Does EHR Use Lead to Lower or Higher Costs? An Overview




By Peter Basch, MD, FACP and Michael Zaroukian, MD, PhD, FACP, FHIMSS


Incorrect Assumptions Lead to Incorrect Conclusions That Misinform Policy


While Democrats and Republicans may each have their own take on how best to move our country’s physicians and hospitals from paper to electronic records, there has generally been broad consensus that widespread adoption of electronic health records (EHRs) is a necessary, if as yet imperfect, step in the evolution to better, safer and more affordable care.  While there have been challenges to the logic and effects of the HITECH Act and the CMS EHR Meaningful Use Incentive Program that have catalyzed greater EHR use, stories published in in two high impact newspapers last week raised additional serious questions about the conventional wisdom that EHR use will help contain increases in health care costs. 

The first, published as an op-ed piece in the Wall Street Journal, takes the position that there is no evidence from four to five decades of research that EHRs save money, and that the current policy to use financial incentives to accelerate EHR adoption is misdirected.  The second, a report published in the New York Times, does not question the rationale behind the policy; instead, it presents the startling and disturbing finding that whatever policy makers were hoping for, there is evidence that doctors and hospitals with EHRs are using them to bill for more and higher complexity services – and are thus increasing costs.  Worse, the New York Times article suggests that much of the higher billing may be due to widespread electronic ‘science fiction’ in which doctors electronically record findings without actually ascertaining them, while their EHR coding software actively encourages documentation resulting in unethically and even criminally inflated bills.

Our initial reactions to these two pieces were mixed.  On the one hand, we were troubled by the possibility that EHRs could be directly and significantly contributing to billing fraud, a criminal activity that hurts everyone.  We applauded the speed and clarity with which the US Attorney General and Health and Human Services Secretary indicated that such fraud will be vigorously pursued and prosecuted. 

On the other hand, as practicing primary care physicians with longstanding informatics leadership roles and experience in our respective organizations and professional societies, we have presented and published data from our own practices showing cost savings and quality gains from EHR implementation and optimization.  We have seen EHR coding decision support software from multiple vendors designed to facilitate appropriate billing rather than fraud.  We have reviewed enough of the primary literature over many years of research and have deployed enough of it in our own practices to believe that well- designed, implemented and optimized EHR systems used by trained healthcare professionals in a “meaningful” way can support improvements in health care quality and value. 

In light of our experiences, we wondered how a recent study could yield such strong and conclusive evidence that the WSJ editorialists could reasonably conclude that the “savings claimed by government agencies and vendors of health IT are little more than hype.”  We were also curious to understand the  evidence base behind the claims in the NYT report that there may be a widespread problem with EHRs encouraging doctors to document work they did not do to inflate their bills, and then using their EHR coding software to discover and document in a manner that would support fraudulent billing practices.  We also were mindful about other forces unrelated to billing fraud or EHR design that drive increased physician documentation but did not see much discussion of these forces in either the WSJ or NYT stories. 

The product of our reflections about these issues is a six-part blog post that can be found on the Doctors Helping Doctors Transform Healthcare web site.  Below we provide a brief summary of each of the six parts and our conclusions for each, with links to each.  We invite others to add their comments, questions and suggestions for moving the conversation forward to ensure that the increasingly widespread adoption of EHR systems and associated health IT results yields its intended goals of improving the quality, safety and value of health care.    

Part 1: Critique of the Wall Street Journal op-ed piece

In Part 1, we examine the evidence behind the claim that the “savings promised by the government and vendors of information technology are little more than hype.”  We identify three major problems with the WSJ story: 1) the op-ed authors drew inaccurate and misleading conclusions from the systematic review; 2) they used conflated logic to represent several decades of research as support for their broad conclusions when only a tiny fraction (<0.1%) were relevant to the much more limited question addressed in the systematic review, all of which were completed prior to the start of the EHR Meaningful Use Incentive Program; and 3) even though the editorialists arrived at their conclusions for the wrong reason, most policy makers have recognized that health IT is at only an enabling infrastructure to the delivery of higher quality, higher value care, which is why the HITECH Act is intentionally NOT an incentive program for the mere implementation of EHRs; it is a program that incents EHR adoption of particular capabilities that are used in meaningful ways that were deemed through the process of literature review and expert opinion to enable better, safer and more efficient care. 

Part 2: Critique of the New York Times story

In Part 2, we look at the recent Center for Public Integrity and Office of the Inspector General reports referenced in the NYT piece in which it was found that physicians and hospitals using EHRs show higher charges than their colleagues using paper records.  Notwithstanding the legitimate concerns, anecdotal examples and significant questions regarding the prevalence and contributing factors to physician billing fraud – including features and functionalities in EHR systems – we did not find evidence from any of these reports that answer the specific question of whether use of EHR systems or their coding decision support software contributed to the billing patterns of the 1,669 (0.38%) of 442,000 physicians who had Medicare billing profiles that differed significantly from their colleagues.  The OIG study also did not look at whether this small fraction of physicians billing at the highest levels were doing so inappropriately or fraudulently.  We conclude that while all causes of billing fraud and abuse in any type of documentation system (paper or EHR) should be explored and reasonable steps taken to prevent and remedy them, we do not accept the premise that EHRs or their coding decision support systems promote fraud and abuse.

Part 3:  Exploring the impact of EHR systems on costs and physician billing

Part 3 starts with the assumption that there actually is an association between EHR use and higher billing levels, whether or not informed by E/M coding decision support software.  We explore the legitimate and even expected reasons why billing levels might increase in the short-term but also are associated with longer-term savings.  We review examples of how EHRs can curb costs, why some of the gains would not be seen in the short-term, and that increases in certain costs (better access, more preventive services, greater attention to chronic disease management) would be an early indicator that EHRs were being used as more than mere documentation tools, and instead are being used to inform and support improved quality and safety. We use the example of one organization’s launch of its Million Hearts™ program which will result in higher costs in the short-term (more visits, needed testing, and some higher complexity visits) but should yield long-term savings by preventing new heart attacks and strokes that would otherwise occur.

Part 4:  Is repetitive or “boiler plate” documentation suggestive of fraud?

In Part 4, we challenge the inference in the Center for Public Integrity report and NYT story that the finding of repetitive language (“boilerplate” documentation) in medical notes or similar appearing notes is prima facie evidence of documentation fraud and illegal billing behavior.  We talk about the degree to which standard terminology has been encouraged in medical school and residency training since long before EHRs were used and that it is fortunate for patients that physicians are trained to use standard terms to describe and clearly communicate positive and negative findings to each other.  We acknowledge that physicians can document in notes work that they did not do but that is neither unique to EHRs nor will be cured by further limiting the ability to use efficient documentation tools wisely and well.  While we believe there is some role for looking at physician documentation patterns to screen for potential fraud, we write of our concerns that software that looks for documentation patterns needs to be designed carefully because much of the repetitive documentation in notes is appropriate rather than fraudulent.

Part 5:  Should EHR coding decision support software be banned or further regulated?

In Part 5, we respond to the suggestion by some who were interviewed for the WSJ and NYT reports that billing fraud is best remedied by regulating and/or removing E/M coding software from EHRs and other health IT systems so they will not tempt physicians to over-document and subsequently overcharge for their visits. We provide another experience-based perspective in which the physician uses the E/M coding software to ensure use of a code that supported the current reasonable and necessary care and to ensure that the provider has not forgotten to include this documentation before signing the note.

We talk about the problems with the complex 1995 and 1997 E/M billing guidelines and why decision support is helpful to ensure compliance and designate an appropriate billing code with a high level of reliability.  We share our view that as long as we retain the current payment system, E/M coding decision support will help responsible physicians ensure their compliance with E/M coding requirements without having coding be a major distraction during patient care. 

Part 6:  E/M documentation guidelines, “note bloat,” and a constructive path forward

In Part 6, we share our perspective that EHRs did NOT cause the documentation clutter and verbosity that is now commonly called “note bloat” but rather that the E/M documentation guidelines are the principal contributor to this problem, which is a not only a significant waste of time for the author and readers of the note, but also a barrier to more efficient use of EHRs.  We argue that while EHRs make it easier to document profusely and are therefore implicated in the “note bloat” problem, EHRs are simply using their capability to assist with the documentation required for billing.  We contend that documentation requirements and coding complexity contained in the 1995 and 1997 E/M Guidelines catalyzed the very conditions that the Center for Public Integrity and the New York Times reports have recently raised concerns about.  We close with a summary of our observations and suggestions for moving forward.

We encourage those of you who are interested in exploring additional details or sharing your views to go to the Doctors Helping Doctors Transform Healthcare web site and let us know what you think.

About Peter Basch, MD

Medical Director, Ambulatory EHR and Health Information Technology Policy
MedStar Health

Washington, DC


  1. Sherry Reynolds @Cascadia says:

    It is critical to ask the right question when we frame a discussion. Cost and quality are important considerations but so is “value” to the patient. For some reason we often forget that it is not only one of the six pillars of IOM but also a key component to the triple AIM as well.

    Health IT by itself is simply the tool you use to acquire data and share it but without a set of core values about your objectives – lower cost, higher quality, better outcomes and patient centered care you will often simply get what you design the system to provide.

    When however you combine health IT with workflow redesign (that takes into account both health care providers AND patients needs) and change funding mechanisms then you start to see the cost savings, improvements in quality, and have the tools necessary to meet the patients needs anywhere, anytime in a more cost effective manner..

    Framing the question as implement A to get B is a mechanistic approach to a systems level challenge

    • Epi says:

      Michael, one of the challenges I see with pacrtices that implement an EHR is trying to achieve excellence too quickly. By this I mean that implementing and using an EHR is like learning to walk. It is hard going initially and takes time to become proficient at specific tasks.I enjoy watching glass-blowers turn out intricate designs consistently and fast. Similarly, watching a chef prepare ingredients for a meal they are about to cook. What amazes me is their speed and consistency in performing these actions.For each of these experts, there were thousands of hours of practice before they were able to perform’ the way they do.The same lessons apply to EHR getting good requires lots of practice. Achieving reductions in diagnostic errors also requires that the EHR operator is able to perform basic EHR tasks such as prescribing without procrastination. Once the basic tasks are second nature, it is much easier to focus on the clinical tasks that lead to improved quality.It is also useful to develop processes that can easily be re-duplicated and applied to different disease states or tasks. For example, the format of templates should have a certain consistency so that anticipating where icons and fields are going to be is completely natural and consistent.

  2. Anwar Hussain says:

    I believe that your arguments about the WSJ article are wrong. Simply complying with process measures (which is what “meaningful use” means) has not proven to lead to reduction in costs or improved outcomes. Just increasing compliance with process measures does not mean we have better quality, until we have improved outcomes. I can easily refer readers to a recent NEJM article showing how widespread compliance in hospitals has not led to improved outcomes.

    In fact, using EHRs, as they are currently designed will simply lead to increased fixed costs of delivering care. I do not see how your arguments show that adoption will increase the cost-effectiveness of care. Put another way, there is no evidence anywhere that a health system/hospital has seen reduced costs and improved outcomes in their patients once any EHR was implemented. If it has not been proven at an institutional level, how does making adoption more widespread going to do that?

    • Peter Basch, MD says:

      Thank you for your comments Dr. Hussain. I believe you have misinterpreted our critique of the WSJ piece. Our points… (1) the review article that was the basis for the WSJ op-ed did NOT conclude that tens of thousands of studies of EHRs over 5 decades failed to show economic benefit. It stated that of these thousands of studies, the vast majority could not be analyzed, and the few that could be analyzed showed mixed results. Lack of evidence is not the same as lack of benefit. (2) Even if the McMaster’s article did state what the editorial claimed (and it did not) – that is not, in our opinion, relevant. EHRs are infrastructure, and it is inappropriate to look at how infrastructure is used without controlling for the business case in which it is used. Thus, even with advanced EHRs, one should not expect to see them being associated with overall lower spending as long as the payment system rewards volume and redundancy. However, with the same EHRs used in a setting of payment aligned with quality outcomes (not strictly volume), we would expect cost savings to be demonstrated. (3) We believe it is also illogical to look even in those settings SOLELY at short term savings. In fact, everyone would agree that particularly in the arena of primary care, where cancer screenings and chronic care management is not optimally delivered approximately 50% of the time, primary care docs using advanced EHRs would be expected to increase costs in the short term.

      Lastly, we have never argued that compliance with Stage 1 of Meaningful Use will lead to a reduction in costs or improved outcomes. I would hope nobody has ever made that statement. Stage 1 of Meaningful Use is about learning the basics of EHR competence. Stage 2 of Meaningful Use is about achieving a higher level of EHR competency, improving competence with patient engagement, and starting to design new workflows that are informed by clinical decision support and reports. It is not until Stage 3 that we are expected to have clear metrics assessing the maintenance of high quality, or quality improvement.

      To your last statement, if I may take the liberty of rephrasing your statement. Where there are health systems or hospitals that have a sustainable business case for reduced volume of services and achievement of quality outcomes (and there are such examples), it is abundantly clear that they could never achieve their missions with advanced health IT – which includes EHRs.

      • Anwar Hussain says:

        Dr. Basch,
        Thanks for your reply. I’d like to address your points here:
        1. “Lack of evidence is not the same as lack of benefit”. I can see that this would not be an issue if we were trying a pilot project that is not expensive. Health IT systems are expensive (added to our already overburdened system), have negative effects (commonly create new types of errors), and negatively affect providers in many ways. If we are going to spend $20 billion to do something (i.e. meaningful use policy), then it is reasonable to assume there is some tangible evidence of value before diving in to increase adoption. I do not see where you have provided evidence of value (either cost reductions in care or outcomes improvement [not the same as process measure compliance]) of EHR implementations.

        2. “EHRs are infrastructure”. While EHRs may be seen as infrastructure, they are directly tied to patient care and have an impact at the point of care. So for that reason alone we should tread carefully. In addition, there is an implication that we can modify these systems after implemented. That is also not true. Case in point, may systems can’t even connect within one health system even during the “optimization” phase. Systems are not interoperable in may cases, clearly. In addition, how much flexibility does the user have in modifying these systems? Very little. In fact, these systems force users to change their behavior, and not necessarily for the better. There are a large number of workarounds. Ultimately, my point is that EHRs are not benign and their inflexibility post-implementation remains a major problem.

        I am not sure how reimbursement models (i.e. reimbursement is volume driven) that focus on volume would address these issues I bring up. You state that reimbursement for quality outcomes would reduce costs. Exactly the point. Nowhere is there proof that outcomes improve with EHRs in the first place (even at advanced IT implementation sites). Process measure compliance may improve, but that is a very narrow view of “quality”. As you know, process measures are not always tied to outcomes. You also suggest improvement in screenings would help, and this would be a long-term benefit. This may be the case, but healthcare delivery is much larger, in terms of cost, on the treatment side; and that is where benefits of EHRs need to be seen if we are to truly bend the cost curve.

        3. You suggest that by Stage 3 we will see cost reductions and quality improvement. I am not sure how ramping up the number of process measures and percentage of compliance with those measures would predictably lead to improved outcomes. Hospitals have attempted this even without EHRs and there have been no improvement in outcomes. There are many examples of this in the literature. Though the later stages of meaningful use are not solidified, the transition from Stage 1 to Stage 2 provides an idea of what to expect in Stage 3.

        What health IT needs to do is to see quality from an outcomes viewpoint, and not process measure viewpoint. Impact on outcomes is required if we are looking to use HIT as a way to build a cost-effective health system overall. HIT should be expected to perform at a much higher level (i.e. improve its ability to be safe, and show true evidence of impact) before we spend billions of dollars to adopt it. I believe we are in the thick of a massive experiment, one that will definitely increase costs of care delivery (from EHR implementation, and upkeep). Whether we see any cost savings or improvement in outcomes seems to be little more than a shot in the dark. I really hope it works.


  1. [...] would also like to call readers’ attention to an outstanding series of six blog postings by Peter Basch, MD, FACP and Michael Zarourkian, MD, PhD, FACP, FHIMSS.  This series of postings [...]

Speak Your Mind