Does EHR Use Lead to Lower or Higher Costs? Part 5




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


Incorrect Assumptions Lead to Incorrect Conclusions That Misinform Policy

Please refer to Part 1 of this post for our comments on the recent Wall Street Journal op-ed questioning the policy behind the EHR incentive program; Part 2 for our comments on the extent to which EHR use was associated with higher short term charges, as published in the Center for Public Integrity report and recently summarized in the New York Times; Part 3 for our comments on where one would expect EHR use to show short-term cost increases; and Part 4 for our comments on repetitive or boilerplate verbiage.

Some have suggested (and some rather vehemently) that whatever the cause of billing fraud, it is best remedied by regulating and/or removing all coding software from the hands of all physicians.  The rationale behind this demand is that whether or not physicians intend to misuse existing coding decision support tools, there is simply too much temptation to expect them to behave ethically and resist the urge to pad notes and inflate billing codes by documenting findings that were never elicited or actions that were not actually taken. Little if any mention is made of the possibility that the physician is simply using the coding software to ensure she uses a code that is supported by current documentation or to provide a reminder to the provider before signing the note that findings or actions that were done and medically necessary (e.g., focused physical examination findings or elements of the Review of Systems) have not yet been documented, whether due to interruptions, distractions or competing priorities during the visit.  

While it may seem reasonable to some to remove or severely restrict the functional capacity of coding decision support software, we contend that doing so is not warranted simply because it is possible to use the decision support for fraudulent purposes.  As physicians who have been in practice continuously since the release of the 1995 and 1997 Evaluation and Management (E/M) billing guidelines, the casual observer might expect that we would know them cold, and thus have no need for guidance.  For those who have never actually seen these guidelines, while straightforward in principle, in practice they are exceedingly complex, and for paper notes it typically requires a professional coder to manually review a note and determine the appropriate billing level with a high level of reliability.  This is reason enough to suspect that there is a huge problem with the guidelines.  Doctors are trained and trusted to make life and death decisions on a daily basis, but not to understand the level of service they provide to a patient.

The truth is that doctors do have a feel for the level of service they provide, based on how many problems are addressed during a visit, the amount of history gathering necessary to make appropriate decisions, the amount and depth of physical examination necessary, the amount and type of information that needs to be reviewed, the number of tests that need to be performed or ordered, and the amount and complexity of medical decision-making required. Under certain circumstances that are uncommon for primary care physicians, all of these elements may be ignored, with the billing level determined strictly by time, assuming that the documentation of time spent and topics covered in patient counseling or care coordination is sufficiently documented.  However, when time-based billing criteria are not met, accurate billing code determination requires counting and categorizing specific elements of history, physical examination and medical decision-making using sufficiently complex specifications and combinations that it is very difficult for many physicians to manually determine the appropriate E/M code from a note rapidly and accurately.  This is true whether the note is generated on paper or created in an EHR. 

Prior to EHRs and even now, many doctors must turn to laminated cards,  pocket guides or desktop documents to remind them of the documentation elements required to satisfy a particular E/M level of service.  One of us is required as part of his large group practice to review a number of office visit EHR notes of practice colleagues every six months and audit them for E/M coding accuracy.  The process of auditing the notes requires counting and organizing the elements to determine if the E/M service is justified. It is a dreaded task that if done carefully is painstakingly slow, requiring one’s undivided attention and concurrent use of a coding guide to minimize the chances of manual error. Accurate coding is not an activity that lends itself to multi-tasking and interruption, but it is precisely what most doctors must contend with regularly in their offices or in the hospital.  

That fact alone is reason to conclude that something is terribly wrong with the E/M system, and thus it bears restating.  The essence of medical practice is providing thoughtful advice and care to the very best of our abilities; and more often than not in severely time-limited settings which are further infused with seemingly continuous distractions and interruptions.  Yet having no choice if we want to get paid for our work, we tolerate a coding system that further adds to our thinking load and distracts our time and attention from patients during and after encounters.   

There are several, mostly reactive responses physicians take to manage the stress of this dysfunctional coding system:

  1. Always use “safe” codes (ones that fit the billing mean for your specialty, or even one code below), and thus remove E/M coding rules from the list of distractions during care.  Consciously or not, this is what some doctors routinely do, underpaying them for their work and exposing them to accusations of technical billing fraud. Overdoing and over-documenting (to stay within the boundaries of “safe” codes) wastes provider and patient time that could be used for more important purposes, while under-billing deprives the provider of resources that could be used to improve patient care services.
  2. Mostly use “safe” codes but keep an E/M coding guide document or laminated card available and use it occasionally to submit a higher billing code when a quick review of the guidelines gives the provider sufficient confidence to do so. Although we do not have systematic data, we contend that these two coding approaches remain extremely common in care settings lacking coding decision support.  The OIG report focused on the 0.38% of doctors who almost always use the two highest billing codes.  They did not report on what we believe might be the typical practice for many if not most individual physicians – they bill the great majority of their services at the same billing code (for example, >70% of E/M codes are level 3).
  3. Retire the E/M billing guidelines.  We believe they have not served their intended purpose of making coding fairer.  They distract attention away from direct patient care and thus decrease the effectiveness, efficiency and perhaps even safety of care.  They divert money away from actual care by necessitating expansion of the coding industry.  Sounds crazy?  That actually was the nearly unanimous conclusion in 2002 of the HHS Advisory Committee on Regulatory Reform.  Section 941 of the Medicare Modernization Act of 2003 required the Secretary of HHS to conduct a series of pilots of alternative payment systems with the idea that one or more would emerge to replace E/M coding.  However, these pilots were never conducted.

As long as we retain the current payment system, there unfortunately remains the need for a fourth option – use coding decision support to better ensure compliance with E/M coding requirements without having coding be a major distraction during patient care.  This approach, as distasteful as we find it to be, reflects the view that coding accurately is as important as safe prescribing.  Safe prescribing (which needs to be done every day in settings of continuous distractions) requires some level of decision support, which is why the EHR Meaningful Use incentive program requires activating drug-drug and drug-allergy alerts. “Safe” coding that is highly accurate, efficient and mitigates risk of inadvertent over- or under-coding for E/M services requires computerized coding decision support and can be integrated into EHR-supported clinical workflows), yet some have called for removing coding decision support from EHRs.

Current Medicare regulations allow such coding software to exist, but appropriately restrict its function to ‘advice only’, rather than to independently determine and automatically record the level of service and associated E/M code.  This ‘advice only’ restriction is important because EHR systems can at best determine the required elements of documentation, but not the elements of appropriate documentation.  One of us (MZ) routinely uses his EHR’s E/M coding advisor software, but only after first completing the documentation felt to be necessary and appropriate.  The coding decision support is then used to confirm that the billing level is never higher than that supported by the advisor and his judgment of medical necessity (which no advisor can assist with today). Occasionally a lower code will be selected than supported by the E/M advisor because medical necessity was felt to be lower than what the documentation itself supported.

Whether used regularly or occasionally, we have both found the coding advice helpful; the majority of the time the advisor agrees with the E/M code we intend to use.  Occasionally it indicates that the documentation unexpectedly supports a much lower than expected code based on the work we did.  In those circumstances, we have found the E/M advisor rapidly points out required by missing documentation fields (such as a structured chief complaint) – something that was obviously obtained, but not documented in the right field.  We can then go back and complete the missing documentation regarding the care that was provided and medically necessary.  We believe this is how most physicians use coding advisors (whether embedded in EHRs or otherwise).  That said, for the small fraction of providers who intend to commit fraud, coding decision support can make it easier for them to do so.

Just like the OIG did not investigate whether any of the 0.38% of providers using the highest two billing codes at least 95% of the time committed billing fraud, it is also unclear that no billing fraud exists among the remaining 99.62% of providers.  However, we believe that the vast majority of providers are not attempting to bill Medicare for higher levels of E/M services than were actually delivered to patients, and that they attempt only to bill for services that are reasonable and necessary.  At the same time, we also do believe that done manually, E/M coding is error-prone and that EHR coding decision support assists in the selection of a level of service that is more consistently an accurate reflection of the appropriate services delivered and documented.  This addresses the pervasive problem we saw in our own practices prior to EHR adoption in which providers systematically under-coded their paper chart-based encounters because of: 1) the time inefficiency of documenting all of the relevant findings; 2) the challenges of locating and documenting the review of all relevant results and reports; 3) difficulty in accurately coding visits according to complex billing regulations; and 4) fear of prosecution and reputational harm from accidentally over-coding visits. 

With accurate real-time coding software in their EHRs, such providers find themselves no longer caught in the regulatory ‘speed trap’ of complex coding rules. They now have the equivalent of a coding ‘speedometer’ that allows them to confidently bill at the level where they should have been billing for such visits for years but had been afraid to.  The same decision support can also alert them when they are coding ‘over the speed limit’ and encourage them – not to go back and fraudulently document  – but to change to a lower E/M code if that is all that can be supported by the work done and the medical necessity of doing it.  We do not want physicians to believe that using the EHR to better document what they are appropriately doing, and then accurately coding for their services will expose them to a new ‘speed trap’ in which such behaviors are considered prima facie evidence of billing fraud, simply because their average coding level is higher since they started using an EHR.

Please continue reading the conclusion of this post, Part 6 , for our comments on the E&M documentation guidelines and “note bloat” – a significant waste of time for note author and reader, as well as a barrier to more efficient use of EHRs; and our conclusions and recommendations for a constructive path forwards.



About Peter Basch, MD

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

Washington, DC


  1. Abi says:

    I want to try on a few ideas on you concerning the EMR.In both Law and Banking prosofsiens the computer is used extensively to generate the information required to document the transaction between the institution and the client. They use this computerized data for quick reference when the client presents for a service. However, no one has ever proposed that they become totally paperless in the records of these transaction. There is always a hard copy !!Now what makes Medicine unique that it should become paperless ?It is certainly wonderful to have a digital copy on the computer for quick reference by Physicians, office personnel, Insurance filers, Office manager, Auditors, Lawyers, etc.However, if you consider the computer as only a “tool” in creating the patient’s medical record you immediately have a quick affordable solution to the EMR.By using the computer to generate the “True or Permanent Record” which is Paper, you gain all the advantages of the EMR without any of deficits. It is legible, not subject to electronic alterations or deletions, Requires no expensive Electronic Input Gadgets, No learning curve on how to create an EMR, Allows the physician to bring to the record increased information (i.e. Differential Diagnosis, Patient information, Drug Interaction).If you find one morning you have a computer glitch right in the middle of a busyClinic, no problem, the paper record is still working fine and the computer generated hard copies can be accomplished at later date.The Hardware for the computer is already available at the level required by the various types of practices. Some solo practices could generate an EMR with only asimple desk top computer. Larger practices should be able to find computer power they require with minimal effort and cost. The soft ware can be purchase off the shelf. After all what is unique about typing someone’s Name, Address, Age, SS# etcI believe our documentation by Diagnosis is quick, complete, accurate and ready .i.e. Each diagnosis is formatted with ICD code, Etiology Check boxes,Symptoms and Signs-check boxes, Lab StudiesImagining, Management ProtocalsIn “Going Paperless”We might be “ Throwing the Baby (i.e. Paper Record) out with the bath water.”I am almost certain that we will see many unintended consequences if we follow this path.The most elegant solutions are the simplestOne of the fundimental tenets of process improvement is to work out a process using low tech solutions before you invest in high tech solutions which can make the process harder to improve and can have a negative impact on productivity.Have a large capital expense?- None.Have a large monthly upkeep -None.

Speak Your Mind