Higher PCI Physician Supply Linked to More Low-Value PCI

Key Takeaways

• A national Medicare analysis of PCI at US hospitals found that higher regional PCI-provider density was associated with higher hospital-level lower-value PCI provision.
• Mean hospital lower-value PCI provision rose from 17.3% in the lowest provider-density quintile to 22.6% in the highest.
• The gap between the highest- and lowest-density regions remained statistically significant after multivariable adjustment.
• Overall PCI use and lower-value PCI use were both higher in regions with more physicians performing PCI.
• Sensitivity analyses using alternate provider definitions and restricted samples were broadly consistent with the main association.

Stable coronary artery disease (CAD) is a setting in which percutaneous coronary intervention (PCI) can shift from lifesaving care to potentially lower-value care when it occurs outside acute or unstable presentations such as acute myocardial infarction (AMI) or unstable angina. Whether regions with more physicians performing PCI also deliver a larger share of procedures for stable disease remains a clinically relevant policy question. Among Medicare beneficiaries, 100% fee-for-service (FFS) claims and Medicare Advantage (MA) encounter data from 2019 through 2021 were used to examine that relationship.

In Gaffney and colleagues’ analysis of PCI-provider density and lower-value PCI, investigators examined 1,580 hospitals across 306 hospital referral regions (HRRs) and identified 5,505 PCI-performing physicians. Exposure was HRR PCI-provider density, measured as the number of PCI-performing physicians per million population. The primary outcome was the hospital-level lower-value PCI provision rate, defined as the share of PCI performed for stable CAD rather than for AMI or unstable angina. Multilevel mixed-effects linear regression models used random intercepts by HRR and adjusted for hospital characteristics, regional population factors, and Medicare enrollee characteristics.

Hospitals in more saturated regions showed a higher share of lower-value PCI across provider-density quintiles. After full adjustment, hospitals in the highest provider-density quintile had a 5.77 percentage point higher lower-value PCI provision rate than those in the lowest quintile (95% CI 3.70 to 7.84; p < 0.01). The association persisted after measured hospital and regional differences were considered.

Regional utilization moved in the same direction. Total PCI use rose from 15.1 to 31.3 per 1,000 Medicare beneficiaries across the lowest to highest provider-density quintiles, and lower-value PCI use rose from 2.6 to 7.9 per 1,000. Sensitivity analyses using alternate provider definitions, exclusion of small or rural hospitals, exclusion of 2020, PCI-weighted ordinary least squares models, and continuous exposure modeling were broadly consistent.

Because the analysis was observational and cross-sectional, it could not establish causation. The claims-based lower-value measure supported hospital- and region-level comparisons but could not determine whether an individual PCI was appropriate because coronary anatomy, prior cardiac interventions, and diagnostic testing results were unavailable. Generalizability beyond Medicare beneficiaries is uncertain, and some regional covariates were model-based rather than directly observed.

The authors concluded that greater regional supply of PCI-performing physicians was associated with both higher overall PCI use and a higher lower-value PCI share among Medicare beneficiaries. They did not interpret the findings as proof that higher physician supply caused those patterns.

Clinician Questions

How did the analysis define lower-value PCI in stable coronary artery disease?

Lower-value PCI was defined as PCI performed in Medicare beneficiaries with stable coronary artery disease rather than acute myocardial infarction or unstable angina. Stable CAD required ischemic heart disease or angina at least 6 months before PCI, no associated emergency department visit, and exclusion of PCIs performed within 14 days of any emergency department visit or diagnostic-code entry for unstable angina or AMI.

How was PCI-provider density measured across hospital referral regions?

Investigators measured PCI-provider density as the number of PCI-performing physicians per million population in each hospital referral region. In the main specification, PCI providers were clinicians performing 11 or more fee-for-service PCIs in 2020, and each hospital was assigned the density of its own hospital referral region before regions were grouped into quintiles.

Which patients do these Medicare PCI findings apply to?

The analysis used Medicare fee-for-service claims and Medicare Advantage encounter data from 2019 through 2021, but the primary hospital-level analysis was limited to 1,580 included hospitals after excluding several hospital types and hospitals with fewer than 10 total or 10 lower-value PCIs over 3 years. The authors noted that the findings may not generalize to patients with other forms of coverage, so the observed association is most directly applicable to the Medicare population studied.

Why could this analysis not determine whether an individual PCI was appropriate?

The lower-value PCI measure was built from claims data, which let investigators study patterns across hospitals and regions but did not include the full clinical detail needed for patient-level appropriateness decisions. The authors specifically cited the absence of coronary anatomy, prior cardiac interventions, and diagnostic testing results.