| Summary: | Background - The Gateway Hypothesis argues that there is a sequence in the use of various drugs and that the use of one substance serves as a gateway for the use of another substance. The aim of this thesis was to examine various aspects of the Gateway Hypothesis using data from the Avon Longitudinal Study of Parents and Children (ALSPAC) (unless stated otherwise). Individual summaries of the multiple methodologies applied are described below. Systematic Review and Meta-Analysis - Aim: To conduct a systematic review of current Gateway Hypothesis literature. Methods: Longitudinal studies examining empirical data were selected and study characteristics extracted. Results were combined using metaanalysis. Results: Nineteen articles from 14 studies were included reporting 17 drug exposure and outcome combinations. Meta-analysis provided evidence for association of tobacco use (OR=2.60, 95% CI 1.39-4.85, p=0.003) and alcohol use (OR=1.56, 95% CI 1.16-2.10, p=0.003) with cannabis use, and between cannabis use and other illicit drug use (OR=2.79, 95% CI 2.01-3.87, p<O.OOl). Conclusion: There is consistent evidence for association at some stages of the Gateway hypothesis. An influence of confounding, reverse causation and/or bias cannot be ruled out. Recanting and Hair Drug Testing - Aim: To examine the extent of recanting and the potential of hair drug testing as a biological measure of illicit drug use for use in studies of the Gateway Hypothesis. Methods: Hair and self-report measures of cannabis and other illicit drug use were available at age 18 years. Recanting was defined as reporting use at any of the earlier time points and denying ever use at age 18. Recanting and self-reported responses in hair positive individuals were compared on a range of predictors. Previously reported sensitivity and specificity values were used to compare expected numbers of false positives and false negatives. Results: Cannabis and other illicit drug use was recanted by 176 (14.4%) and 99 (25.2%), respectively. Recanting cannabis use decreased with the reporting of other substances and anti-social behaviour. For the hair testing of cannabis use, we would expect to see 0 (95% CI: 0 to 169) false positives and 394 (95% CI: 323 to 449) false negatives, where 362 potential false positives and 436 potential false negatives were observed. Conclusion: Self-reported use of illicit substances could be affected by recanting. Hair is an unreliable measure of substance use in a general population sample. Mendelian Randomisation (MR) - Aim: To use MR to examine the causal effect of on alcohol consumption. Methods: I used rsl051730 as a proxy for heaviness of smoking and smoking cessation. MR methods were used to assess the relationship between: [1] continued smoking and alcohol consumption during pregnancy; and [2] smoking heaviness weekly alcohol consumption (also using data from The Nord-Trondelag Health Study (HUNT) and the Copenhagen General Population Study (GCPS) to increase statistical power). Results: There was little evidence for an association between rsl051730 and alcohol consumption during pregnancy in ALSPAC (OR = 0.95, 95% CI 0.82 to 1.10, P = 0.473) or between rsl051730 and units of alcohol per week in ALSPAC, HUNT or CGPS (SO increase in units of alcohol per additional copy of the risk allele = -0.01, 95% CI -0.04 to 0.01, P = 0.328, 12 = 0.0%). Conclusion: Previously reported associations between smoking and alcohol are likely to have been over-estimated and the result of confounding and/or reverse causation. Baseline models - Aims: To examine associations between tobacco phenotypes in adolescence and alcohol and cannabis use in early adulthood, accounting for evidence of pre-existing problems. Methods: I examined associations between exposures (smoking status, smoking heaviness and tobacco dependence) and outcomes (hazardous drinking, ever use of cannabis and cannabis abuse) taking account of baseline problems. Multiple imputation was utilised to take account of missing data. Results: After taking account of pre-existing problems, there was evidence for association between: smoking status and ever use of cannabis (OR in former smokers = 1.98,95% 1.53 to 2.57; OR in current smokers = 1.66,95% CI 1.13 to 2.44; P≤O.OOl); smoking status and cannabis abuse (OR in former smokers = 1.57, 95% 0.73 to 3.23; OR in current smokers = 2.79,95% CI 1.40 to 5.65; P = 0.009); cigarettes per day and harmful alcohol consumption (OR = 0.91, 95% CI 0.84 to 0.98, P = 0.018); and tobacco dependence and ever use of cannabis (OR = 0.72, 95% CI 0.53 to 0.97, P = 0.030). Conclusion: Previously reported associations between tobacco and alcohol/cannabis might be due, in part, to underlying psychopathology. Latent Class Analysis - Aims: To characterise patterns of cannabis-use across adolescence, and assess their influence on substance in early adulthood. Methods: longitudinal latent class analysis was used to derive trajectories of cannabis use. 1 investigated whether latent class membership predicted subsequent nicotine dependence, harmful alcohol use and other illicit drug use. Results: Cannabis-use patterns were captured as four latent classes: 'non-users' (80.1%); 'late-onset occasional' (14.2%); 'early-onset occasional' (2.3%); and 'regular' users (3.4%). Compared with the non-user class late-onset occasional, early onset occasional and regular cannabis user class had higher odds of nicotine dependence (OR=3.5, 95%CI 0.7-17.9; OR=12.1 95%CI 1.0-150.3; and OR=37.2, 95%CI 9.5- 144.8, respectively); harmful alcohol consumption (OR=2.6 95%CI 1.5-4.3; OR= 5.0 95%CI 2.1-12.1; and OR=2.6 95%CI 1.0- 7.1, respectively); and recent illicit drug use (OR=22.7, 95%CI 11.3-45.7; OR=15.9, 95%CI 3.9- 64.4; and OR=47.9 95%CI 47.9- 337.0, respectively). Conclusion: One-fifth of UK adolescents follow a pattern of occasional or regular cannabis-use. These people are more likely to progress to harmful substance use behaviours in early adulthood. Conclusion - Evidence that smoking does not causally influence alcohol consumption suggests that future interventions targeting both drugs should include a common factor. Strong associations between various other drugs provide further evidence for the Gateway Hypothesis, however, other theories of drug use cannot be ruled out as causality cannot be determined
|
|---|
