Longitudinal Study Comparing an Italian and a Californian Population of Mentally Ill Offenders: Identification of Forensic Treatment Effectiveness Factors

This longitudinal, multicentre, non-randomized cohort study compares outcomes for mentally ill offenders entering Italian and Californian forensic care systems for the first time. The Italian cohort (Group I) comprises individuals subject to custodial security measures in REMS and to non-custodial security measures (for example supervised liberty and dedicated community residential facilities), managed by Departments of Mental Health in residential, semi-residential, and community-based settings. The Californian cohort (Group C) comprises patients admitted to secure state hospitals or community forensic facilities under incompetency to stand trial (IST) or not guilty by reason of insanity (NGRI) provisions of the California Penal Code.

The observation period extends over three years from baseline (T0), with follow-up assessments at 12 months (T1), 24 months (T2), and 36 months (T3), or at discharge from the forensic system if discharge occurs earlier. At T0, data collection includes a structured form covering sociodemographic characteristics, clinical history, substance use, previous psychiatric admissions, self-harm and hetero-aggressive behaviour, current and past pharmacological treatment, and judicial information such as type of offence, duration and type of security measure, previous convictions, and prior detention periods. Baseline assessment also includes BPRS 4.0 for psychiatric symptom severity, WHODAS 2.0 for global functioning, HCR-20 V3 for violence risk, SAPROF for protective factors, and DUNDRUM-1, -3, and -4 for therapeutic security level, treatment programme completion, and recovery status, respectively.

During follow-up, the same battery (BPRS 4.0, WHODAS 2.0, HCR-20 V3, SAPROF, DUNDRUM-1/3/4) is administered at T1, T2, and T3, and clinicians record sentinel events, including new offences, compulsory psychiatric hospitalisations, seclusion or restraint episodes, suicide attempts, suicides, and unauthorised absences or escapes, with violent incidents coded using the Modified Overt Aggression Scale (MOAS). Time windows for each follow-up assessment are restricted (maximum approximately 2 weeks) and are applied uniformly across participating sites to minimise temporal bias in outcome measurement. At the end of the three-year observation period, or at discharge, a Clinical Global Impression - Change (CGI-C) interview is conducted with treating health professionals to capture their structured judgement regarding the quality and effectiveness of the overall treatment pathway.

Inclusion criteria encompass adult offenders (≥18 years) with partial or total mental infirmity placed under forensic measures in Italy or admitted to forensic care in California, whereas exclusion criteria include age under 18 years, severe cognitive impairment (for example Mini-Mental State Examination score <18) preventing valid assessment, inability to provide informed consent in the absence of a legal guardian, and acute medical conditions incompatible with study procedures. Target sample size is approximately 300 participants per group, which provides adequate statistical power to detect clinically meaningful differences between systems while allowing for attrition during follow-up, although the study remains informative even with lower enrolment in line with observational research standards.

Primary outcomes include violent recidivism, length of stay in forensic settings, and changes in clinical and functional status, while secondary outcomes focus on changes in structured risk and protective factors, treatment adherence, therapeutic progress, and recovery indicators. Harmonised instruments and procedures across Italian and Californian sites allow direct comparison of trajectories, with particular attention to subgroups with psychotic disorders and mental disorders related to the index offence, in order to disentangle the impact of system-level organisational and legal features from individual clinical profiles.