Sammendrag
Recently, psychiatric hospitals in Norway have been criticised for premature discharges and for poor assessments of patients’ risks of violence or self-harm. Due to the high turnover of patients and the obligation to admit all acutely ill patients who are in need of hospitalisation, the acute wards are particularly exposed to such criticism. Unstructured clinical judgements alone still appear to be the dominant approach to risk assessments. Currently used instruments for risk assessment are time consuming, and their use require special expertise. The goal of the “Risk Project” at Ålesund Hospital was early identification of patients in need of risk assessments. Various screening methods were tested in a prospective, naturalistic design. This dissertation is a part of the larger Risk Project. In the dissertation, different approaches towards risk analyses of violence, suicide and self-injury were examined: (i) Biologically based, in which lipids and serotonin are measured in the blood; (ii) Patient based, in which patients’ self-reports of risk (SRS) are employed, and (iii) Structured professional, by applying a violence screening instrument developed for use in acute psychiatry (V-RISK-10).
SRS and V-RISK-10 were scored both at the time of admittance and discharge. Blood was drawn at admission. These measures were then compared with the episodes of violent, suicidal and self-injurious behaviour recorded during the patient’s hospital stay and also, during the first year after discharge. The study sample for the biological markers and the SRSs included all of the acutely admitted patients during one year (n=489) in one acute psychiatric unit. The V-RISK-10 study included all admissions of two units (n=1017). The numbers of patients who completed the study were 254 during their hospital stay and 196 after discharge, all with regard to lipids/serotonin, 429 and 266 for SRS, and 980 and 381 for V-RISK-10, respectively.
Similar to findings from other studies, low concentrations of total cholesterol were found to predict inpatient suicidal and violent behaviour and also, violent behaviour three months after discharge. Low HDL levels were predictive of violence repeaters. High triglyceride levels predicted self-injurious behaviour (SIB) during the hospital stay, and both suicidal behaviour and SIB during the first year after discharge. The best results were obtained among patients who had recorded episodes of both suicidal behaviour and SIB. This finding contradicts prior studies on self-harm but concurs with three recent studies that look at the effects of short-term and chronic psychological stress. No significant correlations between serotonin levels and violence, suicidal behaviour or SIB were found in our sample.
SRS predicted violence, suicidal behaviour and SIB both during the hospital stay and at three months after discharge. In addition, SRS of violence was also significant at 12 months after discharge. The accuracy was higher for self-harm during the hospital stays and for violence after discharge. We found no other validation studies of patients’ self-reported risk of suicide or violence.
The predictive validity of the V-RISK-10 was equal to or better than those of the comprehensive risk assessment instruments, and it was very high in relation to severe violence. The screening instrument was sensitive to the risk of violence independent of gender and even for patients without any known history of violence. Despite significant findings for the lipids and SRS, these methods are not recommended for regular clinical use as single predictor variables, but they may be used together in combination with established risk procedures or clinical judgements. V-RISK-10 demonstrated good psychometric properties as a violence screening tool for acute psychiatry. Further research should be undertaken to confirm the findings.
Vis fullstendig beskrivelse
