117 PROTEIN PROVISION AND EXERCISE THERAPY TO IMPROVE OUTCOMES IN ICU-SURVIVORS 7 7.2.7 Statistical analysis Effect measures of mortality were presented as adjusted odds ratios (OR) and hazard ratios (HR), andnumerical data fromother outcomes asmeandifferences (MD). In the case of diversity in the measurement of the same outcome, standardised mean difference (SMD) was presented to derive a comparable effect size. SMD was interpreted as follows: ≤0.2 small, 0.3-0.7 moderate, and ≥0.8 large effect. If necessary, we calculated CI from the mean, SD, and n as presented in the study. Median and IQR were transformed into mean and SD.43,44 The results of MD, OR and HR were presented in tables and SMD in forest plots, with similar data grouped, where possible. Clinical heterogeneity was explored by comparing the interventions and study methodologies of at least three studies that provided similar data. Pooling of the results was appropriate if clinical heterogeneity was absent and if the I-square statistic (I2) of the random-effects model was <50%. If I2 was ≥50%, subgroup analysis based on intervention or patient characteristics was planned to explore heterogeneity. When pooling of the results was justified, we planned to perform a sensitivity analysis with studies that were judged as low ROB to determine the robustness of the pooled estimate. We reported information regarding missing data as was stated by the authors. Any other missing or unclear information was requested from authors if necessary. Statistical calculations and risk of bias traffic light and summary plots were conducted with R studio (Version 1.4.1106; 2019-2021).45 7.3 RESULTS 7.3.1 Study selection Figure 1 summarises the study process according to the PRISMA flow diagram.35 Inter-observer agreement of assessing the eligibility of studies was moderate- good (к=0.41-0.72). A total of 107 full-text articles were retrieved for protein provision and 14 for combined protein-exercise therapy. Finally, 15 studies for protein provision were eligible for inclusion,46-61 including one for combined protein-exercise therapy.57 One study was excluded because of conflicting results and lack of scientific quality61 and another after contacting the author for additional information without success.62 Other reasons for exclusion were insufficient information on protein target or delivery, lack of difference between intervention and comparator, or the outcomes were not in line with this review. Not including NRSI occurred mainly because of a lack of comparator or insufficient control for relevant confounders (“not eligible NRSI”). Reasons for all excluded studies are reported in Appendix 2; supplement 2.
RkJQdWJsaXNoZXIy MjY0ODMw