ICD-10 Shouldn’t Dictate Patient Care or EHR Design

I have recently been excoriated by Carl Natale, the editor of ICD10Watch / Government Health IT, for poking some fun at and criticizing ICD-10. Several of his statements warrant a comment.

First, he says that I suggest that “health insurers will be able to reduce reimbursements” under ICD-10. This may be, but my actual comment was that health insurers that are also in the life insurance business may discover an incentive to use detailed medical codes to make decisions about life insurance premiums (an area where pre-existing conditions MAY be taken into account when setting rates).

Second, he addresses the fact that I was not able to locate a code for spousal abuse. In fact, as he notes, such codes do exist in ICD-10. Mea culpa. Yet I submit that my inability to find them, despite spending more than an hour searching ICD-10 using the best keywords I could dream up, underscores a basic problem for any actual clinician who is not an ICD-10 expert: There are so many codes that locating the correct one can be extraordinarily burdensome. Dr. S. K. Nath, the Deputy Director General of the Indian Central Statistical Organization, also finds this to be an issue and has summarized what he considers to be the major problems with ICD-10. They are:

• The size of the classification scheme
• No formal training
• No user-friendly software to guide the selection of the appropriate code
• No abridged version based on local requirements
• No online helpline system

In other words, ICD-10 is not something that is appropriate for (or designed for) physicians to be using in a raw form while they are in the process of caring for patients. They don’t have the necessary training. Coding is a job for coders, not doctors.

Finally, Natale addresses the comments I made about coding for a diagnosis of latent tuberculosis infection (LTBI). I could find no diagnosis codes for LTBI in ICD-10, and still can’t. Natale wonders why I don't think I could use ICD-10 code R76.1, abnormal reaction to tuberculin test. This is another code that my extensive searching and browsing failed to discover; it is not grouped with the other tuberculosis diagnosis codes. Regardless, I would not use that code because “abnormal reaction to tuberculin test” is not a diagnosis. LTBI is a diagnosis. There are several criteria that can be used to establish the diagnosis of LTBI, only one of which is an abnormal reaction to the tuberculin test. Merely coding one or more of the criteria is not the same thing as making a diagnosis - which one can't do because there is no code for it.

LTBI is not some rare disease or condition. In the U.S. the vast majority of individuals that have any TB infection at all have LTBI and, as I said before, ICD-10 appears to be useless in this case - however, if the patient happened to have been struck by a macaw, well by God there's a code for that.

None of this should matter from the limited perspective of billing, which is a major focus of the coding system. As long as both payer and provider agree on the same ground rules, it really makes no difference whether the code that is assigned accurately describes the patient's illness or diagnosis. It only has to be close enough that the payer can be satisfied that it is paying for something appropriate.

The problems start when others come along later and use those billing codes as the basis for other decisions, either about the patient or about public health or healthcare policy.

The final problem with ICD-10 is that it fails to solve the problem it was invented to solve, described nicely by the American College of Emergency Physicians: “Frankly, ICD-9-CM is running out of codes. Hundreds of new diagnosis codes are submitted by medical societies, quality monitoring organizations and others annually. ICD-10-CM will allow not only for more codes but also for greater specificity and thus better epidemiological tracking.”

So, does ICD-10 solve the problem that necessitated its creation? The answer is no. The number of digits in a code has been expanded from five to as many as seven. While this allows room for more codes, it also means that that ICD-10, whatever else it may do, has effectively set the stage for a repeat of this healthcare Y2K problem when ICD-10 fills up and ICD-11 becomes necessary.

Coding, no matter how accurate, is never going to help physicians care for patients and it isn't going to help patients communicate with their doctors, understand their medical conditions or treatments, or improve the coordination of care. Getting paid is important but the codes should be relegated to and kept confined in the billing and statistical departments. They should not be allowed to dictate the design of EHR systems that are to be used by practitioners as an integral part of the process of delivering care.

Find out more about Dan Essin and our other Practice Notes bloggers.