All statistical analyses were reported using two-tailed unpaired t tests to determine significance (P < 0.05).
Safety analyses were reported using the safety population, comprising all patients who received ≥ 1 dose of AI chemotherapy.
Efficacy analyses were reported using the efficacy population, comprising all patients who received ≥ 1 dose of eltrombopag and had at least 1 platelet count measurement in each of Cycles 1 and 2. Eltrombopag PK was analyzed and will be reported elsewhere.
Whole-brain analyses for the pain > no-pain contrast are reported using a cluster-forming threshold of p <.001 (uncorrected, cluster size > 10), with cluster-level family-wise error (FWE) correction.
Data were also analysed as a continuum and correlations were reported using the Pearson Correlation (r), the slope and the significance (P) for a two-tailed analysis.
Bivariate and multivariate analyses were conducted and multivariable associations were reported using odds ratios and 95% confidence intervals.
High throughput and comprehensive genome analyses have recently been reported using B-LCL collections, for example, genotyping of blood cell alloantigens (Kroll et al. 2001), detection of CNV (Redon et al. 2006), and population genetics (Li et al. 2008).
In ancillary analyses, baseline characteristics will be reported using percentages for categorical data, and means and SDs (or medians and IQRs) for quantitative data.
However, because both analyses yielded similar patterns, results are reported using the raw scores.
The data recorded on the observation tool (4D&4R) was analysed using content analysis [ 20] and is reported using frequencies.
