it is still very uncertain what the overall health outcomes will be. Academic literature shows an ongoing polarized debate [14]. The limited empirical evidence on the multiple links between globalization and health poses a problem [15]. Many scholars urge for elaboration and possible quantitative evidence to support the hypothesized relationships [9, 10, 14‐21]. In this paper we try to answer the question if the process of globalization improves the health of populations (or not).

In this paper we use an indicator-based approach [22] linking the Maastricht Globalization Index (MGI) (a measure of globalization) to important health indicators, correcting for possible confounding factors. The MGI as well as the selected health indicators and confounders will be discussed in the following sections. Subsequently, the performed statistical analyses will be clarified.

As this research focuses on indicators of mortality to highlight an important side of global health outcomes, it is interesting to look at some of the drivers directly related to mortality (or factors linking globalization and mortality) identified in the current body of research in this field. Martens [7] claims that increased income levels can result in a decrease in mortality rates, which ultimately impacts life expectancy rates positively. Burns, Kentor, and Jorgenson [35] focus on infant mortality and discuss a country's level of internal development and the related dependencies on the world economy (affecting domestic institutional structures) as a main driver. However, the level of a country's development and the resulting impact on infant mortality is not fully uncovered. Other factors they found to be related to infant mortality are the macro level effect of export commodity concentration, GDP per capita, health expenditures per capita, secondary education, and organic water pollution. They identified several mediating factors between global dependence and infant mortality: quality of water and health care, level of internal development such as GNP per capita, the role of ecology (pollution and misuse of land) as well as public health factors (lack of resources for public health can be seen with indicators such as scarcity of inoculation to childhood diseases, and the lack of trained medical personnel for pre-and post-natal care and for assistance with birth process itself) [35]

Cornia et al. [34] associate globalization mainly with economic changes, such as economy policy, protectionism, costs of technological transfer, privatization, market liberalization, trade and financial liberalization. Looking at the slow progress in infant mortality rates over the past decades, the authors suggest that many factors can be responsible for these slow improvements such as slow growth of household incomes, greater income volatility, shifts in health financing, amongst others. In this study, the effects of globalization are captured by comparing the timeframe of 1980-2000 (the era of globalization) with other timeframes, indicating changes in the following indicators: growth of GDP per capita, economic stability, income inequality, inflation and prices of basic goods, taxation and public health expenditure and health financing, migration and family arrangements, technical progress in health, smoking drinking and obesity, and random shocks [34].

The results of our analysis (Spearman's correlations, and simple and multiple linear regression analyses) indicate that the infant morality rate, under-five mortality rate and adult mortality rate all show a negative association with the process of globalization (as measured by the MGI). Specifically, technological globalization and socio-cultural globalization are shown to have strong associations with the selected health indicators. The multivariate analyses show that different confounders have been found to be significant in the three final models. Specifically, for Ln Infant morality rate confounders accounting for primary and secondary education and public health expenditures have been found to be significant. For the Ln Under-five mortality rate, next to the confounders for primary and secondary education, smoking prevalence under females have shown to be significant in the final model. Lastly, for the model of Ln Adult mortality rate, only a confounder on access to improved sanitation facilities has been significant. These factors, thus can possibly function as confounders in the relationships between the respective mortality rates with the MGI. However, the confounders in the final models could also be important mediating/causal factors in the association between the mortality rates and the MGI. Either way, in all multivariate models, the association between globalization and the mortality indicators remains significant after controlling for confounding factors.

Given the limited existing quantitative information on the association between globalization and health, the results might provide a crude initial indication of the potential advantageous effect of globalization on health. In view of the arguments that globalization provides winners and losers, and might be seen as a disequalizing process, we should perhaps be careful in interpreting the observed positive association between the MGI and health, as simple evidence that globalization is mostly good for our health. Important to note is that all indicators and data are on the country level, without a specific spatial dimension. Globalization interacts with health at levels that make measurement difficult, e.g., trans-border environmental issues, cultural transformations and a so-called 'global consciousness'. For example, the data do not show us that the most globalized countries might have lower mortality rates because they have exported their unhealthy pollution and other externalities of the production of goods and services they enjoy (and which contribute to their health) to people and environments in other parts of the world. Hence, some of the winners might be benefiting from their high levels of globalization at the expense of others. Importantly, it should also be noted that he MGI represents actual levels of globalization across different domains, rather then the mere implementation of neoliberal policies.

In this paper, we consider the impact of the recent process of globalization on the health of populations. Looking at the results, globalization can be characterised as both more complicated and more surprising than was anticipated. One clear lesson can be learned from the many global assessments that have been produced over the past decades: dogmatic predictions regarding the earth's future are unreliable, ill-founded and misleading, and can be politically counterproductive. So, this analysis is beset with the uncertainties and assumptions that apply to any global statistical indicator analysis [41]. For example, if consumerism and global economic processes do have polluting and other unhealthy negative side-effects for some, it needs to be asked which direction these dynamics need to take for sustainable health for all. Furthermore, this analysis is based on 'present day data'. As the globalizing processes intensify over time, the indirect impacts of human-induced disruption of global biogeochemical cycles and global climate change, and their impacts on human health, may start to become more apparent [42, 43]. Borghesi and Vecelli [44] also state that the available empirical evidence suggests that the current process of globalization is unsustainable in the long run unless we introduce new institutions and policies able to govern it, a similar claim being made by Tisdall [45] and Watanabe [46] looking at economic globalization only. Schrecker et al. [47] reject furthermore the presumption that globalization will yield health benefits as a result of its contribution to rapid economic growth and associated reduction in poverty.

Hence, for future research we hypothesize that a country performance might be classified into four categories (adopted from [48]: vicious cycle (low globalisation, high mortality), globalisation-lopsided (high globalisation, high mortality), health-lopsided (low globalisation, low mortality) or virtuous cycle (high globalisation, low mortality).

We hypothesize that a country performance might be classified into four categories (adopted from [48]: vicious cycle (low globalisation, high mortality), globalisation-lopsided (high globalisation, high mortality), health-lopsided (low globalisation, low mortality) or virtuous cycle (high globalisation, low mortality). In the vicious cycle, any efforts to properly integrate into the global process are yet unsuccessful, but might even result in (temporary) adverse health effects (e.g. Ghana). Globalization-lopsided may happen when integration into the globalization process has not yet resulted in major health benefits, or might have even resulted in increasing health problems (e.g. Egypt). Health-lopsided might happen, when health improvements occur that are not related to any globalization benefits, but due to other domestic polices or developments (e.g. Peru). In a virtuous cycle, countries might have benefited from their integration into the globalization process, while averting any associated health risks. It is important to note, however, that for some countries the virtuous cycle could be the result of bias due to causal sequence (i.e. did all the major improvement in health already occurred prior to the modern-day globalization process?) (e.g. the Netherlands).

Example countries:

The important issue for policy purposes, of course, is how a country may move towards the virtuous cycle and several important research questions can be identified. How have countries changed their location over time and due to which underlying mechanisms? If countries find themselves in a viscous cycle, should they first focus on enhancing their health status or on enhancing their integration into the globalization process? Looking at the health-lopsided countries and the globalisation lop-sided countries, which have a higher chance of reaching a virtuous circle and which are most at risk from shifting to a vicious circle? How can health-lopsided countries make sure that their health status is not compromised by any efforts to improve their integration in the globalization process? How can globalisation-lopsided countries increase their health benefits of globalisation? And finally, will the countries that now experience a virtuous cycle also persist to remain in this category in the future?

What is clear is that the increasing complexity of our global society means that sustainable health cannot be addressed from a single perspective, country, or scientific discipline. Changes in human health in the context of globalization are far more complex than health issues that had to be tackled in the past. As addressed by others (e.g., Borgesi and Vecelli [44]), it is our hope that a further analysis of health impacts of globalization may help in adjusting and optimising the process of globalization on every level in the direction of a sustainable and healthy development [57]. To this end, extensive empirical work is needed to identify the relevant causal mechanisms underlying the influence of globalization on human health.

i The variables excluded from the analysis based on these reasons are: from WHOSIS [30, 31]: adult literacy rate (%); adolescent fertility rate (%); antenatal care coverage - at least four visits (%); births attended by skilled health personnel (%); prevalence of HIV among adults aged ≥15 years (per 100 000 population); population with sustainable access to improved drinking water sources (%) total; population with sustainable access to sanitation (%) total; prevalence of current tobacco use amongst adolescents (13-15 years (%) both sexes; prevalence of current tobacco use amongst adults (≥15 years) (%) both sexes; deaths amongst children under 5 years of age due to malaria (%); deaths due to HIV/Aids (per 100 000 population per year). Confounders assessed and excluded for the same reasons from the World DataBank [36] include: malnutrition prevalence, weight for age (% of children under 5); literacy rate adult female (% of females ages 15 and above); literacy rate adult male (% of males ages 15 and above); total enrolment, primary, female (% net); total enrolment, primary, male (% net); pregnant women receiving prenatal care (%); and births attended by skilled health staff (% of total).

ii Variables that did not satisfy the criteria of functioning as a confounder on the MGI-health indicator relationships are: 'Smoking prevalence, males (% of adults) 2006'; and 'Prevalence of HIV, total (% of population ages 15-49), 2007' [36]

iii The following tests have been used to assess whether the indicators used display a normal distribution: Frequency histograms (for a graphical assessment of normality of distribution); P-P plots and Q-Q plots (have been used as a complementary graphical assessment tool for the normality of the distribution of the variable, thus in addition to the frequency histograms); Boxplots (to graphically check for outliers and skewness); the Shapiro-Wilk's W-test (as a formal test for normality has been used[37]. However, results of the W-test have been treated with care and placed within the context of the insights gained from all the other normality tests performed); descriptive statistics have been used to numerically assess skewness and kurtosis (criterion used for skewness: the skewness-statistic must lie between +2 and -2; criterion used for kurtosis: the kurtosis-statistic must lie between +2 and -2)[38].

iv All assumptions of least squares regression analysis have been checked and could be met by the models. The assumption of linearity has been checked with scatterplots and linear curve estimation. The normality of the probability distribution of the error terms of prediction have been tested by generating frequency histograms of the standardized residuals. To test for homoscedasticity, the standardized residuals and the standardized predicted values have been plotted in a scatterplot to observe a random pattern. For the assumption of mean independence, residual statistics and scatterplots of the residual against the predicted values have been used to verify that the mean of the residuals would be approximately zero. In addition, all models have been checked for multivariate outliers by generating Cook's Distances[58]. When the Cook's Distance is higher than 1.0, a case is considered an outlier and is deleted from the analysis.

v Note: The step-wise backward linear regression analyses have been performed manually.

vi When reporting on statistical results, the term 'significance' refers to 'statistical significance'.

vii 'GDP per capita growth (annual%) 2008'; 'Health expenditure, public (% of GDP), 2007'; 'Prevalence of undernourishment (% of population) 2006)'; Immunization, measles (% of children ages 12-23 months, 2008'; 'Improved water source (% of population with access) 2000-2006'; 'Total enrollment, primary (% net) 2000-2008'; 'Smoking prevalence, females (% of adults) 2006'; 'Literacy rate, adult total (% of people ages 15 and above) 2000-2008'; 'Total fertility rate (per woman), 2006'.