|
| Study | Adequate sequence generation | Allocation concealment | Incomplete outcome data addressed | Blinding of assessors | Blinding of participants |
|
| Bichescu et al., 2007 [28] | Yes: assignment through a random selection procedure (name-cards) to either NET or PED group | No | Yes: no dropout reported among participants who started trial | No: an attempt was made. Blinding was finally impossible due to the large differences in procedures and number of sessions between the 2 groups | No: an attempt was made. Blinding was finally impossible due to the large differences between the two groups |
|
| Hinton et al., 2004 [17] | Yes: participants were all randomly assigned but the method was not described | No | Yes: no dropout reported among participants who started trial | No: no attempt was made | No: no attempt was made |
|
| Hinton et al., 2005 [24] | Yes: patients were stratified by gender with random allocation to either the IT or DT group decided by a coin toss | No | Yes: no dropout reported among participants who started trial | Yes: but blinding’s integrity was not tested | No: no attempt was made |
|
| Neuner et al., 2004 [16] | Yes: patients were randomly assigned to either NET, SC or PED group by using a dice | No | Yes: missing data were estimated with a restricted maximum likelihood procedure. | Yes: interviewers were blinded for participant’s treatment condition. | No: no attempt was made |
|
| Neuner et al., 2008 [30] | Yes: patients were randomly allocated to a group by altering allocation of randomly ordered participants. However, method was not described | No | Yes: but partly. Authors reported a high global attrition rate, 23%, 53.1% and 61% at, respectively, 3 months, 6 months and 9 months. Authors chose to apply mixed-effects models instead of a last-observation-carried-forward (LOCF) procedure, considered too conservative. | Yes | No: no attempt was made |
|
| Neuner et al., 2010 [29] | Yes: participants were randomized to NET or TAU group with a block permutation procedure with blocks of 4 patients | No | Yes: low dropout rate (6.3%). Authors used mixed effects models instead of an LOCF procedure to handle missing data. This method did not probably introduce a significant bias because of the small number of drop outs (2) | No: blindness could not be maintained in all cases so we cannot rule out an assessor bias | No: no attempt was made |
|
| Otto et al., 2003 [25] | Yes: participants were randomly assigned to either Sertraline alone or Sertraline + CBT, but method was not described | No | Yes: no dropout was reported during trial | No: no attempt of blinding assessors was made | No: no attempt was made |
|
| Ertl et al., 2011 [31] | Yes | No | Yes: mixed effects model was used | Yes: psychologists were blinded to treatment conditions | No |
|
| Hinton et al., 2009 [27] | Yes: random allocation by a coin toss | No | Yes | No | No |
|
| Kruse et al., 2009 [26] | Yes: first 35 patients assigned to intervention group | No | Yes | No | No |
|
