Medical Student University of British Columbia Victoria, British Columbia, Canada
Background: The management strategy for myocardial infarction (MI) is commonly informed by diagnostic angiogram. Many diagnostic angiograms will “proceed” to “ad hoc” percutaneous coronary intervention (PCI), meaning interventional revascularization is pursued during the same encounter. Whether to proceed to ad hoc PCI, and the number of vessels to treat, is largely determined by the interventionalist. However, the impact of the interventionalist on the treatment a patient with acute MI receives has not previously been examined. We analyzed practices in British Columbia by leveraging the natural experiment where acute MI inpatients with unknown coronary anatomy are assigned to an interventionalist by cath lab schedule alone
METHODS AND RESULTS: We linked data from the provincial Medical Services Plan billing records and Discharge Abstract Database using the British Columbia Ministry of Health Data Platform. We included 39,966 patients aged 20-79 that were urgently hospitalized for their first MI, had unknown coronary anatomy, and received diagnostic angiogram with or without angioplasty (balloon inflations with or without stents) between 2001 and 2024 at any of the five adult cardiac cath labs in the province.
Variables included patient age group, shock on admission, diabetes, cancer, cerebrovascular disease, pulmonary edema, dysrhythmia, renal failure, the number of angioplasties performed, and the identity of the treating interventionalist. Population characteristics are provided in Table 1.
We used inverse-probability-of-treatment-weighted hierarchical regression models to calculate the effect of the interventionalist on extensiveness of PCI treatment controlling for patient-specific factors. The patient-factor adjusted proportion of angiogram cases that proceeded to PCI ranged from 81.7% to 96.3% (STEMI), and from 56.1% to 85.1% (NSTEMI), across interventionalists. Mean ad hoc PCI angioplasties per case ranged from 1.12 to 3.82 (STEMI), and from 1.16 to 4.24 (NSTEMI).
We used Gaussian mixture models to identify clusters of practice patterns. Three archetypes emerged: extensive (mean angioplasties > 2.3), moderate (mean angioplasties < 2.3, NSTEMI ad hoc PCI proportion of cases > 75%), and conservative (mean angioplasties < 2.3, NSTEMI ad hoc PCI proportion < 75%). The archetypes clustered by cath lab (Fisher’s Exact Test P=.03). There was an exceptional significant correlation between the mean number of angioplasties per STEMI case and per NSTEMI case, for each interventionalist (R=0.99, P<.001). The distributions of practice characteristics are shown in Figure 1.
Conclusion: Interventionalists vary significantly in use of ad hoc PCI for MI independent of patient factors.
