A Medical Economics Web Exclusive

The Way I See It

Coding is hazardous to medicine’s health

Is a misplaced emphasis on numbers and statistics skewing our view of disease?

By John R. Egerton, MD
Family Practitioner/Friendswood, TX

I’m worried that coding is changing the course of medicine–not just clinically, but epidemiologically, as well. In compelling us to dance to their tune, insurers sometimes force us to make up our own steps. As a result, we doctors may end up misrepresenting the true incidence of disease. Here’s what I mean.

Mrs. Abbott (I’m not using real names) reports that she has a runny nose, pain in her knee, and a "funny feeling when I turn my head this way." I have to translate these symptoms into a diagnosis, and then find the "appropriate" code.

I write that she has an upper respiratory infection (knowing she could just as well have vasomotor rhinitis), arthritis in the knee, and vertigo. Strictly speaking, she doesn’t have vertigo, but how else could I code that "funny feeling"?

"What sort of arthritis?" my receptionist asks.

"Osteo," I mutter, thinking I’m probably–but not certainly–right.

Thus, Mrs. Abbott’s three diagnoses are entered into the computer, gaining electronic certitude. I believe I treated her appropriately, and I fully charted both her symptoms and my findings. But I’m aware that these diagnoses are not exact–and may not even be correct. Am I practicing bad medicine?

Another patient, Mr. Dees, is admitted to the hospital with chest pain. Initially, a cardiologist sees him, and various other experts weigh in as his testing progresses. By the time he’s discharged, his diagnoses include ischemic heart disease, hyperlipidemia, renal insufficiency, gallstones, and prostatic hypertrophy.

These indisputable diagnoses, made after careful testing and investigation, join my diagnostic conjectures about Mrs. Abbott in the data pool. There, the accuracy of both patients’ diagnoses is deemed equivalent.

A third patient, Mr. Hay, complains that he doesn’t have as much strength as he used to; he’s also a little out of breath and has a dull ache in his left arm when he exerts himself. I can’t spare the time to figure out all the complicated codes that will justify giving Mr. Hay the ECG I’m sure he needs. It’s far simpler to code him for chest pain.

Is it wrong to bend the codes to make life in the real world easier? Physicians do this all the time. As long as we are not trying to defraud, does it matter how we arrive at our conclusions? Who among us has the time or inclination to debate every symptom and diagnosis just to make sure the code comes out right?

There’s a perception among physicians that insurers become suspicious if the same diagnostic code keeps reappearing–and when insurers are suspicious, they sometimes do not pay.

So we become creative. Plain old sinusitis turns into maxillary sinusitis, frontal sinusitis, or even ethmoidal sinusitis–a clinical definition that’s difficult to prove or disprove. Since "rule out" no longer counts, it’s best to code a suspicion of a urinary tract infection–even though the culture may be negative, and a subsequent diagnosis may prove more accurate.

What if I think my patient has exercise-induced asthma? There’s no code for exercise-induced asthma; therefore, it doesn’t exist. That may explain why intrinsic asthma–for which there is a code–is on the increase.

Sometimes, even though the diagnosis is clear-cut, coding for it is untenable. Mrs. Kalem is suffering from fatigue, sleeplessness, memory loss, and tearfulness. Strictly speaking, her diagnosis is depression. After many years of treating depression and anxiety in the office, I feel comfortable handling such cases, and I’m quite sure a dose of counseling and an antidepressant will cure her in a few weeks. But her insurance company doesn’t cover mental illness, so I code her diagnosis as "fatigue," "insomnia," or "amnesia." So far, no insurer has questioned why I prescribe antidepressants for "fatigue."

When dealing with an insurer that insists all psychiatric illness be treated by the mental health specialists under contract, I’ll diagnose "headache," "chest pain," or whatever the presenting symptom seems to be, and then prescribe an antidepressant. I don’t think it’s fair to listen to the patient describe a variety of symptoms that I feel confident mean depression–and then tell him or her to call the 800 number on the insurance card so someone else could prescribe the necessary medication.

I imagine similar scenarios are occurring all over the country, and I wonder what effect all this is having on the accuracy of our statistics about depression.

Reducing the complexities of life to a string of numbers fits well into our modern ways, but medicine is more complicated than codes and digits. How do you relegate to a group of figures the concerns of a young woman who is worried about her husband’s job, is trying to get along with her mother-in-law, and has just found a lump in her breast? What about the 45-year-old man who has just lost his father to cancer, realizes he hasn’t fulfilled his own career aspirations, and has developed chest pain? Or the child who sees her parents fighting, is being teased at school, and is suddenly wetting her bed?

I realize we need numbers to measure, analyze, and act on the basics of what we do, but numbers are merely a framework. The computer can say only "yes," or "no"–it cannot deal with the vast gray areas encompassed by "maybe." I am saddened to think that medicine may become a series of digital codes ruled by rote, bound by the binary system.

Some things do adapt well to coding, such as bodily injury due to a falling object. Add more numbers, and you can identify the part of the body that’s been injured, the nature of the falling object, and whether the object’s fall was accidental or deliberate.

These are all clear-cut specifics. But when the same idea is applied to coding the complexity of an office visit, the result is a ludicrous list of systems, inquiries, and parts of the body examined. Taken together, this list is intended to set a value on the clinical encounter–but it does not, and it can not.

The vast numbers of codes that are entered every day–some carelessly, others craftily, incompetently, or even criminally–must be having a profound effect on our formal perceptions of diagnoses and treatments. Once these entries are electronically stored and analyzed, they become gospel.

This gospel then, decides the commandments that control the prophets of 21^st century medicine.

Or should that be profits?

The author is a contributing editor to Medical Economics.

John Egerton. Coding is hazardous to medicine's health.

Medical Economics

2001;11.