Auditory hallucinations in the population: what do they mean and what should we do about them?

There has been a dramatic surge in research on auditory hallucinations over the past 15 years. Of the more than 100 papers identified in a recent meta-analysis of studies on definition, description and causes of auditory hallucinations (1), >90% were published since the year 2000. This glut of publications points to an upheaval that has taken place in psychiatry. Simply put, there has been a paradigm shift: a movement away from the Schneiderian view of auditory hallucinations as (predominantly) symptoms of psychotic disorder, towards an increasingly accepted view that these are experiences that occur in the full range of mental disorders and, indeed, none. Research on auditory hallucinations was galvanised in 2000 by the results of the Dunedin study, which showed that 11-year-olds who reported psychotic symptoms (predominantly auditory hallucinations) were at increased risk of psychotic disorder at age 26 (2). As a result of this and subsequent research, individuals in the community with psychotic symptoms came to be perceived as an epidemiological complement to the clinic-based ‘ultra high-risk’ approach to psychosis (3). Extensive epidemiological research since that time has hugely broadened this focus to show that individuals who report hallucinations are at high risk for a wide range of poor outcomes across the whole spectrum of mental disorders, but in particular multimorbid (non-psychotic) mental disorders, suicidality, poor socio-occupational functioning and neurocognitive deficits (4, 5). The findings that auditory hallucinations in the population are far more predictive of affective, anxiety and behavioural disorders (and indeed of no specific mental disorder) than psychotic disorders, have caused a conundrum for clinicians. If these symptoms are so non-specific, what (if anything) can we, or should we, do about them? In one of the longest follow-up studies to date, Bartels-Velthuis et al. (6) found that only 18% of children and adolescents with auditory hallucinations continued to experience these 6 years later, and only 6% a decade later. This demonstrates that hallucinations become a chronic problem for only a minority of children when followed to adulthood. Notwithstanding the low persistence rate of hallucinations over time, we know that a report of auditory hallucinations even at one point in adolescence is associated with poor mental health outcomes in a substantial proportion of individuals. This presents an important opportunity to identify a vulnerable group for early intervention. The question, then, is how do we improve our identification of the specific individuals who are likely to have poor outcomes and how do we differentiate them from auditory hallucinations that will take a benign course? Many clinicians suspect that specific aspects of the hallucinations are likely to identify those who are truly at risk of poor outcomes, including (but not limited to) pathways to psychosis. Just as there are clinical features of headache that point to risk of a non-benign course (e.g. nuchal rigidity and photophobia), so too might there be clinical features of hallucinatory experiences that would predict a non-benign course – be it, for example phenomenological aspects of the experience, degrees of reality testing, associated distress or functional impact. Both in research and in the clinic, however, characterisation of these aspects of hallucinatory experiences is typically poor and, without doubt, is poorly standardised from one site to the next. That is despite the fact that we have, in psychiatry, a rich descriptive system for hallucinatory experiences. As highlighted by Upthegrove et al. (1) in their recent meta-analysis, even when specific psychosis interview instruments are used, they take a rather atheoretical approach to determining what aspects of hallucinations to assess, ‘without clear evidence that these are either the aspects research should be focusing on or those which define clinically significant experiences from “normal” voice hearing’. At