EHRs and Audits: The Role of Voice Recognition in Avoiding “Innocent Fraud”

A brief article by Michael Stearns, MD

The FBI, Justice Department, Office of the Inspector General of CMS, multiple congressmen and others have expressed concern over the tendency of electronic health records (EHRs) to increase billing rates for patient encounters. EHRs are excellent business tools if used correctly, as they will match the medically necessary information documented in the clinic visit to the appropriate level of evaluation and management (E&M) service.  Historically, providers tend to undercharge when they document on paper for a variety of reasons; not the least of which are the complex rules used to determine what CPT™ E&M code to choose for the encounter.  EHRs, when properly used, help providers navigate through these rules and lead to more accurate coding.  However, this “correction” has led to higher levels of coding, albeit supported by appropriate levels of documentation.

One of the challenges of EHR usage from a coding and auditing perspective are the presence of “shortcuts” that allow for an unlimited amount text to be entered into the clinical record from prior records, macros or templates.  This has the potential of leading to documentation that does not accurately represent information obtained from the patient during the encounter, or information that was placed in the note that is not relevant for a particular visit.  If information that was not actually obtained by the patient or information that was not medically necessary to obtain are used by the EHR’s E&M coding software to justify the amount charged for the encounter, it is considered to be fraudulent activity committed by the provider.

There are common patterns that emerge from inappropriate EHR usage.  The first is the tendency to pull information from a prior visit into a current visit, a process that has been referred to a “cloning.”  This has a great deal of value, as it compels the provider to review findings from the prior visit and improves the efficiency of documentation.  However, if it is not used carefully, it tends to insert irrelevant information from a prior encounter that is either nonsensical or clearly out of context for the visit at hand.  An example might be a full physical examination that is pulled forward from the patient’s first visit to the clinic.  The full physical is likely not justified for a follow-up visit, and if the level of documentation results in a higher E&M level of service it is considered to be in excess of the medical necessity consideration.  To avoid this, providers must make sure that they change, in particular, the components of the encounter note that are most likely to vary between visits.  These include the history of present illness (HPI), social history, family history, physical examination, assessment and plan.  A logical approach would be to pull notes forward from a prior visit, and add text to make the note truly unique for that visit.  The used of voice recognition software in this context is a popular approach, and it avoids the “cloning” issue and protects the provider from the verbiage of templates that can be highly repetitive.

The use of voice recognition in other sections of the note, in particular the assessment and plan, also serves to make the note truly unique and a far better protection from a negative audit than notes generated from cloning older notes and/or using templates.   Auditors will focus on the assessment to determine the level of complexity of medical decision making, which many feel should be the principle component used to determine the level of service for the encounter.   Notes that contain little information other than bulleted lists do not convey the complexity of the provider’s thought processes.  The use of voice recognition, which now has a greater than 99% accuracy rate, is recommended for many encounters as is far better allows the providers to document “what I was thinking” when the patient returns for the next visit.  Voice recognition tools also support the use of macros that allow for simple voice commands to create paragraphs of text.  An example might be a macro called “Marco: lumbar puncture discussion with patient.”  This command could generate several paragraphs of text created by the provider and clearly illustrate the risk associated with this procedure to a potential auditor.  With relative ease the macro text can be edited to make it specific to the patient encounter.

In summary, the use of voice recognition tools, in combination with pulling forward older notes and the use of templates, is the most efficient method of documenting exactly what happened during a patient visit.  It also allows providers to attain fair and accurate reimbursement based on documentation that cannot be challenged by auditors.

Leave a Reply

Your email address will not be published. Required fields are marked *