Risk of bias of the included study [21]
| Authors’ judgement | Support for judgement | |
| Random sequence generation | Inadequate | Allocated by alternation |
| Allocation concealment | Inadequate | No concealment |
| Blind outcome assessment | Inadequate | No blinding |
| Incomplete outcome data | Adequate | All participants who commenced the study are accounted for |
| Selective reporting (reporting bias) | Adequate | Data for all primary and secondary measures (detailed in the methods) has been reported |
| Other bias | Inadequate | Percentage predicted forced expiratory volume in 1 s was lower in the intervention group and fewer patients started pulmonary rehabilitation in the winter than in the control group. These covariates were reported as not being associated with outcomes but data was not provided. Pulmonary rehabilitation did not deliver similar improvements in routine clinical outcome measures between groups |
| Overall risk of bias | High | More than one criterion deemed inadequate |