Reasoning around moving to ICD10 and our subset of it

This article explains our choice of ICD10 as a coding schema and of the subset of codes we include.

Why are we moving to a new diagnostic coding schema?

Our old diagnosis list was made in-house designed mostly for the Critical Care setting where our database started out. We are moving to a new diagnosis set primarily to make our data set more comparable to other data sets, and to include more diagnoses that are relevant to Internal Medicine physicians outside of the ICU setting.

Why are we moving to ICD10 rather than a different coding schema?

ICD10 is one of the most widely used diagnosis coding standards. Among others, our hospitals' medical records departments use it. Using the same system as used elsewhere means that diagnoses don't need to be translated to another system, where often a 1-to-1 translation is not possible. It also means that our definitions are more likely to be equivalent to those used by other data sets.

Another benefit of moving to ICD10 will be that it should eliminate diagnoses for which we don't have a code, since ICD10 includes codes for diagnoses that have no specific code.

Why are we including the subset of diagnoses that we do

We are seeing increased interest by Internal Medicine physicians in our data set. Since they fund our program to a large extent, we need to make sure our data includes what they are interested in, in a format that is usable for them.

Why do we not include the entire set of ICD10 codes

The full ICD10 includes over 16000 diagnosis codes. We decided to only use a subset of this to make it easier to learn the new list. Any diagnoses whose codes we omitted can be assigned to the next more general NOS code.

Should we remove some?

Template:Discuss@task Laura is working on a list of dxs we might want to remove. Please put here when ready. (emailed Laura Ttenbergen 16:15, 2017 November 6 (CST))

• suggested we remove the following:
• sunburn,
• alopecia,
• albinism,
• disorder of the nail.--LKolesar 07:30, 2017 November 7 (CST)

Concerns

Do we really compare our data to other data sets

• What is the rationale for converting to the ICD10 system? Is it so that we can compare data to a universal system? How much is this going to be done and how will this bring value to our database?

• I want to propose that other alternatives exist. Another idea would be to fix our old system to include the list we already have of codes that are missing...
  • That would not make it any more comparable to other data sets.

• ... and organizing it properly. I see major problems with the ICD10 system for ease of entry and even the list itself is not user-friendly or system-based.
  • see ICD10_Chooser_form#Adding_filter_blocks

• Julie will also have major difficulty deriving clear information from the ICD 10 and especially the CCI system.
  • Julie is fully involved in our move over to the new system, we will make sure she gets what she needs.

• I am currently working on a list of problems with both of these structures. Would appreciate feedback from collectors now that they are looking at ICD10 and CCI and trying to use them. -LKolesar 12:48, 2017 December 13 (CST)
  • Please put your problems on the wiki as they come up. That way we can start working on them rather than wait until you have a final list you are ready to bring forward. Also, other sites can know which concerns have already been expressed, so they don't need to re-hash things separately. Ttenbergen 16:22, 2017 December 23 (CST)/--Trish Ostryzniuk 11:03, 2017 December 24 (CST)

Problems seen with ICD 10 as of December 2017

ICD 10 Diagnosis and CCI review: (Laura Kolesar) Dec 8, 2017

Goals for our database should be: Data must be easily, efficiently and consistently entered. The data must be readily identifiable to retrieve for statistical purposes. The data must meet the needs of most users of the data. Right now, the ICD 10 codes can only be found by a word search method (cannot use type- recognition because the first words of items are often duplicated). This method, while possible, is very slow especially taking into account the alternate codes that have to be checked to ensure we have the “best” code.

• The idea to look on the wiki when trying to determine the best code is not an efficient system and wifi is not available at all times in all locations. Even when we become more familiar with the codes, the search word and pick system is way too slow for efficient data entry.

The linking system will be difficult for Julie to define the “primary diagnosis”, especially with multiple linkings. Also, in order to code many of the diagnosis, additional links must be added which also makes the process of data entry painstakingly slow and inefficient. Have all the people who use the database been given a list of the new codes and examples of the CCI codes to see if this data is sufficient for what they need? Some issues with ICD 10: I started writing a list of potential issues. This list is just the beginning and is in no way complete or thorough. Pathogens: Streptococcus: have Groups A, B, D, G but would need to list the names in each group for accurate data collection. Klebsiella: only have K.pneumoniae, not K. Oxytoca Serratia: no sub categories Candida: no sub categories Enterococcus: no sub categories Citrobacter: not in list Enterobacter: not in list

• Stenotrophomonas: not in list
• Acinetobacter: not in list

We need to be quite specific in our database in order to properly code VAP criteria, CLI etc. I don’t think the pathogen list for bacteria is extensive enough. No mixed option so would need to put in each pathogen separately. Is Dr Kumar OK with this list? If more than one pathogen is linked to a site infection and you also need to link resistance, how do you know which pathogen is the resistant one? They would all have the same priority #. Especially if “resistant drug NOS” code is used. Diagnosis not in list: Hospital acquired pneumonia. No pneumonia in the iatrogenic codes at all. Could put 3 combined codes: pneumonia, bacterial, the pathogen code, Iatrogenic problem NOS related to surgery or procedure but this is not necessarily true. Ruptured chordae tendinae Vasculitis: only options are limited to skin, NOS or necrotizing vasculopthy/vasculitis NOS Wegener’s is on list but Sjogruns syndrome is not Aneurysm in aorta: does not differentiate between abdominal and thoracic aorta. No option for dissection of aortic aneurysm (only arterial aneurysm or dissection NOS) Dr Manji wants to have the dissection of aortic aneurysms and the differentiation of abdominal and thoracic included.

• Are there other issues with ICD10 codes that would be a problem for those who utilize the database?

Right heart failure ( just cor pulmonale listed) No acute renal insufficiency code: can use ATN or “renal tubular disorder NOS” or “kidney or ureter disorder, NOS”, or “Kidney , infarction or ischemia” so many options, it may be inconsistently coded. Often the reason for the issue is not known right away. We do have a very general code: “Kidney, renal failure/insufficiency/uremia, unspecified as acute or chronic”. How helpful would that code be? Have the renal doctors who use our database seen the options for their codes? Is this the list they want? Intra-abdominal abscess: only options are intestine, anus, liver, peritonitis. No pancreatic abscess (do have pancreatitis and pseudocyst). Retroperitoneal hemorrhage or abscess not in list. Have “retroperitoneal area, diagnostic imaging abnormal” Does not differentiate between witnessed or unwitnessed arrest. Isolation is on the tiss and in tasks, do we need to include this in the diagnosis list? Cholecystitis, acute is a code. Related to gallstones is also a code, which we would have to combine to get the whole picture. However if the cholecystitis is not from gallstones, do we just put acute cholecystitis for acalculous type? How do we know the collector just did not miss the cause if it is not combined with another code? Other issues with ICD10:

• There is a risk of data collectors putting in a code like seizures NOS and then not combining it with the cause. Will this be a problem for our database to lose the cause of some diagnosis?
• Are there going to be set guidelines (numeric limits) for the metabolic derangements like hyper or hypo kalemia, natremia, etc. like we have now?

• Comorbid discussion:
• GFR is not readily available to use for the categorization of Chronic Kidney disease STAGES 1-5. I do not see it listed in the EPR labs. Is there an easy place to find this? They sometimes comment upon the stage or the GFR in the progress notes but this is not the ideal place to find this information. It is easy to miss comments in the notes. It is time consuming to find the Stage using this parameter.
• We currently put in a level of severity in our comorbids. Ie COPD- mild, mod or severe. Is it OK to not have this information?
• Are all previous conditions being put in this list? If so, this is a significant increase in numbers of codes required.
• How is chronic drug abuse put in comorbids if we don’t know the specific drug type? ICD10 requires the specific drug type.