Outliers and inliers

‘There can scarcely be a surer sign of the importance attached to scientific statistics, than the anxiety with which they are received, and the estimation in which, if worthy, they are held’ ^{Reference Hawkes1} (p. 516). So wrote Dr John Hawkes in one of the first articles describing statistics in the British Journal of Psychiatry. Unfortunately, we have not always remembered this wise dictum and have sometimes been taken to task for the quality of our statistics. ^{Reference McGuigan2} One reason why our statistics may not be top notch is that most of our readers do not like them very much and quickly jump over the analysis section of most papers as though each was written in a foreign language, one that they dimly remember learning at some time in the past but without ever really grasping its vocabulary and syntax. But do not think you can completely escape statistics by reading this column instead of the main papers. Outliers, in statistical terms, constitute data that are numerically too distant from the rest of the data-set to be regarded as part of normal variation for that data-set. To detect an outlier you can use the conveniently named Grubbs’ test. ^{Reference Grubbs3} This test might identify several outliers in this issue. Barry et al (pp. ) describe a form of psychosis, Patel et al (pp. ) show the value of collaborative stepped care, and Farooq et al (pp. ) describe the results of adherence therapy in schizophrenia. The subject matter is not unusual; what makes each of these papers outliers is their context. Much argument has been spent over the early development and fundamental origin of psychoses; ^{Reference Barrett, Mulholland, Cooper and Rushe4–Reference Taylor7} anti-NMDA receptor encephalitis unequivocally is a psychosis but with a clear and unambiguous cause. The outlying aspect of collaborative care in Patel et al’s paper is the use of lay health counsellors as the main therapists for the treatment of depression and anxiety, and Farooq and colleagues go one step further to involve a family member as the main ‘adherence supervisor’ in their exciting trial, a very different strategy from that used in the UK. ^{Reference Staring, Van der Gaag, Koopmans, Selten, Van Beveren and Hengeveld8} So if these papers are so outside the mainstream, why do we bother to publish them? Put simply, they convey a message for the rest of the data-set: ‘I may be an outlier but that does not mean I can be dismissed; what I describe may be relevant to you, if not now, at least in the not too distant future’.

Other papers in this issue are clearly inliers: they are right in the mainstream. These include Oyebode & Humphreys’ (pp. ) challenging editorial, and the papers by Meadows & Bobevski (pp. ) and Wahlbeck et al (pp. ), as they could not be closer to the middle of a normal distribution, describing as they do the mental health of nations and the future of our subject. Although they might seem to have contrasting messages they are more similar than they at first appear. Good, professionally led mental health services in countries that are relatively stable in socio-economic terms lead to better outcomes, but in Australia the results do not appear quite as good, as those who get more tend to ask for more, and the big effort to promote prevention of mental illness there has had limited success. ^{Reference Walker, Mackinnon, Batterham, Jorm, Hickie and McCarthy9} Our Scandinavian colleagues deserve a small pat on the back for getting on with mental health reform quietly and efficiently compared with some of their more flamboyant European cousins, but the Wahlbeck data show that Swedish men cannot completely escape from the stereotyped view, created by the films of Ingmar Bergman, that one of their favourite pastimes is playing chess with Death. This is a pity, and an error, as I’m sure Ingmar was a true outlier.

Recovery and rehabilitation

The recovery model has now taken hold in psychiatry but I have never been sure whether this is a triumph of policy over science, or vice versa. Perhaps the term ‘rehabilitation’ had been damaged in psychiatry by its former association with industrial mental health units attached to old mental hospitals, but the ‘recovery model’, although now widely embraced, is overused and detracts from the rare and exciting occasions when it genuinely happens against a setting of hopelessness. ^{Reference Bewley10} David Brunskill’s apt comment that the recovery model’s scope ‘can make a cow-catcher on the front of a road train look discriminating’ ^{Reference Brunskill11} chimes with my own view but a few more articles such as that by Leamy et al (pp. ) may help to change my mind. True, a narrative synthesis is only the first step on its road to rehabilitation, but a few more in the same vein, especially if they can provide solid data to show achievement of agreed outcomes, could help me to change my mind. This is clearly not going to be at all easy. The personal recovery journey often seems to be a tortuous one, and the ending unplanned. ‘A man does not know how far he has to go until he starts walking’, runs a Ghanaian proverb, and before he disappears into the jungle and is lost to oblivion we need to give him some sense of direction without being accused of taking over control.