In a forensic psychiatric department Mudde et al. investigated if scores on the Historical Clinical Risk Management-20 (hcr-20) predicted aggressive behaviour. Risk assessment may be used to protect the hospital staff. Of 102 patients about 42% (43 patients) had caused one or more aggressive incidents as recorded by the Staff Observation Aggression Scale-Revised (soas-r).

On average HCR-scores differed 4 points (scale range 0-40) between aggressive versus non-aggressive patients. The authors calculated the area under the curve (auc)-value and in addition performed a “logistic aggression analysis” (something added in translation here). However, this is the world upside-down. The auc-value is a relative prediction assessment tool used as a general measure for the goodness-of-fit of a logistic regression model. It is not a useful measure of how well the model classifies individuals. And the model in this "aggression analysis" didn't perfom well.

Psychiatric ServicesVan der Post and colleagues studied lack of social support as a possible predictor of emergency compulsory admission. For a random sample of 252 patients (netto n=244) of two psychiatric emergency teams, data were collected on social networks, social interactions and the number of emergency compulsory admissions during a two-year follow-up period. 

The researchers concluded that of the social support variables, living alone was the only predictor of emergency compulsory admission and readmission. Patients who lived alone had fewer people in their social network (4.6 versus 6.1, p ≤.001) compared with patients who lived with others.

Among patients who lived alone, the percentage of patients with a compulsory admission was significantly higher among the patients with a high score for negative interactions than among patients with a low score (34% versus 13%, p≤.05).

However, the stepwise logistic regression analyses does not support these conclusions.

Geogieva thesisIn a study on seclusion and restraint in psychiatry in the Netherlands, Georgieva c.s. aimed to find an empirically and clinically relevant prediction model for identifying patients at risk. The authors conducted a forward stepwise logistic regression analysis, which will result in the smallest model, not necessarily a relevant model. But they claim to have found "a simple, accurate and highly predictive model". However, the data suggest a simpler, more accurate and better predictive model than the one reported.

The authors' model, including psychological impairment, uncooperative behavior and involuntary commitment, is not very accurate as the Nagelkerke R-square is .35, indicating a medium model-fit. And the model is not highly predictive. The paper states that the model predicted 72% of the cases correctly, based on the SPSS classification table in logistic regression which is supposed to show how many cases are correctly predicted. However, that model is not the best model that fits the data ...

Penterman and Nijman tested the predictive validity of the Checklist of Risks to the Crisis team (crc). This instrument was specifically designed to assess the risks of aggression from severely disturbed psychiatric patients before they are seen by the crisis team. Any perceived aggression following outreach contact was recorded by means of the Staff Observation Aggression Scale-Revised (soas-r).

The authors performed a forward stepwise logistic regression analyses, but reported no regression coefficients and CI's - only the less informative p-values. Three factors in the CRC were important: the clinical assessment on a visual-analogue scale, an estimate of the number of aggressive persons in the vicinity of the crisis patient concerned and the reporting of crises by the patient himself. Based on these indicators it was possible to predict outwardly directed aggression with a sensitivity of 74% and a specificity of 84%. However, these proportions are inflated because the same data was used to build the model and to evaluate the fit of the model (more background information here).

Aggressive behaviour by patients was observed during 51 of 499 crisis contacts with members of the outreach crisis team. In other words, care workers of the outreach psychiatric crisis team can expect aggression from severely ill patients in about 10% of all cases. If they expect non-aggressive contacts they will be right more often then the model.The authors conclude that an instrument such as the crc will help to get more experienced and more standardization in evaluating risk factors. But for that conclusion no hypothesis testing is warrented.

Penterman B, Nijman H. Assessing aggression risks in patients of the ambulatory mental health crisis team. Community Ment Health J. 2011 Aug;47(4):463-71. doi: 10.1007/s10597-010-9348-7. Epub 2010 Oct 1.

Penterman EJM, Nijman HLI.Assessing aggression in patients of the mental health crisis service. Tijdschrift voor psychiatrie 51(2009)6, 355-364