Medical Necessity Documentation – the Critical Component for CAC

August 15th, 2014 / By Barbara Aubry, RN

Whether or not you can quote chapter and verse of the Medicare statute that first detailed medical necessity, most of us in healthcare are familiar with its premise1. But from this basic tenant we begin to diverge widely in our understanding of the concept. This is especially true for Medicare inpatient services since CMS does not have specific standards the industry can follow. This issue dates back to the late 1980s when then HCFA admitted, “Current regulations are general and we have not defined the terms ‘reasonable’ and ‘necessary’ nor have we described in regulations a process for how these terms must be applied…”2

CMS goes on to explain that their reimbursement contractors are largely responsible for determining when a service is medically necessary during claim adjudication or while conducting utilization and quality reviews. This practice resulted in the preponderance of post care reviews. As data mining capabilities improved, CMS begin to scrutinize certain clinical areas and their corresponding DRGs. They learned how to identify certain DRGs that have a preponderance of errors. This led to the creation of the often-overlooked-by–the-facility, CERT and PEPPER reports.

Now, it is widely understood that CMS via the RACs et al, audit specific DRGs, including those for acute respiratory failure, sepsis and pneumonia among a list of others. CDI and HIM professionals are well aware of the audit targets. They focus on activities to ensure the capture of severity of illness, presence of comorbidities and complications, risk of mortality and resource intensity, all important data required to accurately assign the diagnoses which leads to DRG assignment.

Documenting humans and medical necessity

So what does this have to do with computer-assisted coding (CAC)? When I think of CAC, I am reminded of my grandmother’s concern regarding frost-free freezers. She was not sure she could trust a freezer to defrost itself, and if it did, it certainly could not do it as well as she could. Plus, she worried that the freezer itself wouldn’t work as well.

I don’t know if you can purchase a new freezer now that needs to be manually defrosted – the ‘frost fee’ models worked so well. Who wants to defrost a freezer if they don’t have to? Similarly, many coders were initially concerned that CAC could not assign codes automatically as well as humans manually using code books. Thankfully the results are in on CAC—it works very well indeed. I don’t know about you, but I prefer making my tasks easier and faster. Of course, I would still want to be able to double check the automated coding work, which is a task that coders using CAC perform at this point. There are coding rule nuances a human still needs to take into consideration for the best outcome.

CAC is dependent on the documentation skills of the human provider since it uses natural languageprocessing – and here, in my opinion lies the challenge – the human factor. CAC is dependent on the provider’s documentation skills, or lack thereof. Consider this example:

A patient presents to the ED with complaints of shortness of breath over the last three days. He relates trouble sleeping at night; he has no chest pain, no history of COPD or CHF. Medical history includes hypertension, hyperlipidemia and smoking. Prior surgical history includes a carotid artery procedure. EMS began CPAP on the way to the ED which was switched to BIPAP on arrival.

On exam there was bilateral edema of the lower extremities, rales where heard in the lung bases bilaterally, oxygen saturation was 99 percent on BIPAP – all other systems were negative. The chest x-ray showed cardiomegaly, pulmonary vascular congestion and bibasilar opacification. He was able to speak 4-5 word sentences and was tripoding.

The documented clinical diagnosis in the ED was acute respiratory failure with CHF.

In this case, it is likely that CAC would suggest acute respiratory failure and CHF because it was documented by the ED physician. Now, let’s look at an example of suggested clinical criteria to support assignment of the diagnosis of acute respiratory failure:

1). Data is expected to be present that includes abnormal ABGs (arterial blood gases) and abnormal C02 level s. Neither of these values was documented.3

2). In addition to respiratory compromise which the patient had, other clinical signs and symptoms commonly associated with ARF include:

  • Cyanosis
  • Dizziness
  • Confusion
  • LOC

3). Historically, certain clinical conditions are often associated with acute respiratory failure. These include:

  • Sepsis
  • Trauma
  • Aspiration
  • Pneumonia
  • Pancreatitis
  • Drug/medication toxicity
  • Multiple transfusions
  • Cardiogenic pulmonary edema (accompanying hx of chest pain/paroxysmal nocturnal dyspnea (+) orthopnea (+)

4). Presence of clinical criteria found on examination required to establish a diagnosis of ARF:

  • Tachypnea and dyspnea, crackles upon auscultation
  • Direct insult causing or systemic process causing lung injury
  • Radioloic appearance – 3-quadrant or 4-quadrant alveolar flooding
  • Gas exchange – severe hypoxia refractory to 02 therapy
  • Normal pulmonary vascular prosperities – pulmonary wedge pressure lower than 18 mm Hg

As we can see, the documented presentation of the patient does not appear to support the clinical criteria required to assign the diagnosis of acute respiratory failure. This is not to suggest the patient was not ill or should not have been admitted, but rather a different, more precise diagnosis should have been chosen by the physician that was more closely aligned with the clinical findings.

My opinion – medical necessity and clinical criteria

In the example above, the computer assisted coding system worked precisely as it should, but its efficacy was impaired by the imperfect human. To get the most from a CAC system, it is critically important to educate the documenting humans.

CMS does not use a specific set of clinical criteria, but has instructed physicians to use their best efforts to make an accurate determination. I’ve always felt that was a bit unfair to the provider, since I assume the auditors are using criteria. But that does not prevent a facility from working with its medical staff to choose clinical criteria it believes can help clarify ambiguity in code assignment by providers.

My colleague Cheryl Manchenton recently posted an excellent blog titled “Oh what a tangled web we weave…The impact of CDI and Coding part II.” If you haven’t read it I recommend it, as it includes some eye-opening information. In the blog, Cheryl proposes the “development of standard definitions of acute respiratory distress and acute respiratory failure (for all patients regardless of whether the patient had surgery) for consistent and compliant documentation and reporting.”

Perhaps my thinking is even more radical; don’t wait for regulators to agree on standards – choose standards from a recognized entity now. Many medical specialty societies and respected organizations have criteria available. Work with the medical staff to choose those they are most comfortable with. I believe the use of clinical criteria will assist physicians in accurately selecting the appropriate diagnosis. There is no question that the best person to choose a diagnosis is the attending physician, but sometimes it’s challenging to determine the difference between respiratory insufficiencies versus failure. Use of criteria can assist in the determination, and if audited, following criteria is a reasonable defense.

If you are lucky enough to be using a CAC system, be sure to review your facility’s high-risk DRGs. Are you receiving any denials or requests for medical necessity reviews? If the answer is ‘yes,’ I suggest an internal audit to determine if your humans are documenting appropriately. Focus on clinical criteria – is expected data documented to support the assignment of the diagnoses? Use clinical criteria to help make that determination.

If you are in the market for a CAC system, consider the vendor who also offers robust consulting services. Look for skilled documentation improvement professionals who can determine your provider’s documentation ‘IQ.’ Expect them to work with clinicians to help make them documentation geniuses. This critical step will improve ROI and promote the most robust use of your system.

Barbara Aubry is a Regulatory Analyst for 3M Health Information Systems.

1 For more visit and see Medicare Medical Necessity: Avoiding Overpayments, Penalties and Fraud Allegations.

2 Medicare Program; Criteria and Procedures for Making Medical Services Coverage Decisions That Relate to Health Care Technology, 54 Fed. Reg. 4302, 4304, 4308, 4312 (1989)

3 ‘Respiratory Failure Clinical Presentation’, Ata Murat Kaynar, MD, Chief Editor, Michael R Pinsky, MD, CM, FCCP, FCCM, Sat Sharma, MD, FRCPC , Cory Franklin, MD, Harold L Manning, ND, Francisco Talavera, PharmD, PhD.