DIY ICD-10 conversion – Part 3

December 2nd, 2013 / By Ron Mills, PhD

In DIY Part 2 we digressed from learning how to convert a policy to answer, “What is a code?” We continue with:

Common confusion 2: What is ICD-10?

Wikipedia says , “ICD-10 is the 10th revision of the International Statistical Classification of Diseases and Related Health Problems” developed by the World Health Organization (WHO) and released in 1992. Countries, including the U.S., use it to report mortality and morbidity statistics to the U.N. and other international bodies.

Hold on. This isn’t the ICD-10 we’re talking about. What comes out of WHO is viewed as insufficiently detailed to support payment systems, so countries extend it. The U.S. extension is called ICD-10-CM (for “Clinical Modification”). Developed over ten years with input from many stakeholders, mostly professional societies, ICD-10-CM extends the 14,400 WHO codes to 69,823 codes. It is maintained by the Centers for Disease Control and Prevention,

Further, the WHO does not publish a classification of procedures, so every country is free to make up their own. Come October 2014, the U.S. will have two: For doctor’s office visits and hospital outpatient claims, we’ll be using HCPCS/CPT, most of which (the CPT part) is a proprietary product of the American Medical Association. This is what we have been using all along for those settings. For hospital inpatient claims we’ll be using a brand new system, ICD-10-PCS (for “Procedure Coding System”) with about 4,000 concepts that are assembled into 71,924 codes. ICD-10-PCS is maintained by the Centers for Medicare and Medicaid Services,

Strictly speaking, every time I want to refer to HIPAA compliant coding systems before and after October 2014, I should say ICD-9-CM and ICD-10-CM/PCS respectively. But out of consideration for your eyesight, I’ll usually just say “ICD-9” and “ICD-10” and you’ll know what I mean.

There are two lessons to take away from this digression. First, if you are not concerned with hospital inpatient data, then you can ignore everything I have to say about procedures. Only diagnoses are changing for you. Second, if someone shows up claiming to be an expert in ICD-10 because she has worked with it in other countries – beware. ICD-10-CM/PCS has about the same relationship to WHO ICD-10 as a leopard has to a house cat.

Common confusion 3: There’s just one ICD-9-CM, right? Sorry. ICD-9-CM is updated every October (so is ICD-10-CM/PCS). New codes may be added, obsolete codes removed. The most common change is to expand a code with more digits, so what was a code before October is a header after. One thing they do not do, thank goodness, is re-use a deleted code to mean something new. It may disappear, or become a header, but it never re-appears later meaning something else. Generally, we label each version with the Federal Fiscal Year, which begins in October and takes the number of the following January. For example, we are now using FY2014 ICD-9-CM for coded claims, which became valid on October 1, 2013 and will remain so through September 30, 2014.

Why do you care? To correctly interpret a policy given as a list of ICD-9 codes, you may need to know in which Fiscal Year those ICD-9 codes were written down. Some may now be gone; some may be headers and will need expansion into today’s codes. Second, the ICD-10 codes you convert your policies into will need maintenance. Everyone is trying their best to avoid any changes to ICD-10-CM/PCS between FY2014 (now) and FY2015 (when they become official), but that means demand for changes is building up, so look out for FY2016 and 17.

Common confusion 4: So, ICD-9 codes are numbers, right? ICD-9 codes may look like numbers, but they are NOT numbers. For example, ICD-9 code 003.0 or 0030 is Salmonella gastroenteritis; 030.0 or 0300 is Lepromatous leprosy; and 300.0 or 3000 is a header, Anxiety states, with 300.00 or 30000, Anxiety state, unspecified, as the first code under it. (The decimal point is optional, another source of confusion, and not all codes are the same length.) If you casually load a list of codes into a spreadsheet, the software may “help” you by removing leading and/or trailing zeros and thereby completely destroying the validity of the list. Always designate any column with ICD codes in it as text.

Common confusion 5: ANDs and ORs. I’m sorry to have to do this to you. Try this. How do you gather all the records for patients with diabetes? You take those with ICD-9 codes 250.00 or 250.01 or … or 250.93 (40 codes in all). “No, I want all the patients with 250.00 and those with 250.01 and … those with 250.93.” Or try this. In ICD-9 if you want uncontrolled diabetes with ketoacidosis you can do it with just one code, 250.12. In ICD-10, it takes two: both E11.69, Type 2 diabetes mellitus with other specified complication and E11.65, Type 2 diabetes mellitus with hyperglycemia, on the same record. These words “and” and “or” can be a major cause of confusion. Those of us with a mathematical background use them differently than the way they are often used in casual speech.

In our work, we’ve tried to get around the problem by designating codes in a policy as “alternatives” or “clusters.” Alternatives are codes, any of which, (or more than one of which) on a record, identifies a condition. Alternative codes for diabetes in ICD-9 are 250.00, 250.01, …, 250.93. A cluster is a set of codes all of which must be on the record to specify a condition. It takes the ICD-10 cluster E11.59 with E11.65 to specify “type 2 uncontrolled diabetes with ketoacidosis.”

Another possibility to this may include clusters. For example, in ICD-10, to designate Cardiac Pacemaker Device Replacement, your list might have the single code 0JH83PZ, Insertion of Cardiac Rhythm Related Device into Abdomen Subcutaneous Tissue and Fascia, Percutaneous Approach, as one alternative. It may also have the cluster, 0JH604Z, Insertion of Pacemaker, Single Chamber into Chest Subcutaneous Tissue and Fascia, Open Approach, with 0JPT0PZ, Removal of Cardiac Rhythm Related Device from Trunk Subcutaneous Tissue and Fascia, Open Approach, as another alternative.

Alternatives are like synonyms in a natural language: You can be hungry, peckish, famished, starving. Clusters are like phrases: You can “need food” or “want to eat.”

In part four, we’ll end this series of digressions and get down to converting your policy.

Ron Mills is a Software Architect for the Clinical & Economic Research department of 3M Health Information Systems.