Skip to main content
Innovative Methods – New Tools

A Community Hie-based Hospital Readmission Risk Prediction And Notification System

  • PI:               Hadi Kharrazi, MD, PhD
  • Funder:      AHRQ
  • Status         Results published 

Current readmission risk predictive models (RRPMs) are developed based on post-facto health plan claims and hospital administrative databases. RRPMs are rarely derived or validated using real-time Health Information Exchange (HIE) data. The aim of this research study is to develop and evaluate HIE-based real-time readmission probability risk scores for patients discharged from Maryland hospitals. This research also aims to assess and improve the RRPM’s predictive accuracy via an iterative process that will integrate and evaluate non-HIE data sources in order to explore data soon to be available to the HIEs.

Common HIE data streams/structures are a viable source of information to develop and evaluate RRPMs; however, pilot findings indicate that the variability of data patterns received from different HIE stakeholders (i.e., various hospitals) may limit the generalizability of fixed predictive models. The study team has envisioned a dynamic approach to generate a library of RRPMs customized to various stakeholders of HIEs.

The policy outcomes of this research concentrate on two non-economic perspectives: (a) Inter-provider readmissions that is when a patient is readmitted to another hospital within 30 days of discharge: CMS does not clarify the policies of reimbursement adjustment for inter-provider-caused readmissions. The results of this study will shed light on the effect of inter-provider readmissions when calculating readmission risk scores. (b) Future data integration: HIEs are constantly incorporating, or planning to incorporate, new health data sources. This project will address the value of expanded data sources for HIE integration to improve the performance of HIE-derived RRPMs. 


M.J. Swain, H. Kharrazi. Feasibility of 30-day hospital readmission prediction modeling based on health information exchange data Int. J. Med. Inform., 84 (12) (2015), pp. 1048-1056.

Kharrazi, H., Weiner, J.Z., Sylvia, M., Brotman, D., Lemke, K.W., Richards, T., Lasser, E.C., Gharghabi, F., & Cropp, B. (2016). Development of a community-wide real time health information exchange-based hospital readmission risk prediction and notification system for office-based practices : A feasibility and evaluation study.