乌鸦传媒

SAN DIEGO -- The growing population of adults with congenital heart disease (CHD) now has a dedicated operative risk algorithm for cardiac surgery, model developers reported.

Incorporating 47 newer diagnosis- and procedure-related variables, the tool showed "excellent" discrimination upon validation, with its C-statistic of 0.815 significantly better than the 0.791 of a model based on 57 older Society of Thoracic Surgeons (STS) adult risk variables (P<0.0001).

"Operationalizing the risk model and creating an online calculator are important next steps," Jennifer Nelson, MD, MS, of Nemours Children's Health in Orlando, Florida, told the audience at the STS annual meeting. A full report of the risk model's development and validation was published in the .

Nelson emphasized that the novel risk calculator may be applied to the entire adult CHD cardiac surgical population regardless of operative procedures performed.

"Patients with CHD often require many operations over a lifetime, and the timing of reoperation is commonly dictated by somatic growth and/or physiological changes. For example, prosthetic valves placed in childhood are not expected to last through adulthood," her group wrote.

"Also, ACHD [adult CHD] patients may have a higher burden of traditional cardiovascular risk factors compared to adults without CHD," the authors continued. "Thus, ACHD patients have the potential need for congenital cardiac operations to treat primary, residual, or sequelae CHD lesions, cardiac operations for acquired pathology such as coronary artery disease, or both."

Until now, the available adult STS models had excluded congenital procedures, while existing congenital models excluded operations for acquired disease.

Adult CHD patients pose unique challenges for risk modeling, given their heterogeneity and lack of a perfect data collection tool, commented Frank Scholl, MD, of Joe DiMaggio Children's Hospital in Hollywood, Florida, speaking as the STS session's invited discussant.

The U.S. does not track the number of adults surviving with CHD, but this is believed to be a growing population due to neonatal and childhood treatment.

Risk calculations for this population are important for patient counseling, clinical practice improvement, case mix adjustment, and quality measurement. It is estimated that there are now living with CHD.

Thus, the STS National Database is introducing a new ACHD module for enhanced data collection for this group.

"As additional preoperative factors become available for ACHD, recalibration of the mortality risk model, iteration, and expansion to additional endpoints will further enhance risk-adjusted performance metrics for this unique population," Nelson and colleagues predicted.

In particular, Scholl remarked on the importance of disease chronicity in CHD patients who often undergo multiple surgeries over a lifetime.

Creation of a good chronicity score would "start to get at the holy grail that is understanding the cumulative impact of patients' physiology of living decades with CHD, whether repaired or unrepaired," Nelson agreed.

The present ACHD surgical mortality model was developed and validated in the ACHD population, counting over 192,000 individuals, included in the STS Adult Cardiac Surgery Database from 2011 to 2019.

Participants had undergone various combinations of coronary artery bypass grafting and/or heart valve repair or replacement. The cohort was split into 60:40 groups for risk model development and validation, respectively.

The resulting model was well calibrated overall and performed well in 38 subgroups, Nelson reported.

However, certain high-risk subgroups were represented by relatively small samples, she said.

What's more, she cautioned, there were nearly 21,000 adult patient records from the STS Congenital Heart Surgery Database that could not be folded into the development of the present risk algorithm due to incompatible data elements between databases. This excluded cohort tended to be younger and more likely to have complex CHD compared with the study population.

Comment