Chan 2006a.

Methods	Country: Hong Kong, China Setting: hospital (paediatric wards and outpatient departments) RCT
Participants	1483 mothers of sick children admitted to the ward or attending the outpatient department from all participating trial centres, November 1997 to September 1998
Interventions	Intervention: Mothers received information from nurses including standardised health advice, booklet about preventing child exposure to passive smoking, booklet to give to fathers on quitting smoking, a no smoking sign to place in the home to remind the father not to smoke, and a telephone reminder 1 week later. Control: normal care by nurses
Outcomes	3‐, 6‐, and 12‐month follow‐up: • Mother self‐reports actions taken to reduce child passive smoke exposure.
Type of intervention	Child with health problems (ill‐child health care)
Notes	Retention: 1273/1483 (86%)
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Random numbers were generated by the investigator using the computer and assigned to intervention (even) and control (odd) groups."
Allocation concealment (selection bias)	Low risk	"Nurses then randomized the subjects into the intervention or control group by opening a sealed envelope with serial numbers."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Low loss to follow‐up, ITT analysis used, similar percentage lost in both groups: 86% intervention and 85% control retention
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Self‐report only; differential misreport possible, but no difference found between groups, so unlikely
Other bias	High risk	Contamination of the control group possible: open ward setting "...the mothers in the control group could have by chance read the health education booklet from the mothers in the intervention group... furthermore, the nurses' health education could be easily overheard."