TABLE 3.

Linear Least Square Regressions of Opioid Prescriptions on Each Local Economic Condition Separately Among Medicare Part D Disabled Beneficiaries Aged 65 or Below in Fee-for-Service Programs, 2014

Coefficients of the Regression	Opioid User	Long-term Opioid User	Total Days of Usage	Log Mean Daily MME	Log Total MME
Log median household income†	−0.0532 (0.0000)***	−0.0433 (0.0000)***	−16.0250 (0.0000)***	−0.1181 (0.0030)***	−0.4570 (0.0000)***
Unemployment rate	0.4331 (0.0000)***	0.3401 (0.0080)***	139.9007 (0.0010)***	0.8300 (0.0670)*	3.8887 (0.0010)***
Gini index	−0.3957 (0.0000)***	−0.3118 (0.0000)***	−91.8825 (0.0000)***	−1.6716 (0.0000)***	−3.6182 (0.0000)***

There are 3,493,551 observations for each regression. All models controlled for age, age squared, sex, race/ethnicity, medical condition, state dummies, rural-urban dummies, and individual income status (proxied by cost share information). Reference group for the urban-rural classification is large central metro.

Below each estimated coefficient, its P-value is calculated based on robust SEs clustered by county.

^*P < 0.1.

^**P < 0.05.

^***P < 0.01.

^†The natural logarithm of median household income.