Appendix Table 2:

Association between Primary Care Physician Characteristics and Low Value Care Spending of Attributed Medicare Patients

Characteristic	Panel A: OLS Estimates		Panel B: Logit Odds Ratios
Physician Characteristics
Allopath	−1.604^†	0.461	0.655^†	0.088
Female	−0.570	0.453	1.012	0.149
Non-US Medical Graduate	−0.232	0.415	0.985	0.113
Patient Size
<50	−4.121^‡	0.572	0.323^‡	0.053
50 to 149	−2.775^‡	0.468	0.417^‡	0.056
150 to 299	−1.809^‡	0.417	0.542^‡	0.070
300+	Reference		Reference
Specialty
Internal Medicine	Reference		Reference
Family Medicine	−1.063^†	0.318	0.702^‡	0.067
Other Specialty	−1.057	0.654	0.818	0.193
Graduation Year
Pre-1980	Reference		Reference
1980 to 1989	−0.161	0.383	1.090	0.124
1990 to 2000	−0.730	0.409	0.973	0.112
Post-2000	−1.564^*	0.634	0.720	0.159
Practice Region
Northeast	Reference		Reference
Midwest	−2.765^‡	0.456	0.407^‡	0.060
South	0.330	0.487	1.059	0.136
West	0.028	0.573	0.998	0.153
Rural	−1.715^‡	0.331	0.522^‡	0.080
Patient Characteristics
Age (years)
65 to 69	Reference		Reference
70 to 74	−0.035	0.040	0.995	0.010
75 to 79	−0.063	0.035	0.989	0.010
80 to 84	−0.129^‡	0.034	0.983	0.010
85+	−0.334^‡	0.025	0.923^‡	0.007
Female	−0.042^†	0.015	0.984^†	0.005
Race/Ethnicity
White	Reference		Reference
Black	−0.067^‡	0.012	0.985^‡	0.004
Other	−0.038	0.020	0.991	0.005
ElixHauser Comorbidity Index
0	Reference		Reference
1 to 2	0.050^*	0.024	1.015	0.009
3 to 5	0.213^‡	0.031	1.058^‡	0.007
6+	0.307^‡	0.088	1.049^‡	0.011
Practice Location Characteristics
<12 years schooling	−0.097^†	0.034	0.978^*	0.011
Black	0.052^‡	0.012	1.015^‡	0.004
Hispanic	0.104^‡	0.016	1.028^‡	0.004
<200% FPL	−0.077^‡	0.015	0.979^‡	0.005
Number of Observations	6,905

FPL, federal poverty line; OLS, ordinary least squares. Note: The dependent variable for Panel A was per-patient low value care (LVC) Medicare spending in dollar amounts, while the OR ratios in Panel B were calculated from logit regression results, where the dependent variable was equal to 1 if the primary care physician's (PCP's) per-patient LVC Medicare spending was in the top quintile. Both models were estimated using sample weights that reflect the oversampling of physicians in smaller states. For graphical representations of the estimates of PCP characteristics, see Figure 2 for Panel A and Figure 3 for Panel B.
↵* P < .05.
↵† P < .01.
↵‡ P < .001.