Identification of hospital cost drivers using sparse group lasso

Area of Science:

• Health Economics
• Health Services Research
• Public Health Policy

Background:

• Public hospitals represent a significant government expenditure.
• Variability in hospital costs between and within institutions suggests potential financial waste.
• Identifying cost drivers is crucial for efficient healthcare resource allocation.

Purpose of the Study:

• To identify prime cost drivers in Australian public hospitals.
• To differentiate between warranted and unwarranted cost variability.
• To estimate the proportion of healthcare spending that may represent waste.

Main Methods:

• Principal Component Analysis (PCA) for dimension reduction and noise separation.
• Sparse group lasso technique to adjust for co-linearity of cost drivers.
• Statistical modeling of over 50,000 hospital admissions with 32 cost predictors.

Main Results:

• Identified key cost drivers contributing to hospital expenditure variability.
• Estimated that 33.7% of cost variability may not be clinically warranted.
• Demonstrated a novel approach to analyzing hospital costs using advanced statistical methods.

Conclusions:

• A substantial portion of public hospital spending may be inefficient.
• Findings have significant implications for national health funding policy and resource optimization.
• The methodology offers a reliable framework for future hospital cost analyses.