ICD-10 is Risky Business

The risks associated with the changeover to ICD-10 on Oct. 1 have been numerous and varied. Will there be increased allegations of false claims? Did the recent Centers for Medicare & Medicaid Services (CMS) clarification regarding “families” of codes alleviate concerns? Will the change really create a surge in improper payments? Potential legal ramifications aside, will there be increases in issues that concern patients? Will improved clinical documentation result in optimal coding? 

Like almost everything else related to this rapidly approaching transition, the correct answer is, “It depends.” Who or what is doing the coding? How do they or it determine the most accurate code? Can we safely assume that if things are going well now, we are in excellent position for the future? Have you reevaluated your risk areas? Based on very recent experiences, even more new concerns are rearing their ugly heads.

In this day of automation and electronic health records, native coding is typically assisted by digital means. In many cases, these programs have been working very well and assigning codes with excellent accuracy. However, some of us are learning that historical precedent doesn’t always translate to future reliability. A recent case study demonstrates how quickly and completely things can go wrong.

The physicians were doing their job well. In fact, the clinical documentation was excellent. It included laterality, specificity, and in some cases, the exact verbiage in the code descriptions. With very few exceptions, there were no major documentation changes required. The electronic health record (EHR) and code-assist programs were correctly assigning the ICD-9 diagnosis more than 95 percent of the time. In some cases, code assignment was correct 100 percent of the time. Because the verbiage in the medical record included the necessary details to assign appropriate and accurate ICD-10 codes, no one was prepared for how bad the audits look now when reviewing ICD-10 code assignment in test files or for dual coding! It’s not as though anatomy or basic terminology has changed. How can the exact same verbiage be coded correctly in ICD-9 and completely wrong in ICD-10? There’s no concern about being in the same “family” of codes here – it isn’t even the same species.

It is highly doubtful that this affliction affects only one or two artificial intelligence programs. It would behoove all of us who use these typically excellent products to redo a targeted risk assessment immediately. These errors are far-reaching. There are coding errors that could lead to improper payments. There are coding errors that could lead to denials of legitimate claims. Perhaps most importantly, there are errors that will lead to patients being misdiagnosed. As anyone who has ever tried to correct errors in a medical record and/or insurance file knows, it is essentially an impossible task. You may correct one, but the likelihood of correcting all is slim to none. 

Some of the more prevalent examples include coding personal history as current conditions or diseases, coding family histories as current patient disease, assigning confirmed diagnoses when none exist, coding completely inaccurate anatomic findings or body areas, assigning “junk” codes when the details needed to assign the most specific code are clearly stated, coding a completely inaccurate procedure; the list goes on and on. 

So, what does this tell us? Progress can be painful. We will need proficient human coders to review the codes assigned by electronic health records and code-assist products. We will need time to work out the bugs, and precious little time remains. We will need to work closely with our vendors to troubleshoot and resolve issues quickly and accurately. We will need to work closely with our physicians. We will need to be on high alert for any patient concerns about errors. 

The road to success will feature ongoing risk assessments, robust and frequent audits, and dynamic corrective action plans. The best way to fix claim errors is to not make them in the first place.

If you have not reassessed your risk, it’s time to make it a high priority.

About the Author 

Holly Louie is the president-elect of the Healthcare Billing and Management Association (HBMA) and chairs the ICD-10 committee.  The committee developed definitions for readiness and end-to-end testing for successful ICD-10 implementation. 

Contact the Author


Comment on this Article


Incorporate Coding Quality Audits into ICD-10 Preparations

As organizations continue to search for ways to mitigate coder productivity loss following implementation of ICD-10, they also must consider methods to monitor and enhance coding accuracy. After all, productive coders are only effective when those coders assign accurate and complete codes. A low DNFB doesn’t denote coding compliance, nor does it imply that organizations are on the right track in terms of ICD-10 preparations.

Despite the best preparations, organizations likely will see a shift in accuracy rates once ICD-10 takes effect. Why? It will take coders time to get acclimated to working in a production environment.

Those who have had sufficient training and practice certainly will experience less of a shift; however, all coders undoubtedly will feel the effects. That’s why it’s important to establish a process for ICD-10 validation and auditing, particularly during the first few months after implementation.

Following are some tips to help coding managers and directors focus on implementing quality checks and audits:

1.  Strive to get back on track as soon as possible. Today, in ICD-9, coders strive for at least a 95-percent accuracy rate. Although this should be the goal in ICD-10 as well, managers and directors must understand that achieving this goal may take time. However, every organization should target a 95-percent accuracy rate (or better) as soon as possible to mitigate ongoing denials. Dual coding helps identify coders who may be struggling with certain aspects of ICD-10. It also helps identify larger trends (e.g., a root operation with which all coders struggle).

2.  Identify a facility ICD-10 lead who can research ICD-10 questions as they arise. Although Coding Clinic has published ICD-10 guidance exclusively since the first quarter of 2014, many questions pertaining to specific scenarios continue to surface. An ICD-10 lead can compile these questions, submit them to AHA’s Coding Clinic Advisor, and disseminate information once it is received. Many organizations are combining this role with that of ICD-10 auditor. Consider someone who is an American Health Information Management Association (AHIMA)-approved ICD-10 trainer and who has worked with ICD-10 directly for at least one year. To maximize efficacy, this individual should not be involved in coding active records. Rather, he or she should be able to focus entirely on coding education and auditing. 

3.  Establish a plan for how to handle gray areas. Although organizations can submit questions to Coding Clinic, it often takes time to receive a response. It may not be realistic to suspend claims indefinitely while waiting for a response. Determine how you will maintain these cases in a separate queue and how you will bill them if a timely response from Coding Clinic is not possible. In lieu of formal guidance, consider the following:

    1. Ask for guidance from clinical documentation improvement (CDI) specialists or physician champions.
    2. Reach out to state and regional health information management (HIM) associations to inquire how other organizations are handling the scenario.
    3. Ask for input from coding vendors.
    4. Discuss scenarios during weekly coding department meetings. 

4.  Make internal coding guidelines a dynamic document. When issues surface, keep track of how the department ultimately decides to handle them. What is the consensus until formal guidance is received? Document this information and update it when more formal information is received. Be sure to include the specific date on which formal guidance is received, and disseminate this information immediately. Doing so ensures that all coders will code consistently, and it also provides an audit trail should denials occur. Organizations may be able to rebill these cases. It’s helpful to be able to quickly identify a specific date on which the coding process changed. 

5.  Prioritize cases for quality reviews. Most organizations do not have the time or resources to perform a 100-percent manual review of all cases in ICD-10. Instead, focus on the following:

    1. Current cases that require a manual review in ICD-9 (e.g., auditor targets, payor targets, DRGs that are denied frequently). Some hospitals may use editing/validation software to identify cases that are more prone to error.
    2. High-volume cases.
    3. DRGs or procedures with which coders have struggled during dual coding (and for which refresher training has been provided).

 Other tips

When it comes to accuracy and quality, think outside the box. Consider the following other ways in which MRA’s clients are working diligently and collaboratively to ensure quality coding: 

  • Dividing the coding auditor/validation role among various hospitals in a multi-hospital system. This individual splits his or her time among various hospitals, each of which contributes to the cost of supporting this role.
  • Hiring a temporary coding auditor/validator for an 18-month period post-implementation to help the organization get through the initial transition period.
  • Working with an outsourcing company to secure assistance with quality audits before and after implementation. When pursuing this route, be sure to inquire about consultants’ training and direct hands-on experience with ICD-10. It’s easy to understand ICD-10 concepts; however, being able to apply those concepts to actual medical records requires an entirely different level of expertise. 

Remember, when it comes to ICD-10 coding, accuracy and quality must go hand in hand with productivity.

About the Author

Cathie Wilde, RHIA, CCS, is the director of coding services for MRA. Ms. Wilde has been active in the healthcare industry for more than 30 years. Her previous positions have included assistant director of HIM, DRG coordinator at the Massachusetts Hospital Association, and DRG validator at Blue Cross Blue Shield. She has extensive experience in ICD-9-CM and CPT coding, auditing, data analysis, development and testing of coding products, specialized reporting, and in-service training. As director she is responsible for overseeing the coding division, providing the strategic direction of MRA as a local industry leader of quality coding, auditing, and denial management services. Ms. Wilde is an American Health Information Management Association (AHIMA)-approved ICD-10-CM/PCS trainer.

Comment on this Article


ICD-10: Race Against Time; No Turning Back

In many cases at larger health systems, however, it may still be six or more months. I am not sure when the last time you prepared for an exam was, but you certainly don’t stop studying a month before your score actually counts. If you are a good test taker, you probably would spend that last few weeks taking practice exams and reviewing your own scoring data to identify personal knowledge gaps that you would attempt to close in a highly targeted fashion.  

When studying for a board exam, if you got a question wrong regarding a patient with HIV, you won’t necessarily reread the entire chapter on infectious diseases, though.  

As with any training initiative, your ICD-10 training curriculum must account for a post-implementation phase to close these inevitable specialty-specific (and often individual-specific) gaps that will become apparent during the transition period. With over 80 percent of physicians currently being at or beyond full capacity at work, the post-implementation training phase must be targeted and not wasteful of the physician’s time. In fact, it should be based on at least several weeks of billing data to identify physician-specific knowledge gaps.  

Your training curriculum and system should be capable of delivering targeted training interventions at 3-5 minute increments or less, and it certainly should not include lengthy online courses lasting 30 minutes, 60 minutes, or even longer.  

We physicians, including physicians from the AMA (of which I am a member), just don’t have time for lengthy online courses that are not targeted to our individual needs.  

The American Health Information Management Association (AHIMA) ICD-10 training program for physicians, which I am proud to disclose I helped develop, and the new mobile-only enterprise training app they are about to release, are great examples of systems able to effectively deliver this type of targeted training to physicians based on their own billing data. Embracing these resources will help organizations transition to the most detailed ICD-10 codes before the transition period comes to an end.

About the Author

Andres Jimenez, MD is the founder and CEO of ImplementHIT, a physician and PhD candidate in adult education. His company’s training software is used by hospitals to train physicians on ICD-10 and their EHR.

Contact the Author


Comment on this Article



Clinical Labs Likely to Face Challenges Under ICD-10

If none of these scenarios is the case, it will be up to laboratories to educate referring providers, then hopefully they can apply the knowledge received, which will require updating their physician office electronic medical record (EMR) systems and interfaces. Although the sophistication and capability of these systems vary, they must at the very least be able to communicate using ICD-10 codes.

The laboratory requisition generated from the system may or may not provide relevant diagnostic information pertinent to the testing. It is not unusual for requests to include a diagnosis related to the reason the patient was seen by the physician – but this may not be the reason that supports the medical indicator for testing.

Visit Codes

One example of this can be found when ICD-9 diagnosis code V70.0 (routine general medical examination at a healthcare facility) appears on the requisition (on Oct. 1, that diagnosis code will change to Z00.00 – general adult medical examination without abnormal findings). These diagnosis codes are only for the routine evaluation and management of a patient, not for testing related to existing diseases or other conditions that require additional follow-up. 

Medicare provides coverage when a physician bills V70.0. However, when a clinical laboratory submits a claim containing this code, it is statutorily non-covered, and will be denied.

If during the course of the scheduled visit for a routine exam a condition is identified and laboratory testing is ordered to determine a definitive diagnosis, the physician must update the encounter form with the appropriate sign or symptom that prompted the physician to order the test or tests.

In a worst-case scenario, the physician’s EMR system would not be sophisticated enough to print or generate a test request with the related diagnosis in addition to the diagnosis for the CPT visit code for the encounter.

Ideally, the physician’s EMR system will accommodate the two different encounters and the two different diagnoses for them, and it should allow for the physician encounter diagnosis to be separated from the lab test order. 

The Usual Solution

Front-office staff or accessioning staff at laboratories process the insurance forms and test requisitions and enter the diagnosis codes into the library information system (LIS). The specimens are processed and testing is initiated. Results are generated and charges drop once tests are resulted.

Because clinical laboratory staff members are not interpreting professionals, the test results will not impact the diagnosis submitted. Health information management (HIM) professionals do not review or validate most diagnoses submitted on outpatient claims for clinical laboratory tests, nor do they apply the standard coding guidelines.

The claim will not pass through the edits for reimbursement if it is a statutorily non-covered diagnosis, and the billing department is left with only one option: write off the claim.

As the preventive care benefits of commercial and Medicare payors expand, more and more plan members and beneficiaries will receive preventive care and diagnostic screening benefits, such as those identified by V70.0 and soon Z00.0. As a result, laboratories will see more and more requisitions submitted with these codes, but not the code that will allow the lab to be paid.

Outdated Diagnosis Code

Test requisitions that cannot be processed because the diagnosis is missing or outdated (e.g., with ICD-9 rather than ICD-10) will raise administrative questions such as the following: 

  • Will the laboratory hold the specimen and delay testing until the appropriate diagnosis can be solicited from the physician?
  • Or will the laboratory process specimens, perform testing, and hold the claim until it has the diagnosis? If so, what steps will the laboratory need to take to obtain a compliant diagnosis?
  • If available, will the test results be released or held until the diagnosis is obtained?
  • If the decision is made to always perform testing and to expend the labor and reagent costs, what action will laboratories take when they cannot obtain an ICD-10 diagnosis code and, therefore, cannot submit claims? Even though labor and reagent costs already will have been expended, the claim will have to be written off.
  • What impact will the decision to delay testing have on patient care?

Proactive Steps

Adoption and implementation of ICD-10 will impact productivity at all testing phases (pre-analytic, analytic, and post-analytic) within the clinical laboratory. To lessen the repercussions, consider the following.

  • Educate referring providers about the need to supply comprehensive documentation and to comply with coding guidelines.
  • Be sure that laboratory staff research and share diagnosis coverage requirements of the various payer plans.
  • Track coding errors and noncompliant test orders that impact specimen processing and turnaround times, especially the ones that delay patient care. Focus on the high-dollar and frequent-offender clients first.
  • Develop a reporting metric to track denials that occur because of missing, incomplete, or non-covered diagnoses.
  • Communicate the above results to ordering providers. If they make improvements, communicate those as well.

Tough Love

Laboratory directors may have to make tough business decisions regarding client relationships when providers demonstrate a track record for noncompliance and when the level of noncompliance increases due to lack of correct ICD-10 coding. Determine how to proceed with provider accounts that show no or minimal improvements and include it in the business plan.

Work with providers to ensure that updates and system improvements are implemented in a timely manner. Communicate and share information with providers related to interface changes being implemented through laboratory systems.

On and After Oct. 1

With the proper reporting and established metrics in place prior to Oct. 1, you should be able to identify increases in specimen processing and claim processing delays, as well as suspended charges.

Ensure that personnel in strategic positions are adequately trained and increase their number during the last quarter of 2015. Knowledgeable staff can identify trends in a timely manner, develop and implement a corrective action plan, and follow through with validation and reconciliation.

About the Author

Robin Zweifel is a senior vice president of revenue capture services for Panacea Healthcare Solutions, Inc.

Contact the Author


Comment on this Article


Encephalopathy and Delirium: Tomato-Tomatoe, Potato-Potatoe

Further adding to the confusion, ICD-10 includes a code for delirium resulting from other physiological cause (F05). ICD-10 seems to classify all infective psychosis (with the exception of septic), ICU psychosis, cases of delirium superimposed on dementia, sun downing, and acute brain syndrome/acute confusional state as code F05, with metabolic encephalopathy being reserved for everything else.  Special carve-out codes are still available for hypoxic, toxic, hypertensive, alcoholic, and unspecified encephalopathy, with hepatic encephalopathy now being reported in ICD-10 only for cases of hepatic coma.

The Centers for Medicare & Medicaid Services (CMS) assigns a lower severity to the nonspecific behavioral diagnosis of delirium than for the pathophysiological diagnosis of encephalopathy.

Let’s do a quick review of the best advice from the most official sources. Per Coding Clinic, encephalopathy has not been viewed as a form of delirium since 1988. More current guidance in coding clinic is on page 58 of the edition published in the fourth quarter of 2003:

“Metabolic encephalopathy is always due to an underlying cause,” the segment reads. “There are many causes of metabolic encephalopathy, such as brain tumors, brain metastasis, cerebral infarction or hemorrhage, cerebral ischemia, uremia, poisoning, systemic infection, etc. Metabolic encephalopathy is also a common finding in 12-33 percent of patients suffering from multiple organ failure. The development of metabolic encephalopathy may be the first manifestation of a critical systemic illness and may be caused by various reasons — one of the most important being sepsis.”

Note the above definition does not allow for the reporting of encephalopathy in cases that are clearly psychiatric in nature. 

The National Institutes of Health (NIH) further defines encephalopathy as:

“Encephalopathy is a term for any diffuse disease of the brain that alters brain function or structure. Encephalopathy may be caused by infectious agent (bacteria, virus, or prion), metabolic or mitochondrial dysfunction, brain tumor or increased pressure in the skull, prolonged exposure to toxic elements (including solvents, drugs, radiation, paints, industrial chemicals, and certain metals), chronic progressive trauma, poor nutrition, or lack of oxygen or blood flow to the brain. The hallmark of encephalopathy is an altered mental state. Depending on the type and severity of encephalopathy, common neurological symptoms are progressive loss of memory and cognitive ability, subtle personality changes, inability to concentrate, lethargy, and progressive loss of consciousness.”

Once again, the above definition does not allow for the psychiatric causes to be reported as encephalopathy. The reverse is also true. We should not be reporting delirium for scenarios that clearly have a basis in an underlying medical origin, at least not with the (291.x to 293.x) codes, as that is not what these codes convey.

Recovery Auditors (RACs) have been known to reference the DSM IV as an argument for either removing the diagnosis of encephalopathy or as an argument for downgrading the condition to the diagnosis of delirium. As a psychiatric classification system, the DSM IV does not even address the medical diagnosis of encephalopathy – a fact which RACs no doubt find very convenient.

A review of the definitions of delirium on Medline and the NIH reveal similar definitions of delirium that overlap with the same causes and treatments as encephalopathy. Neurologists are perfectly fine using the two terms interchangeably, and why shouldn’t they? This only really impacts things like risk adjustment, hospital profiling, and payment, after all. 

Fortunately, we are in a transition period in which this risk adjustment also will be of concern for attending physicians. I would like to argue here that the convention is to define delirium as that which is occurring from a psychiatric origin and reserve encephalopathy for brain malfunction occurring as a result of an acute or chronic medical disease state. That is, after all, essentially how the terms are being classified when we report them to CMS and the insurance carriers. Since the two diagnoses are not reported with the same codes and have vastly different levels of expected severity, morbidity, and mortality associated with their classification, we really have no choice but to fall in line with the definitions as defined in ICD-9 and ICD-10. 

Physician advisors and physician champions should work with their staffs to provide a framework of reporting based on the severity of illness reflected in the patient’s condition. Encephalopathy should be reserved for patients who are experiencing confusion and altered mental status, etc. from an underlying pathophysiological cause. Delirium should be reserved for patients with an underlying psychiatric condition. Drug-induced mental status changes can be either, but I would advise using drug-induced delirium for mental status changes occurring as a direct result of brain chemistry due to drugs while urging encephalopathy for patients with mental status changes occurring as a result of systemic changes that have an indirect result on the functioning of the brain. Toxic encephalopathy is certainly a valid diagnosis, but unfortunately it is limited to cases of a poisoning (by drugs or external chemicals) for the purposes of reporting in ICD-9 and ICD-10. “Toxic-metabolic,” which defines patients who are essentially suffering from a buildup of natural toxins in the body due to a failure of metabolism or excretion, should be reported as simply “metabolic encephalopathy.”

Regardless of which diagnosis is chosen, the physician will need to commit to a diagnosis and stop using them interchangeably. If physicians are documenting the same condition as both delirium and encephalopathy, RACs simply will state that the documentation is “inconsistent” and either remove both diagnoses or downgrade the reporting to the lowest severity possible. It is also important to link the brain malfunction to the underlying medical cause. For example: encephalopathy due to severe hyponatremia, encephalopathy in a septic patient with severe sepsis and AKI, encephalopathy due to hypoxia with respiratory failure, encephalopathy due to poisoning from an overdose of narcotics and benzodiazepines, encephalopathy superimposed on chronic dementia in an Alzheimer’s patient with multiple metabolic disturbances, etc.

We will have to adapt to the changes suggested by the notes for F05 when ICD-10 is finally implemented, of course, but at least that will bring us a bit more clarity.

About the Author

Allen Frady is a consultant with experience in management, implementation, education and clinical practice.  With 20 years in healthcare, he provides his clients assistance in the areas of documentation, program implementation and compliance.  His background includes critical care nursing, coding, auditing, utilization review, and documentation improvement.

Contact the Author


Comment on this Article