Evidence-based medicine in general often feels like a quagmire. Reference WallaceWallace (2011) addresses the perils and pitfalls of evidence-based psychiatry in particular. Going through journals such as The Psychiatrist and the British Journal of Psychiatry, a nervous reader might become bogged down in the statistical methods. Twenty years ago, t-tests and ANOVA were common; today, we have SPSS computer programs and all the baffling complexities of logistic regression and factor analysis (Reference Howitt and CramerHowitt 2008).

What articles such as Wallace’s attempt is a perhaps impossible task – to make the ultra-complicated welding of mathematics and medicine comprehensible.

Statistics, as much as theology, is prone to internecine debate, with different contingents disagreeing over the concepts involved. Wallace refers to both P-values and confidence intervals: nowadays, the former have been replaced to a large extent by the latter (Reference Gardner and AltmanGardner 1990). Wallace’s statement that the ‘number needed to treat is regarded as the most useful measure of the benefit of a treatment’ is open to debate. The number needed to treat is still embedded in some controversy, and many prefer to rely on the underlying absolute risk difference.

So how much is evidence-based psychiatry like hazardous quagmire? Numbers and data per se are not enough. The psychiatrist without emotional insight and an intuition of the heart regarding each individual patient is a bad psychiatrist. Forcing psychiatry into the mould of the computer through statistical methods is not always for the best.