Individual and population risk of drug use among adolescents attending an English Youth Offending Team: An epidemiological approach

Abstract

This study aims to show how epidemiological methods can help service providers and policy makers to understand how risk factors for drug use differ at individual and population levels. The study sample comprised 180 young people referred to the Wolverhampton Youth Offending Team, of whom 99 (55%) had never used drugs, 35 (19%) used drugs occasionally (less than weekly) and 46 (26%) used drugs regularly. The probability of an individual's drug use was determined for each of the 64 combinations of the six risk factors in the final multivariate model. An individual with all six risk factors has a 78% chance of being a regular user, a 21% chance of being an occasional user and a 0% chance of being a non-user. These findings do not require assumptions regarding causality; they simply express the relationship between respondents' characteristics and their probability of being a drug user. Policy makers need to be able to combine this information with the relative occurrence of risk factors in the population. Furthermore, they need to be able to make an assessment of the causal status of risk factors in relation to drug use. Variables such as familial drug use are high risk factors for individuals (adjusted OR: 22), but are relatively uncommon in the population (6%). In contrast, current smoking is a relatively weak risk factor for individuals (adjusted OR: 4), but occurs more frequently in the population (32%). After adjusting for multiple risk exposures using the appropriate formulae, the population attributable risk (PAR) for familial drug use is 27% compared to 64% for smoking. While we advocate caution in relation to causality, the current analyses provide the first estimates of how exposure to risk can predict individual and population risk of drug use in a vulnerable population. Given the extensive literature on risk factors for drug use, it should be possible to apply epidemiological methods to other data sets in the field of drug addiction.