Health Information Management

Coding as an art: Another way to look at coding

JustCoding News: Inpatient, August 15, 2012

Want to receive articles like this one in your inbox? Subscribe to JustCoding News: Inpatient!

by Joel Moorhead, MD, PhD, CPC

Every so often I search medical databases for articles on coding and clinical documentation. One such search returned an interesting article, “The coding masterpiece: a framework for the formal pathways and processes of health classification.” The article, published in the Australian Health Information Management Journal in 2011 (40(1):14-20), looked at coding as a process of creation.

Authors Emily Price and Kerin Robinson discussed the creation of a new way of looking at art (cubism) as a metaphor for the "complex, exacting, and knowledge-dense work" needed for accurate coding. The authors define cubism as "a form of art in which objects are broken up, analyzed, and re-assembled in an abstracted form." Artists including Pablo Picasso experimented with breaking up objects and reassembling them into abstract visual forms.

In a section entitled, "Applying the concepts of cubism to clinical coding," the authors cited three essential steps that the "coding artist" performs in reassembling medical record elements into abstracted form. We all follow these steps, although maybe not consciously and maybe not as consistently as we could.

Step 1: Deconstruct the episode of care

This step takes only a minute or two, but is very important.

Note the admission and discharge dates. The chronology of the admission often depends on referring back to these dates.

Check demographic data, including age and gender.

Look at the type of discharge. Accurate MS-DRG assignment may depend on whether the patient was discharge to home or to another acute care hospital, or whether the patient died or was discharged alive. If you are reviewing a case at an accepting tertiary care facility, you need to know whether the correct principal diagnosis is a subsequent episode of care, a late effect, or a continuation of the care for the acute condition.

Step 2: Analyze and extract the relevant information

Take a satellite view of the admission. Does the length of stay correspond to the documented conditions? If not, one or more significant diagnoses or procedures affecting the length of stay may be lurking beneath the surface. Or the opposite may be true—the apparent diagnoses and procedures may overstate the severity of the patient's medical condition.

Assign codes corresponding to each significant diagnosis and procedure. This is a process of abstraction. More about this later.

Many wise people (including Yogi Berra) have advised that we see basically what we are looking for. Some of the conditions that we extract from the medical records may be explicitly named. Others may be implied but not named.

The 2008 American Health Information Management Association (AHIMA) Query Guidelines state that we may initiate queries "to obtain a condition for which clinical indicators are present but the provider does not indicate the actual condition itself." Clinical advisors to the coding department can be very helpful in pointing out significant diagnoses which are supported by clinical indicators but not named in the medical record. The 2008 AHIMA Practice Brief "Managing an Effective Query Process" and the 2010 AHIMA Practice Brief "Guidance for Clinical Documentation Improvement Programs" provide examples of compliant queries for situations including respiratory failure when clinical indicators are present but the physician does not specify the condition by name.

Step 3: Reconstruct the admission using codes that are supported by medical record documentation

Images in cubist paintings communicate essential realities about the objects on which they are based by creating abstractions of those objects. The artist's goal is to express the underlying reality of the elements in the painting, using valid rules of composition. The coder's goal is to express the underlying reality of the elements of the hospital admission, using valid rules governing the assignment and sequencing of ICD-9-CM codes.

Be sure to take full advantage of your expanding clinical knowledge and the expertise of your clinical advisors as well. The goal is to make sure that the selection and sequencing of reported codes express the underlying reality of the admission with the greatest possible accuracy and specificity. At the same time, we must conform to the structure imposed on the coding process.

The following Coding Clinic guidance supports efforts to make sure that the codes submitted are accurately represent the underlying realities of the admission:

  • Coding Clinic January-February 1985, pp 8-11, advises that “a basic coding principle is to refer to the ICD-9-CM Tabular List to verify that the code number selected is in accord with the desired classification of the diagnosis.”
  • Coding Clinic January-February 1985 further advises “referencing the Alphabetic Index so that the physician can make a classification selection. One approach is to look at the modifier terms enclosed in parenthesis following the main term entry, and to reference those terms in the Alphabetic Index… The coder should show the physician the main entry terms referenced and the code assignments in the tabular list. ”


In all of the situations below, coders may need to report conditions that are not explicitly named in the medical record to capture the clinical realities of the admission. A compliant query could support the most accurate and specific code. Consider the following examples.

  • Hypercoaguable state. A patient on lifelong Coumadin® is admitted for his third deep vein thrombosis after a minor surgical procedure for which Coumadin had to be stopped. He had a history of pulmonary embolus. His homocysteine level is elevated. These circumstances support the presence of a hypercoaguable state, although the physician may not document that term. The hypercoaguable state had clear clinical significance to the patient's care.
  • Compression of the brain. MRI and CT scans of the brain consistently report terms such as "midline shift" and "mass effect" when the position of intracranial structures within the skull is affected by cerebral edema, hemorrhage, or space-occupying lesion. For years, coders struggled with capturing these terms. Coding Clinic Third Quarter 2011 brought welcome clarity to coding for these terms, advising that coders should query the provider to determine if the documented midline shift or mass effect indicates that the patient suffered compression of the brain (348.4). Other patients with intracranial conditions may have cerebral edema documented on their imaging studies, and may be under treatment with Decadron® for their cerebral edema, but no treating physician specifies a diagnosis of cerebral edema. A query under these circumstances may be helpful as well.
  • Functional quadriplegia. Physicians rarely document the words "functional quadriplegia." Since physicians do not treat "functional quadriplegia," they are unlikey to begin documenting this term any time soon. But functional quadriplegia (code 780.72) captures the clinical reality of a patient who is not able to participate meaningfully in his or her own care, a circumstance that greatly increases the nursing acuity of the admission. Similarly, Coding Clinic Third Quarter 2011, "Clinical significance of obesity," states that obese individuals are at increased risk for certain medical conditions, and that obesity is always clinically significant and reportable when documented. Documentation of clinical significance is therefore not necessary to report obesity codes. Likewise patients with functional quadriplegia are at increased risk for pressure ulcers and for a number of conditions that are complications of immobility. Nurses may document frequent turning of the patient and protection of bony prominences, as well as the need for total care in the areas of mobility and self-care. Functional quadriplegia is a good example of a condition for which the 2008 AHIMA Query Guidelines state would again support a query "to obtain a condition for which clinical indicators are present but the provider does not indicate the actual condition itself."

Price and Robinson make two other important points. First, the development of clinician-coder partnerships adds "enormous value and depth to the quality of the coding." Find ways to take advantage of nurses and physicians who may be available to the coding department.

Second, internal and external audits can be enjoyable and can promote understanding and respect for the work that coders do. Admittedly, embracing audits may require some attitude readjustment. Auditing at its best is a two-way educational process, bringing benefits to both the original coder and the auditor. If the coder can document solid support for his or her coding position based on ICD-9-CM and other published guidelines, the audit process holds no fear. If the auditor presents a better-supported coding position, give the devil his or her due and learn from the experience. If your coding position looks better than the auditor's, consider getting support from your department and appealing any denials to the highest level.

Two or more different coding interpretations of the same clinical data in an admission can commonly be compliant and supported by coding guidelines. Experience tells me that the best-supported coding position is likely to be affirmed on appeal. The coding department often presents the best-supported coding position, which wins over the less well-supported position of the auditor who may have rejected the original coding of the case.


I spoke once with a managed care medical director who used the term "creative coding" in a dismissive way.

Consider the alternative view expressed by Price and Robinson. Coding is creative at its core. The codes are a numerical abstraction of narrative text and test results, not a clerical transfer of data from one document to another. The final coding abstract at its best is a creation that expresses the underlying reality of the admission with maximum accuracy and specificity, while following all of the rules that govern clinical classification. As Price and Robinson say, this is truly "complex, exacting, and knowledge-dense work."

Editor’s Note: Joel Moorhead, MD, PhD, CPC, is an adjunct assistant professor at the Rollins School of Public Health at Emory University in Atlanta. He is also Clinical Director of Research and Development for FairCode Associates in Baltimore, Md. E-mail him at

Want to receive articles like this one in your inbox? Subscribe to JustCoding News: Inpatient!

Most Popular