Table 3.

The methodological quality of the trials

Type of bias	Criteria	WISP			O'Brien et al	COPE			COSMIC
Selection	Method used to generate randomisation sequence?	'Patients were allocated with minimisation to intervention using the programme MINIM v1.3'			'Randomisation was performed by the clinical pharmacist who randomised an odd number dice roll to placebo and an even number dice roll to drug'	'Randomisation was performed in blocks of six by computer generated allocation'			'A randomisation schedule generated by the patient allocation for clinical trials (PACT) program'
	Method used to generate allocation concealment?	'Inhalers were given an alphanumeric code to conceal allocation'			Inadequate	Unclear			Unclear
Performance	Double-blinding?	'Study nurses and regular clinicians were blind to allocation throughout the study'			'The subject and pulmonary physician were blinded to the treatment regimen. Placebo and drug MDI canisters were identical, and the placebo mist was flavoured to make the treatments indistinguishable'	'This study was a randomised, double-blind parallel-group single centre study...'			'the study medication was packed in identical inhaler devices to ensure both the patient and investigator were unaware of the allocated treatment'
Attrition	Loss of patients accounted for?		Steroid	Placebo	Unclear		Steroid	Placebo		Steroid	Placebo
		Number randomised	128	132		Number randomised	123	121	Number randomised	189	184
		Number analysed	128	132		Number analysed	122(1 died)	120(1 died)	Number analysed	unclear	unclear
Detection	Outcome assessor blind	Unclear			Unclear	Unclear			Unclear
Reporting	Any evidence of reporting bias?	No			No	No			No