Predicting attrition of men with a history of violence from randomised clinical trials
Trials volume 24, Article number: 740 (2023)
Preventing dropout (attrition) from clinical trials is vital for improving study validity. Dropout is particularly important in justice-involved populations as they can be very challenging to engage and recruit in the first instance. This study identifies factors associated with dropout in a double-blind, placebo-controlled randomised control trial of a selective serotonin reuptake inhibitor (SSRI) aimed at reducing reoffending in highly impulsive men with histories of violent offending. Age, education, social support, psychiatric history, and length of previous incarceration were identified as factors that predict attrition. These findings are consistent with previous research examining variables associated with attrition in clinical trials for community and offender populations. We also explored referral source and treatment allocation as attrition predictors. Although neither significantly predicted attrition, we identified that there are discernible differences in the median time to attrition among the referral source subgroups. Understanding factors that predict treatment completion and attrition will allow researchers to identify participants for whom additional provisions may optimise retention and inform development of targeted interventions.
In 2016, 2.2 million adults in Australia were estimated to have been victims of domestic violence by age 15 . For victims of violent crime, the impacts can be profound, including an increased risk of anxiety, depression, post-traumatic stress disorder, and substance use [2,3,4]. Additionally, there is a significant financial burden of such crime, with estimates suggesting the cost of assault, robbery, and homicide in Australia is $1800, $3600, and $1.6 million per incident, respectively  and the cost of domestic violence against women and children to be $20 billion per year . Thus, given the health, social, and economic impacts of violence and domestic violence, research aimed at prevention is vital.
Randomised control trials (RCT) represent the gold standard in medical research . They are posited as necessary to offender studies, particularly pharmacological clinical trials that demonstrate the effectiveness of interventions aimed at offender rehabilitation . In light of this, it might be expected that RCTs are widely used to investigate crime-reduction interventions. However, such studies are relatively uncommon in this field [9, 10].
One reason proposed to account for the lack of RCTs with offenders is that they fall into a group dubbed “hard to treat” by many researchers . This is due, in part, to ethical challenges of conducting research with offenders and their chaotic lifestyles characterised by itinerancy, insecure accommodation, lack of reliable contact methods (e.g. frequent changing of mobile phones or limited internet access), and turbulent social connections [12,13,14]. These factors make research completion particularly challenging, resulting in offender studies having higher rates of attrition (i.e. “1-loss to follow-up”; Brueton, 2014, p.14)  than non-offender studies . Consequently, it has been suggested that aspiring to conduct RCTs in the justice system is too challenging, and lesser levels of evidence should be considered .
Typically, RCTs with offenders report that around half of the sample is lost to attrition. For example, Butler et al., Coccaro et al., Cullen et al., Kraanen et al., Lardén, and Stone et al. report an average attrition rate of 53% [18,19,20,21,22,23]. This is substantially higher than the average attrition rate of 35% reported in community health research [17, 24, 25]. One consequence of high attrition is the negative impact it can have on the validity of clinical trials [26, 27]. Perhaps even more pressing is evidence suggesting offenders who drop out of research interventions and trials are more likely to re-offend, suggesting a need for a better understanding of reasons for dropout from trials in the justice field .
In Australia, those who commit violent crimes such as homicide, assault, sexual assault, and robbery, tend to commit repeat offences [29, 30]. Forty-four percent of individuals convicted of assault and 47% convicted of robbery returned to prison within two years of release . Maximising retention in programmes aimed at reducing reoffending via an improved understanding of factors that predict and affect attrition will be highly useful for reducing violence.
Efforts to predict attrition of violent offenders in clinical trials and treatment programmes have identified “attrition risk factors” among offenders convicted of domestic violence or sex crimes (e.g. sexual assault or child sex offences). These include a lack of social support; a history of alcohol abuse or poor mental health; being younger, unemployed, single, childless, expelled from school, or an ethnic minority; and having low incomes, little formal education, a learning disability, insecure accommodation, dysfunctional expressions of anger, or prior history of criminality [32,33,34,35,36,37,38]. Referral source has been proposed as a risk factor; however, there is a lack of consensus as to whether court-mandated diversion into treatment programmes protects against attrition or is a risk factor [39,40,41].
A limited number of studies have looked at predicting attrition from clinical trials and treatment programmes among individuals who commit violent crimes other than domestic violence and sex crimes (e.g. armed robbery, grievous bodily harm, manslaughter, malicious wounding, or assault). These studies show that impulsivity, depression, and indigenous heritage are associated with attrition [42,43,44]. To our knowledge, only two studies report treating violent populations with pharmacological means [18, 19]. While both of these studies experienced high treatment attrition (41% from Butler’s pilot study and 45% Coccaro’s study with individuals with impulsive aggressive behaviour) [19, 45] little consideration was given to describing factors associated with attrition. Coccaro et al. reported no difference between completers and non-completers regarding impulsivity and aggression scores . Butler et al. suggested that attrition was related to the challenging nature of the client group (i.e. frequently changing phone numbers, transient living arrangements, poor time management, and impulsivity) . Identifying factors that can predict attrition of violent offenders in pharmacological-based trials has significant implications, particularly given the promising results of both studies [19, 45].
This study aimed to identify factors that predict attrition from pharmacological-based treatments. Specifically, we investigated predictors of attrition in a double-blind, randomised, placebo-controlled trial (ReINVEST) by men with histories of repeat violent offending .
The study design, which has been extensively described in previous publications [18, 45], utilises data obtained from the ReINVEST clinical trial. The ReINVEST trial is a two-arm, parallel-group, double-blind, placebo-controlled RCT that was conducted to evaluate the effectiveness of sertraline (a selective serotonin reuptake inhibitor (SSRI)) in reducing reoffending rates among individuals with a history of violent offending.
This study has received ethical approval from the University of New South Wales (HC17771), Aboriginal Health & Medical Research Council (AHMRC; 822/11), Corrective Services NSW (09/26576) and the NSW Justice Health (G8/14) and the ReINVEST study is registered with the Australian and New Zealand Clinical Trials Registry (ACTRN12613000442707).
The study population is comprised of 628 (Mage = 32.25 years; S.D = 9.88) male participants enrolled in the ReINVEST clinical trial. The primary sources of referrals for participant recruitment in the study were Australian local magistrates’ courts, Legal Aid, New South Wales (NSW) solicitors, private lawyers, Corrective Services NSW Community Corrections Officers, and passive recruitment methods (i.e. flyers available at courts and other locations, word-of-mouth, self-referral, a free call number and study website) from October 2013 to June 2021. Medically fit males over the age of 18 with the ability to communicate in English, an impulsivity score (measured by the Barratt Impulsiveness Scale ) of 70 or more, two or more prior convictions for violent offences (excluding homicide or a child sexual offence), no serious mental illness (e.g. schizophrenia), and were not already on an SSRI were eligible for ReINVEST. Eligible participants were offered $50 for the initial screening visit, $20 per subsequent assessment (every 4 weeks), and $10 per visit when collecting medication to cover expenses incurred because of participation.
After screening, eligible participants then underwent a comprehensive psychiatric assessment and medical examination, administration of a number of behavioural measures, and collection of demographic data. A detailed list of all study assessments (including abbreviations) used during the recruitment phase of the ReINVEST clinical trial is available in Butler et al. (2021) . Of these, this study used assessments of factors reported to predict attrition (Table 1). All participants completed a four-week run-in phase when all received the active medication prior to randomisation. During the run-in, all received a daily dose of 100 mg of sertraline taken orally. This was to identify those who react poorly to the medication or are not willing to commit to 12 months follow-up after initially consenting. Participants were then randomised into either the treatment or control arms of the study, with controls receiving an identical-looking placebo tablet. After study completion (12 months) all participants were offered the option to remain in the study or cease protocol treatment. A number of steps were taken to avoid potential sources of bias including employing objective, validated, standardised behavioural measures (Table 1), using an online randomisation system for treatment allocation, and used the STROBE checklist to enhance transparent reporting .
Outcome and definitions
The primary outcome of interest in this study was time to attrition post-randomisation from the ReINVEST trial. The secondary outcomes were socio-demographic, justice-related and mental health factors associated with attrition from ReINVEST trial. Attrition was defined as the termination of a participant's involvement in the trial before its completion. Attrition was assessed based on several reasons, including the occurrence of adverse events, loss to follow-up (three consecutively missed appointments with no further contact), physical or mental health concerns, reincarceration, perceived lack of benefit, and personal choice.
For this study, the follow-up period was defined as the timeline from randomisation (baseline) until the earliest of the following: attrition, death, and completion of the study at 24 months post-randomisation. Although participants could cease protocol treatment at 12 months, 24 months was used to capture those participants who opted to remain in the study beyond 12 months.
The following covariates were included in the present analysis: randomisation status (main exposure), age, Aboriginal status, relationship status, accommodation type, number of children, highest educational level, requiring educational support during school, being expelled from school, history of juvenile violent offending, time spent in juvenile detention, number of violent offences within 5 years prior to randomisation, length of incarceration within 5 years prior to randomisation, psychiatric history, referral source, scores on the DSSI , BIS , BDI-II , K10 , AIAQ , and AUDIT .
Descriptive statistics were used to summarise participant characteristics at randomisation (baseline). The time at risk was calculated as the duration from randomisation until the earliest of the following events: attrition, death, or 24 months post-randomisation. Kaplan–Meier method was used to estimate the cumulative incidence of attrition, overall and stratified by referral source. To identify factors associated with attrition, univariable and multivariable Cox regression models were fitted. Death was considered as a competing risk in this study. However, no participants died during the follow-up period. The multivariate model incorporated all covariates of interest.
The initial multivariate model incorporated all covariates with a P-value of < 0.2 on univariate analysis. Covariates with a P-value of < 0.05 were retained in the final model. Crude and adjusted hazard ratios (aHR) were derived along with their corresponding 95% confidence intervals (95% CI) as measures of association. To evaluate the underlying assumptions and goodness of fit of the Cox proportional hazards regression model, several diagnostic techniques were utilised. The proportional hazards assumption was assessed by analysing Schoenfeld residuals. The cumulative hazards based on Cox-Snell residuals were plotted against the Nelson-Aalen estimate of the cumulative hazard, and the unit slope was visually examined to assess model fit. All statistical analyses were conducted using a two-tailed significance level of 0.05. Data analysis was conducted using Stata version 17 (Stata Corporation, College Station, TX, USA).
The characteristics of the study population at baseline are summarised in Table 2.
Given the interest in the source of referral as an attrition predictor [39, 40], we also assessed the association between referral source and attrition. Those who were referred by a magistrate/legal services tended to exit the study earlier compared to other referral sources (Fig. 1). Cumulative incidence of dropout at 24 months post-randomisation was 50.6% for self/others referral, 62.5% for community corrections referral, and 65.6% for magistrate/legal services referral. The median time to attrition is 11 months (IQR: 3, 19) for self/others, 8 months (IQR: 2, 13) for community corrections and 5 months (IQR: 1, 14) for magistrate/legal services.
Loss to follow-up was the main reason for attrition overall (56.3%) but the proportion was higher in those referred by magistrates/legal services (63.9%) compared to self/other referred (58.8%) and community corrections referred (47.0%). Offence-related reasons for attrition overall were 9.7%, magistrate/legal service referred (5.8%), self/other referred (11.8%) and community corrections referred (13.9%). Unwanted side effects accounted for the least numbers in terms of attrition overall (3.3%), magistrate/legal services referred (4.7%), self-referred (0%) and community corrections referred (2.4%).
The multivariable analysis identified the following predictors associated with attrition (Table 3): age, educational attainment, incarceration history, psychiatric history, and DSSI score.
Being older, higher levels of education, and higher levels of social support serve as protective factors against attrition (Table 3). A psychiatric history requiring medication or hospital admission reduced the risk of dropout, and a smaller cumulative duration of incarceration is associated with dropout.
The current study explored factors associated with attrition from a clinical trial involving pharmacotherapy aimed at treating impulsive individuals with a history of violent offending. Specifically, we investigated whether attrition from the ReINVEST trial could be predicted by demographic and psychosocial factors identified as attrition risk factors from the literature. Our results found that older age, higher levels of social support, higher levels of education, and psychiatric history were all protective factors against attrition at 24 months, whereas less time spent incarcerated in the past 5 years is associated with dropout. We found no association between referral source or treatment allocation on attrition.
Our findings that older age is associated with decreased attrition at 24 months are in line with past studies [61,62,63]. Sampson and Laub’s  (p.37) revised age-graded theory proposes that older offenders’ ability to display “purposeful execution of choice” is essential in criminal desistance. This purposeful choice execution may also explain older adults’ increased likelihood to remain in a trial perceived as being helpful in desistence. Neuroimaging studies show that older adults, compared to younger counterparts, perform better on decision-making tasks when previous choices influence an outcome or reward . According to Grossman et al. older people are better at making decisions requiring higher-order processing of relational dependencies between recent choices and the available rewards . Concerning retaining offenders in trials and programmes, older participants may be more likely to consider the consequences of previous anti/pro-social choices when deciding whether to drop out. Tailoring retention methods to account for the impact of age on decision-making may support retention more effectively. The NSW Behavioural Insights Unit (BIU)  (2018, p.15), recommends adopting a “bespoke approach adapted for offenders” that takes age into consideration. For those under 30 years, they recommend making contact early and emphasising benefits, such as programme completion being regarded in upcoming sentencing. For older offenders, who often report being “weary of prison, but are equally distrustful of programs” (BIU, 2018, p.15) they suggest offering clear information on how the trial can offer hope and help them reclaim a sense of control over their life. Qualitative research with participants and research staff exploring how age-dependant difference in decision-making impacts attrition from clinical trials is recommended.
Our finding that increased social support is protective against attrition is also in line with past research . One explanation is that social support alleviates the impact of other attrition risk factors (e.g. poor mental health) [69, 70]. For example, being able to depend on others for emotional support reduces depressive symptoms, which have been linked to attrition . Past research has linked attrition to the number of supports offenders have . However, it is likely the quality, rather than the number, of social support that is important . Socioemotional Selectivity Theory suggests that social network satisfaction decreases adverse lifestyle fluctuations , likely enhancing the stability needed to remain in trials and programmes.
Adopting a holistic approach when designing RCTs that foster quality in participants' social networks would enhance study retention. For example, employing case management or peer support strategies for those who report low social support at baseline may minimise dropout and enhance successful rehabilitation. Klaehn et al. (2022, p.294)  have demonstrated that case management is a promising “cost-effective, or even cost-saving” method to support retention of participants with complex needs, such as offenders. Many offenders are under judicial supervision (e.g. probation officers or Community Safety Case Managers). Collaboration between existing supervision providers and research teams offers the optimal chance to meet the shared goal of rehabilitation. It also offers justice-health researchers a unique opportunity to harness existing case-management networks. Further research exploring the impact of case-management collaboration and peer support strategies on the retention of people participating in justice system trials and programmes is suggested.
Our finding that lower levels of education are associated with increased attrition is consistent with other research [76,77,78,79,80,81,82,83]. There are several possible explanations why education is related to attrition. For example, higher levels of education are associated with increased cognitive function in areas such as attention, memory, and problem-solving skills  and enhanced ability to manage complex tasks . It may be that individuals with higher levels of education are better able to manage the requirements of participation, whereas lower levels of education may be associated with the perceived complexity of the trial procedures, language, and concepts, leading to feelings of being overwhelmed. This sense of overwhelm, known as cognitive overload, may be one factor that drives the relationship between low education and attrition. Continued investigation into the relationship between cognitive overload and attrition may help elucidate this relationship further. Our findings suggest that providing additional support to participants who report lower levels of education at baseline may be an approach to minimising attrition.
We also found that participants who had spent less time incarcerated in the 5 years prior to randomisation had an increased rate of attrition. While this finding contradicts Zanis et al.’s (2009)  report that offenders whose most recent incarceration was longer were less likely to complete treatment, it is worth noting that their study was conducted with offenders completing community-based drug treatment rather than community offenders aiming to reduce impulsive violent behaviour. Since attrition is the result of an interaction between a programme and its participants , the lack of consistency of this predictor of attrition is not entirely surprising. It does, however, emphasise that a “one size fits all” approach should be avoided when attempting to reduce offender attrition. Whether incarceration length serves as a predictor of dropout may differ depending on the nature of offending behaviour or programme management and delivery.
Zapryanavo’s (2020)  report that each month spent incarcerated results in a 1.12 percentage decrease in the probability of reoffending while on parole offers context for our findings on incarceration and attrition. It may be that more lengthy experiences of incarceration leave participants determined to avoid recidivism; they perceive trial participation as a valuable way to avoid future incarceration. In contrast, individuals who have spent less time incarcerated may perceive less need for intervention, leading to early dropout. Alternatively, it could be that those who have spent less time in prison have more opportunities to become engaged in community activities such as employment, education, or family commitments, limiting available time to be involved in the trial.
Our findings also indicate that those with a history of psychiatric conditions requiring treatment, medication, or hospital admission are significantly less likely to drop out. This is contrary to past research suggesting that psychiatric disorders predict the attrition of violent offenders from programmes [87, 88]. Hoverer, those studies were reporting on dropout from behavioural interventions employing cognitive-behaviour therapy conducted in a group setting, rather than a clinical trial. Dishion et al. (1999)  report some individuals show increases in criminality after group interventions, possibly because of peer reinforcement of criminal thinking and behaviour. Reinforcement processes may also underlie attrition from group therapy. In contrast, being involved in a pharmacological trial delivered one-on-one with research staff offers participants access to mental health clinicians that they might not otherwise have access to, maintaining privacy and confidentiality, and avoiding reinforcement of undesirable behaviours. Additionally, participating in a clinical trial such as ReINVEST might be viewed as less stigmatising, less confronting, and less anxiety provoking then attending group-based behaviour therapy.
Our finding that referral source does not predict attrition is consistent with past studies that also report a null effect regarding referral source [90,91,92] and contradicts reports that court referral supports retention [93, 94]. It is often suggested that participation in trials like ReINVEST, which had a policy developed enabling referral from the bench by a magistrate, is motivated by a simple desire to avoid prison . The fact that we saw the lowest proportion of attrition among those who were self-referred and over half (51%) of those enrolled were serving community sentences and not facing the prospect of prison suggests motivation for participation is more than a simple desire to avoid prison. It is also patronising to suggest that this is the prime and only motivation of this group. Indeed, a pre-study survey  of prisoner's potential willingness to join a trial such as ReINVEST found that over half the respondents were interested in learning more about the study, and of those respondents, 80% said they would still take part despite a 50% chance of receiving a placebo. This suggests that these offenders have insight into the negative impact of their anti-social behaviour, and they are interested and willing to be involved in treatment to reduce problematic violence.
If a relationship between referral source and attrition exists, it is possible the relationship is mediated by other factors. For example, research on why offenders dropout of psychological treatments indicates that younger, less educated men are more likely to remain in treatment if they are court mandated. In contrast, older, better-educated men may be more likely to remain in treatment if they are not court-ordered . Further investigations into an interaction between referral source and ReINVEST attrition variables may help clarify whether a referral source and attrition relationship moderated by age and education in psychological treatments also exist in clinical trials.
Interestingly, we did not see an effect of treatment allocation on attrition. While this finding is consistent with Coccaro  who also report no effect of treatment condition on attrition from their fluoxetine/placebo trial, it contradicts reports of past RCTs that note attrition differences among treatment allocation groups [97,98,99]. However, these studies describe attrition from prison-based behaviour therapies (e.g. yoga, cognitive behaviour therapy, and exercise) rather than pharmacotherapy trials. Olver et al.  report attrition is impacted by treatment modality, with the highest attrition observed in behavioural therapy programmes. Our findings, and those of Coccaro  call to question suggestions that offering a placebo can impact attrition by introducing uncertainty sufficient to decrease the magnitude of response or motivation to remain engaged [99, 100].
It is possible that the unique nature of ReINVEST as a community-based trial that offered free access to follow-up appointments with research clinicians and a psychiatrist influenced retention, regardless of treatment allocation. It allowed participants, irrespective of their treatment allocation, to receive ongoing professional healthcare attention and support. This could be appealing, given services can be difficult to access and costly outside of the trial setting. Further, regular appointments provided an opportunity for participants to discuss any concerns or adverse effects, and to feel heard and validated. This sense of support may have enhanced motivation to remain in the trial regardless of treatment allocation. It is possible that for control participants regular appointments with clinicians and a psychiatrist reinforced any experienced placebo effect. The positive psychosocial benefits of attention and care offered as part of an effective treatment team may have been perceived as an improvement in their condition. This suggests that providing offenders access to frequent, consistent, high-quality healthcare can support retention in trials. Further, offenders may be more likely to remain committed to the study, regardless of treatment allocation, if offered empathetic communication that ensures they feel observed and cared for.
Previous research indicated that Aboriginal status, relationship status, number of dependents, accommodation type, impulsivity, anger, irritability, and alcohol abuse predict offender dropout from research interventions and clinical trials [63, 101, 102]. Despite this, we found no support for these factors as predictors of attrition in the ReINVEST trial. Null findings in a study exploring attrition can be beneficial information to have. It may indicate that researchers conducting pharmacological trials do not need to be overly concerned that these participant characteristics will bias outcomes. However, the unique nature of our sample (all were highly impulsive men with histories of violence) may explain why our findings differ from past studies.
A limitation of this study is it is based on a sample of highly impulsive repeat-violent offenders. Additionally, reliance on self-report to ascertain the impact of some of the variables on attrition (e.g. relationship status, number of dependents, accommodation type, impulsivity, anger, irritability, and alcohol abuse) may be subject to recall or response bias. It is possible that individuals who have a history of committing violent offences may feel the need to manage their public identity and may have modified their responses. Notwithstanding these limitations, this study also possesses considerable strengths. We have identified age, social support, education, time spent incarcerated, and psychiatric history are associated with attrition.
The problem of offender attrition from clinical trials and programmes is ongoing and a serious concern given it is linked to recidivism. Identifying ways to reduce the attrition from clinical trials of offenders who commit violent crimes is vital. This study identified a number of risk factors that support researchers to predict attrition of men with a history of violence from clinical trials. Awareness of these factors can support researchers working with samples that are justice-involved to determine, at baseline, participants for whom additional provisions may optimise retention.
Availability of data and materials
Records management is guided by systems implemented by the NHMRC Clinical Trials Centre, in line with requirements for record keeping in the NSW private sector, medico-legal requirements and the NSW State Records Act 1998. Trial documents are stored under medical records conditions for 15 years post publication in the secure UNSW Shared File storage, with access restricted to the appropriate researchers who have access to UNSWnetworks. On the case report forms and other documents, a unique study code is used to identify the records. The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Australian Bureau of Statistics. Personal Safety Survey, Australia, 2016. Canberra: ABS; 2017.
Aizer A, Doyle JJ. Juvenile incarceration, human capital, and future crime: Evidence from randomly assigned judges. Q J Econ. 2015;130(2):759–803.
Brown SE, Esbensen F, Geis G. Criminology: explaining crime and its context. New York: Routledge; 2018.
Gann S, Sullivan C, Ilchi O. Elaborating on the effects of early offending: a study of factors that mediate the impact of onset age on long-term trajectories of criminal behavior. J Dev Life Course Criminol. 2015;1(1):63–86.
Mayhew P. Counting the Costs of Crime in Australia. Trends Issues Crime Crim Justice. 2003;247:1.
PricewaterhouseCoopers (PwC). A high price to pay: The economic case for preventing violence against women. Melbourne: Victorian Health Promotion Foundation; 2015.
Harper G, Chitty C. The impact of corrections on re-offending: a review of ‘what works.’ London: Home Office Research Study; 2005.
Prichard J, Payne J. Alcohol, drugs and crime: a study of juveniles in detention. Canberra, ACT: Australian Institute of Criminology; 2005.
Kouyoumdjian FG, McIsaac KE, Liauw J, et al. A systematic review of randomized controlled trials of interventions to improve the health of persons during imprisonment and in the year after release. Am J Public Health. 2015;105(4):e13–33.
Lennox C, Leonard S, Senior J, et al. Conducting Randomized Controlled Trials of Complex Interventions in Prisons: A Sisyphean Task? Front Psychiatry. 2022;13: 839958. https://doi.org/10.3389/fpsyt.2022.839958.
Brunner F, Neumann I, Yoon D, et al. Determinants of dropout from correctional offender treatment. Front Psychiatry. 2019;10:142.
DeVall KE, Lanier CL. Successful completion: An examination of factors influencing drug court completion for White and Non-White male participants. Subst Use Misuse. 2012;47(10):1106–16.
Goshin LS, Byrne M. Predictors of post-release research retention and subsequent reenrollment for women recruited while incarcerated. Res Nurs Health. 2012;35(1):94–104.
Hamilton M. Risk-needs assessment: Constitutional and ethical challenges. Am Crim Law Rev. 2015;52:231–91.
Brueton VC. Strategies to improve retention: Effectiveness and use in randomised trials [dissertation]. [London, England]: ProQuest; 2015. https://discovery.ucl.ac.uk/id/eprint/1460522/.
Olver ME, Wong SCP. Treatment programs for high risk sexual offenders: Program and offender characteristics, attrition, treatment change and recidivism. Aggress Violent Behav. 2013;18(5):579–91.
Tilley N. What’s the “what” in “what works?” Health, policing, and crime prevention. In: Knutsson J, Tilley N, editors. Evaluating crime reduction initiatives. Monsey, NY: Criminal Justice Press; 2009. p. 121–46.
Butler T, Schofield PW, Knight L, et al. Sertraline hydrochloride for reducing impulsive behaviour in male, repeat-violent offenders (ReINVEST): protocol for a phase IV, double-blind, placebo-controlled, randomised clinical trial. BMJ Open. 2021;11(9):e044656.
Coccaro E, Lee R, Kavoussi R. A double-blind, randomized, placebo-controlled trial of fluoxetine in patients with intermittent explosive disorder. J Clin Psychiatry. 2009;70(5):48–53.
Cullen AE, Soria C, Clarke AY, Dean K, Fahy T. Factors predicting dropout from the reasoning and rehabilitation program with mentally disordered offenders. Crim Justice Behav. 2011;38(3):217–30.
Kraanen FL, Scholing A, Emmelkamp PMG, et al. The comparative effectiveness of Integrated treatment for Substance abuse and Partner violence (I-StoP) and substance abuse treatment alone: a randomized controlled trial. BMC Psychiatry. 2013;13(1):189.
Lardén M. Serious and violent offending: psychological risk factors and treatment effectiveness [dissertation]. [Stockholm, Sweden]: ProQuest; 2021. https://www.proquest.com/dissertations-theses/serious-violent-offending-psychological-risk/docview/2566223818/se-2.
Stone A, Jason LA, Stevens E, Light JM. Factors affecting the stability of social networks during early recovery in ex-offenders. Am J Drug Alcohol Abuse. 2014;40(3):187–91.
Booker CL, Harding S, Benzeval M. A systematic review of the effect of retention methods in population-based cohort studies. BMC Public Health. 2011;11(1):249.
McMurran M, Huband N, Overton E. Non-completion of personality disorder treatments: A systematic review of correlates, consequences, and interventions. Clin Psychol Rev. 2010;30(3):277–87.
Watt K, Shepherd J, Newcombe R. Drunk and dangerous: a randomised controlled trial of alcohol brief intervention for violent offenders. J Exp Criminol. 2007;4(1):1–19.
Shrier I, Boivin JF, Steele RJ, et al. Should meta-analyses of interventions include observational studies in addition to randomized controlled trials? A critical examination of underlying principles. Am J Epidemiol. 2007;166(10):1203–9.
Mossière A, Serin R. A critique of models and measures of treatment readiness in offenders. Aggress Violent Behav. 2014;19(4):383–9.
Yukhnenko D, Sridhar S, Fazel S. A systematic review of criminal recidivism rates worldwide: 3-year update. Wellcome Open Res. 2019;4:28.
Ng S. Australian crime: Facts and figures 2013. Canberra: Australian Institute of Criminology; 2014.
Fuller G. The serious impact and consequences of physical assault. Trends Issues Crime Crim Justice. 2015;496:1–8.
Anaforian CL. Predicting treatment completion of adolescent sex offenders in a community outpatient setting [dissertation]. [Place unknown]: ProQuest; 2009.
Buttell FP, Powers D, Wong A. Evaluating predictors of program attrition among women mandated into batterer intervention treatment. Res Soc Work Pract. 2012;22:20–8.
Gover A, Jennings W, Davis C, Tomsich E, Tewksbury R. Factors Related to the Completion of Domestic Violence Offender Treatment: The Colorado Experience. Victims Offenders. 2011;6(2):137–56.
Henwood KS, Chou S, Browne KD. Corrigendum to “A systematic review and meta-analysis on the effectiveness of CBT informed anger management.” Aggression Violent Behav. 2016;27:121.
Larochelle S, Diguer L, Laverdière O, Greenman PS. Predictors of psychological treatment noncompletion among sexual offenders. Clin Psychol Rev. 2011;31(4):554–62.
Tzoumakis S, Lussier P, Blanc M, Davies G. Onset, Offending Trajectories, and Crime Specialization in Violence. Youth Violence Juvenile Justice. 2013;11(2):143–64.
Willis GM, Ward T. Striving for a good life: The good lives model applied to released child molesters. J Sex Aggression. 2011;17(3):290–303.
Coviello D, Zanis D, Wesnoski S, Palman N, Gur A, Lynch K. Does mandating offenders to treatment improve completion rates? J Subst Abuse Treat. 2013;44(4):417–25.
Rivera D, Dueker D, Amaro H. Examination of referral source and retention among women in residential substance use disorder treatment: a prospective follow-up study. Subst Abuse Treat Prev Policy. 2021;16(1):21.
Carl LC, Schmucker M, Lösel F. Predicting attrition and engagement in the treatment of young offenders. Int J Offender Ther Comp Criminol. 2020;64(4):355–74.
Listwan SJ. Reentry for Serious and Violent Offenders: An Analysis of Program Attrition. Crim Justice Policy Rev. 2009;20(2):154–69.
O’Brien K, Daffern M. An exploration of responsivity among violent offenders: Predicting access to treatment, treatment engagement and programme completion. Psychiatry Psychol Law. 2017;24(2):259–77.
Polaschek D. Treatment non-completion in high-risk violent offenders: Looking beyond criminal risk and criminogenic needs. Psychol Crime Law. 2010;16(6):525–40.
Butler T, Schofield PW, Greenberg D. Reducing impulsivity in repeat violent offenders: an open label trial of a selective serotonin reuptake inhibitor. Aust N Z J Psychiatry. 2010;44(12):1137–43.
Barratt E, Stanford MS. Impulsiveness. In: Costello CG, editor. Personality Characteristics of the Personality Disordered. New York: Wiley; 1995. p. 91–118.
Erik von, Elm Douglas G, Altman Matthias, Egger Stuart J, Pocock Peter C, Gøtzsche Jan P, Vandenbroucke. Strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. BMJ. 2007;335(7624):806–8. https://doi.org/10.1136/bmj.39335.541782.AD.
Stanford MS, Mathias CW, Dougherty DM, Lake SL, Anderson NE, Patton JH. Fifty years of the Barratt Impulsiveness Scale: an update and review. Personality Individ Differ. 2009;47(5):385–95.
Reid RC, Cyders MA, Moghaddam JF, Fong TW. Psychometric properties of the Barratt Impulsiveness Scale in patients with gambling disorders, hypersexuality, and methamphetamine dependence. Addict Behav. 2014;39(11):1640–5.
Koenig HG, Westlund RE, George LK, Hughes DC, Blazer DG, Hybels C. Abbreviating the Duke Social Support Index for use in chronically ill elderly individuals. Psychosomatics. 1993;34(1):61–9.
Wardian J, Robbins D, Wolfersteig W, Johnson T, Dustman P. Validation of the DSSI-10 to Measure Social Support in a General Population. Res Soc Work Pract. 2013;23(1):100–6.
Beck A, Steer R, Ball R, Ranieri W. Comparison of Beck Depression Inventories-IA and-II in psychiatric outpatients. J Pers Assess. 1996;67(3):588–97.
Wang Y, Gorenstein C. Psychometric properties of the Beck Depression Inventory-II: a comprehensive review. Rev Bras Psiquiatr. 2013;35(4):416–31.
Kessler RC, Andrews G, Colpe LJ, Hiripi E, Mroczek DK, Normand SL, et al. Short screening scales to monitor population prevalences and trends in non-specific psychological distress. Psychol Med. 2002;32(6):959–76.
Blanc S, et al. How much distress is too much on deployed operations? Validation of the Kessler Psychological Distress Scale (K10) for application in military operational settings. Mil Psychol. 2014;26(2):88–100.
Silva BF, Santos-Vitti L, Faro A. Kessler Psychological distress scale: internal structure and relation to other variables. Psico-USF. 2021;14(26):91–101.
Coccaro EH, Kupsaw-Lawrence E, Bernstein D. Anger, irritability, and assault questionnaire. In: Handbook of Psychiatric Measures. 2000. p. 694–7.
Saunders J, Aasland O, Babor T, de la Puente J, Grant M. Development of the Alcohol-Use Disorders Identification Test (Audit) - Who Collaborative Project on Early Detection of Persons with Harmful Alcohol-Consumption. Addiction. 1993;88(6):791–804.
Daeppen J, Yersin B, Landry U, Pecoud A, Decrey H. Reliability and validity of the Alcohol Use Disorders Identification Test (AUDIT) imbedded within a general health risk screening questionnaire: results of a survey in 332 primary care patients. Alcohol Clin Exp Res. 2000;24(5):659–65.
Maisto S, Carey M, Carey K, Gordon C, Gleason J. Use of the AUDIT and the DAST-10 to identify alcohol and drug use disorders among adults with a severe and persistent mental illness. Psychol Assess. 2000;12(2):186–92.
Bennett T, Holloway K, Farrington D. The statistical association between drug misuse and crime: a meta-analysis. Aggress Violent Behav. 2008;13(2):107–18.
Richards TN, Jennings W, Murphy C. Risk and protective factors for batterer intervention treatment program attrition: how completers are distinct from dropouts and no-shows. J Interpers Violence. 2021;36(15–16):7351–70.
Olver ME, Stockdale K, Wormith J. A meta-analysis of predictors of offender treatment attrition and its relationship to recidivism. J Consult Clin Psychol. 2011;79(1):6–21.
Sampson RJ, Laub J. A life-course view of the development of crime. Ann Am Acad Pol Soc Sci. 2005;602(1):12–45.
Worthy DA, Gorlick MA, Pacheco JL, Schnyer DM, Maddox WT. With age comes wisdom: decision making in younger and older adults. Psychol Sci. 2011;22(11):1375–80.
Grossmann I, Na J, Varnum ME, Park DC, Kitayama S, Nisbett RE. Reasoning about social conflicts improves into old age. Proc Natl Acad Sci U S A. 2010;107(16):7246–50.
Behavioural Insights Team. How to Increase Voluntary Participation in Justice Programs. NSW: Department of Premier and Cabinet; 2018.
Spohr SA, Livingston MD, Taxman FS, Walters ST. What’s the influence of social interactions on substance use and treatment initiation? A prospective analysis among substance-using probationers. Addict Behav. 2019;89:143–50.
Cohen S, Wills TA. Stress, social support, and the buffering hypothesis. Psychol Bull. 1985;98(2):310.
Jacoby JE, Kozie-Peak B. The benefits of social support for mentally ill offenders: prison-to-community transitions. Behav Sci Law. 1997;15(4):483–501.
Wojciechowski T. Moderation of the relationship between hostility and offending: is social support a protective factor? J Crime Justice. 2022;ahead-of-print(ahead-of-print):1–15.
Zwecker NA, Harrison AJ, Welty LJ, Teplin LA, Abram KM. Social support networks among delinquent youth: an 8-year follow-up study. J Offender Rehabil. 2018;57(7):459–80.
Wrzus C, Hänel M, Wagner J, Neyer FJ. Social network changes and life events across the life span: a meta-analysis. Psychol Bull. 2013;139(1):53–80.
English T, Carstensen LL. Selective narrowing of social networks across adulthood is associated with improved emotional experience in daily life. Int J Behav Dev. 2014;38(2):195–202.
Klaehn AK, Jaschke J, Freigang F, Arnold M. Cost-effectiveness of case management: a systematic review. Am J Manag Care. 2022;28:e271–9.
Taft CT, Murphy CM, Elliott JD, Morrel TM. Attendance-enhancing procedures in group counseling for domestic abusers. J Couns Psychol. 2001;48(1):51–60.
Wormith JS, Olver M. Offender treatment attrition and its relationship with risk, responsivity, and recidivism. Crim Justice Behav. 2002;29(4):447–71.
Larsen MF. Does closing schools close doors? The effect of high school closings on achievement and attainment. Econ Educ Rev. 2020;76:101980.
Novak A, Fagan A. Expanding research on the school-to-prison pipeline: examining the relationships between suspension, expulsion, and recidivism among justice-involved youth. Crime Delinq. 2022;68(1):3–27.
Rosenbaum JE. School suspension predicts trichomoniasis five years later in a matched sample. BMC Public Health. 2020;20(1):88.
Achilles GM, McLaughlin M, Croninger R. Sociocultural correlates of disciplinary exclusion among students with emotional, behavioral, and learning disabilities in the SEELS national dataset. J Emot Behav Disord. 2007;15(1):33–45.
Lockwood B, Harris P. Kicked out or dropped out? Disaggregating the effects of community-based treatment attrition on juvenile recidivism. Justice Q. 2015;32(4):705–28.
Loeber R, Farrington D. Serious & violent juvenile offenders risk factors and successful interventions. Thousand Oaks: SAGE; 1999.
Clouston SA, Smith DM, Mukherjee S, Zhang Y, Hou W, Link BG, et al. Education and cognitive decline: an integrative analysis of global longitudinal studies of cognitive aging. J Gerontol B Psychol Sci Soc Sci. 2020;75:e151–60.
Guerra-Carrillo B, Katovich K, Bunge SA. Does higher education hone cognitive functioning and learning efficacy? Findings from a large and diverse sample. PLoS ONE. 2017;12(8):e0182276.
Zanis DA, Coviello DM, Lloyd JJ, Nazar BL. Predictors of drug treatment completion among parole violators. J Psychoactive Drugs. 2009;41(2):173–80. https://doi.org/10.1080/02791072.2009.10399910.
Polaschek DL. High-intensity rehabilitation for violent offenders in New Zealand: Reconviction outcomes for high-and medium-risk prisoners. J Interpers Violence. 2011;26(4):664–82. https://doi.org/10.1177/08862605103658.
Tollefson DR, Gross E, Lundahl B. Factors that predict attrition from a state-sponsored rural batterer treatment program. J Aggress Maltreat Trauma. 2008;17(4):453–77. https://doi.org/10.1080/10926770802463495.
Dishion TJ, McCord J, Poulin F. When interventions harm: Peer groups and problem behavior. Am Psychol. 1999;54(9):755.
Feder L, Dugan L. A test of the efficacy of court-mandated counseling for domestic violence offenders: the Broward experiment. Justice Q. 2002;19(2):343–75.
Gondolf EW, Foster R. Pre-program attrition in batterer programs. J Family Violence. 1991;6(4):337–49.
Christopher P, Garcia-Sampson L, Stein M, Johnson J, Rich J, Lidz C. Enrolling in clinical research while incarcerated: what influences participants’ decisions? Hastings Cent Rep. 2017;47(2):21–9.
Bowen E, Gilchrist E. Predicting dropout of court-mandated treatment in a British sample of domestic violence offenders. Psychol Crime Law. 2006;12(5):573–87.
Jewell LM, Wormith JS. Variables associated with attrition from domestic violence treatment programs targeting male batterers: a meta-analysis. Crim Justice Behav. 2010;37(10):1086–113.
Saunders DG, Parker JC. Legal sanctions and treatment follow-through among men who batter: a multivariate analysis. Soc Work Res Abstr. 1989;25(3):21–9.
Hamberger LK, Lohr J, Gottlieb M. Predictors of treatment dropout from a spouse abuse abatement program. Behav Modif. 2000;24(4):528–52.
Bilderbeck AC, Farias M, Brazil IA, Jakobowitz S, Wikholm C. Participation in a 10-week course of yoga improves behavioural control and decreases psychological distress in a prison population. J Psychiatr Res. 2013;47(10):1438–45. https://doi.org/10.1016/j.jpsychires.2013.06.014.
Bradley RG, Follingstad DR. Group therapy for incarcerated women who experienced interpersonal violence: a pilot study. J Trauma Stress. 2003;16:337–40. https://doi.org/10.1023/A:1024409817437.
Cashin A, Potter E, Butler T. The relationship between exercise and hopelessness in prison. J Psychiatr Ment Health Nurs. 2008;15(1):66–71. https://doi.org/10.1111/j.1365-2850.2007.01207.x.
Kunz R, Oxman A. The unpredictability paradox: review of empirical comparisons of randomised and non-randomised clinical trials. BMJ. 1998;317(7167):1185–90.
Howard MV, de Almeida Neto AC, Galouzis JJ. Relationships between treatment delivery, program attrition, and reoffending outcomes in an intensive custodial sex offender program. Sex Abuse. 2019;31(4):477–99.
Daly JE, Pelowski S. Predictors of dropout among men who batter: a review of studies with implications for research and practice. Violence Vict. 2000;15(2):137.
Initial funding for ReINVEST was provided from an Australian National Health and Medical Research Council partnership grant in 2009 (NHMRC; No. 533559). From 2018, subsequent funding has been provided by the NSW Department of Communities and Justice.
Ethics approval and consent to participate
This study has received ethical approval from the University of New South Wales (HC17771), Aboriginal Health & Medical Research Council (AHMRC; 822/11), Corrective Services NSW (09/26576) and the NSW Justice Health (G8/14)and the ReINVEST study is registered with the Australian and New Zealand Clinical Trials Registry (ACTRN12613000442707).
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Doring, N., Hwang, Y.I.(., Akpanekpo, E. et al. Predicting attrition of men with a history of violence from randomised clinical trials. Trials 24, 740 (2023). https://doi.org/10.1186/s13063-023-07774-3