How well do antidepressants work in older people? A systematic review of Number Needed to Treat

Objectives: To make direct comparisons between studies of antidepressant treatment of older people using Number Needed to Treat (NNT) analysis. Methods: Medline and Embase Search 1966–1999 and contact with manufacturers of antidepressant drugs asking for any relevant data on file; NNT analyses of outcome in terms of efficacy and adverse effects. Results: Most antidepressant trials show efficacy; this is less clear for moclobemide and fluoxetine. Head-to-head comparisons between antidepressants showed significant superiority for paroxetine over fluoxetine and a trend in favour of SSRIs and venlafaxine over tricyclics. Conclusions: NNT analysis is one way of providing intelligible information on antidepressant efficacy and adverse effects which can inform clinical decisions. Limitations: Many studies do not present data in a form amenable to NNT analysis. Most head-to-head comparisons between antidepressants are underpowered.