In this study we used data from Ireland’s 2010/11 and 2014/15 National Drug Prevalence Surveys to determine relationships between age at first use of alcohol, tobacco and cannabis and patterns of cannabis use, frequency of use within the last 30 days and whether age at first substance use was related to having a CUD. For the full sample, when compared to former users, the odds of being a current cannabis user were found to be reduced by 11% and 4% for each year of delayed alcohol and cannabis use onset, respectively. Among subjects who indicated current cannabis use, significant inverse linear relationships were noted, with increasing age of first use of tobacco and cannabis being associated with a decreased frequency of cannabis use within the last 30 days. In addition, fully adjusted models demonstrated relationships between age at first use of tobacco and cannabis and having a CUD among subjects aged 15–34 years of age.
Although previous studies on the relationship between alcohol and cannabis are conflicting, our finding that earlier onset of tobacco and cannabis was associated with heavier current cannabis use has been observed in the literature [9, 12, 28, 29]. We also noted that mean age of first use of alcohol and tobacco was lower than cannabis among survey participants. Consequently, these findings could be interpreted as a supporting the sequential initiation pattern of alcohol and tobacco preceding cannabis use. However, although substance use in Western societies normally begins with alcohol or tobacco use, studies have suggested that many other factors also influence patterns of cannabis use, thus limiting the predictive ability of age of onset as a risk factor for later use of the drug . These factors include context of first use, type of use, timing of use, family substance use, parental education, externalising behaviours and conduct problems [31,32,33].
Nevertheless, the linear associations observed in this study showing that earlier use of tobacco and cannabis was related to heavier cannabis use among current users are important in light of research demonstrating that a dose-response relationship may exist between frequency of cannabis use and adverse health effects , including mental health disorders, acute psychotic symptoms, abnormal cognitive development, chronic lung disease and cardiovascular disease [1, 35]. Research has also suggested that earlier onset substance use may lead to assimilation into deviant and substance-using peer groups, which in turn may lead to increased risk of escalation of substance use . Proposals to liberalise cannabis laws are currently being explored in many countries, with some jurisdictions having commenced legalisation of cannabis . In 2019, legislation was passed to allow for a Medical Cannabis Access Programme to come into operation in Ireland on a pilot basis for 5 years. While there has been concern that legalising cannabis for medical or adult use may increase adolescent use [38, 39], the early evidence on this is certainly not conclusive . Given our finding that earlier onset use was associated with increased likelihood of progressing to a CUD, the impact of policy changes upon age of cannabis initiation will need to be monitored.
Our results showing that age of cannabis use onset was independently associated with having a CUD is also supported by the literature. In a cross-sectional sample of 8068 participants, Le Strat et al.  found that the younger respondents were when they began using cannabis, the greater their likelihood of experiencing a CUD. In a representative sample of 1520 youth aged 14.9–17.4 years, Swift et al.  found early use onset of cannabis to be related to the risk of becoming cannabis dependent. Interestingly, we additionally found earlier onset of tobacco use to be associated with having a CUD. This finding contrasts with previous research; in a prospective-longitudinal community study of 3021 subjects, Behrendt et al.  observed no direct cross-substance momentum of younger age at first tobacco use for the risk of developing a CUD.
It should be noted, however, that relationships between tobacco and cannabis use onset and CUD were only observed among younger survey respondents in our sample. Nevertheless, a majority (85.0%) of subjects in our study who indicated having a CUD were aged between 15 and 34 years and mean age at first use of tobacco and cannabis was lower among younger subjects compared to participants aged 35 years or older. This may partly account for observed findings. However, we also noted that relationships between tobacco and cannabis use onset and frequency of cannabis use were noticeably strong among subjects aged 15–34 years of age. Heavier cannabis use is a known risk factor for having a CUD . Consequently, earlier tobacco and cannabis use may be a contributor to the more intense cannabis use necessary for CUD development  and this may be of particular importance among younger adults.
In Ireland, the type of cannabis being used has changed in recent years, from relatively low potency resin (‘hash’) to higher potency herbal cannabis (‘weed’) [7, 44]. This trend in use of increasingly potent herbal cannabis has also occurred across Europe . High-potency cannabis is associated with an increased severity of dependence, especially in young people . The increase in use and move towards higher potency products in Ireland has coincided with a substantial increase in treatment presentations . Drug treatment services have traditionally been funded in the absence of comprehensive, quantitative planning models . However, planning models based on the needs of the population are important for the successful implementation of treatment services and to adequately plan these services requires an understanding of the population in need of treatment . From a secondary prevention perspective, it is also necessary to know who is at greatest risk of progressing from substance initiation to riskier patterns of future use. Finding from this research suggest that in Ireland, younger adult cannabis users who report a more precocious pattern of early use of substances, including alcohol and especially tobacco and cannabis, should be targeted.
Strengths and limitations
This research has several strengths. This is the first study to examine relationships between age of alcohol, tobacco and cannabis use onset and cannabis use patterns, frequency of cannabis use and CUD, using a sample of lifetime cannabis users from Ireland. A further strength of this research is the large sample size which utilised data from two population studies that used the same field survey procedures and data collection methods and which were weighted and adjusted for the year of data collection; thus, our findings are generalisable to the whole population. We also controlled for important potential confounders in analyses and used valid and reliable measures of cannabis abuse and dependence, defined using the DSM-IV, and Ireland is one of the first countries in Europe to include these variables in two general population surveys. Research on relationships between substance use onset and cannabis use is important for preventative work and for informing and assessing drug policy. This is especially relevant at the present time given the debate regarding the decriminalisation of cannabis in many countries.
Despite these strengths, several limitations should be noted. The cross-sectional study design limits inference with regard to causality and precludes drawing conclusions regarding the temporal direction of relationships. Nevertheless, while a cross-sectional study has reduced ability to identify direction of causality in relationships, it has the advantage of including a much wider age range than typical prospective studies. Another limitation of this research is the use of self-reported questionnaires which are subject to potential inaccuracies, recall and reporting bias. Therefore, residual confounding arising from imprecise measurement of variables should also be considered. Also, we did not have data on other substance use disorders or psychiatric disorders for either survey and these may be important potential confounders. In addition, due to sample size constraints, we analysed cannabis abuse and dependence together rather than as separate constructs. Research examining abuse and dependence separately may be warranted as abuse or dependence criteria may be differentially predictive of adverse outcomes.