EXHIBIT 2.

Association between seven levels of physician burnout and Medicare patient outcomes in the US, study of physician burnout, 2017

Physician burnout frequency	ACSA	ACSED	Readmission	Cost
ALL PATIENTS
A few times a year	−0.18	1.02**	−0.00	1,006.51
Once a month	−0.81	0.33	−3.75*	−78.85
A few times a month	−0.54	0.22	−2.89*	−530.18
Once a week	−1.30**	0.48	−3.53*	76.75
A few times a week	−0.94*	0.16	−4.00**	−120.72
Every day	−0.46	0.42	−0.51	229.45
Mean of dependent variable	3.66%	3.98%	13.71%	$10,897.38
No. of attributed patients	32,413	32,413	5,040	32,413
DUAL-ELIGIBLE PATIENTS
A few times a year	−0.89	2.59	−4.83	1,027.82
Once a month	−3.40	1.02	−12.02*	−3,065.66
A few times a month	−0.42	1.36	−10.41**	−1,792.11
Once a week	−3.42	3.74*	−11.46*	−1,837.77
A few times a week	−1.69	0.23	−12.36**	−3,047.74
Every day	0.22	0.37	0.90	138.98
Mean of dependent variable	7.53%	8.05%	18.1%	$16,242.94
No. of attributed patients	3,068	3,068	713	3,068

SOURCE Authors’ analyses of Medicare claims data from January 1 to December 31, 2017, linked to the American Board of Family Medicine survey of all recertifying physicians in 2017. NOTES Dual-eligible patients are those eligible for both Medicare and Medicaid. Coefficients shown are marginal effects—that is, adjusted differences in rates of quality measures (or cost) relative to the omitted category of burnout (never; not shown), estimated from regressions adjusting for physician, practice, and patient characteristics identical to those in exhibit 1. Rates were the annual number of patients with ambulatory care-sensitive admissions (ACSAs), ambulatory care-sensitive emergency department admissions (ACSEDs), and readmissions per 100 patients. Logistic regressions were estimated for quality outcomes, and a generalized linear model was estimated for the cost outcome.

^*p < 0.10

^**p < 0.05