Peng 2014.

Study characteristics
Methods	RCT Unit of randomisation: hospital
Participants	Place of recruitment: hospital Numbers randomised: total ‐ participants 3821; I: 1795; C:2026, hospitals I: 23; C: 24 Completing final follow‐up: 1 hospital withdrew before the study began Inclusion criteria: > 18 years old, proven ischaemic stroke confirmed by CT or MRI, TIA, hospitalisation within 30 days after the index event; clinical stability, independence in daily activities Exclusion criteria: CT/MRI evidence of intracerebral haemorrhage, stroke/TIA unrelated to atherosclerosis, severe co‐morbid illness/unstable medical condition, significant memory/behavioural disorders requiring daily care, concurrent participation in another clinical trial, pregnancy Type of stroke (%): not reported Mean age (SD): I: 61.48 (11.47); C: 60.36 (11.66) Gender (% men): I: 67 C: 69 Ethnicity: not reported Socio‐economic or socio‐demographic status: not reported
Interventions	Intervention details (components, length, frequency): the intervention consisted of lifestyle modification with the patients, including smoking cessation, healthy diet, and regular exercise. Patient education included an interactive website based education session emphasising the importance of adhering to the SMART program including information discussing risk‐factor control through medication and lifestyle changes Location: outpatient Mode of delivery: outpatient and online Personnel responsible for delivery: clinical researcher Timing post‐stroke: within 30 days Control: participants "received only those interventions chosen by their attending neurologist‐clinician, without the use of the algorithm or interactive education and access to the educational website"
Outcomes	12 months: medication adherence
General Information	Funding: funded was provided by the National Key Technology Research and Development Program in the 11th 5‐year plan of China Country of origin: China Publication language: English and Chinese
Notes	Analysis method: linear regression model Risk of bias: Unclear
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Simple cluster sampling method applied
Allocation concealment (selection bias)	High risk	Concealment was not discussed therefore assumed no blinding occurred
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Does not discuss missing data
Selective reporting (reporting bias)	Low risk	Study protocol has been published in a previous publication
Other bias	Unclear risk	Unclear if recurrent events were presented as number of events rather than number of people with one or more event