Nathell 2005.

Study characteristics
Methods	Study design: RCT Total duration of study: 4‐week intervention with follow‐up up to 3 years Details of any run‐in period: None stated Number of study centres and location: Single site. Sweden Study setting: Inpatient clinic Withdrawals: Not stated Date of study: Not stated
Participants	Number recruited: 197 (101 randomised to intervention group. 96 randomised to control group) Number completed: Not stated Mean (SD) age: Intervention group 40.8 years. Control group 42.9 years. No SD was provided for age. Age range: Not stated Gender (M/F): Intervention group 43/58. Control group 44/52 Mean (SD) BMI: Intervention group 26 kg/m2. Control group 26.6 kg/m2. No SD was provided for BMI. Severity of condition: Not stated Diagnostic criteria: Clinical examination Baseline lung function: Mean % predicted FEV1: Intervention group 89.6%. Control group 89.2%. No SD provided. Smoking history: 49.7% of participants were smokers, and 18.8% of participants were ex‐smokers at randomisation. Asthma treatment: Not stated Inclusion criteria: Not stated Exclusion criteria: None stated
Interventions	Intervention: 4‐week inpatient PRP. The main components were education, pharmacological optimisation, physical training, and coping skill acquisition. Comparison: Usual care Concomitant medications: Intervention group: 50.5% of participants had used inhaled steroids, and 17.8% of participants had used oral steroids in year prior to randomisation. Control group: 54.2% of participants had used inhaled steroids, and 19.8% of participants had used oral steroids in year prior to randomisation. Excluded medications: Not stated
Outcomes	Primary outcomes: Sick leave days (number of days with any type of sick leave in the 3 years after randomisation) Secondary outcomes: Use of inhaled steroids (questionnaire) and smoking habits (questionnaire) in the 3 years after randomisation Time points reported: 1, 2, and 3 years
Notes	Funding: Financial funding received from the research department of AFA insurance company, Sweden. Notable conflicts of interest: None stated Other: No outcomes of interest to current review are reported.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	“A person outside the project, without knowledge of the subjects, executed a computerized randomization”.
Allocation concealment (selection bias)	Low risk	“A person outside the project, without knowledge of the subjects, executed a computerized randomization”.
Blinding of participants and personnel (performance bias) Subjectively reported outcomes	High risk	Rehab programme‐ not able to blind participants or personnel delivering it.
Blinding of participants and personnel (performance bias) Not subjectively reported outcomes	High risk	Rehab programme‐ not able to blind participants or personnel delivering it.
Blinding of outcome assessment (detection bias) Subjectively reported outcomes	Unclear risk	“The outcome assessor was not blinded to treatment allocation”. (Outcome: smoking status)
Blinding of outcome assessment (detection bias) Not subjectively assessed outcomes	Unclear risk	“The outcome assessor was not blinded to treatment allocation”. (Outcome: sick days ‐ administrative data)
Incomplete outcome data (attrition bias) All outcomes	Low risk	“All analyses were based on intention to treat”. Data available on all but 2 of 197 participants.
Selective reporting (reporting bias)	Unclear risk	No published protocol available.
Other bias	Unclear risk	Sickness insurance scheme from which subjects were recruited is mainly for manual workers and includes only private workers. Persons living in “certain communities” were selected. Diagnosis of asthma made on clinical examination. The proportion of current smokers at randomization was very high (50% and 45% subjects still smoking at 3 year follow up