Background: Many observational studies that predict patient outcomes have examined the use of myocardial perfusion imaging results. However, a referral pattern for radionuclide testing could bias these analyses and should be determined. These patients may also differ with regard to the extent of coronary artery disease (CAD). All of these differences must be incorporated into proper outcomes examinations. We sought to identify the nuclear perfusion imaging referral pattern for patients with left ventricular (LV) dysfunction and significant CAD.
Methods and results: Patients with LV dysfunction and CAD (n = 2951) meeting our inclusion criteria were compared by receipt or absence of radionuclide perfusion testing within 6 months before or after angiography. Pearson chi2 and Kruskal-Wallis analyses were used to examine differences in baseline characteristics and catheterization results, whereas logistic regression modeling was applied to predict nuclear imaging referral before and after catheterization. Precatheterization nuclear cohort patients were more likely to be minority patients (odds ratio [OR], 1.34; P =.0083) with previous cardiac revascularization (OR, 2.27; P =.0001), Charlson comorbidity index greater than 1 (OR, 1.146; P =.0091), and heart failure symptoms (OR, 1.62; P =.0001) than those without imaging. They were less likely to have a myocardial infarction (OR, 0.464; P =.0001). After catheterization, the nuclear patients were more likely to have had congestive heart failure (OR, 1.452; P =.0019), a myocardial infarction (OR, 1.353; P =.0371), an ejection fraction lower than 30% (OR, 1.058; P =.0002), and prior revascularization (OR, 1.880; P =.0001). In addition, they had fewer diseased vessels (OR, 0.731; P =.0001).
Conclusions: Bias exists in nuclear referral for patients with LV dysfunction and significant CAD and must be considered when interpreting observational studies on this topic.