According to Drs Lee and Cleare ‘many clinicians’ regard the current evidence for lithium augmentation in treatment-refractory depression as ‘compelling’. They are correct in repeating one of the principles of evidence-based medicine, that all levels of evidence need to be taken into account when making clinical decisions.

Previous systematic reviews of this area have included patients who have had ≤3 weeks' treatment with an antidepressant or who have bipolar disorder. We do not think that many UK psychiatrists would consider lithium augmentation in unipolar depression that had not responded to an antidepressant for only 3 weeks. For patients with bipolar disorder, most UK psychiatrists, we think, would in any case be treating with lithium or another mood-stabiliser. Our inclusion criteria, which were set before the review started, were based therefore upon sensible and pragmatic clinical considerations.

We too were surprised and shocked by the lack of randomised evidence to support lithium augmentation; but it is also important to remember that lithium may well be effective, even though the evidence to support its use is extremely weak.

Lithium has a number of potentially serious side-effects, even at normal therapeutic doses (Reference Bell, Cole and EcclestonBell et al, 1993). When we discuss the advantages and disadvantages of lithium with our patients we are unable to provide them with much more than clinical anecdote in its favour. We certainly have no idea from empirical research about the severity of depression for which lithium augmentation might be effective.

We have a collective responsibility to our patients to provide them with good-quality research evidence to justify the treatments we recommend. As a profession we need to address areas of uncertainty such as this using well-designed RCTs that will inform clinical practice.