A critical assessment of NICE guidelines for treatment of depression

Contradictions, methodological flaws, and potential for misinterpretations in ranking treatments of depression

“…Consistent with these results a comprehensive meta‐analysis available to both the NICE committee and Mahli et al did not find CBT to be superior to other psychotherapies in depressive disorders 5 As described in the previous article, 3 the cost‐effectiveness analyses additionally used by NICE to establish a hierarchy of treatment recommendations 2 may not be valid as well, since they were based on the questionable indirect comparisons described above 2 Finally, an explicit link between evidence and recommendations is missing.…”

“…The reasons for this omission are not clear. In less severe depression the NICE direct comparisons found a few clinically significant differences including only one for STPP, which showed a statistically and clinically significant superiority of STPP over counseling (SMD = − 0.61, 95% CI −1.05, −0.17) 2,3 . NICE, however, ranked STPP below counseling 2 .…”

“…Furthermore, as described in more detail elsewhere, 3 the validity of the NICE treatment ranking itself is questionable because of the methodological procedures applied by the NICE committee and the committee's interpretation of results. Our concerns about the NICE treatment ranking can be summarized as follows: Based on effect sizes between each treatment and placebo or treatment as usual (TAU), the NICE committee concluded that some treatments appeared to be more effective than others 2 .…”

“…Our concerns about the NICE treatment ranking can be summarized as follows: Based on effect sizes between each treatment and placebo or treatment as usual (TAU), the NICE committee concluded that some treatments appeared to be more effective than others 2 . However, assuming differences between two treatments if one shows a descriptively larger effect size compared to a control condition than another treatment without comparing the effect sizes directly by a statistical test depicts a statistical flaw 3,4 Even following the NICE committee's approach, the differences between the treatment‐control effect sizes of STPP compared to CBT or SSRIs (0.16–0.25) are below the minimal clinically significant difference defined by the NICE committee itself (SMD, standardized mean difference = 0.50) 3 .…”

“…However, assuming differences between two treatments if one shows a descriptively larger effect size compared to a control condition than another treatment without comparing the effect sizes directly by a statistical test depicts a statistical flaw 3,4 Even following the NICE committee's approach, the differences between the treatment‐control effect sizes of STPP compared to CBT or SSRIs (0.16–0.25) are below the minimal clinically significant difference defined by the NICE committee itself (SMD, standardized mean difference = 0.50) 3 . With differences in effect sizes below the threshold of clinical significance defined by the NICE committee itself, treatment ranking carries considerable uncertainty 3 …”