Can we prevent psychosis? The ineffective paradigm of indicated psychosis prevention

‘... progress in science is not a simple line leading to the truth. It is more progress away from less adequate conceptions of, and interactions with, the world’.

Thomas S. Kuhn, The Structure of Scientific Revolutions

Major improvements in the effectiveness of care for people with psychosis are essential, because these disorders still cause tremendous disability, despite advances in treatment to date. There are three sub-domains of primary prevention for mental disorders: universal prevention, selective prevention and indicated prevention (https://www.who.int/mental_health/publications/prevention_mh_2004/en/). Universal prevention targets the general public or a whole population. Selective prevention targets individuals or subgroups of populations whose risk of developing mental disorder is significantly higher than average. Indicated prevention targets high-risk people identified as having minimal or detectable signs or symptoms foreshadowing mental disorder or biological markers indicating predisposition to mental disorder. Universal and selective prevention for psychosis have not appeared possible to date. Accordingly, researchers and clinicians have explored the potential to prevent psychosis, through indicated prevention, which is targeted at people with minimal or detectable signs or symptoms foreshadowing psychosis. However, the weight of evidence to date indicates that the paradigm of indicated prevention has not been found to be effective. In this context, research into improved care for people with psychosis might benefit from new paradigms that address secondary (early detection and treatment of diagnosed disorders) and tertiary prevention (reduction of recurrence/relapses, disability and enhancement of rehabilitation).

Much time and effort has been expended on research into indicated prevention of psychosis. To intervene to prevent development of psychosis, such methods necessarily must effectively predict transition to psychosis in those at high risk of developing psychosis, to identify those to be treated. However, a systematic review of 91 studies of prediction of transition to psychosis found poor methodology and reporting (Studerus et al., 2017). This review concluded that most studies relied upon small sample sizes, failed to perform external/cross validation and utilised poor model development strategies, with the result that predictive models were overfitted and disproportionately optimistic (Studerus et al., 2017). It was also observed that due to deficiencies of the model presentation, there was insufficient detail to apply the models to other datasets, such that external validation could not be performed. A recent umbrella review also concluded that for people at chronic high risk of psychosis (CHR-P) semi-structured interviews for prediction of development of psychosis had high sensitivity (96%) but unfortunately low specificity (46%), reinforcing that prediction of transition to psychosis has not been effective to date (Fusar-Poli et al., 2020).

A recent umbrella review of 42 meta-analyses indicates that there is no evidence that any specific intervention is particularly effective compared to others, including control conditions, in preventing transition to psychosis (Fusar-Poli et al., 2020). Furthermore, there was no evidence supporting the superiority of any interventions affecting an array of outcomes for CHR-P including reduction of attenuated positive or negative psychotic symptoms, alleviating depression, improving symptom-related distress or quality of life, improving social or overall functioning, and acceptability (Fusar-Poli et al., 2020).

Furthermore, a Cochrane meta-analysis published subsequent to the completion of the review by Fusar-Poli et al. (2020) concluded that there was no evidence that interventions are successful in preventing transition to psychosis, remarking upon the disparate and poor quality of the evidence (Bosnjak Kuharic et al., 2019). The Cochrane meta-analysis usefully classified the interventions into comparisons of absolute effects (specific interventions directly compared vs placebo, for example, amino acids vs placebo compared to omega-3 fatty acids vs placebo), comparisons where complex interactions are probable (combinations of interventions compared, for example, comparisons of medication plus specific care packages vs treatment-as-usual [TAU] or active control), and comparison of differential effects (comparing specific psychosocial treatment interventions vs TAU or active control, for example, cognitive-behavioural therapy [CBT] vs supportive therapy). For absolute effects, that is, interventions compared to placebo, there was very low-quality (amino acids) to low-quality evidence (omega-3 fatty acids). For comparisons of combinations of medication, needs-based interventions, CBT and supportive therapy versus TAU or active controls, there was very low-quality evidence. Differential comparisons of CBT (with and without medication), supportive therapy, cognitive training, family and integrated therapy versus TAU or active controls similarly yielded very low-quality evidence.

Despite these empirical data indicating lack of effectiveness, there remain those who argue that indicated prevention of psychosis is nonetheless effective and worth ongoing effort. For example, some question the validity of the meta-analysis and Cochrane reviews findings, asserting that the well-established tools of evidence-based medicine are subject to errors and interpretation bias specific to research on persons at high-risk for psychosis (McGorry and Nelson, 2020). Furthermore, these authors continue to assert that ‘... it is possible to predict and delay the onset of fully fledged psychosis ...’ (McGorry and Nelson, 2020).

It should be acknowledged that the strong research interest and advocacy for effective care of people with psychosis has been a boon of the indicated prevention paradigm and that clinicians and researchers in that field have been striving towards improved care. However, despite immense research efforts, on the basis of the weight of these recent meta-analytic studies, it is concluded that the present paradigm of indicated prevention of psychosis is ineffective. The findings of the lack of efficacy of control conditions, some of which include existing TAU (Bosnjak Kuharic et al., 2019), indicate that improvements in current psychosis care are still needed.

Primary universal, selected and indicated prevention of psychosis appear to remain beyond the reach of our scientific knowledge at present. Through exploring new paradigms of research, there remains the potential of reducing prevalence by early detection of people with established/diagnosed psychosis and intervention (secondary prevention) as well as reduction of recurrence/relapses, disability and enhancement of rehabilitation (tertiary prevention). Based on the very substantial foundations of pioneering studies into indicated prevention by Australian and international researchers (Fusar-Poli et al., 2020), as well as the extensive research in treating early-diagnosed and first-episode psychosis, and reducing the duration of untreated psychosis (Oliver et al., 2018), extending these research efforts to early detection of people with established psychosis and reduction of symptoms and disability may more effectively assist with improving care presently. Thus, progress in care for people with psychosis can be built on the basis of improved conceptions of, and interactions with, the world’s current research evidence.