EXHIBIT 4.

Changes in the county-level supply of home health agencies, by urban-rural status, June 2002–December 2017

		Change in average number of agencies per 1,000 beneficiaries per county (versus urban counties)
Time period	Rural add-on payment	All rural counties	Urban-adjacent rural counties	Non-urban-adjacent rural counties
Jun 2002–May 2003	10%	(ref)	(ref)	(ref)
Jun 2003–May 2004	0%	0.001	0.06*	−0.08
Mar 2004–Feb 2005	5%	0.04	0.11**	−0.07
Mar 2005–Feb 2006	0%	−0.04	0.04	−0.14**
Jan–Dec 2006	5%	0.02	0.15	−0.15
Jan–Dec 2007	0%	−0.04	0.11	−0.21*
Jan–Dec 2008	0%	−0.11	0.08	−0.35**
Jan–Dec 2009	0%	−0.20	0.04	−0.50****
Apr 2010–Mar 2011	3%	−0.25*	0.07	−0.62****
Jan–Dec 2011	3%	−0.21	0.11	−0.63****
Jan–Dec 2012	3%	−0.18	0.19	−0.64****
Jan–Dec 2013	3%	−0.20	0.16	−0.66****
Jan–Dec 2014	3%	−0.20	0.21	−0.69****
Jan–Dec 2015	3%	−0.19	0.24	−0.67****
Jan–Dec 2016	3%	−0.22	0.23	−0.74****
Jan–Dec 2017	3%	−0.22	0.22	−0.72****

SOURCE Authors’ analysis of data from Home Health Compare and the Geographic Variation Public Use File of the Centers for Medicare and Medicaid Services, and of state certificate-of-need regulations. NOTES The results represent the interaction between a county’s urban-rural status and time period, with June 2002–May 2003 as the reference time period and urban as the reference for urban-rural status. The analysis accounts for correlation and adjusts for certain factors, as explained in the notes to exhibit 3. Significance refers to differences in the change in the average number of home health agencies between each time period and the reference time period.

^*p < 0:10

^**p < 0:05

^****p < 0:001