Safety and Efficacy of PCI at California Hospitals Without Onsite Cardiac Surgery

Design: The PCI California Audit Monitored Pilot with Offsite Surgery (PCI-CAMPOS) program is a prospective, multi-center pilot trial allowing elective PCIs at hospitals without on-site cardiovascular surgery in California. The pilot program was established by California Senate Bill No. 891 (SB 891) which was enacted in January 2009 authorizing the California Department of Public Health (CDPH) to set up a pilot program to allow California hospitals without surgery on-site to perform elective PCIs. The PCI-CAMPOS program was designed to compare PCI outcomes at the pilot hospitals with offsite cardiac surgery to PCI outcomes at established California hospitals with onsite cardiac surgery.

The CDPH requested pilot hospital applications in December 2009, selected 6 hospitals out of over 30 applicant hospitals in May 2010 and contracted with University of California, Davis (UCD) to train, audit, and monitor the program. The CDPH appointed a 12 member Advisory Oversight Committee (AOC) with 6 at-large members nominated by the California American College of Cardiology (CAACC) and 6 members from the pilot hospitals (1 principal investigator from each participating pilot hospital). The pilot program was funded by CDPH from fee assessments of participating pilot hospitals. The study was approved by the Institutional Review Board (IRB) at each pilot hospital and at the University of UCD coordinating center, and each patient is asked to provide written informed consent for pilot participation (unless unavailable in critically ill patients).

The National Cardiovascular Data Registry (NCDR®) CathPCI Registry (v. 4.3, 4.4) was selected as the data entry format for both pilot hospitals and non-pilot hospitals. The pilot PCI data is entered using the NCDR® CathPCI Registry v.4.3 and v.4.4 on an internet accessible UCD Velos server (Velos eResearch v9.1.2, Velos, Inc., Fremont, CA). The non-pilot hospitals enter their data using various NCDR® approved onsite software or the central NCDR® CathPCI v.4.3 and v.4.4 website. The pilot hospital data is entered within 72 hours of each procedure and immediately available to the PCI-CAMPOS coordinating center. The non-pilot data is entered individually or obtained from quarterly data harvests submitted to NCDR®. After masking patient, operator, and hospital identity, this non-pilot California NCDR® CathPCI registry data is downloaded annually from the NCDR® center to the UCD PCI-CAMPOS coordinating center. The UCD coordinating center performs all pilot audits and has full access to the pilot information. The UCD coordinating center performs all pilot and comparative analyses. Interim data summaries and analyses are presented to the full AOC for review at 6-12 month intervals throughout the study.

Outcomes: The pilot hospital outcomes are compared with hospital outcomes from non-pilot California hospitals performing either elective PCIs with surgery on-site or ST segment elevation myocardial infarction (STEMI) only PCIs. The non-pilot clinical data is obtained from the 120 California hospitals that submit their clinical data directly to the NCDR® data center and is used for the analysis of the data. Both the non-pilot and pilot data sets are analyzed and compared using SAS statistical software (version 9.3 for Windows, SAS Institute, Cary, NC).

Study Participants Hospitals: The non-pilot hospital group consists of all 120 California hospitals which submit PCI data directly to the NCDR® central database. For all elective PCIs, these hospitals have on-site surgery available for any post-PCI patients who require emergency coronary artery bypass graft (CABG) surgery.

The pilot hospital group consists of all 6 hospitals selected by the CDPH, which met the eligible hospital criteria as required by SB891 SEC.2. Section 1256.01 (c) (09/25/2008).

Operators: Operators are approved to perform procedures under the pilot program based on the operator criteria as required by SB891 SEC.2. Section 1256.01 (c) (09/25/2008). Pilot operators underwent training for the assessment of the operator, and the assessment of patient inclusion and exclusion requirements by UCD staff and were also were required to pass a competency exam. Non-pilot operators did not receive special training and their identity was masked to comply with NCDR® requirements.

Emergency (and salvage) pilot hospital PCIs are performed by any credentialed interventionalists at the pilot hospitals. Non-emergent (elective, urgent) PCIs are redirected to the pre-qualified pilot operators.

Coders: Data entry coders at each pilot hospital received an initial 1 day training session, followed by a competency exam. On-going training, telephone, and email access to the coordinating center are available throughout the program. Non-pilot coding is also performed by coders at each onsite hospital participating in the NCDR® CathPCI Registry.

Patients: All pilot hospital patients who consent and meet the SB891 SEC.2. Section 1256.01 (A through D) (09/25/2008) selection criteria for PCI at pilot hospitals are enrolled in the program. All patients with STEMI who present to the pilot hospitals are included in the STEMI PCI group. STEMI-excluded patients who are low to moderate risk for complications are included in the STEMI-excluded group. Patients deemed high-risk (both high lesion and patient risk) for complications are excluded from the pilot program. These high-risk patients are either discharged or transferred to non-pilot hospitals with onsite surgery for revascularization with PCI or CABG.

Pilot Protocol After completion of each PCI, the required information is entered online into the NCDR® CathPCI Registry v.4.3 and v.4.4 UCD server. All cases are then reviewed by the central coordinating site for data field completion and data consistency. All cases with complications [death, stoke, CABG, shock, congestive heart failure (CHF), cardiac arrest, IABP (intra-aortic balloon pump), perforation, significant dissection, tamponade, and new requirement for dialysis] and a randomized 10% of non-complication cases are then selected for a pilot-site audit that includes a review of all medical records and angiographic images. Audit reviews produce multiple queries that are transmitted to the hospital coders for discussion with operators. An iterative process eventually leads to a consensus agreement in most cases. In rare cases, when auditor-operator agreement is not achieved, the case is referred to the angiographic committee or to the AOC for a final determination. The patient and procedure data is stored on the Velos Server; the angiographic images are saved on an Xcelera R3.3L1 Server (Koninklijke Philips Electronics N.V.); and the statistical analyses are stored and performed on SAS software, version 9.3 for Windows (SAS Institute, Cary, NC).

Non-pilot Audits: California hospital PCI data that is submitted individually or at 3 month harvests to the NCDR® central database is reviewed for completeness and data accuracy by NCDR® and receives a red, yellow, or green light before acceptance. The vast majority of the hospitals in California that perform PCI participate in the NCDR CathPCI registry. The NCDR® in addition to employing robust data quality strategies for its participating hospitals also performs a limited number of site audits to directly assess data accuracy. The complete NCDR® non-pilot California data is downloaded annually to the UCD central site for comparison analysis. The downloaded data is masked to protect patient, operator, and hospital identity.

Statistical analysis: A PCI risk model has been developed and risk-adjusted endpoints are compared for the 6 pilot and 120 non-pilot hospital PCI procedures. To compare the demographic and clinical profiles and observed outcomes between CA-NCDR® (non-pilot)and PCI-CAMPOS (pilot), all continuous and categorical variables are reported as mean ± standard deviation (SD) or percentages, and compared with the t test or chi-square test (two tailed), respectively. The primary safety endpoint is a composite event which includes in-hospital death (regardless of length of stay), stroke, and/or patients who were transferred for an emergent CABG. The non-pilot data and the pilot data are merged as a single data source for development of the risk adjustment models for the composite outcome. With the combined data, we perform a bi-variable analysis to identify significant demographic and clinical risk factors for the composite outcome and serve as candidate risk factors for the multivariable logistic risk models. We developed both parsimonious and refined risk models with the combined data and use only non-pilot data for model validation purpose. All models are evaluated with the Hosmer-Lemeshow goodness-of-fit statistic. The c- statistic is reported as a measure of predictive power. The refined model is then used for computation of provider's expected and risk-adjusted composite event rates. A general linear model for analysis of variance (GLM/ANOVA) is used to compare observed, expected and risk-adjusted composite event rates between CA-NCDR® and PCI-CAMPOS hospitals. We further compute 95% confidence interval (CI) of provider's risk-adjusted composite event rate using the Poisson exact probability method for pilot and non-pilot hospitals and pilot operators. We determine provider's performance rating based on a comparison of the 95%CI of provider's risk-adjusted composite event rates and to the California average composite event rate that includes both CABG on-site and off-site hospitals. Two performance rating analyses are conducted for hospitals (or operators in pilot hospitals): (1) based on all PCIs and (2) based on PCIs with STEMI excluded. We also assess the correlation between provider's total and STEMI excluded PCI volume and risk-adjusted composite event rate among hospitals and operators in pilot hospitals. All reported p-values are 2-sided, and p-value < 0.05 are considered to be statistically significant. All statistical analyses are conducted at the UCD with the use of SAS software, version 9.3 for Windows (SAS Institute, Cary, NC).