Health Care Use and Expenditures Associated With Cardiac Rehabilitation Among Eligible Medicare Fee-for-Service Beneficiaries

Abstract

Background: Cardiac rehabilitation (CR) can improve cardiovascular health. We identified whether CR participation was associated with fewer subsequent inpatient hospitalizations and emergency department visits and less Medicare and out-of-pocket expenditures, and whether outcomes varied by amount of participation.

Methods: This retrospective study used Medicare fee-for-service claims data, including beneficiaries with a CR-qualifying event in 2016. Participants attended ≥2 sessions of CR within 365 days of the event. Propensity score matching was used to identify CR-eligible nonparticipants. Difference-in-differences analyses were used to compare differences in outcomes before (2014-2015) and after (2018-2019; 2-year CR period=2016-2017) the CR period between participants and nonparticipants.

Results: We identified 57 668 CR-eligible beneficiaries after matching, with equal numbers of participants and nonparticipants. Nearly 65% of beneficiaries had a percutaneous coronary intervention, 33.5% had an acute myocardial infarction, 17.5% had a coronary artery bypass graft, and 16.8% had a heart valve repair/replacement. Compared with nonparticipants, participants had 47.6 fewer subsequent annual inpatient hospitalizations per 1000 beneficiaries (95% CI, -58.8 to -36.3) and $1005 lower subsequent annual Medicare expenditures per beneficiary (95% CI, -$1352 to -$659). Compared with no participation, medium participation (12-23 sessions), high participation (24-35 sessions), and CR completion (≥36 sessions) were associated with fewer inpatient hospitalizations and lower Medicare expenditures per year.

Conclusions: CR was associated with fewer subsequent annual inpatient hospitalizations and lower subsequent annual Medicare expenditures. A higher amount of participation was associated with a further reduction in hospitalizations and expenditures. These findings can inform programs and policies that encourage CR participation.

Keywords: cardiac rehabilitation; emergency department visits; expenditures; inpatient hospitalizations.

Figure 1. Association between CR participation and subsequent annual inpatient hospitalizations and ED visits (per 1000 beneficiaries) and expenditures (per beneficiary), for any CR‐qualifying event and stratified by event type (adjusted DID estimates, matched cohort).

DID analyses were used to compare differences in inpatient hospitalizations (A), ED visits (B), Medicare expenditures (C), out‐of‐pocket expenditures (D), and total medical expenditures (E) before (2014–2015) and after (2018–2019; 2‐year CR period=2016–2017) the CR period between CR participants and nonparticipants (reference group). Models for any CR‐qualifying event controlled for age, sex, race and ethnicity, dual enrollment status, urbanicity, US Census Division, comorbidities, CR‐qualifying event (AMI [with or without procedure], CABG [with or without AMI], combination event [with or without AMI], heart valve repair or replacement [with or without AMI], PCI [with or without AMI]), HCC risk score, and primary qualifying event LOS. Models for PCI, AMI, CABG, heart valve repair or replacement, and combination procedure controlled for age, sex, race and ethnicity, dual enrollment status, urbanicity, US Census Division, comorbidities, HCC risk score, and primary qualifying event LOS. *PCI, CABG, and heart valve repair or replacement can be with or without AMI or other procedures. ^†AMI can be with or without procedure. ^‡Combination event can be with or without AMI. AMI indicates acute myocardial infarction; CABG, coronary artery bypass graft; CR, cardiac rehabilitation; DID, difference‐in‐difference; ED, emergency department; HCC, hierarchical condition category; LOS, length of stay; and PCI, percutaneous coronary intervention.

Figure 2. Association between amount of CR participation and subsequent annual inpatient hospitalizations and emergency department visits (per 1000 beneficiaries) and expenditures (per beneficiary), for any CR‐qualifying event (adjusted DID estimates, matched cohort).

DID analyses were used to compare differences in inpatient hospitalizations (A), ED visits (B), Medicare expenditures (C), out‐of‐pocket expenditures (D), and total medical expenditures (E) before (2014–2015) and after (2018–2019; 2‐year CR period=2016–2017) the CR period between CR participants (by amount of CR participation: 2–11, 12–23, 24–35, ≥36 CR sessions) and nonparticipants (reference group [n=28 834, 50.0%]). All models controlled for age, sex, race and ethnicity, dual enrollment status, urbanicity, US Census Division, comorbidities, CR‐qualifying event (AMI [with or without procedure], CABG [with or without AMI], combination event [with or without AMI], heart valve repair or replacement [with or without AMI], PCI [with or without AMI]), hierarchical condition category risk score, and primary qualifying event LOS. AMI indicates acute myocardial infarction; CABG, coronary artery bypass graft; CR, cardiac rehabilitation; DID, difference‐in‐difference; ED, emergency department; LOS, length of stay; and PCI, percutaneous coronary intervention.