Inpatient mortality after orthopaedic surgery.

Area of Science:

• Orthopaedic Surgery
• Health Services Research
• Biostatistics

Background:

• Accurate comorbidity risk adjustment is crucial for reliable outcome prediction and provider performance evaluation in orthopaedic surgery.
• Existing methods like Charlson and Elixhauser scores were not specifically validated for this patient population.
• There is a need for a refined risk-adjustment tool tailored to orthopaedic patients.

Purpose of the Study:

• To develop and validate a novel, single numeric comorbidity score for predicting inpatient mortality in orthopaedic surgery.
• To compare the predictive performance of the new score against the standalone Charlson and Elixhauser measures.
• To evaluate the new score's efficacy across specific orthopaedic subspecialties, including spine surgery, adult reconstruction, hip fracture, and musculoskeletal oncology.

Main Methods:

• Utilized data from the National Hospital Discharge Survey (1990-2007).
• Developed a combined comorbidity score by integrating and reweighting conditions from the Charlson and Elixhauser indices.
• Derived weights from an 80% training cohort (n=26,454,972) and validated internally on a 20% hold-out cohort (n=6,739,169).
• Assessed and compared score performance using the area under the receiver operating characteristic curve (AUC) from logistic regression models.

Main Results:

• The new combined comorbidity score demonstrated superior predictive performance (AUC=0.858) compared to the Charlson score (AUC=0.794) and the Elixhauser score (AUC=0.845).
• The combined score outperformed Charlson by 58% and Elixhauser by 12%.
• The new score exhibited enhanced discriminatory power across all evaluated orthopaedic admission subgroups.

Conclusions:

• A single, combined comorbidity score derived from Charlson and Elixhauser conditions offers improved prediction of inpatient mortality in orthopaedic surgery.
• This novel score surpasses the predictive capabilities of its individual constituent models.
• Further validation in diverse populations and data settings is recommended.