Social inequalities in changes in health-related behaviour among Slovak adolescents aged between 15 and 19: A longitudinal study

Whereas adolescence is traditionally viewed as an age period of good somatic health, psychosocial health variables, e.g., psychosomatic health complaints or health-related behaviour (HRB), play a decisive role in determining adolescents' self-perceived health.

Over the last thirty years a number of studies describing the relationships between socioeconomic position (SEP), health-related behaviour HRB and health have been performed [1‐7]. Undoubtedly HRB is a very important determinant of health as well as a contributor to socioeconomic inequalities in health [2, 8].

Since adolescents' psychosocial health problems may have major implications for adult morbidity and mortality, investigating their correlates, such HRB deserves priority. With respect to HRB, adolescence is one of the most important periods of life. Adolescence is characterised by a strong tendency to experiment with risk behaviour. The desire for novelty and the courage for experiment are much greater in adolescence than in later life [9]. Despite it being illegal, many young people have experience with drinking alcohol before turning 18, likewise with using drugs such as marijuana [9, 10]. Most adult smokers took up regular smoking in the period of adolescence [11, 12]. Even if most students are physically active at school, their compulsory school involvement often fails to translate itself into leisure time physical activity later [13, 14]. Moreover, the influence of peers and youth subcultures on HRB is statistically significant [15, 16]. HRB established during this period tends to be maintained into adulthood [17‐19]. Due to the characteristics mentioned above, adolescence is also a very sensitive period for interventions and policies aimed at promoting health by focusing on risk behaviour [20].

Previous research has consistently documented social class gradients in child and adult health [21]. Low SEP adults are more likely to engage in risky health behaviours [22].

Findings among adolescents are not so consistent. Previous research has shown a very strong traditional (consistent with adult behaviours) socio-economic gradients in insufficient physical activity of adolescents [6, 14, 23‐25]. Also regarding smoking by adolescents, mostly traditional socio-economic gradients were found [21, 26], though there are few studies which reporting no [6, 27, 28], or a reversed socio-economic gradient [24]. On the other hand, no consistent socioeconomic differences in alcohol consumption have been confirmed among adolescents. The relationship between SEP and alcohol consumption is usually weak [6, 24, 27, 29] or reversed (compared to adult socioeconomic gradients) [26, 30]. While binge drinking is associated with lower socioeconomic groups, some studies report that regular but moderate drinking is more common in higher socioeconomic groups [31, 32]. Similarly in marijuana use among adolescents, mostly no [6, 29] or reversed socioeconomic [29] gradients have been reported among adolescents.

Furthermore, considerable gender differences can be found with relation to health-related behaviour, both in adults and in adolescents. Generally, males exhibit more health-risk and less health-protective behaviour than females [33, 34]. However, in recent years some studies have reported a remarkable increase in smoking among women [33‐35]. A sedentary lifestyle is also more common for women [13, 36‐38].

Several studies have shown differences by gender in socioeconomic gradients in HRB. In a sample of Australian adolescents, Scragg et. al.[39] found a negative association between SEP and smoking only among females; the lower was the SEP the higher was the smoking occurrence. Nevertheless, the majority of the studies focusing on gender differences for socioeconomic gradients in smoking found no difference [21]. We have also found no study showing gender differences in the socioeconomic gradient of alcohol and marijuana consumption among adolescents. On the other hand, several adolescent studies report that they only found associations between physical activity and SEP for females [22, 40, 41]. However, the majority of the studies found no significant differences in the association between SEP and physical activity between boys and girls [21].

Several studies have also gender differences in the relationship between health behaviours and adolescents' self-perceived health. A recently published study on a representative sample of Hungarian adolescents aged 14 to 19 suggests that among boys drug use and the lack of physical activity are significant predictors of self-perceived health, but not among girls. Among girls, smoking may act in a similar way [42]. Gender differences in health perceptions are often reported. This is probably because women, in contrast to men, consider a broader set of factors when making general ratings of health, e.g., psychological factors and minor subjective health complaints [43].

It should be noted that there may be shortcomings in the use of parental SEP markers as measures of social status during adolescence [44]. Traditionally, adolescent studies assess the SEP of the parents (e.g., parental education, parental occupation, family income) as indicators of SEP. As adolescents spend less time at home and experience transition into the independence of adulthood, parental SEP markers may not be accurate indicators of adolescents' social status [21]. It is possible that, during adolescence, HRB is more strongly influenced by peer social status (i.e. the social standing of an adolescent within his/her school), as opposed to family social status. An adolescent's family social status is an assigned status and its impact may be too distant, as teens gain independence, to impact their health behaviour choices [45]. Status among their peers, however, is an earned status, and may better capture the experience of placement within a social hierarchy during adolescence [46].

Most of these studies of socioeconomic differences in HRB in adolescents, however, were cross-sectionally designed. The use of a longitudinal design may be highly significant in exploring changes in HRB, particularly in adolescents. The aim of this study was to analyze the changes in HRB in relation to SEP in a cohort of Slovak adolescents aged between 15 and 19. Our attention is accordingly focused on the following research questions. Did the HRB of adolescents change between the ages 15 and 19? Were the changes in HRB during adolescence related to SEP? If yes, which socioeconomic indicators showed the steepest graduation in HRB?

This study describes changes in HRB according to SEP in a cohort of Slovak young adults aged between 15 and 19. Between these ages a greater increase in alcohol consumption and experience with marijuana was found for males than in females. However, the increase in insufficient physical activity was greater for females. Since there was already a clear gender difference at the study baseline, the gap between the genders in relation to these HRBs became wider. The finding that more males drank alcohol and used marijuana was unsurprising as similar outputs had been published in earlier studies [13, 37, 50‐53]. A sedentary lifestyle is more common among females than in males in late adolescence and young adulthood [13, 37, 52, 53]. High consumption of alcohol is likely to be linked to the young males' lifestyle, associated with a normative peer pressure to drink [34]. The reasons for marijuana use among men may be similar to those for alcohol, and peer attitudes play an important role in explaining this [54].

Studies performed over the last few decades present a well-documented equalisation trend in the smoking behaviour of males and females. This has resulted in an increasing number of smoking females in the community [34, 35], while the proportion of smoking males is decreasing [35] or remaining stable [11].

As no socioeconomic differences in changes in the HRB studied were found among males, the socioeconomic gradients in HRB described at T1 were similar to those at T2 for males. The number of smokers was highest in the lowest socioeconomic groups for all socioeconomic indicators. A similar outcome was obtained when using marijuana and insufficient sporting activity among males, but not for every socioeconomic indicator. No socioeconomic gradient was found for alcohol consumption by males. These findings are consistent with the results of previous studies [32, 55, 56].

Results on experience with marijuana are somewhat difficult to understand. While in relation to women's educational level and to their employment status no socioeconomic gradient was found, in relation to their parental educational and occupational level a reversed gradient was discovered. On the other hand, among men traditional gradients were found in relation to their education level and to their employment status, and no gradients in relation to parental socioeconomic indicators. However, this ambiguous result fits with those of previous studies on the association of SEP and marijuana experience during adolescence. According to the current literature review by Hanson and Chen [21], these studies also found varying associations. The most frequently reported finding was of no socioeconomic gradients, while some reported traditional socioeconomic gradients: higher SEP associated with less use; and some reported reversed socioeconomic gradients: the higher the SEP, the greater the use. The character of the socioeconomic gradients was usually determined by the type of socioeconomic indicator used. The findings of previous studies suggest that the relationship between the social status of parents (e.g. educational or occupational status) as socioeconomic indicators and marijuana use is more likely to show no [29, 57] or a traditional association, [58, 59] whereas the relationship between financial resources [60] or self-assessed SEP [29] as indicators and marijuana use is more likely to be reversed. However, there are studies with results not fully consistent with these findings [21].

According to Luthar and Latendresse, [61] high-SEP adolescents engage in negative health behaviours in order to combat the stress, anxiety, and depression they experience from achievement pressures. It is possible that this type of pressure could be common in relation to highly-educated parents with a less academically successful child. In combination with more available money and negative peer influence, the group having parents with high social status could become more susceptible to negative health-related behaviour. These explanations require more attention in future.

Using SEP based on the parents' characteristics yielded an inconsistent or reversed pattern of socioeconomic differences in smoking among females, while using SEP based on the adolescents themselves – their current position – yielded socioeconomic differences in smoking unfavourable for females of lower SEP. An explanation may be that measuring SEP using the parents' characteristics loses validity in adolescence [3, 7]. On the other hand, using the socioeconomic characteristics of adolescents is also problematic [4]. Methodological problems related to measuring SEP may be a source of inconsistency in findings related to socioeconomic differences in HRB. Glendinning et al. [62], with the aim of measuring SEP based on respondents' instead of their parents' characteristics, used young adults' economic activity. It is suggested that subjects who continue to study will differ as regards SEP from those who enter the labour market and succeed (employed) or fail (unemployed). Hanson and Chen [21] conclude in a current literature review of socioeconomic position and health related behaviour in adolescence that future studies should employ alternative measures of SEP, such as an adolescent's perception of social status relative to others in their peer group, due to recent findings which suggest that such measures may be a better predictor of adolescent health than the traditional objective measures. By employing alternative measures of social status, future studies may be able to further clarify the socioeconomic patterns for adolescent health behaviours.

Findings related to socioeconomic differences in HRB may depend on the way HRB is measured. Sweeting and West explored socioeconomic differences in smoking among adolescents with respect to the definition of smokers. The stricter the definition of smoking, the clearer the socioeconomic gradients in smoking that were found. Moreover, reversed relationships between SEP and HRB were found when occasional smoking was explored [5]. Our definition of smokers includes both daily and occasional smokers. We therefore repeated post-hoc our analyses for daily smokers. The character of socioeconomic gradients in smoking among males remained stable. On the other hand, results for females were more in line with findings of Sweeting and West; the traditional gradients previously found according to their employment status and according to their education level became steeper and the reversed socioeconomic gradient previously found according to their parents' educational level was no more statistically significant.

Changes in alcohol consumption in females were mostly related to the education levels of their parents and the respondents' own education levels. Females with higher SEP reported greater increase in alcohol consumption compared to females with lower SEP. Moderate drinking may be part of a high SEP lifestyle geared towards pleasure and comfort [31, 32, 55, 56].

No statistically significant socioeconomic differences among males were found at T2. On the other hand, obvious socioeconomic gradients regarding insufficient physical activity were found for females at T2. This is consistent with the results of several previous studies [22, 40, 41]. Female respondents' school type and the education level of their parents appeared to be statistically significant predictors of physical inactivity. The largest increase in insufficient physical activity was observed among females with low education and with low-educated parents. Females may be more likely to exercise if they are enrolled in a formal activity, such as a dance class or swimming lessons [63]. Even though lower education is usually associated with lower income, the inaccessibility of sports facilities due to high costs could explain this effect only partially. Overall, these findings suggest that higher family social status or prestige may be a stronger influence on physical activity than financial resources in high SEP adolescents [21]. The explanation for this relationship is more likely to be a kind of normative behaviour towards damaging behaviours [64] and better ability to make informed choices [14, 63].

We have not adjusted the analysis of socioeconomic gradients in one health-related behaviour with regard to the other health-related behaviours. Health-related behaviours may indeed cluster, but it is open to discussion whether they should be considered as confounders of each other, i.e. having a causal relationship with both the determinant (change of age) and the outcome under study, i.e. another health-related behaviour. We think that the most valid approach is to consider them as outcomes with partially identical causes (such as SEP), but having no direct causal effects on each other. For that reason, we have not adjusted our analyses for the effects of other outcomes.

This study has several strengths and limitations. A major strength of our study is its longitudinal design. The main limitation of this study, just as in every longitudinal research project using self-administered postal questionnaires in adolescents, is the relatively low response rate. Compared to females and better-educated males, low-educated males responded slightly less. However, differences in response rates by SEP were relatively small, thereby biased results due selective non-response are less probable. The period of young adulthood is associated with changes of permanent residence. Some of the respondents became independent of their parents and changed residence; large numbers of respondents became simply unreachable because of study or work in another part of the country or abroad. According to data from the Statistical Office of Slovak Republic, about 7.5% of all employed Slovak citizens were employed abroad. A very substantial part of this group consists of young people aged from 18 to 30 years. This may explain some of the differences in response rate we found in terms of gender and SEP, but probably only a part of them. Another limitation stems from the way in which we performed the McNemar test. Data was stratified not only by gender but also by every indicator of SEP and this resulted in 48 different p-values. Multiple testing may have affected some of our findings.

In conclusion, our follow-up study contributes to the debate on health inequality by investigating the relationship between SEP and health-related behaviour in late adolescence and early adulthood, which is relatively rarely investigated. The results show that the dynamics of HRB change are related to SEP and gender. Socioeconomic differences in HRB established in adolescence remained stable until young adulthood among males but not among females. Some diminution of the gender gap in smoking was confirmed.

Initiation into HRB takes place during the turmoil of adolescence, which is characterized by many personal and social changes. It is difficult to capture and understand the dynamics within which the uptake of HRB takes place. More longitudinal research is needed to fully understand the process by which age, SEP and HRB influence health.