OR15-1 * THE GATEWAY HYPOTHESIS, COMMON LIABILITY TO ADDICTIONS OR THE ROUTE OF ADMINISTRATION MODEL? AN ORIGINAL MODELLING PROCESS LINKING THE THREE THEORIES

Abstract

Background. The aim of this study was to describe the patterns of transition across tobacco, cannabis and other illicit drugs (OID) use by investigating simultaneously the Gateway theory (GT), the Common liability model (CLM) and the Route of administration model (RAM). Method. A French nationwide retrospective cohort on drug use was reconstituted from the reports of 1,063 young adults on their lifetime experience with tobacco, cannabis and OID. A multi-state model was fitted, modelling all possible pathways from no use to initiation of the three substances. This model was adjusted on drug-use liability, a latent variable constructed with an item response theory model and reflecting early onset, high levels of use and poly-substance use. Results. The most likely onset sequence was the gateway sequence: no use-tobacco-cannabis-OID. The liability variable was mainly associated with the beginning of the sequence whatever the substance used first: an increase of 1 for this variable was associated with a 1.9 [1.7-2.0] times greater likelihood of starting the sequence with tobacco, a 3.0 [2.3-3.9] times greater likelihood of starting with cannabis and a 5.1 [2.1-125.0] times greater likelihood of starting by OID. The liability variable was also associated, for subjects who had not yet initiated OID, with the initiation of cannabis after tobacco and the initiation of tobacco after cannabis. Conclusions. This study, using an original modelling procedure to describe onset sequences, is in line with a substance use process involving simultaneously GT, CLM and RAM. This suggests shared influences of individual and environmental characteristics. The common liability could contribute to initiation of the use sequence, whatever the first substance used. The rest of the sequence seems to follow a gateway pattern (tobacco, cannabis and then OID) in most cases, which could be the consequence of variations in substance availability during the life. Our drug use liability variable was finally associated, for subjects who had not yet initiated OID, with the transitions between tobacco and cannabis, which would suggest reciprocal relationships between these two substances in line with RAM.