Where insufficient data were reported, first authors were contacted by email to request data. The PEDro scale was used to assess trial quality and it is a reliable BMS-354825 solubility dmso tool for the assessment of risk of bias of randomised controlled trials in systematic reviews.14 The PEDro scale consists of 11 items, 10 of which contribute to a total score.12 In the

present review, PEDro scores of 9 to 10 were interpreted as ‘excellent’ methodological quality, 6 to 8 as ‘good’, 4 to 5 as ‘fair’, and < 4 as ‘poor’ quality.15 Two reviewers (DS and ES) independently assigned PEDro scores and any disagreements were adjudicated by a third reviewer (TH). The number of participants, their ages and genders, and the types of cardiac surgery were extracted for each trial. The country in which each trial was performed was also extracted. To characterise the preoperative interventions, the content of the intervention, its duration and the health professional(s) who Epacadostat chemical structure administered it were extracted for each trial. The data required for meta-analysis of the outcome measures presented in Box 2 were also extracted

wherever available. Meta-analysis aimed to quantify the effect of preoperative intervention on the relative risk of developing postoperative pulmonary complications, on time to extubation (in days), and on the length of stay in ICU and in hospital (also in days). An iterative analysis plan was used to partition out possible heterogeneity in study results by sub-grouping studies according to independent variables of relevance, eg, age, type of

intervention or type of outcome. Due to the differences in clinical populations and therapies being investigated across the studies, random effects meta-analysis and meta-regression models were used. The principal summary measures used were the pooled mean difference (95% CI) and the pooled relative risk (95% CI). Where trials included multiple intervention groups, the meta-analyses were performed using the outcome data of the most-detailed intervention group. Sensitivity about analyses were conducted for length of stay using meta-regression to examine: the influence of population differences (age as a continuous variable); study design (randomised versus quasi-randomised); global geographical region (Western versus Eastern); intensity of education (intensive, defined as anything more than an educational booklet, versus non-intensive, defined as a booklet only); and type of intervention (breathing exercises versus other). Thresholds for sensitivity analyses were defined according to median values (eg, age) or defined using investigator judgment and clinical expertise. Two studies could only be included in analyses for outcomes assessable until time to extubation, as they provided postoperative physiotherapy intervention following extubation in ICU.16 and 17 To aid interpretation of the effect on postoperative pulmonary complications, the relative risk reduction and number needed to treat were also calculated.