Friday, October 24, 2008

Decoupling Disease Coding from Disease Classification

We have hinted at a proposed alternative to ICD-10-CM for coding diagnoses: each diagnosis should receive its own code. Or, to be more specific, since a diagnosis is a conclusion about what disease a patient has, each disease should receive its own code.

In a prior post, we pointed out that ICD-10-CM is a disease classification, and thus the two diseases typhoid endocarditis and typhoid myocarditis both get the same code A01.02, which is really the code for the class of diseases called typhoid fever with heart involvement.

Instead, we propose that all three diseases get a code. Typhoid fever with heart involvement (perhaps inflammation of heart tissue due to typhoid fever is a better term) should also get a code in case the doctor doesn't yet know whether the myocardium or endocardium is affected, so she can still assign a disease code to the patient prior to having full knowledge.

These three diagnoses are at different levels of precision: typhoid endocarditis is a more precise diagnosis, and typhoid fever with heart involvement is a less precise diagnosis. It is the nature of medicine that as doctors gather more information through interviewing, examining, testing, and assessing the response of a patient to treatment or watchful waiting, they are able to improve the precision of their diagnosis.

For example, the doctor may diagnose pharyngitis and then, when she gets the results of a culture two days later, improve the precision of the diagnosis to Streptococcal pharyngitis (that is, if the culture grows a type of bacterium called Streptococcus. If the culture does not grow any bacteria, then the precision of her diagnosis does not change significantly).

Proponents of ICD-10-CM usually refer to increased precision of diagnoses as improved "specificity" or increased "detail". However, in the medical field, the word "specificity" has another, more-commonly used, mathematical meaning. So it is best to avoid its use altogether in this discussion.

We have already pointed out that much of the alleged, improved precision in ICD-10-CM is not improved diagnostic precision at all, but rather just combinations of diseases, their manifestations, and other information about the patient and their encounters with the healthcare system (as opposed to other information about disease itself). Thus, although ICD-10-CM has 290 codes for diabetes mellitus, these codes do not represent different types of diabetes mellitus at different levels of diagnostic precision.

Doctors should simply code each disease at the highest level of diagnostic precision possible at the time. As they acquire new information that improves that level of precision, they should update the disease code.

Doctors should NOT classify the disease according to complex rules and inclusion and exclusion criteria, then assign the code for that class of diseases. Especially when the particular classification of diseases--ICD-9-CM or ICD-10-CM--is not intended for helping doctors care for patients, but rather for helping researchers, policy makers, and insurance companies analyze patient data for other purposes.

Instead, those who need to classify disease for research, infectious disease surveillance, setting health care policy, improving health care quality, and so on, should take the disease codes that physicians assign (for the purposes of patient care) and classify those disease codes as they see fit for their unique purposes. We doubt that a single disease classification such as ICD-10-CM can meet all those so-called secondary uses of data, anyway.

Allowing physicians and other providers to code each disease individually has the following benefits:

1. Health care providers (hospitals, doctors, etc.) never have to change coding systems, they will always use the individual disease coding system.

2. Those entities who need to organize patient data by classes of disease can organize diseases into whatever classes they need. If they need new classes of disease, they can simply re-organize the disease codes into new classes as needed.

3. No one will need to propose or be burdened by expensive, complicated, far-reaching upgrades to a one-size-fits-all, complicated, poorly designed, and technically obsolete diagnosis classification.

No comments: