After review of abstracts and full-text articles, 17 trials were included in the review. Data from 13 of the trials were included in the meta-analyses. The flow of studies through the review is presented in Figure 1. The 17 included trials involved 2689 participants. The characteristics of these trials are presented in Table 1. All trials except one18 satisfied the first item on the PEDro scale, which relates to the eligibility criteria and source of participants and does not contribute to the total score. The remaining PEDro item ratings and total scores for

the included selleck inhibitor trials are presented in Table 2. The median PEDro score of the included trials was 6 (range 3 to 8), indicating that the methodological quality of the included trials varied selleck products from poor to good. The sample sizes of the included trials ranged from 41 to 406, consisting mainly of male participants. The experimental interventions included exercise training, inspiratory muscle training, education, relaxation, counselling, and complex/multiple interventions. Outcome data from at least one trial were available

for postoperative pulmonary complications, time to extubation, length of stay in ICU and the hospital, physical function and costs. Based on data from six trials (661 participants), there was a significant reduction in the relative risk of developing postoperative pulmonary complications with preoperative intervention, old as presented in Figure 2. When the results from trials included in this meta-analysis were pooled, no heterogeneity was present and the pooled relative risk of developing postoperative pulmonary complications was 0.39 (95% CI 0.23 to 0.66). The relative risk reduction was 61% and the number needed to treat was 12 (95% CI 8 to 27). Preoperative intervention shortened the time to extubation by a pooled mean difference of 0.14 days (95% CI 0.01 to 0.26), based on data from four trials (291 participants). There was moderate heterogeneity in the analysis, which is presented in Figure 3. Meta-analysis of data from three trials (233

participants) indicated a non-significant reduction in ICU length of stay due to preoperative intervention, with a pooled mean difference of -0.15 days (95% CI -0.37 to 0.08) and low heterogeneity, as presented in Figure 4. Data from ten trials (1573 participants) showed no significant effect on hospital length of stay, with a pooled mean difference of -0.55 days (95% CI -1.32 to 0.23) and moderate heterogeneity, as presented in Figure 5. Exploratory meta-regression demonstrated no influence on this outcome by study design, geographical region, or type of intervention (either intensive education versus booklet only, or breathing exercises versus no breathing exercises). Age, however, had a significant effect (I2 = 26%, co-efficient = -0.08 (SE 0.03), p = 0.04).