It's time to face the facts: Our nationwide network of Health Information Exchanges (HIEs) is an unmitigated disaster. The President's Council of Advisors on Science and Technology (PCAST) said in December, 2010, that HIE efforts through the states "will not solve the fundamental need for data to be universally accessed, integrated, and understood while also being protected." A recent survey of 179 HIEs found only 13 (covering just 3 percent of hospitals and 0.9 percent o
It's time to face the facts: Our nationwide network of Health Information Exchanges (HIEs) is an unmitigated disaster. The President's Council of Advisors on Science and Technology (PCAST) said in December, 2010, that HIE efforts through the states "will not solve the fundamental need for data to be universally accessed, integrated, and understood while also being protected." A recent survey of 179 HIEs found only 13 (covering just 3 percent of hospitals and 0.9 percent of physician practices) capable of meeting Stage 1 Meaningful Use criteria. Of those, only six were reported to be financially viable. More importantly, none of the HIEs surveyed had the capabilities of a comprehensive system as specified by an expert panel, calling "into question whether RHIOs in their current form can be self-sustaining and effective."
Not surprisingly, multiple HIEs have been shut down, e.g., Washington (DC), Kansas, Tennessee, CalRHIO, and CareSpark (Kingsport, TN, once touted as a national leader). According to ONC, in 32 states more than 90 percent of hospitals have not even exchanged a single patient record. Finally, how many patients today get their care with comprehensive information from all sources? Zero.
One consequence of all this is that the promise of lower healthcare costs from HIT is unfulfilled – in fact, costs are increasing because of the improved documentation produced by EHRs. HIT cost savings will come only from eliminating duplicate tests and avoiding medical errors – this requires comprehensive records for each patient, not just provider EHRs.
Why is this happening? It's certainly not for lack of funding – states have been working on HIEs with $564 million in Federal funds since March, 2010. It's not that we don't know the key obstacles: Privacy, stakeholder cooperation, and financial sustainability have been identified as key issues again and again in report after report. But these barriers have not been successfully overcome.
The problem is that we're on the wrong path. We're trying to build institution-centric systems that leave patient information where it's created and retrieve and integrate it in real time only when it's needed. This seriously flawed approach is:
1. Complex and expensive. Requires all EHR systems to be online 24/7 and to respond to queries. Requires real-time reconciliation of records from multiple sources. Needs unique identifier (politically impractical and a threat to privacy). Needs a 24/7 network operations center to monitor and ensure availability of all possible sources of information.
2. Prone to error. A recent simulation study showed that, in comparison to a central repository, the institution-centric approach requires exponentially more transactions and is exponentially more likely to provide incomplete records due to EHR response failures.
3. Insecure. Maintaining security in a highly distributed system such as this is an unsolved computer science problem – the first step in securing data is to put them in a repository where they can be protected.
4. Not financially sustainable. A widely acknowledged problem.
5. Unable to protect privacy. Leaving data at their source makes managing privacy preferences impossibly complex since consumers would need to set and maintain permissions at each location where they have ever received care.
6. Unable to assure stakeholder cooperation, since providing records is totally voluntary.
7. Unable to facilitate searching the data, e.g., for research and public health purposes. Since each record must be fetched from all sources and integrated every time it is needed, searches must be sequential, which is prohibitively inefficient.
Einstein famously defined insanity as doing the same thing over and over and expecting different results. The current approach to HIEs does not and will not work. If we want to succeed, we must try something else.
The good news is that there is at least one viable alternative that is simpler, scalable, less expensive, more secure, and can provide lifetime records: patient-centric community health record banks. Each health record bank provides individual accounts containing copies of medical records and additional information that optionally may be added by the consumer. The consumer explicitly controls who may access which parts of the information in the account (for details, see the Architecture and Business Model white papers from the Health Record Banking Alliance). This approach solves the problems of privacy (through patient control), stakeholder cooperation (since the patients request their own records, HIPAA requires every stakeholder to provide them), and financial sustainability (with revenue from optional apps for patients and research use of the data with permission).
Are we going to try this (or some other promising new approach) or stubbornly continue our inexorable march to failure?
William Yasnoff, MD, PhD, FACMI, is a health IT consultant and President of the Health Record Banking Alliance. His prior work at HHS resulted in the creation of ONC in 2004. He is the author of the Health Information Infrastructure chapter in the textbook Biomedical Informatics (Shortliffe & Cimino, eds.).
1. President's Council of Advisors on Science and Technology.  Realizing the Full Potential of Health Information Technology to Improve Healthcare for Americans: The Path Forward. Washington, DC: Executive Office of the President, p. 40.
2. Adler-Milstein J, Bates DW, and Jha AK.  A survey of health information exchange organizations in the United States: Implications for meaningful use. Ann Intern Med 154:666-671.
3. Lapsia V, Lamb K, Yasnoff WA.  Where should electronic records for patients be stored? Int J Med Informatics 81(12):821-7.