Provider Profiling Using Mixed Effects Models on a Case Study concerning STEMI Patients

Keywords

Statistical learning
Health Analytics
Code:
21/2010
Title:
Provider Profiling Using Mixed Effects Models on a Case Study concerning STEMI Patients
Date:
Wednesday 30th June 2010
Author(s):
Grieco, Niccolò; Ieva, Francesca; Paganoni, Anna Maria
Download link:
Abstract:
Provider profi ling is the evaluation process of the performance of hospitals, doctors, and other medical practitioners to enhance the quality of medical care. In this work we describe statistical analyses conducted on MOMI2 (MOnth MOnitoring Myocardial Infarction in MIlan) survey, a collection of data concerning patients admitted with STEMI (ST-Elevation Myocardial Infarction) diagnosis in one of the hospitals belonging to the Milan urban area Network, with the aim of pointing out process indicators to be used in health-care evaluation. An ff ective statistical support to decisional process for clinical and organizational governance is then obtained analyzing and modelling data coming from clinical registers and administrative data banks. Typically this kind of data are grouped at first level by structure where patients are admitted to, so multilevel models have been considered and tted to catch and explain overdispersion phenomena. Moreover, we compare three di fferent techniques of hospitals classi cation based respectively on traditional survival rates comparison, on analysis of variance components in fitted generalized linear mixed e ects models and non parametric random e ects estimation. Key words: Generalized Linear Mixed Models In-hospital survival ST-Elevation Myocardial Infarction Provider Pro filing
This report, or a modified version of it, has been also submitted to, or published on
IMA Journal of Management Mathematics ( first published online May 27, 2011)