Share this article on:

Use of the Electronic Health Record for Coding in Outpatient Neurology

Weathers, Allison, L.

doi: 10.1212/CON.0000000000000443
Practice Issues

ABSTRACT Diagnostic coding now factors into reimbursement, quality assessments, reputational metrics, and epidemiologic analysis; therefore, it is more critical than ever that neurologists are accurate and precise in their coding. In addition to being the means though which most neurologists are capturing this information, the electronic health record can offer multiple tools to assist in these efforts. With conscientious design, build, and implementation of the electronic health record, diagnostic coding can be effortless, even for the most complex and varied of conditions seen in an outpatient neurology visit.

Address correspondence to Dr Allison L. Weathers, Cleveland Clinic, 25900 Science Park Dr, AC220, Beachwood, OH 44122,

Relationship Disclosure: Dr Weathers serves on the editorial board of Continuum and as chair of the adult neurosciences specialty steering board for Epic.

Unlabeled Use of Products/Investigational Use Disclosure: Dr Weathers reports no disclosure.

Back to Top | Article Outline


The broad range of conditions and symptoms covered in this issue of Continuum highlights the challenges of accurately capturing all the diagnoses that are potentially encountered in an outpatient neurology visit. In a recent survey conducted by the American Academy of Neurology (AAN), 93% of respondents answered “yes” when asked about electronic health record (EHR) use in their practice, signifying that most neurologists are now both documenting and coding outpatient visits in an EHR.1 A frequent concern from providers regarding EHRs is that these systems were not designed for patient care, but rather to support billing and coding, and that they end up being ill-suited for either task. In fact, the ability of EHRs to improve quality and safety of care has been repeatedly shown,2–4 and when optimally designed, built, and used, EHRs are actually powerful coding tools as well.5 This is even more apparent in the current era of the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM), which significantly increased the volume and complexity of diagnostic codes.6

Back to Top | Article Outline


Much emphasis has been placed on the importance of accuracy in inpatient coding and documentation because of the direct impact on measures such as a hospital’s case mix index (CMI), severity of illness, present-on-admission reporting, and expected risk of mortality. Upon discharge, a patient is assigned to a diagnosis-related group (DRG) based on the severity and complexity of his or her illness, as well as the presence of comorbid and major comorbid conditions. The accurate assignment of the DRG is therefore reliant on precise diagnostic coding. DRGs are used to calculate the CMI, which directly factors into a hospital’s reimbursement. Although much emphasis is placed on timely and accurate Evaluation and Management (E/M) coding, correct diagnostic coding is just as crucial in the determination of hospital reimbursement.

In addition to being used to drive reimbursement and to determine achievement of inpatient quality measures, these metrics have a significant bearing on a hospital’s reputation as they are used to calculate reputational scores generated by organizations such as Vizient (previously known as the University Health System Consortium) and publications such as US News & World Report. Poor documentation and coding can result in underrepresentation of disease severity and risk of mortality and, as a result, falsely lower these scores, with potential negative impact to patient volumes and physician recruitment efforts.

Although outpatient diagnostic coding has not historically factored directly into these measures and ranking calculations, a considerable downstream effect can occur, especially in health systems in which a single shared or fully interfaced EHR between the outpatient clinics and the hospital exists. In EHRs, visit diagnoses are often pulled directly from the problem list, and the reverse is also true in that visit diagnoses either automatically populate the problem list or can be added by the provider with a single click. In an integrated EHR, the outpatient problem list then becomes the starting point of the inpatient problem list, from which the principal hospital problem can be selected. Integrated problem lists should have the ability to mark problems as “hospital problems,” labeling them as pertinent to the hospitalization. The identification of the hospital diagnoses, including the principal diagnosis, has billing implications, as coders will identify these as the main diagnoses for billing purposes. To ensure accuracy in diagnostic coding, it is therefore critical that the outpatient problem list be regularly reviewed and meticulously maintained. Incorrect (whether they are fully inaccurate or simply older and less specific) entries should be deleted, and resolved problems should be designated as such. These resolved problems can usually be filtered out from regular view, yet easily found if necessary. Alternatively, “personal history of” codes can be used. These codes denote that the patient has experienced the illness, but that the problem was in the past. For example, there is a code for personal history of transient ischemic attack and cerebral infarction without residual deficits.

The entire problem list is frequently incorporated into the clinical documentation, including the history and physical, daily progress notes, and the discharge summary. Coders use this documentation as supporting evidence for the principal diagnosis and to identify comorbid conditions, both of which are then used to calculate the metrics discussed previously. Failure to precisely capture a diagnosis code in the outpatient setting can therefore have a profoundly negative impact on the accuracy of inpatient coding and billing should the patient ever get admitted.

The Medicare Access and CHIP Reauthorization Act of 2015 (MACRA) reformed Medicare payments by creating the Merit-Based Incentive Payment System (MIPS), which combined aspects of several quality programs (Physician Quality Reporting System, Value Modifier, and the Medicare Electronic Health Record incentive program) into one single quality program.7 This transition away from fee-for-service reimbursement toward value-based payments that are dependent on compliance with outpatient quality measures has made precision in outpatient coding critical in its own right. EHRs certified under this program will have the ability to capture and report electronic clinical quality measures (eCQMs); for many of the measures, this functionality is dependent on accurate diagnostic coding. Failure to code properly now could result in significant negative adjustments to Medicare reimbursements as early as 2019. At least one payor has already started incentivizing clinicians financially to code more precisely, with the justification that insufficient coding not only impacts a physician’s ability to understand the complexity of their patient population and therefore provide appropriate follow-up care, but that it also makes it more difficult for the insurer to track these patients.8 More than $5 million has been pledged by one payor toward this effort.

Back to Top | Article Outline


In addition to supporting the submission of diagnostic coding data to insurers and government agencies, EHRs can facilitate the capture of this information by the clinician through a variety of features. Although specific functionality and content may vary among EHR vendors, generally, the tools that support coding can be found in most systems. These tools vary greatly in how much support they provide, how intrusive they are to the clinician’s workflow, and how difficult they are to implement within the EHR.

Of these tools, the most basic end of the spectrum is simply the ability to quickly and effortlessly access a complete database of the ICD-10-CM codes, whether this is built into the software or integrated through a third-party content provider. For maximum utility, the database should be searchable through a variety of methods, including alphabetically and numerically with synonym recognition as well. It must be easily accessible wherever needed (such as from the problem list or visit diagnosis fields). Most EHRs can create both departmental and personal preference lists of these codes. However, this functionality must be used with caution to avoid the creation of an incomplete list that inadvertently leads to inaccurate coding. Preconfigured lists of codes can also be grouped with commonly associated orders, documentation templates, and E/M codes for the purpose of expediting the charting of an outpatient visit. This functionality is used generally for very discrete patient populations, such as a return visit in a patient with an established diagnosis of seizures; caution must be taken with this functionality as well.

A more sophisticated coding tool is the integrated diagnosis calculator. Diagnosis calculators will prompt the clinician for additional information, when indicated, to drive to the most complex and therefore accurate code possible. For example, when a provider enters a diagnosis of trigeminal neuralgia, the calculator may automatically display (without additional clicks being needed) a request for additional information such as laterality and duration. The options presented can be configured to be required, preventing the clinician from adding a more generic diagnosis to the problem list or using it as the visit diagnosis. However, if selections are required, then generic options such as “other,” “unknown,” or “not otherwise specified” must be included, as often a more specific diagnosis is unable to be made, especially at the initial outpatient neurology visit. This is particularly true for some of the topics covered in this issue of Continuum, such as episodic vertigo and neuropathic pain.

Precise diagnostic coding may be supported by alerts that trigger based on other data entered into the record and that prompt the clinician to enter a specific code. These alerts can be passive, in which they simply display and don’t require any further action on the part of the end user, or they can be designed to be more intrusive, in which they will interrupt the workflow and an action is required to dismiss them. These alerts can be potentially triggered by laboratory data, such as a low vitamin B12 level precipitating an alert suggesting that a diagnosis of vitamin B12 deficiency is added. The entering of specific medications on the patient’s current medication list or placement of a medication order could also trigger an alert. For example, an order for meclizine could trigger an alert for a diagnosis of vertigo. However, as a number of drugs are used in outpatient neurology for multiple diagnoses, caution should be exercised in the creation of these alerts to avoid the suggestion of inaccurate or even blatantly incorrect diagnoses. Awareness is also growing around the issue of alert fatigue, in which clinicians must deal with so many alerts that critical ones are inadvertently ignored, resulting in potentially significant patient safety issues. Alerts can be a powerful tool to support coding, but should be built and implemented judiciously.

While not traditionally viewed as a coding tool, documentation templates can support the capture of those additional details that are required in ICD-10-CM, yet are often easily missed by providers. This includes the laterality and specific anatomic location if applicable, the acuity and phase of the patient’s diagnosis, and how the patient’s symptoms relate to the diagnosis. Traditional EHR note templates are not directly linked to the problem list or the visit diagnosis fields, in that a problem entered in the note will not automatically appear in the correct field. However, if the template is followed, the clinician will at least have acquired the additional details necessary to code accurately and with an appropriate level of complexity. The potential downside of the use of documentation templates is that if overly templated, the patient’s true story can be lost, and the documentation may be meaningless.

Many EHR companies and several third-party vendors are developing natural language processing capabilities that are either integrated into or overlay an EHR’s existing documentation module. These programs can look for key words and phrases in blocks of text and fire specific alerts that advise the clinician to add more details to their documentation or capture a specific diagnosis on the problem list. For example, the documentation of headache, photophobia, and phonophobia could prompt the clinician to add a diagnosis of migraine to the patient’s problem list. This mining ability of free text allows clinicians to accurately capture their patients’ stories and examination findings, thereby achieving the benefits of highly templated “checkbox” capture of information without the inherent downsides. Although the use of natural language processing tools is not yet widespread, it is likely that the move toward value-based payments will drive more rapid implementation, and in addition to supporting more accurate diagnostic coding, this functionality can also be used to support quality measures such as tobacco cessation counseling.

Back to Top | Article Outline


The role of diagnostic coding has expanded rapidly. It factors into reimbursement, quality assessments, reputational metrics, and epidemiologic analysis; therefore, it is more critical than ever that neurologists are accurate and precise in their coding. In addition to being the means though which most neurologists are capturing this information, the EHR can offer multiple tools to assist in these efforts. With conscientious design, build, and implementation of the EHR, diagnostic coding can be effortless, even for the most complex and varied of conditions seen in an outpatient neurology visit.

Back to Top | Article Outline


1. American Academy of Neurology. Neurology compensation and productivity report 2015. Accessed January 27, 2017.
2. Kern LM, Barrón Y, Dhopeshwarkar RV, et al. Electronic health records and ambulatory quality of care. J Gen Intern Med 2013;28(4):496–503. doi:10.1007/s11606-012-2237-8.
3. McGinn TG, McCullagh L, Kannry J, et al. Efficacy of an evidence-based clinical decision support in primary care practices. JAMA Intern Med 2013;173(17):1584–1591. doi:10.1001/jamainternmed.2-13.8980.
4. Jones SS, Rudin RS, Perry T, Shekelle PG. Health information technology: an updated systematic review with a focus on meaningful use. Ann Intern Med 2014;160(1):48–54. doi:10.7326/M13-1531.
5. Samaan ZM, Klein MD, Mansour ME, DeWitt TG. The impact of the electronic health record on an academy pediatric primary care center. J Ambul Care Manage 2009;32(3):180–187. doi.10.1097/JAC.0b013e3181ac9667.
6. Centers for Disease Control and Prevention. International classification of diseases, tenth revision, clinical modification (ICD-10-CM). Updated August 22, 2016. Accessed January 27, 2017.
7. Centers for Medicare & Medicaid Services. MACRA: delivery system reform, Medicare payment reform. Accessed January 27, 2017.
8. Goedert J. Health data management. Highmark to pay more for better diagnostic documentation. Published May 5, 2016. Accessed January 27, 2017.
© 2017 American Academy of Neurology