Fig. 2. An hour-by-hour plot of cumulative patient-controlled analgesia (PCA) morphine consumption for each group showing the best-fitting least squares regression line calculated separately for day 1 (0–24 h) and day 2 (25–48 h). Each regression line accounts for at least 98% (r 2) of the total variance. The significantly higher Y intercept for group 3 versus group 1 and group 2 on day 1 reflects the greater morphine requirements for the control group upon arriving in postanesthetic care unit due to the absence of active epidural analgesia (§P < 0.0009 for Bonferroni corrected tests of the intercept comparing group 1 with group 3 and group 2 with group 3 on day 1). Despite these initial differences, the hourly rate of morphine consumption was significantly lower for group 1 than group 3 across the 2-day study period and for group 2 than group 3 on day 2, reflecting the benefits of preincisional and postincisional epidural analgesia. In addition, on day 2, the rate of morphine consumption was significantly lower for group 1 than group 2, reflecting a late preemptive analgesic effect. The rate of morphine consumption across the 2 days was lower by approximately 17% among group 1 compared with group 3. Bonferroni corrected significance tests of the regression line slope compared the following: group 1 versus group 2 and group 1 versus group 3 on day 1, *P < 0.003; group 1 versus group 2 and group 1 versus group 3 on day 2, ‡P < 0.0009; and group 2 versus group 3 on day 2, †P < 0.0009.
