Invited commentary on … NICE v. SIGN on psychosis and schizophrenia

Summary

How contemporaneous guidelines interpret similar data is interesting but not relevant here. Conflicts of interest are important and various. Two clinically important recommendations in the National Institute for Health and Care Excellence (NICE) guideline CG178 deserve re-visiting.

How contemporaneous guidelines interpret similar data is interesting but not relevant here. Switching the discussion from an analysis of the National Institute for Health and Care Excellence (NICE) guideline CG178 ¹ to a comparison of CG178 with the Scottish Intercollegiate Guidelines Network guidance 131 ² (SIGN 131) is simply an obfuscatory diversion. Our original editorial ³ only mentions SIGN 131 twice – once in conjunction with the British Association for Psychopharmacology schizophrenia guideline – and we did not ‘claim’, as Kendall et al ⁴ wrongly state, that SIGN 131 ‘was unbiased’. Furthermore, SIGN 131 was not an ‘update’ or ‘largely based’ on the 2009 NICE guideline CG82, ⁵ as also stated twice, beyond noting that NICE did kindly furnish SIGN 131 with their meta-data to 2007, on which CG82 was derived. The SIGN 131 multidisciplinary group then conducted their own systematic reviews and interpretation programme of any subsequent relevant literature. SIGN commenced in 1993 and lent its methodology to NICE when they began in 1999, so it is unclear which is ‘more rigorous’.

Conflicts of interest and bias

Impugning the opposition is another classic debating trick. Kendall et al suggest that we have ‘succumbed to bias’ without further explanation. Examining the basis or validity for the recommendations in CG178 does not mean the questioner is biased. To question is to be scientific, and our editorial merely states that CG178 is ‘open to a critique of bias’. Conflicts of interest are important in any discussion of bias, and can be varied. ^6,7 Potential conflicts of interest here might include: training fees, grant applications and publication royalties for the recommended interventions; researcher or guideline allegiance; receiving money from NICE to promote NICE; and putting forward research questions that mirror the authors' own career interests. NICE itself ⁷ and others ⁸ have expressed concern regarding the governance of NICE guideline groups.

Recommendations re-visited

Two clinically important CG178 recommendations deserve re-visiting. First, authorities ⁹ regard at-risk mental states (ARMS) – for which CG178 has a whole new chapter – as an unreliable category with little predictive validity. Kendall et al state that CG178 recommended to ‘offer CBT for people with ARMS’ based on a meta-analysis they themselves conducted, without commenting on the ARMS construct validity, fidelity of the intervention or resource implications.

Second, advocating cognitive behavioural therapy for psychosis (CBTp) as sole therapy (i.e. no medication of any sort) during the first month of first-episode psychosis arguably lacks clinical face validity and could potentially be dangerous given the high suicide rate in this population. ¹⁰ Extrapolating recommendations from ’sparse’ data is also ill-advised. Moreover, the practical issues of what happens when CBTp is refused or not available locally are simply not addressed by CG178.

In a serious disorder such as schizophrenia, all effective treatments are welcome. CBTp may well have a small benefit (effect size 0.2) in psychosis, ¹¹ although recently there have been concerns regarding the reproducibility of psychology studies. ¹²

Our original editorial used tendentious language with the aim of provoking rational debate – a pillar of scientific progress. It seems we have partially succeeded in that aim.