We read with interest the commentary by Roose et al regarding number needed to treat (NNT) and the concern that this metric is difficult to interpret given the high placebo response rates observed in contemporary clinical trials. ^{Reference Roose, Rutherford, Wall and Thase1} The principal objection of Roose and colleagues is that ‘NNTs derived from clinical trials are not directly relevant to clinical decision-making, because they are based on control conditions that do not exist in standard practice’. Although we agree that this can limit the utility of NNTs from some studies, we contend that NNTs commonly remain ‘indirectly’ relevant, as explained below.

Indirect comparisons of effect sizes among different medication choices can be quite helpful in ranking interventions for both efficacy and common tolerability challenges, provided that the studies used for these calculations are similar enough. Number needed to harm (NNH) values may be even more helpful when distinguishing among treatments that are relatively otherwise similar. ^{Reference Ketter, Miller, Dell'Osso, Calabrese, Frye and Citrome2} The NNH can be for overall tolerability (discontinuation because of an adverse effect) or the occurrence of specific adverse effects of concern for individual patients being treated (such as sedation, weight gain or akathisia). Moreover, ratios of NNH to NNT can provide overall estimates of the risk–benefit trade-offs involved. Finally, we suggest that all of the above concepts are straight-forward enough for average clinicians to calculate and understand. ^{Reference Citrome and Ketter3,Reference Citrome and Ketter4}