Columbia Technology Ventures

Statistical model for congenital heart surgery risk assessment

This technology is a statistical risk model for pediatric and congenital heart surgery combining ICD-10 (International Statistical Classification of Diseases and Related Health Problems-10) administrative data with previous risk models to allow for greater predictive performance.

Unmet Need: Risk adjustment model based on recent administrative data

Current methods of performing patient-risk assessments rely on administrative data collected prior to 2015, limiting their accuracy and sensitivity. The combination of current models and updated administrative data can improve the quality of the risk assessment.

The Technology: Data-driven risk model for congenital heart surgery

This risk model not only is based upon previous technology such as RACHS-1, STS-CHSD, and AHRQ models, but also incorporates additional steps to work with current administrative data and to incorporate a more complete list of procedures. It can be used to assess pediatric cardiac surgery risk more accurately and precisely than existing models. Moreover, with the model being compatible with current administrative data, this adaptable technology can be further refined in its risk assessment for cardiac surgery and other medical procedures.

This technology has been validated with data provided by the NYS Congenital Heart Surgery-Collaborative for Longitudinal Outcomes and Utilization of Resources (CHS-COLOUR), NY State Department of Health, the Pediatric Health Information Systems, NewYork-Presbyterian Hospital, Cincinnati Children’s Hospital, and Cleveland Clinic.

Applications:

  • Statistical risk evaluation for pediatric and congenital heart surgery
  • Healthcare quality improvement
  • Hospital benchmarking assessment
  • Public outcome reporting of hospital performance
  • Research tool for public health outcomes

Advantages:

  • Improves accuracy, sensitivity, and specificity compared to existing risk adjustment models
  • Saves time and money for hospitals
  • Better prediction of healthcare costs for patients
  • Compatible with conventional ICD-10 administrative data
  • Can be expanded to other patient characteristics and social determinants of health

Lead Inventor:

Brett R Anderson, M.D. M.B.A. M.S.

Related Publications:

Tech Ventures Reference: