California Cardiac Surgery and Intervention Project
A comprehensive report on California hospitals performing heart surgery and percutaneous coronary intervention
not here
not here


  1. Summary of Statewide Findings
  2. Strategy and Methods
  3. Supporting Documents (Definitions, Process, Statewide Data Tables, and Statistical Models)

Summary of Statewide Findings

  • In hospital mortality (IHM) for PCI performed for acute MI diagnoses (ICD-9-CM codes 410.xx) was 3.6% (774 of 21,581) in 2009, and IHM for all other PCI was 0.7% (189 of 27,137).
  • IHM for isolated CABG was lower in 2009 (2.0%) compared to 2008 (2.4%). The volume of isolated CABGs continued to decrease while predicted risk in isolated CABG patients was similar in 2008 and 2009.
  • The one-year combined outcome of Mortality, Acute MI, unplanned Reintervention, or Stroke (MARS) for PCI w/o MI decreased from >20% prior to 2004 to 17.6% in 2008.
  • One-year MARS for isolated CABG remained at approximately 10% over the last 10 years; the one-year in-hospital mortality rate decreased from >5% prior to 2000 to 3.9% in 2008, while the one-year reintervention rate increased from 2.6% to 3.4%.
  • Off-pump technique was used in 24% of CABG procedures in 2009; the one-year re-intervention rate was 4.9% for off-pump and 3.1% for on-pump cases.
  • The majority of hospitals perform a relatively low annual volume of revascularization procedures. Those with higher volumes (>500 PCI w/o ACS, >250 PCI w/ACS, >200 CABG) had more consistent outcomes (most had O/E Ratio less than 1.0). Many of the low volume sites have no mortality, so there is little statistical relationship between volume and in-hospital mortality.
  • Repair procedures for mitral valve disease increased from 32% (2000) to 53% (2009). Statewide IHM for isolated MV repair was 2.2% in 2008-2009.
  • The wide variation in outcomes across hospitals for valve procedures continued.
  • Race and gender differences were apparent; IHM and one-year mortality rates for women were higher in all procedure categories; patients of black race had similar IHM as white race, but one-year acute MI rates were twice as high compared to whites.

Strategy and Methods

The California patient Discharge Database (PDD) includes data on all hospitalized patients, and as such is capable of providing long-term information about re-hospitalization after a procedure is performed, including the reason for as well as the urgency of the hospital visit. Outcomes in the present reports are obtained for hospitalized patients only. In the future we plan to link additional information published by the OSHPD public reporting program (California CABG Outcome Reporting Program - CCORP) and through linkage to the vital statistics death file, which locates out-of-hospital deaths. Our measure of "in-hospital mortality" (IHM) does include all patients who die without ever leaving the hospital, even those dying more than 30 days after surgery in another facility; however it does not capture death within 30 days outside the hospital after discharge. Note that currently STS and CCORP do not capture deaths that occur after a transfer out of the primary hospital and later than 30 days after surgery.

The disadvantage of using the PDD is the reliance on ICD-9-CM codes to capture clinical patient risk factors such as the extent of respiratory or cardiac dysfunction. These codes lack quantifiers to stratify the magnitude of co-morbidities. Linkage to clinical information contained in the electronic databases of the CCORP (for CABG data) and individual hospital STS-NDB and American College of Cardiology data files (for other procedures) could be used to more accurately capture patient characteristics. The CASTS is planning to organize regional hospital consortiums within California to request aggregation of STS data for linkage to the PDD. This would provide an electronic audit process for hospital volumes and mortalities, as well as improved subset selection for more patient-specific data on longer term outcomes.

The PDD is, however, the only repository that provides procedure volumes and a number of in-hospital and post-hospital outcomes at the hospital level for all cardiovascular procedures and interventions. Risk adjustment, while imperfect, is reasonably satisfactory as indicated by the statistical analyses posted on this website. The current process for identification of procedures by ICD-9-CM codes and development of risk models is available at the Supporting Documents link

In future iterations of these reports we propose to enlarge the body of administrative and clinical information in our repository to include:

  1. Vital statistics death data: While the discharge information and our ability to track patients across hospitalizations in the patient discharge data provides in-hospital death status, linking information on death outcomes outside the hospital is needed for operative mortality measures commonly used in the study of outcomes after surgery.
  2. Emergency Department data: OSHPD began collecting ED data in 2005. Encounters in the emergency room pre- and post surgery / intervention can be linked through deterministic methods to the patient discharge data. Adding emergency department data to our data repository will enhance our ability to track patient risk factors as well as patient outcomes.
  3. CCORP data: The California CABG Outcome Reporting Project has been collecting detailed clinical data on all patients undergoing isolated CABG surgery in California acute care hospitals since 2003. We intend to add this data to our repository to add clinical information required for improved risk modeling and creating patient subsets.
  4. STS/ACC data: The STS/ACC data offer additional clinical data on PCI, CABG, valve and other types of cardiac surgeries. With the support of CASTS hospital members, this data will be obtained and linked to our data repository of PDD, ED, CCORP, and vital statistics death data. Linkage can be performed using probabilistic methods based on data common to both data sources (patient age, patient gender, patient birth date, patient ZIP code of residence, surgery date, and surgery hospital).
  5. Direct observation: The site visit program that was conducted in 2007 highlighted a number of structure and process variables that were common to successful institutions, such as the importance of a collaborative environment, with strong surgical leadership and good relationships with referring physicians and cardiologists. Critical challenges faced by providers were also noted, including decreasing case volume and reimbursement, and the advancement of percutaneous technologies. The CCSIP will expand this program to build on the strengths of better performers, and transmit findings to less fortunate sites.

Supporting Documents (pdf files)

Documentation and Definitions for Hospital Dashboards

California Totals for all Procedures and Outcomes by Year (1997 - 2009)

Summary of Statistical Models used to Generate Predicted Risk of Outcomes

Details of Logistic Regression Model by Procedure

© 2011 California Society of Thoracic Surgeons