Participating OMs will be surveyed before and after training in each condition of the study to assess attitudinal factors and factors influencing implementation of screening and brief intervention procedures.
Attitudes will be assessed via the shortened Alcohol and Alcohol Problems Perception Questionnaire (SAAPPQ) . A list of all OMs that can deliver alcohol screening and brief intervention in each study site will be compiled. A self-administered SAAPPQ will be distributed to participating staff on three occasions: pre- and post-training and post-study. SAAPPQ has five subscales covering role adequacy, role legitimacy, self-esteem, motivation, and work satisfaction. Role adequacy and role legitimacy are concerned with role security, i.e., how individuals perceive the adequacy of their skills and knowledge in relation to problem drinkers and how appropriate it is for them to work with such offenders. The subscales relating to self esteem, motivation and work satisfaction, are concerned with worker's therapeutic commitment, i.e., the extent to which they seek to engage drinkers in treatment and the extent that they find the work rewarding on both a professional or personal level .
In addition to the SAAPPQ, the post-training and post-study questionnaires will contain a number of semi-structured and open questions developed to elicit information on staff attitudes towards alcohol screening and brief intervention; previous experience of delivering alcohol screening and brief intervention; readiness to undertake these activities; the training needed to conduct screening and brief intervention; the suitability of each site to provide SBI; and potential barriers to effective implementation.
Factors relevant to implementation of screening and brief intervention have been found to be divided into predisposing, enabling and reinforcing factors . Predisposing factors relate to Offender Managers' willingness to implement screening and brief intervention. Enabling factors are the skills and resources needed to implement screening and brief intervention. Reinforcing factors are visible results, feedback from peers and patients and other factors that encourage continuation of screening and brief intervention.
In this study we will modify the above  to survey Offender Managers before and after training and compare the above factors between different implementation models.
The research team will identify the total number of offenders aged over 18 years who attended appointments with the Offender Managers in the trial during the recruitment period, the total number of offenders screened, the number screening positive and the number receiving an alcohol intervention in each of the 3 implementation models. This will allow calculation of the overall screening rate, the screen conversion rate (proportion of positive screens) and the intervention rate in the different settings. We will also compare these measures and the FAST versus M-SASQ screening tools.
Offenders' re-attendances at probation offices over the 6 and 12 month follow-up periods will be assessed using computerised records and compared with attendances by participating offenders in the 6 and 12 months before entry into the study. The sustainability of the screening and intervention approaches will be assessed by examining the extent to which screening and intervention activity continues after the end of the formal study recruitment period.
Immediately before receiving the initial CIL and/or brief advice intervention, participants will be invited by the Offender Manager to provide contact details and complete the Extended AUDIT , Euroqol (EQ-5D) , a short service use questionnaire (S-SUQ)  and modified Readiness Ruler . Participants in the extended intervention will complete the baseline at the same stage as those in other groups.
AUDIT is normally used as a screening test for alcohol use disorders . However in this context the AUDIT will be used as a means of establishing the severity of alcohol use disorders at baseline, in a way that is least intrusive to naturalistic aim of the trial in the probation setting and as a means of measuring the adequacy of matching between the intervention groups at baseline. The AUDIT contains 10 items to measure alcohol consumption, alcohol problems and dependence over, in this case, the previous 6 months, and the sum of the item scores provides a measure of severity which has been used in several previous studies, allowing comparability with other primary care samples . We felt that the use of more elaborate baseline alcohol consumption measures would interfere with the naturalistic aims of the study and possibly would contribute a form of intervention in itself, so introducing bias into the evaluation of the interventions by reducing the difference between trial interventions. In addition, participants will complete the EQ-5D as a brief 5-item measure of quality of life . Use of health, social criminal justice services and wider societal costs will be measured via a shortened version of Service Use Questionnaire  which allows estimation of health care and wider social costs for health economic analysis in the six months prior to intervention. A modified Readiness Ruler  containing a zero to ten scale of the extent to which participants think about their drinking as a problem or have addressed this issue will assess participants' motivational state regarding changing their drinking behaviour.
At 6 and 12 months after intervention, all participants will be contacted via telephone, post or email as preferred by research staff that will be blind to their intervention condition. Participants will be offered telephone, postal or face-to-face follow-up as preferred. Researchers will administer the shorter Alcohol Use Disorders Identification Test (Extended-AUDIT) . Alcohol-related problems will be assessed via the brief Alcohol Problems Questionnaire (APQ) . We will also re-administer an extended version of the Service Use Questionnaire SUQ , EQ-5D  and the modified Readiness Ruler . Participant satisfaction with the advice/help received during the intervention will be assessed using a modified version of the Patient Satisfaction Questionnaire (short form) at 12 months .
At follow-up, each participants will also be asked if, and how often, they made use of the Drink-Line telephone number.
We propose to incentivise research participation in one of two ways dependent on individual offices choice. They will either receive via payments of £1,000 to each probation site geographical area subject to successful offender recruitment or the participating OMS will receive £20.00 of high street vouchers for each eligible case recruited.
Participants will receive a £10 retail voucher after completing the baseline research interview and a £10 voucher for completing each of the 6 and 12 month research follow-up interviews.
The economic component of the study comprises a cost-effectiveness and cost-utility analysis. The study aims to identify, quantify and value resources related to alcohol screening and intervention by OMs in probation and the subsequent use of health, social care, and criminal justice services by offenders following each type of intervention. Resources utilised in the identification and brief intervention delivery or control condition will be recorded by OMs involved on an ongoing basis. This will allow the calculation of costs related to implementation of different models of screening and brief intervention. Local costs will be used to calculate the costs of the interventions, which will include staff costs, premises costs and costs of leaflets and other consumables. In addition, specific training costs for staff will be calculated, in terms of staff time, premises costs and the cost of training materials.
Offenders' self-reported use of health, social care and criminal justice services will be identified retrospectively using a short form of the service use questionnaire previously used to evaluate costs associated with interventions for alcohol use disorders  and applying a common set of national unit cost estimates. The economic analysis will calculate the incremental cost-effectiveness of the control condition with the AHW condition under study, using measures of clinical outcome and quality of life, EQ-5D  responses at baseline and at 6 and 12 months follow up. The use of EQ-5D enables the estimation of Quality Adjusted Life Years. Data will be bootstrapped to account for the expected skewness evident in economic cost data . The analysis will include the construction of cost-effectiveness acceptability curves to illustrate the probability that the brief intervention is more cost-effective than usual care, based on different monetary values being attached to QALYs. The use of QALYs follows the recommendations of NICE and enables the value for money afforded by treatment to be compared to a range of other health care interventions. Furthermore, combination of the economic cost data and outcome data with offender data collected in the trial will enable a secondary analysis of various offender characteristics that may influence the cost-effectiveness of the intervention.
Cost effectiveness of crime
The economic analysis currently is designed to consider cost effectiveness in terms of incremental cost per QALY. However, one of the key outcomes of interest to criminal justice providers is re-offending. We will therefore use these additional data to undertake subsidiary cost-effective analysis and consider whether the choice of principal outcome measure alters the recommended decisions. The two outcomes chosen will be number of offences and time to failure and the analysis will be conducted on the sample participants with both QALY and CJS data.
Sample size calculation
The sample size calculation is designed to account primarily for intervention level outcomes. Powering the study in this way will also account statistically for appropriate outcomes for screening approach and screening method. The primary outcome for this study is the proportion of offenders who are consuming alcohol within recommended levels at 6 and 12 month follow-up. Recent meta-analysis  suggests that the difference between brief intervention and control is of the order 13%, 5% reduction in the control group and 18% in the brief intervention group. In order to detect a difference of this magnitude at the 5% significance level with 80% power, for a 2-sided test, requires 109 offenders in each of the 3 groups, a total of 327. Our experience with other multi-centre randomised controlled trials of interventions for alcohol use disorders suggests that with assiduous follow-up the potential loss to follow-up across groups is of the order 25%. Taking this loss into accounts inflates the sample required to 145 in each group, a total of 435 offenders.
The proposed study involves a cluster design and requires a statistical adjustment to account for any potential cluster effect. The literature is unclear regarding an appropriate estimation of an intra-class correlation coefficient in this population. Our experience leads us to assume the ICC should be similar to primary care correlation coefficients, 0.04. Assuming a cluster size of the order 3 offenders this inflates the sample size calculation by a factor of 1.1 requiring 160 offenders in each group, a total of 480, with an expectation that at least 365 will be followed up at 12 months.
As the study is pragmatic in design, the planned analysis will be by intention to treat. The primary outcome is dichotomous in nature, drinking within or above recommended levels, and will be analysed with logistic regression adjusting for all known prognostic factors; data will be presented as odds ratios and their corresponding confidence intervals. Secondary analyses will be undertaken using the appropriate method for the outcomes, controlling where appropriate for intake values and other known prognostic variables using analysis of covariance.
Due to the nested factorial nature of the study, we will use multi-level modelling to explore potential interactions between each of the levels nested within the trial (screening method and intervention). Offender manager and offender factors will be utilised as part of regression model to explore possible prognostic factors that impact on outcome. Interaction analysis will explore any possible interactions between site and offender characteristics and outcome.
The efficacy of interventions will be explored using a per protocol approach to the analysis. A sub-group, of those who received their allocated treatment will be utilised for this analysis.
Ethical and Research Governance Approval
The study has been granted ethical approval by multi-centres research ethics committee (MREC reference number: 08/H0903/2).
The trial duration is 30 months and it commenced in May 2008