Mental Health Intervention and Social Service in Pretrial Detention

2023-12-04

2026-12-31

The main outcomes of interests are:
1. Mental health: anxiety (GAD-7), depression (PHQ-9) and suicidal ideation, prescription of mental health medications, and mental health related doctor/hospital visits, along with their associated costs
2. Economic situation: employment status, wage earnings, hours worked, subjective and objective indicators of financial distress, and reliance on social welfare.
3. Social relationships: Multidimensional Scale of Perceived Social Support (MSPSS), and relationship status, and household composition
4. Housing situation: housing stability, homelessness, and housing satisfaction
5. Future criminal behavior: crime registry entries after release, and depending on availability criminal charges after release
6. Institutional behavior in jail: disciplinary offenses during pretrial incarceration, social behavior, as assessed by guards

We analyze these primary outcomes within the scope of multiple work packages (see supporting documentation “Work packages”)

We additionally study the impact of PRISMA and proactive social services on the following secondary outcomes:

1. Physical health: self-assessed general physical health (GHP-2), prescription of non-mental health medication, and non-mental health related utilization of health care services.
2. Procedural outcomes: duration of pretrial incarceration, reason for end of pre-trial detention (e.g., transfer to correction facility, release, etc.)
3. Education: highest completed education, discontinuation of educational track

While we do not expect that the treatments impact any of the procedural outcomes, we nevertheless analyze them as a robustness check.

We will also analyze data from a brief survey we administer to all participants shortly before their release from pretrial detention. While this survey has the disadvantage that there are differences across participants in the amount of time that has elapsed since randomization, we expect a relatively high response rate. In particular, we are interested in the following outcomes:

4. Mental health: anxiety and depression (PHQ-2 and GAD-2))
5. Confidence in receiving social support from family and friends immediately after release
6. Expected housing situation in the next couple of days after release
7. Confidence in finding a job within four weeks after release
8. Perception of procedural fairness
9. Trust in the justice system and prison staff

In addition, we collect the following secondary outcome measures to better understand the potential underlying mechanisms how PRISMA and proactive social services affect our primary outcomes.
10. Jail visits
11. Perception of procedural fairness and trust in justice system
12. Self-reported criminal attitudes and legal cynicism
13. Self-evaluations: locus of control, self-efficacy, and self-esteem
14. Cognitive bandwidth: working memory assessed through forward digit span

Because participation in both interventions is voluntary it is also important to assess treatment take up. We will analyze take-up of PRISMA based on the following measures: the number of completed PRISMA sessions, the trainers’ assessment of whether participants were practicing the PRISMA exercises,

We analyze take-up of social services by the frequency and duration of interactions with social services, number of immediate measures taken, amount of time social service workers invested into immediate measure, and the life domains addressed by immediate measures.

Recruitment and informed consent

The model trial will be carried out during the 34-month study period (enrollment window) between December 4, 2023 and September 30, 2026. Study participants will be recruited in 11 pretrial detention facilities in the cantons of Zurich and Bern. All seven pretrial detention facilities in the Canton of Zurich (Prison Dielsdorf, Prison Horgen, Prison Limmattal, Prison Pfäffikon, Prison Winterthur, Prison Zürich, Prison Zürich West) and all four pretrial detention facilities in the Canton of Bern (Prison Bern, Prison Biel, Prison Burgdorf, Prison Thun) will participate in the model trial. Based on historical data, we expect around 5’100 people to enter pretrial detention in the cantons of Zurich and Bern during the 34-month duration of the model trial. See supporting document “Eligibility criteria” for a description of the eligibility criteria for the two interventions.


During the informed consent procedure, guards will explain to each participant the nature of the study, its purpose, the procedures involved, the expected duration, the potential risks and benefits and any discomfort it may entail. Each participant will be informed that the participation in the study is voluntary and that he or she may withdraw from the study at any time, and that non-participation or withdrawal of consent will not have consequences for how they are treated during pretrial detention, nor will it be considered in any legal procedure. To ensure consistent quality, a video is available in all 14 study languages, explaining the above mentioned key features in an accessible way.

Subjects eligible for the PRISMA intervention or for both interventions will be asked to consent to the following (consent form “A”):
1. Being randomized into a PRISMA treatment or a control group
2. Voluntarily participating in the PRISMA intervention if being randomized into the treatment group
3. Being invited to multiple survey assessments
4. Sharing their prison internal medical data from prison health services and doctors
5. Linking the additionally collected survey data and the shared prison internal medical data with the administrative data (i.e., prison internal non-medical data and external administrative register data) collected by administrative organizations (prisons, Swiss Federal Statistical Office, Central Compensation Office)
In addition, subjects can give consent (in consent form “A”) on whether they are willing (or not) to voluntarily disclose their external health insurance provider data.

Subjects eligible for only the proactive social service intervention will be asked to consent to the following (consent form “B”):
1. Being invited to multiple survey assessments
2. Sharing their prison internal medical data from prison health services and doctors
3. Linking the additionally collected survey data and the shared prison internal medical data with the administrative data (i.e., prison internal non-medical data and external administrative register data) collected by administrative organizations (prisons, Swiss Federal Statistical Office, Central Compensation Office)
In addition, subjects can give consent (in consent form “B”) on whether they are willing (or not) to voluntarily disclose their external health insurance provider data.

Subjects not eligible for neither the PRISMA intervention nor the extended social service intervention are not randomized and only serve as an observational group. They are asked to consent to the following (consent form “C”):
1. Sharing their prison internal medical data from prison health services and doctors
2. Linking the shared prison internal medical data with the prison internal non-medical data


Randomization samples and treatment assignment

Depending on whether participants fulfill eligibility and provide consent for the PRISMA intervention, the social service intervention, or both, they will be allocated into one of the following three randomization samples:

Randomization sample A: Subjects who are only eligible for PRISMA but not for proactive social services and who consent are randomized into one of the following two conditions with equal probability:
1. PRISMA: participants in this group are offered the psychological intervention PRISMA.
2. REGULAR PRETRIAL: participants in this group experience the same system and conditions that they would have experienced if the RCT did not exist. In particular, they will not be offered PRISMA, but they can still access medical support through the jails’ health services.

Randomization sample B: Subjects who are only eligible for proactive social services but not for PRISMA are randomized into one of the following two conditions with equal probability:
1. SOCIAL: participants in this are offered extended support from social services. If during the intake interview the social services identify the need for (immediate) measures to preserve external resources, these measures will be discussed with the inmate. For the social services to take action, the inmate (and the prosecution office) must consent to every intended measure.
2. REGULAR PRETRIAL: participants in this group experience the same system and conditions that they would have experienced if the RCT did not exist. In particular, they will not be offered proactive social services, but they can still access social services upon request.

Randomization sample C: Subjects who are eligible for both interventions and who consent are randomized into one of the following four conditions with equal probability:
1. PRISMA: participants in this group are offered the psychological intervention PRISMA.
2. SOCIAL: participants in this are offered extended support from social services. If during the intake interview the social services identify the need for (immediate) measures to preserve external resources, these measures will be discussed with the inmate. For the social services to take action, the inmate (and the prosecution office) must consent to every intended measure.
3. PRISMA & SOCIAL: participants in this group are offered both the psychological intervention PRISMA and the extended support from social services.
4. REGULAR PRETRIAL: participants in this group experience the same system and conditions that they would have experienced if the RCT did not exist. In particular, they will not be offered PRISMA or extended social services, but they can still access social services upon request as well as medical support through the jails’ health services.


Subjects who are only eligible for the proactive social service intervention (randomization sample B) will be randomized into the group SOCIAL or the group REGULAR PRETRIAL irrespective of consenting to consent form “B” or not. For subjects not consenting to consent form “B”, The effects of the proactive social service intervention can still be evaluated using only prison internal non-medical data and external non-medical administrative register data. Federal regulation permits the use of this pseudonymized data without consent.

Subjects who are eligible for both interventions (randomization sample C), but do not provide consent using consent form “A” will be transferred from randomization sample C to randomization sample B. They will thus be treated as if they were not eligible for PRISMA, and the effects will be evaluated using only prison internal non-medical data and external non-medical administrative register data. The reason for this transfer from randomization sample C into randomization sample B is the following: Without consent, we cannot evaluate survey data nor any medical data, which both are essential for evaluating the effects of the PRISMA intervention. However, without consent, we can still evaluate prison internal non-medical data and external non-medical administrative register data.

Outcome Data Collection

The research team will collect outcome data from administrative data sources and surveys (see supporting documents “Survey questions and schedule” and “Administrative data source”. The first (baseline) survey will be administered before randomization. Subsequently, surveys will occur according to a pre-specified schedule (at the day of exit from pre-trial detention, 3 weeks after intake, 10 weeks after intake, 6 months after intake, and 1 year after intake).

Not available

Randomization done by statistical software

Individual

No

N/A

The sample size is determined by the enrollment window (from December 4, 2023 to September 30, 2026). Based on historical data, we expect roughly 5,100 subjects to enter pretrial detention in the cantons of Zurich and Bern during this period. We expect that 80% (i.e., 4082 subjects) will be eligible for at least one of the interventions, with the samples containing the following numbers of subjects: 766 subjects in sample A, 1’888 subjects in sample B, 1’428 subjects in sample C. As it was impossible to conduct a pilot, we expect implementation issues, particularly in the beginning of the study. We therefore keep the option to drop all initial observations from prisons as long as i) less than 80% of the participants from SOCIAL and PRISMA & SOCIAL who are still in jail were approached by social services for an interview within the first three workdays since randomization, and ii) less than 80% of the participants from PRISMA and PRISMA & SOCIAL who are still in jail got their first PRISMA session booked within two workdays since randomization.

Of the projected 5’100 people entering pretrial detention during the enrollment window, we expect that roughly 43% of (or 2’194 individuals) will be eligible for PRISMA and roughly 65% (or 3’316 individuals) will be eligible for the proactive social services intervention, respectively. Thus, we expect 1’428 individuals (or 28% = 43% of 65%) to be eligible for both interventions and, hence, constitute sample C. This means that 766 individuals (= 2’194 – 1’428) will be eligible for PRISMA only (sample A) and 1’888 individuals (= 3’316 – 1’428) for proactive social services only (sample B). 1’018 (= 5’100 – 1’428 – 766 – 1’888) individuals will not be eligible for any of the two interventions. If samples are split with equal probability into experimental groups, we expect the following group sizes:

Randomization sample A:
o REGULAR PRETRIAL: 383
o PRISMA: 383

Randomization sample B:
o REGULAR PRETRIAL: 944
o SOCIAL: 944

Randomization sample C:
o REGULAR PRETRIAL: 357
o PRISMA: 357
o SOCIAL: 357
o PRISMA & SOCIAL: 357