Van der Post ThesisIn assessing the risk of involuntary admission, emergency psychiatric services evaluate formal criteria (diagnosis, possible danger and lack of motivation for treatment). In addition, however, psychiatrist and social-psychiatric nurses  supposedly take into account previous compulsory admission and referral by the police or mental health services. In the journal article, Van der Post and colleagues tentatively labelled the results as surprising. However, in his PhD-thesis the first author suggests that personal bias and external social pressure could be in the equation. In other words, the "mobile psychiatrists" are on the right track, but they have a tunnel vision and bent over backwards on their bycicles (whereas they should "keep their backs straight" - stand firm).

The analyses, however, give no grounds for these "accusations". Naturally, the prediction model classified 88% of all cases correctly, but that is not much of an improvement compared to 82% of patients that were not involuntarily admitted. Besides, the percentage correctly classified is inflated (read here why) and other information on the model's Goodness-of-Fit is lacking. And what's more, the results can be interpreted in a different way - there's an other interpretation of the variables in the model that fits the data quite well.

 

When we "turn over" the logistic regression model, the independent predictors can be interpreted as a cluster of characteristics of the risk assessment. A priory, we would expect more relevant variables than the three Severity of Psychiatric Illness (SPI) items that were used to operationalize three out of five judicial criteria: dangerousness, mental disorder, and noncompliance. The other criteria: the causal link between danger and disorder and admission as a last resort, are more difficult to grasp. However, when psychiatrist take these other two criteria seriously, this will be expressed in some other variables. In my view, psychiatric history and referral patterns are very likely proxies for the causal link between disorder and danger and for coercion as a last resort. In this interpretation of the results the emergency service does what it is supposed to do: assessments are based on judicial criteria. Not very surprising and no reason to chase these cyclists on suspicion of bias and back ache.

 

Emergency compulsory admission in crisis situations; the Amsterdam Study of Acute Psychiatry – L.F.M. van der Post, C.L. Mulder, C.M.H. Bernardt, R.A. Schoevers, A.T.F. Beekman, J.J.M. Dekker – Tijdschrift voor psychiatrie, 54(2012)04, 317-327.

Chapter 4 of Post, L.F.M. van der (2012). IBS admission as an outcome: Factors predicting the probability of patients qualifying for compulsory emergency admission. Dissertation Vrije Universiteit Amsterdam.

 

Reply to Wierdsma

In his reply Van der Post writes that he was looking for a “hypothetical explanation” of the increased risk of compulsory admission in patients referred to emergency services by the police or mental healt care service compared to patients referred by the general practitioner. Alternative explanations are difficult to disentangle and Van der Post refers to his thesis wherein he concluded: “It cannot be excluded that, by using the SPI items, we have not been able to make entirely perfect and comprehensive assessments of the factors severity of the disorder, danger and lack of motivation for treatment. The possibility should therefore be kept in mind that the effect of some part of those three factors can be attributed to the referral pattern through residual confounding. It is possible, for example, that the nature of the danger plays a role that is not covered by the severity of the danger assessed using the SPI. ...“. My suggestion that referral patterns could be proxies for the causal link and last resort criteria is mentioned as one of many possible explanations in Van der Post’s thesis. He ends his reply observing that “indeed, because of all these suppositions, hypotheses and related uncertainties there is no solid basis for criticizing the workers of emergency psychiatric services. On that point I completely agree with Wierdsma.”

 

Reply to a reply

Somehow it became custom to end a scientifc debate after an author’s reply. But not on this site and not when the author only agrees with a comment that was intended to be funny. Van der Post missed my point. I only suggested that the interpretation of the results should stay close to the statistical model. We don’t need fancy post hoc interpretations or explanations involving “residual confounding”.

 

Reaction on ‘Emergency compulsory admission in crisis situations; the Amsterdam Study of Acute Psychiatry. Wierdsma AI (2012) - Tijdschrift voor psychiatrie, 54(2012)9, 845-846.

 

Also see Van der Post et al. tested an impossible null-hypothesis

And Van der Post et al. support the classification table

And Van der Post et al. (2013) didn’t get patients in perspective