Effect of democracy on health: ecological studyBMJ 2004; 329 doi: https://doi.org/10.1136/bmj.329.7480.1421 (Published 16 December 2004) Cite this as: BMJ 2004;329:1421
- 1 Observatory of Public Policies and Health, Departamento de Salud Pública, Universidad de Alicante, Edificio de Ciencias Sociales, APDO 99, 03080 Alicante, Spain
- Correspondence to: C Álvarez-Dardet
Although the influence of democracy in preventing famines has been reported,1 there have been no empirical studies on the relation between the extent of freedom allowed by political regimes and the effect on a nation's health. We explored the effect of democracy on life expectancy and maternal and infant mortality in most countries, taking into account a country's wealth, its level of inequality, and the size of its public sector.
Politics and health
Since Virchow's seminal work, in which medicine was first proposed as a political science,2 politics has often been referred to in the medical literature, although mostly at a rhetorical level.3 Studies of political epidemiology are therefore needed, with research focusing on the effects on health of the institutions derived from political power.
Some authors have tried to determine empirically whether governments can have an effect on the incidence of specific health problems. Studies in the United Kingdom and elsewhere have measured the effect of Labour and Conservative governments on suicide rates.4 More recently, welfare state policies have been associated with health benefits in people from countries belonging to the Organisation for Economic Cooperation and Development.5
Data are now available to enable the measurement of the global impact on health of a wide range of political and economic variables. As a result the World Health Organization commission on macroeconomics and health has produced valuable information on associations between health and wealth.6 Yet information is still lacking on the relation between the extent of freedom of a particular country and the health of its people. Each year, Freedom House, a non-profit making, independent organisation promoting democracy, publishes a freedom rating for most countries, classifying them as free, partially free, or not free.7 These ratings could be used as a proxy to explore the effects of democracy on health, as has been done recently with democracy and the provision of public services.8
Key informers of data for freedom ratings
Competitive parties or political groupings
Opposition with actual power
Self government of minority groups or their participation in the government
Freedom of expression, assembly, association, education, and religion
System of rule of law
Free economic activity
Equality of opportunity
High income countries tend to have democratic governments; dictatorships and lack of civil liberties and political rights tend to be concentrated in low income countries. The level of inequality within a country may be an important determinant of health.9 10 The potential confounding effect of wealth and its distribution within a country should therefore be taken into account in research on the impact of democracy on health.
Global database of political epidemiology
We created a database from countries with data available on per capita gross national product, total government expenditure, the Gini coefficient, freedom ratings, life expectancy, and maternal and infant mortality. Information was obtained from the Human Development Report and publications of the International Monetary Fund and Freedom House. All data relate to 1998.
Freedom House generates freedom ratings for each country on the basis of data from key informers (box). The methods are described elsewhere.11
Firstly we did a simple analysis of the relations of the freedom ratings with health indicators. Then we stratified the analysis, using the World Bank's classification of economies (low, middle, and high income countries). We used a multiple linear regression model to control for the potential confounding effect of wealth (measured as per capita gross national product), level of inequality (measured with the Gini coefficient), and size of the public sector (measured as total government expenditure). To detect the additional effect of democracy, we introduced two dummy variables (demo1, passage from free country to partially free country; demo2, passage from free country to not free country) to obtain adjusted coefficients of association in the model:
Our final sample represents 98% of the world's population in 170 countries—75% of the countries and territories of the world, and 85% of those in the United Nations. Overall, 45% of the countries were free, 32% partially free, and 24% not free. Around 61% of the world's inhabitants are therefore exposed to lack of freedom by living in partially free countries (29%) or not free (32%) countries. For our sample we had available data on freedom ratings and gross national product. We obtained data on life expectancy for 158 countries, on infant mortality for 162 countries, and on maternal mortality for 140 countries.
The health indicators showed a statistically significant relation with freedom ratings: the highest levels of health were in free countries followed by the partially free countries, and the worst levels of health were in countries that were not free. The relation between health indicators and freedom ratings we observed seemed to remain along the stratum of income by countries (figure).
After adjustment in our multiple linear regression analysis, the associations persisted, with a determination coefficient near to 50%; values for life expectancy, infant mortality, and maternal mortality were 0.51, 0.47, and 0.36, respectively (see bmj.com). The inclusion of the freedom ratings in the model produced changes in the coefficient of 13% for life expectancy, 11% for infant mortality, and 6% for maternal mortality, with statistically significant coefficients.
Democracy shows an independent positive association with health, which remains after adjustment for a country's wealth, its level of inequality, and the size of its public sector.
Democracy, political rights, and civil liberties are politically modifiable variables that seem to be associated with health status. In our study, democracy showed a stronger and more significant association with indicators of health (life expectancy and infant and maternal mortality) than indicators such as gross national product, total government expenditure, or inequality in income. When all these variables were taken into account, the economic ones lost their weight, thereby increasing the importance of the effect of democracy.
Maternal mortality showed less of a linear relation with democracy than did the other indicators of health. Countries that were not free seemed to have a small advantage over those that were partially free. In our study, data on maternal mortality were more limited than the other variables. The less developed and less democratic countries tended to have insufficient data or data that were inaccurate.
One important limitation of our study was its cross sectional design. Ideally the effect of democracy or other political constructs on health needs to be examined from a historical perspective to show its cumulative effect. We acknowledge that our study has some non-differential misclassification of exposure to democracy—for example, many countries, such as Spain, Portugal, and Greece, contain populations that have lacked freedom in the past. Freedom House classified these countries as free in 1998, along with democracies of long standing, such as in Sweden and the United Kingdom. This could create a bias towards the null hypothesis, thus favouring our assumption.
Another limitation of our study is the quality of the data on health. Data from the United Nations for life expectancy and mortality are estimates from a mix of sources and methods and are therefore not real data. These were, however, the only data available from which we could test our hypothesis at a global level. Nevertheless the probability of a differential misclassification bias explaining our results produced by dictatorships worsening their health figures for international agencies seems remote.
The underlying mechanisms for the association between democracy and health are still unknown. Democracies allow for more space for social capital (for example, social networks, pressure groups),12 opportunities for empowerment, better access to information, and better recognition by government of people's needs.13 As we describe a new relation in the literature, our finding should be confirmed using longitudinal designs and potential causal pathways explored. If the relation is confirmed, the extent of freedom of a country could provide a new approach to decreasing national mortality.
The way societies organise themselves through their political regimes and their egalitarian policies could have a more important role in health than structural variables such as wealth and the size of the public sector. Increasing democratisation may be a way to counteract the deleterious effect on health of the unequal distribution of economic resources on a global scale.14
Data now available make it possible to measure the global impact on health of a wide range of political and economic variables
Freedom ratings can be used as proxies to explore the effects of democracy on other variables
After a country's wealth, level of inequality, and the size of its public sector are adjusted for, democracy has a beneficial effect on health
Adjusted models for effect of democracy in 170 countries are on bmj.com
We thank Adrian Buzzaqui, Diana Gil, Tomás Pascual, Jaime Latour, and Miquel Porta for shaping the hypothesis, making it testable, and analysing the data.
Funding The PhD programme of Public Health, University of Alicante, holds a grant from the Centro de Estudios Mario Benedetti. The Observatory of Public Policies and Health is a research consortium of the universities of Alicante, Antioquia, Porto-Alegre, and San Salvador, which holds grants from the Spanish Agency for International Cooperation and the Generalitat Valenciana.
Contributors CA-D and MTR developed the hypothesis. All the authors analysed the data, interpreted the results, and drafted the manuscript. CA-D is guarantor.
Competing interests None declared.