In the last lesson I stressed the importance of standardizing technology. For this lesson let's start with the most ubiquitous coding schema that has been used to classify diseases and conditions, The International Classification of Diseases or ICD. I will first introduce version nine of ICD and then we'll move on to ICD-10. By the end of the lesson you will be able to analyze a dataset with ICD-9 or ICD-10 codes and then be able to select which values refer to specific diseases. So, what is ICD-9? Originally published by the World Health Organization in 1959, ICD was short for International Classification of Diseases adapted for indexing hospital records and operational classification. Yes, this is a very long term. Thankfully it was abbreviated as ICD and was developed to help support the development of a standardized classification of hospital records. In 1978, the ninth revision of ICD was released, they called it ICD-9. The benefits of ICD-9 include improved opportunities for reporting and research. With this greater analytical capability, you can monitor patient care more closely and streamline communications. If you are all talking about the same terms, there's no need to explain them in detail. ICD has gone through many modifications. In 1979, clinical modifications were included known as ICD-9-CM coding schema. These were used for standardized hospital billing. While ICD-9 is updated annually to remain clinically accurate, major structural changes are generally quite rare. There have been nine revisions up to 1979 but major revisions don't happen very frequently. These revisions are maintained and agreed to by the National Center for Health Statistics, The Center for Disease Control and The Centers for Medicare and Medicaid Services all under the Department of Health and Human Services here in the United States. As you can imagine three or more very large federal agencies working together don't particularly move quickly. These are glacial changes that take place over decades which is why the ICD-10 revisions were only recently implemented in the United States in 2015 or 36 years later. Now let's look at the structure of an ICD-9 code. An ICD-9 code is a three, four, five digit code that can be broken down. The first two digits represent the category of the disease or condition, and the third digit helps identify where on the body the disease occurs. The first digit after the decimal point, the fourth digit gives an indication as to the cause or ideology of the disease manifestation, and the fifth digit is an additional sub-classification. An example is diabetes mellitus 249 which represents secondary diabetes mellitus. After the dot, the 00 indicates that there is no complication and is not stated uncontrolled or unspecified. This one is a very simple ICD-9 code, 249 is the high level category. You can then have subcategorizations with additional codes after the dot. We've taken a look at ICD-9 which is ubiquitous in health care. ICD-10 on the other hand was published in 1990 and it was not adopted in United States until 2015. Thus ICD-10 increases the five digit code to a seven digit alphanumeric code. ICD-10 offers quite a few more codes for diagnostic coding to offer more specific disease reporting. It gives explicit characterization for which each side of the body is affected, right side, left side or both and there are specific codes for indicating causation, complications, detailed anatomical location, and severity and diagnostic coding. ICD-10-PCS is primarily for inpatient procedure coding which represents an increase to 87,000 codes and as a drastic increase in the number of codes available from ICD-9 giving much better granularity in describing inpatient procedures. What's wrong with ICD-9? Why can't we just continue to use ICD-9 codes if they're so ubiquitous? One answer is that ICD-9 codes are big buckets of information. The point of ICD-9 codes was not to provide the best way to understand the nuances of the clinical reality documentation, it was a way to get the bill out the door of the hospital by characterizing conditions, diseases, and procedures in the most efficient way to get paid for them. With increase of diagnostic codes and procedure codes in ICD-10, we get a lot more granularity to better characterize what's actually occurring. There were 13,000 diagnostic codes in ICD-9 versus 68,000 codes in ICD-10. Likewise the 11,000 procedure codes in ICD-9 correlate 87,000 procedure codes in ICD-10. In some ICD-10 provides a much more granular descriptive framework for characterizing the clinical reality while still retaining the same basic concepts represented in ICD-9. Some people would say that this explosion in new codes available is almost comical. Some of the characterizations available in ICD-10 are really odd, it makes you wonder the intent. For example, the code Y9272 is an ICD-10 code that indicates a chicken coop as a place of the occurrence of the external cause. You can then aggregate data based on how many injuries occurred in a chicken coop. This is probably very helpful to someone's research in agricultural areas but this is not going to be generally usable by clinical population. Another of my favorites is Z63.1 or problems in relationships with your in-laws. Having more specific data available means improved opportunities for measuring quality, safety, and the efficacy of care which is the goal of quality improvement efforts at every hospital. Retrospective research and epidemiological studies can be enhanced with higher quality more granular data. Clinical financial and administrative performance can be both measured and improved and public health risks can be better tracked and reacted to at the state local and federal level. Of course, improved analytical capabilities is why you're taking this course so ICD-10 might be good for you. Here's a snapshot of the ICD-10 format, you notice two additional digits. There is still a category, the prefix before the dot separator is now three digits long. As a result, a category can be more explicit. The ideology is still the first disease after the dot, now the atomic site has its own digit code and the third digit after the dot references the severity of the disease. The seventh code represents an extension. It's an additional modifier that can better characterize or more explicitly characterize the disease code. A sample here is S52 which represents a fracture of the forearm and the digits after the dot give more specificity as to where the fracture took place, the displaced fracture of the head of the left radius. So precisely which bone and where on that bone the fracture occurred. The last code here, C in this case, represents the initial encounter for open fracture type IIIA, IIIB, or IIIC. There can be dozens and dozens of extra codes to very explicitly identify what type of visit this is, you can then really get into nuances of what was occurring at this point in the clinic. This is where some of the codes are pointed out as being comical such as the chicken coop example. Does it always matter in a clinic when a patient is seeing someone with a broken arm, whether or not the arm was broken in a chicken coop or just outside the chicken coop or maybe while they're water skiing? These are all codes that are available in the ICD-10 coding system. You may notice that historical data is going to be harder to retrieve in ICD-10 formats. If it's been coded in a health system with ICD-9, there's not a great likelihood that historical data will be recorded to ICD-10. It's a tremendous effort and not likely to occur anywhere. So, how do you then aggregate data prior to October 1st, 2015 and data after October 1st, 2015? There are conversion maps that exist between ICD-10 versions. However, not all conversion approaches are equal and these should be assessed with caution. To illustrate the complexity of ICD-9 to ICD-10 crosswalks, consider the following, many codes can't be converted without additional detail and this is not usually provided in the ICD-9 code. How would you know how to characterize an ICD-9 code with the chicken coop modifier when it's not represented in the clinical note? Some detail is just not going to be available on the retrospective data. So, be careful looking at coding systems that map from one version of ICD to another, think of the purpose and necessity of doing so and proceed with caution. Great. You're now familiar with one of the most common and important healthcare data terminologies, ICD. We will now move on to other coding systems in our next lesson.
