Abstract
Today’s adolescents are far less likely to smoke, drink, use drugs or be sexually active than their 1990s counterparts. My doctoral project set out to describe and explain trends in adolescent risk behaviours in the early 21st century, considering not only New Zealand but also other high-income countries that have seen dramatic declines in adolescent risk behaviours over the past 15-20 years. To date, this megatrend in adolescent behaviour has not been widely recognised within public health, much less explained.
My research questions were:
1. What are the key trends in adolescent sexual behaviour and substance use (tobacco, alcohol, cannabis) in New Zealand, Australia, England and USA, 1990-2017? What do the patterns suggest about the possible drivers of risk behaviour decline?
2. What does analysis of repeat cross-sectional data from New Zealand reveal about the drivers of declining risk behaviour in secondary school students in this country?
3. What are the possible explanations for the international decline in adolescent risk behaviours? How plausible are these explanations, based on existing theory and evidence and the findings of my own empirical analyses?
4. Do the observed declines represent separate trends with separate drivers; a single trend with common underlying driver(s); or knock on effects with declines in one risk behaviour leading to declines in others?
Key trends
I collated data from New Zealand, Australia, England and the USA on trends in smoking, drinking, cannabis use, and sexual behaviour among adolescents aged less than 16. I found strong declines in all of these behaviours over the past 15 to 20 years. However, there were no corresponding improvements in adolescent fruit and vegetable intake, physical activity, condom use or mental health. In fact, there were significant declines in many of these indicators. These trends were common to most (but not all) high income countries, and were distinct from adult trends.
The patterns observed provide clues as to possible drivers. For example they suggest adolescents are not becoming healthier or more health-conscious in general. Countries with very different regulatory environments have similar trends, suggesting that public health interventions are probably not the primary driver. Similarities across countries and behaviours are suggestive of broad socio-cultural changes impacting on many behaviours simultaneously. However, other clues (e.g. a lag of several years between smoking decline and declines in alcohol indicators in many countries) point to the importance of behaviour-specific factors.
Changing social context
I investigated the changing social context, identifying some of the broad social changes that have influenced the experience of adolescence over the past 30 years. These include:
- greater social concern about health and safety
- greater awareness of the harms of substance use
- more involved and protective parenting
- increasing pressure on adolescents to prepare for a competitive job market
- increasing exclusion of young people from public space
- increasing time spent engaged in digital media use
- declining exposure to pro-smoking influences.
My literature review revealed that many hypotheses for declines in adolescent risk behaviours had been discussed in the literature but few had been empirically tested using rigorous methods.
Drivers of risk behaviour decline in New Zealand
Having identified potential contributory factors, I tested those for which repeat cross-sectional data was available in nationally representative New Zealand surveys: the annual ASH Year 10 Snapshot (2003-2015) and the Youth 2000 series (2001, 2007, 2012). Using regression analyses, I investigated the extent to which each hypothesised contributor accounted for trends in adolescent smoking, cannabis use, binge drinking and sexual activity in New Zealand secondary students aged less than 16 years.
My findings provide evidence against several hypotheses. Factors that did not make a statistically significant contribution to adolescent trends in New Zealand included:
- parental smoking, drinking and cannabis use in the home
- sibling smoking
- exposure to others’ smoking in the home
- parental monitoring
- family connectedness
- school connectedness
- time hanging out with peers
- engagement in paid work.
The main factors that influenced trends were i) large declines in the proportion of adolescents who thought smoking and drinking were acceptable in people their own age, and ii) impacts of risk behaviour trends on one another. Trends in adolescent smoking and binge drinking appear to be primarily driven by tobacco-specific and alcohol-specific factors respectively, in particular attitude changes. In contrast, declines in cannabis use and sexual activity were largely explained by declines in smoking and binge drinking. Common underlying drivers in home, school, and leisure contexts only made a very minor contribution in combination. However, the pattern of findings in the latter half of the study period (2007-2012) is consistent with a decline in going out at night with friends leading to fewer opportunities for all four risk behaviours.
I used survival analysis to investigate whether age of initiation (i.e. the age that young people try smoking, drinking, using cannabis and having sexual intercourse for the first time) changed over the study period among secondary school students. I found that, consistent with US and Australian studies, age of initiation for all four outcomes increased between 2007 and 2012 in New Zealand. These findings indicate that increasing age of initiation (though not a causal explanation) helps to explain the observed risk behaviour trends in secondary students.
Explanations for the international decline in risk behaviour
While a full explanation for international risk behaviour trends remains elusive, some key drivers are beginning to emerge from the international evidence and my own findings. The empirically-established contributors identified thus far are mainly tobacco-specific and alcohol-specific factors such as parental rules and expectations, decreasing ease of access, and decreasing adolescent approval of smoking and drinking. Knock on effects from one risk behaviour to another also appear to be important, with evidence both from my own research and international studies indicating that declining smoking and drinking significantly explain declining cannabis use in this age group. A common driver underlying declines in many risk behaviours is a decline in unsupervised time and space. In particular, adolescents are going out with their friends in the evenings less frequently, thereby reducing opportunities for risk behaviours. It seems likely that broad social changes have created a milieu in which these more proximal factors have emerged.
International evidence does not support the hypothesis that digital media use has displaced risk behaviour in terms of time use. However, the possibility that the digital revolution has contributed to a shift in the social meaning and function of substance use and sex (e.g. as rites of passage) cannot be excluded.
Separate trends, single trend, or knock on effects?
Behaviour-specific factors, common drivers, and knock on effects all appear to have played a role in the decline of adolescent risk behaviours. It is likely that broad cross-national influences (e.g. the digital revolution, labour market changes) are interacting with behaviour-specific factors (e.g. less permissive parental attitudes to youth alcohol use) and the existing culture in each country to produce the pattern of trends that has been observed.
Conclusion
The findings suggest that the picture is complex, with multiple drivers and bi-directional relationships operating in a dynamic system. The finding that risk behaviour trends impact strongly on one another underlines the importance of investigating more than one risk behaviour at a time, and exploring the relationships between them. My thesis highlights both the importance for public health of tackling complex questions about how and why behaviour changes at the population level, and also the limitations of our current tools for dealing with complexity. Overcoming these limitations may require new interdisciplinary collaborations, new methods, and new ways of thinking, but the rewards are potentially transformational.