Table 2 Meta-analyses based on individual patient data

Thase et al. (2007) [8]

6 placebo-controlled studies, 1833 patients

Remission rates, statistical significance and size of the interaction term (depression severity × treatment group) not reported

HAMD 15–18: duloxetine 46.5%, SSRI‘s: 51.7%, placebo: 42.7%

HAMD ≥19: duloxetine 35.9%, SSRI‘s: 28.6%, placebo: 17.7%

Yes, but not definitely

Fournier et al. (2010) [9]

Systematic review, 6 studies (paroxetine, imipramine), 434 patients in AD groups, 284 in placebo groups

Mild to moderate depression (HAMD ≤18): d = 0.11 (0.9 HPD)^a

Severe depression (HAMD 19–22): d = 0.17 (1.4 HPD)

Very severe depression (HAMD ≥23): d = 0.7 (3.8 HPD)

Yes

Khan et al. (2011) [10]

15 trials of one center, 262 patients treated with AD, 140 with placebo

HAMD score was a significant predictor of a reduction of depression scores for patients treated with AD, but not so for patients in the placebo groups. However, the statistical significance and the size of the interaction term (depression severity × treatment group) is not reported

?

Gibbons et al. (2012) [6]

Fluoxetine studies (Eli Lilly & Co), one study on adolescents, venlafaxine studes (Wyeth), total of 31 studies and 9185 patients

HAMD ≤20: 2.2 HPD

HAMD > 20: 2.8 HPD

Similar results for different AD and age-groups

No

Nelson et al. (2013) [7]

Second generation AD, 10 studies with 2283 older patients (≥60 years)

Significant effect only for the AD group. No statistically significant interaction between depression severity and treatment in the multivariate analysis. Differences of response rates for HAMD > 23 ≈ 18%, for HAMD 21–23 ≈ 8%, for HAMD 19–20 ≈ 12%, and for HAMD < 19 ≈ 0%. No mean-values are reported, except for the chronically depressed subgroup (d ≈ 0.7 for HAMD > 23 (5.6 HPD), d ≈ 0.4 (3.2 HPD) for HAMD 21–23, d < 0.1 (0.8 HPD) for HAMD < 21.

?

or only in one subgroup

Harada et al. (2015) [11]

4 studies with duloxetine and different SSRIs, total of 1694 patients

HAMD ≥15:1.4–1.5 HPD

HAMD ≥19: 2.1–2.2 HPD

No

Rabinowitz et al. (2016) [16]

34 studies with second generation AD or quetiapine (4 studies), total of 10,737 patients

HAMD < 22: 2.04 HPD

HAMD 22–25: 1.82 HPD

HAMD > 25: 2.41 HPD

No

Cuijpers et al. (2017) [12]

4 studies, total of 333 patients, SSRI vs. placebo vs. psychotherapy

Comparison of melancholic depression (with an increased HAMD score of about 1.5 points) with other types of depression. No significant interaction effects (0.53 HPD melancholic type vs. 0.33 HPD for other types of depression)

No

Debray et al. (2018) [13]

18 studies of older generation AD vs. Placebo, 2456 patients

HAMD = 21.8: 2.2 HPD

HAMD = 25: 3.1 HPD

?

Furukawa et al. (2018) [14]

Systematic review of pre-registered Japanese trials, 6 studies and 2464 patients

No significant interaction of depression severity and treatment group.

Ca. 1.6 HPD across the whole spectrum of depression severity

No

Nakabayashi et al. (2018) [15]

5 studies used for approval of AD in Japan, 1898 patients

No significant interaction of depression severity and treatment group.

HAMD 8–13: − 0.36 HPD; HAMD 14–18: − 1.50 HPD.

HAMD 19–22: 3.60 HPD; HAMD ≥23: − 1.26 HPD

No for most severely depressed, yes for HAMD 19–22