Oxelmark 2007.

Study characteristics
Methods	Study design: prospective RCT Study duration: The first intervention group (UC) was started in 1996, the second (CD) in 1997, the third (UC) in 1999, and the fourth (CD) in 2000. Setting: IBD‐outpatient clinic at the Karolinska University Hospital Huddinge, Stockholm, Sweden
Participants	State of disease at beginning of study per IG/CG: all patients were in remission or had low disease activity at inclusion, but numbers were not specified. Disease type per IG/CG: mix of UC and CD. UC: IG: 11; CG: 6 CD: IG: 13; CG: 14 Inclusion criteria: people with UC or CD who had had at least one serious flare and had been treated at least once with glucocorticosteroids (GCS) orally or intravenously. All patients had to be in remission or have only mild disease activity at inclusion. Exclusion: patients with high‐dose steroid treatment (more than 10 mg prednisolone or equivalent), blood in stools, and previous bowel surgery. Age at beginning of study per IG/CG: mean (range) years: IG: 36.3 (18‐71); CG: 38.5 (21‐59) Sex per IG/CG: Male: IG: 11; CG: 7; Female: IG: 13; CG: 13 Disease duration per IC/CG: mean (range) years IG: 4.6 (1‐11); CG: 5.2 (1‐10) Number randomised per IG/CG: IG: 24; CG: 22 Number reaching end of study per IG/CG: At 6 months: IG 18; CG 15. At 12 months: IG 20; CG 15
Interventions	IG: nine different sessions (once a week, each session lasted for 1½ hours) for about 3 months, with lectures alternating with group therapy. CG: patients in the control groups received conventional “on demand” medical and psychosocial/psychological treatment during the study period.
Outcomes	Duration of follow‐up: 12 months Primary outcomes as defined by study authors: HRQoL measured using the IBDQ and coping ability measured using the Sense of Coherence scale (SOC) Secondary outcomes as defined by study authors: patient evaluation of the intervention using a visual analogue scale (VAS)
Notes	Funding source: NR Conflicts of interest: NR
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Author response: "For each time point patients’ participation in either control or treatment group was decided by a simple lottery. Ten patients who had accepted to participate (for each time point) and were carefully informed that they would be allocated to either intervention or control group. The patient’s names were written separately on small paper notes. Each note was folded separately and placed into a hat, and one of the researchers then took out one note at a time without looking into the hat. Every second note was allocated to the control group and the others to the intervention group"
Allocation concealment (selection bias)	High risk	Patients were "pre‐allocated", as explained above by the author, and there was no concealment of the allocation list.
Blinding of participants and personnel (performance bias) All outcomes	High risk	It is not possible to blind the participants to the intervention.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Author response: assessors were not blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Reasons for dropouts were given and no imbalance between IG and CG
Selective reporting (reporting bias)	Unclear risk	No protocol or trial registration. Outcomes reported as per the methods section.
Other bias	Low risk	No baseline imbalance between IG and CG. No other concerns.