How coding is like Scrabble

One of the great nuisances in medicine is diagnosis coding. According to Medicare and insurance companies across the U.S., each and every disease must have a unique number. Everything must be quantified and recorded. Why? To facilitate analysis, number-crunching, regulations, reimbursements and, of course, we sometimes joke, to perpetuate the jobs of the coders. They usually know the nuances better than the doctors.

Is it truly possible to describe Mrs. Brown’s frequently upset stomach as a letter, number combination? But alas, we must be accountable for what we do. Value and quality are the buzzwords these days.

Electronic health records (EHR) are supposed to help us, but they really don’t. So here’s a quick tour of the International Classification of Diseases-10.

Let’s talk about diabetes:

E11.9 means diabetes without complications.

If we add a system with a problem to the DM code, we get additional points:

  • E11.2 DM with kidney
  • E11.3 DM with eye
  • E11.4 DM with nerve
  • E11.5 DM with vascular
  • E11.6 DM with other (A1c > 7.0)

The goal is to get the points up, as high as you can without fraud. If you add a second and third digit to the E11.6 you get:

  • E11.62 DM with skin complications
  • E11.620 DM with diabetic dermatitis
  • E11.621 DM with foot ulcer
  • E11.628 DM with skin ulcer

So if your diabetic patient has high blood pressure, high cholesterol, their HBA1C is over 7, you should code: I10 (high blood pressure) and E78.3 (high cholesterol) and the DM code becomes E11.65. Code choices are like playing scrabble: You want to be sure you use the boxes that give you the double and triple letter score as well as the double word score.

Now to add icing to this layered cake, Medicare pays more for certain combinations of codes or hierarchical condition category (HCC). Introduced a few years ago, there are 83 HCC codes that map to over 9,500 ICD-10 codes today, but over 68,000 ICD-10 diagnosis codes. So only a fraction of ICD-10 codes carry any HCC weight. You can probably guess which ones: congestive heart failure, diabetes, heart arrhythmias, stroke, asthma, depression, to name a few.

HCCs were developed to adjust payments in accordance with the complexity of the patient. For example, a 60-year-old with high blood pressure, should be less complex than a 60-year-old with high blood pressure, heart disease, and diabetes. It is called a risk-adjustment model for payment. If you add the fact that the person is homeless, doesn’t speak English, or has a learning disability, it gets more complicated. Those issues are what we often refer to as social determinants of health. We should stand up and cheer because if you care for this population, you know the challenges.

However, if you don’t code it right, it doesn’t count. Risk adjustment drives the new physician payment models that emphasize quality instead of quantity.

Ironically, I learned even more about coding as a home visit provider for a health plan. During our webinar, we were told coding is critical to preserve the prosperity of the health plan. My translation, bill Medicare for everything you can so the health plan gets as much money as possible. I learned some new codes: Most patients with heart failure have secondary hyperaldosteronism (E26.1). If a patient has the arrhythmia atrial fibrillation (I48.0), you should add other thrombophilia (D68.69). A patient with weight loss (greater than 10 percent in 6 months), can have protein-calorie malnutrition at any BMI, based on nutritional status: E44.1 (mild), E44.0 (moderate).

Another difference between the health plan (HP) computer and clinic’s, is the HPs is programmed to calculate diagnoses for me, based on what I enter for symptoms, medications, physical findings, screening assessments such as depression or falls.

My clinic’s electronic record just isn’t that fancy. Secondly, our coders limit us to four diagnoses at a time, for the health plan I often have at least a dozen. Because Medicare has amnesia (Andrew’s phrase), all codes need to be reentered annually.

The World Health Organization (WHO) copyrighted, owns, and publishes the classification ICD-10. WHO authorized the adaptation of ICD-10 for use by the U.S. government, or CMS (Center for Medicare and Medicaid Services).

Some interesting history: The first international classification edition, originally known as the International List of Causes of Death, was adopted by the International Statistical Institute in 1893. And for futurists: ICD-11 is here!

ICD-10 is used internationally. Here is a link to the countries using ICD-10. We do a lot of complaining in the U.S., so I asked one of my British colleagues about how they managed coding. Her reply:

In British General Practice (GP) we use READ codes. They been around since 1985 are not based on ICD-10, but crafted around diagnoses in general practice. They have been added to and modified over time. The system is complicated and about to radically change. For the last 20 years, we are paid using a Quality and Outcomes Framework — where GPs are paid on a sliding “points earned” scale for care of patients with specific conditions. This relied on READ coding the disease — e.g., heart failure, COPD , asthma, etc., but also on the measured parameters, eg. BP, HBA1c, and how close we are to targets. Payment is also linked to if patients attended check-ups, screenings occurred, received advice about smoking, diet, etc. — all of which had to be coded. Over 1,000 different parameters are measured. It starts all over on April 1st each year. GPs income depended on achieving points.

Sound familiar? No easy answers here, but I want my clinical EHR to do the fancy calculations that the health plan’s do. And more importantly, in this digital age, can’t we craft a computer system that keeps the patient at the center of the interaction. Clinicians waste precious time clicking boxes and finding the right diagnosis.

Therese Zink is a family physician and can be reached at her self-titled site, ThereseZink.comAndrew Saal is chief medical officer, Providence Community Health Centers, Providence, RI.

Image credit:

View 1 Comments >

Most Popular

✓ Join 150,000+ subscribers
✓ Get KevinMD's most popular stories