The role of healthcare and education systems in co-occurrence of health risk behaviours in 27 European countries

The role of healthcare and education systems in co-occurrence of health risk behaviours in 27... Abstract Background Contextual factors play an important role in health and related behaviours. This study aims to examine the association of co-occurrence of five health-risk behaviours with healthcare and education contextual factors using nationally representative samples from 27 European countries. Methods Data were from Eurobarometer 72.3, 2009. The outcome was a count variable ranging from 0 to 5 indicating co-occurrence of five health-risk behaviours, namely smoking, excessive alcohol consumption, non-frequent fresh fruit consumption, physical inactivity and non-dental check-ups. Public expenditures on healthcare and education as a percentage of GDP and quality of healthcare and education at a country-level were used as contextual factors. A set of multilevel Poisson regression models were conducted to examine the associations between co-occurrence of health-risk behaviours and each of the contextual factors considering age, gender, marital status, urbanisation, individual socioeconomic positions (education, subjective social status or difficulty in paying bills) and GDP per capita. Results The total population was 23 842. Greater expenditures on healthcare and education, and better quality of healthcare systems had negative associations with co-occurrence of health-risk behaviours in the model adjusted for all individual demographic indicators. However, statistical significance disappeared after adjusting for socioeconomic indicators and GDP per capita. Conclusion While the study highlights the importance of developing high-quality healthcare and education systems generously supported by public fund in relation to co-occurrence of health-risk behaviours, the influence of contextual factors in adopting health-related behaviours is probably attenuated by individual socioeconomic factors. Introduction Modifiable health-related behaviours tend to be grouped together in the same sectors of the population, with those at the bottom of the socioeconomic hierarchy more likely to engage in health-risk behaviours and less likely to adopted health-promoting ones.1 Among these behaviours, the ‘big four’ behaviours, namely smoking, excessive alcohol consumption, poor diet and physical inactivity have excessively been studied in terms of their devastating impact on mortality and morbidity, particularly cancer, cardiovascular disease and diabetes.2–4 The same set of health-related behaviours is also related to oral health5–8 in addition to dental check-ups.9 Traditionally, epidemiological studies have focused on the significant impact of a single health-risk behaviour on a specific health outcome.10 However, co-occurrence of health-risk behaviours in the same individuals, defined as a count number of health-risk behaviours,11 exerts multiplicative effects on disease risk.12 The social environment influences behaviour by shaping norms, enforcing patterns of health-related behaviours, providing or obstructing environmental opportunities to engage in certain behaviours.1 Earlier studies have reported an association between co-occurrence of health-risk behaviours and contextual factors, such as neighbourhood socioeconomic factors,13 area-level deprivation14 and workplace social capital.15 Although the evidence has shown the influence of contextual factors on co-occurrence of health-risk behaviours, to the best of our knowledge, no other study has examined the association between healthcare/education systems and co-occurrence of health-risk behaviours. Better healthcare systems could promote equitable access to preventive services,16 and enhance perception of health security,17 and subsequently reduce anxiety related to availability and quality of healthcare services at the time of sickness.18 Such anxiety could impact on individual’s psychosocial wellbeing which has an established association with adopting health-risk behaviours.19 Similarly, equitable access to high-quality education system will enable equal opportunities to access high quality education, hence increasing the probabilities of having better jobs, and improving their chances for social mobility, and consequently reducing the social gradients in behaviours and health.20 Furthermore, it would possibly impact on individual’s ability to access and interpret information, improve sense of learnt effectiveness and personal control, particularly among the least affluent, hence leading to healthier lifestyles.21 Thus, it is reasonable to hypothesize that healthcare and education systems can potentially influence co-occurrence of health-risk behaviours. Methods Data source and study sample This study is a secondary analysis of cross-country population-based data from Eurobarometer 72.3, which is a cross-sectional interview survey in Europe. This survey used nationally representative samples from each of the European countries and was conducted in October 2009.22 A 2-stage, random (probability) sampling design was used for sample selection. The data were collected by face-to-face interview in people’s home from 2 to 19 October 2009 by the TNS Opinion and Social through its network of national institutes in the respective national language.22 The regular sample size was 1000 participants from each country. Intentionally more participants were selected from the respective group than would typically be done if everyone in the sample had an equal chance of being selected.25 We included participants from 27 European countries who answered all questions pertaining to health-related behaviours and demographic/socioeconomic indicators. Given that smoking and drinking are illegal for those under 18 years in most of the European countries, the analysis was limited to those aged 18 and over. Ethical approval was not required for this secondary data analysis. Variables Outcome The survey had the questions related to check-up and medical screening, oral health, alcohol habits, smoking habits and sport and physical activity. We selected the five behaviours, namely smoking, alcohol use, diet, physical activity and dental check-ups because of their relationships with a number of chronic diseases and with oral health. In addition, these five health-related behaviours had similar socio-demographic determinants. The co-occurrence of health-risk behaviours was estimated based on how many of the following five health-risk behaviours participants reported: smoking, excessive alcohol consumption, infrequent consumption of fresh fruits, physical inactivity and not attending dental check-ups. The behaviours were dichotomised into binary options (health-risk behaviours were coded as 1, while health-promoting behaviours were coded as 0) (table 1). Smoking was indicated by current smokers (vs former- or never-smokers). Excessive alcohol consumption was defined as having five or more drinks on one occasion at least once a week in the last 12 months. Infrequent consumption of fresh fruits was indicated by consuming fresh fruits “from time to time”, “rarely” or “never”. Physical inactivity was created by combining two questions. EU Working Group indicated that physical activity at least five times a week is recommended for European adults.23 As the dataset did not allow using this definition, in this study at least four times a week was used as a cut-off point for physical activity.24 Dental check-ups were indicated by use in the past 12 months. According to the NICE guideline, people who are not at risk of or from oral disease may not need dental check-ups once a year.25 However, given this is a sample of the whole population including those with at higher risk of oral diseases, we thought dental visit during the past 12 months would be a reasonable cut-off point. Table 1 Measurement of health-risk behaviours   Measurement in the survey  Classification of health-risk behaviours for this study  Smoking  Regarding smoking cigarettes, cigars or a pipe, which of the following applies to you?: (1) you smoke at the present time, (2) you used to smoke but you have stopped, (3) you have never smoked and (4) don’t know  Smoking at the present time  Excessive alcohol consumption  How often in the past twelve months have you had five or more drinks on one occasion?: (1) several times a week, (2) once a week, (3) once a month, (4) less than once a month, (5) never, and (6) don’t know/refusal  Having five or more drinks on one occasion at least once a week in the last twelve months  Infrequent consumption of fresh fruits  How often do you eat fresh fruit?: (1) often, (2) from time to time, (3) rarely, (4) never and (5) don’t know  Consuming fresh fruit from time to time, rarely or never  Physical inactivity  There are two questions: (a) how often the person exercises or plays sport (b) how often the participant engages in a physical activity outside sport such as cycling or walking from a place to another, dancing, gardening etc.  Doing physical activity less than four times a week (Except the participates who answered: (1) five times a week or more for either the two questions regardless of any other answer; those who answered (2) three to four times a week for either question and (2) three to four times a week or (3) one to two times a week for another question)  For each of the questions; (1) five times a week or more, (2) three to four times a week, (3) one to two times a week, (4) one to three times a month, (5) less often, (6) never, and (7) don’t know  Not attending dental check-ups  Did you have dental check-ups in the last twelve months, whether or not as part of any treatment?: (1) yes, own initiative, (2) yes, doctor’s initiative, (3) yes, screening programme, (4) no, and (5) don’t know  Having dental check-ups in the last 12 months regardless own initiative, doctor's initiative or screening programme    Measurement in the survey  Classification of health-risk behaviours for this study  Smoking  Regarding smoking cigarettes, cigars or a pipe, which of the following applies to you?: (1) you smoke at the present time, (2) you used to smoke but you have stopped, (3) you have never smoked and (4) don’t know  Smoking at the present time  Excessive alcohol consumption  How often in the past twelve months have you had five or more drinks on one occasion?: (1) several times a week, (2) once a week, (3) once a month, (4) less than once a month, (5) never, and (6) don’t know/refusal  Having five or more drinks on one occasion at least once a week in the last twelve months  Infrequent consumption of fresh fruits  How often do you eat fresh fruit?: (1) often, (2) from time to time, (3) rarely, (4) never and (5) don’t know  Consuming fresh fruit from time to time, rarely or never  Physical inactivity  There are two questions: (a) how often the person exercises or plays sport (b) how often the participant engages in a physical activity outside sport such as cycling or walking from a place to another, dancing, gardening etc.  Doing physical activity less than four times a week (Except the participates who answered: (1) five times a week or more for either the two questions regardless of any other answer; those who answered (2) three to four times a week for either question and (2) three to four times a week or (3) one to two times a week for another question)  For each of the questions; (1) five times a week or more, (2) three to four times a week, (3) one to two times a week, (4) one to three times a month, (5) less often, (6) never, and (7) don’t know  Not attending dental check-ups  Did you have dental check-ups in the last twelve months, whether or not as part of any treatment?: (1) yes, own initiative, (2) yes, doctor’s initiative, (3) yes, screening programme, (4) no, and (5) don’t know  Having dental check-ups in the last 12 months regardless own initiative, doctor's initiative or screening programme  Country-level independent variables Public expenditure on health and education and quality of healthcare and education systems were used as the contextual exposures of interest. Five-year average public expenditure on health as a percentage of Gross Domestic Product (GDP) in each country from 2005 to 2009 reported by the World Bank26 was used as health expenditure. Quality of healthcare system was indicated by Euro Health Consumer Index (EHCI).27 EHCI consisted of 38 indicators of 6 sub-disciplines: (1) Patient rights and information, (2) e-Health, (3) Waiting time for treatment, (4) Outcomes, (5) Range and reach of services and (6) Pharmaceuticals. For each of the sub-domains, the country score was calculated as a percentage of the full score.27 The scores were divided by 100 for better interpretations of the results and used as an indicator of the quality of healthcare system. Five-year average public expenditure on education as a percentage of GDP in each country from 2005 to 2009 by Eurostat28 was used as education expenditure. Quality of education system was indicated by the Education for All Development Index (EDI) in 2006.29 EDI is used as a proxy measure for each of the four “Education for All” goals, and each of these EDI components is assigned equal weight in the overall index in accordance with the principle of considering each goal as being of equal importance. The EDI value for a given country is the arithmetic mean of the four proxy indicators, namely universal primary education (the percentage of primary-school-age children who are enrolled in either primary or secondary school), adult literacy (the adult literacy rate for those aged 15 and above), quality of education (the survival rate to Grade 5) and gender (a simple average of the three gender parity indexes for primary education, secondary education and adult literacy, with each being weighted equally). The percentage of the score was used in this study. Given that expenditure on healthcare and education can be susceptible to economic condition of the country, we included GDP per capita in the analysis. Individual-level independent variables Socioeconomic position was indicated by three measures, namely education, subjective social status and difficulty in paying bills. Education was measured by the age when participants stopped full-time education, and was categorised into three groups; 20 years and older, 16–19 years old and 15 years or less. This is because in most European countries the minimum age for compulsory education is 15 years or older, and secondary school education is usually completed when individuals are under 20 years.30 Participants still studying were included in the category corresponding to their age. Education is an important indicator of socioeconomic position as it is comparable across countries.31 Subjective social status reflects one’s perception of own status in the respective community/country, and is an important determinant for health-related behaviours.32 In this survey, participants were asked to place themselves on a ladder indicating their perception of own positions in their respective society on a scale of 1–10, hence reflecting perception of social standing. For better interpretation of the data and to distinguish between the upper and the lower halves of the scale, subjective social status was categorised into quartiles. The quartiles indicted: highest (7–10), second highest (6), second lowest (5) and lowest (1–4). Difficulty in paying bills reflects financial ability to pay bills at the end of the month during the last 12 months, and has three categories; most of the time, from time to time and almost never/never, which was used in previous studies.33 Demographic factors included gender, age, urbanization and marital status. Age was used as a continuous variable. Urbanization has three categories (rural area or village, small or middle sized town and large town). Marital status was dichotomized to indicate; married/living with a partner, vs single/divorced/separated/widowed. The selection of the variables included in the models was based on their relationships with the outcome variable and with the main exposure. For example, expenditures on health and education are relevant to the country GDP; therefore, inclusion of such a variable could affect the relationship between the exposure and the outcome variable. Statistical analysis In order to examine the associations between co-occurrence of health-risk behaviours and each of contextual factors, a two-level Poisson regression model with log link function34 was conducted because the response variable was a count and the mean (1.70) was almost the same as variance (1.41) for this variable (range 0–5). The fixed- and random-parameter estimates for the two-level Poisson regression models were calculated by MLwiN version 2.35 using the iterative generalized least squares algorithm with the penalized quasi-likelihood, second-order approximation procedure.35 We did not use weight for the multilevel analyses because when the cluster sizes are large (∼750 individuals in each cluster), the standard errors achieved remarkable consistency either weighted or unweighted.36 Three multilevel models were constructed for each of contextual factors. The construction of the models aimed at examining binary association between contextual factors and behaviours, then adjusted for demographic factors and finally, included socioeconomic factors at the country- and individual-levels. Model 1 examined the associations between co-occurrence of health-risk behaviours and each of contextual factors adjusted for all demographic characteristics (age, gender, marital status and urbanization). Model 2 was adjusted for all demographic characteristics and the three socioeconomic indicators. Finally, Model 3 was additionally adjusted for GDP per capita. In each model, rate ratios were reported for fixed parts. For random parts, the proportional change in inter-country variance estimates (σ2) of the different model was imputed. A median rate ratio, which is the median of the rate ratios of pair-wise comparisons of countries, was calculated as e^(0.954*√σ2).37 Results After excluding the participants with missing values, 23 842 individuals were included in this study (the valid percentage: 91.7%). The weighted percentages of each health-risk behaviour among the total population were 30.0% for smoking, 22.4% for excessive alcohol consumption, 36.9% for infrequent fruit consumption, 50.7% for physical inactivity and 36.8% for non-use of dental check-ups. Participants reported no health-risk behaviours were 14.9% (n = 3897). The most frequent co-occurrence numbers were one and two health-risk behaviours with 29.3% (n = 7311) and 29.8% (n = 6835) respectively. Those with three health-risk behaviours accounted for 17.9% (n = 4025) with 6.5% (n = 1440) for four health-risk behaviours, 1.7% (n = 334) for five health-risk behaviours and 16.4% (n = 3897) for no risk behaviour. The mean number of health-risk behaviours was 1.70 (SD = 1.19, Variance = 1.41, Median = 2.5). The individual- and country-level predictor variables are presented in table 2. The results of multilevel Poisson regression models examining the associations between co-occurrence of health-risk behaviours and contextual factors are presented in tables 3 and 4. In an empty model, variance at country-level was 0.023 (SE: 0.006), hence the median of all pairwise comparisons between countries gives a rate ratio of 1.16 (95%CI 1.14–1.17). This indicated that there were variations in co-occurrence of health-risk behaviours between countries. A unit increase in health expenditure was associated with lower risk of co-occurrence of risk behaviours with the rate ratio of 0.96 (95%CI 0.93–0.99) in the model accounting for all demographic factors (table 3). After accounting for socioeconomic positions, the association lost significance. Individuals in the countries with higher quality of healthcare system were less likely to have health-risk behaviours with the rate ratio of 0.93 (95%CI 0.88–0.97) in the model accounting for all demographic factors, again statistical significance disappeared after adjusting for socioeconomic position and GDP per capita. Table 2 Data description for analytical multilevel models Level 1: Individual-level predictor variable (N = 23 842)    N  Weighted %  (95%CI)  Mean±SE  Gender    Male  10 453  48.00  (47.00–49.01)    Female  13 389  52.00  (50.99–53.00)  Age in years  23 842  47.71±0.18 (range 18–96)  Marital status    Married/with a partner  15 608  65.77  (64.82–66.71)    Single  8234  34.23  (33.29–35.18)  Urbanisation    Rural area/village  8624  34.79  (33.84–35.75)    Small/middle-sized town  8360  39.94  (38.95–40.93)    Large town  6858  25.28  (24.44–26.13)  Education    20 years or older  7574  29.48  (28.58–30.39)    16-19 years old  11 103  46.48  (45.48–47.48)    15 years old or less  5165  24.05  (23.20–24.91)  Subjective social status    Highest  6889  27.21  (26.34–28.10)    Second highest  5091  24.28  (23.42–25.17)    Second lowest  6934  29.94  (29.03–30.88)    Lowest  4928  18.56  (17.81–19.34)  Difficulty in paying bills    Almost never/never  15 292  66.16  (65.22–67.09)    From time to time  6351  26.16  (25.30–27.05)    Most of the time  2199  7.67  (7.18–8.19)  Level 2: country-level predictor variable (N = 27)  Mean  Standard Deviation  Variance  Range  Healthcare expenditure (% of GDP)  6.32  1.50  2.26  2.8–8.7  Quality of healthcare  6.65  1.06  1.12  4.48–8.75  Education expenditure (% of GDP)  5.25  1.01  1.02  6.45–8.10  Quality of education  98.15  0.95  0.91  95.5–99.4  GDP per capita (in US dollars/1000)  31.93  17.98  323.43  6.74–100.74  Level 1: Individual-level predictor variable (N = 23 842)    N  Weighted %  (95%CI)  Mean±SE  Gender    Male  10 453  48.00  (47.00–49.01)    Female  13 389  52.00  (50.99–53.00)  Age in years  23 842  47.71±0.18 (range 18–96)  Marital status    Married/with a partner  15 608  65.77  (64.82–66.71)    Single  8234  34.23  (33.29–35.18)  Urbanisation    Rural area/village  8624  34.79  (33.84–35.75)    Small/middle-sized town  8360  39.94  (38.95–40.93)    Large town  6858  25.28  (24.44–26.13)  Education    20 years or older  7574  29.48  (28.58–30.39)    16-19 years old  11 103  46.48  (45.48–47.48)    15 years old or less  5165  24.05  (23.20–24.91)  Subjective social status    Highest  6889  27.21  (26.34–28.10)    Second highest  5091  24.28  (23.42–25.17)    Second lowest  6934  29.94  (29.03–30.88)    Lowest  4928  18.56  (17.81–19.34)  Difficulty in paying bills    Almost never/never  15 292  66.16  (65.22–67.09)    From time to time  6351  26.16  (25.30–27.05)    Most of the time  2199  7.67  (7.18–8.19)  Level 2: country-level predictor variable (N = 27)  Mean  Standard Deviation  Variance  Range  Healthcare expenditure (% of GDP)  6.32  1.50  2.26  2.8–8.7  Quality of healthcare  6.65  1.06  1.12  4.48–8.75  Education expenditure (% of GDP)  5.25  1.01  1.02  6.45–8.10  Quality of education  98.15  0.95  0.91  95.5–99.4  GDP per capita (in US dollars/1000)  31.93  17.98  323.43  6.74–100.74  Table 3 Rate ratios for the associations between healthcare domain and co-occurrence of health-risk behaviours from multilevel Poisson regression models   Health expenditure  Quality of healthcare    Model 1  Model 2  Model 3  Model 1  Model 2  Model 3    RR  RR  RR  RR  RR  RR    (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  Fixed  Country-level variables  Health expenditure  0.96*  0.98  0.99          (0.93–0.99)  (0.96–1.01)  (0.96–1.03)        Quality of healthcare    0.93*  0.97  0.98          (0.88–0.97)  (0.93–1.01)  (0.92–1.04)  GDP per capita  0.99      0.99        (0.99–1.00)      (0.99–1.00)  Individual-level variables  Age in years  0.99*  0.99*  0.99*  0.99*  0.99*  0.99*    (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  Gender (ref: male)  Female  0.74*  0.74*  0.74*  0.74*  0.74*  0.74*    (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  Marital status (ref: married individuals)  Single  1.14*  1.11*  1.11*  1.14*  1.11*  1.11*    (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  Urbanisation (ref: rural area or village)  Small/middle sized town  1.02  1.04*  1.04*  1.02  1.04*  1.04*    (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  Large town  1.04*  1.07*  1.07*  1.04*  1.07*  1.07*    (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  Education (ref: 20 years or older)  16–19 years    1.15*  1.15*    1.15*  1.15*      (1.13–1.18)  (1.12–1.18)    (1.12–1.18)  (1.12–1.18)  15 years or less    1.23*  1.23*    1.23*  1.23*      (1.19–1.27)  (1.19–1.27)    (1.19–1.27)  (1.19–1.27)  Subjective social status (ref: highest)  Second highest    1.08*  1.08*    1.08*  1.08*      (1.05–1.11)  (1.05–1.11)    (1.05–1.11)  (1.05–1.11)  Second lowest    1.07*  1.07*    1.07*  1.07*      (1.04–1.11)  (1.04–1.11)    (1.04–1.11)  (1.04–1.11)  Lowest    1.17*  1.17*    1.17*  1.17*      (1.13–1.21)  (1.13–1.20)    (1.13–1.21)  (1.13–1.20)  Difficulty in paying bills (ref: never)  From time to time    1.16*  1.16*    1.16*  1.16*      (1.14–1.19)  (1.14–1.19)    (1.14–1.19)  (1.14–1.19)  Most of the time    1.29*  1.29*    1.29*  1.29*      (1.25–1.33)  (1.25–1.33)    (1.25–1.33)  (1.25–1.33)  Random  σ2  0.018  0.011  0.010  0.016  0.010  0.010  (SE)  0.005  0.003  0.003  0.004  0.003  0.003  MRR  1.14*  1.11*  1.10*  1.13*  1.10*  1.10*  (95% CI for MRR)  (1.13–1.15)  (1.10–1.11)  (1.09–1.11)  (1.12–1.14)  (1.09–1.11)  (1.09–1.11)    Health expenditure  Quality of healthcare    Model 1  Model 2  Model 3  Model 1  Model 2  Model 3    RR  RR  RR  RR  RR  RR    (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  Fixed  Country-level variables  Health expenditure  0.96*  0.98  0.99          (0.93–0.99)  (0.96–1.01)  (0.96–1.03)        Quality of healthcare    0.93*  0.97  0.98          (0.88–0.97)  (0.93–1.01)  (0.92–1.04)  GDP per capita  0.99      0.99        (0.99–1.00)      (0.99–1.00)  Individual-level variables  Age in years  0.99*  0.99*  0.99*  0.99*  0.99*  0.99*    (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  Gender (ref: male)  Female  0.74*  0.74*  0.74*  0.74*  0.74*  0.74*    (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  Marital status (ref: married individuals)  Single  1.14*  1.11*  1.11*  1.14*  1.11*  1.11*    (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  Urbanisation (ref: rural area or village)  Small/middle sized town  1.02  1.04*  1.04*  1.02  1.04*  1.04*    (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  Large town  1.04*  1.07*  1.07*  1.04*  1.07*  1.07*    (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  Education (ref: 20 years or older)  16–19 years    1.15*  1.15*    1.15*  1.15*      (1.13–1.18)  (1.12–1.18)    (1.12–1.18)  (1.12–1.18)  15 years or less    1.23*  1.23*    1.23*  1.23*      (1.19–1.27)  (1.19–1.27)    (1.19–1.27)  (1.19–1.27)  Subjective social status (ref: highest)  Second highest    1.08*  1.08*    1.08*  1.08*      (1.05–1.11)  (1.05–1.11)    (1.05–1.11)  (1.05–1.11)  Second lowest    1.07*  1.07*    1.07*  1.07*      (1.04–1.11)  (1.04–1.11)    (1.04–1.11)  (1.04–1.11)  Lowest    1.17*  1.17*    1.17*  1.17*      (1.13–1.21)  (1.13–1.20)    (1.13–1.21)  (1.13–1.20)  Difficulty in paying bills (ref: never)  From time to time    1.16*  1.16*    1.16*  1.16*      (1.14–1.19)  (1.14–1.19)    (1.14–1.19)  (1.14–1.19)  Most of the time    1.29*  1.29*    1.29*  1.29*      (1.25–1.33)  (1.25–1.33)    (1.25–1.33)  (1.25–1.33)  Random  σ2  0.018  0.011  0.010  0.016  0.010  0.010  (SE)  0.005  0.003  0.003  0.004  0.003  0.003  MRR  1.14*  1.11*  1.10*  1.13*  1.10*  1.10*  (95% CI for MRR)  (1.13–1.15)  (1.10–1.11)  (1.09–1.11)  (1.12–1.14)  (1.09–1.11)  (1.09–1.11)  N = 23842; Eurobarometer 72.3, 2009 Model 1: considering individual-level demographic characteristics, Model 2: considering individual-level demographic and socioeconomic characteristics, Model 3: considering individual-level characteristics and country-level variable (GDP per capita). * P < 0.05. Table 4 Rate ratios for the associations between education domain and co-occurrence of health-risk behaviours from multilevel Poisson regression models   Education expenditure  Quality of education    Model 1  Model 2  Model 3  Model 1  Model 2  Model 3    RR  RR  RR  RR  RR  RR    (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  Fixed  Country-level variables  Education expenditure  0.95*  0.97  0.98          (0.90–0.99)  (0.94–1.01)  (0.94–1.01)        Quality of education        0.95  0.98  1.00          (0.90–1.00)  (0.94–1.03)  (0.95–1.05)  GDP per capita      0.99      0.99        (0.99–1.00)      (0.99–1.00)  Individual-level variables  Age in years  0.99*  0.99*  0.99*  0.99*  0.99*  0.99*    (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  Gender (ref: male)        Female  0.74*  0.74*  0.74*  0.74*  0.74*  0.74*    (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  Marital status (ref: married individuals)        Single  1.14*  1.11*  1.11*  1.14*  1.11*  1.11*    (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  Urbanisation (ref: rural area or village)        Small/middle sized town  1.02  1.04*  1.04*  1.02  1.04*  1.04*    (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  Large town  1.04*  1.07*  1.07*  1.04*  1.07*  1.07*    (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  Education (ref: 20 years or older)  16–19 years    1.15*  1.15*    1.15*  1.15*      (1.12–1.18)  (1.12–1.18)    (1.13 1.18)  (1.12–1.18)  15 years or less    1.23*  1.23*    1.23*  1.23*      (1.19–1.27)  (1.19–1.27)    (1.19–1.27)  (1.19–1.27)  Subjective social status (ref: highest)  Second highest    1.08*  1.08*    1.08*  1.08*      (1.05–1.11)  (1.05–1.11)    (1.05–1.11)  (1.05–1.11)  Second lowest    1.08*  1.07*    1.08*  1.07*      (1.05–1.11)  (1.04–1.11)    (1.05–1.11)  (1.04–1.11)  Lowest    1.17*  1.17*    1.17*  1.17*      (1.13–1.21)  (1.13–1.20)    (1.13–1.21)  (1.13–1.21)  Difficulty in paying bills (ref: never)  From time to time    1.16*  1.16*    1.16*  1.16*      (1.14–1.19)  (1.14–1.19)    (1.14–1.19)  (1.14–1.19)  Most of the time    1.29*  1.29*    1.29*  1.29*      (1.25–1.34)  (1.25–1.33)    (1.25–1.33)  (1.25–1.33)  GDP per capita    0.99      0.99      (0.99–1.00)      (0.99–1.00)  Random  σ2  0.019  0.010  0.010  0.020  0.011  0.010  (SE)  0.005  0.003  0.003  0.006  0.003  0.003  MRR  1.14*  1.10*  1.10*  1.14*  1.11*  1.10*  (95% CI for MRR)  (1.13–1.15)  (1.09–1.11)  (1.09–1.11)  (1.13–1.16)  (1.10–1.11)  (1.09–1.11)    Education expenditure  Quality of education    Model 1  Model 2  Model 3  Model 1  Model 2  Model 3    RR  RR  RR  RR  RR  RR    (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  Fixed  Country-level variables  Education expenditure  0.95*  0.97  0.98          (0.90–0.99)  (0.94–1.01)  (0.94–1.01)        Quality of education        0.95  0.98  1.00          (0.90–1.00)  (0.94–1.03)  (0.95–1.05)  GDP per capita      0.99      0.99        (0.99–1.00)      (0.99–1.00)  Individual-level variables  Age in years  0.99*  0.99*  0.99*  0.99*  0.99*  0.99*    (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  Gender (ref: male)        Female  0.74*  0.74*  0.74*  0.74*  0.74*  0.74*    (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  Marital status (ref: married individuals)        Single  1.14*  1.11*  1.11*  1.14*  1.11*  1.11*    (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  Urbanisation (ref: rural area or village)        Small/middle sized town  1.02  1.04*  1.04*  1.02  1.04*  1.04*    (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  Large town  1.04*  1.07*  1.07*  1.04*  1.07*  1.07*    (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  Education (ref: 20 years or older)  16–19 years    1.15*  1.15*    1.15*  1.15*      (1.12–1.18)  (1.12–1.18)    (1.13 1.18)  (1.12–1.18)  15 years or less    1.23*  1.23*    1.23*  1.23*      (1.19–1.27)  (1.19–1.27)    (1.19–1.27)  (1.19–1.27)  Subjective social status (ref: highest)  Second highest    1.08*  1.08*    1.08*  1.08*      (1.05–1.11)  (1.05–1.11)    (1.05–1.11)  (1.05–1.11)  Second lowest    1.08*  1.07*    1.08*  1.07*      (1.05–1.11)  (1.04–1.11)    (1.05–1.11)  (1.04–1.11)  Lowest    1.17*  1.17*    1.17*  1.17*      (1.13–1.21)  (1.13–1.20)    (1.13–1.21)  (1.13–1.21)  Difficulty in paying bills (ref: never)  From time to time    1.16*  1.16*    1.16*  1.16*      (1.14–1.19)  (1.14–1.19)    (1.14–1.19)  (1.14–1.19)  Most of the time    1.29*  1.29*    1.29*  1.29*      (1.25–1.34)  (1.25–1.33)    (1.25–1.33)  (1.25–1.33)  GDP per capita    0.99      0.99      (0.99–1.00)      (0.99–1.00)  Random  σ2  0.019  0.010  0.010  0.020  0.011  0.010  (SE)  0.005  0.003  0.003  0.006  0.003  0.003  MRR  1.14*  1.10*  1.10*  1.14*  1.11*  1.10*  (95% CI for MRR)  (1.13–1.15)  (1.09–1.11)  (1.09–1.11)  (1.13–1.16)  (1.10–1.11)  (1.09–1.11)  N = 23842; Eurobarometer 72.3, 2009 Model 1: considering individual-level demographic characteristics, Model 2: considering individual-level demographic and socioeconomic characteristics, Model 3: considering individual-level characteristics and country-level variable (GDP per capita). * P < 0.05. In the analysis pertaining to education domain, people in the country with higher education expenditure had less number of health-risk behaviours with the rate ratio of 0.95 (95%CI 0.90–0.99) for 1% increase of public expenditure on health in the model accounting for all demographic factors (table 4). On the other hand, the association between quality of education system and the co-occurrence of health-risk behaviours was not significant in any models. In the fully adjusted models, there were clear socioeconomic gradients in the number of health-risk behaviours for all three socioeconomic indicators. Those with the lowest socioeconomic position had higher number of health-risk behaviours than those with the highest socioeconomic position (tables 3 and 4). Discussion In this study, we examined the associations between co-occurrence of health-risk behaviours and contextual factors related to healthcare and education systems in 27 European countries. The results indicate the presence of a relatively modest and inverse association between better quality of healthcare system, and greater public expenditure on healthcare and education systems and co-occurrence of health-risk behaviours. However, these associations were attenuated and lost statistical significance after accounting for individual-level socioeconomic factors and country-level GDP per capita, thus highlighting a possibility of mediation effect of socioeconomic factors in the associations. It is worth noting that different indicators of individuals’ socioeconomic position reflecting perception of social position, material ability and education were all significantly associated with co-occurrence of health-risk behaviours. The findings also highlighted the importance of other individual level characterises such as gender, living area, and age. The observations about these individual characteristics were in line with the previous studies.38,39 While these findings clearly demonstrate the persistence of socioeconomic inequalities in adopting a number of health-risk behaviours, it also indicates a potential role of these country-level contextual factors in individual-level socioeconomic factors, which are often neglected in research on co-occurrence of behaviours. Expenditures on healthcare and education reflect the countries’ commitments to the population welfare that would enhance a sense of security and equality. These features were found to be associated with adopting health-risk behaviours.1 Healthcare system enables equitable access and effective use of health-promoting resources by eliminating health insecurity, and related anxieties and financial distress.18 Similarly, high-quality education can improve health literacy and acquisition of knowledge related to enhancing health-related behaviours.21 This in turn enhances the population abilities to adopt health-promoting behaviours and avoid health-risk behaviours. The findings of this study to some extent were in line with previous studies which examined the association of co-occurrence of health-risk behaviours with other contextual factors, such as neighbourhood socioeconomic factors, social capital at workplace and area-level socioeconomic deprivation; where worse contextual conditions were related to higher number of health-risk behaviours.13–15 While a previous study found a positive association between health and education spending at country-level and physical activity regardless of socioeconomic status in Europe,33 the study however did not account for GDP and used different definition of physical activity from the one used here. In the current study, adjusting for GDP per capita and individual socioeconomic factors eliminated the significant association between expenditure of health and education and co-occurrence of health-risk behaviours. Although a negative association between contextual factors related to healthcare and education systems and co-occurrence of health-risk behaviours was implied, the association was not strong. A possible explanation would be the known correlation between contextual factors related to education and individual-level socioeconomic factors. In other words, greater public expenditure is highly correlated with education at the individual level, hence the inclusion of education in the analytical model attenuates the association between public expenditure on education and health behaviours. It is also possible that in countries with generous healthcare systems, some people take less responsibility for their own health and engage in health-risk behaviours. Finally, there is very small variation in the contextual factors in the EU countries. This undoubtedly contributed to the relatively weak association observed in this study. The findings on co-occurrence of health-risk behaviours and their association with contextual factors and individual socioeconomic position also emphasise the need to adopt a holistic approach addressing multiple health behaviours rather than a single behaviour approach. There are some limitations. Firstly, this is a cross-sectional study, therefore cannot establish causality. Secondly, because the survey was self-reported, the response might be over-reported or under-reported, especially, health-risk behaviours and socioeconomic position. Thirdly, this study included 27 countries, which might not be a sufficiently large number for assessing variation in multilevel analysis. Furthermore, the terms of joining the European Union include having consistent policies, hence the variations between EU countries are very small. The cut-off points of some of the health-related behaviours did not perfectly match the guidelines or optimal behaviours; however, we used the closest cut-off points to the guidelines. Finally, participants with missing data were excluded throughout the analyses, which might a potential for selection bias, although the excluded number was relatively small. This is the first study, which examined the associations between co-occurrence of health-risk behaviours and healthcare and education systems at the country level using multilevel analysis. While the study highlights the importance of developing high quality healthcare and education systems, generously supported by public fund in relation to co-occurrence of health-risk behaviours, the association might be attenuated by individual socioeconomic factors because of the possible association with country-level contextual factors. Conflicts of interest: None declared. Key points Better quality of healthcare system appeared to be inversely related to co-occurrence of health-risk behaviours. Greater public expenditure on healthcare and education appeared to be inversely related to co-occurrence of health-risk behaviours. Individual socioeconomic positions might attenuate the associations between co-occurrence of health-risk behaviours and healthcare/education systems in Europe. The findings will potentially enable identifying population/countries with greater needs for health promotion policies aiming at reducing multiple health-risk behaviours. References 1 Berkman LF, Kawachi I. Social Epidemiology . New York: Oxford University Press, 2000. 2 Danaei G, Vander Hoorn S, Lopez AD, Group CRAc, et al.   Causes of cancer in the world: comparative risk assessment of nine behavioural and environmental risk factors. Lancet  2005; 366: 1784– 93. Google Scholar CrossRef Search ADS PubMed  3 Paffenbarger RSJr, Hyde RT, Wing AL, et al.   The association of changes in physical-activity level and other lifestyle characteristics with mortality among men. N Engl J Med  1993; 328: 538– 45. Google Scholar CrossRef Search ADS PubMed  4 Hu FB, Manson JE, Stampfer MJ, et al.   Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engl J Med  2001; 345: 790– 7. Google Scholar CrossRef Search ADS PubMed  5 Hanioka T, Ojima M, Tanaka K, et al.   Causal assessment of smoking and tooth loss: a systematic review of observational studies. BMC Public Health  2011; 11: 221. Google Scholar CrossRef Search ADS PubMed  6 Goldstein BY, Chang S-C, Hashibe M, et al.   Alcohol consumption and cancer of the oral cavity and pharynx from 1988 to 2009: an update. Eur J Cancer Prev  2010; 19: 431. Google Scholar CrossRef Search ADS PubMed  7 Moynihan P. The Role of Diet in the Prevention of Dental Diseases . Comprehensive Preventive Dentistry: Wiley-Blackwell, 2012. 99 p. 8 Tada A, Watanabe T, Yokoe H, et al.   Relationship between the number of remaining teeth and physical activity in community-dwelling elderly. Arch Gerontol Geriat  2003; 37: 109– 17. Google Scholar CrossRef Search ADS   9 Nowak MA. Physical activity and its associations with other lifestyle elements in Polish women. J Hum Kinet  2011; 29: 161– 72. Google Scholar CrossRef Search ADS PubMed  10 Glasgow RE, Goldstein MG, Ockene JK, Pronk NP. Translating what we have learned into practice: principles and hypotheses for interventions addressing multiple behaviors in primary care. Am J Prev Med  2004; 27: 88– 101. Google Scholar CrossRef Search ADS PubMed  11 McAloney K, Graham H, Law C, Platt L. A scoping review of statistical approaches to the analysis of multiple health-related behaviours. Prev Med  2013; 56: 365– 71. Google Scholar CrossRef Search ADS PubMed  12 Khaw K-T, Wareham N, Bingham S, et al.   Combined impact of health behaviours and mortality in men and women: the EPIC-Norfolk prospective population study. Obstet Gynecol Surv  2008; 63: 376– 7. Google Scholar CrossRef Search ADS   13 Halonen JI, Kivimäki M, Pentti J, et al.   Quantifying neighbourhood socioeconomic effects in clustering of behaviour-related risk factors: a multilevel analysis. PLoS One  2012; 7: e32937. Google Scholar CrossRef Search ADS PubMed  14 Lakshman R, McConville A, How S, et al.   Association between area-level socioeconomic deprivation and a cluster of behavioural risk factors: cross-sectional, population-based study. J Public Health  2011; 33: 234– 45. Google Scholar CrossRef Search ADS   15 Väänänen A, Kouvonen A, Kivimäki M, et al.   Workplace social capital and co-occurrence of lifestyle risk factors: the Finnish Public Sector Study. Occup Environ Med  2009; 66: 432– 7. Google Scholar CrossRef Search ADS PubMed  16 World Health Organisation. Quality of care: a process for making strategic choices in health systems 2006. Available at http://www.who.int/management/quality/assurance/QualityCare_B.Def.pdf. 17 Heymann DL, Prentice T, Reinders LT. The World Health Report 2007: a Safer Future: Global Public Health Security in the 21st Century: World Health Organisation, 2007. 18 Ensor T, Cooper S. Overcoming barriers to health service access: influencing the demand side. Health Policy Plann  2004; 19: 69– 79. Google Scholar CrossRef Search ADS   19 Park CL, Iacocca MO. A stress and coping perspective on health behaviors: theoretical and methodological considerations. Anxiety Stress Coping  2014; 27: 123– 37. Google Scholar CrossRef Search ADS PubMed  20 World Health Organisation Commission on Social Determinants of Health. Closing the gap in a generation. Health equity through action on the social determinants of health: World Health Organization; 2008. Available at http://whqlibdoc.who.int/publications/2008/9789241563703_eng.pdf. 21 Johnson W, Kyvik KO, Mortensen EL, et al.   Education reduces the effects of genetic susceptibilities to poor physical health. Int J Epidemiol  2010; 39: 406– 14. Google Scholar CrossRef Search ADS PubMed  22 European Commission. ICPSR Codebook: Inter-university Consortium for Political and Social Research (ICPSR); 2013. Available at http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/32441. 23 EU Working Group. EU Physical Activity Guidelines Recommended Policy Actions in Support of Health-Enhancing Physical Activity. Brussels; 2008. 24 Filippidis FT, Agaku IT, Vardavas CI. Geographic variation and socio-demographic determinants of the co-occurrence of risky health behaviours in 27 European Union member states. J Public Health  2016; 38: e13– 20. Google Scholar CrossRef Search ADS   25 National Institute for Health and Care Excellence. Dental checks: intervals between oral health reviews 2004. Available at https://www.nice.org.uk/guidance/cg19/chapter/1-guidance. 26 The World Bank. 2014. Available at http://data.worldbank.org/. 27 Health Cosumer Powerhouse. Euro Health Consumer Index 2009. Available at http://www.healthpowerhouse.com/files/Report-EHCI-2009-090925-final-with-cover.pdf. 28 Eurostat. Your key to European statistics; Public expenditure on education 2016. Available at http://ec.europa.eu/eurostat/tgm/table.do?tab=table&init=1&language=en&pcode=tsdsc510&plugin=1. 29 UNESCO. Education for All Development Index 2009. Available at http://en.unesco.org/gem-report/sites/gem-report/files/efagmr2009_Annex1_EDI.pdf. 30 West A, Stokes E, Edge A. Secondary education across Europe: Curricula and school examination systems: Centre for Educational Research; 1999. Available at http://www.leeds.ac.uk/educol/documents/00001195.htm. 31 Liberatos P, Link BG, Kelsey JL. The measurement of social class in epidemiology. Epidemiol Rev  1988; 10: 87– 121. Google Scholar CrossRef Search ADS PubMed  32 Adler NE, Epel ES, Castellazzo G, Ickovics JR. Relationship of subjective and objective social status with psychological and physiological functioning: preliminary data in healthy, White women. Health Psychol  2000; 19: 586. Google Scholar CrossRef Search ADS PubMed  33 Lera-López F, Wicker P, Downward P. Does government spending help to promote healthy behavior in the population? Evidence from 27 European countries. J Public Health  2015; 38: e5– e12. Google Scholar CrossRef Search ADS   34 Cameron AC, Trivedi PK. Regression Analysis of Count Data . Cambridge, MA: Cambridge University Press; 2013. Google Scholar CrossRef Search ADS   35 Snijders TAB, Bosker RJ. Multilevel Analysis: an Introduction to Basic and Advanced Multilevel Modeling , 2nd edn London: Sage Publishers, 2012. 36 Carle AC. Fitting multilevel models in complex survey data with design weights: Recommendations. BMC Med Res Methodol  2009; 9: 1. Google Scholar CrossRef Search ADS PubMed  37 Assareh H, Achat HM, Stubbs JM, et al.   Incidence and variation of discrepancies in recording chronic conditions in Australian hospital administrative data. PLoS one  2016; 11: e0147087. Google Scholar CrossRef Search ADS PubMed  38 Noble N, Paul C, Turon H, Oldmeadow C. Which modifiable health risk behaviours are related? A systematic review of the clustering of Smoking, Nutrition, Alcohol and Physical activity ('SNAP') health risk factors. Prev Med  2015; 81: 16– 41. Google Scholar CrossRef Search ADS PubMed  39 Meader N, King K, Moe-Byrne T, et al.   A systematic review on the clustering and co-occurrence of multiple risk behaviours. BMC Public Health  2016; 16: 657. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png The European Journal of Public Health Oxford University Press

The role of healthcare and education systems in co-occurrence of health risk behaviours in 27 European countries

Loading next page...
 
/lp/ou_press/the-role-of-healthcare-and-education-systems-in-co-occurrence-of-NtVH20wAN6
Publisher
Oxford University Press
Copyright
© The Author 2017. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.
ISSN
1101-1262
eISSN
1464-360X
D.O.I.
10.1093/eurpub/ckx071
Publisher site
See Article on Publisher Site

Abstract

Abstract Background Contextual factors play an important role in health and related behaviours. This study aims to examine the association of co-occurrence of five health-risk behaviours with healthcare and education contextual factors using nationally representative samples from 27 European countries. Methods Data were from Eurobarometer 72.3, 2009. The outcome was a count variable ranging from 0 to 5 indicating co-occurrence of five health-risk behaviours, namely smoking, excessive alcohol consumption, non-frequent fresh fruit consumption, physical inactivity and non-dental check-ups. Public expenditures on healthcare and education as a percentage of GDP and quality of healthcare and education at a country-level were used as contextual factors. A set of multilevel Poisson regression models were conducted to examine the associations between co-occurrence of health-risk behaviours and each of the contextual factors considering age, gender, marital status, urbanisation, individual socioeconomic positions (education, subjective social status or difficulty in paying bills) and GDP per capita. Results The total population was 23 842. Greater expenditures on healthcare and education, and better quality of healthcare systems had negative associations with co-occurrence of health-risk behaviours in the model adjusted for all individual demographic indicators. However, statistical significance disappeared after adjusting for socioeconomic indicators and GDP per capita. Conclusion While the study highlights the importance of developing high-quality healthcare and education systems generously supported by public fund in relation to co-occurrence of health-risk behaviours, the influence of contextual factors in adopting health-related behaviours is probably attenuated by individual socioeconomic factors. Introduction Modifiable health-related behaviours tend to be grouped together in the same sectors of the population, with those at the bottom of the socioeconomic hierarchy more likely to engage in health-risk behaviours and less likely to adopted health-promoting ones.1 Among these behaviours, the ‘big four’ behaviours, namely smoking, excessive alcohol consumption, poor diet and physical inactivity have excessively been studied in terms of their devastating impact on mortality and morbidity, particularly cancer, cardiovascular disease and diabetes.2–4 The same set of health-related behaviours is also related to oral health5–8 in addition to dental check-ups.9 Traditionally, epidemiological studies have focused on the significant impact of a single health-risk behaviour on a specific health outcome.10 However, co-occurrence of health-risk behaviours in the same individuals, defined as a count number of health-risk behaviours,11 exerts multiplicative effects on disease risk.12 The social environment influences behaviour by shaping norms, enforcing patterns of health-related behaviours, providing or obstructing environmental opportunities to engage in certain behaviours.1 Earlier studies have reported an association between co-occurrence of health-risk behaviours and contextual factors, such as neighbourhood socioeconomic factors,13 area-level deprivation14 and workplace social capital.15 Although the evidence has shown the influence of contextual factors on co-occurrence of health-risk behaviours, to the best of our knowledge, no other study has examined the association between healthcare/education systems and co-occurrence of health-risk behaviours. Better healthcare systems could promote equitable access to preventive services,16 and enhance perception of health security,17 and subsequently reduce anxiety related to availability and quality of healthcare services at the time of sickness.18 Such anxiety could impact on individual’s psychosocial wellbeing which has an established association with adopting health-risk behaviours.19 Similarly, equitable access to high-quality education system will enable equal opportunities to access high quality education, hence increasing the probabilities of having better jobs, and improving their chances for social mobility, and consequently reducing the social gradients in behaviours and health.20 Furthermore, it would possibly impact on individual’s ability to access and interpret information, improve sense of learnt effectiveness and personal control, particularly among the least affluent, hence leading to healthier lifestyles.21 Thus, it is reasonable to hypothesize that healthcare and education systems can potentially influence co-occurrence of health-risk behaviours. Methods Data source and study sample This study is a secondary analysis of cross-country population-based data from Eurobarometer 72.3, which is a cross-sectional interview survey in Europe. This survey used nationally representative samples from each of the European countries and was conducted in October 2009.22 A 2-stage, random (probability) sampling design was used for sample selection. The data were collected by face-to-face interview in people’s home from 2 to 19 October 2009 by the TNS Opinion and Social through its network of national institutes in the respective national language.22 The regular sample size was 1000 participants from each country. Intentionally more participants were selected from the respective group than would typically be done if everyone in the sample had an equal chance of being selected.25 We included participants from 27 European countries who answered all questions pertaining to health-related behaviours and demographic/socioeconomic indicators. Given that smoking and drinking are illegal for those under 18 years in most of the European countries, the analysis was limited to those aged 18 and over. Ethical approval was not required for this secondary data analysis. Variables Outcome The survey had the questions related to check-up and medical screening, oral health, alcohol habits, smoking habits and sport and physical activity. We selected the five behaviours, namely smoking, alcohol use, diet, physical activity and dental check-ups because of their relationships with a number of chronic diseases and with oral health. In addition, these five health-related behaviours had similar socio-demographic determinants. The co-occurrence of health-risk behaviours was estimated based on how many of the following five health-risk behaviours participants reported: smoking, excessive alcohol consumption, infrequent consumption of fresh fruits, physical inactivity and not attending dental check-ups. The behaviours were dichotomised into binary options (health-risk behaviours were coded as 1, while health-promoting behaviours were coded as 0) (table 1). Smoking was indicated by current smokers (vs former- or never-smokers). Excessive alcohol consumption was defined as having five or more drinks on one occasion at least once a week in the last 12 months. Infrequent consumption of fresh fruits was indicated by consuming fresh fruits “from time to time”, “rarely” or “never”. Physical inactivity was created by combining two questions. EU Working Group indicated that physical activity at least five times a week is recommended for European adults.23 As the dataset did not allow using this definition, in this study at least four times a week was used as a cut-off point for physical activity.24 Dental check-ups were indicated by use in the past 12 months. According to the NICE guideline, people who are not at risk of or from oral disease may not need dental check-ups once a year.25 However, given this is a sample of the whole population including those with at higher risk of oral diseases, we thought dental visit during the past 12 months would be a reasonable cut-off point. Table 1 Measurement of health-risk behaviours   Measurement in the survey  Classification of health-risk behaviours for this study  Smoking  Regarding smoking cigarettes, cigars or a pipe, which of the following applies to you?: (1) you smoke at the present time, (2) you used to smoke but you have stopped, (3) you have never smoked and (4) don’t know  Smoking at the present time  Excessive alcohol consumption  How often in the past twelve months have you had five or more drinks on one occasion?: (1) several times a week, (2) once a week, (3) once a month, (4) less than once a month, (5) never, and (6) don’t know/refusal  Having five or more drinks on one occasion at least once a week in the last twelve months  Infrequent consumption of fresh fruits  How often do you eat fresh fruit?: (1) often, (2) from time to time, (3) rarely, (4) never and (5) don’t know  Consuming fresh fruit from time to time, rarely or never  Physical inactivity  There are two questions: (a) how often the person exercises or plays sport (b) how often the participant engages in a physical activity outside sport such as cycling or walking from a place to another, dancing, gardening etc.  Doing physical activity less than four times a week (Except the participates who answered: (1) five times a week or more for either the two questions regardless of any other answer; those who answered (2) three to four times a week for either question and (2) three to four times a week or (3) one to two times a week for another question)  For each of the questions; (1) five times a week or more, (2) three to four times a week, (3) one to two times a week, (4) one to three times a month, (5) less often, (6) never, and (7) don’t know  Not attending dental check-ups  Did you have dental check-ups in the last twelve months, whether or not as part of any treatment?: (1) yes, own initiative, (2) yes, doctor’s initiative, (3) yes, screening programme, (4) no, and (5) don’t know  Having dental check-ups in the last 12 months regardless own initiative, doctor's initiative or screening programme    Measurement in the survey  Classification of health-risk behaviours for this study  Smoking  Regarding smoking cigarettes, cigars or a pipe, which of the following applies to you?: (1) you smoke at the present time, (2) you used to smoke but you have stopped, (3) you have never smoked and (4) don’t know  Smoking at the present time  Excessive alcohol consumption  How often in the past twelve months have you had five or more drinks on one occasion?: (1) several times a week, (2) once a week, (3) once a month, (4) less than once a month, (5) never, and (6) don’t know/refusal  Having five or more drinks on one occasion at least once a week in the last twelve months  Infrequent consumption of fresh fruits  How often do you eat fresh fruit?: (1) often, (2) from time to time, (3) rarely, (4) never and (5) don’t know  Consuming fresh fruit from time to time, rarely or never  Physical inactivity  There are two questions: (a) how often the person exercises or plays sport (b) how often the participant engages in a physical activity outside sport such as cycling or walking from a place to another, dancing, gardening etc.  Doing physical activity less than four times a week (Except the participates who answered: (1) five times a week or more for either the two questions regardless of any other answer; those who answered (2) three to four times a week for either question and (2) three to four times a week or (3) one to two times a week for another question)  For each of the questions; (1) five times a week or more, (2) three to four times a week, (3) one to two times a week, (4) one to three times a month, (5) less often, (6) never, and (7) don’t know  Not attending dental check-ups  Did you have dental check-ups in the last twelve months, whether or not as part of any treatment?: (1) yes, own initiative, (2) yes, doctor’s initiative, (3) yes, screening programme, (4) no, and (5) don’t know  Having dental check-ups in the last 12 months regardless own initiative, doctor's initiative or screening programme  Country-level independent variables Public expenditure on health and education and quality of healthcare and education systems were used as the contextual exposures of interest. Five-year average public expenditure on health as a percentage of Gross Domestic Product (GDP) in each country from 2005 to 2009 reported by the World Bank26 was used as health expenditure. Quality of healthcare system was indicated by Euro Health Consumer Index (EHCI).27 EHCI consisted of 38 indicators of 6 sub-disciplines: (1) Patient rights and information, (2) e-Health, (3) Waiting time for treatment, (4) Outcomes, (5) Range and reach of services and (6) Pharmaceuticals. For each of the sub-domains, the country score was calculated as a percentage of the full score.27 The scores were divided by 100 for better interpretations of the results and used as an indicator of the quality of healthcare system. Five-year average public expenditure on education as a percentage of GDP in each country from 2005 to 2009 by Eurostat28 was used as education expenditure. Quality of education system was indicated by the Education for All Development Index (EDI) in 2006.29 EDI is used as a proxy measure for each of the four “Education for All” goals, and each of these EDI components is assigned equal weight in the overall index in accordance with the principle of considering each goal as being of equal importance. The EDI value for a given country is the arithmetic mean of the four proxy indicators, namely universal primary education (the percentage of primary-school-age children who are enrolled in either primary or secondary school), adult literacy (the adult literacy rate for those aged 15 and above), quality of education (the survival rate to Grade 5) and gender (a simple average of the three gender parity indexes for primary education, secondary education and adult literacy, with each being weighted equally). The percentage of the score was used in this study. Given that expenditure on healthcare and education can be susceptible to economic condition of the country, we included GDP per capita in the analysis. Individual-level independent variables Socioeconomic position was indicated by three measures, namely education, subjective social status and difficulty in paying bills. Education was measured by the age when participants stopped full-time education, and was categorised into three groups; 20 years and older, 16–19 years old and 15 years or less. This is because in most European countries the minimum age for compulsory education is 15 years or older, and secondary school education is usually completed when individuals are under 20 years.30 Participants still studying were included in the category corresponding to their age. Education is an important indicator of socioeconomic position as it is comparable across countries.31 Subjective social status reflects one’s perception of own status in the respective community/country, and is an important determinant for health-related behaviours.32 In this survey, participants were asked to place themselves on a ladder indicating their perception of own positions in their respective society on a scale of 1–10, hence reflecting perception of social standing. For better interpretation of the data and to distinguish between the upper and the lower halves of the scale, subjective social status was categorised into quartiles. The quartiles indicted: highest (7–10), second highest (6), second lowest (5) and lowest (1–4). Difficulty in paying bills reflects financial ability to pay bills at the end of the month during the last 12 months, and has three categories; most of the time, from time to time and almost never/never, which was used in previous studies.33 Demographic factors included gender, age, urbanization and marital status. Age was used as a continuous variable. Urbanization has three categories (rural area or village, small or middle sized town and large town). Marital status was dichotomized to indicate; married/living with a partner, vs single/divorced/separated/widowed. The selection of the variables included in the models was based on their relationships with the outcome variable and with the main exposure. For example, expenditures on health and education are relevant to the country GDP; therefore, inclusion of such a variable could affect the relationship between the exposure and the outcome variable. Statistical analysis In order to examine the associations between co-occurrence of health-risk behaviours and each of contextual factors, a two-level Poisson regression model with log link function34 was conducted because the response variable was a count and the mean (1.70) was almost the same as variance (1.41) for this variable (range 0–5). The fixed- and random-parameter estimates for the two-level Poisson regression models were calculated by MLwiN version 2.35 using the iterative generalized least squares algorithm with the penalized quasi-likelihood, second-order approximation procedure.35 We did not use weight for the multilevel analyses because when the cluster sizes are large (∼750 individuals in each cluster), the standard errors achieved remarkable consistency either weighted or unweighted.36 Three multilevel models were constructed for each of contextual factors. The construction of the models aimed at examining binary association between contextual factors and behaviours, then adjusted for demographic factors and finally, included socioeconomic factors at the country- and individual-levels. Model 1 examined the associations between co-occurrence of health-risk behaviours and each of contextual factors adjusted for all demographic characteristics (age, gender, marital status and urbanization). Model 2 was adjusted for all demographic characteristics and the three socioeconomic indicators. Finally, Model 3 was additionally adjusted for GDP per capita. In each model, rate ratios were reported for fixed parts. For random parts, the proportional change in inter-country variance estimates (σ2) of the different model was imputed. A median rate ratio, which is the median of the rate ratios of pair-wise comparisons of countries, was calculated as e^(0.954*√σ2).37 Results After excluding the participants with missing values, 23 842 individuals were included in this study (the valid percentage: 91.7%). The weighted percentages of each health-risk behaviour among the total population were 30.0% for smoking, 22.4% for excessive alcohol consumption, 36.9% for infrequent fruit consumption, 50.7% for physical inactivity and 36.8% for non-use of dental check-ups. Participants reported no health-risk behaviours were 14.9% (n = 3897). The most frequent co-occurrence numbers were one and two health-risk behaviours with 29.3% (n = 7311) and 29.8% (n = 6835) respectively. Those with three health-risk behaviours accounted for 17.9% (n = 4025) with 6.5% (n = 1440) for four health-risk behaviours, 1.7% (n = 334) for five health-risk behaviours and 16.4% (n = 3897) for no risk behaviour. The mean number of health-risk behaviours was 1.70 (SD = 1.19, Variance = 1.41, Median = 2.5). The individual- and country-level predictor variables are presented in table 2. The results of multilevel Poisson regression models examining the associations between co-occurrence of health-risk behaviours and contextual factors are presented in tables 3 and 4. In an empty model, variance at country-level was 0.023 (SE: 0.006), hence the median of all pairwise comparisons between countries gives a rate ratio of 1.16 (95%CI 1.14–1.17). This indicated that there were variations in co-occurrence of health-risk behaviours between countries. A unit increase in health expenditure was associated with lower risk of co-occurrence of risk behaviours with the rate ratio of 0.96 (95%CI 0.93–0.99) in the model accounting for all demographic factors (table 3). After accounting for socioeconomic positions, the association lost significance. Individuals in the countries with higher quality of healthcare system were less likely to have health-risk behaviours with the rate ratio of 0.93 (95%CI 0.88–0.97) in the model accounting for all demographic factors, again statistical significance disappeared after adjusting for socioeconomic position and GDP per capita. Table 2 Data description for analytical multilevel models Level 1: Individual-level predictor variable (N = 23 842)    N  Weighted %  (95%CI)  Mean±SE  Gender    Male  10 453  48.00  (47.00–49.01)    Female  13 389  52.00  (50.99–53.00)  Age in years  23 842  47.71±0.18 (range 18–96)  Marital status    Married/with a partner  15 608  65.77  (64.82–66.71)    Single  8234  34.23  (33.29–35.18)  Urbanisation    Rural area/village  8624  34.79  (33.84–35.75)    Small/middle-sized town  8360  39.94  (38.95–40.93)    Large town  6858  25.28  (24.44–26.13)  Education    20 years or older  7574  29.48  (28.58–30.39)    16-19 years old  11 103  46.48  (45.48–47.48)    15 years old or less  5165  24.05  (23.20–24.91)  Subjective social status    Highest  6889  27.21  (26.34–28.10)    Second highest  5091  24.28  (23.42–25.17)    Second lowest  6934  29.94  (29.03–30.88)    Lowest  4928  18.56  (17.81–19.34)  Difficulty in paying bills    Almost never/never  15 292  66.16  (65.22–67.09)    From time to time  6351  26.16  (25.30–27.05)    Most of the time  2199  7.67  (7.18–8.19)  Level 2: country-level predictor variable (N = 27)  Mean  Standard Deviation  Variance  Range  Healthcare expenditure (% of GDP)  6.32  1.50  2.26  2.8–8.7  Quality of healthcare  6.65  1.06  1.12  4.48–8.75  Education expenditure (% of GDP)  5.25  1.01  1.02  6.45–8.10  Quality of education  98.15  0.95  0.91  95.5–99.4  GDP per capita (in US dollars/1000)  31.93  17.98  323.43  6.74–100.74  Level 1: Individual-level predictor variable (N = 23 842)    N  Weighted %  (95%CI)  Mean±SE  Gender    Male  10 453  48.00  (47.00–49.01)    Female  13 389  52.00  (50.99–53.00)  Age in years  23 842  47.71±0.18 (range 18–96)  Marital status    Married/with a partner  15 608  65.77  (64.82–66.71)    Single  8234  34.23  (33.29–35.18)  Urbanisation    Rural area/village  8624  34.79  (33.84–35.75)    Small/middle-sized town  8360  39.94  (38.95–40.93)    Large town  6858  25.28  (24.44–26.13)  Education    20 years or older  7574  29.48  (28.58–30.39)    16-19 years old  11 103  46.48  (45.48–47.48)    15 years old or less  5165  24.05  (23.20–24.91)  Subjective social status    Highest  6889  27.21  (26.34–28.10)    Second highest  5091  24.28  (23.42–25.17)    Second lowest  6934  29.94  (29.03–30.88)    Lowest  4928  18.56  (17.81–19.34)  Difficulty in paying bills    Almost never/never  15 292  66.16  (65.22–67.09)    From time to time  6351  26.16  (25.30–27.05)    Most of the time  2199  7.67  (7.18–8.19)  Level 2: country-level predictor variable (N = 27)  Mean  Standard Deviation  Variance  Range  Healthcare expenditure (% of GDP)  6.32  1.50  2.26  2.8–8.7  Quality of healthcare  6.65  1.06  1.12  4.48–8.75  Education expenditure (% of GDP)  5.25  1.01  1.02  6.45–8.10  Quality of education  98.15  0.95  0.91  95.5–99.4  GDP per capita (in US dollars/1000)  31.93  17.98  323.43  6.74–100.74  Table 3 Rate ratios for the associations between healthcare domain and co-occurrence of health-risk behaviours from multilevel Poisson regression models   Health expenditure  Quality of healthcare    Model 1  Model 2  Model 3  Model 1  Model 2  Model 3    RR  RR  RR  RR  RR  RR    (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  Fixed  Country-level variables  Health expenditure  0.96*  0.98  0.99          (0.93–0.99)  (0.96–1.01)  (0.96–1.03)        Quality of healthcare    0.93*  0.97  0.98          (0.88–0.97)  (0.93–1.01)  (0.92–1.04)  GDP per capita  0.99      0.99        (0.99–1.00)      (0.99–1.00)  Individual-level variables  Age in years  0.99*  0.99*  0.99*  0.99*  0.99*  0.99*    (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  Gender (ref: male)  Female  0.74*  0.74*  0.74*  0.74*  0.74*  0.74*    (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  Marital status (ref: married individuals)  Single  1.14*  1.11*  1.11*  1.14*  1.11*  1.11*    (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  Urbanisation (ref: rural area or village)  Small/middle sized town  1.02  1.04*  1.04*  1.02  1.04*  1.04*    (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  Large town  1.04*  1.07*  1.07*  1.04*  1.07*  1.07*    (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  Education (ref: 20 years or older)  16–19 years    1.15*  1.15*    1.15*  1.15*      (1.13–1.18)  (1.12–1.18)    (1.12–1.18)  (1.12–1.18)  15 years or less    1.23*  1.23*    1.23*  1.23*      (1.19–1.27)  (1.19–1.27)    (1.19–1.27)  (1.19–1.27)  Subjective social status (ref: highest)  Second highest    1.08*  1.08*    1.08*  1.08*      (1.05–1.11)  (1.05–1.11)    (1.05–1.11)  (1.05–1.11)  Second lowest    1.07*  1.07*    1.07*  1.07*      (1.04–1.11)  (1.04–1.11)    (1.04–1.11)  (1.04–1.11)  Lowest    1.17*  1.17*    1.17*  1.17*      (1.13–1.21)  (1.13–1.20)    (1.13–1.21)  (1.13–1.20)  Difficulty in paying bills (ref: never)  From time to time    1.16*  1.16*    1.16*  1.16*      (1.14–1.19)  (1.14–1.19)    (1.14–1.19)  (1.14–1.19)  Most of the time    1.29*  1.29*    1.29*  1.29*      (1.25–1.33)  (1.25–1.33)    (1.25–1.33)  (1.25–1.33)  Random  σ2  0.018  0.011  0.010  0.016  0.010  0.010  (SE)  0.005  0.003  0.003  0.004  0.003  0.003  MRR  1.14*  1.11*  1.10*  1.13*  1.10*  1.10*  (95% CI for MRR)  (1.13–1.15)  (1.10–1.11)  (1.09–1.11)  (1.12–1.14)  (1.09–1.11)  (1.09–1.11)    Health expenditure  Quality of healthcare    Model 1  Model 2  Model 3  Model 1  Model 2  Model 3    RR  RR  RR  RR  RR  RR    (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  Fixed  Country-level variables  Health expenditure  0.96*  0.98  0.99          (0.93–0.99)  (0.96–1.01)  (0.96–1.03)        Quality of healthcare    0.93*  0.97  0.98          (0.88–0.97)  (0.93–1.01)  (0.92–1.04)  GDP per capita  0.99      0.99        (0.99–1.00)      (0.99–1.00)  Individual-level variables  Age in years  0.99*  0.99*  0.99*  0.99*  0.99*  0.99*    (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  Gender (ref: male)  Female  0.74*  0.74*  0.74*  0.74*  0.74*  0.74*    (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  Marital status (ref: married individuals)  Single  1.14*  1.11*  1.11*  1.14*  1.11*  1.11*    (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  Urbanisation (ref: rural area or village)  Small/middle sized town  1.02  1.04*  1.04*  1.02  1.04*  1.04*    (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  Large town  1.04*  1.07*  1.07*  1.04*  1.07*  1.07*    (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  Education (ref: 20 years or older)  16–19 years    1.15*  1.15*    1.15*  1.15*      (1.13–1.18)  (1.12–1.18)    (1.12–1.18)  (1.12–1.18)  15 years or less    1.23*  1.23*    1.23*  1.23*      (1.19–1.27)  (1.19–1.27)    (1.19–1.27)  (1.19–1.27)  Subjective social status (ref: highest)  Second highest    1.08*  1.08*    1.08*  1.08*      (1.05–1.11)  (1.05–1.11)    (1.05–1.11)  (1.05–1.11)  Second lowest    1.07*  1.07*    1.07*  1.07*      (1.04–1.11)  (1.04–1.11)    (1.04–1.11)  (1.04–1.11)  Lowest    1.17*  1.17*    1.17*  1.17*      (1.13–1.21)  (1.13–1.20)    (1.13–1.21)  (1.13–1.20)  Difficulty in paying bills (ref: never)  From time to time    1.16*  1.16*    1.16*  1.16*      (1.14–1.19)  (1.14–1.19)    (1.14–1.19)  (1.14–1.19)  Most of the time    1.29*  1.29*    1.29*  1.29*      (1.25–1.33)  (1.25–1.33)    (1.25–1.33)  (1.25–1.33)  Random  σ2  0.018  0.011  0.010  0.016  0.010  0.010  (SE)  0.005  0.003  0.003  0.004  0.003  0.003  MRR  1.14*  1.11*  1.10*  1.13*  1.10*  1.10*  (95% CI for MRR)  (1.13–1.15)  (1.10–1.11)  (1.09–1.11)  (1.12–1.14)  (1.09–1.11)  (1.09–1.11)  N = 23842; Eurobarometer 72.3, 2009 Model 1: considering individual-level demographic characteristics, Model 2: considering individual-level demographic and socioeconomic characteristics, Model 3: considering individual-level characteristics and country-level variable (GDP per capita). * P < 0.05. Table 4 Rate ratios for the associations between education domain and co-occurrence of health-risk behaviours from multilevel Poisson regression models   Education expenditure  Quality of education    Model 1  Model 2  Model 3  Model 1  Model 2  Model 3    RR  RR  RR  RR  RR  RR    (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  Fixed  Country-level variables  Education expenditure  0.95*  0.97  0.98          (0.90–0.99)  (0.94–1.01)  (0.94–1.01)        Quality of education        0.95  0.98  1.00          (0.90–1.00)  (0.94–1.03)  (0.95–1.05)  GDP per capita      0.99      0.99        (0.99–1.00)      (0.99–1.00)  Individual-level variables  Age in years  0.99*  0.99*  0.99*  0.99*  0.99*  0.99*    (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  Gender (ref: male)        Female  0.74*  0.74*  0.74*  0.74*  0.74*  0.74*    (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  Marital status (ref: married individuals)        Single  1.14*  1.11*  1.11*  1.14*  1.11*  1.11*    (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  Urbanisation (ref: rural area or village)        Small/middle sized town  1.02  1.04*  1.04*  1.02  1.04*  1.04*    (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  Large town  1.04*  1.07*  1.07*  1.04*  1.07*  1.07*    (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  Education (ref: 20 years or older)  16–19 years    1.15*  1.15*    1.15*  1.15*      (1.12–1.18)  (1.12–1.18)    (1.13 1.18)  (1.12–1.18)  15 years or less    1.23*  1.23*    1.23*  1.23*      (1.19–1.27)  (1.19–1.27)    (1.19–1.27)  (1.19–1.27)  Subjective social status (ref: highest)  Second highest    1.08*  1.08*    1.08*  1.08*      (1.05–1.11)  (1.05–1.11)    (1.05–1.11)  (1.05–1.11)  Second lowest    1.08*  1.07*    1.08*  1.07*      (1.05–1.11)  (1.04–1.11)    (1.05–1.11)  (1.04–1.11)  Lowest    1.17*  1.17*    1.17*  1.17*      (1.13–1.21)  (1.13–1.20)    (1.13–1.21)  (1.13–1.21)  Difficulty in paying bills (ref: never)  From time to time    1.16*  1.16*    1.16*  1.16*      (1.14–1.19)  (1.14–1.19)    (1.14–1.19)  (1.14–1.19)  Most of the time    1.29*  1.29*    1.29*  1.29*      (1.25–1.34)  (1.25–1.33)    (1.25–1.33)  (1.25–1.33)  GDP per capita    0.99      0.99      (0.99–1.00)      (0.99–1.00)  Random  σ2  0.019  0.010  0.010  0.020  0.011  0.010  (SE)  0.005  0.003  0.003  0.006  0.003  0.003  MRR  1.14*  1.10*  1.10*  1.14*  1.11*  1.10*  (95% CI for MRR)  (1.13–1.15)  (1.09–1.11)  (1.09–1.11)  (1.13–1.16)  (1.10–1.11)  (1.09–1.11)    Education expenditure  Quality of education    Model 1  Model 2  Model 3  Model 1  Model 2  Model 3    RR  RR  RR  RR  RR  RR    (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  (95%CI)  Fixed  Country-level variables  Education expenditure  0.95*  0.97  0.98          (0.90–0.99)  (0.94–1.01)  (0.94–1.01)        Quality of education        0.95  0.98  1.00          (0.90–1.00)  (0.94–1.03)  (0.95–1.05)  GDP per capita      0.99      0.99        (0.99–1.00)      (0.99–1.00)  Individual-level variables  Age in years  0.99*  0.99*  0.99*  0.99*  0.99*  0.99*    (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  (0.99–0.99)  Gender (ref: male)        Female  0.74*  0.74*  0.74*  0.74*  0.74*  0.74*    (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  (0.73–0.76)  (0.72–0.75)  (0.72–0.75)  Marital status (ref: married individuals)        Single  1.14*  1.11*  1.11*  1.14*  1.11*  1.11*    (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  (1.11–1.16)  (1.09–1.13)  (1.09–1.13)  Urbanisation (ref: rural area or village)        Small/middle sized town  1.02  1.04*  1.04*  1.02  1.04*  1.04*    (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  (0.99–1.05)  (1.01–1.06)  (1.01–1.06)  Large town  1.04*  1.07*  1.07*  1.04*  1.07*  1.07*    (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  (1.01–1.07)  (1.04–1.09)  (1.04–1.09)  Education (ref: 20 years or older)  16–19 years    1.15*  1.15*    1.15*  1.15*      (1.12–1.18)  (1.12–1.18)    (1.13 1.18)  (1.12–1.18)  15 years or less    1.23*  1.23*    1.23*  1.23*      (1.19–1.27)  (1.19–1.27)    (1.19–1.27)  (1.19–1.27)  Subjective social status (ref: highest)  Second highest    1.08*  1.08*    1.08*  1.08*      (1.05–1.11)  (1.05–1.11)    (1.05–1.11)  (1.05–1.11)  Second lowest    1.08*  1.07*    1.08*  1.07*      (1.05–1.11)  (1.04–1.11)    (1.05–1.11)  (1.04–1.11)  Lowest    1.17*  1.17*    1.17*  1.17*      (1.13–1.21)  (1.13–1.20)    (1.13–1.21)  (1.13–1.21)  Difficulty in paying bills (ref: never)  From time to time    1.16*  1.16*    1.16*  1.16*      (1.14–1.19)  (1.14–1.19)    (1.14–1.19)  (1.14–1.19)  Most of the time    1.29*  1.29*    1.29*  1.29*      (1.25–1.34)  (1.25–1.33)    (1.25–1.33)  (1.25–1.33)  GDP per capita    0.99      0.99      (0.99–1.00)      (0.99–1.00)  Random  σ2  0.019  0.010  0.010  0.020  0.011  0.010  (SE)  0.005  0.003  0.003  0.006  0.003  0.003  MRR  1.14*  1.10*  1.10*  1.14*  1.11*  1.10*  (95% CI for MRR)  (1.13–1.15)  (1.09–1.11)  (1.09–1.11)  (1.13–1.16)  (1.10–1.11)  (1.09–1.11)  N = 23842; Eurobarometer 72.3, 2009 Model 1: considering individual-level demographic characteristics, Model 2: considering individual-level demographic and socioeconomic characteristics, Model 3: considering individual-level characteristics and country-level variable (GDP per capita). * P < 0.05. In the analysis pertaining to education domain, people in the country with higher education expenditure had less number of health-risk behaviours with the rate ratio of 0.95 (95%CI 0.90–0.99) for 1% increase of public expenditure on health in the model accounting for all demographic factors (table 4). On the other hand, the association between quality of education system and the co-occurrence of health-risk behaviours was not significant in any models. In the fully adjusted models, there were clear socioeconomic gradients in the number of health-risk behaviours for all three socioeconomic indicators. Those with the lowest socioeconomic position had higher number of health-risk behaviours than those with the highest socioeconomic position (tables 3 and 4). Discussion In this study, we examined the associations between co-occurrence of health-risk behaviours and contextual factors related to healthcare and education systems in 27 European countries. The results indicate the presence of a relatively modest and inverse association between better quality of healthcare system, and greater public expenditure on healthcare and education systems and co-occurrence of health-risk behaviours. However, these associations were attenuated and lost statistical significance after accounting for individual-level socioeconomic factors and country-level GDP per capita, thus highlighting a possibility of mediation effect of socioeconomic factors in the associations. It is worth noting that different indicators of individuals’ socioeconomic position reflecting perception of social position, material ability and education were all significantly associated with co-occurrence of health-risk behaviours. The findings also highlighted the importance of other individual level characterises such as gender, living area, and age. The observations about these individual characteristics were in line with the previous studies.38,39 While these findings clearly demonstrate the persistence of socioeconomic inequalities in adopting a number of health-risk behaviours, it also indicates a potential role of these country-level contextual factors in individual-level socioeconomic factors, which are often neglected in research on co-occurrence of behaviours. Expenditures on healthcare and education reflect the countries’ commitments to the population welfare that would enhance a sense of security and equality. These features were found to be associated with adopting health-risk behaviours.1 Healthcare system enables equitable access and effective use of health-promoting resources by eliminating health insecurity, and related anxieties and financial distress.18 Similarly, high-quality education can improve health literacy and acquisition of knowledge related to enhancing health-related behaviours.21 This in turn enhances the population abilities to adopt health-promoting behaviours and avoid health-risk behaviours. The findings of this study to some extent were in line with previous studies which examined the association of co-occurrence of health-risk behaviours with other contextual factors, such as neighbourhood socioeconomic factors, social capital at workplace and area-level socioeconomic deprivation; where worse contextual conditions were related to higher number of health-risk behaviours.13–15 While a previous study found a positive association between health and education spending at country-level and physical activity regardless of socioeconomic status in Europe,33 the study however did not account for GDP and used different definition of physical activity from the one used here. In the current study, adjusting for GDP per capita and individual socioeconomic factors eliminated the significant association between expenditure of health and education and co-occurrence of health-risk behaviours. Although a negative association between contextual factors related to healthcare and education systems and co-occurrence of health-risk behaviours was implied, the association was not strong. A possible explanation would be the known correlation between contextual factors related to education and individual-level socioeconomic factors. In other words, greater public expenditure is highly correlated with education at the individual level, hence the inclusion of education in the analytical model attenuates the association between public expenditure on education and health behaviours. It is also possible that in countries with generous healthcare systems, some people take less responsibility for their own health and engage in health-risk behaviours. Finally, there is very small variation in the contextual factors in the EU countries. This undoubtedly contributed to the relatively weak association observed in this study. The findings on co-occurrence of health-risk behaviours and their association with contextual factors and individual socioeconomic position also emphasise the need to adopt a holistic approach addressing multiple health behaviours rather than a single behaviour approach. There are some limitations. Firstly, this is a cross-sectional study, therefore cannot establish causality. Secondly, because the survey was self-reported, the response might be over-reported or under-reported, especially, health-risk behaviours and socioeconomic position. Thirdly, this study included 27 countries, which might not be a sufficiently large number for assessing variation in multilevel analysis. Furthermore, the terms of joining the European Union include having consistent policies, hence the variations between EU countries are very small. The cut-off points of some of the health-related behaviours did not perfectly match the guidelines or optimal behaviours; however, we used the closest cut-off points to the guidelines. Finally, participants with missing data were excluded throughout the analyses, which might a potential for selection bias, although the excluded number was relatively small. This is the first study, which examined the associations between co-occurrence of health-risk behaviours and healthcare and education systems at the country level using multilevel analysis. While the study highlights the importance of developing high quality healthcare and education systems, generously supported by public fund in relation to co-occurrence of health-risk behaviours, the association might be attenuated by individual socioeconomic factors because of the possible association with country-level contextual factors. Conflicts of interest: None declared. Key points Better quality of healthcare system appeared to be inversely related to co-occurrence of health-risk behaviours. Greater public expenditure on healthcare and education appeared to be inversely related to co-occurrence of health-risk behaviours. Individual socioeconomic positions might attenuate the associations between co-occurrence of health-risk behaviours and healthcare/education systems in Europe. The findings will potentially enable identifying population/countries with greater needs for health promotion policies aiming at reducing multiple health-risk behaviours. References 1 Berkman LF, Kawachi I. Social Epidemiology . New York: Oxford University Press, 2000. 2 Danaei G, Vander Hoorn S, Lopez AD, Group CRAc, et al.   Causes of cancer in the world: comparative risk assessment of nine behavioural and environmental risk factors. Lancet  2005; 366: 1784– 93. Google Scholar CrossRef Search ADS PubMed  3 Paffenbarger RSJr, Hyde RT, Wing AL, et al.   The association of changes in physical-activity level and other lifestyle characteristics with mortality among men. N Engl J Med  1993; 328: 538– 45. Google Scholar CrossRef Search ADS PubMed  4 Hu FB, Manson JE, Stampfer MJ, et al.   Diet, lifestyle, and the risk of type 2 diabetes mellitus in women. N Engl J Med  2001; 345: 790– 7. Google Scholar CrossRef Search ADS PubMed  5 Hanioka T, Ojima M, Tanaka K, et al.   Causal assessment of smoking and tooth loss: a systematic review of observational studies. BMC Public Health  2011; 11: 221. Google Scholar CrossRef Search ADS PubMed  6 Goldstein BY, Chang S-C, Hashibe M, et al.   Alcohol consumption and cancer of the oral cavity and pharynx from 1988 to 2009: an update. Eur J Cancer Prev  2010; 19: 431. Google Scholar CrossRef Search ADS PubMed  7 Moynihan P. The Role of Diet in the Prevention of Dental Diseases . Comprehensive Preventive Dentistry: Wiley-Blackwell, 2012. 99 p. 8 Tada A, Watanabe T, Yokoe H, et al.   Relationship between the number of remaining teeth and physical activity in community-dwelling elderly. Arch Gerontol Geriat  2003; 37: 109– 17. Google Scholar CrossRef Search ADS   9 Nowak MA. Physical activity and its associations with other lifestyle elements in Polish women. J Hum Kinet  2011; 29: 161– 72. Google Scholar CrossRef Search ADS PubMed  10 Glasgow RE, Goldstein MG, Ockene JK, Pronk NP. Translating what we have learned into practice: principles and hypotheses for interventions addressing multiple behaviors in primary care. Am J Prev Med  2004; 27: 88– 101. Google Scholar CrossRef Search ADS PubMed  11 McAloney K, Graham H, Law C, Platt L. A scoping review of statistical approaches to the analysis of multiple health-related behaviours. Prev Med  2013; 56: 365– 71. Google Scholar CrossRef Search ADS PubMed  12 Khaw K-T, Wareham N, Bingham S, et al.   Combined impact of health behaviours and mortality in men and women: the EPIC-Norfolk prospective population study. Obstet Gynecol Surv  2008; 63: 376– 7. Google Scholar CrossRef Search ADS   13 Halonen JI, Kivimäki M, Pentti J, et al.   Quantifying neighbourhood socioeconomic effects in clustering of behaviour-related risk factors: a multilevel analysis. PLoS One  2012; 7: e32937. Google Scholar CrossRef Search ADS PubMed  14 Lakshman R, McConville A, How S, et al.   Association between area-level socioeconomic deprivation and a cluster of behavioural risk factors: cross-sectional, population-based study. J Public Health  2011; 33: 234– 45. Google Scholar CrossRef Search ADS   15 Väänänen A, Kouvonen A, Kivimäki M, et al.   Workplace social capital and co-occurrence of lifestyle risk factors: the Finnish Public Sector Study. Occup Environ Med  2009; 66: 432– 7. Google Scholar CrossRef Search ADS PubMed  16 World Health Organisation. Quality of care: a process for making strategic choices in health systems 2006. Available at http://www.who.int/management/quality/assurance/QualityCare_B.Def.pdf. 17 Heymann DL, Prentice T, Reinders LT. The World Health Report 2007: a Safer Future: Global Public Health Security in the 21st Century: World Health Organisation, 2007. 18 Ensor T, Cooper S. Overcoming barriers to health service access: influencing the demand side. Health Policy Plann  2004; 19: 69– 79. Google Scholar CrossRef Search ADS   19 Park CL, Iacocca MO. A stress and coping perspective on health behaviors: theoretical and methodological considerations. Anxiety Stress Coping  2014; 27: 123– 37. Google Scholar CrossRef Search ADS PubMed  20 World Health Organisation Commission on Social Determinants of Health. Closing the gap in a generation. Health equity through action on the social determinants of health: World Health Organization; 2008. Available at http://whqlibdoc.who.int/publications/2008/9789241563703_eng.pdf. 21 Johnson W, Kyvik KO, Mortensen EL, et al.   Education reduces the effects of genetic susceptibilities to poor physical health. Int J Epidemiol  2010; 39: 406– 14. Google Scholar CrossRef Search ADS PubMed  22 European Commission. ICPSR Codebook: Inter-university Consortium for Political and Social Research (ICPSR); 2013. Available at http://www.icpsr.umich.edu/icpsrweb/ICPSR/studies/32441. 23 EU Working Group. EU Physical Activity Guidelines Recommended Policy Actions in Support of Health-Enhancing Physical Activity. Brussels; 2008. 24 Filippidis FT, Agaku IT, Vardavas CI. Geographic variation and socio-demographic determinants of the co-occurrence of risky health behaviours in 27 European Union member states. J Public Health  2016; 38: e13– 20. Google Scholar CrossRef Search ADS   25 National Institute for Health and Care Excellence. Dental checks: intervals between oral health reviews 2004. Available at https://www.nice.org.uk/guidance/cg19/chapter/1-guidance. 26 The World Bank. 2014. Available at http://data.worldbank.org/. 27 Health Cosumer Powerhouse. Euro Health Consumer Index 2009. Available at http://www.healthpowerhouse.com/files/Report-EHCI-2009-090925-final-with-cover.pdf. 28 Eurostat. Your key to European statistics; Public expenditure on education 2016. Available at http://ec.europa.eu/eurostat/tgm/table.do?tab=table&init=1&language=en&pcode=tsdsc510&plugin=1. 29 UNESCO. Education for All Development Index 2009. Available at http://en.unesco.org/gem-report/sites/gem-report/files/efagmr2009_Annex1_EDI.pdf. 30 West A, Stokes E, Edge A. Secondary education across Europe: Curricula and school examination systems: Centre for Educational Research; 1999. Available at http://www.leeds.ac.uk/educol/documents/00001195.htm. 31 Liberatos P, Link BG, Kelsey JL. The measurement of social class in epidemiology. Epidemiol Rev  1988; 10: 87– 121. Google Scholar CrossRef Search ADS PubMed  32 Adler NE, Epel ES, Castellazzo G, Ickovics JR. Relationship of subjective and objective social status with psychological and physiological functioning: preliminary data in healthy, White women. Health Psychol  2000; 19: 586. Google Scholar CrossRef Search ADS PubMed  33 Lera-López F, Wicker P, Downward P. Does government spending help to promote healthy behavior in the population? Evidence from 27 European countries. J Public Health  2015; 38: e5– e12. Google Scholar CrossRef Search ADS   34 Cameron AC, Trivedi PK. Regression Analysis of Count Data . Cambridge, MA: Cambridge University Press; 2013. Google Scholar CrossRef Search ADS   35 Snijders TAB, Bosker RJ. Multilevel Analysis: an Introduction to Basic and Advanced Multilevel Modeling , 2nd edn London: Sage Publishers, 2012. 36 Carle AC. Fitting multilevel models in complex survey data with design weights: Recommendations. BMC Med Res Methodol  2009; 9: 1. Google Scholar CrossRef Search ADS PubMed  37 Assareh H, Achat HM, Stubbs JM, et al.   Incidence and variation of discrepancies in recording chronic conditions in Australian hospital administrative data. PLoS one  2016; 11: e0147087. Google Scholar CrossRef Search ADS PubMed  38 Noble N, Paul C, Turon H, Oldmeadow C. Which modifiable health risk behaviours are related? A systematic review of the clustering of Smoking, Nutrition, Alcohol and Physical activity ('SNAP') health risk factors. Prev Med  2015; 81: 16– 41. Google Scholar CrossRef Search ADS PubMed  39 Meader N, King K, Moe-Byrne T, et al.   A systematic review on the clustering and co-occurrence of multiple risk behaviours. BMC Public Health  2016; 16: 657. Google Scholar CrossRef Search ADS PubMed  © The Author 2017. Published by Oxford University Press on behalf of the European Public Health Association. All rights reserved.

Journal

The European Journal of Public HealthOxford University Press

Published: Feb 1, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 12 million articles from more than
10,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Unlimited reading

Read as many articles as you need. Full articles with original layout, charts and figures. Read online, from anywhere.

Stay up to date

Keep up with your field with Personalized Recommendations and Follow Journals to get automatic updates.

Organize your research

It’s easy to organize your research with our built-in tools.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

Monthly Plan

  • Read unlimited articles
  • Personalized recommendations
  • No expiration
  • Print 20 pages per month
  • 20% off on PDF purchases
  • Organize your research
  • Get updates on your journals and topic searches

$49/month

Start Free Trial

14-day Free Trial

Best Deal — 39% off

Annual Plan

  • All the features of the Professional Plan, but for 39% off!
  • Billed annually
  • No expiration
  • For the normal price of 10 articles elsewhere, you get one full year of unlimited access to articles.

$588

$360/year

billed annually
Start Free Trial

14-day Free Trial