Fig. 3

Significant predictors of MME. Comparison of the GLM of MME use, mean estimates with 95% CIs, without and with adjusting for fentanyl protocol differences. Values in parentheses show the comparison group, not the reference. After adjusting for the fentanyl outliers, Treatment Group is no longer a significant predictor of MME. The other predictors remain similar despite the adjustments. Without fentanyl use, there is no significant difference in MME between the MMA and standard opioid groups, which proves that differences in treatment are an artifact of differences in fentanyl use at the different institutions and not differences in the basic protocol