Abstract
<title>Abstract</title> <p>Background In real-world non-ST-elevation acute coronary syndrome (NSTE-ACS), patients with the highest predicted ischemic risk may not uniformly undergo invasive management because of comorbidity burden, frailty, bleeding risk, and perceived procedural futility influence treatment selection. Methods We analyzed 1,461 consecutive patients with NSTE-ACS from the Taiwan ACS Full-Spectrum Registry who underwent diagnostic coronary angiography with percutaneous coronary intervention (DCA + PCI), diagnostic coronary angiography without revascularization (DCA-only), or conservative management. The derived GRACE score was retrospectively computed using the available registry variables. A two-stage propensity score matching framework generated a Stage 1 primary matched cohort (N = 237) and a Stage 2 stringent sensitivity cohort (N = 147). Time-updated Cox models incorporating aspirin, clopidogrel, and statin persistence were used to assess all-cause and cardiovascular mortalities. Results The Derived GRACE score was computable in all 1,461 patients. The mean GRACE score was 121.8 ± 39.0, and 442 patients (30.2%) met the high-risk threshold of GRACE > 140. High-risk GRACE status was substantially more common in conservatively managed patients than in the DCA + PCI group (54.0% vs. 27.0%), illustrating a treatment-selection paradox. Before matching, conservative management was associated with higher crude all-cause mortality rates. After propensity matching, the mortality difference across the management groups was substantially attenuated, and cardiovascular mortality did not differ significantly in the primary matched cohort. Active malignancy and chronic kidney disease were the strongest adverse prognostic markers, whereas time-updated persistence of antiplatelet and statin therapies was associated with lower mortality. Angiography timing was not independently associated with survival in the exploratory angiography-restricted analyses. Conclusions In this real-world NSTE-ACS registry, a high predicted GRACE risk was disproportionately concentrated among conservatively managed patients, supporting a strong treatment-selection bias and competing risk confounding. Comorbidity burden and longitudinal medication persistence were more consistently associated with 12-month outcomes than management strategies or angiography timing. These hypothesis-generating findings support individualized invasive decision-making and structured post-discharge medical therapy optimization.</p>