Variability in Cardiac Stress Test Interpretation: Agreement Between Enrollment Sites and Core Laboratories in the Global ISCHEMIA Trial

BACKGROUND: Cardiac stress testing is a cornerstone of risk stratification and management in patients with chronic coronary disease, yet the consistency and accuracy of its interpretation remain poorly defined. This analysis evaluated variation in the interpretation of myocardial ischemia between enrollment sites and core laboratories in the ISCHEMIA trial (International Study of Comparative Health Effectiveness With Medical and Invasive Approaches). METHODS: ISCHEMIA was a global (37 countries, 2012–2018) randomized trial of an initial invasive versus conservative strategy in patients with chronic coronary disease and moderate or severe ischemia. This analysis included participants with site-interpreted qualifying stress tests—nuclear, echocardiography (echo), cardiac magnetic resonance, or exercise tolerance test—and independent core laboratory adjudication. Core laboratories, serving as the reference standard, reinterpreted tests blinded to site results. A trinary outcome variable (site underestimation, concordance, or overestimation) was defined by comparing site-determined ischemia levels to standardized core lab assessments. Adjusted mixed-effects logistic regression models with random site intercepts assessed variability. RESULTS: Among 6971 participants (mean age, 62.8 years; 73% men), site interpretations showed 0% no/mild (by design), 43% moderate, and 57% severe ischemia. Core labs reclassified these as 8% none, 11% mild, 30% moderate, and 51% severe ischemia. For the imaging modalities, median site-core lab agreement rates were ≈55%; nearly 25% of site-classified moderate/severe cases were downgraded to no or mild ischemia by core labs. Adjusted median odds ratios for site overestimation were 2.36 (95% CI, 2.02–2.82; nuclear), 1.98 (95% CI, 1.62–2.60; echo), 1.89 (95% CI, 1.0–5.41; cardiac magnetic resonance), and 2.15 (95% CI, 1.76–2.79; exercise tolerance test). Adjusted median odds ratios for underestimation ranged from 1.25 to 1.77. CONCLUSIONS: In ISCHEMIA, enrollment sites frequently overestimated or underestimated the severity of myocardial ischemia compared with core laboratory assessments, highlighting the need for strategies to improve the consistency and accuracy of stress testing interpretation in patients with chronic coronary disease. REGISTRATION: URL: https://www.clinicaltrials.gov ; Unique identifier: NCT01471522.