Correlates of self-rated health in Southern Europe: evidences from national representative samples in Italy and Serbia

Abstract

Background

Self-rated health (SRH) is a widely adopted tool to compare health across countries. Our aim was to examine SRH differences between Italy and Serbia and to observe the role of predictors of SRH referring to health behaviors within and between both countries.

Methods

We used cross-sectional population-based data from Italian and Serbian national health surveys carried out in 2013. Post hoc cross-standardization was undertaken to ensure that the information from both data sets was comparable.

Results

Univariate and multivariate multinomial logistic regressions showed that Serbians reported bad-SRH significantly more often than Italians. Moreover, consistently across national groups, younger participants, males, higher educated participants and participants with lower body mass index (BMI) had more chances than older, lower educated and higher BMI participants, respectively, to report better SRH. Finally, smoking and drinking behaviors did not correlate with SRH, while the frequency of fruits and vegetables intake was differently associated with SRH across countries.

Conclusion

Health assessments based on SRH in Italian and Serbian national surveys are directly comparable and show similar relationships with socio-demographic correlates and BMI. However, the effect of health behaviors on SRH may differ according to national and cultural contexts.

Introduction

Nowadays, we are facing a growing consensus in support of the use of self-rated health (SRH) to improve policy making in public health and to highlight correlates of health inequalities across countries.^1,2 SRH is a significant predictor of mortality³ and of several chronic diseases.⁴ International organizations such as the World Health Organization (WHO)⁵ and the Organization for Economic Co-operation and Development (OECD)⁶ have been advocating the use of standardized measures of SRH to foster cross-national comparisons. Several studies have looked at between-countries differences^7,8 as well as the change patterns in the role of protective and risk factors within-nations.^9,10 However, while the effects of socio-demographic characteristics have been observed across countries worldwide,^11,12 comparisons of the impact of health behaviors (i.e. eating habits, smoking and alcohol consumption) and health characteristics, such as body mass index (BMI), on SRH are lacking.

Although Italy and Serbia belong to the same geographical macro-area, these two countries differ depending on various demographic and developmental indexes. Italy is ranked first among Southern European countries according to the Human Development Index (HDI)¹³ and third according to the Quality of Life Index (QLI)¹⁴ while is expected to experience one of the largest growths in persons older than 65 years (over 20% by 2020)¹⁵ being already among the top-10 oldest populations.¹⁶ This rapid population aging has social and economic consequences. Nevertheless, total health expenditure in Italy has decreased in recent years following the aftermaths of the global financial crisis of 2008.¹⁷ On the other hand, Serbia is listed as the third worst Southern European country according to HDI,¹³ and the worst according to QLI¹⁴ (data referring to Albania, Montenegro, Macedonia and Moldova are not available). Since early 2000s, major transformations have taken place following years characterized by wars and imposed economic and diplomatic sanctions.¹⁸ In sum, in countries such as Italy and Serbia, there is an impellent need to advise new welfare strategies to face present and forthcoming challenges that may also be relevant for other regions in the Southern European zone.

Both Italy and Serbia dispose of nationally representative data that include measures of health-related habits as a result of the alignment to international standards in health surveys. Previous studies have adopted SRH as a dependent variable in both countries.^19,20 More specifically, while correlates of SRH have been studied more extensively in Italy,^21,22 research in Serbia is more scarce.²³ This study aimed at (i) comparing SRH between these two countries and (ii) analyzing associations of socio-demographic characteristics and health behaviors with SRH within and between Italy and Serbia. Despite the explorative nature of this study, based on the existing research literature on this topic^7–9,12 we expected socio-demographic characteristics (i.e. age, gender, education) to be similarly correlated with SRH across Italian and Serbian national samples, while the relationships between health behaviors and SRH may differ according to specific context-related differences.

Methods

Samples

We used cross-sectional population-based data from national health surveys in Italy and Serbia. More specifically, for this secondary analysis we used data from the 2013 Italian National Health Survey, section Multi-scope Survey on the Family Aspects of Daily Life²⁴ (henceforth, the Italian study) and from the 2013 National Health Survey for the population of Serbia²⁵ (henceforth, the Serbian study).

Yearly, the Italian Institute of Statistics (ISTAT) collects a wide range of information on Italian households. A sample is selected through a complex stratified multistage design within strata of geographic regions (North, Center, South and Islands), municipalities and household sizes. The population of interest is composed by household members residing in Italy. The data are part of European Health Interview Survey wave 2 (EHIS wave 2).²⁶ Out of 25 697 registered households, 20 275 accepted or were able to be interviewed, creating a response rate of 78.9%. Ethical approval was obtained and written informed consent was signed.

The Serbian study was conducted by the Serbian Ministry of Health. The data were harmonized with those of EHIS wave 2. In line with the EUROSTAT recommendations, this survey adopted a stratified two-stage cluster probability sampling. The representative sample was selected within subpopulations at the level of geographic areas/statistical regions (Vojvodina, Belgrade, Šumadija, Western Serbia and Southern-Eastern Serbia) and the level of cities and other settlements/areas. Field researchers informed the participants about objectives as well as ethical and privacy issues. Out of 16 474 registered household members aged 15 and over, 14 623 of them accepted or were able to be interviewed, creating a response rate of 88.8%.

The databases’ comparison has been possible after a post hoc cross-standardization, given the alignment to EHIS wave 2 standards.²⁶ Reading from previous studies that investigated the correlates of SRH,^11,12,19,20 we selected the most relevant variables present in both surveys. The items were translated and back translated by native speakers to ensure they were comparable. Subjects below age 18 were excluded from the analysis, giving a total sample of 16 832 Italians (8016 men and 8816 women) and 14 082 Serbians (6478 men and 7595 women).

Measures

As dependent variable, SRH was measured using a single-item question, namely ‘How is your health in general?’, with five possible answers: very good, good, fair, bad and very bad. For the purpose of analysis, given the skewed distribution of responses (2.4% in the lowest category), the first two were then grouped as ‘good’ and the last two were grouped as ‘bad’. This procedure has been adopted by previous studies.^7,27

The independent variables were selected after reviewing the research literature on factors influencing SRH. The selected items were divided into two groups: socio-demographics and health behaviors. Socio-demographics included: age, gender and education level. Age was categorized into the following groups: 18–29, 30–44, 45–64 and 65+. Gender was classified as males and females. Education level was reclassified into three levels according to the UNESCO International Standard Classification of Education (ISCED 2011):²⁸ ‘high’ (ISCED 5–8), ‘medium’ (ISCED 3–4) and ‘low’ (ISCED 0–2). Items regarding health behaviors included: intake of fruits, intake of vegetables, binge drinking, smoking and BMI. The latter was measured using the formula weight divided by height squared (kg/m²) and then grouped into three groups, according to the WHO classification:²⁹ under/normal weight (<25), overweight (25–29.99) and obese (≥30). Participants were wearing light clothing and no shoes during the measurements. The frequencies of vegetables and fruits intake were measured with the two following questions: ‘How often do you eat vegetables or salad (excluding juice and potatoes)?’ and ‘How often do you eat fruits (excluding juice)?’ Answers were then rearranged into two levels, following the WHO report recommending a minimum of 400 g of fruits and vegetables per day for the prevention of several diseases: inadequate (‘never’, ‘less than once a week’, ‘1 to 3 times a week’ and ‘4 to 6 times a week’), and adequate (‘once or more a day’).³⁰ Risky single occasion drinking, or binge drinking, was measured with a single question: ‘During the past 12 months, how often did you have six or more drinks on one occasion?’ Answers were dichotomized into non-binge drinkers (‘never’ and ‘not during the past 12 months’), and binge drinkers (‘less than once monthly’, ‘once a week’ and ‘every day’). Finally, smoking habit was measured with a single question: ‘Do you smoke at all nowadays?’ Answers were dichotomized into non-smokers (‘no’) and smokers (‘yes, daily’ and ‘yes, occasionally’).

Analysis

The Italian and the Serbian databases were merged to facilitate analyses. Descriptive statistics, univariate and multivariate binary and multinomial logistic regressions analyses were used to explore SRH across national samples.

First, all the study variables were examined using Chi-square tests to test for significant differences between Italy and Serbia. Then, SRH was analyzed as a potential discriminator between Italy and Serbia by univariate (unadjusted) and multivariate (adjusted for age and gender and all study variables) logistic regression analysis. In this instance, binary logistic regression analysis was used as a classification tool taking the country as a dependent variable and SRH as an independent variable (unadjusted and adjusted).

In order to determine predictors of good-SRH, we employed univariate and multivariate multinomial logistic regressions analyses using SRH as dependent variable with fair- and bad-SRH as referent categories in separated analyses. By using multivariate modeling, we were able to distinguish between effects due to socio-demographics and effects due to health behaviors. These analyses were carried out separately for Italy and Serbia because we expected that the predictors and their effects were different for Italians and Serbians. Odds ratios (ORs) in univariate analysis were calculated, then further models were tested adjusting for age and gender first and finally for all significant variables. Only the significant variables in the adjusted for age and gender model entered the last multivariate multinomial logistic regression model. By entering successive blocks of variables, it is possible to assess how each group mediates the effects of previously introduced variables. A mediation effect is suggested when the association between the independent and dependent variables is weakened after the proposed mediators are accounted for.

At last, the data from both countries were pooled together and all significant variables generated in previous country separate models were included in a multinomial model with SRH as dependent variable and country as dummy variable. Serbia was chosen as a reference group because of lower prevalence of good-SRH. To examine whether the association between the independent variables and the outcome measure differed across countries, interaction terms (the differences in the predictor coefficients between countries) between each independent variable and country were included in the multivariate multinomial models.

In all the analyses, a P-value of <0.05 was considered statistically significant. Data analysis was carried out with the SPSS version 21.0 software (SPSS, Inc., Chicago, IL, USA).

Results

Individual-level characteristics of the total sample of 30 914 participants are shown in Table 1.

According to the descriptive statistics, in both samples, the majority of the participants were older than 45 and females outnumbered males. The two countries significantly differed on all analyzed variables. Serbians tended to have worse levels of SRH, to eat more vegetables and to have higher levels of binge drinking in comparisons with Italians. On the other hand, Italians received lower education, had lower BMI and declared to eat more fruits and to smoke less than what Serbian participants did.

Comparison of SRH in both countries (OR: 3.11, 95% CI: 2.88–3.35 and OR:1.26, 95% CI: 1.19–1.33, for bad- and fair-SRH, respectively, compared to good-SRH in favor of Italian participants) remained significant after adjusting for age and gender structure of the respondents (OR: 3.93, 95% CI: 3.60–4.30; OR: 1.48, 95% CI: 1.40–1.56) as well as for all study variables (OR: 3.00, 95% CI: 2.57–3.48; OR: 1.35, 95% CI: 1.24–1.48) (see Table 2).

Univariate and multivariate multinomial logistic regression models for correlates of good-SRH versus bad-SRH and good-SRH versus fair-SRH for Italy are shown in Table 3.

The youngest group of respondents, 18–29, reported bad- and fair-SRH significantly less often than oldest respondents, 65+. The same also hold for the 30–44 and the 45–64 groups. Males, high-educated respondents, normal weight respondents and respondents who declared to adequately eat vegetables reported good-SRH significantly more often than females, low-educated respondents, obese and those who did not declare to adequately eat vegetables, respectively.

Univariate and multivariate multinomial logistic regression models for correlates of good-SRH versus bad-SRH and good-SRH versus fair-SRH for Serbia are shown in Table 4.

In the Serbian sample, age, gender, education level and BMI showed a similar relationship with SRH as in Italy. However, respondents who declared to adequately eat fruits reported good-SRH significantly more often than bad-SRH with regards to those who declared to inadequately eat fruits.

The significant interaction terms between country and each predictor for model comparison of good-SRH versus bad-SRH were: country (Italy) by age (3), by BMI (2) and by vegetables (1). More specifically, middle-aged participants (aged 45–64), overweight participants and those who adequately consumed vegetables had better SRH when compared with their reference categories especially in the Italian sample. Regarding the model comparison of good-SRH versus fair-SRH, the significant interactions were: country (Italy) by age (1), by age (2), by age (3), by gender (1) and by vegetables (1). Results of model comparison of good-SRH versus fair-SRH between Italy and Serbia yielded that youngest participants and males had better SRH especially among Serbians, while young and middle-aged participants (aged 30–64), and those males who adequately consumed vegetables had better SRH especially among Italians.

Discussion

Main finding of this study

This study had two main objectives, namely (i) to compare SRH between two representative samples of Italian and Serbian adult populations and (ii) to analyze associations between socio-demographic characteristics and health behaviors with SRH in both countries. Our main findings were four. First, in both countries, socio-demographic characteristics had similar impacts on SRH. Consistently across the two national groups, males had more chances than females to report good-SRH versus bad-SRH and good-SRH versus fair-SRH. The same hold true for high-educated participants in comparison with medium and low-educated participants. Finally, with age increasing, it also increased the chance to report worse SRH. These findings are consistent with those of previous cross-national³¹ and national-based studies.^32,33 In addition, low BMI was also consistently showing significant positive associations with reporting better SRH across countries. This latter result is aligned with previous research findings.³⁴

Second, Serbian participants reported bad-SRH significantly more often than Italian participants. Our results confirmed previously highlighted findings by Bobak et al.:³⁵ former communist countries report worst self-perceived health than countries which have never been directly ruled by communist governments. In fact, the percentage of the sample that reported bad-SRH in Serbia is in line with the data from countries with similar political histories.³⁵ Indeed, Serbia is a country that still lives the consequences of the 1990s war period and health implications for the defeated aggressor group are well known.³⁶ Moreover, within the 18–34 population, the health discrepancy between Southern and Central European countries seemed most enhanced within inhabitants of former communist countries, and this pattern is consistent even after adjustments for socio-economic factors.³⁷ Even a comparison between Central European adolescents and Eastern European adolescents reported worst-SRH for the latter.³⁸ Therefore, the discrepancy created in the past still lives on today and seems to influence also future generations. Efforts need to be put by all the parts involved in order to fill this gap.

Third, the frequency of fruits and vegetables intake was differently associated with SRH in the countries of interest. For what concerning the probability of reporting good-SRH versus bad-SRH, Italians adequately eating vegetables significantly correlated with reporting good-SRH, while no significant effect was observed for frequencies of fruits intake. On the other hand, Serbians adequately eating fruits significantly reported better SRH, while frequencies of vegetables intake were not correlated with the variable of interest. This finding could be rooted in the two different national geographical configurations and climate standards which, in turn, affect the direct availability of seasonal fruits and vegetables. Further research is needed to assess more precisely the particular effect of fruits and vegetables on health in the Southern European region.

What is already known on this topic

At last, smoking and binge drinking did not influence SRH in this analysis. This result is in conflict with those of previous general studies, which clearly showed the negative impact of both smoking³⁹ and binge drinking⁴⁰ on health. Regarding smoking in Italy, it can be noticed the constant decrease of smokers, mainly due to high efficacy of the so-called Sirchia Law (by that time, the fourth of the same kind in Europe) that from 2005 banned smoking in all indoor public places, causing an 8% decrease in tobacco consumption.⁴¹ About binge drinking instead, it is noteworthy to report the notable fall in alcohol intake in Italy since the 1970s and that peaked in positive in the last decade, with a 23% decrease in per capita alcohol consumption (from 5.6 to 4.4 drinks/week).⁴² On a related note, considering the consumption of alcohol beverages only as a risky behavior may further collide with its cultural value in contexts such as Italy where, for example, a moderate wine consumption is encouraged by physicians as part of a Mediterranean diet⁴³ and Serbia where the cultural attitude toward alcohol is tolerant and is connected with many social rituals.⁴⁴

What this study adds

Nevertheless, one of the strengths of the present research is the use of data collected through standardized sampling, recruitment and data collection protocols. We were allowed to make a proper international comparison for a wide set of variables given also the presence of high response rates in both countries. Thus, implications for health and social care policies can be drawn. Our findings could also be informative for other countries in the Southern European region. In the future, it would be valuable to carry out similar studies including more countries from this geographical area.

Limitations of this study

Finally, it is important to notice that this study was not without limitations. First, given the cross-sectional nature of the data analyzed here, it is not possible to establish causal relationships between levels of SRH and studied factors. Second, other variables that have been found to be significant predictors of SRH in adult populations could not be included in this study since they were omitted or differently measured in the two national databases analyzed here, such as physical activity or a past history of smoking. Last, measure of health status was subjective and it could be biased.⁴⁵ In particular, due to limitations typical in analysis of secondary survey data, there may be recall bias within the self-reported data while some measures remain inaccurate. For example, as pointed out above, the way in which alcohol consumption was measured can be prone to underestimation of this phenomenon for a social desirability effect.⁴⁶ The same may apply to the questions regarding vegetables and fruits consumption. We consider this an important topic for further research.

Acknowledgements

The authors gratefully acknowledge the support received by the ERAWEB (Erasmus Mundus–Western Balkans) joint mobility program (further details can be found on www.erasmus-westernbalkans.eu).

Conflicts of interest

The authors have declared that no competing interests exist.

Authors’ contributions

FL conceived the study and drafted the manuscript. GP contributed to conceive the study and to draft the manuscript. JM helped the analysis and interpretation of data. VB critically revised the manuscript.

References