Cardiovascular surgery risk prediction from the patient's perspective

Area of Science:

• Cardiovascular Surgery
• Health Services Research
• Medical Informatics

Background:

• Previous cardiovascular surgery outcome prediction models focused solely on patient risk factors.
• Patient-reported outcomes in cardiovascular surgery can vary significantly between healthcare facilities.
• There is a need to integrate hospital-specific factors into surgical outcome prediction.

Purpose of the Study:

• To develop novel cardiovascular surgery outcome prediction models.
• To incorporate preoperative patient risks alongside hospital processes and structures.
• To enhance the accuracy and patient-centricity of surgical outcome prediction.

Main Methods:

• Utilized data from the Japan Cardiovascular Database (2005-2007).
• Analyzed 33,821 cardiovascular surgery procedures across 102 hospitals.
• Employed logistic regression for risk model generation and validated using split-sample validation.

Main Results:

• Hospital factors like annual surgical volume and recommended staffing/equipment significantly impacted mortality.
• Public hospital status and surgeon workload (e.g., hours worked per week) were associated with patient outcomes.
• The developed mortality prediction model demonstrated strong discrimination (C-index=0.85).

Conclusions:

• The developed models offer robust discrimination and calibration for predicting cardiovascular surgery outcomes.
• These models can aid in hospital selection by considering individual patient risks and hospital characteristics.
• Optimizing surgeon work environments is crucial for improving both surgical team well-being and patient outcomes.