Bourbeau 2003.

Methods	A multicentre randomised clinical trial
Participants	191 patients were randomised. All patients in each participating hospital, of 3 in Quebec, who had been hospitalised for at least once in the preceding year for an acute exacerbation of COPD were screened for participation. The eligibility criteria included: be a current or previous smoker with FEV1 after bronchodilator between 25% and 70% of predicted normal value; have no evidence of asthma, left congestive heart failure, terminal disease, dementia or uncontrolled psychiatric illness.
Interventions	Intervention: The intervention group received a COPD self management program, consisting of 1 hour a week teaching at home for 7 to 8 weeks, in English or French. Supervised by experienced nurses or respiratory therapists who acted as case managers. Follow‐up consisted of weekly telephone calls for 8 weeks during the educational period then monthly calls for the remainder of the study. The patients could also contact their case manager for advice during this time. Control: Both groups continued their usual care by their respective general practitioners and specialists and there was no restriction on their access to the regional universal health program.
Outcomes	1. Medication profile. 2. Spirometry., 3. 6 minute walk test. 4. Dyspnoea measurements after exercise. 5. Health related quality of life as measured by the St George Respiratory Questionnaire (SGRQ). 6. Healthcare utilisation. 7. Costs. 8. Cost effectiveness.
Notes	There were 469 eligible patients; however, 251 refused to participate and 27 who agreed to participate were deemed to live too far away thus 191 patients were randomised. This level of exclusion risks introducing selection bias especially as certain socioeconomic groups are perhaps more likely to refuse to participate in studies. However, the authors do state that those who refused were similar to the study group with respect to sex, age and level of airflow obstruction.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated list of random numbers
Allocation concealment (selection bias)	Low risk	“randomisation with the use of a central computer generated list of random numbers”, “The blocking factor was not known by the investigators or their staff in each participating centre.
Blinding (performance bias and detection bias) All outcomes	Low risk	"As a double blind design was impossible, an independent evaluator unaware of the patient assignment was responsible for the evaluation process in each centre. The evaluator was cautioned not to ask about the workbook modules and types of contact"
Incomplete outcome data (attrition bias) All outcomes	Low risk	26 patients dropped out after randomisation, 1 was lost to follow up and 11 found that the burden of evaluation was too great. 14 patients in total died.
Selective reporting (reporting bias)	Low risk	No evidence of selective reporting
Other bias	High risk	Possible selection bias