O Fortuna

We must have confidence in what we read in scientific journals. We accept varying methodologies, study limitations and differences of professional opinion; we debate peer-review processes and conflicts of interest; but the science underneath must be robust. Thankfully frank scientific fraud is relatively rare, but more common is concern about the way in which some trials are undertaken and reported. Nevertheless, articles are retracted relatively rarely, and normally with much consideration. This month our Editor-in-Chief leads an editorial (pp. 381–382) describing an account of one such recent retraction in the BJPsych, reader feedback having identified some concerning issues peer reviewers missed. Having sat through several journal board meetings where this was discussed at length, I was struck by how complex the underpinning issues were, how seriously the Journal took these, and our prime commitment to have confidence in what we publish. The BJPsych has implemented changes going forward, including proposed training for researchers. Describing what occurred, including failings in our processes and proposed redress, are part of the transparency we believe are crucial in such instances.

In the UK, the Royal College of Psychiatrists is reviewing its postgraduate curriculum, in line with General Medical Council direction. How would you shift it to better meet contemporary training needs? A future editorial will argue for an enhancement of evolutionary biology training, and this month analysis by Steele & Paulus propose (pp. 404–408) a more pragmatic approach to teaching neuroscience. It is hard not to be struck by their comment on how jarringly neuroscience advances in the last half century contrast with our somewhat static clinical practice over the same time. Their call to develop objective biomarkers to assist clinical decision-making is well made, and they provide several thoughtful examples of how this might occur. An excellent piece for trainees and supervisors to share, I would suggest. It also aligns well with the Royal College's Wellcome Trust and Gatsby Foundation programme, led by our President, and featuring initiatives such as the ‘Brain Camp’ teaching days for trainees. We are producing a podcast on this area with Professor Steele and the Mental Health Foundation that will be released soon.

Velut luna

Steele & Paulus might appreciate two papers in this month's Journal that look at clinical risk markers. Using schizotypal and behavioural scales, Stanfield et al (pp. 422–427) assessed adolescents with special educational needs (SEN), and prospectively followed them up over 6 years. Compared with matched controls without SEN, they were more symptomatic on the Positive and Negative Symptom Scale for psychosis throughout. Those above schizotypal and behavioural cut-off scores at baseline were significantly more likely to display positive symptoms and over a quarter developed psychoses – which is five to ten times the rate seen more broadly in those with intellectual disability. The authors propose this relatively simple screening may help identify those who require more monitoring.

Francesca Solmi and colleagues note (pp. 428–433) how there has been some evidence showing a shared genetic liability between psychoses and eating disorders, but this has generally been overlooked in the wider literature. Exploring the UK Avon Longitudinal cohort (ALSPAC) they found, in the 7000 or so with available genetic data, a significant association between polygenetic risk scores for schizophrenia and self-reported eating disorders and body mass index at ages 14, 16 and 18. The authors remark upon some shared phenotypes between these conditions, notably impaired social cognition and irritability, and posit that the genetic liability might also account for the greater rates of metabolic syndrome in those with schizophrenia.

Early intervention in psychosis (EIP) services are universal across the UK and much of the world, and their utility is usually just taken as read. Aceituno et al (pp. 388–394) challenge orthodoxy, saying that their evidence base is not definitive, certainly with regards to cost-effectiveness. They systematically reviewed the economic evaluations from 16 suitable EIP services, most of which modelled cost versus standard care, although a minority also included quality-adjusted life-years. The findings support cost-effectiveness and were replicated across healthcare systems, but strongest in high-income countries. The authors note the heterogeneity and flaws in the existing literature on the topic; further, few tackled more meaningful social recovery aspects, and the wider societal costs to carers, social care and the criminal justice system have largely been excluded.

Statu variabilis

Like elbow patches on jackets, lithium is effective but unfashionable. Lyall et al (pp. 415–421) decide to replace anecdote with evidence, and report on actual prescribing data in bipolar disorder. Their sample was the electronic Scottish Morbidity Records from 2009 to 2016, encapsulating a cohort of over 23 000 patients. Antidepressant monotherapy – you know, that thing all guidelines consistently caution against doing – was the most common treatment, occurring in about 25% of cases. Lithium, however, does indeed appear to be going the way of corduroy; it was the fifth most common medication prescribed for bipolar disorder (under 6%), and the only one that showed a decrease in use over the time measured. Deenan Edward and Su Ahmed from South West London and St George's NHS Trust write more in this month's Mental Elf blog: https://elfi.sh/bjp-me18.

Chen et al (pp. 409–414) report on a retrospective analysis of a cohort of almost 20 000 Taiwanese patients on mood stabilising medication to evaluate any association with cardiovascular events. Both carbamazepine and valproate were associated with a significantly higher risk of stroke in those with bipolar disorder, but guess which amazing ion was not? Lithium has the best effectiveness data of any mood stabiliser, has evidence for antisuicidal properties, and newer data show fetal and renal risks to be lower than erstwhile reported. One for your peer-group reflective practice?

On our spectrum model, bipolar disorder's similarities to both psychoses and borderline personality disorder have long been noted. Valli et al (pp. 383–385) explore the former, noting that bipolar disorder has traditionally been less considered as a neurodevelopmental disorder. The editorial describes how cluster analyses show different cognitive profiles in bipolar disorder, inferring that there may be different illness developmental subtypes. The question opens as to whether this will translate to clinical utility. Yu et al (pp. 395–403) tacked the bipolar disorder/borderline personality disorder data, specifically grey matter volume and density. Meta-analysis of 60 relevant studies reveal quite distinct patterns between the two conditions that, the authors argue, run against a continuum argument.

Finally, Kaleidoscope (pp. 437) explores the neuroscience of the causes for which you are willing to die. Quod per sortem sternit fortem, mecum omnes plangite.