While we were testing the next version of the ICD-10 MS-DRG grouper, we came across a real world record I want to share with you. It is a great example of how the detailed codes in ICD-10 will work in the data-driven world of healthcare, and why their less-detailed ICD-9 counterparts in ICD-9 don’t work well at all.
About 270,000 records into the MedPAR data we were using, our attention was drawn to the record of a severely ill patient who had pressure ulcers—bedsores—all over the place. From the ICD-9 codes on the record, we learned there were pressure ulcers at six separate sites —the upper back, lower back, hip, ankle, heel, and “other site.” We also learned there were stage 2 pressure ulcers and stage 4 pressure ulcers. Stage is essentially a measure of the depth of an ulcer, stage 4 being deepest—open craters with exposed bone or tendon or muscle.
Knowing the stage of a pressure ulcer is key information. The depth of an ulcer tells how sick the patient is, what treatment is indicated, and what the predicted costs and outcomes are. In MS-DRGs, stage affects MS-DRG assignment. Pressure ulcer staging has its own set of standard definitions, maintained by an organization created for this sole purpose—the National Pressure Ulcer Advisory Panel (NPUAP). In short, pressure ulcers are a big deal, and stage is an important thing to know.
Unfortunately, because of ICD-9’s imprecision we cannot correlate this patient’s six pressure ulcers with their correct stage. Pressure ulcer stage is only available in ICD-9 as an “adjunct code” –one of those standalone things ICD-9 started using when they realized they were running out of space for new codes. ICD-9 has a total of 15 pressure ulcer codes—six codes for pressure ulcer staging and nine codes for pressure ulcer site, with no way to correlate between the two.
The MedPAR record we were looking at had eight codes for recording this patient’s pressure ulcers: six codes for the various sites and two codes for the fact that one or more ulcers were stage 2 and one or more ulcers were stage 4. There could have been five ulcers at stage 4 and one ulcer at stage 2, or five ulcers at stage 2 and one ulcer at stage 4, or any combination in between. In clinical terms, this patient is anywhere from someone who has several manageable pressure ulcers to someone who is unquestionably at death’s door. From the ICD-9 data, we have no idea where this patient belongs on the severity of illness spectrum, and the codes give only the vaguest notion what it should take to treat this patient effectively and how much it should cost. As a specimen of data, it is pathetically inadequate.
Using ICD-10, we could have had the data we needed. There are 150 ICD-10 codes for pressure ulcers. Each code tells you both the site and the stage of any pressure ulcer, like this:
L89.624 Pressure ulcer of left heel, stage 4
Physicians do not have to change their habits at all to reap the benefit of these precise ICD-10 codes. They are already documenting both site and stage in ICD-9, they just aren’t getting good codes for their trouble. While our MedPAR record needed eight codes in ICD-9, it will only take six codes in ICD-10, and those six codes will link the site and stage of the ulcers together, something we are unable to do in ICD-9. If you were the physician treating this Medicare patient, and this patient had five stage 4 pressure ulcers, wouldn’t you want credit for that with Medicare and with the organizations out there publishing their analysis of the quality of your work?
ICD-9 was designed in an age when the industry was not using the ICD-9 codes for much, and it shows. Over the past 30 years, ICD-9 has become such a hodgepodge that it is severely limited as tool for extracting clear information from coded records. ICD-10, on the other hand, was designed and developed in the information age. Where important concepts like pressure ulcer site and stage are linked together, computers and their users can extract clear information from the coded records. Bottom line: Nonspecific ICD-9 codes make lousy coded data. Specific ICD-10 codes make good coded data. They can frequently be used to identify best clinical practices and reward the best providers for the work they do.
Rhonda Butler is a Senior Clinical Research Analyst with 3M Health Information Systems.