i n t e r n a t i o n a l j o u r n a l o f m e d i c a l i n f o r m a t i c s 8 1 ( 2 0 1 2 ) 821–827 j ourna l homepage: www.ijmijournal.com Where should electronic records for patients be stored? Vijay Lapsia a,* , Kenneth Lamb b , William A. Yasnoff c,d,e a Department of Medicine, Mount Sinai School of Medicine, New York, United States b The Chronic Disease Research Group, Minneapolis, MN, United States c NHII Advisors, Arlington, VA, United States d Division of Health Sciences Informatics, Johns Hopkins University, Baltimore, MD, United States e Institute for Healthcare Informatics, University of Minnesota, Minneapolis, MN, United States a r t i c l e i n f o Article history: Received 4 May 2012 Received in revised form 10 August 2012 Accepted 22 August 2012 Keywords: National Health Information Infrastructure Electronic Medical Record Health Information Exchange Health Record Bank Personal Health Record Patient-Centered Medical Home a b s t r a c t Introduction: The importance of a nationwide health information infrastructure (NHII) is widely recognized. Patient data may be stored where it happens to be created (the dis- tributed or institution-centric model) or in one place for a given patient (the centralized or patient-centric model). Minimal data is available regarding the performance implications of these alternative architectural choices. Objective: To help identify the architecture best suited for efficient and complete nationwide health information exchange based on the large-scale operational characteristics of these architectures. Design: We used simulation to study the impact of health care record (data) fragmentation and probability of encounter on transaction volume and data retrieval failure rate as markers of performance for each of the above architectures. Results: Data fragmentation and the probability of encounter directly correlate with trans- action volume and are significantly higher for the distributed model when the number of data nodes >4 (p < 0.0001). The number of data retrieval failures increases in proportion to fragmentation and is significantly higher for the distributed model when the number of data nodes ≥2 (p < 0.0059). Conclusion: In simulation studies, the distributed model scaled poorly in terms of data avail- ability and integrity with a higher failure rate when compared to the centralized model of data storage. Choice of architecture may have implications on the efficiency, usability, and effectiveness of the NHII at the point of care. © 2012 Elsevier Ireland Ltd. All rights reserved. 1. Introduction The Health Information Technology for Economic and Clinical Health (HITECH) Act calls for the Office of the National Coor- dinator for Health Information Technology (ONC) to develop “a nationwide health information technology infrastructure that allows for the electronic use and exchange of health ∗ Corresponding author at: 1468 Madison Ave, Box 1243, New York, NY 10029, United States. Tel.: +1 212 241 2264; fax: +1 212 987 0389. E-mail address: vijay.lapsia@mssm.edu (V. Lapsia). care related information” [1]. Through this legislation, the fed- eral government has committed unprecedented resources on a multiyear incentive program to support the adoption and use of electronic health records (EHRs) [2]. Getting health- care data in an electronic format is foundational to eventually enabling patient data sharing. Ultimately, the goal of this nationwide health information infrastructure (NHII) is to provide secure access to comprehensive electronic patient 1386-5056/$ – see front matter © 2012 Elsevier Ireland Ltd. All rights reserved. http://dx.doi.org/10.1016/j.ijmedinf.2012.08.008