Table 3.

The Effects of ECHO Pain Participation on Prescribing: Results Adjusted for Selection Bias and for Sex, Age, Beneficiary Category Covariables (Alternative Analysis)

Theme/variable	Comparison baseline	ECHO Pain baseline	Comparison annual change percentage, P value	ECHO Pain annual change percentage, P value	Interaction P value
Opioid analgesic prescriptions (Rx)
Avg. number of opioid RX/patient/year	0.56	0.31	− 6.4, P < 0.001	− 22.5, P < 0.001	< 0.001
Percent opioid users	29.0	13.5	− 8.0, P < 0.001	− 20.1, P < 0.001	< 0.001
Morphine milligram equivalents (MME)
Avg. MME/patient /year	362	176	− 10.6, P < 0.001	− 27.5, P = 0.002	< 0.001
Percent opioid users	29.0	13.5	− 8.0, P < 0.001	− 20.1, P < 0.001	< 0.001

Alternative analysis using a propensity scoring transformation with the ECHO distribution as target, i.e., ECHO Pain is unchanged but the comparison group is weighted attempting to reduce the selection bias (see SOM). The addition of sex, age, beneficiary category to model adjusts for lack of balance in Table 1. Importantly, annual change percentages are little affected validating the original analysis. These alternative analyses also do reduce baseline bias