View source for Predicting the Adoption of Electronic Health Records by Physicians: When Will Health Care be Paperless?

Eric W. Ford, Ph.D., MPH, Nir Menachemi, Ph.D., MPH, M. Thad Phillips, MSHA, MSHI, MBA. 

J Am Med Inform Assoc. 2006;13:106-112. DOI 10.1197/jamia.M1913

'''Intro:'''  The authors analyze historic EHR adoption trends for small practices, and then project future adoption rates to identify a target date by which EHR adoption will be complete.  They then discuss policy options to facilitate earlier adoption for solo/small practitioners.  

'''Disclosures:''' Affiliations of the authors include the Center for Healthcare Leadership and Strategy, Texas Tech University, Lubbock, TX (EWF); Center on Patient Safety, Florida State University, Tallahassee, FL (NM); Department of Clinical Information Systems, University of South Alabama Hospitals, Mobile, AL (MTP).

'''Background:''' In 2004 President Bush signed the executive order creating the Office of the National Coordinator for Health Care Technology (ONCHIT) with the mission of implementing EHRs nationwide by 2014.  Progress towards that goal has been slow, especially among solo/small practices.  Slow progress calls into question whether the President’s target can be met. The authors base their research approach on diffusion of innovation theory.  Specifically, the Bass technology diffusion statistical model was employed.  This empirical model predicts how many customers will eventually adopt a new product and when, based on early market penetration rates.  The approach predicts the diffusion patterns of new technology as a function of external (innovation) and internal (social system) influences.  

'''Question:''' 1) Will the U.S. health system achieve universal EHR adoption by 2014?
	     2) If not, what is the likely time horizon?

'''Objective:''' The purpose of the study was to answer the questions by constructing three models that project likely adoption trends using historic data from published studies 2001-2003.  

'''Methods:''' The Bass statistical Model of Technology Diffusion was used.  First, historic data of small practice (10 or fewer practitioners) EHR adoption was quantified and graphically depicted.  Data were derived from 6 previously published studies.  Based on this analysis of historic trends, future adoption patterns were extrapolated and then discussed in terms of the two factors that drive innovation adoption processes- internal and external social influences.  The statistical methodology that allows forward projection is explained.

'''Key Findings:'''  The authors discern that market penetration will be achieved.  Three different approaches each yield the anticipated S shaped curve pattern associated with full adoption.  They predict that diffusion will plateau around 2024, reflecting adoption rates of 87-95%.  Tipping points occur between 2009-2012 at when adoption rates greater than 50% are achieved.
 
'''Discussion:'''  Concluding that full adoption of EHR use in small practices will be achieved, but not until about 2024, the authors turn discussion to areas of policy that could alter that timeline.  Noting that 60% of all physicians practice in groups of 10 or fewer, the authors argue that small ambulatory practices constitute the critical pathway for nationwide adoption of EHR.   

The authors compare adoption trends for EHR to those of other medical technologies such as ultrasound, mammography and computer assisted tomography, all of which diffused quickly.  These are instances of specific purpose technology.  EHR technology, alternatively, is a general purpose innovation which disrupts workflows and does not deliver productivity gains to the practitioners in the short term.  Thus a slower adoption curve is anticipated.  In areas such as Western Europe and Australia, where use of EHR has been widely adopted, there have been significant partnerships between government and physicians, or the costs of innovation have been highly subsidized.  In the U.S. the strongest external driving forces for EHR adoption are CMS and other large payer sources.  The move to P4P is likely to meet with strong resistance unless the needs of practitioners are accommodated.

The authors argue that relying on such external influencers is not likely to be a successful strategy.  Internal influencing factors (social contagions) appear to be more powerful in driving adoption.  Among the potential social contagions, several factors operate to slow adoption:

1) ROI does not accrue to the provider in the short term under many reimbursement plans, but typically flows to insurers as a reduction in use.
2) Initial purchase and operating costs are high, particularly given low ROI in the short term.
3) Vendor transiency creates risk borne by the small practitioner.  Most EHR companies are small and may not persist long in the industry.
4) Vendor transiency combined with limited adoption of standards create risk to the data.  If a vendor goes out of business, the practitioner is left with the costs of system migration, and the possibility that not all data will transfer.
5) Both initial system adoption and system changeovers affect workflows negatively and require a period of  adjustment, with anticipated loss in revenue.

Physicians tend to rely on their professional peers for information related to new technology.  These tend to be close-knit networks, and can view EHRs as a threat or annoyance.  The authors identify that there is extensive research on ways to influence physician social networks.  These include passive information dissemination via journals, articles, mailings (not very effective), use of “thought leaders” or champions (somewhat successful), and interactive educational strategies (most likely to be successful in penetrating social networks).  Among the educational strategies, integrating experience with EHRs in medical school and residency will graduate practitioners more acculturated toward the technology.  Provision of CME may have some impact, though research has shown limited impact on changing practice behaviors.  Lastly, academic detailing or in-depth focused sessions with individual practitioners, has the highest likelihood of changing behavior.  The authors conclude that these interactive educational strategies have the highest likelihood of hastening universal EHR adoption.

'''Reviewer Comments:'''  The application of diffusion of innovation theory, a broad and deep body of knowledge, to the adoption curve for EHR technology provides insight into the management of expectations and opportunities for policy intervention to accelerate technology adoption.  The adoption curve is predicated on studies conducted in 2001-2003.  It would be useful to validate this study by comparing its findings to those of more contemporary studies of EHR adoption, and to assess whether policy changes since 2003 are having an impact on the acceleration of EHR adoption.  This study provides a framework and benchmarks for ascertaining whether there is progress being made in the adoption of this important technology.

The study methodology does have its short comings, which the authors identify and discuss.  The study’s conclusion is based on the assumption that diffusion will follow an S-shaped trajectory.  However, especially given the potential to increase external drivers, and the possibility of leap-frogging advances in technology, it is possible that a discontinuous curve will be followed and the assumption may not hold.  Secondly, the study relies on previously conducted surveys of EHR adoption.  These have the potential for bias including: differing definitions of EHR, social desirability of response bias, and a possibility that early adopters were more likely to respond to a survey about technology than non-adopters.  These can be addressed through more comprehensive studies of adoption in small practices that use strong sampling methodologies.

[[user:milleas|Susan Millea]]
[[Category:OHSU-SP-07]]