Heymen 2009.

Methods	Randomised controlled trial. Analysis: intention to treat.
Participants	Setting and country: single centre. Tertiary referral hospital in USA. Inclusion criteria: consecutive series of "chronically incontinent patients referred to University of North Carolina Hospitals". At least 1 teaspoon of faeces at least weekly; anatomic defects on anal ultrasound permissible. Exclusion criteria: flatus incontinence and staining only; psychotic disorder and severe cognitive impairment Number of participants: 168 randomised. After run‐in period of conservative management, 108 started intervention as randomised. Age: average 59.6 years Sex: 83 female, 25 male Other characteristics of participants:
Interventions	PFMT alone 63 participants. PFMT plus manometric biofeedback 45 participants. Both groups taught behavioural strategies to avoid incontinence.
Outcomes	Primary: Faecal Incontinence Severity Index (FISI) Secondary: Days per week incontinent; complete continence (no staining); squeeze pressure and abdominal tension; Faecal incontinence quality of life score (FI‐QoL); telephone report to therapist of "adequate relief"
Notes	Randomised patients prior to run‐in period which resulted in unbalanced group sizes: 35 patients reported adequate relief from run‐in and 24 withdrew, resulting in 109 patients only entering treatment phase.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Co‐investigator produced the randomisation table using random number generator
Allocation concealment (selection bias)	Unclear risk	Group membership was reported to the therapist after patient arrived for initial visit. Patients randomised prior to unblinded run‐in phase
Blinding (performance bias and detection bias) All outcomes	High risk	Therapist telephoned participant to ask if they had experienced "adequate relief"
Incomplete outcome data (attrition bias) All outcomes	Low risk	Intention‐to‐treat analysis reported: authors replaced missing values with zero