Comparison of randomized groups (PPA vs. GA) on primary outcome variables (N = 247).

^ 98.75% confidence intervals to maintain overall 0.025 significance level: one-tailed testing for superiority for four outcomes with one-tailed alpha of 0.00625 for each outcome.* P-values from one-sided Wilcoxon rank sum test; Bonferroni multiplicity correction [i.e., significance criterion P ** Note: sevoflurane was administered to 2% of regional and 99% of general anesthesia patients (relative risk, 0.02; 98.75% CI, 0.001 to 0.1).# Mean ranks: values for an outcome variable were first ordered from smallest to largest across all patients, and the ordering for a particular patient was considered as the “rank” for that outcome (i.e., there were a total of 121 + 126 = 247 ranks). The ranks were then averaged within each randomized group and the ratio of those means reported as “ratio of mean ranks”. Confidence intervals were obtained by bootstrap resampling.Abbreviations: PPA, propofol and paravertebral block anesthesia; GA, general anesthesia; CI, confidence interval; MAC, minimum alveolar concentration. VAS = 10-cm visual analog scale. Morphine equivalents were calculated from long-acting opioids given intraoperatively or postoperatively, but excluded intraoperative fentanyl, which is reported separately. Results presented as median [first, third quartiles], median difference and ratio of mean ranks (98.75% confidence interval).