Provider-directed imaging stress testing reduces health care expenditures in lower-risk chest pain patients presenting to the emergency department

Abstract

Background: Among intermediate- to high-risk patients with chest pain, we have shown that a cardiac magnetic resonance (CMR) stress test strategy implemented in an observation unit (OU) reduces 1-year health care costs compared with inpatient care. In this study, we compare 2 OU strategies to determine among lower-risk patients if a mandatory CMR stress test strategy was more effective than a physicians' ability to select a stress test modality.

Methods and results: On emergency department arrival and referral to the OU for management of low- to intermediate-risk chest pain, 120 individuals were randomly assigned to receive (1) a CMR stress imaging test (n=60) or (2) a provider-selected stress test (n=60: stress echo [62%], CMR [32%], cardiac catheterization [3%], nuclear [2%], and coronary CT [2%]). No differences were detected in length of stay (median CMR=24.2 hours versus 23.8 hours, P=0.75), catheterization without revascularization (CMR=0% versus 3%), appropriateness of admission decisions (CMR 87% versus 93%, P=0.36), or 30-day acute coronary syndrome (both 3%). Median cost was higher among those randomly assigned to the CMR-mandated group ($2005 versus $1686, P<0.001).

Conclusions: In patients with lower-risk chest pain receiving emergency department-directed OU care, the ability of a physician to select a cardiac stress imaging modality (including echocardiography, CMR, or radionuclide testing) was more cost-effective than a pathway that mandates a CMR stress test. Contrary to prior observations in individuals with intermediate- to high-risk chest pain, in those with lower-risk chest pain, these results highlight the importance of physician-related choices during acute coronary syndrome diagnostic protocols.

Clinical trial registration: URL: http://www.clinicaltrials.gov. Unique identifier: NCT00869245.

Figures

Figure 1

CONSORT Diagram

Figure 2

Boxplot of interval times by study group No significant differences in time intervals were detected among groups (all p>0.05, Kruskal-Wallis tests). *excludes 3 CMR participants who received no cardiac imaging