Procedure-based severity index for inpatients: development and validation using administrative database.

Area of Science:

• Health Services Research
• Medical Informatics
• Clinical Epidemiology

Background:

• Risk adjustment is crucial for studies using administrative healthcare databases.
• The utility of procedure records for assessing patient severity in risk adjustment is not well-established.
• Developing a validated severity index from procedure data is needed.

Purpose of the Study:

• To develop and validate a severity index derived from procedure records for risk adjustment.
• To assess the usability of procedure data in administrative databases for predicting patient severity.
• To improve the accuracy of risk adjustment models in healthcare research.

Main Methods:

• Utilized the Japanese Diagnosis Procedure Combination database for acute-care hospitals.
• Randomly assigned 539,385 hospitalizations to derivation (n=270,054) and validation (n=269,331) cohorts.
• Applied multivariable logistic regression to identify procedures associated with in-hospital death and created a severity index.

Main Results:

• Identified 19 significant procedures, yielding a severity index range of -13 to 69.
• The mortality-predicting model incorporating the index achieved a c-statistic of 0.767 in the validation cohort.
• The index demonstrated significant contribution (ω-statistic: 1.09) to the model's predictive performance.

Conclusions:

• A procedure-based severity index effectively predicts in-hospital mortality.
• Procedure records within administrative databases are valuable for enhancing risk adjustment.
• This index offers a practical tool for improving severity assessment in healthcare research.