Data extraction, synthesis and statistical analyses

Important information including outcome measures and outcomes, primary and secondary endpoints, participants’ demographic and clinical characteristics and other relevant information were obtained from the selected research articles of the respective studies and organized on datasheets. Data were extracted by two researchers independently who later cross checked the work of each other. Inter-rater reliability was good (kappa = 0.94).

Random effects meta-analyses were performed with STATA software (version 12; Stata Inc. Texas, USA) to achieve overall effect sizes of the prevalence of OSA, OSA risk, and SDB in asthma patients and to achieve a summary estimate of the odds ratio of the prevalence of OSA/OSA risk/SDB between asthma and non-asthma patients observed in the individual studies.

To assess the significance of differences in FEV (% predicted), BMI and ESS between asthma patients with and without OSA/OSA risk/SDB, meta-analyses of mean differences were carried out with RevMan software (version 5.3.5; Cochrane Collaboration) under random effects model. Between studies statistical heterogeneity was tested by tau² and I² indices.