Everything I know about ICD-9, I learned from ICD 10

I may have already mentioned to you, in the strictest confidence, that I have been a lazy diagnosis coder in the past. The reason for that was simple: most physician services were paid based on the CPT code not the diagnosis code. Payers used the diagnosis code to deny the claim for medical necessity reasons. Years ago I was working with a general surgeon who would list all of the patient’s underlying medical problems on the charge slip. I told him “Doctor, you know we don’t get paid anymore if we put one diagnosis or four diagnoses on the claim form.”  But he said to me, “Betsy I want the insurance company to know just how sick this patient is.”  Of course, the bubble over my head was “the insurance company doesn’t care.” But, that was then and this is now. Diagnosis codes aren’t just for claims payment anymore but will have an effect on reconciliation for risk-based contracts including the Medicare shared savings program.

This brings me to ICD 10. I think only the AMA was less happy than I when CMS didn’t postpone ICD-10 another time.  But, when they didn’t, I was forced to learn ICD-10 coding. And that is when I really learned ICD-9. We consultants will tell you the many ways that ICD-10 is more complex and difficult than ICD-9. I certainly won’t deny that going from 14,000 diagnosis codes to 70,000 diagnosis codes will be a challenge. However, although ICD-10 does have unique features, there are many things that are similar between the two coding sets.

Yes, it is an entirely new system, not one built on ICD-9 and yes every code is changing.  And it’s true, a complete ICD-10 code can be 3, 4, 5, 6, or even 7 characters long.  But, many things remain the same. The ICD-10 instructions for selecting a code for physician/outpatient services are similar to those instructions in ICD-9.  Select a diagnosis code that is chiefly responsible for the service.  If there is a known diagnosis or condition, select it.  If the diagnosis is not confirmed, do not use “possible” or “rule out,” use the symptom.  Assign codes to the highest degree of specificity.  Do not code conditions that no longer exist. Code all documented conditions, which coexist at the time of the visit that require or affect patient care or treatment.  If a patient with a history of neoplasm is seen and is no longer being treated for the condition and has no signs of the disease, use “personal history of neoplasm” codes.

In addition, the book will look familiar to coders.  It starts with an alphabetic index, includes a neoplasm table and a table of drugs and chemicals.   There is a tabular list from which to select the code, after being directed to it from the index.  There are external cause codes (these are the codes the popular press loves, “pecked by a turkey”) and factors influencing health status and contact with health services that includes examination and personal history of cancer or long term use of anticoagulants.   The ICD-10 book has general instructions that give sequencing rules for many chapters.  Of course, on the physician side, we don’t always pay as close attention to sequencing as we should and as they do on the facility side. (What can I say, I already admitted I was a lazy diagnosis coder in my past.)  Physician coders should pay attention to code sequencing. Some codes may not be submitted in the first position. When you see the notation “in diseases classified elsewhere” in ICD-9 or ICD-10 it means this should not be submitted as the principle diagnosis.  Code first the condition that is classified elsewhere.  Similarly, some manifestation codes instruct the coder to “code first” the underlying condition or etiology.

Am I making it sound too similar to ICD-9?  Well, there are differences.  Besides the sheer number of codes, there are changes.  Complications of a procedure are no longer relegated to a small section of codes in the 995-999 series but are located at the end of each chapter.  Complications of a digestive surgery are in the digestive chapter, complications of eye surgery are in the eye surgery chapter.  There is a seventh character extender for some chapters, and it means something different depending on the chapter in which it is found.  In the maternity care chapter, it can refer to trimester or fetus.  In the nervous system chapter, the Glasgow coma scale.  I know you’ve heard about the seventh character extender for injuries.  And, as for coding fractures and the many seventh character extenders for those, I think most of us will concede that Orthopedics will have a difficult time.  Long term after effects from injuries are indicated by a seventh character extender.  There is a new concept of underdosing in ICD-10.  And what does “Excludes” mean? There are two meanings in ICD-10.  And, let’s not forget the placeholder code X.

What about mapping programs?  Some groups are hopeful that the crosswalk built into their systems will be sufficient, that there won’t be a need for expensive training days away from the office.  I’ve looked at some of these and here’s what I think.  The automatic crosswalks will very often yield an unspecified ICD-10 code.  Sometimes, this is because the ICD-9 code was not specific, and so the program could only map to an unspecified ICD-10 code.  Or it is because the code options in ICD-10 are much more specific than in ICD-9, such as right, left, bilateral (such as for many eye and ear complaints or musculoskeletal complaints) or foot, calf, knee, thigh, hip (rather than limb.)  In that case, an automatic mapping will always yield a code without specificity.  If these are going into the problem list, then when the patient is seen the clinician needs to review these and select a more specific diagnosis code.  I’ll write more about these mappings in the near future.

I don’t know if the conversion to ICD-10 will be more like Y2K or the end of the world as we know it.  I do know that experienced coders can learn ICD-10.  If you can code in ICD-9, you can code in ICD-10.