Blog | Wednesday, February 13, 2013

The health IT "Grand Experiment": mid-study check-up

It seems that whenever there is something negative about health information technology (HIT) in the popular press, I get emails from people inside and outside the field, asking "What's wrong?"

A case in point is a recent article in the New York Times 1, reporting on a "negative" point of view from two researchers from the RAND Corp. that was published in the journal Health Affairs 2.

One of the interesting twists of the Health Affairs piece was that it was written by researchers from RAND, the same organization that published a modeling study in 2005 that concluded that investment in HIT could provide potential annual savings to the healthcare system of $142-$371 billion 3. About the same time, another model-based study from the Center for Information Technology Leadership (CITL) found similar potential savings 4. This data in part led to the inclusion of HIT in the Health Information Technology for Economic and Clinical Health (HITECH) Act, the program from the American Recovery and Reinvestment Act (ARRA) of 2009, also known as the economic stimulus bill, that invested up to $29 billion in the adoption and "meaningful use" of the electronic health record (EHR) and other HIT 5.

What can we conclude from this recent publication and reporting about it in the popular press? As always, it is best to look at exactly what has been claimed, what evidence supports it, and where it fits in the larger picture of this topic.

The 2005 RAND study modeled savings that could occur from HIT adoption 3:
--reduced adverse drug events that extend hospital length of stay in the inpatient setting and avoid hospitalization in the outpatient setting,
--increased used of cost-effective immunizations and screening interventions, and
--improved efficiency of chronic disease management.

Their model, however, noted that HIT adoption alone would not be enough; also required would be "interconnected and interoperable systems" that were "adopted widely" and "used effectively." This had an implicit assumption of change in the healthcare delivery system away from payment for volume toward payment for value. The paper describing this work was published in Health Affairs, along with several dissenting views 6-8. An analysis by the Congressional Budget Office also took issue with the conclusions 9.

The CITL study used a somewhat similar modeling approach and drew similar conclusions. The CITL model focused on different types of health information exchange (HIE), from simple transmission of documents to full semantic interoperability of EHR systems. The latter approach was shown to achieve the most benefit, up to $77 billion per year.

Can we assess the correctness of these modeling studies, now that we have substantially increased EHR adoption through HITECH? The recent paper from RAND noted that the question is not simple to answer, but that HIT probably has fallen short of its promises, especially in terms of reducing costs 2.

Of course, one of the challenges in answering the question of cost-reduction is that it is difficult to attribute avoidable cost in the healthcare system. We do know that healthcare costs have reduced their rate of growth in the last few years, probably mainly due to the economic recession 10. But we cannot know for sure how much of that reduction might be due to HIT adoption.

But an even bigger reason why we cannot know if the modeling studies are true is that we have achieved the kind of HIT environment that these studies assumed in the development of their models. The original RAND study assumed, as noted above, interconnected and interoperable systems that were adopted widely and used effectively. The authors of the new RAND paper note that HIT failure has come in large part because of failure to reach those assumption. In particular:
--We do not have interconnected and interoperable systems. In part, this is because many EHR systems are still closed and proprietary. In addition, HIE efforts are still early and nascent.
--We also do not have wide adoption yet of systems, especially advanced systems. While HITECH has led to increased adoption, there is still a long ways to go.
--And probably the biggest shortcoming has been lack of EHRs being used effectively. The adoption incentives in Stage 1 of meaningful use focus (by design) on building the data foundation. More effective use will come based on that foundation in Stage 2 and beyond.

The RAND authors conclude that the potential of HIT in reducing costs is still very real, but critical focus on interoperability, patient-centeredness, and usability must be prioritized.

Therefore my view echoes that of the RAND researchers in the new Health Affairs piece, which is that yes, HIT has not yet delivered on its promise to improve efficiency and reduce cost in the health care system.

But the proposition that it inherently is not able to do so is also not known. As such, if we hope for that improvement, the grand experiment should go on. There is no question that the required time will be longer, the resources required will be larger, and the cultural change will be more difficult.

There is also quite valid concern that there are some untended consequences of the staged approach in HITECH, which may be locking clinicians and hospitals into monolithic systems that are difficult to use and expand. I sympathize with the notion of current market-leader systems locking us into an "EHR trap," where the EHR should not be a monolithic application but instead a platform on top of which we can build apps that provide innovative functions and/or make new use of the data 11.

Over the last few years, I have ended many a talk on informatics noting that a "grand experiment" in our field was taking place, with the complete results unlikely to be years away. This study can be viewed as a mid-study assessment, and we can conclude that the benefits have not yet accrued, but that it may be too early to conclude that they will not occur.

Although I agree that we probably need some mid-course correction in our approach, I also argue that we cannot go back nor should we end the experiment prematurely. We also must remember the motivations for implementing HIT and reforming healthcare in the first place, which is the error-prone and financially dysfunctional existing system, which both undermines competitiveness of U.S. companies globally due to high employee healthcare costs as well as threatening to bankrupt the U.S. government through unsustainable Medicare cost increases.

1. Abelson R and Creswell J, In Second Look, Few Savings From Digital Health Records, New York Times. January 10, 2013.
2. Kellermann AL and Jones SS, What will it take to achieve the as-yet-unfulfilled promises of health information technology? Health Affairs, 2013. 32: 63-68.
3. Hillestad R, Bigelow J, Bower A, Girosi F, Meili R, Scoville R, et al., Can electronic medical record systems transform health care? Health Affairs, 2005. 24: 1103-1117.
4. Pan E, Johnston D, Walker J, Adler-Milstein J, Bates DW, and Middleton B, The Value of Healthcare Information Exchange and Interoperability. 2004, Center for Information Technology Leadership: Boston, MA.
5. Blumenthal D, Launching HITECH. New England Journal of Medicine, 2010. 362: 382-385.
6. Himmelstein DU and Woolhandler S, Hope and hype: predicting the impact of electronic medical records. Health Affairs, 2005. 24: 1121-1123.
7. Goodman C, Savings in electronic medical record systems? Do it for the quality. Health Affairs, 2005. 24: 1124-1126.
8. Walker JM, Electronic medical records and health care transformation. Health Affairs, 2005. 24: 1118-1120.
9. Orszag P, Evidence on the Costs and Benefits of Health Information Technology. 2008, Congressional Budget Office: Washington, DC,
10. Hartman M, Martin AB, Benson J, and Catlin A, National health spending in 2011: overall growth remains low, but some payers and services show signs of acceleration. Health Affairs, 2013. 32: 87-99.
11. Mandl KD and Kohane IS, Escaping the EHR trap--the future of health IT. New England Journal of Medicine, 2012. 366: 2240-2242.
This post by William Hersh, MD, FACP, Professor and Chair, Department of Medical Informatics & Clinical Epidemiology, Oregon Health & Science University, appeared on his blog Informatics Professor, where he posts his thoughts on various topics related to biomedical and health informatics.