Analyzed the data: IB. Wrote the paper: IB AM RM JB. Designed the study: IB AM RM JB.
The authors have declared that no competing interests exist.
Smoking prevention should be a primary public health priority for all governments, and effective preventive policies have been identified for decades. The heterogeneity of smoking prevalence between European Union (EU) Member States therefore reflects, at least in part, a failure by governments to prioritise public health over tobacco industry or possibly other financial interests, and hence potentially government corruption.
The aims of this study were to test the hypothesis that smoking prevalence is higher in countries with high levels of public sector corruption, and explore the ecological association between smoking prevalence and a range of other national characteristics in current EU Member States.
Ecological data from 27 EU Member States were used to estimate univariate and multivariate correlations between smoking prevalence and the
In univariate analysis, smoking prevalence was significantly higher in countries with higher scores for corruption, material deprivation, and gender inequality; and lower in countries with higher
Corruption appears to be an important risk factor for failure of national tobacco control activity in EU countries, and the extent to which key tobacco control policies have been implemented. Further research is needed to assess the causal relationships involved.
Since cigarette smoking is the leading preventable cause of death, disability and social inequality in health in high and middle income countries
Differences in current smoking prevalence between countries in part reflect inevitable differences in stage of progression of the smoking epidemic, but also reflect the extent to which past and current governments have implemented World Health Organisation Framework Convention on Tobacco Control policies
Failure to reduce smoking prevalence may arise either from failure to enact effective tobacco control policies, or from failure to ensure compliance with them. It has previously been reported that smoking prevalence reflects the extent to which effective tobacco control policies are implemented, and that support for and the success of smoke-free policies is greater in EU countries with more advanced tobacco control policies
We investigated ecological associations between smoking prevalence in the 27 EU Member States and variables describing various national characteristics identified from existing evidence
Smoking prevalence data were taken from the Eurobarometer survey, which measures smoking prevalence in all current 27 EU Member States from samples of around 1,000 respondents (500 in smaller Member States) aged 15 years and older. Since the most recent available data for other country characteristics (below) were available for the years 2007 or 2008, we used 2008 Flash Eurobarometer data for the present analysis
We used the
We measured national wealth as
We used the ratio of total equivalised disposable income, defined as total household income divided by its age-weighted equivalent size (to take into account the size and composition of household), in the highest relative to the lowest quintiles of income
Material deprivation was measured as the proportion of the population receiving an equivalised income below 60% of the median income, using 2008 data from the Eurostat database (data for the UK and France were provisional)
Data on national spending on social benefits (transfers in cash and in kind to households and individuals, other social protection spending and administration costs) in purchasing power standards (PPS) were obtained from the Eurostat database for 2007 (values for Germany, Spain, France, Italy, Cyprus, Latvia, Lithuania, the Netherlands, Slovenia, Slovakia, Sweden, UK were provisional)
We used national average life satisfaction scores, measured on a scale from 1 to 10 from least to most satisfied, from the Second European Quality of Life Survey for 2007
The Human Development index is a composite index of national human development which combines data on life expectancy at birth, adult literacy, educational enrolment and
We used the Gender Empowerment Measure, a composite index of gender inequality in economic and political participation, and power over economic resources, provided for 26 Member States (Luxembourg unavailable) by the United Nations for 2006
Data on the proportion of the labour force (age 15–74) unemployed in 2008 were obtained from the Eurostat database
Data on the proportion of the population aged 18–24 with at most lower secondary education (early school leavers) were taken from the Eurostat database for 2008
Data on the proportion of respondents in each country reporting that religion is among three of their most important personal values were obtained from the Standard Eurobarometer survey for 2008
We used data on total quantity of raw tobacco delivered by Member States in the year 2008 provided by European Commission Directorate General for Agriculture and Rural Development
Data on the proportion of people who used to smoke but have stopped were included as a proxy indicator of the current stage of smoking epidemic
We also assessed the extent of overall national tobacco control policy enactment in individual Member States using the Joossens and Raw Tobacco Control Scale (TCS) for 2007 (max 100), and as a specific example of implementation of a currently topical policy we used smoke-free policy TCS scores for smoke free work and other public places (maximum score 22)
We measured enforcement of smoke-free policy using 2008 Flash Eurobarometer survey
We used SPSS v.17 to estimate univariate Spearman Rank correlations, and partial correlation and multiple regression with backwards exclusion to identify associations with smoking prevalence that were independently significant at p<0.05.
Mean and standard deviation values, ranges and countries at the extremes of the ranges for the variables studied are summarised in
Variable | Mean (SD) | Range | |
Minimum (Country) | Maximum (Country) | ||
Smoking prevalence (%) | 31.4 (4.8) | 22.6 (SI) | 42.1(EL) |
Per capita GDP (Euros) | 24,293 (15,923) | 4,500 (BG) | 80,500 (LU) |
Corruption Perceptions Index | 6.5 (1.7) | 3.6 (BG) | 9.3 (DK) |
Income inequality | 4.7 (1.2) | 3.4 (CZ) | 7.3 (LV) |
Material deprivation (%) | 42.2(19.4) | 14.1 (SE) | 92.8 (BG) |
Social budget (PPS |
5,615.0 (3,064.5) | 1352.2 (RO) | 13,231.3 (LU) |
Life satisfaction | 7.0 (0.8) | 5.0 (BG) | 8.5 (DK) |
Human development | 0.921 (0.041) | 0.837 (BG) | 0.965 (IE) |
Gender inequality | 0.700 (0.121) | 0.497 (RO) | 0.906 (SE) |
Unemployment rate (%) | 6.2 (1.9) | 2.8 (NL) | 11.3 (ES) |
Education (Early school leavers, %) | 14.3 (8.5) | 5.0 (PL) | 39.0 (MT) |
Religion as personal value (%) | 8.3 (6.9) | 2.0 (PT) | 27.0 (CY) |
Proportion of ex-smokers | 20.9 (4.2) | 12.7(CY) | 29.2(NL) |
Overall Tobacco Control Scale scores | 50.7 (12.8) | 35.0 (AT) | 93.0 (UK) |
Tobacco Control Scale scores for smoke free public places | 10.5 (5.2) | 2.0 (DE) | 21.0 (IE) |
Proportion of people who work away from home exposed to tobacco smoke in the workplace (%) | 22.59 (11.93) | 8.0 (SE) | 60.0 (EL) |
Proportion of indoor workers with no smoking restriction in the workplace (%) | 10.8 (7.78) | 3.0 (UK) | 38.0 (EL) |
*
Smoking prevalence was significantly correlated with the Corruption Perceptions Index (R = −0.583; p = 0.001),
To explore the possibility that this finding might differ between the EU countries that became Member States before 2004 (old EU countries- Belgium (BE), Denmark (DK), Germany (DE), Ireland (IE), Greece (EL), Spain (ES), France (FR), Italy (IT), Luxembourg (LU), the Netherlands (NL), Austria (AT), Portugal (PT), Finland (FI), Sweden (SE), the United Kingdom (UK)) and those that joined in 2004 and 2007 (new EU countries- Czech Republic (CZ), Estonia (EE), Cyprus (CY), Latvia (LV), Lithuania (LT), Hungary (HU), Malta (MT), Poland (PL), Slovenia (SI), Slovakia (SK), Romania (RO) and Bulgaria (BG)) we ran the backward regression analysis separately in these groups of countries. In the new EU Member States Corruption Perceptions Index was the only independently significant predictor of smoking prevalence (p = 0.001), and accounted for 63% of the variance in smoking prevalence. In old EU countries, the last variable retained was Gender Empowerment (p = 0.078). Since the Human Development Index is a composite measure that includes components of GDP and educational enrolment, measures of which were also included in our analysis as independent variables (in the form of
TCS scores were significantly inversely correlated with smoking prevalence (R = −0.41; p = 0.034) suggesting that smoking prevalence tends to be lower in countries with more comprehensive tobacco control policies in place. Analysis of scores for smoke-free policy revealed that these were significantly and inversely correlated with the proportion of the population reporting no smoking restrictions at work (R = −0.411; p = 0.033), but were not significantly correlated with the proportion reporting exposure to tobacco smoke in the workplace (R = −0.255; p = 0.198). Corruption Perceptions Index scores were unrelated to overall TCS scores (R = 0.130; p = 0.57) or TCS scores for the existence of smoke-free policy (R = −0.027; p = 0.892), but were strongly correlated with the prevalence of workplace exposure (R = −0.769; p<0.01) and an absence of smoking restrictions in the workplace (R = −0.454; p = 0.017). The correlation between the Corruption Perceptions Index and workplace exposure remained significant (R = −0.451; p = 0.021) after controlling for the effect of smoking prevalence. TCS scores for smoke-free policy were also not significantly correlated with any other country characteristic variables (
Smoking prevalence, and the extent to which policies to prevent smoking have been implemented, varies substantially across the EU
Our findings are based on cross-sectional ecological analyses and therefore need to be interpreted with caution, particularly in relation to any causal inference. The data we used were all collected at a time in which EU countries were entering a substantial economic recession, and in absence of more detail and more frequent observations, we are unable to determine whether these unusually stringent economic times influenced our findings. We were prevented from carrying out a more robust analysis of the longitudinal relation between corruption and smoking prevalence by the fact that the methods and sources used to construct the Corruption Perceptions Index vary from year to year, and are therefore not directly comparable over time
The heterogeneity of smoking prevalence between countries arises in part from their being at different stages of smoking epidemic
Our primary objective in this study was to determine whether corruption predicts smoking prevalence; our secondary aim was to provide some insight into the likely mechanism. Although we were only able to identify one tobacco control measure for which suitable data were available out of many that would be interesting to study, our findings for smoke-free policy indicate that the effect of corruption is not to inhibit the passage of measure to restrict smoking, but instead to reduce the extent to which these measures are observed and indeed enforced. One inference that can be drawn from this is that corrupt governments are willing to act to be seen to do the right things for health, but then choose or neglect to ensure that those measures are observed. In addition, it may be that populations in corrupt countries are, for many potential reasons, less likely to feel obliged to observe public health measures. Data from analysis stratifying countries as with high and low levels of smoke free policy implementation confirmed that the univariate association between prevalence of smoking and corruption was true independently of level of smoke free policy implementation. However multivariate analysis suggested that corruption is a significant predictor of smoking prevalence only in countries with low level of smoke free policy implementation. On the data available to us we were unable to study the implementation of other tobacco control policies in a similar way, though the World Bank has reported that in countries with higher corruption, tobacco smuggling is more common
Tobacco companies have a vested interest in and a history of inhibiting both enactment of and compliance with tobacco control policies
Correlations between variables. *Only 11 countries included.
(DOC)