Implications for Achieving the 2014 EHR Diffusion Goal
Two questions motivated this study. First, can the U.S. health care system achieve universal EHR adoption by 2014? Based on recent assessments of small practices' EHR adoption rates and prospective innovation diffusion modeling, we present evidence to suggest the answer is no. The question then becomes what is the magnitude of the challenge facing policy makers, or what is the most likely time horizon for universal EHR adoption? The most conservative estimate is that 86.6% of physicians in small practices will be using EHRs in 2024. In other words, at the current adoption rate, the goal of universal adoption will take more than twice as long as desired. This gives rise to a third question: Where should policy makers focus their efforts to accelerate EHR adoption?
Identifying the Providers to Target for EHR Adoption
Nearly 60% of all physicians practice in groups with ten or fewer doctors, and a significant number of all patient encounters therefore occur in these settings.25,26
Thus, small ambulatory practices are where the patient histories essential to an effective nationwide EHR system are generated. Further, physicians in small practices are likely to be among the late majority and laggards in the product adoption life cycle. Therefore, the critical path for effectively achieving the EHR capabilities sought in the 2014 universal adoption goal runs through small group and solo practices. The potential to accelerate the EHR adoption rate by increasing the external (p
) and internal (q
) influence coefficients among these providers is considered next.
External Factors Influencing EHR Adoption
The external diffusion coefficient estimates for EHRs are already relatively large compared to other medical equipment technologies, such as ultrasound imaging (p = 0.000), mammography (p = 0.000), and use of computed tomography scanners (p = 0.036), all of which diffused quickly.27,28
Compared to other consumer electronics designed to support decision making, such as calculators (p = 0.143) and personal computers (p = 0.121), the external influence coefficient is relatively small. However, reaching the tipping point in EHR adoption is qualitatively different from electronic tools such as calculators and computers.
Electronic health record implementations represent a disruptive change in the health care workplace. In addition to the introduction of new equipment, the job design of interconnected health professionals must be reengineered to effectively and efficiently accommodate the technology. In this respect, EHRs may follow the slower adoption pattern of “general purpose” technologies that are pervasive today, such as electric motors in manufacturing, which required the transformation of entire industries. General purpose technologies typically take relatively long periods to reach the diffusion tipping point and do not deliver productivity gains immediately upon arrival.29
The latter point has been a frequently identified barrier to EHR adoption.9
Nevertheless, other countries have been able to effectively promote EHR diffusion.
In other mature health care systems, such as Australia and Western Europe, various forms of EHRs have been widely adopted.30
In those systems, there have been significant governmental efforts to partner with physicians or subsidize the cost of the new technology, respectively. The policy mechanism most commonly discussed for increasing EHR's external influence coefficient in the United States is the introduction of clinical reporting mandates. The Centers for Medicare and Medicaid Services (CMS) has introduced several new reporting requirements for hospitals with quality improvement and cost control as the primary objectives.31
As reporting requirements increase, the only feasible mechanism for gathering such data will be the EHR. While such programs may be of some use, they may not advance the goal of full
EHR adoption significantly, because U.S. providers tend to respond negatively to such mandated-use policies,32,33
particularly in comparison to their international counterparts.34
Therefore, some external stakeholders are taking a more positive approach to accelerating EHR adoption rates.
Pay-for-performance (P4P) programs would reward physicians for using EHRs in their clinical practices. There are currently over 100 P4P programs in the United States designed to improve the quality of care and adherence to best demonstrated practices, often relying on EHRs to provide the required documentation.35
CMS, under its demonstration authority, intends to carry out P4P demonstration programs in the future related to EHRs. Cisco, a computer networking company, introduced a program that paid California physicians' groups more than $50 million for achieving key quality metrics and investing in EHR technology in 2004.36
Despite these positive incentives, some physicians see P4P programs as a third-party attempt to overly influence medical practice, decrease costs, and increase profits for payers.35
As such, relying solely on external influences to achieve full EHR diffusion by 2014 is unlikely to be a successful strategy. The internal influence factor appears to be more powerful for accelerating diffusion than the external one. This phenomenon is apparent by comparing the optimistic scenario's external and internal coefficients () to the other two scenarios' values. The external influence factor in the optimistic model is slightly lower than the other two models' values while the internal coefficient is markedly higher. This suggests that increasing the internal influences (e.g., social contagions) has a far greater impact on the overall adoption rate than a similar increase in external factors. Furthermore, it is possible for external stakeholders to have a positive impact on social networks' internal technology diffusion mechanisms, as noted below.
Internal Factors Affecting EHR Adoption
Compared to other medical technologies that diffused rapidly, such as ultrasound imaging (q = 0.510) and mammography (q = 0.738), the internal influence coefficients for all the EHR models are relatively low. In order to rapidly accelerate a technology's diffusion, it is essential to increase the internal or social contagion factors that influence adoption decisions. Otherwise, EHR adoption rates among small practices will remain relatively low and time horizons for complete adoption will remain distant.
One aspect of adopting EHRs that physicians in small practices have had to internalize is the system's initial purchase and ongoing operational costs. The return on investment for an EHR system does not accrue to the provider in the short run under many reimbursement schemes.37
Instead, the savings from improved care efficiency and quality typically flow back to health care insurers or payers as a reduction in service use.38
Another significant barrier to adoption has been vendor transience; many early EHR companies are no longer in business or are in precarious financial positions.39
The adoption risk associated with vendor volatility could be mitigated if a common data standard were implemented across the sector. There would still be significant changeover costs in the event of a vendor failure, but the initial cost of creating the EHRs would not be totally lost. In addition to the monetary costs, system changeovers negatively affect physicians' workflows, something they are keen to avoid.40
Physicians have historically relied on their professional peers as their primary source of information related to new technologies.41,42
The medical community's professional culture makes it a very close-knit social network that views external attempts at instituting controls as an assault on its autonomy.43
Further, the physician community does not, in general, have a strong grasp of the quality improvement processes that are being targeted at them.44
Collectively, the medical community's social mechanisms that influence adoption decisions view EHRs as a potential threat to professional autonomy. This may be particularly true among physicians in small practices who value the freedom and autonomy they provide.
There is extensive research on ways to influence physicians' internal social networks. Passive dissemination strategies, such as journal articles and mailings, are typically ineffective.45
The use of “thought leaders” to influence social networks and change clinical behaviors has experienced some success. However, given that many of targeted adopters are working in solo practices, this may not be a broadly applicable intervention. Therefore, an interactive educational strategy is likely to be most influential in penetrating physicians' social networks, particularly those in small practices.
Measuring the Level of EHR Diffusion
The present study's projections are limited in two respects. From a theoretical perspective, the future diffusion of EHRs may follow a discontinuous rather than an S-shaped trajectory. Under such conditions, the EHR adoption rate will not grow in a gradual, evolutionary process, but rather a series of revolutionary leaps forward will occur as external pressures and new product innovations increase.51
Given the significant amount of activity in the health care arena related to increasing EHR adoption, this scenario is one that may occur.
A second limitation in the study's design is that it relies on previously conducted survey estimates regarding historic EHR adoption rates. Consistent with all survey methodologies, the results of those studies may have been biased in an upward direction. This may have occurred because physicians who already used EHRs, “early adopters,” were potentially more likely to respond to inquiries about such systems compared to nonusers. Also, respondents to the previous surveys may have provided answers to questions in a socially desirable manner. In such instances, the inclination would be to respond positively on familiarity and frequency of EHR use. A third potential source of bias lies in how EHRs were defined in previous studies. Respondents may have viewed their nonclinical automated systems (i.e., electronic scheduling and billing) as EHRs. Moreover, users of less robust systems may have responded positively despite the fact that key capabilities of a minimal EHR may not have been present.52
All these biases serve to inflate the previous estimates of EHR adoption. Therefore, even our conservative estimates of future adoption trends may be overstated, creating a need for more rigorous studies.
Both limitations can be addressed by conducting more comprehensive surveys of small practices' EHR use. In particular, tracking the incidence of EHR adoption over time, using accepted statistical approaches and national sampling methodologies, would be helpful. The Agency for Healthcare Research and Quality has established the National Resource Center for Health Information Technology to monitor and disseminate information about EHR diffusion.53
Therefore, it seems likely that improved assessments of adoption rates will be forthcoming.