Table 3.

Risk of bias assessment. Outcome data are considered complete if ≥90% of those randomized were included in outcome data.

First author (year)	Random sequence generation	Allocation concealment	Blinding of participants and personnel	Blinding of outcome assessment	Complete outcome data or intention-to-treat analysis used	All outcomes reported
Gardner-Nix (2008) [34]	No	No	No	No	No	Yes
Buhrman (2013) [29]	Yes	Yes	No	No	Yes	Yes
Davis (2013) [33]	Yes	Yes	No	No	Yes	Yes
Dowd (2015) [30]	Yes	Yes	No	No	Yes	Yes
Trompetter (2015) [31]	Yes	Unclear	No	No	Yes	Yes
Younge (2015) [35]	Yes	Yes	No	Yes	No	No
Ljótsson (2010) [36]	Yes	Yes	No	No	Yes	Yes
Ljótsson (2011) [38]	Yes	Yes	No	Unclear	Yes	Yes
Ljótsson (2011) [39]	Yes	No	No	No	Yes	Yes
Ljótsson (2014) [40]	Yes	Yes	No	Unclear	Yes	Yes
Thompson (2010) [41]	Yes	Yes	No	No	No	Yes
Thompson (2015) [42]	Yes	Yes	No	No	Yes	Yes
Hesser (2012) [43]	Yes	Yes	No	No	Yes	Yes
Johansson (2015) [44]	No	No	No	No	Noa	Yes
Zernicke (2014) [45]	Yes	Yes	No	No	Yes	Yes
Bruggeman-Everts (2015) [47]	No	No	No	No	Yes	Yes

^aIntention-to-treat analysis conducted with a subset of participants but not all.