Tuesday, March 5, 2013

Educating the Enterprise about ICD10

On the list of exciting topics for enterprise-wide motivational meetings, ICD-10 is unlikely to rise to the top.  Starting off your Monday morning with an overview of 79,500 ICD-10-CM and 72,100 ICD-10 PCS codes can be about as exciting as watching grass grow.

Given the impact of ICD-10 on the revenue cycle, quality measurement, and risk adjustment, it's clear that we must educate all stakeholders about the importance of the initiative, the workflow challenges we'll face, and the need to improve our existing documentation.

We kicked off the BIDMC enterprise communication plan in January 2013 and in February, I presented this overview to all directors, managers, and supervisors.

They key take home messages were:
*ICD10 requires that we code and bill differently than we do today
*ICD10 is an FY13 Annual Operating Plan Goal
*The majority of BIDMC revenue is at risk
*Implementation and training will involve every department at BIDMC
*We must be fully live by October 1, 2014

I used several examples to build a lasting impression of ICD-10 such as

*If I go climbing in New Hampshire and crush my wedding ring finger in a rock, my ICD-9 code would be 915.8 "Other and unspecified superficial injury of fingers without mention of infection".   My ICD-10 code would be S60.445A
"External constriction of left ring finger, initial encounter"

*Since injury cause and location are coded separately, it is certainly possible to be struck by a turtle (W5922XA) in a squash court (Y92.311)

*There are initial encounters, subsequent encounters, and sequelae.    Important codes to know are

Bitten by Orca, initial encounter (W56.21XA)

Sucked into jet engine, subsequent encounter (V97.33XD)

and the Hitchcock classic

Bitten by birds, sequelae (W61.91XS)

All recognized the incredible training effort required to get clinicians and coders to apply ICD-10 properly.  More daunting is the need to improve clinical documentation so that it can justify the high degree of granularity possible with ICD-10

As we develop further training materials, posters, and broadcast communications, I'll share them on the blog.


T.Jentz, CPC said...

The climbing scenario example is an excellent illustration of the daunting need to improve clinical documentation so that it can justify the high degree of granularity possible. Depending on the documentation, full and complete ICD-10-CM coding of the initial encounter for this crushed finger injury could look more like this detailed picture:
• S60.445A "External constriction of left ring finger, initial encounter"
Official coding guidelines that apply to all codes in range S00 – T88 state, “Use secondary code(s) from Chapter 20, External causes of morbidity, to indicate cause of injury.” Therefore, we further code from external cause codes V00 – Y99 for cause of injury, activity during which injury occurred, place of injury and injured person’s work/military/volunteer status at the time of the event:
• W23.1XXA “Caught, crushed, jammed or pinched between stationary objects, initial encounter”
• Y93.31 “Activity, mountain climbing, rock climbing and wall climbing”
• Y92.828 “Other wilderness area as the place of occurrence of the external causes” e.g., mountain as opposed to desert (Y92.820) or forest (Y92.821)
• Y99.8 “Other external cause status” e.g., recreation or sport not for income.

GreenLeaves said...

my daughter, an engineer, asks whether ICD-10 can be used to indicate the degree of certainty for the selected diagnosis; i.e. is 100%, 75% or 50% certain that the Dx is correct.

GreenLeaves said...

Another follow on question after seeing T.Jentz's detail.
Are the additional codes for location and cause required by law or is it requested by public health or insurers?
I can see the immediate benefit of rendering a diagnosis but locating the appropriate codes for the causes could be onerous.

Brian said...

I struggle with how to provide real time feedback for documentation improvement in the emergency department, where patients are undifferentiated and time is precious. It seems you would either have to use very structured templates (which introduce precognition bias, force collection of information not relevant to emergency care, and annoy providers) or prompt for additional documentation after provider enters an ICD-10 diagnosis. The second option seems like a particularly bad idea as rushed docs would almost always assign the most generic top level diagnosis, if they assign one at all. How are others planning to confront this?

DrBazUK said...

Having wrestled with the level of misalignment that occurs in ICD-9, one can only boggle at the potential for codes to be inappropriately applied in ICD-10 unless the EHR/coding software provides more direction for those doing the coding. What is the future for semantic extraction of information from free text, allowing dictated notes to be crawled for subsequent coding without impacting on the flow of patient care?

Anonymous said...

There is a lot of software that uses Natural Language Processing to identify the proper ICD10 code. The vendor that I work for (I will omit to avoid any self-promotion) embeds the NLP into the system and reads both structured and unstructured information to provide accurate diagnosis and visit coding. It is working at multiple facilities with ICD9 and is slated to work with ICD10 as well. I believe there will be some additional clarification necessary for the providers once ICD10 is implemented, but it will be a huge advantage to having a human coder or a provider attempt to identify the correct codes.