Risk of bias summary of the included studies
| Random sequence generation | Allocation concealment | Blinding of patients and personnel | Blinding of outcome assessment | Analysed as randomised | Attrition bias and missing data (>10%) | |
|---|---|---|---|---|---|---|
| Wang, 201426 | + | + | + | + | + | −a |
| Zanasi, 201427 | + | ? | + b | ?b | + | −c |
| Woodcock, 201028 | ? | ? | + b | ?b | + | (+)d |
| Zolghadrasli, 200929 | ? | ? | − | − | ? | + |
| Ponsioen, 200530 | + | + | + b | ?b | + | + |
| Pornsuriyasak, 200531 | ? | ? | + | + | ? | (−)e |
↵a <80% of patients with outcome data at week 4 (co-primary outcome). For 2 weeks (co-primary outcome), 87% of patients with primary outcome data.
↵b Study placebo-controlled and reported as double-blind. The authors think it is likely that patients and personnel conducting the intervention were blinded, but this is unclear for the outcome assessor and the outcomes are subjective.
↵c Of 92 randomised patients, nine were excluded from the analysis due to adverse events (n = 5) or ‘increased cough’ (n = 4).
↵d For the primary outcome (cough severity scores), only two of 91 randomised patients dropped out. In 17 out of 91 patients there were technical problems with the device used for objectively monitoring cough (secondary outcomes).
↵e For the 2-week cough score results, data for all 30 randomised patients were available. For the 4-week timepoint, three of 15 randomised patients in the control group (20%) had no data assessment (one of 15 in the experimental treatment group, 7.5%). Green = low risk of bias. Pink = high risk of bias. Yellow = unclear risk of bias.