Comparing the Classification of Percutaneous Coronary Interventions Using the 2012 and 2017 Appropriate Use Criteria: Insights From 245,196 Patients in the NCDR CathPCI Registry.

BACKGROUND: Appropriate use criteria (AUC) have been developed to promote the rational use of percutaneous coronary intervention (PCI) among clinicians and to provide benchmarking feedback to hospitals. The original AUC were published in 2012 and subsequently updated in 2017 to reflect emerging, contemporary evidence however the degree to which the updated guidance re-classifies PCI appropriateness is unknown. METHODS: Elective PCI cases from March 1, 2018 until June 30, 2021 were identified from within the NCDR CathPCI database. PCI cases were classified as 'appropriate,' 'uncertain' or 'inappropriate' under 2012 AUC and 'appropriate,' 'may be appropriate' or 'rarely appropriate' under 2017 AUC; those with missing data elements were termed 'not mappable.' Groups that 'remained appropriate' (appropriate in both 2012 and 2017), 'became non-appropriate' ('appropriate' in 2012 but became either 'may be appropriate' or 'rarely appropriate in 2017) and 'became appropriate' ('appropriate' in 2017 but were 'uncertain' or 'inappropriate' in 2012) were descriptively compared. Concordance was assessed by calculation of Cohen's Kappa. RESULTS: A total of 245,196 patients underwent elective PCI across 1669 centers. By 2012 AUC, 44% were classified 'appropriate,' 28% were 'uncertain' and 16% were 'inappropriate' compared with 2017 AUC which considered 34% 'appropriate', 56% may be 'appropriate' and 4% 'rarely appropriate'. Overall fair agreement was observed with a Kappa statistic of 0.40 (95%CI 0.396-0.403). Compared with PCI that 'remained appropriate' under the 2017 AUC, PCI that 'became non-appropriate' in 2017 were more likely to be asymptomatic, less likely to be on anti-anginals and less likely to have complex lesions. Compared with PCI that 'became non-appropriate', PCI that 'became appropriate' had a higher proportion of atypical and non-anginal symptoms and were less likely to have had positive functional tests. Procedural related outcomes were similar across all groups. A total of 29 429 PCI (12.0%) were not mappable by 2012 AUC while 16 077 (6.6%) were not mappable by 2017 AUC. CONCLUSIONS: In this contemporary analysis of patients undergoing PCI in the United States, only fair agreement between the 2012 and updated 2017 AUC was observed. While some of this reflects the intention of the updated guidance, the large proportion that were considered 'maybe appropriate' or who 'became non-appropriate' reflect the difficulties of documenting and implementing contemporary AUC guidance.