Choice of cardiac ‘operative mortality’ definition affects outcomes reporting

View on the News

Working in a flawed system

“The public, our patients, and their families might be surprised to discover that doctors have difficulty tallying how many people die after cardiac interventions, and that the question merited a piece of published statistical analysis,” Dr. Tom Treasure of University College London and Dr. Samer Nashef of Papworth Hospital, Cambridge, said in their invited commentary (J Thorac Cardiovasc Surg. 2016;151:1110-1).

Dr. Maximus and colleagues showed that the wider they cast their net to account for postoperative death, the more deaths they found, Dr. Treasure and Dr. Nashef said. That may penalize institutions that are more fastidious in collecting outcomes data. “When we require institutions with scarce resources to chase long-term outcome data for the purposes of treatment quality comparison, we immediately penalize those institutions that comply; more extensive data retrieval will find more deaths,” the commentators said.

The “perfect system” would capture death following a procedure no matter where it occurs, they added. “For the sake of compliance and simplicity, and until data systems are adequately robust and comprehensive, quality monitoring of cardiac surgery may have to be based pragmatically on data that are universally available and difficult to falsify, such as death at the base hospital during the same hospital admission as the intervention.”

Dr. Treasure and Dr. Nashef also accounted for another “quirk” in measuring outcomes. “Humans make errors in data entry, transcription and transfer and computers have ‘glitches’ and ‘gremlins,’ ” they said. “An apparent improvement in clinical outcome in fact may be merely the result of better record keeping.”

The commentators reported that they had no relationships to disclose.




Reporting of outcomes has grown in importance as clinical registries collect data and establish benchmarks for quality care, but exactly what those data report can vary depending on the definition of a specific outcome, even for an outcome as seemingly straightforward as death after an intervention.

Dr. Steven Maximus of the University of California Irvine Medical Center, and colleagues reported on this phenomenon in the April issue of the Journal of Thoracic and Cardiovascular Surgery (2016;151:1101-10). Specifically, they showed how rates of postoperative mortality after five different cardiac interventions can vary within the same data set depending on the definition of postoperative mortality.

“A significant percentage of procedural deaths occur after transfer or discharge from the index hospital,” Dr. Maximus and colleagues said. “These findings illustrate the importance of the definition of ‘operative’ mortality and the need to ensure accuracy in the reporting of data to volunteer clinical registries.”

They calculated outcomes based on five different definitions of postoperative mortality for five different cardiac interventions, depending on when and where the patient died, each of which showed slightly different results. The five different definitions of postoperative mortality are: during the index hospitalization only; during the index hospitalization or in a hospital the patient was transferred to on the same day or within 24 hours of discharge; during the index hospitalization or within 30 days after the procedure; during the index hospitalization or transfer hospitalization or during readmission within 30 days; and during the index hospitalization, transfer, and within 30 days after the procedure regardless of readmission.

As an example, mortality rates for percutaneous coronary intervention (PCI) with acute coronary syndrome (ACS) ranged from 3.25% for the first definition listed to 4.51% for the fifth definition. For isolated coronary artery bypass grafting, the rates ranged from 1.71% for the first definition to 2.15% for the fifth.

The fifth definition “is the most encompassing and is the current definition used by the STS [Society of Thoracic Surgeons] and should be applied to other clinical registries, such as the American College of Cardiology National Cardiovascular Data Registry,” Dr. Maximus and colleagues said.

The study noted that the Society of Thoracic Surgeons National Database (STS-NDB) has revised its definition of operative mortality to more accurately measure outcomes. Originally, the STS-NDB definition included all deaths during the index hospitalization, even those after 30 days, and postdischarge deaths within 30 days unless the cause of death was not related to the operation. In 2011, the definition was updated to include patients transferred to other acute care facilities. STS-NDB updated the definition again in 2014 to include all deaths, regardless of cause, during the index hospitalization, even if after 30 days, and including patients transferred to other acute care facilities, and all deaths, again regardless of cause, after discharge and within 30 days of the operation.

The analyses by Dr. Maximus and colleagues used data from the California Office of Statewide Health Planning and Development hospitalized patient discharge database for the year 2009. The first analysis the researchers performed did not exclude any patients and found a significant number of cardiac procedural deaths occurred after transfer at discharge from the index hospitalization, 17% in the surgical group vs. 31% in the PCI group. In the second analysis, which excluded untrackable patients, hospital deaths included 12% for PCI and 4% for surgery.

“PCI mortality was more dependent on the method used to define mortality, compared with the surgical patients, and a larger percentage of deaths occurred after hospital discharge and within 30 days of the procedure,” Dr. Maximus and colleagues said.

Another key factor in calculating mortality rates was the ability of hospitals to follow patients. “We found that up to 20% of hospitals were not able to track their patients long-term,” Dr. Maximus and coauthors said.

Study coauthor Dr. Junaid Kahn disclosed consulting fees from Edwards Lifesciences. The other authors have no financial relationships to disclose.

Recommended Reading

Cardiac cath lab work carries multiple risks
MDedge Cardiology
Angina rates similar across metal stents
MDedge Cardiology
Pre-PCI beta-blockers offer no clinical benefit
MDedge Cardiology
Endovascular thrombectomy procedure volume for stroke may not affect outcomes
MDedge Cardiology
Hybrid option ‘reasonable’ for HLHS?
MDedge Cardiology