Description

The DUNDRUM Forensic Redevelopment Evaluation Study (D-FOREST).

Forensic Mental Health Services have a dual role, to treat mental disorder and reduce violent recidivism. Forensic mental health services are low volume, high cost services and therefore it is vital to conduct robust outcome measure studies to demonstrate benefit and effectiveness. D-FOREST is a study comprising the evaluation of a complete national forensic mental health service at a time of significant change, that of the re-development of Ireland's National Forensic Mental Health Service from a 19th century building to a complete national service in North Dublin.

We hypothesise that the redevelopment of the campus from Dundrum Hospital to a new 170-bed hospital in Portrane will demonstrate improvements in two main areas, firstly service wide improvements and secondly patient benefit related improvements.

We hypothesise that the development of the new campus at Portrane will lead to reduced time on the waiting list for admission and reduced length of stay in the hospital as well as sustainable rates of admissions and discharges maintained over the five year period of this study.

We anticipate that the newly expanded hospital will be able to facilitate the admission of patients with higher security needs from the prison settings, and therefore there may be a rise in the security needs profile of the group, while urgency of need for admission on the waiting list will decrease.

We anticipate that patient engagement with group and individual therapy will improve with greater access to therapists as the staff number will expand and therefore measures of therapeutic programme completion and recovery and measures of risk in a broad sense should improve across the patient group. We also anticipate finding improved measures of patient ratings on environmental measures such as ward atmosphere on moving to the new campus.

Repeated measures in the area of physical health, mental health and violence risk and forensic recovery will be taken, at baseline (time of admission to the service) as well as at six-monthly intervals throughout the patients admission journey. These main areas will include structured interview measures of psychopathology e.g. SCID, PANSS, clinician rated measures violence and self-harm risk assessment instruments e.g. HCR-20, SRAMM, and measures of therapeutic programme completion and recovery (DUNDRUM-3 and DUNDRUM-4 scales) as well as overall functioning MIRECC GAF. Physical health measures will include measures of BMI, frailty and sedentary behaviours.

Other measures including ward atmosphere scales ESSENCES and measures of quality of life (WHO-QUOL) will also be offered to the patient group to complete.

A combination of both self-rated and interview based measures to engage patients, and clinician rated measures will be used. This is due to the high rates of treatment resistant psychoses in forensic hospital settings, and therefore to rely on self-rated or interview rated measures along would mean that many of the most unwell patients may not be able or willing to engage. Thus utilising this combination approach ensures that any bias in the results and outcomes observed will be minimised.

Statistical analysis plan:

Statistical methods for primary and secondary outcomes. Time intervals (time in days to admission from waiting list, length of stay) will be measured by median and 95% confidence interval using Kaplan Meyer and Cox regression analyses. Survival analysis will be used to assess factors affecting length of stay.

Patient level measures will be expressed as changes from baseline (typically from the time of admission) using ANCOVA, MANOVA or logistic regression as appropriate with correction for the effects of static baseline variables such as demographic characteristics (e.g. age, sex and ethnicity) or diagnosis including multi-axial diagnoses, neurocognitive function, and legal status. Measures of effect size will be used alongside measures of statistical significance for primary and secondary outcomes. At the individual level measures of reliable and clinically meaningful change will be calculated and proportions achieving this will be expressed with 95% confidence intervals.

Analysis of repeated measures will be used to compare mean scores on measures of physical health, risk of violence, therapeutic programme completion, recovery and overall functioning, at time points in advance of the move of the NFMHS to the new Portrane Hospital Campus and for five years after the move.

Lagged causal model analysis will be used to assess the existence and significance of potential directed relationships between the baseline measures of symptomatology of schizophrenia and violence risk and final outcome namely length of stay.

The lagged causal model approach will be used also to examine the relationship between amount of 'treatment' received and change in outcome measures.