Insights on the natural history of problem drug user (PDU) offending

Background. Problem drug use (PDU) is thought to be the cause of a very high level of social and economic costs - the total costs of Class A drug use in England and Wales have been estimated to be over £15 billion per year [3] and PDU accounts for most of this cost. The estimated costs are so high because of the crime that is assumed to be caused by PDU - it has been suggested that well over a half of all recorded acquisitive crime is drug-related [31]. The drug-crime link has been explored in the past by research with people following arrest [32] and work with people in prison [33] to estimate levels, and types, of drug use in these groups of offenders. The simplest argument for the link between drug use and offending in the natural histories of these individuals is that drug users start to commit crime in order to pay for drugs. But, research also points to criminal activity often happening before PDU starts [9,34]. So, the direction of the drugs-crime link is not always clear. In addition, sometimes there is no drugs-crime link at all - not all PDUs are offenders and not all offenders are PDUs [3,33].

Proposed Research. There are likely to be a number of different pathways that could describe the natural history of PDU and offending. We will review research done already on the drug-crime link and on the nature of drug use/ offending pathways. Then, we will use data that have been collected already for administrative purposes and, anonymously, link information about individuals. This will give us data about the crime histories of a very large number of people, which we will use to explore the natural history of their pathways through PDU and offending on a much larger scale than has been previously possible. We will look for characteristics of offending, such as age at onset, type and frequency, which are different for PDUs prior to drug use and non-PDUs. We will further look at whether offending changes after people start PDU, and whether there are factors (for example, the age when drug use started) that may change (moderate or mediate) these relationships. We think this is very important work to do because, despite all the public money that is spent on trying to break the drugs-crime link, there is not much hard evidence about the nature of this link. Our research will lead to a much greater understanding of routes into PDU and offending. This could lead to early identification of, and intervention with, those offenders likely to go on to develop PDU, making it easier to find, and offer help, to PDUs who are involved in crime, or criminals who are at risk of becoming PDUs. For example, the research may pinpoint types of crime or patterns of offending which predict the start of PDU. Tailored intervention targeted at people with this offending profile could interrupt the pathway from crime to established PDU, thereby preventing significant negative health outcomes for the individual and reducing the very high costs to society associated with drug-related crime. This work is also unique - linking existing datasets to look at this particular problem has never been done before on such a large scale in the field of addiction research.

Initial hypotheses about the natural history of PDU offending will be generated via a systematic review of the quantitative and qualitative evidence base. This will access a wide selection of electronic databases, e.g. MEDLINE, EMBASE, PsycINFO etc. A narrative summary will be supplemented by pooling quantitative findings in a meta-analysis should sufficiently similar outcomes be identified.
The systematic review will inform a further hypothesis generating stage involving analysis of linked data within the Drug Data Warehouse (DDW); this resource contains information about substance misusers identified between 2005 and 2009 via treatment and the criminal justice system, linking data from multiple sources with recorded offending histories from the Police National Computer (PNC). Common patterns of offending and PDU status will be explored using predictive modelling and data mining techniques, such as logistic regression (utilising regression shrinkage and variable selection via the lasso [25] method) and Classification and Regression Trees (CART) [26]. Model validation will be based on bootstrap sampling [22,23,28]. This stage will be based on groups of criminally active PDUs (n=67,966) and criminally active non-PDUs (n=163,200) sampled from the DDW on the basis of information about drug tests performed on arrest.
Longitudinal modelling, such as General Linear Mixed Model (GLMM), will describe the rate of offending for PDUs and non-PDUs over an offending life-course. This work will examine whether the type, and rate, of offending changes following PDU onset. The model will be expanded by including factors that might mediate or moderate these relationships; such as age at start of drug use, polydrug use, ethnicity, or gender.

Non academic impact: Work carried out, to date, in this field has been based on several assumptions about the nature of the drug-crime link. These assumptions have dominated thinking in public policy, but in the absence of robust data and large-scale quantitative analysis to justify common assumptions.
The findings of our work will be highly relevant to government policy in this field. Recent years have seen a huge investment in treatment and interventions designed to limit the assumed crime impact of problem drug use (PDU). Policy making in this field requires a more robust evidence base. It is therefore highly desirable to make best use of the available data to explore the natural histories of PDU and crime and to determine whether predominant precursors of offending and/ or PDU can be identified.
We already have well established links with the main governmental organisations that formulate policy in this area and will use these links to deliver briefings and seminars to inform their work. We will also make lay summaries available to drugs and other relevant professionals and drug user-organisations.
We will use available routes to workshop/ communicate our findings to members of the public via the University of Manchester's Public Engagement in Science activities and the University's Widening Participation activities.

Economic & societal impact: Estimates suggest that PDU incurs very high social and economic costs, with the bulk of the costs being associated with acquisitive crime. In addition, the negative health and social effects of PDU impact severely on society, communities and individuals, including elevated mortality and poor quality of life. Developing a better understanding of the natural histories of PDU and offending could lead to early identification of, and intervention with, those offenders likely to go on to develop PDU. For example, by identifying types, or patterns, of offending which predict the onset of PDU and/ or which enable PDUs to be targeted for intervention more effectively. Tailored intervention targeted at individuals with a particular offending profile could interrupt the pathway from crime to established PDU, thereby preventing significant negative health and social outcomes for the individual and reducing the societal costs associated with crime; even relatively modest changes could lead to significant economic impact due to the very substantial costs involved. In addition, the research could be used to leverage improvement in the effectiveness and cost effectiveness of interventions targeted at specific types of offender, leading to better outcomes and using existing resources more efficiently. Corroboration of the assumption that crime is increased following, and fuelled by, PDU-onset would support continued investment in initiatives designed to reduce PDU-related crime.