Table 1.

Included studies—characteristics and risks of bias

First author (year)	Country	Setting	Participants	Participant age	Study design	Follow-up	Risks of bias
Surrat (2010)	USA	Community	806 410 intervention 396 control	36.8 years (mean) (SD 8.2)	RCT	3 and 6 months	No detail of randomisation or allocation process No blinding of participants or personnel Self-reported non-blinded outcome measures Unclear whether analysis decisions were prospective eg, dichotomisation of outcome results Follow-up rates calculated by combining attendance at either of follow-up appointments which increased follow-up rates Intention to treat analysis not undertaken
Litchfield (2010)	UK	Community	34	Not stated	Before/after clinical records	12 months	All participants received intervention and no control group Self-reported non-blinded outcome measures Missing data not accounted for
Sherman (2006)	USA	Community	54	39 years (median) (IQR 34–45)	Before/after survey	3 months	All participants received intervention and no control group Self-reported and non-blinded outcome measures Missing data not accounted for
Yahne (2002)	USA	Community	27	37.8 years (mean) (SD 8.1)	Before/after survey	4 months	All participants received intervention and no control group Self-reported and non-blinded outcome measures Follow-up in settings likely to influence outcomes and act as confounder
Bellis (1993)	USA	Outpatient	41	31.8 years (mean)	Before/after	12 months	All participants received intervention and no control group Self-reported non-blinded outcome measures No data on participants lost to follow-up
Gunne (1986)	Sweden	In-patient	34	28.1 years (mean) (SD 5.8)	Before/after clinical records	1–220 months	All participants received intervention and no control group Self-reported non-blinded outcome measures