Patient safety analyses using Lombardy administrative archives

The purpose of this paper is to discuss strategies for benchmarking patient safety using Lombardy region administrative archives. Patient safety indicators and statistical methods are presented that allow risk adjustment. The analysis benchmarks regional health structures, focusing on two patient safety indicators: failure to rescue; and death in low mortality diagnostic related group.

Data were drawn from a research project promoted by the Italian Agency of Regional Health Services in 2002 to furnish statistical evidence regarding adverse events based on Agency for Healthcare Research and Quality indicators and methods. Hierarchical models for an equitable benchmark analyses are proposed.

Empirical analysis shows that hierarchical approaches, based on comparing health structures within homogenous specialties, disaggregates and moderates failure to rescue variabilities existing between hospitals, especially in oncology, intensive care and general medicine.

The paper proposes using hierarchical models for properly benchmarking health structures, resolving logistic regression drawbacks and limitations.

The paper strengthens the theory that accurate coding supported by software and administrative databases could provide a valuable and economical source for patient safety research.

The paper analyses and suggests strategies for consistent benchmark analyses based on patient safety outcomes, applicable to several situations and different health structure typologies.