Uncategorized

Diffusion of EHR Innovation

No matter what your opinion of Electronic Health Records (EHR) is, you would probably agree that the concept of computerizing medical records represents an innovation of sorts. The spread of innovation, or its diffusion, has been researched and modeled by Rogers[1] as a bell shaped advancement through populations of Innovators, Early Adopters, Early Majority, Late Majority and Laggards (the blue curve in the figure below). At some point during this spread of an innovative solution a Critical Mass of adopters, or Tipping Point, is reached and the innovation is assured widespread diffusion (Gladwell[2]).  Adoption is usually described by an S-shaped curve of adopters vs. time, and the rate of adoption is the slope of the S-shaped curve at any given time (the red curve in the figure).

The Tipping Point occurs right after the rate of adoption assumes its largest value which will be maintained throughout most of the adoption process. It is worth noting that the diffusion of innovation model is not predictive. Many innovations linger and die within the Innovator circle. Another important aspect of the model is that the time variable is not constrained. Depending on the rate of adoption, it may take weeks, months or many years for an Innovation to spread throughout a given population. There is no question that EHR adoption is slowly moving up on the ascending side of a classic diffusion model bell curve, but is it moving fast enough? Is the tipping point visible? Are we there yet?

DiffusionsIn order to answer these questions and assess where we are on the path to EHR adoption, we must examine the elements affecting diffusion of new ideas, objects or practices, i.e. Innovations, as they pertain to EHRs.

The Innovation – Not all things classified as Innovations are equal and this may explain why some succeed in becoming main stream and others fail. Rogers[1] suggests four defining characteristics of innovation:

  • Relative Advantage – To what degree is the EHR perceived to be better than the practice it aims to supersede – paper charts? The daily debates on the merits of EHR in various health care forums are stark testimony to the fact that potential users are sharply divided on the answer to this question. For some, EHRs are increasing efficiency and improving quality of care for patients. For others, EHRs are impediments to quality of care and an endless money pit for the practice. The biggest advertised advantage of EHRs, interoperability, is little more than a promissory note issued by EHR proponents to prospective adopters. The Government is adding an advantage in the form of stimulus incentives and future financial penalties for non-adoption. This advantage seems to be significant for hospitals and large groups, but less so for small private practices.
  • Compatibility – Are EHRs compatible with existing values, past experiences and needs of potential adopters? Here EHRs are being propelled onto the much larger stage of health care reform. They are no longer a humble replacement of pen & paper and fax machines. EHRs are instruments of change; a change from treating one patient at a time the best you can, to considering value-based strategies for the benefit of entire populations, and considering those right at the point of care. This seems to be a major departure from a value system created and enriched across many generations of medical doctors.
  • Complexity – Simplicity is always a virtue. The complexity of hospital EHRs, with their unwieldy CPOE modules, has created a perception of EHRs being rigid and unduly complicated tools, which take years to master.  The simple ambulatory EHRs available today, have failed to change these perceptions. To be fair, EHRs are inherently more complex than a piece of paper, but that should not necessarily deter adoption. After all, the wheel was more complex than walking, and making fire was extremely hard initially.
  • Trialability – Experimenting with small parts of an innovation before taking the final leap reduces adopters’ risk and anxiety. EHRs can be, and mostly are, implemented in stages, particularly in hospitals. In the ambulatory sector there was a trend to implement electronic prescribing as a trial before complete computerization. Perhaps the best exercise of Trialability for EHRs is the free trials offered by too few vendors.
  • Observability – An innovation is more likely to be adopted if its results are easily visible to others. This of course assumes that the results are positive. Unfortunately, successful implementations of EHRs are uneventful and largely anonymous, while their failing counterparts, usually associated with astronomic losses in funds and sometimes lives, are very visible, heavily advertised and frankly more interesting. There is no news in a 3 physician practice in Omaha installing a mid-priced EHR and having no problems to report.

Communication Channels – The news and evaluation of an innovation are spread throughout the community by various means, from mass media to informal peer to peer communications. As every EHR vendor knows, the latter is the most prevalent and effective method of disseminating messages amongst physicians. It is also a very slow method of creating awareness and unless the innovation is indisputably positive, much confusion is created by conflicting messages from friends and colleagues, thus slowing down the diffusion process. The Government’s intervention, with all its glorious publicity and billions of dollars in support of the innovation, is doing wonders in bypassing the conventional “word of mouth” diffusion mechanism. The downside is that EHR is now associated with Big Government and a particular flavor of politics.

Time – Time is involved in diffusion of innovation in several ways.

  • On an individual level the innovation–decision process goes through five stages known as: knowledge – persuasion – decision – implementation – confirmation. In EHR industry parlance, these translate to: research – assessment – selection – implementation – adoption. The shorter the innovation-decision individual cycle is and the more people actually complete it without dropping out in its midst and the more positive their adoption experience is, the faster an innovation is expected to diffuse. For EHRs this cycle can range anywhere between a few months for a small practice to several years for a large hospital. The Government imposed Meaningful Use schedules are shortening the innovation-decision cycle for those racing to qualify for maximum incentive. Perversely, the inadequate time allowed for implementation will also increase failure rates and adverse events, which does not bode well for long term diffusion rates.
  • Diffusion is also affected by the Innovativeness of the population targeted by the innovation. Here is where we encounter the accusations of physician being inherently opposed to technological advancements, and the counter arguments based on the number of iPhones and iPads already owned by physicians, not to mention all the advanced technologies in imaging, surgery and other medical fields, which are readily embraced by the medical community. When it comes to EHRs though, most docs don’t mind being very late adopters or even laggards.
  • The rate of adoption dictates how much time it will take for an innovation to diffuse throughout the system. Unfortunately for EHRs, the rate of adoption is heavily dependent on the 5 characteristics of an innovation and none of those are particularly stellar for EHRs. This is why the rate of EHR adoption prior to HITECH has been lingering at the bottom of a very wide S-curve. The Government intervention, which as mentioned above is increasing the financial advantage, is making a marked difference in the rate of adoption effectively pushing EHRs up the S-curve.

The Context – Innovations are diffused within the boundaries of a social system. The structure, norms and leadership of a system also affect the diffusion of innovations.  Systems whose members are similar in education, social status and beliefs (homophilous) are not well suited to rapid change and innovation. Physicians arguably do form such system. The historical low rates of EHR adoption could be attributed to lack of accepted opinion leaders in general, and those who view EHRs as a positive innovation in particular. The public personas that have an MD after their name and relentlessly advocate for EHR adoption are usually not practicing physicians and as such, are not accepted as respected opinion leaders by most practicing doctors.

The diffusion patterns described so far are assuming that the innovation is optional for any given individual. However, innovations are not always an individual choice. Sometimes the decision to innovate is Authority Driven. Authority driven innovations are faster to be adopted and depending on the level of coercion, may follow a completely different path. Up to HITECH, individual physicians in private practice considered EHRs optional. Those employed by hospitals or large groups were experiencing the effects of authority driven innovation all along, thus the much larger adoption rates in those sectors. Although EHRs are not yet mandatory, the increasing pressure exerted by Government incentives, regulations and penalties is changing the diffusion patterns of the EHR innovation.

Keep in mind that Government exertions do not need to continue until every physician in the country has purchased an EHR. They only need to ensure a critical mass of EHR adopters is created and the mythical Tipping Point is reached. The Tipping Point is usually observed at about 15% adoption under normal circumstances and is marked by the emergence of opinion leaders who adopted the Innovation. EHRs and health care in general are anything but normal and I would expect a larger percentage of adoption to be required before EHRs “take off” in a self-sustaining fashion. As to respected opinion leaders, there are none.

So are we there yet? I don’t think so, but we are awfully close.

1. Everett M. Rogers (1995), Diffusion of Innovations (Fourth Edition), New York, Free Press.
2. Malcolm Gladwell (2000), The Tipping Point: How Little Things Can Make a Big Difference, New York, Little Brown.

Margalit Gur-Arie blogs frequently at her website, On Healthcare Technology. She was COO at GenesysMD (Purkinje), an HIT company focusing on web based EHR/PMS and billing services for physicians. Prior to GenesysMD, Margalit was Director of Product Management at Essence/Purkinje and HIT Consultant for SSM Healthcare, a large non-profit hospital organization.

Categories: Uncategorized

Tagged as: , ,

5 replies »

  1. Re ROI, good point Margalit. More than one party is now doing the “I.” HHS of course is investing via ARRA and other grants. But the providers themselves, in most cases, have a bigger piece of the “I” in labor + money.
    Re Thought Leaders, what’s your gauge for when someone qualifies as one? Readership? Polls (what % would agree with them)? Are there good examples of physician thought leaders you’d cite, even if their leadership is not in EHRs? There are plenty of people (in politics, etc.) who have strong thought leadership within a particular segment, which may be large, but who may be ignored or even scorned by other segments. I’d posit that those whom HHS is seriously listening for policy and standards to are regarded by them as thought leaders, even though they may not be thought leaders in other communities. I think your point as that it’s not the same people that are accepted in both the physician and the policy maker societies.

  2. David, I do agree with your qualification for “simplicity”. However, I have a question regarding ROI. If ROI is to play a role, the R needs to accrue to the same entity that made the initial I. I’m not certain it does. What is the R measured in? Dollars, quality, or both? This will affect the power of ROI to drive adoption.
    As to practicing docs who are developing EHRs, I know several of those as well, but none of them are considered thought leaders, and most are relatively unknown. Homophilous societies are known to treat those who advocate change, and departure from accepted norms, with great suspicion and physicians are doing just that.

  3. Dr. Levin, I couldn’t agree more. In the short term, I believe the Government intervention is/will speed up adoption. This may backfire in the future though, if implementations are undertaken for all the wrong reasons and if they start cutting corners to make the “deadlines”.

  4. Hi Margalit,
    Interesting post. I agree that we’re not “there yet” but are getting closer.
    I’d qualify the premise that “simplicity is always a virtue” though. I’d say “simplicity to the appropriate degree for the task at hand.” For example, even a “simple” airplane is far more complex than a sophisticated bike, and both are means of transportation. But I don’t want to fly on an airplane that has bike technology, and I don’t want to spend millions of dollars for a bike. So it depends on how ambitious your requirements and your budget are. The differing viewpoints about the merits vs costs of “computable semantic interoperability using standardized structured data” vs “Let’s just move human readable info from point A to point B” are complex. It isn’t just a matter of which one is simpler (clear winner there) or which one has more functionality (a different clear winner there). Rather, where’s the biggest bang for the buck (ROI)? ROI is hard to project, even harder to guarantee. That’s why the trend seems to be to adopt what is known to work, rather than what we “predict” ought to work.
    I’d like to add that, while MD after someone’s name doesn’t guarantee that he/she is a practicing physician, there ARE (and I know several) dedicated practicing physicians who are also EHR developers or product managers. They have to “eat their own cooking” (or someone else’s EHR cooking as it were) and learn from it. I respect them for it.
    Thanks for your thought-provoking posts.

  5. Roger’s formula is heavily skewed in our present setting. The overpowering effects of governmental subsidy (or else) makes his formula and predictablity uncertain in our setting. There is also a serious possibility of a modification of the implementation incentive/penalization equation.