Table 3. Association Between Ambulatory Follow-up and Risk of 30-Day Postdischarge Mortality, Subsequent ED Visit, and Inpatient Hospitalization Among Medicare Beneficiaries Treated in the ED and Discharged From 2011 to 2016, Overall and Stratified by Hospital Follow-up Category^a.

Outcome	HR (95% CI)b	P value
All visits
Mortality	0.49 (0.49-0.50)	<.001
Subsequent ED visit	1.010 (1.003-1.030)	<.001
Inpatient stay	1.22 (1.21-1.23)	<.001
Visits to high follow-up hospitalsc
Mortality	0.47 (0.46-0.48)	<.001
Subsequent ED visit	1.02 (1.01-1.02)	<.001
Inpatient stay	1.22 (1.20-1.25)	<.001
Visits to medium follow-up hospitalsc
Mortality	0.50 (0.49-0.51)	<.001
Subsequent ED visit	1.00 (0.99-1.00)	.82
Inpatient stay	1.22 (1.21-1.23)	<.001
Visits to low follow-up hospitalsc
Mortality	0.60 (0.58-0.63)	<.001
Subsequent ED visit	1.02 (1.01-1.03)	.001
Inpatient stay	1.22 (1.20-1.25)	<.001

Abbreviations: ED, emergency department; HR, hazard ratio.

^aCox proportional hazards model with the time to each postdischarge event as the outcome and ambulatory follow-up as a time-varying covariate as the primary exposure. We incorporated beneficiary age, sex, race, and Medicaid eligibility, year of visit, principal diagnosis category, and beneficiary chronic conditions as covariates and accounted for clustering by hospital. For the outcomes of ED visits and inpatient stays, we also incorporated mortality as a competing risk.

^bAn HR less than 1 indicates a longer time until the outcome event.

^cThree groups of hospitals were created based on their adjusted rates of ambulatory follow-up after ED discharge into high follow-up (top quartile), medium follow-up (middle 50%), and low follow-up (bottom quartile) hospitals. We repeated our main models separately for high, medium, and low follow-up hospitals.