Unsystematic review shows neither that early intervention in psychosis is cost-effective nor cost-minimising

Aceituno et al’s systematic review of economic evaluations of early intervention in psychosis (EIP)¹ is an example of the use of spin to misrepresent advantages of EIP, a topic reviewed elsewhere.² Although it refers to standard protocols for systematic reviews, it does not critically analyse the collected articles, leading to wildly optimistic conclusions.

I provide several examples, noting my earlier review that critically analysed the papers extant at the time.³ Aceituno et al report a cost-effectiveness study based on cognitive–behavioural therapy to prevent transition to psychosis as a positive study, without reference to the fact that the current consensus is that it is not possible to prevent transition to psychosis.⁴ If a treatment is not effective, it cannot be cost-effective.

The first paragraph of Aceituno and colleagues’ discussion concludes: ‘Investing in EIP could, as the best-case scenario, save money and is at least a more cost-effective alternative than treatment as usual’. As it reviewed cost-effectiveness articles, the article did not provide evidence on whether EIP saves money. Indeed, the review excluded one study explicitly for its cost-minimisation approach, the relevant type of study for assessing whether an intervention can save money.

Although Aceituno et al note significant methodological limitations in this literature, the article does not analyse the fact that 11 of 14 studies demonstrated ‘Selective reporting (reporting bias)’, according to the Cochrane risk of bias tool. Alongside the comment: ‘…more rigorous trials have failed to demonstrate clinical or functional differences with standard care…’ this lack of critical interest in evidence of systematic misreporting suggests that Aceituno et al have not actually scrutinised the literature, but only followed protocol.

In addition to ignoring evidence of systematic bias, the failure to identify limitations of specific articles should convince readers of the value of this review. For example, using service records, Tsiachristas et al identified all patients with psychosis in several regions of England, then compared treatment costs of patients managed in EIP units with those of patients in non-EIP units.⁵ As there was no matching on duration of illness or treatment, essentially this study compared the costs of treating patients in the first few years of treatment (EIP) with the costs of treating patients with established, chronic illness. Given Aceituno et al do not mention this extreme confound, it seems fair to wonder what level of methodological compromise would have been enough to conclude that the literature cannot be relied upon.