Reduction of Crime in First-Onset Psychosis: A Secondary Analysis of the OPUS Randomized Trial
ABSTRACT
Objective: Violence and criminality are adverse outcomes for some persons who develop psychotic illnesses. The extent to which treatment can reduce offending has rarely been studied. The aim of this study was to evaluate whether assertive specialized treatment would reduce the rate of crime in patients with a first episode of psychotic illness.
Method: From January 1998 to December 2000, a total of 547 patients aged 18-45 years with a first episode of schizophrenia spectrum disorder (ICD-10 diagnostic code within F2) were randomized to assertive specialized treatment or standard treatment in an outpatient setting. In the current secondary analysis of the data, levels of criminality during the 2-year treatment period and the 3 years following were assessed using official records from Danish registers. Main outcome measures were any offending and violent offending.
Results: No significant reduction in violent offending or any offending was found in the assertive specialized treatment group (adjusted hazard ratio = 1.06; 95% CI, 0.72-1.56) compared with the control group. Prevalence of offending was low and had often commenced prior to inclusion in the trial.
Conclusions: While assertive specialized treatment has shown good treatment effects, it had no impact on rates of offending, thereby calling into question the potential efficacy of universally applied improvements in outpatient services with respect to reducing crime and violence. More specific interventions that address criminogenic needs in a more narrowly defined group of high-risk patients may be considered.
J Clin Psychiatry 2013;74(5):e439-e444
© Copyright 2013 Physicians Postgraduate Press, Inc.
Submitted: September 10, 2012; accepted February 15, 2013 (doi:10.4088/JCP.12m08156).
Corresponding author: Hanne Stevens, MSc, National Centre for Register-Based Research, Aarhus University, Fuglesangs Allé 4, 8210 Aarhus V, Denmark ([email protected]).
Those who suffer from psychotic illnesses have been consistently shown to display elevated rates of violence and criminality,1,2 particularly around the time of illness onset.3 Some researchers4 have argued that the excess risk in this group can be explained by co-occurring antisocial traits and problems with substance misuse, while others5 maintain that the risk is driven by the psychotic symptoms per se, whereby adequate treatment is thought to prevent offending. Whether criminality is driven by direct effects of illness or whether known criminogenic risk factors are occurring at a higher rate, the reduction of such criminality is important both in terms of avoiding the adverse impact on the perpetrator and the victims and in terms of potentially reducing the stigma of those with psychosis.5,6 However, violence and criminality have rarely been considered an outcome of interest for interventions in this group.
Using data from the OPUS trial,7 a randomized controlled trial comparing assertive specialized treatment and standard treatment of first-episode psychosis, the aim of this secondary analysis was to compare rates of offending in the 2 treatment groups and to assess whether assertive specialized treatment prevented offending in the 2-year treatment period and the subsequent 3-year follow-up period. Since the treatment under consideration was not directed at reducing antisocial behavior, the question addressed in this study was whether improved clinical management was sufficient to reduce offending.
The main outcome measures were guilty verdicts in general (any offending) and more specifically for violent offending. Additionally, we utilized the high-quality information on duration of untreated psychosis in the OPUS data to assess whether the risk of offending increased after illness onset, but before treatment.
METHOD
Participants
During the period from January 1998 to December 2000, a total of 547 patients with a diagnosis in the schizophrenia spectrum (ICD-10 code within F2) were included in the OPUS trial and were randomized to either assertive specialized treatment or treatment as usual. They were recruited from both inpatient and outpatient mental health services in the 2 largest Danish cities, Copenhagen and Aarhus; were 18-45 years old; and had not received antipsychotic drugs for more than 12 weeks of continuous treatment at the time of inclusion. Exclusion criteria included presence of mental retardation, organic mental disorder, or psychotic condition due to acute intoxication or withdrawal state, although comorbid substance misuse in itself was not grounds for exclusion. Additionally, familiarity with the Danish language was required. Around 5% of the referred patients refused to participate in the trial8; however, they did not differ from those who did participate with regard to duration of psychosis, severity of psychopathology, or diagnosis.9 Comparisons with national registers revealed that, in Aarhus, 90% of those who had a first diagnosis within ICD-10 diagnostic code F2 in the inclusion period participated in the trial. In Copenhagen, the corresponding number was 63%.
Ethics
All participants gave informed consent. The randomized controlled trial was approved by the Danish Ethics Committee (KF 01-387/97) prior to its initiation and was registered at ClinicalTrials.gov (identifier: NCT00157313).
Treatment Groups
The experimental treatment consisted of assertive community treatment, family involvement, and social skills training and had a duration of 2 years, during which time patients saw their primary staff member usually on a weekly basis, often in their own home. The caseload ratio was 1:10. After 2 years, the patients were transferred to standard treatment. The fidelity to the standard assertive community treatment model, as measured by the Index of Fidelity of Assertive Community Treatment, was 70%.10 The standard treatment offered contact with a community mental health center, with an average caseload ratio of 1:25, less frequent meetings, and no systematic offers of additional treatment elements. In both treatment groups, antipsychotic medication was administered as indicated by and in accordance with Danish guidelines, which recommend a low dose strategy and second-generation antipsychotics as the first drug choice. For a full description of randomization, treatment content, and assessments, see Petersen et al.7 After a 2-year treatment period, patients receiving the assertive specialized treatment were found to have significantly better clinical outcomes with regard to psychotic and negative symptoms, secondary substance misuse, treatment adherence, and success with lower doses of antipsychotic medication.7 At the 5-year follow-up, these differences had equalized between the treatment groups; however, the patients who had received assertive specialized treatment fared better on secondary outcome measures such as living in supported housing and number of days spent in hospital.10
- Assertive specialized treatment alone, when applied universally in first-episode psychosis, did not reduce the risk or volume of violence and other criminality, despite the proven effectiveness of this treatment in reducing levels of symptoms and problems with comorbid substance misuse.
- Almost three-quarters of those who committed offenses after commencing treatment for first-episode psychosis had already begun doing so before inclusion in the treatment trial.
Sources of Data
From the OPUS trial, we obtained baseline information on gender and age at inclusion, primary diagnosis according to ICD-10,11 level of psychotic and negative symptoms (Scale for Assessment of Positive Symptoms12 and Scale for Assessment of Negative Symptoms12), duration of untreated psychosis (Interview for the Retrospective Assessment of the Onset of Schizophrenia13), and presence of substance misuse (Schedule for Clinical Assessment in Neuropsychiatry14), all of which was combined with information from Danish national registers. The Psychiatric Central Research Register15 contains information on all admissions since 1969 and on all outpatient contacts since 1995. From this register, we obtained information for each patient on any periods of admission after inclusion in the trial. The Danish Civil Registration System16 contains data on gender and date of birth and maintains continuously updated information on vital status; from this register, we obtained information on the date of death or emigration, when applicable. Data were linked using the unique personal identification number.
Main Outcome Measure
From the Danish National Crime Register,17 which is essentially 100% complete, we obtained information on all offenses that led to a guilty verdict. This register became electronic in 1978, and, through Statistics Denmark (http://www.dst.dk/en/Statistik/dokumentation/Declarations/convictions-for-criminal-offences.aspx), we had access to all criminal charges from 1980 to 2007. Guilty verdicts include custodial sentences, suspended sentences, conditional withdrawal of charges, fines, and sentences to psychiatric treatment; Danish courts do not use diversion programs for mentally disordered offenders. Offenses against the penal code, special legislation regarding drugs and weapons, or sections of the traffic act dealing with impaired driving were included as "any" offending, while violent offending included all violent and sexual offenses. We used the date of the offense as the time point for the survival analyses, and, in cases for which this date was missing (16 of 904; 1.8%), we used the date of the conviction instead. We also calculated the number and type of offenses and convictions within the 5-year follow-up period. For this analysis, we considered the following types of offending: violent (including sexual), acquisitive (theft, fraud, breaking and entering, and related), substance related (possession, trafficking, and driving under the influence), and other.
Statistical Analysis and Power Calculation
The participants were followed from the time of inclusion in the trial until the time of violent offending or any offending, death, emigration, or the end of follow-up (2 years and 5 years), whichever came first. Attrition from the study was around 45% in both treatment groups after 5 years,10 but, since we used official records to assess offending status, we were able to obtain full follow-up information on all participants in the trial. For generating Kaplan-Meier plots,18 we used the LIFETEST procedure in SAS statistical software, version 9.1.3 (SAS Institute Inc; Cary, North Carolina). Hazard ratios, 95% CIs, Wald statistics, and associated P values were based on the Cox proportional hazards model,18 using the PHREG procedure in SAS. Adjusted models considered the potential confounding effects of age at baseline, gender, offending history (prior to recruitment into the study), level of negative and psychotic dimension symptoms, duration of untreated psychosis, and presence of substance misuse at baseline. Periods during follow-up that were spent as an inpatient in a psychiatric hospital were included as a time-varying covariate. All analyses were performed as intention to treat.
With 270 patients randomized to each group, we would be able to detect a difference equivalent to a hazard ratio of 0.7 for any offending statistically significant at the 5% level with a high probability (power > 80%).19
Analysis of Offending Prior to Treatment
Looking at the period prior to inclusion in the trial, we used information on duration of untreated psychosis to estimate the time of onset of psychosis and analyzed whether the first offense occurred before or after this time. This analysis was restricted to those patients who had experienced psychotic symptoms (excluding 93 patients with a diagnosis of schizophrenia simplex or schizotypal disorder) and to those whose complete criminal record was available (excluding a further 75 patients born in 1964 or earlier). Using a Cox regression model, we followed individuals from their 15th birthday until their first offense or their inclusion in OPUS, whichever came first. Onset of psychosis was entered as a time-varying variable, and the model was adjusted for gender. Note that this analysis involved conditioning on the future, as all participants were later enrolled in the OPUS study.
RESULTS
Of the total OPUS sample, 275 (50.3%) were randomized to the assertive specialized treatment group and 272 (49.7%) to the treatment-as-usual group. There were no significant differences between the groups with respect to sociodemographic and clinical factors. Characteristics of the groups have been described in full elsewhere.20 Offending prior to inclusion in the trial was prevalent in both groups, with 88 (32.0%) of the assertive specialized treatment group and 90 (33.1%) of the standard care group having engaged in such behavior. In both treatment groups, 23 (about 8%) had a conviction for violent offending prior to inclusion in the trial.
Main Outcome: Effect of Treatment on Crime
Figure 1 shows a Kaplan-Meier plot of the 2 treatment groups with respect to their first offense following inclusion in the program. Contrary to what was hypothesized, there did not appear to be any difference between the 2 treatment groups (P = .69). By the end of the 2-year treatment period, 12% in both groups had offended, and, at 5 years after inclusion, 20% of the assertive specialized treatment group and 19% of the standard treatment group had offended. Of those who offended after inclusion, almost 75% had also done so before inclusion (41 of 55 in the assertive specialized treatment group and 35 of 50 in the standard treatment group).
Click figure to enlarge
Violent offending was less prevalent but was of similar magnitude in both treatment groups, with 3% in both groups having committed a violent crime after 2 years, and with 5% in the assertive specialized treatment group and 6% in the standard treatment group having committed a violent crime after 5 years. Hypothesizing that treatment could have differential effects depending on the person’s prior offending history, we tested for equality over strata in restricted models that contained only those with or those without a history of offending. With P values for the log-rank test in the range of .31-.73 for any offending and .65-.97 for violent offending, we found no evidence to support this hypothesis.
In a Cox regression, we found an insignificant hazard ratio of 1.08 (95% CI, 0.74-1.58) for assertive specialized treatment compared to standard treatment. In a fully adjusted model (Table 1), the association remained insignificant, but male gender, younger age, substance misuse at baseline, and a history of offending were identified as risk factors for offending. Although the result was not significant, there was some indication (P = .11) that those who were unwilling or unable to give information on duration of untreated psychosis were at increased risk for offending, while our data did not indicate that those with a long duration of untreated psychosis should be at increased risk for offending. For violent offending, the unadjusted hazard ratio was 0.91 (95% CI, 0.45-1.84). Given the very low prevalence of this outcome, we lacked sufficient data to fit an adjusted model.
Click figure to enlarge
Click figure to enlarge
Frequency of Offending
Although we found no evidence that assertive specialized treatment reduced the occurrence of offending relative to standard treatment, we considered the possibility that assertive specialized treatment could reduce the volume of offending, such that those who offend do so less frequently. Table 2 shows the cumulative number of offenses for both treatment groups during the first 5 years after inclusion, along with frequencies of different types of verdicts and offenses and the total number of convictions within the 5-year follow-up. Again, there were no significant differences between the 2 treatment groups; however, the level of criminality was modest. Most of those who offended did so only once, and many received only a fine.
Relative Time of Onset—Psychosis Versus Offending
Looking to the period preceding inclusion in the trial, we found a hazard ratio of 1.29 (95% CI, 0.82-2.02) for committing the first offense after the onset of psychosis relative to before. Although the result was not significant, this finding does give some indication that the risk of offending may increase after onset of a psychotic disorder.
DISCUSSION
In our analysis of data from a controlled trial of 547 patients with a first episode of psychosis who were randomized to assertive specialized treatment or standard care, we found no significant reductions in violent offending or any offending, both in terms of the number of people who engaged in such behaviors and with respect to frequency of offending. While the sample size may have limited our ability to detect small differences, no trends for differences between the treatment groups were found.
Our finding of no difference is in line with 1 previous study21 of the effect of intensive case management on violent behavior. Utilizing data from the UK700 study, Walsh et al21 found no reduction in violence in an inner city sample of persons with chronic psychosis. In comparison to the UK study, our patients were younger, they were in an earlier stage of illness, and the difference between treatments was larger since our specialized treatment consisted of assertive community treatment, psychoeducational family involvement, and social skills training and not simply a lighter caseload. For these reasons, along with OPUS patients’ having been shown to have significantly better clinical outcomes (psychotic and negative symptoms, secondary substance misuse, treatment adherence, and success with lower doses of antipsychotic medication) and secondary outcomes (living in supported housing and days spent in hospital),7,10 one would have expected better results. Explanations for a lack of effect include the possibility that the intervention was still not intensive enough or that it should have specifically targeted risk of criminal behavior, eg, proviolence attitudes, anger experience, and social problem solving. It is also the case that the prevalence of offending after inclusion in the OPUS study was relatively low, and it might be argued that benefits would more likely be found in interventions that target higher-risk patients, such as those with dual diagnoses.
Periods spent in psychiatric hospitals can be regarded as more intensive interventions, and, for the analyses, we considered whether these times were best conceptualized as time not at liberty to offend or as time with reduced opportunities to offend. On the basis of Danish practice, in which it is not uncommon for violent episodes in inpatient settings to be reported to the police and dealt with by the courts,22 and on the basis of the empirical observation that some patients did in fact offend while hospitalized, we entered time in hospital as a time-varying variable in the Cox regression rather than censoring out those periods.23 As this variable was not significant in multivariate regression, we found no indication that hospitalization reduced offending. Of course, those who are at increased risk of violence would more likely be admitted, which would confound the results; however, we have no reason to believe that increased risk of nonviolent offending, which is by far the most prevalent in our study, should have any association with likelihood of admission.
A key point is that almost three-quarters of those who offended after commencing treatment had already begun doing so before inclusion in the program. Numerically, most of the preinclusion offending took place before illness onset, indicating that non-mental illness factors may be important for offending. Taking time at risk into account, however, we found some indication, although not statistically significant, that the risk of offending increases after onset of psychotic symptoms, which makes programs targeting early detection and early treatment potentially of interest for study. In other words, the timing of treatment may be important; however, this issue remains to be explored. Our failure to find any association between offending and duration of untreated psychosis is consistent with the results of a recent meta-analysis and systematic review by Large and Nielssen,3 in which more serious violence was associated with a longer duration of untreated psychosis, while less serious violence was not.
Strengths and Limitations
Even though the OPUS trial is one of the larger randomized controlled trials comparing assertive specialized treatment with standard care, our analysis still suffers from the possibility of type II error in the Cox regression, although we did not see any indication of trend in our data. A great strength of the present study is the use of national registers for follow-up information, particularly since those with antisocial traits are more likely to be lost to follow-up under usual study conditions. Linking the OPUS dataset to the national registers means that differential attrition was avoided and that we were able to obtain follow-up information on all participants regardless of whether, or for how long, they participated in the trial. Loss to follow-up occurred only in cases of death or emigration, and, in these cases, we had access to the exact dates of loss so as to make relevant adjustments to the analyses.
Using official records caused underestimation of the rates of offending and aggression, which becomes quite apparent when our results are compared to the meta-analysis of Large and Nielssen,3 who reported that 35% of first-episode patients had any degree of violence and that 17% had at least 1 episode of more severe violence (any degree of injury, use of weapon, or sexual assault) prior to treatment contact. Only 8% of patients in our study had a previous conviction for violence. While part of the difference can possibly be related to differences in levels of criminality or demographic compositions in the various studies, our measurement was less sensitive than using self-report or case notes. Apart from avoiding differential attrition, the reliance on official records protects against information bias in a study in which the intervention group had more frequent contact with carers than controls and also ensures standardized definitions.
CONCLUSIONS
While assertive specialized treatment has been shown to improve clinical outcomes in first-episode psychosis, in the current study we found no indication of an effect on offending. The prevalence of offending was low in the study group, and the majority of those who offended after inclusion in the trial had commenced doing so prior to that time, indicating that earlier intervention may be warranted.
Author affiliations: National Centre for Register-Based Research (Ms Stevens and Drs Agerbo and Mortensen) and CIRRAU-Centre for Integrated Register-Based Research (Dr Agerbo), Aarhus University, Aarhus, Denmark; School of Psychiatry, University of New South Wales, Sydney, Australia (Dr Dean); and Department of Psychiatry, Bispebjerg Hospital, Copenhagen, Denmark (Dr Nordentoft).
Author contributions: Ms Stevens analyzed and interpreted the data and drafted the manuscript. Drs Nordentoft, Agerbo, Dean, and Mortensen interpreted the data and critically revised the manuscript. All authors have read and approved the final version for submission.
Potential conflicts of interest: None reported.
Funding/support: Ms Stevens was funded by The Danish Council for Independent Research: Medical Sciences (grant 271-08-0489) and Danish Regions (grant 08/2741). Dr Mortensen was supported financially by The Stanley Medical Research Foundation and The Danish Council for Independent Research: Medical Sciences.
Additional information: The OPUS database is owned by Merete Nordentoft, DrMed; for information on accessing the OPUS data, contact Dr Nordentoft at [email protected]. Data from the Danish National Crime Register and the Central Person Register (Danish Civil Registration System) are available through Statistics Denmark at http://www.dst.dk/en/TilSalg/Forskningsservice.aspx. The Psychiatric Central Research Register is located at Statens Serum Institut, Copenhagen, Denmark; for information on gaining access, contact [email protected].
REFERENCES
1. Fazel S, Gulati G, Linsell L, et al. Schizophrenia and violence: systematic review and meta-analysis. PLoS Med. 2009;6(8):e1000120. PubMed doi:10.1371/journal.pmed.1000120
2. Monahan J. Mental disorder and violent behavior: perceptions and evidence. Am Psychol. 1992;47(4):511-521. PubMed doi:10.1037/0003-066X.47.4.511
3. Large MM, Nielssen O. Violence in first-episode psychosis: a systematic review and meta-analysis. Schizophr Res. 2011;125(2-3):209-220. PubMed doi:10.1016/j.schres.2010.11.026
4. Bonta J, Law M, Hanson K. The prediction of criminal and violent recidivism among mentally disordered offenders: a meta-analysis. Psychol Bull. 1998;123(2):123-142. PubMed doi:10.1037/0033-2909.123.2.123
5. Torrey EF. Stigma and violence: isn’ t it time to connect the dots? Schizophr Bull. 2011;37(5):892-896. PubMed doi:10.1093/schbul/sbr057
6. Mullen PE. Facing up to unpalatable evidence for the sake of our patients. PLoS Med. 2009;6(8):e1000112. PubMed doi:10.1371/journal.pmed.1000112
7. Petersen L, Jeppesen P, Thorup A, et al. A randomised multicentre trial of integrated versus standard treatment for patients with a first episode of psychotic illness. BMJ. 2005;331(7517):602-605. PubMed doi:10.1136/bmj.38565.415000.E01
8. Thorup A, Petersen L, Jeppesen P, et al. Gender differences in young adults with first-episode schizophrenia spectrum disorders at baseline in the Danish OPUS study. J Nerv Ment Dis. 2007;195(5):396-405. PubMed
9. J׸rgensen P, Nordentoft M, Abel MB, et al. Early detection and assertive community treatment of young psychotics: the OPUS Study Rationale and design of the trial. Soc Psychiatry Psychiatr Epidemiol. 2000;35(7):283-287. PubMed doi:10.1007/s001270050240
10. Bertelsen M, Jeppesen P, Petersen L, et al. Five-year follow-up of a randomized multicenter trial of intensive early intervention vs standard treatment for patients with a first episode of psychotic illness: the OPUS trial. Arch Gen Psychiatry. 2008;65(7):762-771. PubMed doi:10.1001/archpsyc.65.7.762
11. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Copenhagen, Denmark: World Health Organization; 1992.
12. Andreasen NC, Olsen S. Negative v positive schizophrenia: definition and validation. Arch Gen Psychiatry. 1982;39(7):789-794. PubMed doi:10.1001/archpsyc.1982.04290070025006
13. Hפfner H, Riecher-Rössler A, Hambrecht M, et al. IRAOS: an instrument for the assessment of onset and early course of schizophrenia. Schizophr Res. 1992;6(3):209-223. PubMed doi:10.1016/0920-9964(92)90004-O
14. World Health Organization Division of Mental Health. Schedule for Clinical Assessment in Neuropsychiatry, Version 2.1: Present State Examination. Copenhagen, Denmark: World Health Organization; 1998.
15. Mors O, Perto GP, Mortensen PB. The Danish Psychiatric Central Research Register. Scand J Public Health. 2011;39(7 suppl):54-57. PubMed doi:10.1177/1403494810395825
16. Pedersen CB, G׸tzsche H, M׸ller JO, et al. The Danish Civil Registration System: a cohort of eight million persons. Dan Med Bull. 2006;53(4):441-449. PubMed
17. Kyvsgaard B. The Criminal Career. London, England: Cambridge University Press; 2003.
18. Hosmer DW, Lemeshow S. Applied Survival Analysis: Regression Modeling of Time to Event Data. Hoboken, NJ: John Wiley & Sons; 1999.
19. Schoenfeld DA. Sample-size formula for the proportional-hazards regression model. Biometrics. 1983;39(2):499-503. PubMed doi:10.2307/2531021
20. Petersen L, Nordentoft M, Jeppesen P, et al. Improving 1-year outcome in first-episode psychosis: OPUS trial. Br J Psychiatry. 2005;187(48):S98-S103. PubMed doi:10.1192/bjp.187.48.s98
21. Walsh E, Gilvarry C, Samele C, et al. Reducing violence in severe mental illness: randomised controlled trial of intensive case management compared with standard care. BMJ. 2001;323(7321):1093-1096. PubMed doi:10.1136/bmj.323.7321.1093
22. Stevens H, Nordentoft M, Agerbo E, et al. [Psychiatric treatment sentences] [article in Danish]. Ugeskr Laeger. 2010;172(35):2366-2370. PubMed
23. Allison PD. Survival Analysis Using SAS: A Practical Guide, Second Edition. Cary, NC: SAS Institute; 2010.
