Table 4

Main effect of treatment and treatment by recruitment source interaction on estimated mean change in self-rated depressive symptoms (PHQ-9) from baseline to post and follow-up and minimally clinically relevant change of PHQ-9 at post-assessment. The sum of participants in the treatment groups (intervention plus CAU) is smaller than the number of participants in each subgroup because some participants did not complete the post or the follow-up assessment and could therefore not be included in the main analysis. A sensitivity analysis using multiple imputation to replace missing values yielded essentially the same results as the main analysis

Recruitment sourceTreatmentEstimated changeEffect SizeMinimally clinically important change of PHQ-9Effect size
MSE95% CId95% CIn%NNT95% CI
Main effect of treatment F1,823 =21.84, p<0.001B=1.18 (SE 51), p=0.021
All (n=1013)Intervention (n=509)2.940.212.53 to 3.350.300.01 to 0.5914328.196 to 15
CAU (n=504)1.510.211.09 to 1.928316.5
Treatment by recruitment source interaction F3,817 =0.29, p=0.83B=−0.16 (SE 16), p=0.31
Clinical (n=105)Intervention (n=42)3.300.542.24 to 4.370.27−0.49 to 1.041729.85−3 to 107
CAU (n=38)2.350.581.21 to 2.49816.7
Internet forums (n=171)Intervention (n=63)2.200.461.30 to 3.100.39−0.23 to 1.012328.053 to 15
CAU (n=62)0.810.45−0.07 to 1.69910.1
Insurance (n=271)Intervention (n=106)3.240.352.56 to 3.930.480.00 to 0.964029.974 to 39
CAU (n=113)1.520.340.85 to 2.192619.0
Other (n=466)Intervention (n=184) to 3.520.460.09 to 0.836326.785 to 28
CAU (n=186)1.340.270.81 to 1.874017.4